The case for jets in cataclysmic variables

Abstract

For decades cataclysmic variables (CVs) were thought to be one of the few classes of accreting compact objects to not launch jets, and have consequently been used to constrain jet launching models. However, recent theoretical and observational advances indicate that CVs do in fact launch jets. Specifically, it was demonstrated that their accretion-outflow cycle is analogous to that of their higher mass cousins – the X-ray Binaries (XRBs). Subsequent observations of the CV SS Cygni confirmed this and have consistently shown radio flaring equivalent to that in the XRBs that marks a transient jet. Based on this finding and the emission properties, several studies have concluded that the radio emission is most likely from a transient jet. Observations of other CVs, while not conclusive, are consistent with this interpretation. However, the issue is not yet settled. Later observations have raised a number of questions about this model, as well as about potential alternative radio emission mechanisms. CVs are non-relativistic and many have well-determined distances; these properties would make them ideal candidates with which to address many of our outstanding questions about fundamental jet physics. Here we review the case for jets in CVs, discuss the outstanding questions and issues, and outline the future work necessary to conclusively answer the question of whether CVs launch jets.

Introduction

Cataclysmic variables (CVs) are compact binary systems in which a low-mass main-sequence secondary transfers material to an accreting white dwarf (AWD) primary via Roche-lobe overflow (for a review see Warner, 1995). The orbital periods of CVs are typically in the range 1.2 h ≲ P_orb ≲ 6 h, with a gap in the period distribution between  ≃ 2 h and  ≃ 3 h. Stable mass transfer in CVs, which is characterized by mass ratios $q=M_2/M_{WD}<1,$ is only possible as a result of angular momentum losses from the binary. In the standard picture of CV evolution (see Hameury, Dubus, Lasota, 2001, Knigge, Baraffe, Patterson, 2011), mass transfer above the period gap is driven by magnetic braking (Eggleton, 1976, Verbunt, Zwaan, 1981), while mass transfer below the gap is driven by gravitational radiation (Faulkner, 1971, Paczynski, Sienkiewicz, 1981).

If the magnetic field of the white dwarf (WD) primary is sufficiently weak (B_WD ≲ 10⁵G), the accretion process takes place via a disk that extends down to the WD surface. These “non-magnetic” systems dominate the overall CV population and are the WD analogues to low-mass X-ray binary systems (LMXBs, which contain accreting neutron stars or black holes). As discussed in Section 2.1 and in Chapter 7 of this volume, the existence of radio jets in LMXBs has been well established for quite some time. By contrast, non-magnetic CVs have only been discovered to be significant radio emitters – and possible jet sources – in the last decade. These systems are the main focus of this review.

Even though we will not be discussing them in any detail, it is worth noting that there are other classes of AWDs. First, ultracompact (P_orb ≲ 1 hr) AWD binaries harbour (partially) degenerate donors and are known as AM CVn stars. No radio emission or other jet signatures have so far been reported in these systems – but we are also not aware of any dedicated searches. Second, AWDs in wide binaries (P_orb ≳ 6 hr) tend to have nuclear-evolved companions, with the longest-period systems (P_orb ~ years) containing red giant secondaries. In these systems with red giant companions, known as symbiotic stars, mass transfer can be wind-driven. Jets have been detected in some symbiotic stars, on the basis of radio emission, optical and X-ray imaging and optical spectroscopy (e.g. Tomov, Munari, Marrese, 2000, Kellogg, Pedelty, Lyon, 2001, Brocksopp, Sokoloski, Kaiser, Richards, Muxlow, Seymour, 2004, Leedjärv, 2004). In at least one symbiotic system, radio flares tend to be preceded by X-ray dips, similar to what is seen in X-ray binaries (Sokoloski and Kenyon, 2003). Third, AWDs with more massive donors (M₂ ≳ 1 M_⊙) and inverted mass ratios ($q=M_2/M_{WD}>1$) tend to display luminous ($L\simeq {10}^{36}-{10}^{37}\text{erg}{\mathrm{s}}^{-1}$), soft (T ≃ 10⁵ K) X-ray emission and are therefore known as supersoft sources (SSS). Jets have also been detected in some SSSs, albeit only indirectly based on blue- and red-shifted “satellite lines” in their optical spectra (Crampton, Hutchings, Cowley, Schmidtke, McGrath, O’Donoghue, Harrop-Allin, 1996, Southwell, Livio, Charles, O’Donoghue, Sutherland, 1996, Tomov, Munari, Kolev, Tomasella, Rejkuba, 1998, Quaintrell, Fender, 1998, Becker, Remillard, Rappaport, McClintock, 1998, Motch, 1998). Fourth and finally, some CVs contain AWDs whose magnetic fields are strong enough to either truncate the accretion disk (intermediate polars:  ~ 10⁵ G ≲ B_WD ≲ 10⁷ G) or prevent disk formation entirely (polars: B_WD ≳ 10⁷ G). These “magnetic CVs” produce radio emission (likely gyrosynchrotron and/or electron-cyclotron maser radiation; see Section 3.7), but this is thought to be associated with the magnetized accretion flow, rather than with a collimated, outflowing jet.

Returning to the non-magnetic systems that are our focus (and hereafter simply referred to as “CVs”), the long-term light curves suggest a natural split into two basic sub-classes. The first, the so-called dwarf novae (DNe), exhibit dramatic outbursts, with recurrence times ranging from weeks to decades (e.g. Coppejans, Körding, Knigge, Pretorius, Woudt, Groot, Van Eck, Drake, 2016, Otulakowska-Hypka, Olech, Patterson). At the onset of a DN eruption, the optical brightness increases quickly by  ~ 2-8 magnitudes, with a characteristic rise time-scale of only a day or so. The following plateau phase can then last from days to weeks. The second class of non-magnetic CVs are known as nova-likes (NLs), and these typically remain at roughly constant brightness.

This split in the non-magnetic CV population is analogous to the split in the LMXB population between “transient” and “persistent” systems. This is no coincidence, as the underlying physics are thought to be the same. More specifically, the outbursts of both DNe and transient LMXBs can be understood in the framework of the “disk instability model” (DIM, for a review see Osaki, 1996, Lasota, 2001). The DIM (Smak, 1971, Osaki, 1974, Hōshi, 1979, Meyer, Meyer-Hofmeister, 1981) is based on the observation that accretion disks are subject to a thermal-viscous instability across a fairly wide range of mass-transfer rates (bounded above and below by ${\dot{M}}_{crit,+}$ and ${\dot{M}}_{crit,-}$ respectively, Smak 1984). This instability is associated with the sensitive and non-monotonic temperature dependence of the opacity in partially ionized plasmas (i.e. for T ≃ 7000 K). In nova-likes and persistent LMXBs, the mass-transfer rate from the secondary star is high enough to avoid the instability, ${\dot{M}}_2>{\dot{M}}_{crit,+}$. However, in DNe and transient LMXBs, the mass-supply rate lies in the unstable range, ${\dot{M}}_{crit,-}<{\dot{M}}_2<{\dot{M}}_{crit,+}$. As a result, their disks cycle between a cool, largely neutral, low-viscosity state (“quiescence”) and a hot, ionized, high-viscosity state (“outburst”). In the quiescent state, the mass-transfer rate through the disk is less than the supply rate, ${\dot{M}}_{tr}<{\dot{M}}_2$. Material therefore piles up in the disk. This eventually triggers the transition to the outburst state, in which ${\dot{M}}_{tr}>{\dot{M}}_2$.

Strong support for this basic picture comes from comparisons between the theoretically predicted ${\dot{M}}_{crit,+}$ and the observationally inferred long-term average accretion rates, ${\overline{\dot{M}}}_{acc}\simeq {\dot{M}}_2$. These confirm that ${\dot{M}}_{crit,+}$ neatly separates steady (${\overline{\dot{M}}}_{acc}>{\dot{M}}_{crit,+}$) and erupting (${\overline{\dot{M}}}_{acc}<{\dot{M}}_{crit,+}$) systems, in both DNe and LMXBs (e.g. Coriat, Fender, Dubus, 2012, Miller-Jones, Sivakoff, Knigge, Körding, Templeton, Waagen, 2013, Dubus, Otulakowska-Hypka, Lasota, 2018). Note that the physics responsible for nova eruptions (a thermonuclear runaway on the AWD) is completely different to that driving dwarf nova eruptions (the DIM). Bipolar outflows associated with nova eruptions are common, as is radio emission (e.g. Sokoloski, Rupen, Mioduszewski, 2008, Ribeiro, Chomiuk, Munari, Steffen, Koning, O’Brien, Simon, Woudt, Bode, 2014). Since nova outbursts are not accretion-powered, they fall outside the scope of this review and we do not discuss them in detail.

Accretion disks are a key component in several astrophysically important systems on all scales. In young stellar objects, the accretor is a still-forming proto-star. In CVs, SSSs and symbiotics, it is a WD. In X-ray binaries, it is a neutron star (NS) or a stellar-mass black hole (BH). In active galactic nuclei and quasars, it is a supermassive black hole.

Remarkably, despite these differences in size and mass scales, a fair amount of the physics governing the accretion process in all of these objects appears to be universal. For example, the statistical properties of accretion-induced variability (“flickering”) appear to scale predictably with accretion rate and system size (see Uttley, McHardy, 2001, Uttley, McHardy, Vaughan, 2005, Gandhi, 2009, Scaringi, Körding, Uttley, Knigge, Groot, Still, 2012, Scaringi, Maccarone, Kording, Knigge, Vaughan, Marsh, Aranzana, Dhillon, Barros, 2015). Similarly (and paradoxically), the accretion process always seems to go hand-in-hand with mass loss from the system, in the form of weakly collimated bipolar disk winds or highly collimated jets (e.g. Greenstein, Oke, 1982, Fender, 2001, Froning, Long, Drew, Knigge, Proga, 2001, Ponti, Fender, Begelman, Dunn, Neilsen, Coriat, 2012, Kafka, Hoard, Honeycutt, Deliyannis, 2009, Higginbottom, Knigge, Long, Matthews, Parkinson, 2019 and other references in this review).

In an effort to shed light on the mechanism that drives the collimated jets in these systems, (Livio, 1997, Livio, 1999) first drew attention to the fact that – at that time – CVs were the only class in which no jet had ever been seen. Weakly collimated disk winds are observed in NL CVs and DNe in outburst, but there was no evidence for the existence of collimated jets in CVs. Based on this apparent failure to launch jets from CVs, Livio, 1997, Livio, 1999 [also see Soker and Lasota 2004] proposed that jet formation requires a powerful energy source associated with the central object, which is present in all disk-accreting systems except CVs. Spruit et al. (1997), as well as Jafari and Vishniac (2018), also noted the absence of evidence for jets in CVs. However, they interpreted it as a consequence of the small dynamic range in radii spanned by CV disks, since their preferred collimation mechanism (via poloidal magnetic fields) requires a large dynamic range to achieve strong collimation. In any case, regardless of any preferred interpretation, a unique inability of CVs to drive jets would be a powerful constraint on theoretical models of jet formation and collimation. In particular, if jets do exist in CVs and display a similar phenomenology to those in neutron star and black hole XRBs, any universal driving mechanism requiring strong gravity and/or ultra-strong magnetic fields would be disfavoured (e.g. Kylafis, Contopoulos, Kazanas, Christodoulou, 2012, Parfrey, Spitkovsky, Beloborodov, 2016).

In evaluating the detectability of jets in CVs, three observational signatures are particularly relevant: (i) spatially resolved emission in optical, radio or X-ray images (as seen, for example, in young stellar objects, as well as some X-ray binaries and AGN); (ii) unresolved radio emission associated with jet synchrotron radiation (as seen, for example, in other X-ray binaries and AGN); (iii) jet “satellite lines”, i.e. red- and blue-shifted emission line components produced by line-formation in the approaching and receding parts of the jet, respectively (as seen, for example, in some SSSs and in the ultraluminous X-ray binary SS 433). Back in 1997, none of these signatures had been seen in non-magnetic CVs. Although radio surveys of non-magnetic CVs in the 1980s yielded five claimed detections [SU UMa (Benz et al., 1983), but see Coppejans et al. (2016b), TY Psc (Turner, 1985), UZ Boo (Turner, 1985), EM Cyg (Benz and Guedel, 1989) and AC Cnc (Torbett and Campbell, 1987)], at least two of these detections were almost certainly false positives [UZ Boo (Nelson, Spencer, 1988, Benz, Guedel, 1989) and AC Cnc (Körding et al., 2011)] and none were detected in follow-up observations.

Of course, absence of evidence is not (always) evidence of absence. Indeed, Livio, 1997, Livio, 1999 already emphasized the importance of establishing the presence or absence of jets in CVs definitively. Given that optical jet satellite lines had been seen in SSSs – another class of AWDs – Knigge and Livio (1998) attempted to estimate the expected equivalent width of such features in non-magnetic CVs. Scaling from the jet satellite lines in SSSs, their estimate for luminous NLs was EW ≃ 0.1 Å. In principle, this is difficult, but perhaps not impossible to detect in high S/N data. At radio wavelengths, the main obstacle was simply the limited instrumental sensitivity. The flux limit for a typical radio observation at this time was  ≃ 0.2 mJy, which corresponds to a specific radio luminosity of L_R,ν ~ 10¹⁶(d/200 pc)² erg/s/Hz. As we shall see in Sections 2.3 and 2.4, this sensitivity turns out to be insufficient to rule out the presence of radio jets in CVs with confidence.

However, in addition to these practical difficulties, the search for jet signatures in CVs was also constrained by a conceptual issue. As noted above, Livio, 1997, Livio, 1999 conjectured that the mechanism responsible for driving astrophysical jets may require the existence of a powerful central energy source. In line with this idea, the thinking about jets in CVs at the time was focused on the most luminous, steady CVs, i.e. the highest accretion rate nova-likes. After all, in these systems, a sufficiently strong and concentrated source of power might be provided by the hot inner disk and/or a luminous boundary layer and/or an accretion-heated WD.

This focus – and Livio’s conjecture – appeared to receive strong observational support when Shahbaz et al. (1997) announced the discovery of blue- and red-shifted Hα jet satellite lines in the nova-like T Pyx. However, T Pyx is not just any old nova-like, but also a recurrent nova. Nova eruptions – as opposed to the dwarf nova eruptions described above – occur when the non-degenerate envelope of an AWD reaches a critical pressure, triggering a thermonuclear runaway. They are always associated with the ejection of significant amounts of material, typically ${10}^{-5}{\mathrm{M}}_{\odot }-{10}^{-4}{\mathrm{M}}_{\odot },$ at velocities of $300\text{km}{\mathrm{s}}^{-1}-3000\text{km}{\mathrm{s}}^{-1}$. In many recent novae – including T Pyx – this material produces a detectable, line-emitting “nova shell” around the system. Following up on the announcement of Shahbaz et al. (1997), O’Brien and Cohen (1998) and Margon and Deutsch (1998) obtained spatially resolved spectroscopy of T Pyx and its nebula. These observations showed that the “jet lines” identified by Shahbaz et al. are, in fact, nebular [N ii] lines associated with the nova shell. The only other search for jet satellite lines in high-accretion-rate NL variables also produced no detections (Hillwig et al., 2004). In the wake of these negative results, progress effectively ground to a half for several years.

The conceptual leap that rekindled interest in the field was made by Elmar Körding. His approach was driven by the work he had done on radio jets in black-hole XRBs and AGN (Körding, 2007). A key development in that field was the realization that strong radio emission is almost exclusively associated with transient systems, and is actually quenched in these systems as they transition towards the peak of their outbursts (Fender, Belloni, Gallo, 2004, Migliari, Fender, 2006, McHardy, Koerding, Knigge, Uttley, Fender, 2006). Given that the outbursts in CVs and XRBs are thought to be driven by the same physics (the DIM), he therefore wondered if these similarities might extend to jet formation. Could it be that jets were actually present in CVs, but only (or mainly) in transient systems? If so, the search for jets in CVs should focus not on the steady, luminous nova-likes, but instead on dwarf novae away from the plateau phase of their eruptions.