Interacting Supernovae: Types IIn and Ibn
aa r X i v : . [ a s t r o - ph . H E ] D ec Interacting Supernovae: Types IIn and Ibn
Nathan Smith
Abstract
Supernovae that show evidence of strong shock interaction between theirejecta and pre-existing, slower circumstellar material (CSM) constitute an interest-ing, diverse, and still poorly understood category of explosive transients. The chiefreason that they are extremely interesting is because they tell us that in a subset ofstellar deaths, the progenitor star may become wildly unstable in the years, decades,or centuries before explosion. This is something that has not been included in stan-dard stellar evolution models, but may significantly change the end product andyield of that evolution and complicates our attempts to map SNe to their progenitors.Another reason they are interesting is because CSM interaction is a very efficient en-gine for making bright transients, allowing super-luminous transients to arise fromnormal SN explosion energies, and allowing transients of normal supernova lumi-nosities to arise from sub-energetic explosions or low radioactivity yield. CSM in-teraction shrouds the fast ejecta in bright shock emission, obscuring our normalview of the underlying explosion, and the radiation hydrodynamics of the interac-tion is challenging to model. The CSM interaction may also be highly non-spherical,perhaps linked to binary interaction in the progenitor system. In some cases, thesecomplications make it difficult to definitively tell the difference between a core col-lapse or thermonuclear explosion, or to discern between a non-terminal eruption,failed supernova, or weak supernova. Efforts to uncover the physical parameters ofindividual events and connections to possible progenitor stars make this a rapidlyevolving topic that continues to challenge paradigms of stellar evolution.
Nathan SmithSteward Obervatory, University of Arizona, 933 N. Cherry Ave., Tucson, AZ 85721, USA, e-mail: [email protected]
Type IIn superovae (SNe IIn hereafter), defined broadly as SNe that exhibit brightand narrow Balmer lines of H in their spectra, were recognized as a distinct class ofobjects relatively late compared to other SN subtypes (see Schlegel 1990 and Fil-ippenko 1997 for early background). This was spurned largely by observations ofthe classic SN IIn 1988Z and connecting it to unusual properties in some other SNe.Our understanding of this class and various subclasses has been evolving continu-ously since then as more SNe are discovered and the diversity of SNe IIn grows. Therelated class of SNe Ibn, showing narrow lines of He in the optical spectrum (andrelatively weak H lines), was clearly identified only 10 years ago following studiesof the prototype SN 2006jc (Pastorello et al. 20017; Foley et al. 2007).Interpreting observations of interacting SNe is fundamentally different fromstudying “normal” SNe. For our purposes here, “normal” SNe are those cases wherethe observed emission comes from a photosphere receding (in mass coordinates)back through freely expanding SN ejecta, or radiation from the optically thin SNejecta at late times that are heated internally by radioactivity. These are the mainsubtypes of II-P, II-L, IIb, Ib, Ic, Ic-BL discussed elsewhere in this volume. SNe IInand Ibn represent a physically very different scenario, where the usual diagnosticsand physical pictures cannot be applied. As a means of introduction to the topic,we begin with two warnings that should be heeded when thinking of SNe IIn andrelated subtypes of interacting SNe:Warning 1. The main engine is conversion of kinetic energy to light in a shockat the outside boundary of the SN. It is not shock-deposited energy leaking outof a homologously expanding envelope (as in early phases of a SN II-P), nor isit internal heating from radioactive decay (most SNe at peak and at late times).Instead, the observed radiation comes primarily from a shock crashing intodense CSM which - unfortunately from the point of view of predictive models- makes the situation complicated and malleable. “Malleable” here means thatit is hard to calculate predictive models because the shock luminosity can riseand fall somewhat arbitrarily depending on the radial density structure of theCSM. One should not take the absence of normal signatures (broad lines in theSN ejecta photosphere or nebular emission lines) as an indication that theseare not true SNe, because CSM interaction can mask these normal signatures.Moreover, radiation from the shock can propagate back into the SN ejecta,changing their appearance.Warning 2. SNe IIn and related categories are not really a supernova type (ormore accurately, not an intrinsic explosion type), but an external phenomenon associated with CSM interaction. Any type of core collapse or thermonuclearSN (or for that matter, any non-SN explosive outflow) can appear as a SN IIn nteracting Supernovae 3 or SN Ibn. All that is required is fast ejecta with sufficient energy crashinginto slower ejecta with sufficient density. This is a cause of much confusionand uncertainty.As a consequence of these differences compared to other SNe, we find a verydiverse range of properties in interacting SNe. There is not a singular evolutionarypath that leads to an interacting SN, nor is there a singular mapping of SNe IInto a specific progenitor type. For example, we usually associate SNe II-P as com-ing from moderately massive red supergiants of 8-20 M ⊙ (Smartt 2009), but suchassignments are not so straightforward for SNe IIn. It can be difficult to tell thedifference between a SN IIn that arises from the core collapse of a very massiveluminous blue variable (LBV) or LBV-like star, and a thermonuclear explosion ofa white dwarf with dense CSM. SNe IIn can arise any time there is dense H-richmaterial sitting around a star that explodes, which might happen in a number of dif-ferent ways. Exploring this diversity in progenitors as well as in CSM and explosionproperties is a main emphasis of current work because it has important implicationsfor mass loss and instability in the late phases of stellar evolution.The detailed physical picture and time evolution of interacting SNe may also bevery different from normal SNe. Instead of a more-or-less spherical photosphere re-ceding through outflowing ejecta, we have radiation from a thin shell, which maystart out as an optically thick sphere, but may transition to a limb-brightened shellthat is transparent in the middle and optically thick at the edges. We may have un-derlying explosions that would otherwise be too faint to be observed (for example,explosions producing very low radioactive yield), and that don’t contribute to thetraditional populations of extragalactic SNe, but nevertheless yield bright transientsthrough their CSM interaction. Thus, we should be mindful that some SNe IIn maybe powered by entirely new (or so far observationally unrecognized) classes of ex-plosions other than SNe II-P, II-L, IIb, Ib, and Ic. Much of this is still little exploredor uncharted territory.For this chapter, it is also worth making a distinction of scope. Actually, all SNeare interacting with surrounding material at some level, because space is not a truevacuum and an interaction arises naturally even when fast SN ejecta expand into lowdensity ISM or their progenitor’s normal wind (this eventually forms a standard SNremnant). Whether this can be detected in extragalactic SNe shortly after explosion— and in what waveband it is detected — depends to a large extent on the CSM den-sity. For the purposes of discussion here, we take “Interacting SNe” to mean thosewith extraordinarily strong CSM interaction that leads to a radiative shock at earlytimes after explosion, producing narrow lines in the visible-wavelength spectrumand enhanced optical continuum luminosity at early times. We exclude otherwisenormal SNe that show signs of interaction based on radio or X-ray emission, al-though it should be kept in mind that these represent the tail of a continuum of CSMdensity stretching from superluminous SNe, through more traditional SNe IIn, andon to lower CSM density. N. Smith
When a SN explodes inside a dense cocoon of CSM, a strong shock is driven intothe CSM, creating a basic structure as depicted in Figure 1. A forward shock (FS)is driven outward into the CSM, and a corresponding reverse shock (RS) is drivenback into the SN ejecta. In the simplest picture, four zones are delineated whereheated material can radiate and contribute to the observed spectrum:Zone 1. The unshocked CSM outside the forward shock (photoionized).Zone 2. The swept-up CSM that has been hit by the forward shock.Zone 3. The decelerated SN ejecta that have encountered the reverse shock.Zone 4. The freely expanding SN ejecta.In normal SNe, we only see emission from Zone 4 at visible wavelengths asthe photosphere recedes back through the freely expanding ejecta, creating broademission and absorption lines whose velocities decrease smoothly with time as the
Fig. 1
Sketch of the basic picture in CSM interacting SNe. Four different zones are noted withnumbers: (1) the pre-shock CSM, (2) the shocked CSM, (3) the shocked SN ejecta, and (4) thefreely expanding SN ejecta. These zones are divided by boundaries corresponding to the forwardshock, the reverse shock, and the contact discontinuity between the shocked CSM and shockedejecta where material cools, mixes via Rayleigh-Taylor instabilities, and piles up. This is oftencalled the cold dense shell (CDS) in a SN IIn or Ibn. The squiggly radial lines are meant as areminder that X-rays and UV radiation generated in the shock can propagate out to the CSM orinward to the unshocked ejecta, changing the physical state of the gas there. A zoom-in of zones2 and 3 is shown at the right. In practice, efficient radiative cooling can cause these two zones tocollapse to very thin layers, and mixing can make them merge into one thin clumpy shell. Thisfigure is adapted from Smith et al. (2008).nteracting Supernovae 5 emitting layer recedes through the monotonic outflow. Zones 2 and 3 are sometimesdetected in X-rays or radio emission if the CSM is dense enough. Zone 1 might beseen in absorption lines, although this is rare.In an interacting SN, any of these 4 zones can contribute strongly to the emittedvisible-wavelength spectrum. In fact, the continuum photosphere moves through all4 zones as time proceeds, and their relative contributions to line emission changewith time. The basic picture of this interaction, which can differ widely in detailfrom case to case, has been described many times in the literature (Chugai 1997,2001; Chugai & Danziger 1994; Chugai et al. 2004; Smith et al. 2008, 2010a, 2015;Dessart et al. 2015).At early times, radiation from the shock propagates upstream and causes an ion-ized precursor. At this stage (sometimes only lasting a day or two, sometimes severalweeks or months), the electron scattering continuum photosphere is actually in theunshocked CSM, obstructing our view of the shock. The spectrum usually appearsas a smooth blue continuum with narrow emission lines that have broad Lorentzianwings. These wings are not caused by Doppler motion of shocked gas, but by elec-tron scattering as narrow line photons work their way out of the optically thickCSM.As the pre-shock CSM density drops at larger radii, eventually the photosphererecedes into zones 2 and 3, and we begin to see a transformation in the spectrum.This usually occurs around the time of peak luminosity. Strong intermediate-width(few 10 km s − ) components of permitted lines appear, sometimes accompanied byP Cygni absorption. Here we are seeing radiation from the post-shock gas. Becauseof strong radiative cooling (which occurs by definition in an interacting SN), Zones2 and 3 can collapse into a geometrically very thin layer, and mixing at the contactdiscontinuity may cause these two zones to merge. The standard picture is that thiscooled gas is very dense and piles up in a “cold dense shell” (Chugai et al. 2004).This gas is continually reheated by the shock, and emits strongly in lines like H a or He I that appear as the strongest intermediate-width lines in the spectra of SNeIIn and Ibn, respectively. This line emission continues to be strong even after thecontinuum photosphere recedes further, because the CDS is reheated by X-ray andUV radiation from the FS and RS.After peak, the continuum photosphere may recede back into Zone 4 in the freelyexpanding SN ejecta. If the CSM interaction is relatively weak or shortlived, an ob-server might then see a fairly normal SN spectrum with broad P Cygni lines atthe end of the photospheric phase, or typical signatures of SN ejecta heated by ra-dioactive decay in the nebular phase. However, if CSM interaction is strong andsustained, an observer might never actually see a normal SN ejecta photosphere ornebular phase, and the P Cygni absorption may be filled in. The underlying SN(heated by the original shock deposition or radioactive decay) may fade in a fewmonths, whereas CSM interaction might remain very bright for many months oryears. How long this CSM interaction stays bright is determined mainly by the du-ration and speed of the pre-SN mass loss, not by the explosion. By the time theCSM fades enough to become transparent, the inner SN ejecta may be too faint todetect on their own, and the ongoing CSM interaction may continue to dominate N. Smith the spectrum at late times. In fact, the fast SN ejecta may be heated predominantlyby radiation from the CSM interaction shock that propagates backward into the out-ermost ejecta, creating a very different time dependent spectrum than is seen in anormal SN heated from the inside-out by radioactive decay. One must bear thesedifferences in mind when interpreting SNe IIn spectra, especially at late phases.Of course, if the geometry is asymmetric, any of these zones can be seen simulta-neously and at different characteristic velocities, potentially making the interpreta-tion quite complicated. Absorption features may or may not be seen, depending onviewing angle. Even if the interaction is spherically symmetric, we might see mul-tiple zones simultaneously because long path lengths through the limb-brightenededges of the CDS may remain optically thick while material along our line of sight inthe CDS has already become transparent. Very strong and non-monotonic changesin density and velocity make this a complicated situation that presents a significantchallenge for radiation transfer codes. For the time being, unraveling the variousclues in the spectral evolution is something of an art form.The total luminosity generated by CSM interaction in this picture can be highbecause a radiative shock is a very efficient engine to convert kinetic energy intovisible-wavelength light. Whereas a normal SN II might have a total radiated en-ergy that is 1% of its kinetic energy, a Type II SN can have a total radiated energycloser to 50% or more of the total explosion kinetic energy. The luminosity of CSMinteraction L depends essentially on the rate at which CSM mass is entering the for-ward shock, which of course depends mostly on the progenitor’s mass loss rate, andis typically expressed as L = wV , (1)where w is the wind density parameter w = p R r , or w = ˙ M / V CSM , and V CDS isthe value for the evolving speed of the CDS. From the intermediate-width lines inoptical spectra we can measure the value of V CDS (be careful not to do this at earlytimes when electron scattering in the CSM is determining the line width; at earlytimes the true value of V CDS is hidden from the observer, and it may be much higherthan when it is seen at later times). We can also measure V CSM from the narrow com-ponent due to pre-shock gas, especially if a narrow P Cygni profile is observed (notehere that higher resolution spectra with R ≃ L ) derived from photometry (note that for V and R bands, the bolomet-ric correction may be quite small if the apparent temperature is around 6000-7000K), we can then turn this equation around to estimate the progenitor star’s mass-lossrate ˙ M CSM = L V
CSM ( V CDS ) , (2)that occurred at some time t = t exp ( V CDS / V CSM ) in the past, where t exp is the timeafter explosion to which these measurements correspond (careful again, as t exp hasits own potentially large uncertainties). nteracting Supernovae 7 Fig. 2
Sketch of the morecomplicated interaction whenthe CSM is highly asymmet-ric; in this example a disk-likedistribution surrounds theprogenitor. (a) The initial pre-explosion state is a progenitorwith a wind, as well as a densedisk at radii of ∼
10 AU. (b)Immediately after explosion,narrow lines may arise eitherfrom a photoionized densewind or from the disk. StrongCSM interaction with the diskoccurs immediately and en-hances the early luminosity,but the emergent spectrummay be dominated by narrowlines with Lorentzian wings.(c) CSM interaction with thedisk has slowed the forwardshock in the equatorial plane,but the SN ejecta expandrelatively unimpeded in theless dense polar regions. Af-ter a few days, the normalSN ejecta photosphere mayexpand so much that it com-pletely envelopes the CSMinteraction occurring in thedisk. The enveloped CSMinteraction can now heat theoptically thick ejecta fromthe inside and contribute sig-nificantly to the total visibleluminosity, even if no nar-row lines are seen, and maycause the photosphere to beasymmetric. (d) At late times,the SN ejecta photosphererecedes. At this time, theCSM interaction in the diskis exposed again (now withlarger radial velocities be-cause it has been accelerated),and intermediate-width andpossibly doubled-peaked linesmay be seen from the ongoinginteraction. This sketch wasoriginally invoked to explainthe observed evolution ofPTF11iqb (Smith et al. 2015). N. Smith
This sort of analysis is oversimplified, but it is sufficient to give us a ballpark ideaof the sorts of mass-loss rates that are required to produce a Type IIn supernova.In order to make the CSM-interaction luminosity high enough that it can competewith the photosphere of a typical SN (say 3 × L ⊙ , or − V CSM = 100 km s − and V CDS = 2500 km s − , we requirea huge mass-loss rate of order 0.01 M ⊙ yr − . SNe IIn tend to be more luminousthan − V CDS is probably faster than 2500 km s − in theearly optically thick phases near peak). This mass-loss rate is well beyond the realmof normal stellar winds (Smith 2014). We return to the implications for progenitorslater.For a constant mass-loss rate in the progenitor wind (and hence, a single valueof w ), L will usually drop with time as the forward shock decelerates. Equation (1)does a pretty good job of describing the visible continuum luminosity as long asthe material is optically thick enough to reprocess most or all of the X-ray and UVluminosity into visible continuum radiation. It doesn’t work so well at very earlytimes during the rise or at peak when diffusion is important, and it doesn’t workat later times when the X-rays begin to leak out and the H a line emission becomesvery strong. At these phases, Equation (2) will underestimate the required progenitormass-loss rate if L is derived from the visible light curve. At later times, one needsto adopt a very uncertain efficiency factor in order to derive physical parametersfrom observations. The efficiency of converting kinetic energy into visible light canvary widely depending on CSM density (van Marle et al. 2010), CSM composition(H-free gas tends to be more transparent), and geometry, as discussed next. Figure 1 is a good place to begin when picturing the physics of CSM interaction,but it may be something of a fantasy. One exceedingly obvious result from 20 yearsof
Hubble Space Telescope imaging of circumstellar shells is that they are oftennot spherically symmetric, and this is especially true of massive stars and eruptivemass loss. Bipolar shells and equatorial disks or rings are common around evolvedmassive stars.This has an impact on our interpretation of observed signatures of CSM interac-tion, because when spherical symmetry is broken, it can affect the qualitative ap-pearance of the spectrum, and it can change line profile shapes. More importantly, itcan dramatically change the inferred total energy and mass budgets, and sometimesit can hide evidence of CSM interaction altogether. If one assumes spherical sym-metry in an interacting SN that is actually not spherically symmetric, one can be ledto astoundingly wrong conclusions. In this case, the “simplest” assumptions may besimply inappropriate.Figure 2 shows an example of how the scenario in Figure 1 might be modifiedif a SN is surrounded by an equatorial disk-like distribution of CSM as opposed nteracting Supernovae 9 to a spherical shell (this is borrowed from Smith et al. 2015, where much of thistopic is discussed in more detail). A key point is that if the dense CSM is disk-like,then strong CSM interaction only occurs in a small fraction of the total solid angleinvolving the equatorial regions (see Mauerhan et al. 2014; Smith et al. 2014). Thevast majority of the SN ejecta (say, 90% if the disk only intercepts 10% of the solidangle seen by the SN) can expand unimpeded in the polar directions. Thus, as faras the majority of the SN ejecta are concerned, their kinetic energy is not tappedby CSM interaction, and they would emit a normal broad-lined spectrum. For theunlucky 10% of the SN ejecta in the equator, their kinetic energy is converted intoheat and radiation by the shock. In this way, CSM interaction with a disk can createSNe that are overluminous compared to normal SNe II-P, for example, but not nearlyas luminous as the class of super-luminous SNe. This may be applicable to manySNe IIn.
Enveloped, Swallowed, and Hidden CSM Interaction.
There is a nonintuitivebut very interesting result that can occur from interaction with asymmetric CSM,which is that the resulting SN might not show any narrow lines even if much of itsluminosity comes from strong CSM interaction.
This can occur in the following way:Imagine that a SN progenitor is surrounded by a dense CSM disk located at a radius10-20 AU. The initial shock interaction might produce strong narrow lines in the firstfew days. In the disk, the fast SN ejecta are decelerated, creating a strong shock thatproduces a moderately high luminosity and bright narrow-line emission from thisCSM interaction. However, the remaining 90% of the SN ejecta expand unimpededas if there was no CSM interaction. The normal SN ejecta photosphere quicklyexpands to a radius of 10s of AU in a few days. At such time, the optically thick SNejecta can completely envelope the equatorial CSM interaction, hiding some or allof the narrow and intermediate-width lines that arise due to CSM interaction. Theextra radiated energy generated by the equatorial CSM interaction is now unableto escape freely because it has been swallowed by the opaque fast ejecta, and thisradiative energy is instead deposited in the interior of the optically thick SN ejectaand must diffuse out. It will therefore add extra heat to the outer envelope, perhapsmimicking extra deposition of radioactive decay energy. The asymmetric injectionof heat in this way will likely make the photosphere asymmetric. Viewed from theoutside at a time of roughly a month after explosion, one might then observe a broad-lined SN photosphere with enhanced luminosity and perhaps a longer than normalduration, unusual line profiles, or significant polarization. At this stage, it wouldbe difficult to tell the difference between some small extra amount of radioactiveheating or buried CSM interaction.All hope is not lost, though, because an observer might then be able to deduce thatCSM interaction was contributing to the main light curve all along by watching untillate times. After the SN recombination photosphere recedes, the enveloped CSMinteraction region is exposed once again. When this happens, the dense CSM diskthat has been swept up into a cold dense shell (or cold dense torus) and acceleratedto a few thousand km s − may now be seen again via strong intermediate-widthline emission. Prominent examples of this are SN 1998S, PTF 11iqb, SN 2009ip,and even the SN IIb 1993J (Leonard et al. 2000; Pozzo et al. 2004; Smith et al. ergs of kinetic energy. The range of luminos-ity from the strongest CSM interaction in super-luminous SNe down to cases withminimal extra luminosity provided by CSM interaction can be explained either byramping down the density of the progenitor’s wind, or by increasing the degree ofasymmetry in dense CSM. The strongest clues that significant CSM asymmetry ispresent are from: (1) spectropolarimetry (e.g., Mauerhan et al. 2014), (2) asymmet-ric line profile shapes (although some asymmetry can also be caused by dust oroccultation by the SN photosphere itself), or (3) the time evolution of velocities (forexample, seeing very fast speeds in the SN ejecta that persist after the strongest CSMinteraction has subsided; Smith et al. 2014). The last point confirms unequivocallythat not all the fast SN ejecta participated in the CSM interaction. There may also besome special observed signatures of asymmetry at certain viewing angles (edge-on,for example) and specific times. If one does recognize such signatures of asymme-try, one must realize that the progenitor mass-loss rate inferred from equation (2) aswell as the SN explosion energy are not just lower limits, but likely underestimatesby a factor of ∼
10. Of course, one can also invoke differences in explosion energy,explosion mechanism, SN ejecta mass, and radioactive yield to also contribute tothe observed diversity.
The main physical parameters that determine the observed properties of a typicalcore-collapse SN are the mass of ejecta, kinetic energy of the explosion (as wellas the mass of synthesized radioactive material), and the composition and structureof the star’s envelope at the time of explosion. This leads to the wide diversity inobserved types of normal ejecta-dominated core-collapse SNe and thermonuclearSNe (II-P, II-L, IIb, Ib, Ic, Ia).For interacting SNe, we have all these same free parameters of the underlyingexplosion, as well as the possibility of SNe with low or no radioactive yield and awider potential range of explosion energy — but to these we must also add the vari-able parameters associated with the CSM into which the SN ejecta crash: CSM massand/or mass-loss rate, radial distribution (speed and ejection time before explosion),CSM composition, and CSM geometry. Given these parameters, it may not be sur-prising that the class of interacting SNe is extremely diverse, and observations arecontinually uncovering new or apparently unique characteristics. Moreover, a SNcan change type depending on when it is seen. An object that looks like a SN IIn nteracting Supernovae 11 in the first few days can morph into a normal SN II-L or IIb if it undergoes en-veloped/asymmetric CSM interaction, and may then return to being a Type IIn atlate times, as discussed in the previous section.There are, however, some emerging trends among interacting SNe. The list belowattempts to capture some of these emerging subtypes among interacting SNe thatappear to share some common and distinct traits. The reader should be advised,however, that this is still a rapidly developing field, so this is neither a definitive nor acomplete list, and moreover, there are observed events that seem to skirt boundariesor overlap fully between different subtypes. Further subdivisions are likely to beclarified with time. The list below includes a descriptive name and some prototypicalor representative observed examples that are often mentioned (again, not a completelist).
Superluminous IIn; compact shell (SN 2006gy).
SN 2006gy was the first super-luminous SN to be discovered, and it remains extremely unusual even though severalother SLSNe IIn have since been found (see Smith et al. 2007, 2010a; Ofek et al.2007; Woosley et al. 2007). It had a slow rise to peak ( ∼
70 days) and faded withinanother 150 days or so, which is unlike the other SLSNe IIn that seem to fade veryslowly and steadily from peak. Also unusual was that it had strong intermediate-width P Cygni absorption features in its spectra, strong line-blanketing absorptionin the blue, and narrow P Cygni features from the CSM. It is thought to have arisenfrom a relatively compact and opaque CSM shell, where CSM interaction mostlysubsided within a year. In this case, the CSM is thought to have arisen from a sin-gle eruption that ejected ∼ M ⊙ about 8 years prior to core collapse (Smith et al.2010a). SN 2006gy is one of the best observed SNe IIn and is often discussed, butreaders should be aware that it is not at all typical. Superluminous IIn; extended shell (SN 2006tf, SN 2010jl, SN 2003ma).
Theseare basically the superluminous version of the SN 1988Z-like subclass, discussednext. Unlike SN 2006gy, they show strong, smooth blue continua without much lineblanketing, and little or no P Cygni absorption in their intermediate-width lines.The also tend to fade slowly and steadily, in some cases remaining bright for severalyears as the shock runs through an extended dense shell (e.g., Smith et al. 2008;Rest et al. 2011; Fransson et al. 2014).
Enduring IIn (SN 1988Z, SN 2005ip).
These SNe IIn show a smooth blue contin-uum superposed with strong narrow and intermediate-width H lines, and in somecases even broad components (Chugai & Danziger 1994; Aretxaga et al. 1999;Smith et al. 2009). Some cases, such as SN 2005ip, have strong narrow coronalemission lines (Smith et al. 2009), implying photoionization of dense clumpy CSMby X-rays generated in the shock. They tend to be more luminous than SNe II-P, butnot as bright as SLSNe. These objects tend to fade very slowly and show signs ofstrong CSM interaction for years or even decades after discovery; SN 1988Z is still going strong. One can picture these as a stretched-out version of the SLSNe IIn, inthe sense that they have somewhat lower luminosities at peak, but they last longer,eventually sweeping through similar amounts of total CSM mass (of order several to 20 M ⊙ ). While SLSNe have CSM that is so dense that it requires eruptive LBV-like mass loss in the years or decades before explosion, these enduring SNe IInlike SN 1988Z and SN 2005ip can potentially be explained by extreme RSG windsblowing for centuries before core colllapse. If there is such a thing as a “standard”SN IIn, this is probably what most IIn-enthusiasts have in mind. Transitional IIn (SN 1998S, PTF11iqb, SN 2013cu).
This class of SNe IIn onlyshows fleeting signatures of CSM interaction that disappear quickly, and may beentirely missed if the SN isn’t discovered early after explosion. SN 1998S was one ofthe nearest and best studied SNe IIn that helped shape our understanding of SNe IIn,and so it has often been referred to as “prototypical” – but it really isn’t. SN 1998Swasn’t very luminous and its spectral signatures of CSM interaction disappearedpretty quickly, transitioning into a broad-lined ejecta-dominated photosphere withina couple weeks (Leonard et al. 2000; Shivvers et al. 2015). This indicated that thetotal mass of CSM was actually quite modest (of order 0.1 M ⊙ or so), substantiallydifferent from the types noted above. If it had not been caught so early, SN 1998Smight not have been classified as a Type IIn at all. An older example of this wasSN 1983K (Nielmela et al. 1985). More recently, a few other related objects havebeen found (SN 2013cu & PTF11iqb; Gal-Yam et al. 2014; Smith et al. 2015),which only showed a Type IIn spectrum for the first few days after discovery (andthese cases were thought to have been discovered very early; within 1 day or soof explosion). PTF11iqb and SN 2013cu then morphed into broad lined SNe withspectra similar to SNe II-L or IIb, respectively. At late times, PTF11iqb showedstrong CSM interaction again, very similar to SN 1998S and the Type IIb SN 1993J.We do not know how common these are. While SNe IIn represent roughly 8-9% ofcore collapse SNe (Smith et al. 2011), this statistic does not include SNe that onlyexhibit Type IIn characterstics for a brief window of time and then morph into othertypes after a few days; thus, early strong CSM interaction with dense inner windsmight be fairly common among the larger population of ccSNe. The corresponding(potentially very important) implication is that many core collapse SNe may suffera brief pulse of episodic mass loss shortly before core collapse. The reason for thisis not yet known, but is probably related to the final nuclear burning stages and mayhave important implications for models of core collapse. Late-time interaction in otherwise normal SNe (SN 1993J, SN 1986J).
Theoverlap with the previous subclass is probably considerable, but it is worth mention-ing that some otherwise normal SNe (perhaps appearing normal simply because wemissed the early IIn signatures in the first day or so) show particularly strong CSMinteraction at late times. Some of the more common cases are SNe IIb and SNe II-L, for which there are well-studied and famous nearby examples like SN 1993J andSN 1986J (Matheson et al. 2000; Milisavljevic & Fesen 2013). For these, the tran-sition from a SN into a SN remnant is somewhat blurry, and some of these are wellknown nearby aging SNe. Because this interaction is most apparent when the SNehave faded after the first year, this phenomenon can only be studied in nearby galax-ies. These may be caused by strong RSG winds, or by equatorial CSM deposited bybinary interaction in objects like SNe IIb that are more clearly associated with the nteracting Supernovae 13 end products of binary evolution. The reason why this CSM still resides close to thestar at the time of death is still unclear, and (as for the previous subclass) may hint atsome rapid and dramatic changes in the final nuclear burning sequences in a widerfraction of SN progenitors than just standard SNe IIn.
Delayed onset, slow rise, multi-peaked (SN 2009ip, SN 2010mc, SN 2008iy,SN 1961V, SN 2014C).
In contrast to the previous subclass, some SNe IIn showlittle or no signs of CSM interaction at first, but then rise to the peak of CSM in-teraction after a delay of months or years. This delay is presumably due to the timeit takes for the fastest SN ejecta to catch up to a CSM shell that was ejected a yearor more before core collapse. The underlying SN ejecta photosphere could be rel-atively faint at first if the progenitor was a more compact BSG, like SN 1987A,or an LBV; the faintness of the initial SN might cause the initial transient to bemissed altogether, or mistaken as a pre-SN eruption rather than the actual SN, sinceit might precede the delayed onset of peak luminosity that actually arises when theejecta overtake the CSM. It must be admitted, however, that from the time of peakonward, these tend to appear as relatively normal SNe IIn; as such, the distinctionbetween this “delayed onset” subclass and other SNe IIn might be artificial, andonly distinguishable in cases with fortuitous pre-discovery or pre-peak data. Theamount of delay in the onset of CSM interaction bears important physical informa-tion about the elapsed time between the pre-SN eruption and core collapse. In thecase of SN 2009ip, the delay of ∼
40 days made sense, as fast SN ejecta caught upto CSM moving at 10% the speed, associated with eruptions that were actually ob-served a year or so before the SN (Mauerhan et al. 2013a, 2014; Smith et al. 2014).SN 1961V had a light curve consistent with a normal SN II-P for the first ∼
100 days,followed by a brighter peak (Smith et al. 2011b). In SN 2014C, the delayed onsethad different chemical properties, where a H-poor stripped-envelope SN crashedinto a H-rich shell after a year (Milisavljevic et al. 2015). In some cases, however,the delayed onset is rather extreme and the CSM interaction strong; SN 2008iy ap-peared to be a relatively normal core collapse SN in terms of luminosity, which roseto very luminous peak as a Type IIn after 400 days (Miller et al. 2010). SN 1987Amight be thought of as an even more extended version of this, where the onset ofCSM interaction was delayed for 10 years as the SN ejecta caught up to CSM ejected ∼ yr earlier. We don’t know what fraction of normal SNe have a delayed onsetof CSM interaction, since most SNe are not monitored continuously with large tele-scopes for years or decades after they fade. Type IIn-P (SN 1994W, SN 2011ht, the Crab).
This is a distinct subclass ofSNe IIn that exhibit IIn spectra throughout their evolution, but have light curveswith a well-defined plateau drop (Mauerhan et al. 2013b). They are not to be con-fused with “transitional” SNe IIn/II-P that may have narrow lines at early times andevolve into an otherwise normal SN II-P. In SNe IIn-P, the narrow H lines withstrong narrow P Cyg profiles are seen for the duration of their bright phase. Thedrop from the plateau is quite extreme (several magnitudes) and the late-time lumi-nosities suggest low yields of Ni. These may be the result of ∼ erg electroncapture SN explosions with strong CSM interaction (Mauerhan et al. 2013b; Smith SN IIn impostors (LBVs, SN 2008S-like, etc.).
These transients have narrow Hlines in their spectra similar to SNe IIn, but are subluminous (fainter than M R ≃− Type Ia/IIn or Ia-CSM (SN 2002ic, SN 2005gj, PTF11kx).
These are transientsthat show spectral features indicative of an underlying SN Ia ejecta photosphere,but with strong superposed narrow H lines and additional continuum luminosity(Silverman et al. 2013). Cases with stronger CSM interaction tend to obscure theType Ia signatures, leading to ongoing controversy about their potential core col-lapse or thermonuclear nature (Benetti et al. 2006). Cases with weaker CSM inter-action (PTF11kx) reveal the Type Ia signatures more clearly, allowing them to beidentified unambiguously as thermonuclear events. The relatively rare evolutionarycircumstance that leads a thermonuclear SN to have a large mass of H-rich CSM isstill unclear.
Type Ibn (SN 2006jc).
These are similar to SNe IIn, except that H lines are weakor absent, so that narrow or intermediate-width He I lines dominate the spectrum(H a is weaker than He I l I lines have similar strength (Smith et al. 2012;Pastorello et al. 2008, 2015). It has been hypothesized that the likely progenitorsmay be Wolf-Rayet stars or LBVs in transition to a WR-like state (Pastorello et al.2007; Foley et al. 2007; Smith et al. 2012). Type Icn (hypothetical).
A discovery of this class has not yet been reported (as faras the author is aware), but as transient searches continue, there may be cases wherea SN interacts with a dense shell of CSM that is both H and He depleted, yieldingnarrow or intermediate-width lines of CNO elements, for example. With sufficientcreativity, one can imagine stellar evolution pathways that might create this; suchSNe are clearly rare, and if they never turn up, their absence will provide interestingphysical constraints for some binary models. nteracting Supernovae 15
The formation of dust by SNe may be essential to explain the amount of dust inferredin high redshift galaxies. SNe with strong CSM interaction provide an avenue fordust formation that is different from normal SNe, and possibly much more efficient.In normal SNe, dust forms in the freely expanding ejecta where there is a competi-tion between cooling to low enough temperatures while the ejecta are also rapidlyexpanding and achieving lower and lower densities. Even if dust forms efficiently,this ejecta dust might get destroyed when it crosses the reverse shock. In interactingSNe, by contrast, evidence suggests that dust can form very rapidly in the extremelydense, post-shock cooling layers (zones 2 and 3 in Figure 1). Moreover, this dust isalready behind the shock and may therefore stand a better chance of surviving andcontributing to the ISM dust budget.The first well-studied case of post shock dust formation was in SN 2006jc, whichwas a Type Ibn. The classic signs of dust formation were seen starting at only 50days post-dicovery, in an increase in the rate of fading, excess IR emission, andan increasing blueshifted asymmetry in emission-line profiles (Smith et al. 2008b).The last point strongly favored post-shock dust formation, since this was seen in theintermediate-width lines that were emitted from the post-shock zones. The Type Ibnevent may have been the result of a Type Ic explosion crashing into a He-rich shell.In this case, one might think that the C-rich ejecta were important in assisting theefficient formation of dust, as is seen for the post-shock dust formation in colliding-wind WC+O binaries (Gehrz & Hackwell 1974; Williams et al. 1990). However,similar evidence for this same mode of post-shock dust formation has also been seenin several Type IIn events, such as SN 2005ip, SN 1998S, SN 2006tf, SN 2010jl,and others (Smith et al. 2008a, 2009, 2012; Fox et al. 2009; Gall et al. 2014; Pozzoet al. 2004). It may therefore be the case that enhanced post shock dust formation isa common outcome of strong CSM interaction, where efficient post-shock coolingcauses the forward shock to collapse and become very dense.
Basic considerations about powering SNe IIn with CSM interaction, noted above,argue that extremely strong pre-SN mass loss is required. What sorts of progenitorstars can give rise to this dense CSM?
The first thing to do is to look around us in the nearby universe and ask what sortof observed classes of stars might fit the bill. Figure 3 makes this comparison byplotting the diversity in wind density of interacting SNe, deduced from the pre-
Fig. 3
Plot of mass-loss rate as a function of wind velocity, comparing values for interacting SNeto those of known types of stars. The solid colored regions correspond to values for various typesof evolved massive stars taken from the review about mass-loss by Smith (2014), corresponding toAGB and super-AGB stars, red supergiants (RSGs) and extreme RSGS (eRSG), yellow supergiants(YSG), yellow hypergiants (YHG), LBV winds and LBV giant eruptions, binary Roche-lobe over-flow (RLOF), luminous WN stars with hydrogen (WNH), and H-free WN and WC Wolf-Rayet(WR) stars. A few individual stars with well-determined very high mass-loss rates are shown withcircles (VY CMa, IRC+10420, h Car’s eruptions, and P Cyg’s eruption). Also shown with X’s aresome representative examples of SNe IIn (and one SN Ibn) that have observational estimates ofthe pre-shock CSM speed from the narrow emission component as measured in moderately highresolution spectra as well as estimates of the progenitor mass-loss required, taken from the litera-ture. The diagonal lines are wind density parameters ( w = ˙ M / V CSM ) of 5 × and 5 × g cm − ,which are typically the lowest wind densities required to make a SN IIn. Values are taken from theliterature; this figure is from a paper in prep. by the author. Note that in some cases, an observa-tionally derived value for the mass of the CSM has been converted to a mass-loss rate with a roughestimate of the time elapsed since ejection. shock wind (or shell) expansion speed and the inferred mass-loss rate, as comparedto expected mass-loss from known types of stars. The shaded and colored regionsin this figure show rough parameters for different classes of evolved massive starsthat are potential SN progenitors with strong winds. These are taken from the recentreview by Smith (2014); see the caption for definitions. In this plot, a given wind nteracting Supernovae 17 density parameter (required for a particular value of a CSM interaction luminosityvia Equation 1) has a diagonal line increasing to the upper right. The solid line showsa typical value for a moderate-luminosity SN IIn, and the dashed line is a typicallower bound for wind densities required to make a SN IIn (although an object canbe slightly below this line and still make a Type IIn spectrum if the wind is clumpyor asymmetric).We can see immediately that the normal classes of evolved stars with strong butrelatively steady winds (WR stars, LBV winds, normal YSGs and RSGs, AGB stars)do not match up to the wind density required for SNe IIn. These could potentiallyproduce SNe that have strong X-ray or radio emission from CSM interaction, butthey are not dense enough to slow the forward shock, and to cause the forwardshock to cool radiatively and collapse into the cool dense shell that is required forintermediate-width H a . The only observed classes of stars that have the high winddensities required are the giant eruptions of LBVs and the most extreme cool hyper-giants (the slower winds of extreme RSGs and YHG make their winds have densitycomparable to the shells in LBV giant eruptions, even though the mass-loss ratesare lower). Both classes of stars are not steady winds, but rather, they are dominatedby relatively short-lived phases of eruptive or episodic mass loss, or extreme denseand clumpy winds. In the case of LBVs, these are eruptions that last a few monthsto a decade. For extreme RSGs and YHGs, these are phases of enhanced mass lossthat may persist for centuries to a few thousand years.From the point of view of their mass loss, LBVs (and to a lesser extent the mostextreme cool hypergiants) are good candidates for the progenitors of SNe IIn. Thisdoes not, however, mean that these stars are necessarily all poised to explode in thenext couple years, nor does it mean that they are the only possible IIn progenitors.The caveat is that the progenitors of SNe IIn undergo strong eruptive or episodicmass loss immediately before core collapse, and may therefore experience signif-icant changes in their internal structure. The stars may have looked very differentin the time period before their pre-SN eruptions began. Hence, these stars in theirimmediate pre-SN state may not actually exist among nearby populations of stars inthe Milky Way and Magellanic Clouds right now.Also note that this is a mass-loss rate , not total mass, and note that these areusually lower limits to the required mass-loss rate (the high rate makes high op-tical depths, and hence, makes an observational determination difficult). In caseswhere we have estimates of the total mass ejected, one must divide by an assumedtimescale to derive a mass-loss rate; for SNe IIn, this timescale is inferred from therelative velocities of the CSM and CDS. Since we are essentially plotting a winddensity, it is important to recognize that two different SNe with the same inferredprogenitor mass-loss rate might have had that high mass loss lasting for very dif-ferent amounts of time, and hence, may have very different amounts of total massejected shortly before the SN. In some cases, the total mass can be more constraint-ing than the rate. For example, the total mass of ∼ M ⊙ required for some super-luminous SNe IIn rules out any progenitor stars with initial masses below about30-40 M ⊙ , because we need a comparable mass of SN ejecta to provide enough mo-mentum for a long-lasting interaction phase (not to mention mass loss due to winds throughout the life of the star). Also, there are cases like SN 2005ip where the in-ferred mass-loss rate is on the low end, but this SN had remarkably long-lastingCSM interaction that suggests a total mass of order 15 M ⊙ of H-rich material (e.g.,Smith et al. 2009; Stritzinger et al. 2012), which rules out lower-mass RSGs andsuper-AGB stars for its progenitor.Warning 3. One must be somewhat cautious in drawing conclusions about theprogenitor based on only the observed pre-shock CSM speed. We generallythink of the outflow speed being proportional to the star’s escape velocity, sowe expect slower outflows from RSGs and YHGs, somewhat faster speedsfor LBVs and BSGs (few 10 km s − ), and very fast outflows for WR stars(few 10 km s − ). This is a good guide if the outflow is a relatively steadyradiation-driven wind. However, remember that the observed CSM may bethe result of eruptive or explosive mass loss driven by a shock wave. If so, itis possible to get a relatively fast outflow even from a bloated RSG; eruptionspeeds need not match steady wind speeds for various types of stars in Figure3. Moreover, the high luminosity of the SN itself can potentially acceleratethe pre-shock CSM (essentially a radiatively driven wind from a temporarilymuch more luminous star), which could also lead to a faster pre-shock CSMspeed than expected for a certain type of star, or multiple velocity componentsin the CSM. The types of inferences described above are all still based on rather indirect and cir-cumstantial evidence. A potentially more direct way to link SNe to their progenitorsis to detect the progenitor star in pre-explosion imaging data, usually with the
Hub-ble Space Telescope , although ground-based imaging has been used for some verynearby cases (see review by Smartt 2009). This method has been used successfullyfor some other types of SNe, especially SNe II-P, IIb, and one particularly famousII-pec event that occurred 20 years ago. The progenitor identification can be con-firmed after the SN fades, to verify that the progenitor star is gone (as opposed tobeing a chance alignment, a host cluster, or a companion star).For interacting SNe, however, the interpretation of pre-SN direct detections canbe a little tricky. First, the “direct” detection of the progenitor might actually be adirect detection of a pre-SN eruption and not the quiescent star. This source mightindeed fade after the SN, but it is hard to tell if the star is really dead, or if theeruption has just subsided and the star returned to its quiescent state. Inferring aninitial mass by comparison with stellar evolution tracks is also complicated if theprogenitor might be in an outburst rather than in its quiescent state (we don’t havemuch choice here and are lucky if there is even one
HST image in the archive,but one just needs to be aware of the caveat). A second complication is that some nteracting Supernovae 19
SNe IIn have persistent CSM interaction for years or decades after the SN, and soone might need to wait a very long time before the SN has faded enough to befainter than the progenitor. Third, a faint progenitor or even a faint upper limit to aprogenitor is very inconclusive in terms of its implication for the mass of the star.This is because with interacting SNe, there is, by definition, a large mass of CSM.Thus, we certainly expect some cases where the progenitor was surrounded by amassive dust shell that should obscure the progenitor star’s visible light output. Thismakes it difficult to place a progenitor detection (or upper limit) on an HR Diagramand infer an initial mass if one has only an optical filter.Despite this ambiguity, there have been a few lucky and important cases thatguide our intuition about the progenitors of SNe IIn and SNe Ibn.
SN 2005gl:
This was a SN IIn where a very luminous progenitor consistent withan LBV star like P Cygni was detected in
HST imaging and the SN had an impliedmass-loss rate of 0.01 M ⊙ yr − (Gal-Yam et al. 2007), and this source then fadedafter the SN (Gal-Yam & Leonard 2009). Moreover, the pre-shock CSM speed of420 km s − inferred from narrow H lines was suggestive of the fast outflow froman LBV (Smith et al. 2010). As noted above, there is some ambiguity as to whetherthe pre-SN detection was a massive quiescent LBV-like star or a pre-SN eruptioncaught by the HST image. In either case an eruptive LBV-like star is likely, althoughthe implied initial mass may be different. Based on its spectral evolution, SN 2005glfits into the subclass of “transitional” SNe IIn like SN 1998S and PTF11iqb, whichshow a Type IIn spectrum at first, but later show broad P Cygni lines indicative ofa SN photosphere, so this can be taken as an argument that it was most likely acore-collapse SN event.
SN 1961V:
Long considered an extreme LBV or SN impostor event, recent ar-guments favor an actual core-collapse SN for the 1961 transient (Kochanek et al.2011; Smith et al. 2011). If this was a SN, then it has one of the best documentedprogenitor detections and progenitor variability among SNe, and it holds the recordfor the most massive directly detected SN progenitor. The pre-1961 variability sug-gests a very massive ∼ M ⊙ blue LBV-like progenitor that was variable beforethe SN. The source at the SN position is now more than 5.5 mag fainter than thisprogenitor (a much more dramatic drop than in the case of SN 2005gl), and there isno IR source with a luminosity comparable to the progenitor (Kochanek et al. 2011),so the extremely massive star is likely dead. SN 2006jc:
An eruption in 2004 was noted as a possible LBV or SN impostor,and then a SN occurred at the same position 2 years later. The pre-SN outbursthad a peak luminosity similar to SN impostors (Pastorello et al. 2007). The SNexplosion two years later was of Type Ibn with strong narrow He I emission lines(Foley et al. 2007, Pastorello et al. 2007). There is no detection of the quiescentprogenitor, but this unusual case implies an LBV-like eruption that occurred in aWR-like progenitor star that was clearly H depleted. The CSM speed was of order1000 km s − , which is consistent with WR stars, and faster than typical LBVs. SN2011ht:
This belongs to the subclass of SNe IIn-P, which are thought to arisefrom lower-energy explosions that may be linked to electron capture SNe in 8-10 M ⊙ super-AGB stars (Mauerhan et al. 2013b; Smith 2013). There is no detection of the quiescent progenitor, although deep upper limits seem to rule out a luminous,blue quiescent star (Mauerhan et al. 2013b; Roming et al. 2012). However, Fraseret al. (2015) reported the detection of pre-SN eruption activity in archival data. Thismay be an important demonstration that non-LBVs can have violent pre-SN erup-tions as well. SN 2010jl:
This was a SLSN IIn with roughly 10 M ⊙ or more of CSM. Smithet al. (2011c) identified a source at the location of the SN in pre-explosion HSTimages that suggested either an extremely massive progenitor star or a very youngmassive-star cluster; in either case it seems likely that the progenitor had an initialmass above 30 M ⊙ . We are still waiting for this SN to fade to see if the progenitorcandidate was actually the progenitor or a massive young cluster/association. SN 2009ip:
This source was initially discovered as an LBV-like outburst in 2009(its namesake) before finally exploding as a much brighter SN in 2012. A quiescentprogenitor star was detected in archival HST data, indicating a very massive 50-80 M ⊙ progenitor (Smith et al. 2010b, Foley et al. 2011). In this case, the HST detectionmay well have been the quiescent progenitor, rather than an outburst, because muchbrighter outbursts came later. It showed slow variability consistent with an S DorLBV-like episode (Smith et al. 2010b), followed by a series of brief LBV-like gianteruptions (Smith et al. 2010b, Mauerhan et al. 2013a, Pastorello et al. 2013). Unlikeany other object, we also have detailed, high-quality spectra of the pre-SN eruptions(Smith et al. 2010b, Foley et al. 2011). The presumably final SN explosion of SN2009ip in 2012 would fall into the “delayed onset” subclass, since at first the faintertransient showed very broad lines indicative of a SN photosphere. Reaching peak,however, it looked like a normal SN IIn, as the fast ejecta crashed into the slow ma-terial ejected 1-3 years earlier (Mauerhan et al. 2013a, Smith et al. 2014). A numberof detailed studies of the bright 2012 transient have now been published, althoughthere has been some controversy about whether the 2012 event was a core-collapseSN (Mauerhan et al. 2013a; Ofek et al. 2013; Prieto et al. 2013; Smith et al. 2014)or some type of extremely bright nonterminal event (Fraser et al. 2013a, Pastorelloet al. 2013, Margutti et al. 2014). More recently, Smith et al. (2014) showed that theobject continues to fade, and its late-time emission is consistent with late-time CSMinteraction in normal SNe IIn. If SN 2009ip was indeed a SN, it provides a strongcase that very massive stars above 30 M ⊙ may in fact experience core collapse andexplode, and that LBV-like stars are linked to some SNe IIn. We will undoubtedly find more examples of direct detections and pre-SN outburstsin the future. One must bear in mind, though, that LBVs and eruptive precursorsare relatively easy to detect because they are brighter than any quiescent stars, sothese cases do not rule out alternatives such as dust-enshrouded RSGs, or faint andhot quiescent stars. From various clues described above such as the CSM mass andmass-loss rate, CSM expansion speed, H-rich composition, and direct detections of nteracting Supernovae 21 progenitors or environments, we can attempt to link certain subclasses of interactingSNe to different possible progenitors. Some associations are more likely than others.
SLSN IIn (compact and extended shell; SN 2006gy or SN 2010jl-like):
Based onthe extreme required masses of (10-20 M ⊙ ) of H-rich CSM, typically expanding at afew hundred km s − , it seems very likely that the progenitors of SLSNe IIn are verymassive, eruptive LBV-like stars (see review by Smith 2014). If they are not trulyLBVs, then they do a very good impersonation. The simple fact that very massivestars above ∼ M ⊙ are exploding as H-rich SNe is a challenge to understand, sincestellar evolution models predict all those stars to shed their H envelopes at roughlysolar metallicity. Most of these have huge mass ejections occurring just a few yearsor decades before the SN, so the connection to nearby LBVs — some of whichhave shells that are hundreds or thousands of years old – is not yet clear. LBV-likeprogenitors are even likely in cases where the progenitor is optically faint, if a pre-SN eruption has obscured the star with a dust cocoon (as is likely to be the case,given the consequent CSM interaction). Since the progenitors must be very massivestars simply because of the mass budget, the physical mechanism of pulsationalpair instability eruptions is a viable candidate for the pre-SN outbursts (Woosley etal. 2007). This is not the case for the remainder of SNe IIn, because they are toocommon (Smith et al. 2014). Normal, enduring SNe IIn (SN1988Z-like):
Likely progenitor types are LBVs orextreme cool hypergiants (YHGs/eRSGs), based mostly on the required mass-lossrates and observed CSM speeds. For these enduring cases, we require either strongmass loss in several centuries preceding the SN, or a large bipolar shell ejectedshortly before the SN with a range of ejection speeds (to accommodate CSM in-teraction over a large range of radii, lasting for a long time). Total CSM massesthat exceed 10 M ⊙ in some cases and integrated radiated energies that exceed 10 ergs (over years) point to relatively massive progenitor stars. Enhanced late-phaseRSG mass loss (Yoon & Cantiello 2010) or instabilities in late nuclear burning se-quences (Quataert & Shiode 2012; Smith & Arnett 2014) are good candidates forthe episodic mass-loss. Transitional or fleeting SNe IIn (SN 1998S-like):
Likely progenitors are RSGs,YHGs, or BSG/LBVs with less extreme pre-SN bursts of mass loss, confined toa relatively short-duration event preceding core collapse (i.e. within a few yearspreceding the SN). These objects also seem to require highly asymmetric CSM in-teraction, to allow the expanding SN photosphere to completely or mostly envelopethe disk of CSM interaction that emits narrow lines (as discussed in Section 3). Forthis reason, there is a strong suspicion that close binary interaction plays a role intheir pre-SN episodic mass loss, although it must still be linked somehow to the fi-nal nuclear burning sequences (see Smith & Arnett 2014 regarding the role of closebinaries in this scenario).
SNe IIn-P (SN 2011ht-like):
The favored scenario for these events involves anintermediate-mass progenitor in a super-AGB phase that suffers a sub-energetic(10 erg) electron-capture SN. This is based on the low Ni yield, the deep ab-sorption features that imply more-or-less spherical symmetry, and the energy/massbudgets inferred (Mauerhan et al. 2013b; Smith 2013). However, observations can- not yet confidently rule out a more massive progenitor that suffers fallback to ablack hole, yielding a smaller ejecta mass and very low Ni yield (although in thecase of SN 2011ht, progenitor upper limits seem to argue against this interpreta-tion). This type of SN also fits the bill for SN 1054 and the Crab nebula, whoseabundances and kinetic energy seem to point to an electron capture SN from anintermediate-mass star (Nomoto et al. 1982). If these are ecSNe, then the pre-SNepisodic mass-loss might be related to nuclear flashes in the advanced degeneratecore buring sequences.
SN IIn impostors : This group of putatively non-SN transients may be quite di-verse, and it may include transients that have narrow lines because they are pow-ered largely by CSM interaction (and weaker explosions than core-collapse SNe)and other transients that have narrow lines because they have slow winds/outflows.This may include LBVs, super AGB stars, binaries with a compact object, stellarmergers, pre-SN nuclear burning instabilities, failed SNe, or all of the above. Anymassive supergiant star enshrouded in dust or with strong binary interaction that suf-fers an instability is a viable candidate. The theoretical mechanisms for this class ofoutbursts is still very poorly understood, and there is probably considerable overlapwith pre-SN outbursts that lead to SNe IIn and Ibn.
SNe Ibn:
The most likely progenitors are massive WR stars that for unknownreasons undergo pre-SN outbursts. This is interesting, because there is no knownprecedent for an observed eruption in a H-deficient massive star.
SNe Ia/IIn or Ia-CSM:
If these really are thermonuclear SNe Ia (some casesseem clear, others are still debated in the literature), then the exploding progenitorsare white dwarfs that have arisen from initial masses below roughly 8 M ⊙ . In thiscase, a single degenerate system is clearly required to supply the large mass of H-rich CSM (several M ⊙ in some cases). There is, as yet, no viable explanation forthe sudden ejection of a large mass of H (by the companion) shortly preceding thethermonuclear explosion of the WD. Observations of interacting SNe present one of the most interesting challenges toour understanding of the end phases of evolution for massive stars. What makesthese stars explode before they explode? Computational resources are only begin-ning to meet the needs of the complex problem of simulating convection and nuclearburning coupled with stellar structure and turbulence in these final phases (Arnett &Meakin 2011; Meakin & Arnett 2007). The fact that ∼
10% of core-collapse eventsare preceded by some major reorganization of the stellar structure and energy bud-get tells us that we have been missing something important, which might be a keyingredient for understanding core collapse SNe more generally. It will be importantto try and understand if this 10% corresponds to the most extreme manifestation of awider instability (for example, if all stars undergo some instability in the final burn-ing sequences, but only the most extreme cases lead to detectably violent mass loss nteracting Supernovae 23 and SNe IIn) or if SNe IIn are the outcome of special circumstances in a particularevolution channel (i.e. interacting binaries within a certain mass and orbital periodrange). There is still very little information available on any trends with metallicity,although this is always good to investigate when mass loss plays a critical role.Recent years have seen something of a paradigm shift in massive star studies.Formerly standard or straightforward assumptions about the simplicity of single-starevolution are giving way to more complicated scenarios as astronomers grudginglyadmit that binary stars not only exist, but are common and influential (e.g., Sana etal. 2012). This may be especially true among transient sources and stellar deaths thatseem otherwise peculiar or difficult to understand. Given the very high multiplicityfraction among massive stars, binary interaction should probably not be considereda last resort or the refuge for an uncreative theorist, but rather, a default assumption.Whether they are binaries or not, all SNe IIn require some major shift in stel-lar structure and mass loss before the SN. The synchronization with core collapsegives a strong implication that something wild is happening in the latest sequencesof nuclear burning. Those cases where a star changes its structure as a result of thesenuclear burning instabilities (i.e. an inflated envelope) in a close binary system seempromising for inducing sudden eruptive behavior for various reasons (Smith & Ar-nett 2014). Single stars may also be able to induce their own violent mass loss in thecouple years preceding core collapse due to energy transported to the envelope fromNe and O burning zones (Quataert & Shiode 2012). However, we do not yet have agood explanation in single-star evolution for the strong mass-loss that leads to the“enduring” class of SNe IIn or some of the more extreme cases of “delayed onset”of CSM interaction, where the strong interaction that lasts for years after explosionsuggests very strong mass loss for decades or centuries before core collapse. If bi-nary interaction is an important ingredient, then this sort of interaction before a SNmight make asymmetric CSM very common, suggesting that we should be payingclose attention to possible observed signatures of asymmetry.
Acknowledgements
My attempt to understand interacting SNe and their connections to massivestars has benefitted greatly from conversations with numerous people, but especially Dave Arnett,Matteo Cantiello, Nick Chugai, Selma de Mink, Ori Fox, Morgan Fraser, Dan Kasen, Jon Mauer-han, Stan Owocki, Jose Prieto, Eliot Quataert, Jorick Vink, and Stan Woosley. While drafting thischapter, I received support from NSF grants AST-1312221 and AST-1515559.
Cross-References
1. Supernova of 1054 and its Remnant, the Crab Nebula2. Observational and Physical Classification of Supernovae3. Supernovae from super AGB Stars (8-12 Msun)4. Supernovae from massive Stars (12-100 Msun)5. Very Massive and Supermassive Stars: Evolution and Fate6. Progenitor of SN 1987A7. Supernova Progenitors Observed with HST8. Interacting Supernovae and the Influence on Spectra and Light Curves9. X-ray and radio emission from circumstellar interaction
10. Dust and Molecular Formation in Supernovae11. Supernova remnant from SN1987A12. The Supernova - Supernova Remnant Connection