HO Puppis: Not a Be Star but a Newly Confirmed IW And-Type Star
Chien-De Lee, Jia-Yu Ou, Po-Chieh Yu, Chow-Choong Ngeow, Po-Chieh Huang, Wing-Huen Ip, Franz-Josef Hambsch, Hyun-il Sung, Jan van Roestel, Richard Dekany, Andrew J. Drake, Matthew J. Graham, Dmitry A. Duev, Stephen Kaye, Thomas Kupfer, Russ R. Laher, Frank J. Masci, Przemek Mroz, James D. Neill, Reed Riddle, Ben Rusholme, Richard Walters
aa r X i v : . [ a s t r o - ph . S R ] F e b Draft version February 22, 2021
Typeset using L A TEX default style in AASTeX62
HO Puppis: Not a Be Star but a Newly Confirmed IW And-Type Star
Chien-De Lee, Jia-Yu Ou, Po-Chieh Yu,
Chow-Choong Ngeow, Po-Chieh Huang, Wing-Huen Ip, Franz-Josef Hambsch,
3, 4,5
Hyun-il Sung, Jan van Roestel, Richard Dekany, Andrew J. Drake, Matthew J. Graham, Dmitry A. Duev, Stephen Kaye, Thomas Kupfer, Russ R. Laher, Frank J. Masci, Przemek Mr´oz, James D. Neill, Reed Riddle, Ben Rusholme, and Richard Walters Graduate Institute of Astronomy, National Central University, Jhongli 32001, Taiwan College of General Studies, Yuan-Ze University, Chung-Li 32003, Taiwan American Association of Variable Star Observers (AAVSO), Cambridge, MA, USA Vereniging Voor Sterrenkunde (VVS), Oostmeers 122 C, 8000 Brugge, Belgium Bundesdeutsche Arbeitsgemeinschaft f¨ur Ver¨anderliche Sterne e.V. (BAV), Munsterdamm 90, D-12169 Berlin, Germany Korea Astronomy and Space Science Institute (KASI), Bohyunsan Optical Astronomy Observatory (BOAO), Youngcheon, Gyungbuk38812, Republic of Korea Division of Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA Caltech Optical Observatories, California Institute of Technology, Pasadena, CA 91125, USA Kavli Institute for Theoretical Physics, University of California, Santa Barbara, CA 93106, USA IPAC, California Institute of Technology, Pasadena, CA 91125, USA
Submitted to ApJABSTRACTHO Puppis (HO Pup) was considered as a Be-star candidate based on its γ Cassiopeiae-type lightcurve, but lacked spectroscopic confirmation. Using distance measured from Gaia Data Release 2 andthe spectral-energy-distribution (SED) fit on broadband photometry, the Be-star nature of HO Pupis ruled out. Furthermore, based on the 28,700 photometric data points collected from various time-domain surveys and dedicated intensive-monitoring observations, the light curves of HO Pup closelyresemble IW And-type stars (as pointed out in Kimura et al. 2020a), exhibiting characteristics such asquasi-standstill phase, brightening, and dips. The light curve of HO Pup displays various variabilitytimescales, including brightening cycles ranging from 23 to 61 days, variations with periods between 3.9days and 50 minutes during the quasi-standstill phase, and a semi-regular ∼ INTRODUCTIONBe phenomena are the photometric and spectroscopic variability seen in the main-sequence luminous rapid rotators,known as Be stars, with a luminosity class III − V. In recent years, we have studied the evolutionary effect on theformation of Be stars in open clusters (Yu et al. 2015, 2016, 2018) using the Palomar Transient Factory (PTF; Law et al.2009) and the intermediate Palomar Transient Factory (iPTF; Kulkarni 2013). The Zwicky Transient Facility (ZTF;Bellm et al. 2019; Graham et al. 2019; Masci et al. 2019) came after iPTF, and its improved data can extend ourinvestigation on the variability of Be stars (Ngeow et al. 2019), especially for the Be stars and Be-star candidatesat the faint end ( m >
13 mag), which were largely excluded in previous works (e.g., in Labadie-Bartz et al. 2017).
Corresponding author: Chow-Choong [email protected]
Lee et al.
Together with accompanying time-series spectroscopic data, we have a new opportunity to explore the fundamentaltime-domain nature of Be stars.Here we report the photometric characteristics of HO Puppis (HO Pup, α J = 7 h m . s δ J = − ◦ ′ . ′′
28) as a result of our investigations of Be-stars variability with ZTF. HO Pup is listed as a Be star inthe
SIMBAD database (Manek 1997) and, hence, is included in our list of Be-star candidates, in which the classificationis based on its γ Cassiopeia (GCAS) type variability recorded by Samus et al. (2017) – a class of variable starsthat exhibit eruptive irregular variability that is not easily classified further. Some early literature even suggestedthat HO Pup was a possible type Ia supernova with V-band photometry varying between 12.7 mag and 14.2 mag(Kukarkin et al. 1971). As presented in Figure 1, two highly unusual ∼ . ∼ . HO Pup could be an additional member of theIW And-type stars (Kimura et al. 2020a, see also vsnet-chat 8162 from VSNET Collaboration).Given the ambiguous nature of HO Pup (either as a Be-star or a IW And-type star), we collected its light-curve dataas much as possible from archival catalogs, some ongoing surveys, and new dedicated observations. These collections oflight curves are presented in Section 2. In addition to light-curve data, we have also made spectroscopic and polarimetricobservations on HO Pup, described further in the same section. Analysis and results based on the observations arepresented in Section 3, in which we also present the first emission-line spectra of HO Pup – confirming its emission-linenature. In Section 4, we discuss the scenarios for the observed unusual light-curve behaviors of HO Pup, and wesummarize our findings in Section 5. OBSERVATIONS AND DATATime-series photometric data ranging from optical to infrared for HO Pup were collected from various survey catalogs,including the ZTF, the ASAS-SN, the Digital Access to a Sky Century @ Harvard (DASCH, Grindlay et al. 2012), thethird phase of the All Sky Automated Survey (ASAS-3, Pojmanski 2002), the Panoramic Survey Telescope and RapidResponse System 3 π survey (Pan-STARRS, Kaiser et al. 2010; Chambers et al. 2016), the Wide-field Infrared SurveyExplorer (WISE, Cutri et al. 2012), and observations available via the American Association of Variable Star Observers(AAVSO). These light-curve data were supplemented with dedicated observations taken at the Lulin Observatory inTaiwan. All of these light curves were merged in Figure 2 and listed in Table 1, which cover years from 1894 to 2020.We also summarized the light curves data in Table 2. Following the terminologies used to describe the light curvesof IW And-type stars (Kimura et al. 2020a), definitions of main features exhibited in the light curves, such as dips,brightenings and quasi-standstills, are demonstrated in the upper panel of Figure 1. Spectroscopic observations withlow and high spectral resolutions were conducted by P60/SEDM, BOAO/BOES, CFHT/ESPaDOnS, and P200/DBSP. More candidates for IW And-type stars can be found in VSNET Collaboration (http://ooruri.kusastro.kyoto-u.ac.jp/pipermail/vsnet-chat/). For example, V526 Ori (vsnet-chat 8474), MGAB-V1252 (vsnet-chat 8101), USNO-A2.0 1275-09782989 (vsnet-chat 8432), RXJ1831.7+6511 (vsnet-chat 24230), V2837 Ori (vsnet-chat 23538), LN UMa (vsnet-chat 24347), and EZ Vul (vsnet-chat 8266).
O Puppis
800 1000 120016151413 dipsbrighteningquasi-standstill1100 1120 1140 1160 118016151413
Figure 1.
The ZTF g-band (red), r-band (blue) and ASAS-SN V-band (black) light curves of HO Pup. The top panel shows aportion of the light curve across ∼
600 days, together with terminologies that are used in this work to describe the light curvefeatures: quasi-standstills (data points in the dotted box) are part of the light curves with magnitudes close to the mean value;brightenings are data points (in the dashed box) showing a brightening of ∼ . ∼ ∼ . Lee et al. −40000 −30000 −20000 −10000 01213141516 M a g DASCHASAS-3DASCHASAS-3 M a g WISE1WISE2ps_gps_rps_ips_zps_yWISE1WISE2ps_gps_rps_ips_zps_y
JD-2450000 M a g Lulin_rLulin_gLulin_iAAVSOZTF_rZTF_gASASSN_gASASSN_VLulin_rLulin_gLulin_iAAVSOZTF_rZTF_gASASSN_gASASSN_V
Figure 2.
The light curve of HO Pup across more than a century from optical to mid-infrared, consisting of data from AAVSO,ASAS-3, ASAS-SN, DASCH, Lulin (with SLT), Pan-STARRS (PS), WISE and ZTF. Each data point (for clarity, error barsare ignored) is noted with data source and filters accordingly. The observation time of follow-up spectroscopic and polarimetricobservations are marked with vertical lines containing LOT/TRIPOL2 (yellow line in 2018), BOAO/BOES (blue line in 2018),CFHT/ESPaDOnS (red line in 2019) and P200/DBSP (black line in 2020). More details can be seen in Figure 6 with a focuson the individual duration.
Additionally, polarization was measured in four different nights in late-October 2018 using the TRIPOL2 instrumentinstalled at the Lulin Observatory. 2.1.
Optical and Infrared Light Curve Data
ZTF is a northern-sky synoptic survey project ( δ > − ◦ ) carried out by the 1.2-m Samuel Oschin Telescope at thePalomar Observatory. With a large field-of-view mosaic CCD camera (47 deg with 1.0 ′′ pixel scale), the GalacticPlane can be scanned once a night with g and/or r filter. For HO Pup, 54 and 494 good quality measurements weretaken in the g- and r-bands, respectively, between 2017 December and 2018 December.ASAS-SN data are taken by a couple of quadruple telescopes located in both hemispheres. The mounted camerashave a 4.5-deg field of view with a pixel scale of 7.8 ′′ . This survey provides us the longest-time baseline from 2012 to2020 containing 960 V-band and 937 g-band measurements.The AAVSO responded quickly to get involved with the monitoring of HO Pup with V-band, right after our reportof this extraordinary event to the ZTF community. On 2018 October 13, the AAVSO began to collect data withextremely good coverage in the time domain, using a ML16803 CCD camera with a pixel scale of 2.06 ′′ equipped ona 0.4-m telescope at the Remote Observatory Atacama Desert (ROAD; Hambsch 2012) in San Pedro de Atacama, O Puppis Table 1.
Collected light curves with 28700 datapoints for HO Pup.
MJD Mag Uncertainty Band Source13146.3190 13 .
330 0 . B DASCH .
510 0 . B DASCH .
670 0 . B DASCH .
600 0 . B DASCH .
950 0 . B DASCH .
890 0 . B DASCH .
580 0 . B DASCH .
680 0 . B DASCH .
430 0 . B DASCH .
830 0 . B DASCH · · · · · · · · · · · · · · ·
Note —The entire Table will be available in its electronicform at the
SIMBAD archive.
Table 2.
Summary of light-curve data.
Band Database N MJD (start) MJD (end) UT Date (start) UT Date (end) Mean Mag. Std.B DASCH 260 13146.3 47616.4 11-14-1894 03-31-1989 13.472 0.463g ZTF 54 58107.4 58482.5 12-20-2017 12-30-2018 14.136 0.288g ASAS-SN 937 58220.2 58933.2 04-12-2018 03-25-2020 13.954 0.130g Lulin/SLT 313 58397.9 58832.7 10-06-2018 12-15-2019 14.128 0.089g Pan-STARRS 12 55593.4 56737.3 02-01-2011 03-21-2014 14.261 0.382V ASAS-3 217 51874.2 54862.2 11-26-2000 01-31-2009 13.793 0.310V ASAS-SN 960 55957.4 58450.6 01-31-2012 11-28-2018 13.860 0.324V AAVSO 10890 58404.3 58915.2 10-13-2018 03-07-2020 14.094 0.134r ZTF 494 58073.3 58476.4 11-16-2017 12-24-2018 14.283 0.311r Lulin/SLT 13786 58397.9 58938.6 10-06-2018 03-30-2020 14.190 0.099r Pan-STARRS 10 55940.4 56709.3 01-14-2012 02-21-2014 14.109 0.175i Lulin/SLT 309 58397.9 58832.7 10-06-2018 12-15-2019 14.293 0.067i Pan-STARRS 13 55195.4 56652.5 12-30-2009 12-26-2013 14.332 0.292z Pan-STARRS 14 55283.3 56641.5 03-28-2010 12-15-2013 13.311 0.336y Pan-STARRS 15 55276.2 56641.5 03-21-2010 12-15-2013 14.327 0.423W1 WISE 208 55298.7 58780.5 04-12-2010 10-24-2019 13.338 0.249W2 WISE 208 55298.7 58780.5 04-12-2010 10-24-2019 13.280 0.273
Chile. Every clear night, ROAD monitored HO Pup more than 5 hours with an average of 50 measurements, resultinga total of 10890 photometric data points. The typical uncertainty of the measurements is about 0.02 mag. Here weused 0.15 mag as our quality threshold after testing several values.Pan-STARRS used five filters , g p , r p , i p , z p and y p , to survey the sky. It has a wide-field 7-deg mosaic camerawith a pixel scale of 0.26 ′′ , mounted to the dedicated 1.8-m Pan-STARRS telescope, which is located at the HaleakaleObservatory in Hawaii. HO Pup was observed over a full five-year time span (2009-2014). In contrast to ASAS-SN orAAVSO, there are only a small number of observations. Two significant dips ( m > For simplicity, we refer to them as g, r, i, z and y in the rest of this paper.
Lee et al.
The 0.4-m SLT telescope, located at the Lulin Observatory in Taiwan, was used to perform near-simultaneous gri-band and intensive r-band monitoring in follow-up observations of HO Pup. Together with the equipped Apogee U42CCD, SLT images has a pixel scale of 0.79 ′′ . We used 77 reference stars located within 0.5 ◦ from HO Pup to performdifferential photometry and calibrated to the Pan-STARRS catalog. The SLT data provides significant support to thevariability investigation in the short-time scale and monitoring the color variation in the optical as the multi-banddata from the optical surveys mentioned earlier were not taken simultaneously or nearly simultaneous (hence no colorinformation).WISE is an all-sky survey mission that mapped the entire sky at 3.4, 4.6, 12, and 22 µ m (hereafter referred as W1,W2, W3, and W4) with spatial resolutions of 6.1 ′′ and 6.4 ′′ in W1 and W2 bands (Mainzer et al. 2011), respectively.The near-earth object WISE (NEOWISE) project is particularly designed to hunt for asteroids using the W1 and W2bands, and it provides infrared (IR) data over nine years from 2010 April to 2019 October. The cadence of the WISEobservations is about twice a year; each observation includes about a dozen of measurements over one day.Finally, we considered the utility of the DASCH and ASAS-3 light-curve data, but they were not used in subsequentanalysis. The DASCH project has digitized photometric measurements from nearly 500,000 glass plates across 100years. With a quality cut of 0.2 mag, 260 Johnson B-band magnitudes were selected for HO Pup from 1894 up to 1989as historical records. One possible dip is included (see subsection 3.2). In case of ASAS-3 data (Pojmanski 1997, 2002),we excluded them in the analysis because of the associated large pixel scale ( ∼ ′′ ). Therefore, issue of blending isunavoidable at the location of HO Pup. Nevertheless, we extracted 217 V-band light curves (grade A and B only) fromthe ASAS-3 archive. These light-curve data were measured using the smallest ASAS-3 aperture size (Pojmanski et al.2005, 2 pixels; corresponding to the MAG 0 in ASAS-3 catalogs).2.2.
Spectroscopic Data
Soon after the 2.5-mag dips found by ZTF in late 2017, we collected spectra of HO Pup based on the observationscarried out in 2018 by P60/SEDM and BOAO/BOES, in 2019 by CFHT/ESPaDOnS, as well as in 2020 by P200/DBSP.Due to the nature of queue observations for these telescopes and instruments, none of the spectra were taken duringthe 2.5-mag dip event of HO Pop, which did not occur again after mid-2018.The Spectral Energy Distribution Machine (SEDM, Ben-Ami et al. 2012; Ritter et al. 2014; Blagorodnova et al.2018; Rigault et al. 2019) is a low-resolution IFU (integral field unit) spectrograph, mounted on the robotic P60telescope at the Palomar Observatory (Cenko et al. 2006), providing efficient follow-up observations. The dispersionon the red side and blue side are 35 and 17.4 ˚A per pixel, respectively. Queued observations of SEDM were carriedout multiple times between 2018 October 19 and 2018 November 07, but only spectra from four nights were usable.The data were automatically reduced using the dedicated SEDM reduction pipeline (Rigault et al. 2019). The low-resolution P60/SEDM spectra can only be used to identify H α emission lines, hence they were excluded in this work.We have also obtained an optical spectrum of HO Pup using the Bohyunsan Optical Echelle Spectrograph (BOES,Kim et al. 2002) in long-slit mode, mounted on the 1.8-m Optical Telescope at the Bohyunsan Optical AstronomyObservatory (BOAO) in South Korea, on 2018 October 20. The observation was conducted using the grating 300V with3 ′′ slit width under 1.8 ′′ seeing, giving a spectral resolution of R ∼ , e.g., flat-fielding,wavelength calibration with FeNeArHe lamp, and flux calibration with the standard star G191-B2B. The normalizedspectrum is shown in the bottom panel of Figure 4. The H α emission line is clearly detected with an equivalent width(EW) of ∼ − . α emission variability, we took high-resolution echelle spectra of HO Pup using theESPaDOnS (Echelle SpectroPolarimetric Device for the Observation of Stars) mounted on the 3.6-m Canada-France-Hawaii Telescope (CFHT), with a spectral resolution of R ∼ − . ′′ ) conditions. CFHT provided a set of fullyreduced spectra from Libre Esprit, an automatic ESPaDOnS reduction package/pipeline (Donati et al. 1997, 2007).One of the brightenings or brightening events from 14.1 mag to 13.6 mag was fortunately well observed during our IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Researchin Astronomy, Inc., under cooperative agreement with the National Science Foundation.
O Puppis Table 3.
Polarization values of HO Pup
Date P r θ r P i θ i (%) ( ◦ ) (%) ( ◦ )2018 October 24 0 . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . observing runs. As shown in Figure 3, all H α emission lines are clearly seen, despite that the parts of the continuumwere barely observed.Finally, a single 600-second spectrum was obtained using the Palomar 200-inch Hale Telescope (P200) with theDouble-Beam Spectrograph (DBSP, Oke & Gunn 1982) on 2020 January 30. A 2 ′′ slit was used with the 600/4000grating in the blue arm and the 316/7500 grating in the red arm, providing R ∼ .Using the pipeline, we performed the standard bias corrections, flat-field corrections, wavelength calibration and flux-calibration. We also automatically combined the red and blue spectra using this pipeline. The reduced spectrum isdisplayed in the lower panel of Figure 4, in which the H α EW value is ∼ − . Polarimetric Data
In addition to the photometric and spectroscopic observations, we also took polarimetric data by using the TRIPOL2(second generation of the Triple Range Imager and POLarimeter) instrument (Sato et al. 2019), equipped on the LulinOne-meter Telescope (LOT), in 2018 October. TRIPOL2 can simultaneously take polarization images in Sloan g’-, r’-and i’-band with the half-wave plate rotating to four angles: 0 ◦ , 22 ◦ .
5, 45 ◦ and 67 ◦ .
5. We measured the flux at eachangle using aperture photometry following standard reduction procedures, and the Stokes parameters (I, Q and U)were then derived. The polarization percentage P = p Q + U /I and the position angle θ = 0 . U/Q ) canbe calculated from these Stokes parameters with a typical accuracy of ∆ P . . RESULTS3.1.
An Emission-Line Object
As mentioned in the Introduction, HO Pup could be considered as a Be star candidate or an IW And-type DN,without any emission-line or even spectral observations reported in the literature. With our spectroscopic follow-upobservations, the H α emission line is clearly seen in each of our observations including the spectra from BOAO/BOES,CFHT/ESPaDOnS and P200/DBSP (as shown in Figure 3 and 4). The H α emission EW is weak, e.g. ∼ − . α lines, other prominent hydrogen, helium, and metal features can be easily identified. During the bright-ening phase, HO Pup shows Balmer lines, from H β to H ζ , which are superposed on weak emission lines (see Figure 4). https://github.com/ebellm/pyraf-dbsp Lee et al.
JD-2450000 M a g I n t e n s i t y + O ff s e t Figure 3.
The top panel is same as Figure 2, except for the epoch with CFHT/ESPaDOnS observations. The bottom panelshows the CFHT/ESPaDOnS spectra of HO Pup taken from 2019 March 15 to 22, centered on the H α line (vertical dashed line).To improve the signal-to-noise ratio in the plot, the spectra were re-binned with a resolution of δλ ∼ α line, derived using the Astrolib PySynphot package (STScI Development Team2013). O Puppis JD-2450000 M a g JD-2450000 M a g Wavelength (Å) N o r m a li z e d I n t e n s i t y HβHγHδHεHζ He IHαtelluricNa IHe IIC III+N III Mg IIFe IIHe I Fe IHe ICr II+Ni I+ Fe I+Y IICr II Cr II
Figure 4.
The top panels are same as Figure 2, except for the epoch with BOAO/BOES (blue line) and LOT/TRIPOL2(yellow lines, see Table 3) observations on the top-left panel. The top-right panel is for the epoch with P200/DBSP spectroscopicobservation (black line), taken right after the photometric maximum of one of the brightening event. The bottom panel shows theBOAO/BOES spectrum (dashed blue curve) and P200/DBSP spectrum (solid black curve) resulted from a normalization usinga 5 th order polynomial. Both of the H α and H β lines are obviously seen in these spectra with either low or medium resolution,hence the emission nature of HO Pup was essentially confirmed. The spectral characteristics at the phase of brightening,including hydrogen Balmer lines, as well as helium, iron, nickel, sodium and chromium lines, are identified. Lee et al. −20 −15 −10 −5 0 5 10 15 20Day13.514.014.515.015.516.016.5 m a g E1E2E3E4E5E6E6E7E7E8
Figure 5.
The dip events observed in the 2017-2018 season, where the diamonds and crosses represent the ZTF and ASAS-SNdata, respectively. Two 2.5-mag dips, E6 and E7, are over-plotted with six other similar dips, including E1, E2, E3, E4, E5and E8. We aligned these events using their own fading slope, which have almost the same decreasing trends. With visualinspection, the minimum of the dips are chosen as the day of zero in the x -axis. The helium features show absorption lines He I I II II II I doublet5889 ˚A and 5893 ˚A; other emission features include Cr II II III +N III II +Ni I +Fe I +Y II α emission for HO Pup will be discussed further in Section 4.3.2. The Dip Events
Two 2.5-mag dips were observed in the ZTF r and g bands (13.5-16 mag; see Figure 1) in late 2017. During thesesignificant events, ASAS-SN also witnessed these two dips in the V band. These dips observed in both independentsurveys are well matched in the time sequence. When examining the long-term light curve taken from ASAS-SN, intotal there are eight dip events found between 2017 September and 2018 January, labeled as E1 to E8 in Table 4. Notethat the dip event E4 was barely observed by ASAS-SN on 2017 October 23 (JD=2458049.806). Within half a day,WISE happened to catch the IR counterpart of the same event down to 14.58 mag in W2 band (JD=2458049.435).Just like other dips in optical bands, the IR brightness drops significantly at this event when compared to the averagebrightness in WISE W2 band (13.280 mag).
O Puppis Table 4.
Faintest measurements of each dip event.
JD Date Surveys Mag error Band Event No.2427525.315 1934 / /
28 DASCH 15 .
170 0 . B B / /
25 Pan-STARR 15 .
119 0 . y B / /
30 ASAS-SN 15 .
526 0 . V B / /
30 Pan-STARR 15 .
646 0 . y B / /
15 ASAS-SN 15 .
576 0 . V B / /
10 ASAS-SN 15 .
017 0 . V E / /
25 ASAS-SN 15 .
175 0 . V E / /
07 ASAS-SN 15 .
224 0 . V E / /
23 ASAS-SN > . V E / /
23 WISE 14 .
581 0 . W E / /
30 ASAS-SN 15 .
178 0 . V E / /
16 ZTF 15 .
958 0 . r E / /
17 ASAS-SN 15 .
211 0 . V E / /
13 ZTF 15 .
975 0 . r E / /
13 ASAS-SN 15 .
05 0 . V E / /
12 ASAS-SN 15 .
041 0 . V E / /
20 ZTF 15 . . g A The E1-E8 dips have very similar shapes, especially their decreasing slopes. In Figure 5, we over-plotted these eventsall together to show their similar shapes. Note that only the E6 and E7 events were found to drop by ∼ . ∼ ∼ ∼ ∼
30 days. Using the entire light-curve coverage from E1to E8, the most likely event duration was found to be around 14.3 days estimated using phase dispersion minimization(PDM, Stellingwerf 1978) method.In the early epochs from 2011 to 2015, the dips B2 to B5 were individually recorded by Pan-STARRS or ASAS-SN,as seen in Table 4. A suspected dip, B1, was also recorded in the DASCH light curve around 1934. In addition, onemore dip, A1, was caught by ZTF during the 2018-2019 season. Before A1 occurred (2018 November 20), we have foursuccessful SEDM observations. Unfortunately, there is a long queue of observing requests on SEDM until the end ofour proposed observation run on 2018 November 27. Thus, we did not have any further observation and missed theopportunity to catch spectroscopic observations of this dip.3.3.
Semi-Regular 0.5-Magnitude Brightening Events
In addition to the dip events mentioned in the previous subsection, the long-term light curves of HO Pup also exhibitsemi-regular ∼ . ∼ ∼ . Lee et al. (a) (b) (c) (d) (e) (f)
JD-2450000 M a g Figure 6.
Same as Figure 2, but time scale enlarged for the different years of interest: (a) 2014-2015, (b) 2015-2016, (c)2016-2017, (d) 2017-2018, (e) 2018-2019 and (f) 2019-2020.
JD-2450000 m a g JD-2450000 m a g Figure 7.
Same as Figure 2, but time scale enlarged for the two ∼ . O Puppis . ± .
03, 31 . ± .
18 and 26 . ± .
34 days, respectively. Finally, in the recentepoch 2019-2020 (JD: 2458720 - 2458933), the estimated period turns out to be 60 . ± .
28 days. It is worth pointingout that these repeatable brightening events resemble the light curve of IW And-type stars (Kimura et al. 2020a).In addition to the brightening events in the optical, two brightening events were also observed in the mid-IR fromWISE. As shown in Figure 7, these brightening events in the mid-IR were almost varying together with the opticalASAS-SN V-band light curve, suggesting the entire continuum rises up during the brightening events.For cases in which spectroscopic observations are possible, we have continuously taken spectra every night usingCFHT/ESPaDOnS before the optical light curve reached its maximum brightness in one of the brightening eventsthat occurred in 2019 March (see Figure 3). We found that the strength, or equivalently the EW, of H α line decreaseswhile HO Pup became brighter, as shown in the spectra of the last two days in Figure 3.3.4. Color Variations
We investigated the color variations of HO Pup using several measurements. To construct the ( g − V ) color, weselected g-band data from ASAS-SN and V-band data from AAVSO. Each g-band point and its corresponding V-bandpoint were observed within 5 minutes. Also, we used gri-band data of the SLT telescope at the Lulin Observatory toconstruct the ( g − r ), ( r − i ) and ( g − i ) colors, where the gri-band images were always taken nearly simultaneously,e.g.,within 5 minutes. To quantify the color variation with brightness, we plotted the magnitudes as a function ofcolor (i.e. the color-magnitude diagram, CMD) in Figure 8. In all these optical colors, the blue data points were thosecorresponding the ∼ . g − i )colors shown in panel (c) of Figure 8. 3.5. Variabilities at Short Time Scale
In addition to the dip events and brightening events, as presented in sub-section 3.2 and 3.3, respectively, short-term variabilities were also found in the light curves of HO Pup during its quasi-standstill phase. For example inlate December of 2018, ZTF performed continuous cadence observations on the Galactic Plane that included HO Pup.Therefore, continuous r-band light curves taken within 2 hours were available from ZTF on 2018 December 22 and 23,with 280 and 136 measurements, respectively (see Figure 9). An hour-scale sinusoidal variability with ∼ . ∼ . ∼ Polarization Variation
We have also observed variations in polarization for HO Pup. Based on polarization level P and position angle θ measured from Lulin Observatory (see Table 3), a small and yet significant polarization was observed on 2018 October24 and 26. However, the polarization drops back to an insignificant value on 2018 October 25 and 2018 October 28.The observed variation in polarization suggested HO Pup exhibits intrinsic polarization. DISCUSSION4.1.
Spectral-Energy-Distribution Fitting
Prior to our work, classification of HO Pup in the literature was done via inspection of its light curves, as eithera Be star with GCAS-type variability (Samus et al. 2017) or as an IW And-type DN (Kimura et al. 2020a). Here,4
Lee et al. −0.3 −0.2 −0.1 0.0 0.1 0.2 g-V g m a g (a) −0.20 −0.15 −0.10 −0.05 0.00 0.05 g-r g m a g (b) −0.35 −0.30 −0.25 −0.20 −0.15 −0.10 −0.05 0.00 g-i g m a g (c) −0.16 −0.14 −0.12 −0.10 −0.08 −0.06 −0.04 −0.02 0.00 r-i r m a g (d) Figure 8.
The color-magnitude diagrams (CMD) of HO Pup. The representative error bars of colors and magnitudes are shownin the upper right corners of these four CMDs. The g- and V-band data in panel (a) were taken from ASAS-SN and AAVSO,respectively. The intense observations from both surveys allow us to separate the data points for those that occurred at the ∼ . we first estimated its spectral and luminous class using spectral-energy-distribution fitting together with distanceestimation based on the Gaia-Data-Release-2 (DR2) parallax measurement (1 . ± . d = 618 . O Puppis JD-2450000 +8.4742e313.813.914.014.114.214.314.414.5 m a g JD-2450000 +8.4752e313.813.914.014.114.214.314.414.5 m a g Figure 9.
Same as Figure 2, but with the time scale covering 0.6 days. Based on the ZTF data taken on 2018 December, somevariabilities in the timescale of hours are clearly observed. On December 22 (upper panel), we see a flipped S-shaped event(indicated by an arrow) lasting about an hour. On the next day, December 23 (lower panel), the ZTF light curve obtained overtwo hours shows a variability with sinusoidal-like profile. Even with lower cadence or photometric precision, similar variationsare also seen in the AAVSO light curve. queried within 1 ′′ radius using the VizieR photometry viewer (Ochsenbein et al. 2000). As shown in the Figure 12,the SED peaked between the NUV and B-bands, which indicates it is either a hot early-type star or a source with ahot component.Assuming HO Pup is a luminous B-type main-sequence star, fitting these broadband photometric data with a the-oretical SED model (Kurucz 1993) suggested that the spectral class of HO Pup has to be B1V, as demonstrated inFigure 12(a), with the corresponding extinction of A V = 0 . M V ∼ . m V = 13 . A V = 0 . M V is consistent with the known IW And-type stars, as summarized in Table 6. Therefore, thepossibility of HO Pup being a Be star is ruled out.As presented in previous sections, the IW And-type phenomenon can be seen in the light curves of HO Pup collectedin our work (as well as in Kimura et al. 2020a). Assuming HO Pup is a IW And-type DN, which, in general, consistof a white dwarf as the primary star, a (low-mass) secondary companion star, and a (hot) disk. In Figure 12(b), arepresentative theoretical spectrum for a white dwarf was shown at a distance provided by Gaia DR2, 618 pc, whichhas negligible contribution to the observed SED of HO Pup (the black points in Figure 12). Assuming HO Pup has ahot disk with a temperature of 11,000 K and a radius of 0 . Lee et al.
Period M u l t i - b a n d s T h e t a −1.00 −0.75 −0.50 −0.25 0.00 0.25 0.50 0.75 1.00 Phase M a g Figure 10.
Multi-band PDM analysis of the oscillatory variation at the timescale of days. A period of 3 . ± .
005 days isclearly detected (upper panel) from light curves based on the superposition of two independent observations. The bottom panelis the phased light curves folded with the detected 3.9-day period, in which the symbols are same as in Figure 2, revealing asinusoidally shaped light curve.
O Puppis Figure 11.
The variability of HO Pup down to timescale of hours. Panel (a) presents the SLT r-band photometric datataken from 2019 December to 2020 March. The blue curve represents the 3.9-day cycle fitted with a sinusoidal function. (seeFigure 10). To uncover the variability shorter than one day, this light curve has been subtracted with the sinusoidal functionof 3.9-day and the residuals are displayed in panel (b). After removing the 3.9-day cyclic variations, we found a new cycle of50-minute variations using Lomb-Scargle periodogram analysis, based on the light curve data taken from 2019 December to 2020January, as shown in panel (c). As can be seen from this panel, the variation signal is well above the false-alarm probability(FAP, indicated by the red and cyan dashed lines for the 95% and 99% FAP, respectively). The variability with 50-minute periodcan be easily seen in the phased light curve as presented in panel (e). However, the 50-minute variations were barely resolvedor undetected for the data collected between 2020 February and March, as evident in panel (d) and (f). Instead, complicatedvariations can be seen in the nightly chunks of light-curve data within this period of observations.
SED data. Given that the observed flux of the SED peaked between the NUV-band and the B-band, it turns outthat the observing time in the NUV band is the key to determining which phase that we fit in the observed SED, andit was found to be taken on 2012 February 02 (JD = 2455959.9603; according to the header information available fromthe corresponding GALEX image). Based on the two data points taken from ASAS-SN with V ∼ . before and after the NUV observation at the quasi-standstill phase, it is very likely that the NUV observationwas taken while HO Pup was at the quasi-standstill phase. For example, the B-band data ranged from 13 . − .
16 mag (see Table 5) suggesting one of the brightest data points might be takenduring the brightening phase. The next brightening event occurred on 2012 February 17, which were ∼ ∼
14 days durationbetween NUV observation and the next brightening was shorter than the typical duration of two consecutive HO Pup brightening events,which is longer than 20 days. Lee et al. −1.00 −0.75 −0.50 −0.25 0.00 0.25 0.50 0.75 1.00 log(μm) −16.0−15.5−15.0−14.5−14.0−13.5−13.0−12.5−12.0 l o g ( W / m ) (a) Fitting with a Be star spectrum at 16 kpc −1.00 −0.75 −0.50 −0.25 0.00 0.25 0.50 0.75 1.00 log(μm) −18−17−16−15−14−13−12 l o g ( W / m ) (b) Fitting with a hot disk black-body at 618 pc −1.00 −0.75 −0.50 −0.25 0.00 0.25 0.50 0.75 1.00 log(μm) −16.0−15.5−15.0−14.5−14.0−13.5−13.0−12.5−12.0 l o g ( W / m ) (c) Fitting with a hot sub-luminous star spectrum at 618 pc Figure 12.
Fitting of the observed SED of HO Pup (black points from broadband photometry; see Table 5) with varioustheoretical spectra. In panel (a), the blue curve is the theoretical spectrum for a B1V dwarf, taken from the Kurucz (1993)stellar atmosphere model, at a distance of 16 kpc with A V = 0 .
5. In panel (b), the light blue curve is a representative theoreticalwhite dwarf spectrum (with effective temperature of 12,000 K) adopted from Tremblay & Bergeron (2009) and Koester (2010),available at the Spanish Virtual Observatory, at a distance of 618 pc with A V = 0 .
1. The purple curve represents the sum ofthe white dwarf spectrum and a hot disk with radius of 0 . A V = 0 .
1. In panel (c), the green curve is a theoretical spectrum for a hot sub-luminous star withan effective temperature of 12,000 K, taken from the Kurucz (1993) stellar atmosphere model, assuming a distance of 618 kpcwith A V = 0 . Alternatively, the observed SED of HO Pup can also be fitted with a hot sub-luminous star as its secondary, asshown in Figure 12(c). We note that an early spectroscopic observation has suggested the prototype object, IW And,could be an early type dwarf or subdwarf (Meinunger 1980).4.2.
Photometric Characteristics
O Puppis Table 5.
Collected broadband photometric data for HO Pupused in SED fitting.
Band λ ( µ m) mag error referenceNUV 0.231 15.135 0.011 Morrissey et al. (2007)Bj 0.435 15.16 0.42 Lasker et al. (2008)B 0.444 14.233 0.57 Zacharias et al. (2012)B 0.444 13.8 0.43 Lasker et al. (2008)B 0.444 13.240 · · · Zacharias et al. (2005)Bf 0.468 13.86 0.44 Lasker et al. (2008)g 0.481 14.079 0.556 Geier et al. (2019)g 0.481 14.237 0.192 Heinze et al. (2018)g 0.481 13.818 0.273 Wolf et al. (2018)g 0.481 14.199 0.204 Chambers et al. (2016) G BP · · · Zacharias et al. (2005)r 0.617 14.085 0.013 Heinze et al. (2018)r 0.617 13.974 0.272 Wolf et al. (2018)r 0.617 14.043 0.008 Chambers et al. (2016)r 0.617 14.074 0.42 Zacharias et al. (2012)R 0.658 14.170 · · ·
Zacharias et al. (2005)G 0.673 14.083 0.009 Gaia Collaboration et al. (2018)i 0.752 14.391 0.003 Wolf et al. (2018)i 0.752 14.150 · · ·
Heinze et al. (2018)i 0.752 14.118 · · ·
Chambers et al. (2016)i 0.752 13.887 0.10 Zacharias et al. (2012)In 0.784 14.04 0.43 Lasker et al. (2008) G RP K s Two major phenomena of HO Pup, i.e., brightenings and deep dips, were found to be consistent with models proposedfrom two different teams – Kimura et al. (2020a) and Hameury & Lasota (2014). The former team explained the usualstatus of HO Pup (without presenting the deep dips) based on a tilted accretion disk model inspired by the ideaproposed by Kato (2019). The latter team reproduced both of the sudden brightening and deep dip by variations ofthe mass transfer rate.In the tilted accretion disk model, the accretion flow from the donor star will directly fall into the inner disk aswell as outer disk. With a steady supply of mass, the inner part of accretion disk of IW And-type stars can stayin a hot state in both quasi-standstill and brightening phases. Once the mass threshold was achieved in the outerdisk, brightening could occur from inside to outside of the disk, also known as inside-out brightening (Kimura et al.2020a; Court et al. 2020). The brightening event, in fact, shows up in the light curve (for example, see Figure 3 at JD2458565). Due to the instability of the accretion rate, the entire disk can be cooled since there is little mass left in thedisk, and hence the brightness can drop dramatically.0
Lee et al.
Table 6.
Absolute magnitude of IW And-type stars
Name Dist.(pc) A V m V M V IW And 835.1 0.152 a c a b a b a d a c b b a c a b Note —Distances were converted from Gaia DR2 paral-laxes (Gaia Collaboration et al. 2018). a Green et al. (2019) b Jayasinghe et al. (2018) c Henden et al. (2015) d Alfonso-Garz´on et al. (2012)
The performance of the thermal-viscous instability in the tilted accretion disk proposed by Kimura et al. (2020a) isconsistent with the brightening events and the 3.9-day cycle seen in the light curve of HO Pup. Indeed, Kimura et al.(2020a) stated the tilted disk models with 10 gs − were similar to the ”heartbeat-type oscillations” of HO Pup.Furthermore, in their highly-tilted disk models with accretion rate of 10 . and 10 gs − (the C1 and C2 model,respectively), the sawtooth-like pattern shown in the quasi-standstill phase resembling the light curve of HO Pupdisplaying the 3.9-day variations (See Figure 10). However, as pointed out by the referee, the 3.9-day variationsof the HO Pup light curve could also be caused by the precession of the tilted disk (the super-orbital modulation).Nevertheless, the tilted-disk models did not include eclipsing-like deep-dip events ( > ∼ . ∼ . O Puppis
Spectroscopic Characteristics
As shown in lower panel of Figure 4, the spectra of HO Pup displayed broad Balmer absorption lines up to H ζ , aswell as other helium and metal lines including He I , He II , Fe I and Mg I . Moreover, we found CV signatures withweak emission cores presented in most of the broad Balmer absorption lines from our P200/DBSP spectrum, whichwas taken at the brightening right after the maximum brightness. Szkody et al. (2013) observed two CVs, IW And& V513 Cas, in their brightening phase, in which their Balmer lines were nearly the same as in HO Pup. Beyond thehydrogen features, there is a double-peaked, bumpy and shoulder-like feature in between H β and H γ , which includethe lines of C III , N
III (known as Bowen fluorescence at 4640 ˚A) and He II (4731 ˚A). This is a unique blended-emission line that can be revealed especially during the phase of CV brightening. In the phase of quasi-standstill, theBowen features become much more insignificant as seen in the BOAO/BOES spectrum, along with shallower Balmerabsorption. Hessman et al. (1984) reported the time-resolved spectroscopic observations from brightening to quasi-standstill for a classical CV SS Cyg, showing the strength of the emissions from Bowen fluorescence declining to thecontinuum level. The similarity of HO Pup spectra features and these CV strongly suggested that HO Pup is indeedan IW And-type star. Two additional spectroscopic differences between brightenings and quasi-standstill phase notseen in the literature are the weaker H I I doublet showingup during the brightenings (See Figure 4).4.4. Polarimetric Characteristics
Based on the polarimetric observations taken during the quasi-standstill phase (see upper panel of Figure 4), HO Pupexhibited an intrinsic optical polarization variability, at which the polarized light included the light from both the starand the disk. Presence of magnetic field or disk scattering may account for the observed polarization. Despite thevariation of polarization level, the polarization percentage P at four different nights were all measured to be less than1% in our r and i band TRIPOL2 data. This level of polarization has been observed in other three DNe during theirquasi-standstill phase (Szkody et al. 1982). 4.5. Similarity to Be stars
Previously, HO Pup was considered as a Be star based on its GCAS-like light curves. However, those incomplete lightcurves were sparsely sampled in the past, as shown in Figure 2, which could be responsible for the mis classificationof HO Pup as a Be star. HO Pup has better observation coverage since 2016, therefore its DN nature is revealed byKimura et al. (2020a) from the dense sampling of the light curve.Since both of the Be stars and IW And-type stars are hot objects with hydrogen and helium lines, as well as exhibitirregular and abrupt photometric variations (e.g., due to presence of a disk), these two types of stars share a number ofcommon observational features. As evident from the case of HO Pup, an IW And-type star could be misclassified as Bestars based on limited observational evidence (and/or vice versa). In Table 7, we compared and summarized some ofthe important observational features of Be stars, IW And-type stars and HO Pup. For example, Be stars could displayregular dip events due to eclipsing as in the case of RW Tau with dips fainter by ∼ ∼ α line strength decreased from2 Lee et al.
Table 7.
The spectroscopic and photometric features shown in Be stars,(confirmed) IW And-type stars (excluding HO Pup) and HO Pup
Be Stars IW And Stars HO PupNumber ∼ a δM V · · · δM V < δ Sco) 15-100 days 23-61 daysbrightening duration 2-1000 days a few days ∼ a The first ”7” represents the confirmed IW And-type stars as mentioned in theIntroduction; the second ”7” is counting the candidates listed in footnote 1. − . − . α emission line strength and photometric brightnesswas also clearly observed in another Be star, V438 Aur (Labadie-Bartz et al. 2017). However, the obscured phase ofHO Pup is up to ∼
36 days which is far less than V438 Aur (more than 2000 days).Therefore, as summarized in Table 7, we cannot easily distinguish IW And-type stars from the field (luminous) Bestars without knowing the distance, because they share the same color, and nearly the same spectral features. Withthe low sampling of light curves and limited spectroscopic observations, neither photometric nor spectroscopic datacan be used to distinguish possible IW And-type stars from a list of faint Be stars. If we observe any other unknownIW And-type stars with a low-sampling light curve (as in the case of HO Pup before 2010), there is a large probabilitythat these stars would be classified as Be stars based on their GCAS-like light curve. Spectroscopic observations mightalso be difficult to separate these two classes of stars, unless the spectra were taken during the brightening phase whichcontain unique features of metal lines (such as Bowen fluorescence) for the CV. Since the brightening phases of a CVlast only a few days, as compared to Be stars that can last for longer than 10 days, there is a narrow observing windowfor taking the spectra for these objects. SUMMARYInstead of classifying HO Pup as a Be candidate, we confirmed that HO Pup is an IW And-type star based on thelight-curve pattern, spectroscopic characterizations along with the Gaia DR2 distance and SED fitting. As in otherIW And-type stars, light curves of HO Pup display various types of stochastic and quasi-periodic variations such asbrightenings, dip events and quasi-standstill phase. To shed light on the stellar physics behind the characteristic lightcurve of IW And-type stars, further well-covered spectroscopic monitoring along with intense photometric observationsare highly desirable.This work is partly supported by the Ministry of Science and Technology (Taiwan) under grants of 104-2923-M-008-004-MY5, 107-2119-M-008-014-MY2, 107-2119-M-008-012, 108-2811-M-008-546, and 109-2112-M-155-001. Wethank Abert Kong for pointing out the DASCH light curve. We also thank the discussions with Yi Chou, Wen-PingChen, Shih-Yun Tang, Michihiro Takami, Chi-Hung Yan, Paula Szkody and Melissa Graham on this work, as well assuggestions from an anonymous referee to improve the manuscript.Based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the PalomarObservatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundationunder Grant No. AST-1440341 and a collaboration including Caltech, IPAC, the Weizmann Institute for Science, theOskar Klein Center at Stockholm University, the University of Maryland, the University of Washington, DeutschesElektronen-Synchrotron and Humboldt University, Los Alamos National Laboratories, the TANGO Consortium ofTaiwan, the University of Wisconsin at Milwaukee, and Lawrence Berkeley National Laboratories. Operations are
O Puppis