arXiv:1711.05180v2 [astro-ph.HE] 19 Dec 2017 0Dcme 2017 December 20 hthsafs uflwvlct of (CSM) velocity matter peaks. outflow circumstellar five fast a least a in at is has has typical gas curve that of absorbing light that The The than light (2) (3) slower the SNe. along II-P of evolution type order The an (1) is three. curve here list fro we properties, which unusual several has 2016bse; iPTF14hls (AT Gaia16aog). iPTF14hls (SN) II-P type enigmatic MNRAS † 2016 al. et Goranskij al. et Arcavi non-spherical be the to the of observed Some with (e.g., radiation. are collide to outbursts to energy thought pre- kinetic convert is and ejecta CSM supernova the cases 2017 2016 deduced was (e.g., outbursts SNe, 2013 pre-explosion other in for explosion fore of energy kinetic estimated an with and ⋆ ( omSoker Noam jets envelope common supernova a as iPTF14hls Explaining INTRODUCTION 1 c 2 1 raie l 2017 al. et Arcavi otc e-mail: Contact otc e-mail: Contact nttt fAtooy nvriyo abig,Madingle Cambridge, of University Astronomy, of Institute eateto hsc,Tcno salIsiueo Techn of Institute Israel – Technion Physics, of Department 07TeAuthors The 2017 jcino astn fyast e asbe- days few to years of tens mass of Ejection ; ; ; elye l 2017 al. et Reilly fke l 2013 al. et Ofek oa rh2017 Groh & Boian ao ta.2017 al. et Yaron 000 , ( 1 2017 – 5 [email protected] [email protected] 21)Pern 0Dcme 07Cmie sn NA L MNRAS using Compiled 2017 December 20 Preprint (2017) eottedsoeyadeouino the of evolution and discovery the report ) 1 ). ⋆ ; vsa Gilkis Avishai , ; agtie l 2017 al. et Margutti vrk aa 2014 Nakar & Svirski o N20i) ncnrs othese to contrast In 2009ip). SN for oe ta.2007 al. et Foley ; atrloe l 2018 al. et Pastorello ; i ta.2017 al. et Liu epooea propose We ABSTRACT icmtla a n h xlso,tenurnsa hudaccre should kinetic the for neutron account the To explosion, itself. the explosion 0 and of the ejection later gas the weeks power circumstellar of few that envelope jets the a launches inside and and spirals-in mass that accretes star star neutron a where iPTF14hls asaceinb h eto tra eisrnpsaepirt prior passage words: periastron orb Key at eccentric star an phase. neutron to envelope the pow common outburst by further pre-explosion photo and accretion supernova 1954 gas mass the the fallback and where accretes attribute core medium star We the the neutron eject and remaining neutron that itself the The jets interaction explosion the launches the of and form hours mass, we last several the accretes to In core, one star. about giant for the orbits of star neutron the while ejected . 3 M v CSM ⊙ h tens The . ; E ; ≈ aehne al. et Mauerhan CSM ; atgi tal. et Tartaglia ; 2 yome al. et Nyholm 00k s km 6000 † oia2015 Moriya .I these In ). omnevlp essupernova jets envelope common tr:jt uenve eea iais close binaries: — general supernovae: — jets : ≈ × 10 M 52 ⊙ erg − m ie abig,C30A UK 0HA, CB3 Cambridge, Rise, y 1 fcruselrgsta consfrsm bopinlnsis lines absorption some for accounts that gas circumstellar of lg,Hia3003 Israel 3200003, Haifa ology, ; , eshsbe icse eoe(e.g., launches and before mass discussed accretes star in- been giant or 1998 envelope a has the of jets inside core orbits the that side companion a where 2013 2017 to close even energy of 10 amount an carry cannot mechanisms 2014 rmtevgru ovcini h oeadi eoie in deposited is and ( core envelope the the in wave convection by vigorous carried the is from that energy involves outbursts explosion col- ejecta-CSM for evidence no is lision. there iPFT14hls in cases, aso ahrta aseeto (e.g., ejection mass than rather pansion oet h neoe( envelope the to core nbet xe uhms,w dp h iwta a mass (e.g., the that outburst eject view to the energy power the the and of and envelope adopt most inflated supplies we the jets launching from mass, be mass much to accreteing companion expel seems to outbursts unable pre-explosion observed commonly II-P type a is enve which hydrogen-rich iPTF14hls entire to SN. the applicable not eject is will and core lope, the to close or 52 n fteepoe ehnsst rge hs pre- these trigger to mechanisms employed the of One ic h eoiino nryi h neoeo the of envelope the in energy of deposition the Since ; ; deposits that mechanism other Any ). the from energy carry also might activity Magnetic ). r,adte r ieyt as anyevlp ex- envelope mainly cause to likely are they and erg, riae&Lvo2000 Livio & Armitage ce oe 2014 Soker & Mcley utet&Sid 2012 Shiode & Quataert cnrofrteeimtcsupernova enigmatic the for scenario oe iks2017 Gilkis & Soker .Tecmo neoeprocess envelope common The ). ; oe 2004 Soker ah oe 2010 Soker & Kashi k nieteenvelope the inside eks h icmtla shell circumstellar the h ne neoeto envelope inner the trmre ihthe with merges star ; h ne fthe of onset the o r h supernova. the ers tadtemporary and it hoe&Quataert & Shiode oe 2013 Soker ; A .Hwvr these However, ). eams of mass a te re Woosley & Fryer ≈ T asv giant massive a hvle 2012 Chevalier peeresides. sphere E nryo the of energy 10 tl l v3.0 file style X 52 r inside erg ; ; Fuller Soker ≈ ). s - 2 N. Soker, A. Gilkis

Fryer & Woosley (1998), for example, propose a gamma ray burst model where a that spirals-in all the way to 0.5 the helium core of a giant accretes mass and launches jets. We propose a scenario to account for the enigmatic type II-P SN iPTF14hls and its pre-explosion mass ejection that 0 is based on the jet feedback mechanism (for review see Soker 2016) that can both facilitate the removal of a common en- -0.5 velope (e.g., Armitage & Livio 2000; Soker 2004; Chevalier 2012; Shiber & Soker 2017) and power the explosion of most (or all) core collapse supernovae (e.g., Papish & Soker 2011). -1 In section 2 we discuss the jet-driven envelope removal of part of the envelope, and in section 3 we discuss the much -1.5 faster removal of the core by jets, i.e., the explosion. We log10 Mlost/M⊙ summarize in section 4. log r/100R⊙ -2 10 0 1 2 3 2 PRE-EXPLOSION ENVELOPE EJECTION t [Myr]

Arcavi et al. (2017) note that the evolution of some spec- tral absorption lines is best explained by coming from pre- Figure 1. Total mass lost and radius of the star as function of explosion ejected gas. However, a velocity of vpre ≃ 4000 − time, for a star with an initial mass of MZAMS = 80M⊙. 8000 km s−1 is not typical for winds. A much more violent event must take place. We take it to be the launching of jets by a companion. with a reduction factor of 0.33 (for a review on mass loss see Arcavi et al. (2017) constrain the pre-explosion lumi- Smith 2014). For the low we take, this results in ≈ × 41 nosity to be below 2 10 erg. This upper bound on a small mass lost upon reaching the giant phase which in- the luminosity and the kinetic energy of the pre-explosion terests us, Mlost ≈ 1M⊙. In Fig. 1 we present the evolution ≈ 52 ejecta of ECSM 10 erg (Arcavi et al. 2017), constrain with time of the stellar radius and total mass lost by winds, the efficiency of converting energy to optical radiation to and in Fig. 2 we present the stellar model when the radius −3 . CE CE be 10 (τ /1 yr), where τ is the duration of the pre- is R1 = 100R⊙. explosion phase. This is a very low effi- To have a rapid spiraling-in process already from the ciency. The ejection of tens of solar masses at a velocity of surface, the binary system should be unstable to the Darwin −1 thousands of km s should be like a supernova explosion instability. For the model presented in Fig. 2, the moment and lead to a detection if observed. 2 of inertia is I1 = ηMenvR1 with η = 0.0072, where Menv = Our proposed solution is to consider a very short pre- 46.5M⊙ is the envelope mass. For an earlier time when the explosion common envelope phase, during the time the SN radius is 50R⊙ we find for the same model η = 0.0137. was not observed and the upper bound on the luminosity We note that ours is just one model, and a more massive does not apply. Before entering the envelope and accreting model might have a more massive envelope at the relevant at a very high rate that enables neutrino cooling, the neutron stage. The condition of Darwin instability for a circular orbit star accreted at its Eddington limit, which is much below the (later we will argue that the orbit was actually somewhat detection limit. We propose that the entered eccentric) reads the envelope within less than 100 days before discovery. The −2 neutron star cannot bring the envelope to synchronization, 3I1 ≃ η Menv aD M2 < 2 1.5 M⊙. (2) and very rapidly, within a few dynamical times or one Keple- aD  0.01   50M⊙ R1  rian orbital time (e.g., Passy et al. 2012; Ivanova & Nandez This shows that we require the binary companion to be a 2016), spirals-in all the way to the core. The spiral-in time neutron star, rather than a black hole. is then The demand for a very low mass companion comes also −1/2 3/2 from the finding that the polarization of iPTF14hls is very M R1 τCE ≈ τKep ≃ 15 day, (1) low, implying an almost spherical explosion (Arcavi et al.  60M⊙   100R⊙  2017). Namely, the companion cannot deform much the gi- where M is the total mass of the binary system, and R1 is ant envelope as it enters the common envelope phase. the radius of the . As the spiraling in might be The binding energy of the hydrogen envelope in the 50 several times the Keplerian time on the surface, we consider model presented in Fig. 2 is Ebind = 5.8 × 10 erg, much the radius of R1 = 100 at the onset of the common envelope below the estimated kinetic energy of the CSM of iPTF14hls 52 phase to be an upper limit. Namely, we expect R1 ≃ 50 − (ECSM ≈ 10 erg). The presence of hydrogen in the enve- 100R⊙ at the onset of the common envelope phase. lope at explosion time implies that the pre-explosion energy To obtain some representative values we evolve a non- deposition cannot be below the base of the hydrogen-rich rotating stellar model with a zero age mass mass. Namely, it cannot come from the core. A secondary of MZAMS = 80M⊙ and metallicity of Z = 0.001. We use star that launches jets inside the envelope can deposit en- the code Modules for Experiments in Stel- ergy inside the envelope. With an efficiency of ǫ ≃ 0.02 of 2 lar Astrophysics (MESA version 10108; Paxton et al. 2011, converting rest energy of the accreted gas Maccc (say a 2013, 2015). Mass loss is according to Vink et al. (2001), fraction of 0.1 of the accreted mass is launched at terminal

MNRAS 000, 1–5 (2017) iPTF14hls as a common envelope jets supernova 3

velocity of vjet = 0.55c), the accreted mass should be 0 10 −1 H ≃ ECSM ǫ Macc 0.3 52 M⊙. (3) He  10 erg   0.02  -1 C The neutron star can stay a neutron star after an accretion 10 N of such an amount of mass. Within 15 days the accretion rate −1 O will be ≈ 10M⊙ yr . Neutrino cooling of the accreted mass allows such a high accretion rate (e.g., Houck & Chevalier 10 -2 1991). In the jet feedback mechanism the relation Ebind ≪ ECSM implies that the feedback efficiency is low. The rea- son is that during the common envelope phase the jets eject 10 -3 envelope gas above and below the location of the compact companion that launches the jets, while the companion ac- cretes mass mainly from envelope zones inner to its orbits. -2 -1 0 1 2 Namely, the jets do not directly remove gas from the reser- log r/R⊙ voir of the accreted gas. This makes the negative feedback cycle inefficient, and explains why the total energy in the jets 80 was much larger than the binding energy of the envelope. We note that the gravitational energy that was released by the neutron star during its spiral-in to an orbital separation of 50 ≃ 1R⊙ is only ≈ 10 erg and does not add much to the 60 energy of the ejected envelope. For the same reason the di- rect binary interaction does not cause much deviation from spherical symmetry. 40 M/M⊙ We adopt the jet-driven envelope ejection over the alter- native scenario of the pair-instability pulsations mechanism. Rakavy & Shaviv (1967) were the first to find that very mas- 20 sive stars are expected to suffer pair instability that results from pressure drop due to electron-positron pair production. A later paper by Barkat et al. 1967 studied the explosion it- 0 self (although it was submitted after the paper by Rakavy -2 -1 0 1 2 & Shaviv, it appeared first, and even today misleads re- searchers on the truly first paper on stellar pair instability). log r/R⊙ Arcavi et al. (2017) consider the possibility that pair insta- bility pulsations lead to the pre-explosion outbursts, but dis- 2 miss it for two reasons. (1) The first pair instability pulsa- tion, that takes place in the , is expected to eject the entire hydrogen-rich envelope. (2) The pair instability 52 0 is not expected to supply the energy of ECSM ≈ 10 erg. We note that our examined 80M⊙ stellar model might be- -2 come pair-unstable, but only after expansion and formation of an oxygen core, while our proposed scenario occurs earlier (during expansion) and prevents a pair-instability explosion. -4

-6 3 THE EXPLOSION log10 ρ 51 log10 Ebind/10 erg There are two possible explanations for the final explosion of -8 iPTF14hls. In the first the core of the giant star experienced -2 -1 0 1 2 a regular core collapse supernova. However, this cannot work log r/R⊙ with our proposed pre-exploison outburst scenario. If the companion is to spiral-in within less than 100 days (con- straint by the non-detection of pre-explosion outburst 100 Figure 2. Top: Mass fraction of the predominant elements of a days before discovery), the giant could not be much larger M M stellar model with an initial mass of ZAMS = 80 ⊙ at the time than about 200R⊙. But a star with a rich hydrogen envelope its radius is 100R⊙. Middle: Enclosed mass as function of radius. will reach larger radii before it explodes (see figure 1). As R R Bottom: Density profile at = 100 ⊙, as well as the binding well, the slow evolution of the supernova suggests that it is energy at each mass coordinate, i.e., the amount of additional not a regular core-collapse supernova. energy needed to unbind matter from the surface down to that We instead follow Chevalier (2012) and Papish et al. point. (2015b) and attribute the explosion to the rapid accre- tion of core material onto the neutron star companion

MNRAS 000, 1–5 (2017) 4 N. Soker, A. Gilkis as the latter merges with the core. For that, Chevalier vation time the resides within the mass that (1996) and then Papish et al. (2015b) in a stronger way con- was ejected in this last phase of neutron star interaction sider impossible the formation of a Thorne–Zytkow object with the dense core. (TZO), where a neutron star settles at the center of the We term this entire explosion a common envelope jets star (Thorne & Zytkow 1975). Overall, the final explosion is supernova. powered by a neutron star that accretes mass at a high rate In the premise where all core collapse supernovae are from the core and launches jets that power the explosion. driven by jets (e.g. Papish et al. 2015a), common envelope In the jet feedback explosion mechanism of most core- jets supernovae should qualitatively be similar to core col- collapse supernovae the explosion energy is about several lapse supernovae, and in the present case a type II-P super- times the binding energy of the core (e.g., Papish & Soker . 2011; review by Soker 2016). Simply, this is the energy that is required for the jittering jets (Papish & Soker 2014) to expel the core and terminate accretion onto the newly born neutron star. When the jets are well collimated they are less 4 SUMMARY efficient in removing the core material from the equatorial plane. The accretion process continues and the jets carry We proposed that both the pre-explosion mass ejection and much more energy than the binding energy of the core. This the explosion of the type II-P SN iPTF14hls were powered inefficient jet feedback process leads to a super-energetic by a neutron star that accreted mass while spiraling-in in- (and/or super-luminous) supernova (Gilkis et al. 2016). side the envelope and the core of a giant massive star. We As discussed in section 2, when the jets are launched argue that this common envelope jets supernova scenario by a neutron star that accretes mass from the envelope the can account for the basic properties of iPTF14hls. feedback is inefficient. While the jets are launched perpen- The process operates by a negative feedback mecha- dicular to the equatorial plane and remove envelope mass nism, where the jets remove mass from the reservoir from from those regions, the mass supply comes from inner enve- which mass is accreted to power them. However, because the lope zones and through the equatorial plane. This inefficient jets are launched perpendicular to the orbital plane while the negative feedback mechanism implies that in total the jets accretion flow onto the neutron star is from inner zones of carry much more energy than the binding energy of the en- the giant and near the equatorial plane, the feedback cycle velope they remove. As discussed by Papish et al. (2015b), is inefficient. This explains the kinetic energy of the outer the core is destructed and forms an around gas that is responsible for some absorption lines being much the neutron star. The disk launches jets that further ener- larger than the binding energy of the envelope. gize the expanding gas. Fallback material can prolong even When the core is destructed by the neutron star, the more the accretion process and augment the explosion en- process is somewhat different as the core is expected to form ergy. Arcavi et al. (2017) find that at very late times in their an accretion disk around the neutron star, and the explosion observations the declines as t−5/3, supporting a energy is about the binding energy of the core. The explosion powering by fallback material. Furthermore, instabilities in takes place when the neutron star is interacting with the the accretion flow and interaction of jets with earlier ejecta helium rich core. However, the inner regions of the hydrogen- might account for the light-curve variability and multiple rich envelope are still with-in hundreds of solar radii, and the peaks. explosion shock catches up with those layer and makes the An alternative explanation for the peaks is the colli- explosion a type II-P SN. sion of the ejecta with previously ejected mass. Arcavi et al. We attribute the outburst of 1954, which is essentially 2017 dismiss collision with circumstellar matter because an intermediate luminosity optical transient (ILOT), to an they see no narrow lines. However, in our model the previ- eccentic orbit of the neutron star before it entered the enve- ously ejected gas moves at thousands of km s−1, and we do lope. The event of high mass accretion rate at periastron pas- not expect to see narrow lines. This surrounding gas moves sage, qualitatively similar to the binary model of the Great only slightly slower that the photosphere, also at thousands Eruption in (Kashi & Soker 2010), could have of km s−1. The collision occurs inside the broad-line form- powered this ILOT event. ing region and hence does not dilute the lines. An earlier ejection of mass can lead to the interac- The entire process from the onset of the common en- tion of the SN ejecta with the earlier ejected mass, the velope phase to the final explosion is a continuous one. But CSM. Andrews & Smith (2017) find evidence for an inter- there is a qualitative difference as the neutron star reaches action with CSM expanding at ≈ 1000 km s−1 at late times the core because the time scales is much shorter now and and suggest that iPTF2014hls can be a regular CCSN in- the density of the core is much larger (see density profile in teracting with the CSM. We see two problems with this Fig. 2). The spiraling-in process inside the envelope lasts for scenario. (1) They do not account for the massive and weeks. The outer parts that were ejected first reach a dis- fast (≈ 6000 km s−1) absorbing material that Arcavi et al. tance of ≈ 1015 cm while the inner envelope region is ejected. (2017) find. (2) Their scenario involves collision deep inside The final spiraling-in process, from a radius of ≈ 5R⊙ to the the ejecta. This has a low radiation efficiency, because part core, lasts for only about a day. The hydrogen-rich envelope of the collision energy goes to accelerating the slowly ex- that is ejected from this volume does not reach a large dis- panding earlier ejecta, and because the long diffusion time tance before the neutron star reaches the core, accretes mass of the X-ray photons out (otherwise X-ray emission will be within several hours and launches energetic jets that drive strong). So even in their scenario the explosion energy is shock waves through the expanding dense envelope. This very large. We do agree with them that ejecta-CSM inter- results in a supernova explosion. During most of the obser- action does take place and does contribute, in particular at

MNRAS 000, 1–5 (2017) iPTF14hls as a common envelope jets supernova 5 late times, but our aim is to account for the massive and Mcley L., Soker N., 2014, MNRAS, 445, 2492 fast absorbing gas that Arcavi et al. (2017) claim for. Moriya T. J., 2015, ApJ, 803, L26 One challenge to the present model is the finding of Nyholm A. et al., 2017, A&A, 605, A6 Arcavi et al. (2017) that the polarization is very low, im- Ofek E. O. et al., 2013, Nature, 494, 65 plying a spherical explosion. We account for this as follows. Papish O., Soker N., 2011, MNRAS, 416, 1697. Papish O., Soker N., 2014, MNRAS, 438, 1027 The low mass companion, with a mass of M2 ≃ 0.02M1, im- Papish O., Nordhaus J., Soker N., 2015a, MNRAS, 448, 2362 plies that it cannot deform much the envelope of the giant Papish O., Soker N., Bukay I., 2015b, MNRAS, 449, 288 star. The jets were launched deep in the envelope and did Passy J.-C. et al., 2012, ApJ, 744, 52 not break out due to the fast orbital motion (Papish et al. Pastorello A. et al., 2018, MNRAS, 474, 197 2015b). Their energy deposition expelled the envelope and Paxton B. et al., 2011, ApJS, 192, 3 then the core to all directions. Paxton B. et al., 2013, ApJS, 208, 4 The evolution of the progenitor binary system leading Paxton B. et al., 2015, ApJS, 220, 15 to such a common envelope jets supernova event is also of in- Quataert E., Shiode J., 2012, MNRAS, 423, L92 terest. One possible channel is through immense mass trans- Rakavy G., Shaviv G., 1967, ApJ, 148, 803 Reilly E., Maund J. R., Baade D., Wheeler J. C., H¨oflich P., fer from a primary of MZAMS ≈ 50M⊙ onto a secondary of Spyromilio J., Patat F., Wang L., 2017, MNRAS, 470, 1491 MZAMS ≈ 40M⊙, with the primary then exploding as a type × Shiode J. H., Quataert E., 2014, ApJ, 780, 96 Ib or type Ic supernova (with a mass of few M⊙) leaving Shiber S., Soker N., 2017, arXiv:1706.00398 behind a neutron star. A 50M⊙ star will collapse before a Smith N., 2014, ARA&A, 52, 487 40M⊙ star leaves the main-sequence, but the mass transfer Soker N., 2004, New Astron., 9, 399 will alter the evolution time, so a detailed binary evolution Soker N., 2013, arXiv:1302.5037 model is required to study the path towards the scenario Soker N., 2016, New Astron. Rev., 75, 1 described in this work. Also, hydrodynamical simulations of Soker N., Gilkis A., 2017, MNRAS, 464, 3249 the interaction between the neutron star and the envelope Svirski G., Nakar E., 2014, ApJ, 788, L14 (and the neutron star and the core) are needed to further Tartaglia L. et al., 2016, MNRAS, 459, 1039 elucidate the scenario. Thorne K. S., Zytkow A. N., 1975, ApJ, 199, L19 Vink J. S., de Koter A., Lamers H. J. G. L. M., 2001, A&A, 369, Additional study of our proposed common envelope jets 574 supernova scenario is required. At present we raise the pre- Yaron O. et al., 2017, Nature Physics, 13, 510 diction that the inner part of the ejecta might show bipolar structure due to the last jet-launching episodes.

ACKNOWLEDGMENTS

We thank Amit Kashi and Erez Michaely for helpful com- ments. This research was supported by the Israel Science Foundation and by the Pazi grant. A.G. is supported by the Blavatnik Family foundation.

REFERENCES

Andrews J. E., Smith N., 2017, arXiv:1712.00514 Arcavi I. et al., 2017, Nature, 551, 210 Armitage P. J., Livio, M., 2000, ApJ, 532, 540 Barkat Z., Rakavy G., Sack N., 1967, Physical Review Letters, 18, 379 Boian I., Groh J., 2017, arXiv:1708.05380 Chevalier R. A., 1996, ApJ, 459, 322 Chevalier R. A., 2012, ApJ, 752, L2 Foley R. J., Smith N., Ganeshalingam M., Li W., Chornock R., Filippenko A. V., 2007, ApJ, 657, L105 Fryer C. L., Woosley S. E., 1998, ApJ, 502, L9 Fuller J., 2017, MNRAS, 470, 1642 Gilkis A., Soker N., Papish O., 2016, ApJ, 826, 178 Goranskij V. P., Barsukova E. A., Valeev A. F., Tsvetkov D. Yu., Volkov I. M., Metlov V. G., Zharova A. V., 2016, Astrophys- ical Bulletin, 71, 422 Houck J. C., Chevalier R. A., 1991, ApJ, 376, 234 Ivanova N., Nandez J. L. A., 2016, MNRAS, 462, 362 Kashi A., Soker N., 2010, ApJ, 723, 602 Liu L.-D., Wang L.-J., Wang S.-Q., Dai Z.-G., 2017, arXiv:1706.01783 Margutti R. et al., 2017, ApJ, 835, 140 Mauerhan J. C. et al., 2013, MNRAS, 430, 1801

MNRAS 000, 1–5 (2017)