Evidence that short period AM CVn systems are diverse in outburst behaviour
C. Duffy, G. Ramsay, D. Steeghs, V. Dhillon, Mark R. Kennedy, D. Mata Sánchez, K. Ackley, M. Dyer, J. Lyman, K. Ulaczyk, D. K. Galloway, P. O'Brien, K. Noysena, L. Nuttall, D. Pollacco
MMNRAS , 1–11 (2021) Preprint 9 February 2021 Compiled using MNRAS L A TEX style file v3.0
Evidence that short period AM CVn systems are diverse in outburstbehaviour
C. Duffy, , ★ G. Ramsay, D. Steeghs, V. Dhillon, , M. R. Kennedy, D. Mata Sánchez, K. Ackley, , M. Dyer, J. Lyman, K. Ulaczyk, D. K. Galloway, , P. O’Brien, K. Noysena, L. Nuttall, D. Pollacco Armagh Observatory and Planetarium, College Hill, Armagh, BT61 9DB, UK Department of Physics, University of Warwick, Gibbet Hill Road, Coventry, CV4 7AL, UK Department of Physics and Astronomy, Hicks Building, The University of Sheffield, Sheffield, S3 7RH, UK Instituto de Astrofísica de Canarias, 38205 La Laguna, Tenerife, Spain Jodrell Bank Centre for Astrophysics, Department of Physics and Astronomy, The University of Manchester, Manchester, M13 9PL, UK School of Physics & Astronomy, Monash University, Clayton VIC 3800, Australia OzGrav: The ARC Centre of Excellence for Gravitational Wave Discovery, Clayton VIC 3800, Australia School of Physics and Astronomy, University of Leicester, University Road, Leicester, LE1 7RH, UK National Astronomical Research Institute of Thailand, 260 Moo 4, T. Donkaew, A. Maerim, Chiangmai, 50180, Thailand Institute of Cosmology and Gravitation, University of Portsmouth, Dennis Sciama Building, Burnaby Road, Portsmouth, PO1 3FX, UK
Accepted 2021 February 05. Received 2021 February 05; in original form 2020 December 18
ABSTRACT
We present results of our analysis of up to 15 years of photometric data from eight AM CVn systems with orbital periodsbetween 22.5 and 26.8 min. Our data has been collected from the GOTO, ZTF, Pan-STARRS, ASAS-SN and Catalina all-skysurveys and amateur observations collated by the AAVSO. We find evidence that these interacting ultra-compact binaries showa similar diversity of long term optical properties as the hydrogen accreting dwarf novae. We found that AM CVn systems inthe previously identified accretion disc instability region are not a homogenous group. Various members of the analysed sampleexhibit behaviour reminiscent of Z Cam systems with long super outbursts and standstills, SU UMa systems with regular, shortersuper outbursts, and nova-like systems which appear only in a high state. The addition of TESS full frame images of one ofthese systems, KL Dra, reveals the first evidence for normal outbursts appearing as a precursor to super outbursts in an AM CVnsystem. Our results will inform theoretical modelling of the outbursts of hydrogen deficient systems.
Key words: accretion, accretion discs – binaries: close – stars: dwarf novae – surveys
AM CVn systems are at the shortest tail of the cataclysmic variable(CV) period distribution, consisting of a white dwarf which accretesmatter from a hydrogen deficient low mass companion that is eitherfully or partially degenerate. The orbital periods of these systemsare in the range ∼ −
65 mins (Solheim 2010). The first AM CVnwas identified in 1967 (Smak 1967, 1975) and by 2018 the knownpopulation consisted of 56 systems (Ramsay et al. 2018), with afew more in subsequent years. It is understood that the accretion isdriven by gravitational wave (GW) radiation as the binary systemloses angular momentum, as first proposed by Kraft et al. (1962).This GW radiation is predicted to be detectable by LISA (Stroeer& Vecchio 2006) which intends to employ AM CVn systems asverification sources. The expected GW signal can be predicted fromthe distance, derived from Gaia parallax data, and the componentmasses, derived from photometric and spectroscopic observations, ★ Contact e-mail: christopher.duff[email protected] (Kupfer et al. 2018). This utility in part drives the desire to improveour understanding of AM CVn systems.In addition to these aforementioned properties understood to becommon to all AM CVn systems, some systems have shown outburst-ing behaviour. These events are characterised by a sudden increasein brightness of 3-4 magnitudes which often leads to their discovery.It has previously been predicted and subsequently observed that sys-tems with orbital periods between approximately 22 and 44 minutesexhibit outburst behaviour (Ramsay et al. 2012). This correlates andagrees well with predictions of the behaviour of the accretion discsof AM CVn systems, and how they vary with orbital period. Thosesystems with the shortest periods are expected to have hot, small andstable accretion discs and those with the longest periods are expectedto have cool, large and stable accretion discs. Those systems withintermediate periods are expected, however, to have an unstable discwhich can act as the source of outbursts (Solheim 2010; Kotko et al.2012).Outbursting AM CVn systems share a number of characteristicswith hydrogen dominated dwarf nova, whereby they both exhibit so-called normal and “super outbursts” (SO). SO last several weeks or © a r X i v : . [ a s t r o - ph . S R ] F e b C. Duffy et al. more, e.g. ∼
10 days in KL Dra or ∼
29 days in V803 Cen, and resultin the AM CVn achieving maximal brightness. Normal outbursts arefar shorter events lasting between one and five days, with a peakbrightness typically 1 magnitude dimmer than their superoutburstcounterparts (Cannizzo et al. 2012). SO generally have more complexprofiles consisting of a sudden increase in brightness that is bluer incolour (when compared to the system in quiescence; Hameury et al.2020), which gradually decreases over the duration of the outburst.Regardless of the duration of the SO, they often exhibit a small dip inbrightness soon after maximum brightness after which the brightnesscan increase again (Ramsay et al. 2012).Some systems also spend extended periods, sometimes severalyears, in “high states” similar to the “standstills” that are observed inZ Cam systems. Such states have been attributed to the mass accretionrate lying close to the critical value for accretion disc instability (Katoet al. 2001). From a theoretical standpoint the behaviour of thesesystems is described by the disc instability model (DIM; Meyer &Meyer-Hofmeister 1983). This model outlines how different massaccretion rates, a key property driving AM CVn behaviour, and disctemperatures arise; which states are stable, and how the unstablesystems can exhibit outbursts. Crucially DIM also predicts that themass accretion rate is strongly correlated to the orbital period of asystem, and thus is a marker for its expected behaviour (Solheim2010).KL Dra was one of the first AM CVn systems that was seen toexhibit regular outbursts (Wood et al. 2002). Follow up work byRamsay et al. (2010) found that KL Dra showed SO approximatelyevery 60 days, which lasted for about two weeks. Conversely, despitetheir almost identical orbital periods, Kato et al. (2000) found thatCR Boo appeared to show a cyclic behaviour moving between highand low states on a timescale of 46.3 days. In order to gain a betterunderstanding of these differences we have selected a sample ofAM CVn systems with orbital periods within 2.5 minutes of thatmeasured for CR Boo and have collated photometric data from arange of All Sky Surveys and amateur measurements taken over aperiod of 15 years. This data allowed us to study the long termoutbursting properties of these AM CVns with many having datafor the entire 15 years considered – although other systems such asCX361were observed less often. We further used
TESS full frameimages to study KL Dra which revealed key detail in the SO for thefirst time.
We combined data from a number of all sky surveys as well asdata gathered by amateurs and made available through the AAVSOInternational Database (Kafka 2020). The surveys which we usedwere Catalina (Drake et al. 2009), ASAS-SN (Kochanek et al. 2017),Pan-STARRS (Flewelling et al. 2020), ZTF (Bellm et al. 2019) andGOTO (Steeghs et al. 2021, in preparation) which brought togetherphotometric data from 2005 through to 2020. Data from Catalina,Pan-STARRS, ASAS-SN and ZTF was accessed through their re-spective public data releases. Additional, more recent, ZTF data wasalso acquired through the Lasair transient broker (Smith et al. 2019).Data from GOTO was accessed through the in-collaboration databasewhich offers access to photometry shortly after observation.As the data we used came from a range of telescopes we had to takecare to select data from suitable band passes to ensure compatibility.For Pan-STARRS, ASAS-SN, and ZTF this was achieved using data
Table 1.
Key observational parameters of the AM CVn systems in this studywhich have periods between 22.5-26.8 minutes. Periods taken from Ramsayet al. (2018); all other data, including uncertainties, established in this work.We show the orbital period, P orb , the SO recurrence time, T
Rec , the SOamplitude, the SO duration, T
Dur and not the presence or absence of normaloutbursts.
System P orb T Rec
Amplitude T
Dur
Normal(mins) (days) (mag) (days) OutburstsPTF1 J1919+4815 22.5 35 ± ± ± (cid:88) ASASSN 14CC 22.5 29 ± ± ± (cid:88) CX361 22.9 - - - × CR Boo 24.5 49 ± ± ± (cid:88) KL Dra 25.0 60 ± ± ± (cid:88) PTF1 J2219+3135 26.1 67.3 ± ± (cid:88) V803 Cen 26.6 74 ± ± ± (cid:88) PTF1 J0719+4858 26.8 - 2.9 ± × taken in their respective g band ( ∼ − r band data which is used illustratively in some figuresbut was not part of the analysis. Similarly for GOTO the data usedwas taken in either the L band ( ∼ − V band which hassignificant overlap with these other filters ( ∼ − V band filter or no filter but calibrated with a V band zeropoint was used. Figure 1 shows subsections of the light curves ofeach of the sources which we have studied in this paper combiningeach of these data sets.In surveys with quality flags associated with each photometricmeasurement, such as in Pan-STARRS, these were used to identifybad data which should not be included. Manual filtering, by e.g.inspecting source images, was also performed with measurementswhich were considered to be the most flawed also removed, e.g.those subject to poor seeing. This conservative approach, usuallyremoving fewer than 5 data points, ensured that measurements whichcould be part of a real feature are not inadvertently removed whilstalso avoiding introducing data that maybe indicative of a feature thatdoes not exist. In order to search for any periodicity in systems, such as regularSO, an Analysis of Variance (AOV) period search (Schwarzenberg-Czerny 1989; Devor 2005) from the vartools suite developed byHartman & Bakos (2016) was employed. AOV, which is well suitedto determining the orbital periods of eclipsing binaries, works on theprinciple of folding and binning the data on different periods andidentifying the period which minimises the difference between datapoints in the same bin from successive cycles. The results of theseperiod searches, from successive observing seasons, provided ourrecurrence times for SO. In addition to this, since there were differ-ences in the sampling frequency of observations, visual inspectionwas used to verify results. The
Lightkurve (Lightkurve Collabo-ration et al. 2018) python module was used to fold the photometricdata according to the identified period. Doing this made it possiblenot only to see the overall cyclic behaviour in an observing seasonbut it also allowed for cycles to be compared as per Figure 2 andFigure 3. Consequently it was possible to develop an understandingof the differences in those features which were present or absent indifferent cycles and observing seasons.
MNRAS , 1–11 (2021)
M CVn Accretion States Figure 1.
Representative subsections of each of the light curves of the sources under investigation. Blue points from GOTO, red and green points from g and rband ZTF data, pink points from ASASSN, brown points from Pan-STARRS and purple points from AAVSO. MNRAS000
Representative subsections of each of the light curves of the sources under investigation. Blue points from GOTO, red and green points from g and rband ZTF data, pink points from ASASSN, brown points from Pan-STARRS and purple points from AAVSO. MNRAS000 , 1–11 (2021)
C. Duffy et al.
Using plots such as these we were able to study the SO cycles ofthe systems in detail visually identifying the duration and amplitudeof SO, the existence of normal outbursts and any dips in SO bright-ness. Table 1 shows the key properties determined for each systemin our study; the mean recurrence time, amplitude, SO duration andthe presence of normal outbursts, the errors on these mean valueswere calculated from their respective standard deviations. This highlevel comparison shows that despite all of these systems having rel-atively similar orbital periods, the property considered to be keyin determining their behaviour, they appear to show quite divergentbehaviour.What follows is a discussion of the observed behaviour of thosesystems with periods within 2.5 minutes of CR Boo as seen in ourdata. Note that when individual observing seasons are discussed inthis paper they are referred to by the year which forms the majorityof the observing seasons.
The behaviour that is exhibited by PTF1 J1919+4815 shows twodistinct states which it cycles between. The SO which marks thistransition lasts on average 17 ± ± ∼
13 days (Levitan et al. 2014). From the dataavailable for this systems it appears that this is a consistent behaviourseen in all observing seasons considered. The SO show a minor dipand then increase in brightness a few days after onset. The low statefor this system is close to the limiting magnitude for the surveys weused; however as we see a number of short lived events of magnitudeless than that seen in SO we are confident that the system does exhibitnormal outbursts, which have also been observed in previous studies.
ASASSN 14CC was discovered in 2014 by ASAS-SN (Jayasingheet al. 2019) and was subsequently observed extensively by amateurastronomers in the proceeding months; consequently there is a wealthof data from 2014, however following this, amateur observation ap-peared to cease leaving only limited data from All Sky Surveys.Nevertheless the data from 2014 is sufficient to allow discussion ofthe observed behaviour. ASASSN 14CC clearly shows distinct highand low states with abrupt changes from one to the other. The 2014observations show 7 distinct high states with 7 corresponding lowstates. ASASSN 14CC appears to spend the majority of its time ina high state with these lasting for a mean duration of 29 ± First discovered by Wevers et al. (2016), CX 361 was identified asan AM CVn with an average apparent i band magnitude of 17.35with short term variability of 0.2 mag over a period of 15 years asseen in the OGLE III and IV surveys. In addition to this a long termmodulation, where the brightness appears to trend downwards wasalso observed. During the 15 years of observations it was not seento exhibit any outbursts and was further identified as being in a highstate from optical spectroscopic observations. This marks the systemas highly unusual as it lies outwith the previously defined zonesof stability for accretion discs and would be expected to outburstregularly.In the data which we considered here CX 361 has only been ob-served by GOTO and Pan-STARRS each providing a small numberof observations. The crowded nature of the field (CX 361 is locatedin the galactic bulge) makes extracting photometry challenging; how-ever our data appears to be consistent with the earlier observationsof Wevers et al. (2016). Although we have only limited data wesee some evidence, particularly in Pan-STARRS, that the previouslyidentified gradual decrease in brightness has continued. Much of thedifferences which we see in our data may be attributed the use a dif-ferent filter and observations crowded fields being highly sensitive tothe effects of variable seeing conditions making it harder to resolvesingle sources. Plots such as those shown in Figure 2 and Figure 3 allowed us toextract the period of the SO cycle and study individual cycles fordifferent observing seasons of CR Boo. Using these plots it waspossible to make a detailed study of observing seasons from 2005through to 2019. From this study it was clear that there are two typesof seasonal behaviour; those which only show a continuous highstate and those which alternate between a high and a low state. Thoseseasons which only show a high state make up ∼
40% of the seasonsstudied and show an absence of other features in their light curvesalthough occasionally they show short-lived dimming events. All butone of these seasons are seen to occur successively with other seasonsof this type, indeed the early time of the 2018 season also exhibitsthis behaviour. The extended duration that this state is maintained forsuggests that it is a stable configuration for CR Boo’s accretion disc.Those seasons which show both high and low states are far morefeature rich in comparison. In those seasons that do exhibit a SO,e.g. at MJD = 58640, we see this to occur on a period that is centredupon 48 days. In those seasons which show low states, we see frequent(every 6 to 10 days) and sudden brightening events which we identifyas normal outbursts similar to those seen by Ramsay et al. (2012).We expect that these probably occur in every low state but due totheir brevity have escaped observation in some low states as a resultof the sampling. These observations compare favourably to those ofKato et al. (2000) who saw the shift from high to low state occuron a period of 46.3 days and normal outbursts occur every 4-8 days.Within our observations, although broadly consistent, there is somedrift season to season in observed SO recurrence time, with someseasons more in line with previous observations.The other prominent feature frequently seen is a dip in brightnessimmediately after the onset of the SO i.e. the start of the high state,after which the brightness again increases. A clear example of this canbe seen in cycle 2019-4 shown in Figure 3. Although our observationsdo not show this feature in every season it is likely that the dip is a
MNRAS , 1–11 (2021)
M CVn Accretion States Figure 2.
Upper: AOV periodogram of the CR Boo 2019 observing season.Middle: light curve of CR Boo in the 2019 observing season. Lower: phasefolded light curve of the 2019 observing seasons. The different colours rep-resent a different cycle and phase 0 corresponds to the mean transition timeto the high state for each of the cycles. We see a remarkable consistency fromcycle to cycle in the high state behaviour, in addition to numerous normaloutbursts during the low state. feature common to SO and that poor data collection, most likely poorsampling has resulted in the missed observation of it in some SO.
KL Dra has been observed by various All Sky Surveys which gaveus access to several years worth of photometric data. It consistentlyexhibits SO which last between 7 and 15 days which recur on atimescale of approximately 60 days with an amplitude of 3.6 mag.This is consistent with the findings of Ramsay et al. (2012) whofound a variable recurrence time centred on 60 days with durationsin the same range that we observed. Similar to many of the other AMCVn systems in our study, the SO which we see also exhibit a dipin brightness shortly after onset. We discuss these SO in more detailusing
TESS observations in §3.1.During the low states between successive SO we see 3-4 normaloutbursts which appear to last approximately 1 day. Our observationsof these normal outbursts are in line with the current understandingthat KL Dra does exhibit normal outbursts, with previous work e.g.Kotko et al. (2012) coming before this discovery and so buildingmodels without accounting for their existence.
The behaviour observed in PTF1 J2219+3135 appears to be similarto that discussed above for KL Dra. Though data does not have
Figure 3.
The light curve of the CR Boo SO cycle 2019-4 extracted using theinferred period as shown in Figure 2 (48.5days). particularly good sampling, a pattern similar to that seen in KL Dracan be identified. Using the AOV method we were able to extracta recurrence of time of approximately 67.3 days which we couldtenuously confirm by visual inspection. Due to the sampling of thedata it was not possible for us to determine an outburst duration,nor was it possible to confirm the presence of any dip in the SO.Furthermore our data does not show convincing evidence of normaloutbursts in PTF1 J2219+3135.
V803 Cen displays behaviour which is very similar to that seen inCR Boo; exhibiting both high and low states as well as extendedperiods where it is only found in a high state, which make up themajority of observations. As is common with other systems that ex-hibit low states, V803 Cen shows normal outbursts during the lowstates. The SO that V803 Cen exhibits also show a dip in their bright-ness immediately after their maximum brightness as is seen in othersystems. The periodicity of the transition between the high and lowstates is somewhat hard to ascertain due to the apparent preferenceshown in many observing seasons towards the high state which limitsthe number of observed transitions. However in those seasons withtransitions, the recurrence time, whilst varying by several days, liesat approximately 74 days. Of the SO that we did see, the majorityhad a duration, as determined by visual inspection of the lightcurves,centred on 28 days, although a few were seen to have durations asshort as only 20 days.
PTF1 J0719+4858 is one of the dimmest sources considered in thiswork regularly being observed at magnitude 20. As a result of thisit is more sparsely observed that many of the other sources we havestudied, nevertheless there is sufficient data to allow for a broaddiscussion of the behaviour observed. Although not densely sampledthe system appears to show distinct high and low states which ittransitions between; however with the data available it not possible todetermine the timescale of these transitions, which state is preferredby the system, or if the SO contains any feature of note. Additionally
MNRAS000
MNRAS000 , 1–11 (2021)
C. Duffy et al. the system does not appear to show any normal outbursts in the lowstates occurring the between the SO.
The
Transiting Exoplanet Survey Satellite (TESS) satellite waslaunched on 18 𝑡ℎ April 2018 into a 13.7 day orbit and has foursmall telescopes that cover a 24 × ◦ area of sky (see Ricker et al.(2015) for details). It has now covered both the southern and northernecliptic hemispheres apart from a strip along the ecliptic plane. Eachsector of sky is observed for around 28 days. In the first two yearsof observations, around 20,000 stars were observed with a cadenceof 2 minute in each sector. However, the full frame images which wemade use of only give a cadence of 30 minute for each sector. KL Drais located in the continuous viewing zone near the northern eclipticpole (data from sector 15 were not available). Ideally we would havegathered data of more than a single system, however CR Boo, theonly other system which could have been seen in more than a singlesector, repeatedly fell in a chip gap or just out of field. One of the challenges of obtaining photometry of KL Dra using
TESS is the large pixel scale (21 (cid:48)(cid:48) /pixel) coupled with the fact that itis spatially nearby (6 (cid:48)(cid:48) ) a galaxy (it was originally misclassified as asupernova – SN 1998di Jha et al. 1998; Wood et al. 2002). To obtainphotometry we used packages eleanor (Feinstein et al. (2019) and lightkurve (Lightkurve Collaboration et al. 2018) which correct forthe background and remove instrumental effects which are present inthe data. Both produced similar results but we present the photometryproduced by lightkurve . We extracted 15 ×
15 pixel postage stampscentred on KL Dra. We took the flux from a 3 × TESS observations,we are confident that
TESS has recorded five SO from KL Dra. Thisis the first time that any SO from a AM CVn has been observed atsuch high cadence over its duration.In the first SO (SO1), observations started mid-way through theSO and there is a short gap near the peak of SO3. In four SO the dipwhich has been seen in previous observations of KL Dra and otherAM CVn systems is also seen. After the flux has recovered from thedip, we find that in two SO, the source becomes highly variable. Inall cases there is a rapid decline in flux at the end of the SO.The difference in the apparent scale of the outbursts between thetwo sets of observations is immediately apparent, with the outburstin
TESS appearing far less prominent. Due to the large pixel size thenearby galaxy (which is significantly brighter than KL Dra) domi-nates the light in the pixels. We confirmed this effect by performingaperture photometry upon the sources in the approximate field ofview of the
TESS pixel using data from Pan-STARRS and simulatingthe effect of a SO on the received flux. We found that a SO producesan approximate increase of 10% in flux received which is in line withthe
TESS observations.In Figure 5 we show the first four days of four of the SO observed. We have manually shifted the time axis so the dip lines up. In eachcase we find a clear drop in flux around 1 day from the initial risein flux. Similar behaviour was first reported in a SO of a hydrogenaccreting dwarf novae using
Kepler data (Cannizzo et al. 2012),where it was identified to be a normal outburst acting as the triggerof a subsequent SO. Due to the similarity, we attribute this finding inKL Dra to be the same feature. Further observations of other dwarfnovae, show this to be a feature of all SO observed in high cadence.However, this is the first time that this precursor has been seen in anoutburst from an AM CVn.We also show one SO from the hydrogen rich accreting dwarfnova WX Hyi (using
TESS ∼
12 days; SO4-SO5 ∼
13 daysand after SO5 ∼
14 days. This appears to be correlated with the SOrecurrence time with a shorter recurrence time leading to a shorterinterval between normal outbursts. Observations such as these willinform future models of outbursts in systems similar to KL Dra.
We have presented optical photometry collected over 15 years of eightAM CVn systems all of which have an orbital period in the range22.5 – 26.8 min. The vast majority of these systems have been seento outburst in a manner consistent with the findings of Ramsay et al.(2012) who identified that systems with an orbital period between20 and 45 minutes undergo outbursts. The properties of these SOhave been identified to be correlated with the period of the system asquantised by Levitan et al. (2015, §4.2) and Cannizzo & Nelemans(2015).These systems have now been sufficiently well sampled to allow usto distinguish between normal outbursts, which only last a few daysat most, and SO, which last at least a week. This has allowed someof these systems to be identified as the hydrogen deficient analoguesof SU UMa dwarf novae which show similar behaviour (Kotko et al.2012). Complete study of the normal outbursts has, however, onlybeen possible with space telescopes such as TESS; offering sufficienttemporal resolution to see these short lived events. Normal outburstsare seen clearly in PTF J1919+4815, ASASSN-14CC, CR Boo, KLDra, PTF J2219+3135 and V803 Cen both in our observations andthat of others. This has allowed for a more complete picture of thebehaviour of these systems to be observed and thus address oversightsin earlier modelling work which omitted features due to their apparentabsence.
In our study of these AM CVn systems we identify two distinctgroupings of outbursting systems (Figure 1). ASASSN 14CC, CR
MNRAS , 1–11 (2021)
M CVn Accretion States Figure 4.
Top panels: ZTF g and r band, in green and red, and GOTO, in blue, data showing SO in KL Dra. Bottom panels: TESS data taken contemporaneouslyshowing each of these SO and the immediately proceeding normal outburst. No ground based data is available for SO3 and SO4 as KL Dra was too close to theSun at these times.
Boo and V803 Cen each show very similar high and low states ofsimilar duration. Reminiscent of the standstills that are seen in ZCam systems (Simonsen et al. 2014), we see that CR Boo and V803Cen have extended periods of time ( > Table 2.
Superhump derived mass ratios from Green et al. (2018). The lefthand columns show those systems which we have identified to have longSO and standstills, whilst the right most columns show CX 361, a high statesystem, and and KL Dra which shows short SO.System q System q CR Boo 0.058 ± ± ± ± this, it provides a possible marker as to the origin of the divergencein outbursting AM CVn behaviour.There are however other possible factors which could be involvedin determining the observed behaviour. The rate of mass transfer fromthe donor star is the fundamental property which determines outburstbehaviours and there are a number of different ways in which thiscan manifest differently. The type of the donor star and the formationchannel that was followed by the AM CVn system at its birth canaffect this mass transfer rate, and incidentally affect the value of q fora system. The mass transfer rate can also be affected the entropy ofthe donor and how it has been effected by irradiation (Deloye et al.2007). These factors are not readily determined through observationalone so future confirmation of this will rely heavily upon modelling.An additional factor which could affect the accretion process is if theprimary white dwarf had a sufficiently high magnetic field to eithertruncate or prevent the formation of an accretion disc. One AM CVnsystem (SDSS J0804+1616) has some evidence for the white dwarfhaving a significant magnetic field (Roelofs et al. 2009).In addition to the two different outbursting states we also seeevidence of high state systems in our period range. In §2.2.3 wediscussed the behaviour which we see in our data of CX 361. Consis-tent with previous observations, this system is unexpectedly deviatesfrom theoretical predictions and is observed in a high state. RecentlyBurdge et al. (2020) identified another high state AM CVn system,ZTF J2228+4949, with a period of 28.6 minutes. Although this sys-tem lies outwith the period range we have studied, it does lie wellwithin the range of periods that are predicted to outburst. Comparisonof the spectra of CX361 and J2228+4949 with the high state spectra MNRAS000
Superhump derived mass ratios from Green et al. (2018). The lefthand columns show those systems which we have identified to have longSO and standstills, whilst the right most columns show CX 361, a high statesystem, and and KL Dra which shows short SO.System q System q CR Boo 0.058 ± ± ± ± this, it provides a possible marker as to the origin of the divergencein outbursting AM CVn behaviour.There are however other possible factors which could be involvedin determining the observed behaviour. The rate of mass transfer fromthe donor star is the fundamental property which determines outburstbehaviours and there are a number of different ways in which thiscan manifest differently. The type of the donor star and the formationchannel that was followed by the AM CVn system at its birth canaffect this mass transfer rate, and incidentally affect the value of q fora system. The mass transfer rate can also be affected the entropy ofthe donor and how it has been effected by irradiation (Deloye et al.2007). These factors are not readily determined through observationalone so future confirmation of this will rely heavily upon modelling.An additional factor which could affect the accretion process is if theprimary white dwarf had a sufficiently high magnetic field to eithertruncate or prevent the formation of an accretion disc. One AM CVnsystem (SDSS J0804+1616) has some evidence for the white dwarfhaving a significant magnetic field (Roelofs et al. 2009).In addition to the two different outbursting states we also seeevidence of high state systems in our period range. In §2.2.3 wediscussed the behaviour which we see in our data of CX 361. Consis-tent with previous observations, this system is unexpectedly deviatesfrom theoretical predictions and is observed in a high state. RecentlyBurdge et al. (2020) identified another high state AM CVn system,ZTF J2228+4949, with a period of 28.6 minutes. Although this sys-tem lies outwith the period range we have studied, it does lie wellwithin the range of periods that are predicted to outburst. Comparisonof the spectra of CX361 and J2228+4949 with the high state spectra MNRAS000 , 1–11 (2021)
C. Duffy et al. . . . . . . . N o r m F l u x WX HyiSO2SO3SO4SO5
Figure 5.
Four SO from KL Dra and one from DN WX Hyi as seen byTESS. Each have been overlapped in time such that the dip following theproceeding outburst occurs at approximately 1 day after onset. The WX HyiSO shows similar features although occurring on a longer time scale; we havecompressed it for straightforward comparison. These observations of WXHyi show evidence of superhumps, although previously observed in KL Drathese observations do not show this as a result of the flux excess from theother sources in the same pixel. of KL Dra (Ramsay et al. 2010) yields no immediate evidence ofdifference. This is in line with expectations as they were identifiedas high state systems initially from their spectra, nevertheless it ispossible that unusual metallicity is a component. Alternatively it isequally possible, as before, that another factor, such as a differentformation channel or donor type has resulted in this permanent highstate.In high state systems, such as AM CVn itself, the state is main-tained because the accretion disc temperature is always in excess ofthe ionisation temperature of Helium. This makes the temperaturesufficiently high so as to have a mass accretion rate above the criticalvalue; this in turn is a stable thermal equilibrium for the system tooccupy and thus the system remains in a high state (Warner 2003,see §3.5.3.2 and 3.5.3.3).
Figure 6.
From top to bottom, the SO recurrence time, duration, and amplitudeplotting as function of orbital period. The lines represent the relationships foramplitude and recurrence time derived by Levitan et al. (2015), and durationderived by Cannizzo & Nelemans (2015) as a function of orbital period.Squares and circles denote CR Boo-like and KL Dra-like systems respectively.Values taken from Table 1, recurrence and duration for PTF1J0719+4858from Levitan et al. (2011, Table 4).
The existence of such systems in the instability region is furtherproof that more than simply the orbital period of system is importantwhen considering their behaviour. These systems, and other subse-quently identified, should be a focus of further observations in orderto identify which physical parameters are important in determiningthe behaviour of these systems and how the disc accretion model canbe altered in light of this.In previous studies (eg. Ramsay et al. 2012; Levitan et al. 2015),considering the AM CVn systems discovered at the time, it was pos-sible to say that all systems within the period range ∼
22 and 44minutes showed outbursts, however with discovery of further sys-tems this is clearly not the case. Systems within this period rangecannot be treated like an homogeneous group and like their hydrogendominated cousins they ought to be subdivided based on their out-burst behaviour. The high state systems appear to be the analoguesof nova-like systems. Likewise, we agree with Kotko et al. (2012)that KL Dra-like systems are most likely the helium dominated ana-logues of SU UMa systems whilst CR Boo-like systems are Z Camanalogues. In Z Cam systems the mass accretion rate lies very closeto the critical mass accretion rate and standstills are believed to occurwhen this critical value is reached. it is equally possible however, thatthe CR Boo systems are analogous to VY Scl systems which showsimilar extended high states. It is still an open question as to whatcauses these variations in mass accretion rate but star spots have beensuggested as a possible origin (Livio & Pringle 1994) and it could bethe case that systems of this type are more likely to host star spots.
MNRAS , 1–11 (2021)
M CVn Accretion States Observations of dwarf nova using
Kepler (Cannizzo et al. 2012)provided evidence of normal outbursts immediately before the onsetof a SO. Subsequent observations of SS Cyg using AAVSO data byCannizzo (2012) also found evidence of such precursor outbursts.These were identified to be associated with the superhump wherethe mass transfer rate undergoes a modulation allowing for a briefcooling front to propagate.Since these discoveries it has been established that such precursoroutbursts are seen in all dwarf nova. Until now these had not beenseen in hydrogen deficient systems. Our observations of KL Dra inTESS provide the first evidence of a normal outburst immediatelyproceeding a SO in an AM CVn. The relative rarity of AM CVn sys-tems has meant that it has only been possible to make this discoverywith high cadence observations.This has significant implications for the application of the discinstability model as applied to AM CVn systems; in dwarf nova theyare believed to be the event that induces the subsequent outburst,their presence here suggests a similar effect. We believe that a similarsuperhump feature should be introduced to modelling of AM CVnSO to account for this discovery.In order to identify if this is a feature common to all outburstingAM CVn systems, or indeed a subset of them, high cadence photom-etry of similar type to that supplied by TESS should be employedin future studies of these systems if possible. We believe, however,that evidence from hydrogen dominated systems points to this beinglikely.
Most of the sources in this study appear to show a pronounced dipin brightness a handful of days after maximum brightness whichlasts for approximately a day. This feature has has previously beenobserved by Ramsay et al. (2012, §4) in various AM CVn systems.In our observations we see this dip feature, to varying degrees, inPTF1919+4815, ASASSN 14CC, CR Boo, KL Dra, and V803 Cen.Examples of these features can be seen in Figure 3 and Figure 4.For those systems where we do not see a dip, there is limited data,and previous work has seen a dip in PTF1 J0719+4858 (Levitanet al. 2011), suggesting they are common in outbursts from AM CVnbinaries.In Figure 7 we compare one SO from KL Dra observed usingTESS with two other systems: ASASSN 14-mv, a 41 minute AMCVn system (Ramsay et al. 2018), and EG Cnc, a SU UMa dwarfnova (Patterson et al. 1998), both of which were identified as showing“echo" outbursts, where multiple, shorter duration, lower amplitudeoutbursts are seen during the decline to quiescence. Although weexpect the physical mechanism which gives rise to the dips seen insome AM CVn systems and the echo outbursts will be different,they highlight the need to obtain photometry covering the durationof the SO so that they can inform and test models which predict theproperties of the accretion disc during an outburst (e.g. Kotko et al.(2012)).
We have investigated the behaviour of 8 AM CVn systems in the discinstability region with orbital periods between 22.5 and 26.8 minutesin order to probe the behaviour of these systems and their outbursts.We present the first evidence of precursor normal outbursts before
Figure 7.
Lightcurves of a SO observed in KL Dra using TESS and ofASASSN 14mv and EG Cnc, where data originates from AAVSO. In KL Dra denotes the peak of the preceding normal outburst, , denotes the dip inbrightness between the normal outburst and the SO (§4.2), denotes the peakof the SO and denotes the dip in the SO. The middle panel shows an echooutburst in 2015 from the AM CVn binary ASASSN 14mv and the lowerpanel shows an echo outburst from the hydrogen accreting binary EG Cncfrom 1996. Such features should inform models which predict outburst fromaccreting binaries. the onset of the SO in KL Dra; which we believe, based on hydrogendwarf nova, are likely to be ubiquitous in AM CVn systems. Ifconfirmed, this finding should prompt changes to the existing modelsfor disc accretion in AM CVn systems.We have further identified that the previously broadly acceptedfinding that systems with orbital periods between 22 and 44 minutesshould exhibit outbursts is not as well defined as previously thought.We have studied one AM CVn and discussed another with periodsinside this range that have only been seen to exist in their high state.It is also likely that more such systems exist but observational biasmeans that they have not been identified; outbursting sources oftenmake more attractive targets for study meaning that these systems maywell have been neglected. This adds to the growing body of evidencethat suggests that the long term behaviour of AM CVn systems in theinstability region is more subtle than previously thought and mirrorsthat of the hydrogen dominated accreting dwarf novae. We believethat further observations of more sources will cement this conclusion,yielding further evidence of these high state systems. MNRAS000
Lightcurves of a SO observed in KL Dra using TESS and ofASASSN 14mv and EG Cnc, where data originates from AAVSO. In KL Dra denotes the peak of the preceding normal outburst, , denotes the dip inbrightness between the normal outburst and the SO (§4.2), denotes the peakof the SO and denotes the dip in the SO. The middle panel shows an echooutburst in 2015 from the AM CVn binary ASASSN 14mv and the lowerpanel shows an echo outburst from the hydrogen accreting binary EG Cncfrom 1996. Such features should inform models which predict outburst fromaccreting binaries. the onset of the SO in KL Dra; which we believe, based on hydrogendwarf nova, are likely to be ubiquitous in AM CVn systems. Ifconfirmed, this finding should prompt changes to the existing modelsfor disc accretion in AM CVn systems.We have further identified that the previously broadly acceptedfinding that systems with orbital periods between 22 and 44 minutesshould exhibit outbursts is not as well defined as previously thought.We have studied one AM CVn and discussed another with periodsinside this range that have only been seen to exist in their high state.It is also likely that more such systems exist but observational biasmeans that they have not been identified; outbursting sources oftenmake more attractive targets for study meaning that these systems maywell have been neglected. This adds to the growing body of evidencethat suggests that the long term behaviour of AM CVn systems in theinstability region is more subtle than previously thought and mirrorsthat of the hydrogen dominated accreting dwarf novae. We believethat further observations of more sources will cement this conclusion,yielding further evidence of these high state systems. MNRAS000 , 1–11 (2021) C. Duffy et al.
ACKNOWLEDGEMENTS
The Gravitational-wave Optical Transient Observer (GOTO) projectacknowledges the support of the Monash-Warwick Alliance; Univer-sity of Warwick; Monash University; University of Sheffield; Univer-sity of Leicester; Armagh Observatory & Planetarium; the NationalAstronomical Research Institute of Thailand (NARIT); Instituto deAstrofísica de Canarias (IAC); University of Portsmouth; Universityof Turku, and the UK Science and Technology Facilities Council(STFC grant numbers ST/T007184/1, ST/T003103/1).We acknowledge with thanks the variable star observations fromthe AAVSO International Database contributed by observers world-wide and used in this research.Armagh Observatory & Planetarium is core funded by the North-ern Ireland Executive through the Department for Communities. C.Duffy acknowledges STFC for the receipt of a postgraduate stu-dentship.D.M.S and M.R.K acknowledge support from the ERC under theEuropean Union’s Horizon 2020 research and innovation programme(grant agreement no. 715051; Spiders).The CSS survey is funded by the National Aeronautics and SpaceAdministration under Grant No. NNG05GF22G issued through theScience Mission Directorate Near-Earth Objects Observations Pro-gram. The CRTS survey is supported by the U.S. National ScienceFoundation under grants AST-0909182.The Pan-STARRS1 Surveys (PS1) and the PS1 public sciencearchive have been made possible through contributions by the In-stitute for Astronomy, the University of Hawaii, the Pan-STARRSProject Office, the Max-Planck Society and its participating in-stitutes, the Max Planck Institute for Astronomy, Heidelberg andthe Max Planck Institute for Extraterrestrial Physics, Garching, TheJohns Hopkins University, Durham University, the University of Ed-inburgh, the Queen’s University Belfast, the Harvard-SmithsonianCenter for Astrophysics, the Las Cumbres Observatory Global Tele-scope Network Incorporated, the National Central University of Tai-wan, the Space Telescope Science Institute, the National Aeronau-tics and Space Administration under Grant No. NNX08AR22G is-sued through the Planetary Science Division of the NASA ScienceMission Directorate, the National Science Foundation Grant No.AST-1238877, the University of Maryland, Eotvos Lorand Univer-sity (ELTE), the Los Alamos National Laboratory, and the Gordonand Betty Moore Foundation.Based on observations obtained with the Samuel Oschin 48-inchTelescope at the Palomar Observatory as part of the Zwicky TransientFacility project. ZTF is supported by the National Science Foundationunder Grant No. AST-1440341 and a collaboration including Caltech,IPAC, the Weizmann Institute for Science, the Oskar Klein Center atStockholm University, the University of Maryland, the University ofWashington, Deutsches Elektronen-Synchrotron and Humboldt Uni-versity, Los Alamos National Laboratories, the TANGO Consortiumof Taiwan, the University of Wisconsin at Milwaukee, and LawrenceBerkeley National Laboratories. Operations are conducted by COO,IPAC, and UW.
DATA AVAILABILITY
The following data used in this article is available in the pub-lic domain at the following locations Pan-STARRS: https://catalogs.mast.stsci.edu/panstarrs/ , ASAS-SN: https://asas-sn.osu.edu/photometry , Catalina: http://nesssi.cacr.caltech.edu/DataRelease/ , AAVSO: , and ZTF: https://irsa.ipac.caltech.edu/Missions/ztf.html and https://lasair.roe.ac.uk .GOTO data products will be available as part of planned GOTOpublic data releases.
REFERENCES
Bellm E. C., et al., 2019, PASP, 131, 018002Burdge K. B., et al., 2020, arXiv e-prints, p. arXiv:2009.02567Cannizzo J. K., 2012, ApJ, 757, 174Cannizzo J. K., Nelemans G., 2015, ApJ, 803, 19Cannizzo J. K., Ramsay G., 2019, AJ, 157, 130Cannizzo J. K., Smale A. P., Wood M. A., Still M. D., Howell S. B., 2012,ApJ, 747, 117Deloye C. J., Taam R. E., Winisdoerffer C., Chabrier G., 2007, MNRAS, 381,525Devor J., 2005, ApJ, 628, 411Drake A. J., et al., 2009, ApJ, 696, 870Feinstein A. D., et al., 2019, PASP, 131, 094502Flewelling H. A., et al., 2020, ApJS, 251, 7Green M. J., et al., 2018, MNRAS, 477, 5646Hameury J. M., Knigge C., Lasota J. P., Hambsch F. J., James R., 2020, A&A,636, A1Hartman J. D., Bakos G. Á., 2016, A&C, 17, 1Jayasinghe T., et al., 2019, MNRAS, 486, 1907Jha S., Garnavich P., Challis P., Kirshner R., Berlind P., 1998, IAU Circ.,6983, 1Kafka S., 2020, Observations from the AAVSO International DatabaseKato T., Nogami D., Baba H., Hanson G., Poyner G., 2000, MNRAS, 315,140Kato T., Stubbings R., Monard B., Pearce A., Nelson P., 2001, InformationBulletin on Variable Stars, 5091, 1Knigge C., 2006, MNRAS, 373, 484Kochanek C. S., et al., 2017, PASP, 129, 104502Kotko I., Lasota J. P., Dubus G., Hameury J. M., 2012, A&A, 544, A13Kraft R. P., Mathews J., Greenstein J. L., 1962, ApJ, 136, 312Kupfer T., et al., 2018, MNRAS, 480, 302Levitan D., et al., 2011, ApJ, 739, 68Levitan D., et al., 2014, ApJ, 785, 114Levitan D., Groot P. J., Prince T. A., Kulkarni S. R., Laher R., Ofek E. O.,Sesar B., Surace J., 2015, MNRAS, 446, 391Lightkurve Collaboration et al., 2018, Lightkurve: Kepler and TESStime series analysis in Python, Astrophysics Source Code Library(ascl:1812.013)Livio M., Pringle J. E., 1994, ApJ, 427, 956Meyer F., Meyer-Hofmeister E., 1983, A&A, 121, 29Patterson J., et al., 1998, PASP, 110, 1290Ramsay G., et al., 2010, MNRAS, 407, 1819Ramsay G., Barclay T., Steeghs D., Wheatley P. J., Hakala P., Kotko I., RosenS., 2012, MNRAS, 419, 2836Ramsay G., et al., 2018, A&A, 620, A141Ricker G. R., et al., 2015, JATIS, 1, 014003Roelofs G. H. A., et al., 2009, MNRAS, 394, 367Schwarzenberg-Czerny A., 1989, MNRAS, 241, 153Simonsen M., et al., 2014, JAAVSO, 42, 177Smak J., 1967, Acta Astron., 17, 255Smak J., 1975, Acta Astron., 25, 371Smith K. W., et al., 2019, RNAAS, 3, 26Solheim J. E., 2010, PASP, 122, 1133Steeghs D., et al., 2021, The Gravitational-wave Optical Transient Observer(GOTO): prototype performance and prospects for transient science, Pa-per in developmentStroeer A., Vecchio A., 2006, CQGra, 23, S809Warner B., 2003, Cataclysmic variable stars. Cambridge University Press,Cambridge New York, N.YMNRAS , 1–11 (2021)
M CVn Accretion States Wevers T., et al., 2016, MNRAS, 462, L106Wood M. A., Casey M. J., Garnavich P. M., Haag B., 2002, MNRAS, 334, 87This paper has been typeset from a TEX/L A TEX file prepared by the author. MNRAS000