A magnetic study of the δ Scuti variable HD 21190 and the close solar-type background star CPD −83°64B

ABSTRACT

1 INTRODUCTION

HD 21190 is a known δ Scuti variable star showing Ap star characteristics. Using time-series photometry obtained by the Hipparcos mission (Perryman et al. 1997), Koen et al. (2001) discovered a variability period of 3.6 h. According to these authors, the spectral classification of HD 21190 is F2 III SrEuSi, making it the most evolved Ap star known. González et al. (2008) used the Ultraviolet and Visual Echelle Spectrograph (UVES; Dekker et al. 2000) time-series of HD 21190 to search for pulsational line profile variations and were able to show that this star presents the best example of a δ Scuti star with moving bumps in its line profiles, which is characteristic for high-degree pulsations. The weak longitudinal magnetic field of HD 21190, 〈B_z〉_hyd = 47 ± 13 G, was detected by Kurtz et al. (2008) from FOcal Reducer low dispersion Spectrograph (FORS 1; Appenzeller et al. 1998) observations in spectropolarimetric mode. Re-observation of this target, 〈B_z〉_hyd = −237 ± 75 G and 〈B_z〉_all = −254 ± 60 G using FORS 2 (Hubrig & Schöller 2016) with the same setup as Kurtz et al. (2008) confirmed the magnetic field detection.

A recent astrometric analysis (Niemczura et al. 2017) using Gaia DR1 data (Gaia Collaboration et al. 2016) showed that the δ Scuti variable HD 21190 forms a physical binary system with CPD −83° 64B at an angular distance of about 19 arcsec. An atmospheric chemical analysis based on observations with the High Accuracy Radial velocity Planet Searcher polarimeter (HARPSpol; Snik et al. 2008) yielded the abundances of 24 chemical elements in the atmosphere of HD 21190 and 28 chemical elements in CPD −83° 64B. Niemczura et al. (2017) obtained the atmospheric parameters T_eff = 6900 ± 100 K and |$\log \, g=3.60\pm 0.20$| for HD 21190 and T_eff = 5850 ± 50 K and |$\log \, g=4.3\pm 0.1$| for CPD −83° 64B. The abundance analysis of HD 21190 showed that all the studied heavy and rare-earth elements appear overabundant, confirming the peculiar nature of this star. The abundance pattern for CPD −83° 64B appears mostly solar with modest enhancements of heavy and rare-earth elements.

With the most recent release of the Gaia DR2 data (Gaia Collaboration et al. 2018), it became possible to re-investigate the physical connection between the two stars. In the following, we present a brief analysis of these results. Further, while a few spectropolarimetric observations were obtained in the past for HD 21190, the presence of a magnetic field in CPD −83° 64B remained unexplored. We present our recently acquired multi-epoch high- and low-resolution spectropolarimetric observations for both objects.

2 GAIA DR2 RESULTS

Based on Gaia DR1 data, Niemczura et al. (2017) considered HD 21190 and CPD −83° 64B to be a wide binary system. However, the new Gaia DR2 (Gaia Collaboration et al. 2018) data clearly rule out a physical connection between these two objects and show that CPD −83° 64B is an unrelated background star appearing close to HD 21190. The Gaia DR2 data of HD 21190 and CPD −83° 64B, including parallaxes (and distance estimates by Bailer-Jones et al. 2018), proper motions, the radial velocity of CPD −83° 64B, and derived physical parameters, are listed in Table 1. These data are discussed in the following subsections.

2.1 Astrometry

The physical association of HD 21190 and CPD −83° 64B previously asserted by Niemczura et al. (2017) was mainly based on the common proper motion derived by these authors using the same three epochs of modern non-photographic positional measurements for both sources, including Gaia DR1 (Gaia Collaboration et al. 2016) positions. The corresponding proper motions agreed within their error bars (about ±2 mas yr⁻¹). Niemczura et al. (2017) also considered the proper motions published in two different catalogues involving Gaia DR1 data, namely the 5th United States Naval Observatory CCD Astrograph Catalog (UCAC5) of Zacharias, Finch & Frouard (2017) and the Tycho-Gaia Astrometric Solution (TGAS) of Lindegren et al. (2016). The difference between the TGAS proper motion of HD 21190 and the UCAC5 proper motion of CPD −83° 64B was found to be even smaller than 1 mas yr⁻¹.

Whereas the Gaia DR2 proper motion of HD 21190 deviates by less than about 0.7 mas yr⁻¹ from the above mentioned previous proper motion measurements, large discrepancies of 6–7 mas yr⁻¹ in RA and 2–5 mas yr⁻¹ in DE are found for CPD −83° 64B (see our Table 1 and table 1 in Niemczura et al. 2017). With the much higher precision of the Gaia DR2 proper motions (errors of the order of 0.1 mas yr⁻¹ for both stars), the two stars cannot be considered a common proper motion pair any longer.

The difference in their Gaia DR2 parallaxes is about 20 times larger than the formal errors (see Table 1). By simply inverting the parallaxes, we derive distances of 200.8 pc for HD 21190 and 248.9 pc for CPD −83° 64B, very similar to the distance estimates of Bailer-Jones et al. (2018) given in Table 1. Therefore, CPD −83° 64B lies about 50 pc further away than HD 21190. The new distances of both objects are much larger than the TGAS-based distance of HD 21190 (165 pc) that was used by Niemczura et al. (2017) as their assumed common distance.

We have also checked how reliable the astrometric data of both stars are with respect to various Gaia DR2 quality flags as discussed by Lindegren et al. (2018). Although the two stars lie beyond the 100 pc horizon used by Lindegren et al. (2018) for defining a high-quality sample (their ‘Selection C’) of 242 582 objects within 100 pc, they otherwise fulfil all their selection criteria applied to astrometric and photometric quality flags.

2.2 Radial velocity

A radial velocity measurement is reported in Gaia DR2 for only one of the two stars, CPD −83° 64B (see Table 1). This radial velocity is only slightly larger than the value of |$+$|19 ± 0.5 km s⁻¹ measured by Niemczura et al. (2017). On the other hand, a much larger radial velocity of |$+$|34.0 ± 1.3 km s⁻¹ was reported by the Radial Velocity Experiment (RAVE) DR5 (Kunder et al. 2017). As we found no hint on possible problems with the RAVE measurements of this star, it is possible that CPD −83° 64B is a close binary unresolved in Gaia DR2.

2.3 Physical parameters

Astrophysical parameters, including stellar effective temperatures, radii and luminosities, were derived for 77 million sources (Andrae et al. 2018) by using the Gaia DR2 photometry and parallaxes. These parameters are part of Gaia DR2. In Table 1, we show the Gaia DR2 effective temperatures and radii of HD 21190 and CPD −83° 64B. The effective temperatures are in good agreement with the values estimated by Niemczura et al. (2017) in their analysis of the chemical abundances using high-resolution HARPSpol spectra.

3 OBSERVATIONS

The high-resolution spectropolarimetric observations of HD 21190 and CPD −83° 64B were obtained with HARPSpol attached to ESO’s 3.6 m telescope (La Silla, Chile). HD 21190 was observed on 2016 June 15, 18, and 19, while CPD −83° 64B was observed on 2016 June 16. Each observation consisted of subexposures with exposure times varying between 23 and 32 min for HD 21190 and between 18 and 30 min for CPD −83° 64B. The quarter-wave retarder plate was rotated by 90° after each subexposure. All spectra have a resolving power of about 110 000 and cover the spectral range 3780–6910 Å, with a small gap between 5259 and 5337 Å. The reduction and calibration of these spectra was performed using the HARPS data reduction software available on La Silla. The normalization of the spectra to the continuum level is described in detail in Hubrig et al. (2013). A summary of all HARPSpol observations is given in Table 2. The columns list the heliocentric Julian date (HJD) for the middle of the subexposures, the exposure time, and the signal-to-noise ratio (S/N) of the spectra.

Due to the rather short pulsation period of HD 21190 of the order of 3.6 h discovered by Koen et al. (2001), the shape of the line profiles in the high-resolution spectropolarimetric observations is changing significantly over the pulsational cycle. Recently, Järvinen et al. (2017) showed that such high-resolution observations of pulsating variable stars frequently fail to deliver credible measurement results, if the whole sequence of subexposures at different retarder waveplate angles has a duration comparable to the timescale of the pulsation variability. To minimize the impact of pulsations on the magnetic field measurements, we asked for observing time with FORS 2 mounted on the 8 m Antu telescope of the Very Large Telescope. FORS 2 is a multimode instrument equipped with polarization analysing optics comprising superachromatic half-wave and quarter-wave phase retarder plates, and a Wollaston prism with a beam divergence of 22 arcsec in standard resolution mode. The exposure time with such an instrument is only of the order of a couple of minutes and no impact of pulsation on the field measurements is expected.

The acquired nine observations of HD 21190 and six observations of CPD −83° 64B summarized in Table 3 were spread between 2017 September and 2018 February. The columns list the modified Julian dates (MJD) for the middle of the exposure and the S/N of the spectra. We used the GRISM 600B and the narrowest available slit width of 0.4 arcsec to obtain a spectral resolving power of R ≈ 2000. The observed spectral range from 3250 to 6215 Å includes all Balmer lines, apart from H α, and numerous helium lines. Further, in our observations we used a non-standard readout mode with low gain (200 kHz,1 × 1,low), which provides a broader dynamic range, hence allowing us to reach a higher S/N in the individual spectra. The position angle of the retarder waveplate was changed from +45° to −45° and vice versa every second exposure, i.e. we have executed the sequence +45° −45°, −45° +45°, +45° − 45°, etc. up to 10 times for HD 21190 and four times for CPD −83° 64B. Using this sequence of the retarder waveplate ensures an optimum removal of instrumental polarization. The exposure time for observations of each pair of subexposures +45° −45° or −45° +45°, including overheads, accounted for about 4 min for HD 21190 and about 12 min for CPD −83° 64B.

4 MAGNETIC FIELD ANALYSIS USING HIGH-RESOLUTION HARPS SPECTROPOLARIMETRY

To study the presence of a magnetic field in HD 21190 and CPD −83° 64B, we employed the Least-Squares Deconvolution (LSD) technique (Donati et al. 1997). This technique combines line profiles (using the assumption that line formation is similar in all lines) centred at the position of the individual lines given in the line mask and scaled according to the line strength and sensitivity to a magnetic field (i.e. to a Landé factor). The resulting average profiles (Stokes I, Stokes V, and N) obtained by combining several lines yield an increase in S/N as the square root of the number of lines used and therefore in sensitivity to polarization signatures. The diagnostic N profiles are usually used to identify spurious polarization signatures. They are calculated by combining the subexposures in such a way that the polarization cancels out.

For HD 21190, we constructed a line mask containing 156 metallic lines with a mean Landé factor |$\bar{g}_{\rm eff} =1.32$| based on the stellar parameters T_eff = 6900 ± 100 K and log g = 3.6 ± 0.2 obtained by Niemczura et al. (2017). Since the subexposure times of 23–32 min of the HARPSpol observations of this δ Scuti variable are rather long, constituting a significant fraction of the pulsation cycle, the observed line profiles exhibit very strong pulsational variability, leading to pulsational wavelength shifts that affect the magnetic field measurements. The impact of the pulsations is clearly visible in Fig. 1, where we present the LSD Stokes V and I profiles calculated for individual subexposures.

Figure 1.

LSD Stokes V (top) and I (bottom) profiles calculated for the individual subexposures obtained in the HARPSpol observations of HD 21190 of three different nights. The numbers next to the profiles indicate the time lapse between the first and subsequent subexposures in minutes.

Open in new tabDownload slide

To exclude the pulsational impact on the magnetic field measurements, we decided to carry out the measurements of the line shifts between the spectra (I + V)₀ and (I − V)₀ in the first subexposure and the spectra (I + V)₉₀ and (I − V)₉₀ in the second subexposure separately (e.g. Hubrig et al. 2011; Järvinen et al. 2017). No definite magnetic field detection with a false alarm probability (FAP) < 10⁻⁵ was achieved in the HARPSpol observations of HD 21190. Only for the first subexposure obtained during the first observing night, we obtain 〈B_z〉 = 87 ± 33 G with a FAP of 5.8 × 10⁻⁴, indicating a marginal detection. We classify the magnetic field measurements making use of the FAP (Donati, Semel & Rees 1992), considering a profile with FAP < 10⁻⁵ as a definite detection, 10⁻⁵ ≤ FAP <10⁻³ as a marginal detection, and FAP ≥ 10⁻³ as a non-detection.

The chemical abundance analysis of CPD |$-83\boldsymbol {^{\circ }}$|64B by Niemczura et al. (2017) revealed that it is a G-type star with T_eff = 5850 ± 50 K and log g = 4.3 ± 0.1. The LSD Stokes I and V profile, together with the diagnostic null profile N, obtained using 730 metallic lines with a mean Landé factor |$\bar{g}_{\rm eff} =1.27$| are presented in Fig. 2. However, as Fig. 2 shows the LSD null profile N does not appear flat, indicating a possible presence of variability also in this star. Similar features were also observed in the null profiles of other G-type stars (see e.g. Marsden et al. 2014). Therefore, to minimize the impact of pulsations on the magnetic field measurements, we analysed the presence of a magnetic field on each individual subexposure in a similar way as for HD 21190. In Fig. 3, we present the LSD Stokes I and V profiles of CPD −83° 64B for each individual subexposure and the comparison of the individual LSD Stokes I profiles with the average LSD Stokes I profile. The distinct line profile variability, probably caused by the presence of pulsations, is clearly visible in the bottom panel displaying differences between the Stokes I profiles computed for the individual subexposures and the average Stokes I profile. As we show later in Section 6, the rotation period of CPD −83° 64B is too long for variability to be caused by spots.

Figure 2.

LSD Stokes I, V, and N profiles of CPD −83° 64B combining all four subexposures.

Open in new tabDownload slide

Figure 3.

Top: LSD Stokes V profiles of CPD −83° 64B calculated for each individual subexposure. The subsequent profiles are shifted vertically for better visibility. Middle: individual LSD Stokes I profiles compared with the average Stokes I profile (red line). The profiles are also shifted vertically for better visibility. Bottom: differences between the Stokes I profiles computed for the individual subexposures and the average Stokes I profile. The dashed lines were added to guide the eye. The time difference in minutes to the first subexposure is given next to the profiles.

Open in new tabDownload slide

Only marginal field detections were achieved for the individual subexposures with field strengths varying from −103 to 218 G, the best FAP value attained was 4 × 10⁻⁵. To summarize, our high-resolution HARPSpol observations with long exposure times are certainly affected by the presence of pulsations and do not allow us to make any definite conclusion on the presence or absence of a longitudinal magnetic field in either star.

5 MAGNETIC FIELD ANALYSIS USING LOW-RESOLUTION FORS 2 SPECTROPOLARIMETRY

A description of the assessment of the presence of a longitudinal magnetic field using FORS 1/2 spectropolarimetric observations was presented in our previous work (e.g. Hubrig et al. 2004a,b, and references therein). Rectification of the V/I spectra was performed in the way described by Hubrig, Schöller & Kholtygin (2014b). Null profiles N are calculated as pairwise differences from all available V profiles so that the real polarization signal should cancel out. From these, 3σ-outliers are identified and used to clip the V profiles. This removes spurious signals, which mostly come from cosmic rays, and also reduces the noise. A full description of the updated data reduction and analysis will be presented in a separate paper (Schöller et al., in preparation, see also Hubrig et al. 2014b).

The longitudinal magnetic field was measured in two ways: using the entire spectrum including all available lines, or using exclusively hydrogen lines. Furthermore, we have carried out Monte Carlo bootstrapping tests. These are most often applied with the purpose of deriving robust estimates of standard errors (e.g. Steffen et al. 2014). The measurement uncertainties obtained before and after the Monte Carlo bootstrapping tests were found to be in close agreement, indicating the absence of reduction flaws. The results of our magnetic field measurements, those for the entire spectrum or only for the hydrogen lines are presented for HD 21190 and CPD −83° 64B in Tables 4 and 5, respectively. Distributions of the mean longitudinal magnetic field values as a function of MJD obtained for both stars are presented in Figs 4 and 5.

Figure 4.

Distribution of the mean longitudinal magnetic field values of HD 21190 as a function of MJD between 2007 and 2018. The black open (already published) and filled (new) circles represent FORS 2 observations, whereas the red filled square is used for the HARPSpol observation on 2016 June 15.

Open in new tabDownload slide

Figure 5.

Distribution of the mean longitudinal magnetic field values of CPD −83° 64B as a function of MJD in 2017 and 2018.

Open in new tabDownload slide

For HD 21190, the values for the longitudinal magnetic field 〈B_z〉_all vary roughly between −250 and 250 G, with the measurement 〈B_z〉_all = 230 ± 38 G at a significance level of 6.1σ. The magnetic field appears to be stronger in CPD −83° 64B although the uncertainties of the field determination in this component are higher due to its faintness with m_v = 10.8. The highest value for the longitudinal magnetic field, 〈B_z〉_all = 509 ± 104 G, is measured at a significance level of 4.9σ.

6 PERIOD ANALYSIS

Based on the new distance estimates presented in Table 1 and the apparent magnitudes of the stars (the ASCC-2.5 catalogue; Kharchenko 2001, V = 7.604 for HD 21190, and V = 10.815 for CPD −83° 64B), we obtain for both targets the absolute visual magnitudes M_V = 1.10 ± 0.05 and M_V = 3.85 ± 0.03, respectively. After applying the corresponding bolometric corrections of 0.028 and −0.067 (Flower 1996), we obtain the luminosity log (L/L_|$\odot$|) = 1.44 ± 0.01 for HD 21190 and the luminosity log (L/L_|$\odot$|) = 0.38 ± 0.01 for CPD −83° 64B. Using the Stefan–Boltzmann law with the effective temperature values derived from high-resolution spectra by Niemczura et al. (2017), we derive stellar radii of 3.69 ± 0.07 and 1.51 ± 0.03 R_|$\odot$|. The obtained radii are identical within the uncertainties with the radii estimated using the Gaia DR2 data presented in Table 1. With these radii and the |$v$| sin i values of 74 ± 3 and 3.1 ± 0.7 km s⁻¹ from Niemczura et al. (2017), we obtain upper limits for the rotation period: P_rot ≤ 2.5 ± 0.1 d for HD 21190 and P_rot ≤ 25 ± 4 d for CPD −83° 64B.

Magnetic Ap and Bp stars usually exhibit photometric, spectral, and magnetic variability with the rotation period. Unfortunately, in our study, the number of available magnetic field measurements is only 11 (including two previously published measurements) for the primary component and only six for the secondary component. We nevertheless tried to detect the most probable rotation period for HD 21190 by calculating the frequency spectrum. Our analysis was performed using a non-linear least squares fit to the multiple harmonics utilizing the Levenberg–Marquardt method (Press et al. 1992). For each trial frequency, we performed a statistical F-test of the null hypothesis for the absence of periodicity (Seber 1977). The resulting F-statistics can be thought of as the total sum including covariances of the ratio of harmonic amplitudes to their standard deviations, i.e. a S/N.

The highest peak in the periodogram calculated for the magnetic field measurements including the previously published values corresponds to a period of 1.02471 ± 0.00045 d with an FAP of 2.89 × 10⁻⁹. The distribution of the mean longitudinal magnetic field values phased with this period is presented in Fig. 6 in the left-hand panel. Using the above estimated stellar radius |$R= 3.69\pm 0.07\, \text{R}_{\odot }$|⁠, |$v$| sin i = 74 ± 3 km s⁻¹, and the rotation period |$P_{\rm rot} = 1.024\, 71\pm 0.000\, 45$| d, we obtain |$v$|_eq = 182 ± 4 km s⁻¹ and an inclination angle of the stellar rotation axis to the line of sight i = 24 ± 2°.

Figure 6.

Left-hand panel: distribution of the mean longitudinal magnetic field values of HD 21190 phased with the rotation period |$P_{\rm rot} = 1.024\, 71\pm 0.000\, 45$| d. The black filled squares represent FORS 2 measurements using the hydrogen lines and the red filled circles the measurements using the entire spectrum. Previously published measurements are presented by open symbols. Right-hand panel: distribution of the mean longitudinal magnetic field values of HD 21190 phased with the rotation period |$P_{\rm rot} = 1.849\, 33\pm 0.000\, 29$| d.

Open in new tabDownload slide

In the case when only the new measurements of the magnetic field are used, the highest peak in the periodogram corresponds to a period of |$1.849\, 33\pm 0.000\, 43$| d with an FAP of 2.01 × 10⁻⁶. The distribution of the most recent mean longitudinal magnetic field values phased with this period is presented in Fig. 6 in the right-hand panel. This period leads to |$v$|_eq = 101 ± 2 km s⁻¹ and an inclination angle of the stellar rotation axis to the line of sight i = 47 ± 1°.

We note that due to the rather low number of magnetic field measurements, our results on the period search are only tentative and should be verified by additional spectropolarimetric observations.

7 DISCUSSION

The recent atmospheric chemical analysis of Niemczura et al. (2017) based on HARPSpol observations confirmed the presence of chemical peculiarities in HD 21190. According to the authors, HD 21190 is currently the only known case of an Ap star with a detected magnetic field that shows δ Scuti pulsations. CPD −83° 64B appeared mostly solar with modest enhancements of heavy and rare-earth elements.

Although previous studies showed that HD 21190 exhibits δ Scuti pulsations, the possible presence of pulsations in CPD −83° 64B was discovered in this work exclusively through spectroscopic observations. Pulsations in solar-like stars are mainly acoustic modes excited by turbulent convection like in the Sun, and are predicted for all main-sequence and post-main-sequence stars cool enough to harbour an outer convective envelope (e.g. Di Mauro 2016). Future spectroscopic and photometric monitoring will be worthwhile to characterize this short-time variability in CPD −83° 64B.

Whereas the previous astrometric analysis (Niemczura et al. 2017) using Gaia DR1 data (Gaia Collaboration et al. 2016) showed that the δ Scuti variable HD 21190 forms a physical binary system with CPD −83° 64B at an angular distance of about 19 arcsec, the more recent Gaia DR2 results clearly indicate that the two stars are not physically associated.

The initial idea to search for magnetic fields in HD 21190 and CPD −83° 64B originated under the aspect that both targets are physical components of a wide binary. A detection of magnetic fields in components of a wide binary could possibly be related to the most important issue related to the nature of magnetic chemically peculiar stars, namely the origin of their magnetic fields, which is still not properly understood. A scenario for the origin of the magnetic fields of Ap stars, in which these stars result from the merging of two lower mass protostars, was suggested by Ferrario et al. (2009). In this scenario, the binaries that are observed now were triple systems earlier in their history. If the hypothesis that the magnetic field originates in dynamos driven by violent mixing of stellar plasma when two stars merge is valid, one would expect that only very few magnetic massive stars are found in close binaries. Indeed, close binary or multiple systems with the particular combination of a magnetic and a non-magnetic star or two magnetic stars are very rare. Among the studied binary systems with A and late B-type primaries, only two systems, HD 98088 and HD 161701, are known to have a magnetic Ap star as a component (Babcock 1958; Hubrig et al. 2014a). Should a wide third component be essential to lead to mergers and thus to Ap stars with a magnetic field, this wide component should still be physically bound to the magnetic Ap star and be detectable as a wider binary component (Mazeh & Shaham 1979). Unfortunately, the new astrometric analysis of HD 21190 and CPD −83° 64B showed that these two objects are not such a wide pair.

ACKNOWLEDGEMENTS

We are grateful for very helpful comments from an anonymous referee. Based on observations made with ESO Telescopes at the La Silla Paranal Observatory under programme IDs 079.D-0241 (PI:Briquet), 191.D-0255 (PI: Morel), 097.C-0277 (PI: Hubrig), and 0100.D-0137 (PI: Järvinen). This work has made use of the VALD database, operated at Uppsala University, the Institute of Astronomy RAS in Moscow, and the University of Vienna. EN acknowledges the Polish National Science Center grant no. 2014/13/B/ST9/00902.

REFERENCES