Flows and Shocks: Some Recent Developments in Symbiotic Star and Nova Research
J. L. Sokoloski, Stephen Lawrence, Arlin P. S. Crotts, Koji Mukai
FFlows and Shocks: Some Recent Developments inSymbiotic Star and Nova Research
J. L. Sokoloski ∗ Columbia Astrophysics LaboratoryE-mail: [email protected]
Stephen Lawrence
Hofstra UniversityE-mail:
Arlin P. S. Crotts
Columbia Department of Astronomy
Koji Mukai
Goddard Space Flight Center and University of Maryland Baltimore CountyE-mail:
There have been several surprising developments in our understanding of symbiotic binary starsand nova eruptions over the last decade or so based on multiwavelength data. For example, sym-biotic stars without shell burning have been revealed through their X-ray emission, UV excess,and UV variability. These purely accretion powered symbiotic stars have much weaker opticalemission lines and radio emission than those with shell burning, and therefor harder to discover,yet may be as numerous as the burning symbiotic stars. Interestingly, both types of symbioticstars are capable of driving strong outflows, leading to colliding wind X-ray emission and spa-tially resolved X-ray jets. For nova eruptions, the most surprising discovery has been that theyare capable of particle acceleration as evidenced by Fermi detection of novae as transient GeVgamma-ray sources. For nova eruptions in cataclysmic variables, this implicates internal shocks,between a slow, dense outflow and a fast outflow or wind. Other signatures of shocks includethermal X-rays and non-thermal radio emissions, and a substantial fraction of optical emissionmay be shock-powered in the early phase of novae. Radio (V959 Mon) and HST (V959 Mon andT Pyx) images of nova shells within a few years of their respective eruptions suggest that novaejecta may commonly consist of an equatorial ring and a bipolar outflow.
Accretion Processes in Cosmic Sources — APCS2016 —5-10 September 2016,Saint Petersburg, Russia ∗ Speaker. c (cid:13) Copyright owned by the author(s) under the terms of the Creative CommonsAttribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0). https://pos.sissa.it/ a r X i v : . [ a s t r o - ph . S R ] F e b lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski
1. Introduction: shocks point the way in symbiotic stars and novae
During the past 5 to 10 years, multiwavelength observations and theoretical studies of symbi-otic binary stars and nova eruptions have led to some surprising developments in these fields. Inboth symbiotic binaries and novae, the key findings are related, at least in part, to flows, shocks,and the high-energy emission that results.Symbiotic stars are wide, interacting binaries in which a compact object accretes from a red-giant (RG) companion. Orbital periods range from years to decades, and binary separations rangefrom AU to tens or even hundreds of AU. Here, we restrict our discussion to symbiotics that containwhite dwarf (WD) accretors. In some such symbiotic stars, the interaction is powered solely by therelease of gravitational potential energy as matter falls onto the WDs. In others, the temperature andluminosity of the WD indicate that the interaction is powered primarily by the quasi-steady burningof hydrogen-rich fuel in a shell on the WD surface (see [56] for a compilation of WD temperaturesand luminosities). The degree to which this shell burning is: 1) residual burning from past novae(which perhaps lasts longer in wide binaries than in cataclysmic variables; CVs); 2) the result ofaccretion at a high rate; or 3) a combination of both is not yet evident. But recent observations atradio through X-ray wavelengths reveal that the presence or absence of shell burning on the WDdetermines the characteristics of a symbiotic at every waveband. X-rays are particularly valuablefor diagnosing the inner accretion flow. Lastly, symbiotics without shell burning, which have beendifficult to identify in the optical, are worth searching for in the X-rays and the ultraviolet (UV) timedomain. This heretofore hidden population offers the opportunity to investigate wind-fed accretiondisks with radii of 10 to 10 cm, their jets, and the evolution and statistics of wide, interactingbinaries — including the production of type Ia supernovae.Nova eruptions occur on accreting WDs in wide, symbiotic binaries and in the much tightercataclysmic variables (CVs). Between 2010 and 2014, the Fermi
Gamma-Ray Space Telescopeastonished the astronomical community with its finding that novae constitute a new class of GeV γ -ray sources (e.g., [1, 2]). It turns out that nova events generate γ -rays both when the eruptingWD is embedded within the wind of a red-giant companion, and when it is not. As we will discussbelow, the hunt for the origin of these γ -rays has exposed the degree to which the outflows fromnovae consist of multiple, distinct flows, and the crucial role of powerful shocks in almost allaspects of novae.
2. Symbiotic stars with and without shell burning
Energetics.
The luminosity of an accreting WD with a red-giant companion is set chiefly bywhether or not accreted material burns quasi-steadily in a shell on the surface of the WD. If materialin the WD envelope is burning at approximately the rate given by the core mass-luminosity relation[67], then the luminosity of the WD is given by L WD ( with burning ) ∼ L (cid:12) . If, on the other hand,shell burning is absent, the luminosity of the WD is roughly that generated by accretion, L WD ( without burning ) ≈ GM WD ˙ M WD / R WD (2.1) ≈ L (cid:12) (cid:18) M WD . M (cid:12) (cid:19) (cid:18) ˙ M WD − M (cid:12) yr − (cid:19) (cid:18) R WD cm (cid:19) − , (2.2)1 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski where M WD is the mass of the WD, ˙ M WD is the rate of accretion onto the WD, and R WD is the radiusof the WD. As we will see below, it is not just the WD luminosities that distinguish WD-plus-RGinteracting binaries with and without shell burning; the two different sources of power for the WDcreate observational differences across the electromagnetic spectrum. Optical spectra.
Because symbiotic stars have traditionally been defined and identified bytheir optical spectra, it is worth starting our discussion with the impact of shell burning (or thelack thereof) on the optical emission. In a CV, the accretion disk usually dominates the opticalemission whether or not shell burning is present, because most light from the WD photosphere isradiated shortward of the optical. In an WD-plus-RG interacting binary, on the other hand, thedonor is quite bright, and the nebula reprocesses some of the UV to soft X-ray emission fromthe burning WD into the optical. Thus, if the WD is hot and luminous due to shell burning, theoptical emission is frequently dominated by light from the red giant donor and/or the ionized nebula[89]. In particular, the strength of the optical line and recombination-continuum emission from thenebula depends on the luminosity (and to some extent temperature) of the accreting WD, which inturn depends mainly upon whether hydrogen-rich fuel is burning quasi-steadily on its surface. Withongoing burning, the temperature of the WD sits above ∼ K, and the luminous WD ionizes alarge portion of the nebula. The nebula then generates the strong, high-ionization state opticalemission lines that are the hallmark of classical symbiotic stars (e.g., [32, 53, 46] and referencestherein). Without shell burning on the WD, a predominantly neutral — and therefore optically faint— nebula allows photospheric emission from the red giant, and in some cases the accretion diskto dominate the optical spectrum, with only weak, if any, emission lines. Extreme examples ofsuch systems include the symbiotic recurrent novae (such as T CrB in its normal, quiet state); thewell-known jet-producing binaries CH Cyg, R Aqr, and MWC 560; and systems with very hardX-ray emission such as RT Cru, V648 Car, and SU Lyn (Fig. 1 shows the UV to optical spectrum ofSU Lyn, from [51]). WD-plus-RG interacting binaries in this state have sometimes been referredto as weakly symbiotic or symbiotic-like binaries. WD-plus-RG binaries without shell burning,however, are not necessarily weakly interacting. What’s in a name?
Although WD-plus-RG interacting binaries without quasi-steady shellburning on the WD often lack the strong optical emission lines that motivated the original defi-nition of symbiotic stars, we argue that it is nevertheless appropriate and useful to include suchsystems among the so-called symbiotic stars . As in past work (e.g., [31, 41, 51]), we thereforedefine a WD symbiotic system as a binary in which a red giant transfers enough material to aWD for the interaction to produce an observable signal at some waveband . Considering WD-plus-RG binaries with and without shell burning together is meaningful in part because the ratesof accretion onto the WDs in systems with and without shell burning can be comparable ( ∼ − – 10 − M (cid:12) yr − (e.g., [23, 91, 15, 41]. For WD-plus-RG binaries without shell burning, however,the interaction is normally more apparent in X-ray spectra and optical-UV fast photometry thanin optical spectra (see below). In fact, at X-ray wavelengths and in optical-UV fast photometry,signatures of binary interaction can be stronger in non-burning than in burning symbiotics. Finally, Though a bit counter-intuitive to extend the definition of symbiotic stars to objects that do not show much of theoptically defined “symbiotic phenomenon" [45], we hope readers will agree that it is more natural than introducing anew name for a what is almost certainly a transient state. Additionally, the name
WD-plus-RG interacting binaries iscumbersome and obscures the connection to past research on symbiotic stars. lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski
Figure 1:
UV through optical spectrum of non-burning symbiotic star SU Lyn, reproduced from [51]. Theyellow UV points for SU Lyn are from
Swi f t /UVOT. The black curve shows
IUE and optical spectra of astandard MIII star, for comparison. Longward of about 4000, the low-resolution optical spectrum of SU Lynis difficult to distinguish from that of a normal red giant. The insets show faint emission lines (of H α and[Ne III] 3869) detected with a high-resolution optical spectrum. SU Lyn shows how difficult it would be todetect such systems with low-resolution optical spectroscopy alone. because shell burning on the WD is almost certainly a transient phenomenon, each WD-plus-RGbinary is likely to spend some time with and some time without shell burning. Shell burning mayeven turn on and off more than once. We thus refer to WD-plus-RG binaries with and without shellburning on the surface of the WD as burning and non-burning symbiotics, respectively. X-rays: revealing the inner accretion flow.
Looking beyond the optical spectrum, the natureof X-rays from a symbiotic star is also dictated in large part by whether or not quasi-steady shellburning is present on the surface of the WD. For example, some non-burning symbiotics emithighly absorbed X-rays with energies greater than several keV and as high as tens of keV (whichwe refer to as hard
X-rays). When such binaries are nearby, their hard X-ray emission can bebright enough to reveal flickering-type variations on time scales of minutes to hours, and spectrathat are well modeled as isobaric cooling flows (as in [50]). This X-ray spectral component wasreferred to by [41] as δ -type X-ray emission, and it most likely emanates from an accretion-diskboundary layer (e.g., [39, 31, 15, 41, 51, 27]). Evidence for a boundary-layer origin includes: 1)rapid, stochastic brightness variations like those for the boundary layers of CVs; 2) cooling flowspectra, like those expected from flows onto the surface of a WD (and observed in CVs (e.g., [69]);3) high intrinsic, partial-covering absorption, indicating that the source of hard X-rays is smalland located behind the nebular and/or disk wind; and 4) a lack of detectable modulation of the3 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski
X-ray brightness at the WD spin period, indicating that the X-rays are unlikely to emanate frommagnetic accretion columns(see [52] for a review of X-ray properties of accreting WDs). Although[13] have suggested magnetic accretion as the origin of hard X-rays in at least one non-burningsymbiotic (RT Cru), we contend that the lack of spin modulations in the hard X-ray emission fromall non-burning symbiotics supports the boundary-layer interpretation for the class as a whole.In burning symbiotics, on the other hand, hard X-ray emission, with energy greater than a fewkeV, seems to be absent, perhaps because the high flux of FUV photons from the luminous, hotWDs Compton-cools the boundary layers out of the X-ray regime (as described in general terms in[16]). So, whereas searches for red giants with strong, high-ionization state optical emission linespreferentially uncover burning symbiotics, searches for red giants with hard X-ray emission wouldpreferentially (or exclusively) uncover symbiotics without shell burning.
Softer X-rays from shell burning, jets, and colliding winds.
If a WD with shell burning ismassive enough for photospheric temperatures to reach at least several hundred thousand degrees,burning symbiotics emit supersoft X-ray emission, which [57] dubbed α -type X-ray emission. Ex-amples of symbiotics with such emission include AG Dra, RR Tel, and StH α
32 [57, 65]. Suchsupersoft X-rays are only detectable if the column of absorbing material is fairly low. Both burningand non-burning symbiotics are able to produce so-called β -type thermal X-ray emission, in whichmost photons have energies less than approximately 2 or 3 keV. The lower column of absorbingmaterial compared to δ -type X-ray emission, the lack of minute-to-hour time scale variability, andthe lower plasma temperatures than typically found in boundary layers all suggest that it arisesfrom colliding winds [57, 41, 61] or within collimated jets (e.g., [29, 17, 30, 60, 28, 96]. R Aqr andCH Cyg provide that most dramatic examples of X-ray emission from shock-heated plasma withinWD jets (Fig. 2). For sources that are faint in the X-rays, it is sometimes difficult to distinguishbetween β -type and δ -type X-ray emission [41]. But even if it is not always clear whether moder-ately hard X-rays come from colliding winds, a jet, an accretion-disk boundary layer, or magneticaccretion columns, observations with Swi f t , INTEGRAL,
Chandra , and most recently,
NuSTAR have certainly made it clear that the X-ray emission from symbiotic stars is complex, variable, bothsoft and hard, and dependent upon the presence or absence of shell burning.
Disk flickering in the UV.
Accretion in CVs, X-ray binaries, and active galactic nuclei leadsto stochastic brightness variations on time scales from roughly the dynamical time near the inneredge of the disk to viscous time scales throughout the disk. UV light curves from
Swi f t /UVOThave now shown that — as long as no shell burning is present to hide the disk — symbiotic starsfit this same pattern. The classic accretion signature of large-amplitude flickering (fractional rmsamplitude greater than about 10%) was not initially found to be pervasive in symbiotic stars becauseearly searches were performed in the optical (e.g., [90, 22]). Although many symbiotics probablycontain accretion disks whose inner regions are similar to disks in CVs [36, 48, 12], the diskis not usually a major contributor to the optical light. Moving to the UV, where the red giantmakes a much smaller contribution, enabled [41] to detect disk flickering from a larger fraction ofsymbiotics than [90] did in the optical. Moreover, based on a comparison between UV variabilityand X-ray spectral properties, they concluded that symbiotics without shell burning had a muchgreater propensity to generate large-amplitude UV flickering. In symbiotics with shell burning,UV disk light is outshined by emission the ionized nebula – which does not generally vary on timescales of minutes to hours. Supporting the idea that flickering from the accretion flow is easier to4 lows and Shocks in Symbiotic Stars and Novae
J. L. Sokoloski
Figure 2:
Chandra images of the X-ray jets from non-burning symbiotics R Aqr (left) and CH Cyg (right),reproduced from [30] and [28]. In the R Aqr image, the 0.2 to 3.5 keV X-rays were smoothed, and the redarrow indicates the location of the central binary. In the CH Cyg image, green contours show the radio fluxdensity at 5 GHz, magenta contours show the hard, 6-7 keV X-rays, and the yellow and orange image showsthe location and strength of soft X-rays. detect if the WD does not support shell burning, all of the symbiotics known to have strong opticalflickering are non-burning symbiotics (e.g., CH Cyg, V407 Cyg, RT Cru, MWC 560, and symbioticrecurrent novae such as RS Oph and T CrB).
Finding the accretion structures in symbiotic stars.
Therefore, by moving beyond the opti-cal band, [41] (and others) identified robust techniques for detecting the accretion disks in at leastsome symbiotics. One can probe the accretion flow by observing in the hard X-rays and/or in theUV time domain. Additionally, Fig. 1 and work by [77] show that even without variability informa-tion, one can identify red giants with UV excess as candidate non-burning symbiotics. Conversely,these same authors found that seeing a classic signatures of accretion — in this case UV or opti-cal disk flickering, or hard X-ray emission — furnishes strong evidence that a given symbiotic ispowered by accretion alone rather than shell burning. Another consequence of these findings isthat there is no evidence that the inner accretion flows in symbiotics and CVs are wildly different;accretion is just usually hidden in symbiotics with shell burning.
Radio emission from burning vs non-burning symbiotics.
Completing our examinationof the observational signatures of burning and non-burning symbiotics across the electromagneticspectrum, clear disparities exist between the radio brightnesses of the two types of symbiotics.In a seminal series of papers, [80, 81, 82] found that most of the symbiotic stars they detectedwith the Very Large Array (VLA) had radio flux densities on the order of mJy, consistent withfree-free emission from the ionized wind of the red giant. Because their target lists consistedprimarily of objects with strong optical emission lines, the majority of the sources they detectedwere likely burning symbiotics. When [105] specifically targeted 11 non-burning symbiotics forradio observations, on the other hand, she found that about half of these objects had radio flux5 lows and Shocks in Symbiotic Stars and Novae
J. L. Sokoloski densities of approximately 10 µ Jy or less (T CrB, ER Del, CD -27 8661, TX CVn, MWC 560, CD -28 3719, and BD -21 3873). The others on her target list (NQ Gem, UV Aur, ZZ CMi, and Wray 15-1470) had flux densities of a few tenths of a mJy (but with no indication that these sources areparticularly distant). The finding that non-burning symbiotics tend to have very faint radio emissionis consistent with the optical emission lines from these sources being weak, and the ionized regionsof the red-giant winds being small compared to those in burning symbiotics. In terms of diagnosingthe source of the WD’s power, a luminosity of L WD ∼ L (cid:12) or greater is a compelling sign thatshell burning is the source of power; it can, however, be challenging to measure lwd (especiallyif UV observations are not available). That non-burning symbiotics generally have much weakerradio emission than burning symbiotics furnishes a more easily attainable observational diagnosticof the burning status.Fortuitously, the low quiescent-state radio luminosity of non-burning symbiotics might alsoallow radio emission to be used as an effective probe of transient, bipolar outflows. For instance,[38] discovered a major radio brightening during the 2016 optical high state of MWC 560 [54].They concluded that the radio brightened as a result of an increase in the power of the well-known(possibly bipolar) outflow in concert with the rise in the accretion rate onto the WD. Selection bias and the missing population of interacting binaries.
Considering our jux-taposition of burning and non-burning symbiotics, it becomes evident that optical spectroscopicsurveys, which excel at finding burning symbiotics, almost certainly miss many (or even most) non-burning symbiotics. Recent and on-going optical spectroscopic surveys are adding to the numberof symbiotics and candidate symbiotics, both in our galaxy (e.g., [9, 10, 100] and in nearby galax-ies [47]. But with a bias toward finding burning symbiotics, these data could tempt us into drawingmistaken conclusions about, for example, number densities and birth rates of symbiotic stars, andthe distributions of their properties. For example, if shell burning is easier to establish and main-tain on low-mass WDs, symbiotics identified by strong lines in their optical spectra would tendto contain low-mass WDs, or higher rates of mass transfer. Indeed, compilations of WD massestimates for optically selected symbiotics do suggest that this sample on average has low-massWDs (e.g., [46]). Conversely, symbiotics with hard X-ray emission are more likely to containhigh-mass WDs (e.g., [39, 40, 15]) and/or lower rates of mass transfer. In the most extreme case,a reliance on optical spectroscopy could mean we have missed almost an entire population of in-teracting binary stars — the non-burning symbiotics. Work by [55] suggesting that an appreciablefraction of the so-called Galactic ridge X-ray emission could be due to WDs accreting from red-giant companions; past detections of X-rays from red giants by [99]); and estimates of the spacedensity of non-burning symbiotics by [51] all raise the intriguing possibility that the population ofnon-burning symbiotics is significant. Finally, because nearby non-burning symbiotic stars powersome of the most dramatic WD jets (e.g., R Aqr, CH Cyg; Fig. 2), unveiling the population ofnon-burning symbiotics could be particularly exciting for the study of astrophysical disks and jets.We are currently laying the groundwork for moving beyond optical spectroscopy to search for thismissing population.
3. Novae: colliding flows explain γ -rays and more Two distinct flows.
Turning to novae, we now direct our attention from WD inflows to out-6 lows and Shocks in Symbiotic Stars and Novae
J. L. Sokoloski flows. At a fundamental level, the ejecta from novae appear to consist of two main components: aslow, dense outflow with a maximum velocity of less than about 1000 km s − and a fast outflowor wind with a maximum expansion speed of several thousand km s − . As a nova remnant ex-pands, the slow flow is observed as a dense core in radio or optical images (e.g., HST images ofV959 Mon and T Pyx, below), whereas the fast flow often takes the form of more extended, bipolarlobes. In addition to the two novae discussed below, the γ -ray bright nova V339 Del affords anice example of this behavior [79, 18]. Because the rapidly expanding outer structures tend to fadewithin a few years, many of the optical images of old nova shells (observed more than a decadeafter the novae eruption (e.g., [19, 21] may record principally the slow-moving component of theejecta. Supporting this idea, these old shells regularly show expansion speeds (size scale dividedby time since eruption) of less than 1000 km s − . In terms of optical line profiles, [18] note that[34, 72, 73, 79] all conclude that high-velocity wings on emission lines detected in early spectraof CO novae “originate in low-density, outlying gas, and that the lower velocity given by the linecore is representative of the expansion rate of the bulk of the ejecta." Critically, the ejecta do notconsist of a structure that expands uniformly from t (the time of the thermonuclear runaway; TNR)– rather, the fast flow plows through and/or around the slow flow, getting shaped by it and givingrise to strong shocks. Equatorial rings.
Within the slower component of the ejecta, the density is often enhancedtoward the equatorial plane (e.g., [70, 7, 14, 92, 8]). This equatorial torus presumably shapes thefaster component of the ejecta into a bipolar morphology. Evidence for slow, equatorial rings andfaster, bipolar outflows comes from the complex profiles of optical emission lines (e.g., [26, 95,20, 75, 87, 76]). Optical and radio imaging support the idea that equatorial rings and bipolar lobesare common (e.g., [19, 107] and the
HST images below). Minor features such as polar blobs,polar rings, general clumpiness, and an early ‘puff’ of ejecta are also often observed, but the mostenergetically important components of the ejecta appear to be the slow and fast flows, which tend toproduce equatorial rings and bipolar lobes, respectively. Furthermore, [8] used radio observationsof the γ -ray bright nova V959 Mon to infer that collisions between a dense equatorial torus anda faster flow led to shocks that accelerated particles, explaining the stunning recent discovery thatmany normal novae produce GeV γ -rays [2, 5]. Generation of γ -rays. Moreover, the scenario for γ -ray production inferred for V959 Monby [8] could be broadly applicable. Relativistic particles that generate γ -rays from classical novae(either via inverse-Compton scattering or the decay of neutral pions) are almost certainly acceler-ated in shocks. Although some nova-producing WDs that are embedded in the wind of a red-giantcompanion (which we refer to as embedded novae ) produce γ -ray emission via external shocksbetween the ejecta and the pre-existing circumbinary material (e.g., in V407 Cyg [1, 58, 63] andV745 Sco [4, 66, 64]), most γ -ray bright novae are not embedded in such dense environments [2].Thus, the location and properties of shocks within nova shells are crucial for understanding hownormal (non-embedded) novae generate γ -rays. With its fortuitously high inclination [68, 88, 76],V959 Mon displayed radio synchrotron-producing shocks between a bipolar flow extending to theeast and west, and a less extended equatorial torus aligned north-south and viewed from the edge[8]. But we know that equatorial rings and shocks are both common in non-embedded novae. Aswe discussed above, imaging at various wavelengths and optical emission-line profiles frequentlyreveal equatorial rings. X-ray emission with energy greater than approximately 1 keV, and radio7 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski synchrotron emission (e.g., [103, 104]), reveal shock-heated gas (e.g., [49]) and particles accel-erated in shocks, respectively. Thus, the ingredients that led to gamma-rays in V959 Mon arecommon. Most novae could therefore perhaps generate γ -rays as in V959 Mon. Figure 3:
HST /WFC3 images of V959 Mon, which was the first classical nova to be detected by
Fermi .Images have been drizzled to 0.03 (cid:48)(cid:48) pixel scale. All frames are 64 pixels = 1.92 (cid:48)(cid:48) square, N up, with identicallogarithmic intensity scaling and stretch (pegged to brightest pixel in F657N in 2014). (a) F657N 2014November. (b) F502N 2014 November. (c) RGB color frame mapping G channel to F502N 2014, R channelto F657N 2014 and B channel left blank. (d) F657N 2015 November. (e) F502N 2015 November. (f) Aswith panel (c), but for the 2015 epoch.
HST observations of V959 Mon.
To examine the dominant structures in the ejecta fromV959 Mon more clearly, we turn to
HST images and imaging spectroscopy. On 2014 November21 and then 2015 November 30, HST observed V959 Mon with the WFC3 camera (and on 19December 2014 and 21 December 2015 with STIS; program 13715 [PI: Sokoloski]). With aninitial discovery date for the nova (in the γ -rays) of 2012 June 19 ([2]; see also [3]), the HST observations captured the state of the remnant approximately 2.5 and 3.5 years after the start of theoutburst. Fig. 3 shows
HST /WFC3 images of V959 Mon through the F657N filter (H α + [NII]; leftcolumn) and F502N filter ([OIII]; right column).The [H α + NII /F657N] images of V959 Mon are dominated by H α emission, which tracesdense, ionized gas. They show the central binary (the unresolved central point source), an overallbipolar structure with a major axis in the east-west direction, and four knots lying on a circular8 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski feature that spans the minor axis (north-to-south) of the remnant. The outermost, bipolar shapeis roughly consistent with the morphology that [76] inferred from the profiles of optical emissionlines and that [8] detected in their VLA images on day 126. The STIS spectroscopic image in-dicates that the eastern lobe is somewhat blueshifted (tilted slightly toward the observer) and thewestern lobe is somewhat redshifted (tilted slightly away from the observer). The two strongestknots, along the north-south axis, are consistent with limb-brightened emission from the edge-onequatorial torus that dominated the radio images on day 615 [8]. Additionally, the kinematics ofthe central circular structure from our two epochs of
HST /STIS imaging spectroscopy suggest thatthe circular ring in the WFC3 images is actually a ring plus caps or a 3-dimensional spherical shellexpanding with a velocity of approximately 1000 km s − [93]. Thus, the HST observations showthat the equatorial torus that presumably shaped the faster, bipolar flow is part of a more complete,spherical core. Interestingly, by three and a half years into the eruption, H α (and [OIII]) emis-sion from the outer, fast flow had almost completely faded, leaving a remnant that was much morespherically symmetric than the original, bipolar morphology. We identify this spherical core —with its equatorial density enhancement — as the slow component of the ejecta.Compared to the H α emission, [OIII] emission from V959 Mon traces more diffuse gas. De-tectable levels of [OIII] emission emanated from the same features as H α — except for the dense,edge-on equatorial torus (see Fig. 3; right side). The extent of the [OIII] remnant in 2014 Novem-ber (day 875 after γ -ray detection) along the major axis was 1.05 (cid:48)(cid:48) (measured from locations thatwere 10% of the peak flux). The spherical shell and two knots to the east and west of the centralpoint source that were quite faint in H α outshined most other features in [OIII]. And some hintsof tails extending outward from the two [OIII] knots are also present. In 2014 December, STISalso detected Ne[V] ( λ α images, the outer lobes almost completely faded in [OIII] by the 2015 November. Wedescribe the HST
WFC3 and STIS observations of V959 Mon in detail in a forthcoming paper[93].
HST observations of T Pyx.
Moving on to another nova-producing binary,
HST observationsof the recurrent nova T Pyx highlight the similarities among even quite disparate novae. We ob-served T Pyx with
HST /WFC3 on 2012 July 12; 2013 January 30, June 19, July 15, and September17-18; and 2014 July 10. We also obtained 9 observations with
HST /STIS (gratings G430L andG750L) spanning this same time period (through programs 12448, 13400, and 13796 (PI: Crotts)).Although T Pyx is generally considered to be a quite unique nova-producing binary, with its shortorbital period (of 1.8 hr; e.g. [71]) and high rate of mass transfer [33, 98], its ejecta contain thesame basic components as that of V959 Mon. Light-echoes and flash ionization in the well-knownold remnant after the recent nova eruption in 2011 demonstrated that its 3-dimensional shape isdominated by an inclined torus [92, 85]. And as with V959 Mon, long-slit spectroscopy of the oldshell by [62] suggests that the inclined torus is part of a more complete, spherical shell. Once theejecta from the 2011 eruption had had a few years to expand,
HST /WFC3 images showed that theyalso contained a central core consistent with an inclined torus (Figure 4; the central torus is mostclearly visible in the latest images, at the bottom of the figure); kinematics of the central structurefrom STIS spectroscopy indicate that the torus is inclined in the same direction as the torus in the9 lows and Shocks in Symbiotic Stars and Novae
J. L. Sokoloski old remnant and expanding with a similar velocity ([94], based on a comparison with the propermotion of knots in the old remnant measured by [78]). Moreover, our
HST /STIS images showthat, as in V959 Mon, the equatorial torus appears to have mechanically shaped a faster flow intobipolar lobes (although they are much fainter in T Pyx than in V959 Mon). The lobes are not easilyrecognizable in
HST images of the old remnant [84, 78]. However, [83] reported an “extendedenvelope” component (in H α + [NII]) in the remnant that could perhaps have been from faster,bipolar lobes.Fig. 4 also shows that the ejecta from T Pyx comprised additional features besides the equa-torial ring and bipolar lobes. HST resolved two bright knots that we will refer to as jet knots from 2013 June onward (in [OIII], although bipolar structure was already evident in the spatiallyresolved STIS spectra from 2012 July), within the eastern and western lobes. Radial velocities ofapproximately ±
800 km s − from STIS spectroscopy combined with proper motions (and takingd=4.8 kpc; [92]) indicates that they moved away from the central binary with a velocity of approx-imately 2000 km s − . They thus moved with a speed of more than twice the average expansionspeed of material in the equatorial torus. Furthermore, if the jet knots moved perpendicular to theorbital plane, then the inclination of the binary must be on the high side of reported range of values.Moreover, the resemblance between the bright jet knots in T Pyx and the relatively fainter knotswithin the lobes of V959 Mon supports conclusions from previous studies that nova remnants haveother pervasive features in addition to equatorial rings and bipolar lobes. We will describe the HST
WFC3 and STIS observations of T Pyx in detail in a forthcoming paper [94].
Secondary features.
Various imaging studies of nova remnants have led to the determinationthat many remnants contain not only bipolar lobes and equatorial rings, but also ‘polar blobs’,‘polar caps’, ‘polar rings’, and knots with tails. Here we speculate that these secondary featuresmight all be related to the main features in our
HST images of the young (few-year-old) remnantsof V959 Mon and T Pyx — if we take into account the possibility that most of the ejecta mass couldbe expelled somewhat later than the initial ’puff’ of material at t (the time of the TNR). Strongevidence for a delay of on the order of a month between the TNR and the expulsion of most ofthe mass in both V959 Mon and T Pyx comes from the evolution of the radio and X-ray emissionfrom shock-heated plasma [59, 8, 35]. In T Pyx, the delayed ejecta clearly overtook and collidedwith slightly slower-moving material ([86] detected P Cyg absorption features during the first withradial velocities of − − − ) about two months after the start of the eruption [59, 8].As discussed by [6], the collision between an initial spherical shell and a later, faster bipolar flownaturally leads to polar caps, blobs, or rings. And some novae experience even more than twoepisodes of mass ejection; V1369 Cen experienced a series of ejections, each expelling materialwith velocities than the previous one (F. Walter, private communication). In any case, the featureswe call jet knots in T Pyx, and the analogous features in V959 Mon, could easily be described aspolar blobs (especially once the bipolar lobes faded). Finally, with faster flows overtaking slowerflows at multiple locations within a given remnant, there are plenty of opportunities for instabilitiesto produce knots. Implications of flow structure on γ -rays. Observations suggesting that the ejecta from novaefundamentally consist of two colliding flows — a slow flow with an equatorial density enhancementand a faster flow that is shaped by the slow flow — have bearing on γ -ray production and mass ejec-tion in these events. For starters, if the V959 Mon scenario of γ -ray production is widespread, then10 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski
Figure 4:
HST /WFC3 imaging of T Pyx. All frames are 34 pixels = 1.3 (cid:48)(cid:48) square, N up, with logarithmicintensity scaling and identical stretch (pegged to brightest pixel in F502N image in each epoch/row). (a) Ato-scale schematic of the potential ejecta structures. (b) F502N 2013 January. (c) A sum of F656N+F658N2013 September (d) F502N 2013 September. This epoch best shows the outer lobes and first reveals the jetknots. (e) F657N 2014 July, this filter provides complete velocity sampling of all components in H α + [NII]. (f) F502N 2014 July. This latest epoch shows significant proper motion in the jet knots. lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski γ -rays are expected to be strongest from systems with slow (less than approximately 1000 km s − ),radiative shocks. Such shocks in novae have high enough densities and low enough expansionspeeds to efficiently trap relativistic protons, which then collide with non-relativistic protons toproduce pions that decay and emit γ -rays [43, 44]. Work by [44] showed this so-called hadronic scenario for the production of γ -rays in novae to be favored over the leptonic scenario of inverseCompton scattering by relativistic electrons. They argued that because the neutral material aheadof the forward shocks can efficiently reprocess X-rays from shock-heated plasma into the optical(at the early times when γ -rays are being produced), the ratio of γ -ray to optical fluxes placesconstraints on the efficiency of particle acceleration. Such constraints for both V1324 Sco andV339 Del were quite high and therefore more consistent with hadronic than leptonic scenarios.Other surprising results from the models explored by [42, 43, 44, 101, 102] are that the observed γ -ray luminosities require shocks in novae to be very powerful (with power comparable to the Ed-dington luminosity for a white dwarf), and that shocks probably convert the kinetic energy of thefast flow into optical emission for at least a few weeks around maximum optical light. Common-envelope interaction.
In addition, by highlighting the importance of material con-centrated in the orbital plane, recent investigations of γ -rays from novae have resurrected past ques-tions about the role of the donor star in unbinding the WD envelope after the TNR (see, e.g., [8]).In other words, is common-envelope evolution important in shaping and/or ejecting the remnantsof classical novae (e.g., [37])? Whether or not common-envelope effects are at play in classicalnovae, this mechanism is unlikely to be important for wide, symbiotic binaries because their donorstars are far from their eruptive WDs. So, does the lack of interaction between the expanded WDphotosphere and its binary companion lead to lower ejection efficiency and therefore lower ejectamasses in some wide, symbiotic binaries? If so, could this difference help explain how quasi-steadyshell burning can persist for up to centuries after novae in some symbiotic stars (such as AG Peg[74, 97]). Regardless of the answers to these questions, the flow structure, γ -ray production, andejection of the WD envelope seem to be strongly linked. Pre-maximum halts, THEAs, and dust.
Finally, new understanding of the flow structureand shocks in novae may also have implications for the shape of optical light curves (includingpre-maximum halts), the transient heavy element absorbing gas (THEA features in optical spectra;[106]), and the formation of dust. Although [24] propose the pre-maximum halts are due to changesin convective energy transfer, and this effect might well have an impact on optical light curves, themulti-component flow structure also naturally leads to a ‘pre-maximum halt’ (as is sometimesobserved in optical light curves; [25]). If the initial optical rise is due to the expansion of anoptically thick photosphere near the outermost edge of the ejecta, then the optical light curve mustturn over or halt as this material becomes optically thin. During the halt, the optical (pseudo-)photosphere recedes with respect to the outer edge of the expanding ejecta. When it reachesthe slow, dense core, with its greater optical depth, the optical brightness would be expected torise a second time along with the increase in size of the expanding core (especially if the opticalbrightness of the core is high due to shock heating). The optical brightness peaks when the corereaches its maximum size before beginning to become optically thin. Near optical maximum, onemight expect to see blueshifted absorption by the material in the (slow) core along with emissionfrom the optically thin fast flow. Such absorption features, with blueshifted velocities of severalhundred km s − , have been observed and described as THEA lines (by [106]). With regard to the12 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski long-standing question of how dust can form in the harsh environment of a nova, [11] have recentlyshown that the dense, cool gas behind the type of radiative shock needed for γ -ray production alsoprovides an ideal environment for the creation of dust grains. Thus, several long-standing problemscould be resolved once we take into account the structure and microphysics of colliding flowswithin the ejecta from novae.
4. Conclusions • We define a WD symbiotic star as a binary in which a red giant transfers enough material toa WD for the interaction to produce an observable signal at some waveband. • The presence or absence of quasi-steady shell burning sets the luminosity of the accretingWD in a symbiotic and affects its appearance across the electromagnetic spectrum. We referto symbiotic stars with and without shell burning on the surface of their WDs as burning and non-burning symbiotic stars, respectively. • Non-burning symbiotic stars are difficult to find in optical spectroscopic surveys. Becausetheir WD masses, mass transfer rates, space densities, and other properties are likely to differfrom those of burning symbiotics, alternative types of searches are needed to draw accurateconclusions about the nature and evolution of this class of wide binaries. • Non-burning symbiotics can be identified in the X-rays, by their UV excess, and by their UVvariability. • Non-burning symbiotics offer a clearer view of the accretion flow from the red giant to theWD than burning symbiotics. • The pervasive production of γ -ray emission shows that shocks are common and importantduring nova eruptions whether they are embedded or not. The emission of γ -rays by non-embedded (classical) novae requires complex outflows that result in internal shocks. • The basic structure of ejecta consists of a core plus halo. Early images of V959 Mon and TPyx share common characteristics of equatorial rings and bipolar lobes that extend to largerangular sizes, even though these novae are dissimilar in many ways. It is likely that thecore has an equatorial enhancement, which interacts with the faster flow (halo) and shapes itbipolar lobes, not only in these novae but most others. • By three and a half years after the start of the eruption, the fast component of the V959 Monhad faded almost beyond detectability. Images of older nova shells might show primarily theslow component, which is more massive and denser than the faster component. • Collisions between the slow-moving material that comprises the core and the faster flowproduce strong shocks that are crucial for accelerating particles and generating gamma-rays.13 lows and Shocks in Symbiotic Stars and Novae
J. L. Sokoloski
The same shocks produce non-thermal radio emission, thermal X-rays, and perhaps a largefraction of optical light, at different phases of the eruption. • Now that they have been recognized as γ -ray sources, novae serve as unique laboratoriesfor the study of particle acceleration. Multiwavelength observations contemporaneous with gamma -ray detection have the potential to allow us to distinguish between leptonic andhadronic processes, and to constrain the shock properties, including the efficiency of par-ticle acceleration. Acknowledgments
We are grateful for conversations with J. Mikołajewska, T. Nelson, G. J. Luna, A. Lucy, M. Ru-pen, L. Chomiuk, and the other members of the ENova collaboration. G. J. Luna and A. Lucy, andgave helpful comments on this manuscript. The authors acknowledge support from HST grantsGO-13400, GO-13796, and GO-13715, as well as NNX15AF19G and AST-1616646.
References [1] The Fermi-LAT Collaboration, A. A. Abdo et. al. 2010,
Science , 329, 817.[2] The Fermi-LAT Collaboration, M. Ackermann, et al. 2014,
Science , 345, 554.[3] C. C. Cheung, et al. 2012, ATel
ApJ , 826, 142.[6] S. M. Chit˘a, et al. 2008,
A&A , 488, L37.[7] D. Chochol et al. 1997
A&A , 318, 908.[8] L. Chomiuk, et al. 2014,
Nature , 514, 339.[9] R. L. M. Corradi et al. 2008,
A&A , 480, 409.[10] R. L. M. Corradi et al. 2010,
A&A , 509, A41.[11] A. M. Derdzinski, B. D. Metzger, and D. Lazzati 2016, arXiv:1610.02401.[12] M. de Val-Borro, M. Karovska, and D. Sasselov 2009,
ApJ , 700, 1148.[13] L. Ducci, et al. 2016,
A&A , 592, A58[14] Eyres et al. 2005,
MNRAS , 358, 1019.[15] R. N. C. Eze, G. J. M. Luna, and R. K. Smith 2010,
ApJ , 709, 816.[16] J. Frank, A. King, and D. J. Raine 2002,
Accretion Power in Astrophysics: Third Edition , (CambridgeUniversity Press).[17] D. K. Galloway and J. L. Sokoloski 2004,
ApJ , 613, L61.[18] R. D. Gehrz et al. 2015,
ApJ , 812, 132.[19] C. D. Gill and T. J. O’Brien,
MNRAS , 300, 221.[20] C. D. Gill & T. J. O’Brien 1999,
MNRAS , 307, 677. lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski[21] C. D. Gill and T. J. O’Brien,
MNRAS , 314, 175.[22] M. Gromadzki, M. Mikołajewski, T. Tomov, L. Bellas-Velidis, A. Dapergolas, and C. Gałan 2006,
AcA , 56, 97.[23] I. Hachisu and M. Kato 2001,
ApJ , 558, 323.[24] Y. Hillman, D. Prialnik, A. Kovetz, M. M. Shara, and J. D. Neill 2014,
MNRAS , 437, 1962.[25] R. Hounsell et al. 2010,
ApJ , 724, 480.[26] J. B. Hutchings 1972,
MNRAS , 158, 177.[27] K. Iłkiewicz, J. Mikołojewska, K. Stoyanov, A. Manousakis, and B. Misalski 2016,
MNRAS , 462,2695.[28] M. Karovska et al. 2010,
ApJ , 710, L132.[29] E. Kellogg, J. A. Pedelty, and R. G. Lyon 2001,
ApJ , 563, L151.[30] E. Kellogg et al. 2007,
ApJ , 664, 1079.[31] J. A. Kennea, et al. 2009,
ApJ , 701, 1992.[32] S. Kenyon,
The Symbiotic Stars ,[33] C. Knigge, A. R. King, and J. Patterson 2000,
A&A , 364, L75.[34] E. A. Kolotilov 1980,
SvAL , 6, 268.[35] J. D. Linford et al. 2015,
ApJ , 805, 136.[36] M. Livio and B. Warner 1984,
Obs , 104, 152.[37] M. Livio, A. Shankar, A. Burkert, and J. W. Truran 1990,
ApJ , 356, 250.[38] A. Lucy, et al. 2017, in preparation.[39] G. J. M. Luna and J. L. Sokoloski 2007,
ApJ , 671, 741.[40] G. J. M. Luna, J. L. Sokoloski, and K. Mukai 2008, in
RS Ophiuchi (2006) and the Recurrent NovaPhenomenon , ASP Conf. Ser., 401, 342[41] G. J. M. Luna, J. L. Sokoloski, K. Mukai, and T. Nelson,
A&A , 559, A6.[42] P. Martin and G. Dubus 2013,
A&A , 551, A37.[43] B. D. Metzger et al. 2014,
MNRAS , 442, 713.[44] B. D. Metzger, T. Finzell, I. Vurm, R. Hascoët, A. M. Beloborodov, and L. Chomiuk, 2015,
MNRAS ,450, 2739.[45] J. Mikołajewska, M. Friedung, S. J. Kenyon, R. Viotti 1988,
The Symbiotic Phenomenon ,Astrophysics and Space Science Library Vol. 145.[46] J. Mikołajewska 2003, in
Symbiotic Stars Probing Stellar Evolution , ASP Conf. Series, 303, 9.[47] J. Mikołajewska, N. Caldwell, and M. M. Shara 2014,
MNRAS , 444, 586.[48] S. Mohamed, Ph. Podsiadlowskin 2007, in , ASP Conf.Ser., 372, 397.[49] K. Mukai and M. Ishida 2001,
ApJ , 551, 1024.[50] K. Mukai, A. Kinkhabwala, J. R. Peterson, S. M. Kahn, and F. Paerels 2003,
ApJ , 586, L77. lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski[51] K. Mukai, et al. 2016,
MNRAS , 461, L1[52] K. Mukai 2017, submitted to PASP.[53] U. Munari and T. Zwitter 2002,
A&A , 383, 188.[54] U. Munari et al. 2016, arXiv:1607.06309.[55] K. Morihana, et al. 2016,
PASJ , 68 (4), 57.[56] U. Mürset, H. Nussbaumer, H. M. Schmid, and M. Vogel,
A&A , 248, 458.[57] U. Mürset, B. Wolff, and S. Jordan 1997,
A&A , 319, 201[58] T. J. Nelson, D. Donato, K. Mukai, J. L. Sokoloski, and L. Chomiuk 2012,
ApJ , 748, 43.[59] T. J. Nelson, et al. 2014,
ApJ , 785, 78.[60] J. S. Nichols, et al. 2007,
ApJ , 660, 651.[61] N. E. Nuñez, T. Nelson, K. Mukai, J. L. Sokoloski, and G. J. M. Luna 2016,
ApJ , 824, 23.[62] T. J. O’Brien and J. G. Cohen 1998,
ApJ , 498, L59.[63] S. Orlando and J. J. Drake 2012,
MNRAS , 419, 2329.[64] S. Orlando, J. J. Drake, and M. Miceli 2017,
MNRAS , 464, 5003.[65] M. Orio, et al. 2007,
ApJ , 661, 1105.[66] M. Orio, V. Rana, K. L. Page, J. L. Sokoloski, and F. Harrison 2015,
MNRAS , 448, L35.[67] B. Paczy´nski and A. N. ˙Zytkow 1978,
ApJ , 222, 604.[68] Page et al. 2013,
ApJ , 768, L26.[69] D. Pandel et al. 2005,
ApJ , 626, 396.[70] Paresce et al. 1995,
ApJ , 442, L57.[71] J. Patterson, et al. 2017,
MNRAS , 466, 581.[72] T. P. Prabhu & G. C. Anupama 1987,
Ap&SS , 131, 479.[73] T. P. Prabhu & G. C. Anupama 1987,
JApA , 8, 369.[74] G. Ramsay, J. L. Sokoloski, G. J. M. Luna, and N. E. Nuñez 2016,
MNRAS , 461, 3599.[75] Ribeiro et al. 2011,
MNRAS , 412, 1701.[76] V. A. R. M. Ribeiro, U. Munari, and P. Valisa 2013,
ApJ , 768, 49.[77] R. Sahai, J. Sanz-Forcada, C. Sánchez Contreras, and M. Stute 2015,
ApJ , 810, 77.[78] B. E. Schaefer, A. Pagnotta, and M. M. Shara 2010,
ApJ , 708, 381.[79] G. H. Schaefer et al. 2014,
Nature , 515, 234.[80] E. R. Seaquist, A. R. Taylor, and S. Button 1984,
ApJ , 284, 202.[81] E. R. Seaquist and A. R. Taylor 1990,
ApJ , 349, 313.[82] E. R.. Seaquist, M. Krogulec, and A. R. Taylor 1993,
ApJ , 410, 260[83] M. M. Shara, A. F. Moffat, R. E. Williams, and J. G. Cohen,
ApJ , 337, 720.[84] M. M. Shara, et al. 1997, AJ , 114, 258 lows and Shocks in Symbiotic Stars and Novae J. L. Sokoloski[85] M. M. Shara, et al. 2015,
ApJ , 805, 148.[86] S. N. Shore, T. Augusteijn, A. Ederoclite, and H. Uthas,
A&A , 533, L8.[87] S. N. Shore et al. 2013,
A&A , 559, L7.[88] S. N. Shore et al. 2013,
A&A , 553, A123.[89] A. Skopal 2005,
A&A , 440, 995.[90] J. L. Sokoloski, L. Bildsten, and W. C. G. Ho 2001,
MNRAS , 326, 553.[91] J. L. Sokoloski and L. Bildsten 2010,
ApJ , 723, 1188.[92] J. L. Sokoloski, A. P. S. Crotts, S. Lawrence, and H. Uthas2013,
ApJ , 770, L33.[93] J. L. Sokoloski et al., 2017, in preparation.[94] J. L. Sokoloski et al., 2017, in preparation.[95] J. Solf 1983,
ApJ , 273, 647.[96] M. Stute, G. J. M. Luna, and J. L. Sokoloski 2011,
ApJ , 731, 12.[97] T. V. Tomov, K. A. Stoyanov, and R. K. Zamanov 2016,
MNRAS , 462, 4435.[98] H. Uthas, C. Knigge, and D. Steeghs 2010,
MNRAS , 409, 237.[99] M. van den Berg, et al. 2006,
ApJ , 647, L135.[100] K. Viironen, et al. 2009,
A&A , 502, 113.[101] , A. Vlasov, I. Vurm, and B. D. Metzger 2015,
MNRAS , 463, 394.[102] I. Vurm and B. D. Metzger 2016, arXiv:1611.04532.[103] J. H. S. Weston, et al. 2016,
MNRAS , 457, 887.[104] J. H. S. Weston, et al. 2016,
MNRAS , 460, 2687.[105] J. H. S. Weston, 2016,
Radio observations as a tool to investigate shocks and asymmetries inaccreting white dwarf binaries , PhD, Columbia University, Department of Astronomy.[106] R. Williams, E. Mason, M. Della Valle, and A. Ederoclite 2008,
ApJ , 685, 451.[107] P. A. Woudt, et al. 2009,
ApJ , 706, 738., 706, 738.