Cataclysmic Variables in the First Year of the Zwicky Transient Facility

The Zwicky Transient Facility (ZTF) is a northern all-sky survey that uses the Palomar 48 inch telescope equipped with a 47 deg² field-of-view camera (Bellm et al. 2019a, 2019b; Graham et al. 2019; Masci et al. 2019). The advantages of the ZTF survey compared to past and ongoing northern sky surveys are the combination of large sky coverage that includes the Galactic plane, along with color information and increased temporal coverage, and the availability of nightly alerts. These aspects are ideal for finding variable stars, especially those that have non-periodic, erratic brightness changes such as cataclysmic variables (CVs), which consist of close binaries with mass transfer from a late main-sequence secondary to a white dwarf (see Warner 1995 for a review of all types of CVs). Combining the results from all-sky surveys along with astrometric data from Gaia will lead to the correct number and density of CVs throughout our Galaxy, thus ultimately constraining population models and close binary evolution scenarios.

The ZTF project uses 40% of the time for the public, 40% for partnerships, and 20% for Caltech. As part of the public portion, the available sky is sampled in ZTF g or r filters every three nights and the available Galactic plane with ∣b∣ ≤ 7° every night in both g and r with mean 5σ limiting magnitudes of 21 and saturation near 15 mag. Alerts on all variable objects are available every night. Commissioning took place in the latter part of 2017 with the official start of the 3 yr survey on 2018 March 18 and the first public data release DR1 occurred on 2019 May 8. DR1 is available from IPAC.¹⁷

The alerts are constructed from difference images between a reference field (consisting of a minimum of 15 images) and the new field. These alerts pass through the GROWTH Marshal (Kasliwal et al. 2019), which uses various filters constructed by participants to select specific types of variables and transients. One such filter was created to select candidate CVs. In the CVs with a non-magnetic white dwarf (B < 1 MG), the mass transfer accumulates in a disk before accreting onto the white dwarf. The amount of mass transfer and accretion determines what the light curve of a specific CV will look like. Those with relatively low transfer rate values will undergo a disk instability that results in a dwarf nova outburst (a rise in brightness of 2–9 mag within 1–3 days, the high brightness lasting for 1–15 days, and a subsequent return to quiescence), with a repetition timescale of weeks to months. The lowest mass transfer systems are those with the largest amplitudes, the longest times at outburst, and the longest recurrence time between outbursts (years). The highest rates of mass transfer result in nova-like (NL) systems, which have no outbursts but sometimes undergo several magnitude transitions between low and high accretion states that can last for weeks at a time. In systems with a magnetic white dwarf, the inner disk region can be channeled to the magnetic poles via accretion curtains (called intermediate polars, IPs) or in the highest field cases (termed polars), the mass transfer goes directly to the white dwarf magnetic poles, and no disk exists. Without a disk, there are no outbursts but high and low states of accretion can occur, resulting in several magnitude changes in their light curves on timescales of weeks. The colors of CVs are generally blue (ZTF g − r close to 0 and even bluer during an outburst) due to the contributions of the hot white dwarf and the accretion disk. However, long-period systems with K-type secondaries or those with magnetic white dwarfs and resulting cyclotron emission from the accretion pole can be redder in color.

To find CVs, a simple GROWTH filter was used to search for non-moving point sources with an amplitude change of 2 mags within a time span of 3 days and a color (from PanSTARRS) of g − r < 0.6. A real-bogus (rb) filter (calibrated from a zooniverse program of human classification on a large data set of ZTF images) was set to be low (0.1) to maximize findings. The filter began its full operation on 2018 June 5. With this filter, each night resulted in anywhere from 30 to 200 objects, the number mostly dependent on the weather and partly on the location of the observed fields. The resulting objects were then searched by eye to identify possible CVs. After many months of data, it became clear that it was most worthwhile to do the eye search if the rb was greater than 0.5. Future refinements to the filter and machine-learning classifiers being developed by other teams in the ZTF project will undoubtedly be able to decrease the amount of human interaction required. This paper presents the resulting previously confirmed CVs and CV candidates found using the simple GROWTH Marshal filter up to the time of the first public data release.

Each night, the combined g, r ZTF light curves of the filtered objects provided by the Marshal were scanned by eye to determine a possible dwarf nova outburst or a change in state of an NL system within the 30 day interval depicted by the Marshal. Likely candidates were saved and checked against known sources via SIMBAD, the Sloan Digital Sky Survey (York et al. 2000), the Catalina Real-time Transient Survey (CRTS; Drake et al. 2009, 2014; Breedt et al. 2014), Mobile Astronomical System of TElescopic Robots (MASTER; Lipunov et al. 2010), and All-Sky Automated Survey for Supernovae (ASAS-SN; Shappee et al. 2014). The saved sources continued to be monitored to help determine the correct classification. In many cases, this allowed full outbursts to be observed, in other cases, only rises or declines from an outburst or a high/low state were viewed. Other ZTF groups also transferred objects that passed their different filters (e.g., nuclear transients, SN, etc.) if they thought they might be CVs. When the weather allowed long stretches of good observations, it was common to end with 50–70 previously known CVs and/or good candidates per month. The most frequent contaminants were RR Lyrae and other large amplitude periodic variables that fell within the color range.

Follow-up spectra on some objects were obtained using several telescopes, sometimes with multiple spectra on the same object (27 systems with the Palomar 60-in, 5 with the Palomar 200-in, 10 with the Keck I0 m, 4 with the William Herschel 4.2 m (WHT), 14 with the Apache Point Observatory (APO) 3.5 m, and 2 with the Liverpool 2 m telescope) to confirm candidates. The presence of Balmer emission (from the decline or quiescent state) or the presence of He ii emission (indicative of a magnetic white dwarf or a very high mass transfer system) were used as confirmation criteria. Spectra obtained too close to an outburst generally showed only Balmer absorption lines and are indicative of an accretion disk but were not used as confirmation.

The Spectral Energy Distribution Machine (SEDM; Blagorodnova et al. 2018; Rigault et al. 2019), a low-resolution (50 Å) integral field spectrograph operating from 3650 to 10000 Å, obtained spectra on the Palomar 60-in telescope. An automatic pipeline reduced the data and uploaded it to the Marshal. The Double Beam Spectrograph (Oke & Gunn 1982) was used on the Palomar 200 in, the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) on the Keck, and the Spectrograph for the Rapid Acquisition of Transients (SPRAT) on the Liverpool telescope (Steele et al. 2004), and uploaded after reduction pipelines. A few spectra were obtained at medium resolution (3.3 Å) with the WHT using the Auxiliary-port CAMera (ACAM) spectrograph (Benn et al. 2008). The APO data were obtained with the Double Imaging Spectrograph using either low-resolution (2 Å) or high-resolution (0.6 Å) gratings to cover the Balmer lines and the data reduced using IRAF routines to calibrate the data via He, Ne, and Ar lamps and flux standards obtained during each night. The APO blue CCD suffered contamination problems throughout the year, so most of the data collected were only from the red side of the dichroic (5500–9000 Å).

The year-long scans of the GROWTH Marshal with the CV filter yielded 90 previously confirmed (from spectra or from the presence of a superhump outburst feature; Warner 1995) CVs, 218 strong candidates based on their light curves, and 21 objects that might be CVs but the data are too limited to tell for sure. Table 1 lists the previously confirmed objects, Table 2 the strong candidates, and Table 3 the remaining possibilities. Some sources were listed as candidates in CRTS or MASTER, but if they have not been confirmed, we placed them in Table 2. In the rest of this paper, we will generally refer to objects by their abbreviated R.A.(HHMM) and decl.(Deg) i.e., ZTF0014+59, but provide R.A.(HHMMSS) if needed to differentiate sources. The tables also provide the Galactic latitude, the range in magnitudes from outburst peak to quiescence, or from high-to-low accretion states observed by ZTF, the Gaia parallax and errors in mas (for those objects with a measurement that was more than three times the error), the number of normal outbursts and longer superoutbursts (SOBs) observed in the Marshal light curves, the interval of actual coverage by ZTF in days up to 2019 May 8, if the source is in the Sloan Digital Sky Survey (SDSS) footprint, the number of outbursts visible in CRTS if covered by that survey, any spectra obtained with the ZTF instruments or available from the SDSS or the literature, and any other relevant information.

The easiest systems to classify as CVs are those that show a long string of observations at quiescence along with one SOB that has a large amplitude (typically 5–9 mag) and a long duration (typically 3 weeks). A typical example (ZTF1727+23) is shown in Figure 1. The SOB often has a distinct shape, with a linear, slowly declining plateau following the peak magnitude that lasts for 1–2 weeks, then a steep decrease toward the quiescent magnitude. In a few cases, there are one or more rebrightening episodes during the decline (examples are ZTF2231+33 with one rebrightening and ZTF184159+37 with six possible rebrightnenings shown in Figure 1). These systems are known as WZ Sge stars (Warner 1995) or Tremendous Outburst Amplitude Dwarf Novae (TOADS; Howell et al. 1995). Modeling of TOADS has shown that they are the systems with the lowest accretion rates among dwarf novae (Howell et al. 1995) and the ones that population synthesis models (Howell et al. 1997, 2001) predict should be the most numerous in surveys of CVs. They should have evolved to the shortest orbital periods in their evolution during the lifetime of the Galaxy. Most dwarf novae that have orbital periods less than 2 hr show both SOBs and normal outbursts. The systems with larger accretion rates are generally those with longer orbital periods and with short recurrence times for outbursts, so they are readily found in sky surveys with coverage of weeks to months. An example of this kind of system is ZTF0613+06 in Figure 1. All objects in Table 2 show outbursts with one of the types described above. There also exist NL systems which do not show outbursts but do exhibit high and low states of accretion that last for many days. Two of these previously known systems, ZTF0854+39 (FR Lyn) and ZTF1631+69 (shown in Figure 2), were found in the Marshal, but none of the candidates in Tables 2 or 3 showed this behavior. The filter works best to detect outburst behavior and will need to be modified to pick up the generally slower transitions between high and low states.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

For the most part, the objects in Table 3 did not have enough data to unambiguously determine if their variability is due to a dwarf nova outburst or to some type of pulsational instability. As the survey proceeds, more observations may provide an unambiguous classification.

3.1. Spectroscopic Confirmations

Many of the SEDM spectra were obtained near outburst, when the accretion disk is dominant and producing a blue continuum or weak absorption lines. Thus, the low resolution usually only showed the continuum. The higher resolution and larger telescopes of the Palomar 200 in, Keck, WHT, and APO enabled better determination of Balmer and Helium emission lines as the systems evolved to quiescence. In addition, two candidates were found to have spectra available from SDSS. In total, 29 of the systems in Table 2 were able to be positively confirmed as CVs with noticeable Balmer or Helium emission lines (those designated with a small letter "e" in the Spec column of Table 2). Figure 3 shows the blue and red emission line spectra from the ZTF Marshal follow-up (coverage from 4000 to 9000 Å) while Figure 4 shows the red region from 5500 to 9000 Å) available from APO data when the blue side was not operational.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.2. The Galactic Plane

One of the unique features of ZTF is its frequent observations within 7° of the Galactic plane (see the spatial distribution of single-exposure epochs in DR1).¹⁸ Previous all-sky surveys such as CRTS (Drake et al. 2014) generally avoided latitudes within 10° of the Galactic plane due to crowding and large pixel sizes. Exceptions include a few targeted surveys, such as the Optical Gravitational Lensing Experiment (OGLE), that were successful in finding large numbers of dwarf novae in the fields toward the Galactic Bulge and the Magellanic Clouds (Mroz et al. 2015) and the All Sky Automated Survey (ASAS) which had large pixels and a faint limit of only 14 mag (Pojmanski 1977). The 1'' pixels of ZTF allow the nightly plane observations to provide variable sources that can be identified in the Marshal. Figure 5 shows the distribution of Galactic latitudes for the objects in Tables 1 and 2. The number of objects lying within 10° of the Galactic plane confirm the higher densities expected there, albeit not as large as the numbers found from OGLE (likely due to the limitations expressed in Section 5). Of the previously identified CVs (Table 1), only 23 (26%) are within 10°, while among the strong candidates (Table 2), the number is 98 (45%) and 7 (33%) for the possible candidates in Table 3. While the more frequent sampling of the plane by ZTF maps the shape of outbursts better than the three-day sampling, the detection rates of outbursts should be about the same over the sky as dwarf nova outbursts normally last longer than 3 days. Thus, the summed number of known plus candidate systems (30–40 per 10° bin in the plane) compared to ≤20 per bin out of the plane does imply a higher density of CVs, with most being new candidates.

Zoom In Zoom Out Reset image size

3.3. Absolute Magnitudes

The available Gaia parallaxes (Gaia Collaboration 2018) provide distances that enable meaningful absolute magnitudes and heights above the plane without relying on average absolute magnitudes at quiescence or outburst. However, the sample is small due to the current limits on the Gaia parallaxes. Table 1 (confirmed CVs) has 57 (63%) with good (S/N > 3) parallax measurements, while Table 2 has only 29 (13%) and Table 3 has 7 (33%). These numbers are consistent with the closest systems being found in previous discoveries and the new candidates having large distances and/or fainter brightness. Figure 6 plots the absolute magnitude of the objects from Tables 1 and 2. In the cases where ZTF did not get a detection for the quiescent magnitude, the upper limits mean that the absolute magnitudes will be even fainter than shown. Past results on absolute magnitudes of dwarf novae at quiescence (Warner 1987, 1995) have shown a range from 7.5 to 11, depending on the orbital period and the outburst recurrence time. Figure 6 shows that the majority of the ZTF sources are between 10 and 12, confirming that most systems are faint. To investigate whether ZTF is predisposed to a particular outburst type that would be related to mass transfer rate (such as the low mass transfer rate TOADS with infrequent, large amplitude outbursts; Howell et al. 1995), the absolute magnitudes are plotted versus detected outburst frequency in the interval covered by ZTF, in Figure 7 (top), as well as versus outburst amplitude (Figure 7 bottom). While the majority of the objects have outburst frequencies less than 0.02 (50 days), there is a large scatter at all brightnesses. This result is consistent with those from smaller data sets (e.g., Howell et al. 1995; Thorstensen et al. 2002) that showed a wide range of outburst behavior for similar orbital periods and luminosities. However, the outburst amplitudes show a clear increase as the absolute magnitudes become fainter. Of the 14 objects fainter than 12th mag, 8 would qualify as candidates for TOADS (Howell et al. 1995), with outburst amplitudes ≥6 mag, and modeled by Howell et al. (1995) with low mass transfer and low disk viscosity. These eight are all in the previously known CVs (Table 1), while candidate CVs all have brighter absolute magnitudes. This is likely related to the low frequency of outbursts evident in this faint magnitude range (Figure 7 top) and the short timescale of the ZTF survey.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Brief descriptions of the systems with interesting features in their spectra or light curves are provided below.

4.0.1. He ii Objects

Figure 3 shows four systems from Table 2 with strong He ii4686, or only helium lines, characteristics of AM CVn systems or those containing a magnetic white dwarf and/or very high accretion. ZTF2128+63 shows only lines of He and so is likely an AM CVn system. ZTF2123+15 has He ii stronger than Hβ and a prominent blue continuum, making it a strong candidate for either an IP or a member of the class of high mass transfer rate NL called SW Sex stars (Thorsensen et al. 1991; Hoard et al. 1998) that have orbital periods between 3 and 4 hr. ZTF0451+16 also shows strong Balmer and He i lines as well as relatively strong He ii, but without the blue continuum that is characteristic of a high mass accretion rate system, so is a candidate for a polar system. ZTF1647+43 is peculiar, as it has a very strong He ii line and a very blue continuum but very weak Balmer emission. Further data are needed to refine its classification.

ZTF1631+69 (Table 1) has been identified as a CV (Appenzeller et al. 1998), and is reported in the ROSAT and XMM catalogs, but there is no detailed study of this system available in the literature. The ZTF light curve shows the existence of high and low states, a common feature of high accretion rate SW Sex systems as well as those containing highly magnetic white dwarfs (polars and IPs). A series of five sequential spectra obtained over the span of an hour at APO on 2019 May 24 shows strongly doubled Balmer lines along with He ii, as well as large changes from one spectrum to the next (Figure 8). The velocities of the Hα and Hβ lines show a portion of a sinusoidal variation over 200 km s⁻¹ during the hour but the time span is too short to determine an orbital period. The doubled lines are a signature of an accretion disk while the high excitation could indicate a magnetic white dwarf. Further data are needed to ascertain if this is an IP system.

Zoom In Zoom Out Reset image size

4.0.2. Strong Balmer Lines

4.0.3. Peculiar Light Curves

ZTF1841+37 (Figure 1) shows an SOB followed by six normal outbursts or rebrightenings within 60 days. If this sequence repeats in future data, this could be a new ER UMa system. The cyclic behavior of SOBs interspersed with normal outbursts in this small group of CVs is thought to be a combination of a high mass transfer rate combined with a tidal instability present in short-period systems (Osaki 1996).

SIMBAD identifies ZTF1752+07 with V982 Oph and classifies it as a long-period variable candidate, but the ZTF light curve (Figure 9) and the blue colors at bright states are more consistent with a dwarf nova classification, as proposed by Antipin & Samus (2002).

Zoom In Zoom Out Reset image size

At this time, it is difficult to obtain a good estimate of how complete the Marshal is in finding all CVs. This first year was hampered by a late start of the Marshal filter, the loss of the month of October due to equipment improvements, significant weather losses in the winter months, and the lack of good reference images for all the fields. In addition, the CVs in any field may not have had a dwarf nova outburst or a change in low/high state during the months they were available in the sky. A rough estimate can be made using the results from SDSS spectra and the recent analysis of Pala et al. (2019) who used the overlap between Gaia and SDSS colors to determine the completeness of the CVs within 150 pc. The ∼300 available spectra from the SDSS Legacy program provided a lower limit estimate of the density of CVs as 0.03 deg⁻² (Szkody et al. 2003, 2011), and Pala et al. (2019) estimated that the spectral completeness of SDSS is ∼57%. ZTF covers about 62% of the sky (25,774 deg² so this should mean about 770 CVs should have been found, while only 308–329 were, which would be about 40% completeness). However, given the down time and weather, the actual usable time in the survey is probably closer to 6–9 months (390–580 CVs) which would be more like 53%–80%. Since the SDSS Legacy program was oriented toward high galactic latitudes to sample quasars, larger numbers for the stellar density are expected with the inclusion of low latitudes, so the lower percentages are likely more realistic. A better estimate will be possible as the survey proceeds and the lost months are re-observed and the reference fields are completed.

Using the GROWTH Marshal to filter nightly alerts from ZTF g and r light curves throughout the first year of operation resulted in the identification of 90 known CVs, 218 strong candidates based on the shape, amplitude and colors of the light curves, and an additional 21 potential candidates that require further data. Follow-up spectra obtained on a variety of 1.5–10 m telescopes allowed spectroscopic confirmation of 27 of the 218 strong candidates from Balmer emission lines, with an additional two with SDSS spectra. Unlike previous surveys, almost half of the new ZTF candidates are located within 10 degrees of the galactic plane, demonstrating the capability of the ZTF camera and software to discover objects in crowded fields. While only 13% of the strong candidates have available and significant Gaia parallaxes, most of their absolute magnitudes are consistent with the faint end of the CV distribution (10–12), similar to the CRTS. The outburst amplitudes increase with fainter absolute magnitudes in this range, with many of those fainter than 12 being good candidates for TOADs (Howell et al. 1995).

Four of the objects with spectra show high-excitation He ii or only helium lines and deserve further time-resolved spectra to determine their correct classification as either a system containing a magnetic white dwarf or an SW Sex system, or an AM CVn type. Hour-long time-resolved spectra of the known CV ZTF1631+69 shows strong He ii along with doubled Balmer emission lines, implying an IP origin. A previously identified long-period variable (V982 Oph) is more consistent with a dwarf nova classification as proposed by Antipin & Samus (2002).

While the available Marshal filter is not complete in finding all the CVs, it does demonstrate that even with several ongoing surveys, i.e., CRTS, ASASSN, and MASTER, there are many systems being missed, especially those in the Galactic plane. Completeness in any ground survey is difficult to obtain due to weather and software ability in crowded fields. Additionally, follow-up spectra of all the candidates is a time-consuming venture and will require increasingly large telescopes to obtain spectra at quiescent magnitudes. Unfortunately, classification is best near quiescence when the emission lines produce the most information from the intensity and shape as to the correct type of CV. Some compromise can be reached by obtaining observations midway from outburst to decline.

P.S. and B.D. acknowledge funding from NSF grant AST-1514737. A.Y.Q.H. is supported by a National Science Foundation Graduate Research Fellowship under grant No. DGE-1144469. M.C. is supported by the David and Ellen Lee Prize Postdoctoral Fellowship at the California Institute of Technology. M.L.G. acknowledges support from the DIRAC Institute in the Department of Astronomy at the University of Washington. The DIRAC Institute is supported through generous gifts from the Charles and Lisa Simonyi Fund for Arts and Sciences, and the Washington Research Foundation. We would like to thank occasional observers on the UW APO ZTF follow-up team, including Brigitta Sipőcz, James Davenport, Daniela Huppenkothen, Dino Bektešević Gwendolyn Eadie, and Bryce T. Bolin. The authors thank the Observatoire de la Côte d'Azur for support. This work was supported by the GROWTH project funded by the National Science Foundation under PIRE grant No. 1545949. Based on observations obtained with the Samuel Oschin Telescope 48 inch and the 60 inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. Z.T.F. is supported by the NSF under grant AST-1440341 and a collaboration including Caltech, IPAC, the Weizmann Institute for Science,the Oskar Klein Center at Stockholm University, the University of Maryland, the University of Washington, Deutsches Elektronen-Synchrotron and Humboldt University, Los Alamos National Laboratories, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, and Lawrence Berkeley national Laboratories. Operations are conducted by COO, IPAC, and The SED Machine is based upon work supported by NSF under grant 1106171. Some observations were made with the Apache Point 3.5 m telescope, which is owned and operated by the Astrophysical Research Corporation. The Liverpool Telescope is operated on the island of La Palma by Liverpool John Moores University in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias with financial support from the UK Science and Technology Facilities Council. The William Herschel Telescope is operated on the island of La Palma by the Isaac Newton Group of Telescopes in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias.

Facilities: APO:3.5m - , Hale - , ING:Herschel - , Keck:I - , Liverpool:2m - , PO:1.2m - , PO:1.5m - .