No velocity-kicks are required to explain large-distance offsets of Ca-rich supernovae and short-GRBs
MMNRAS , 1–8 (0000) Preprint 29 January 2021 Compiled using MNRAS L A TEX style file v3.0
No velocity-kicks are required to explain large-distance offsets ofCa-rich supernovae and short-GRBs
Hagai B. Perets and Paz Beniamini , Physics Department, Technion — Israel Institute of Technology, Haifa 3200003, Israel Theoretical Astrophysics, Walter Burke Institute for Theoretical Physics, Mail Code 350-17, Caltech, Pasadena, CA 91125, USA Astrophysics Research Center of the Open University (ARCO), The Open University of Israel, P.O Box 808, Ra’anana 43537, Israel
29 January 2021
ABSTRACT
The environments of explosive transients link their progenitors to the underlying stellar popu-lation, providing critical clues for their origins. However, some Ca-rich supernovae (SNe) andshort gamma ray burst (sGRBs) appear to be located at remote locations, far from the stellarpopulation of their host galaxy, challenging our understanding of their origin and/or physicalevolution. These findings instigated models suggesting that either large velocity kicks wereimparted to their progenitors, allowing them to propagate to large distances and attain theirremote locations; or that they formed in dense globular clusters residing in the halos. Here weshow that instead, the large spatial-offsets of these transients are naturally explained by theobservations of highly extended underlying stellar populations in (mostly early type) galaxyhalos, typically missed since they can only be identified through ultra-deep/stacked images.Consequently, n o large velocity kicks, nor halo globular cluster environments are required inorder to explain the origin of these transients. These findings support thermonuclear explosionson white-dwarfs, for the origins of Ca-rich SNe progenitors, and no/small-natal-kicks givento sGRB progenitors. Since early-type galaxies contain older stellar populations, transientarising from older stellar populations would have larger fractions of early-type galaxy hosts,and consequently larger fractions of transients at large offsets. This is verified by our results forsGRBs and Ca-rich SNe showing different offset distributions in early vs. late-type galaxies.Furthermore, once divided between early and late type galaxies, the offsets’ distributions ofthe different transients are consistent with each other. Finally, we point out that studies ofother transients’ offset distribution (e.g. Ia-SNe or FRBs) should similarly consider the hostgalaxy-type. Key words: stars: neutron stars – (stars:) gamma-ray burst: general – transients: supernova–transients:) neutron star mergers – (transients:) gamma-ray bursts – (transients:) fast radiobursts
The host galaxies and the positions of explosive transients insidetheir hosts hold critical information about their progenitors. Theprogenitors of supernovae (SNe) and gamma-ray bursts (GRBs)can be potentially linked to the underlying stellar populations intheir close environments.For example, the star formation history of the host galaxy ofa given transient provides a statistical inference of the age of theprogenitor of the transient. Given a statistical sample of transients ofa specific type, one can even infer the delay time distribution for thattype of explosive events (e.g. Maoz et al. 2014). Such inference hasa long history, with Baade noting already early-on that type II SNeexplode in early type (spiral) galaxies, while type I SNe explodeboth in early and late type galaxies, providing the first clue of ourcurrent understanding of type II SNe arising from core-collapse SNe, while type I SNe arising from both core-collapse (most Ib, IcSNe) and thermonuclear SNe (Ia SNe). The local environments oftransients inside their host galaxy provide further information onthe stellar populations of their progenitors, their metallicities andtypical lifetimes.In some cases, transients were found to be located in a differentenvironment than expected from their theorized progenitors. Thosecases assisted in identifying and eventually classifying such objectsas part of a new class of objects arising from different type of pro-genitors. Two examples, which are also the focus of this study areshort-GRBs (sGRBs) and Ca-rich SNe. For instance, long GRBsare typically observed in young environments in late type type star-forming galaxies and close to star-forming regions in these galaxies,pointing to their likely origin from massive young stars (Bloom et al.2002). In contrast, sGRBs were found both in late and in early type © a r X i v : . [ a s t r o - ph . H E ] J a n Perets & Benamini galaxies, and in many of which far from any star-forming region(Fong & Berger 2013; Fong et al. 2017). This was found to beconsistent with the idea of sGRBs arising from the mergers of neu-tron stars (NSs; and confirmed through the recent multi-messengergravitational-wave and electromagnetic counterpart GRB 170817A(Abbott et al. 2017)). Similarly, some faint type Ib SNe were foundin late-type galaxies, and far from star-forming regions, in contrastwith the expected massive progenitors for core-collapse SNe typi-cally thought to produce type Ib SNe (Perets et al. 2010). This gaverise to the identification of a new class of Ca-rich, faint type IbSNe, suggested to arise from thermonuclear He-rich explosions ofwhite-dwarfs (Perets et al. 2010), as we further discuss below.The early type host galaxies and the sometimes large distancesfrom any star-forming regions of both sGRBs and Ca-rich SNe pointto old progenitors for a significant fraction of these transients. Thisis consistent with the suggested NS mergers and thermonuclear WDexplosions which can indeed have very long delay-time since theformation of their progenitor stars. However, in many cases, not onlywere these transients far from any star-forming regions, but they alsoappeared to be located very far from their host galaxies (Perets et al.2010; Kasliwal et al. 2012; De et al. 2020; Fong & Berger 2013;Fong et al. 2017; Lunnan et al. 2017). It was therefore suggested byseveral studies that the remote locations of such transients could beexplained either by large velocity kicks given to their progenitors((Lyman et al. 2014; Foley 2015; Lyman et al. 2016) for Ca-richSNe;(Fong & Berger 2013) for sGRBs), or that the progenitorsform in dense globular clusters, since many of those reside in thehalos (Yuan et al. 2013; Sell et al. 2015; Shen et al. 2019). Inthe former scenario, the post-kick progenitors could then propagatethrough their host galaxies over hundreds of Myrs attaining remotepositions at the time of their explosions. In particular, the observedhigh velocities of pulsars point to natal kicks given to NSs upontheir formation. Since sGRBs are produced through the merger ofNSs, it was thought plausible that sGRBs could be found at largeoffsets from their host galaxies. However, as we discuss below, it isdebated whether indeed sGRB progenitors receive such large kicks.Following a similar reasoning it was suggested that many of theprogenitors of Ca-rich SNe receive some large velocity kicks, eitherdue to interactions of their progenitors with massive black holesin galaxy nuclei leading to their ejection as hypervelocity stars, orpossibly due to their progenitors containing a NS which received alarge natal kick.Already a few years ago, we pointed out that the recent find-ings that the stellar halos of early type galaxies are far larger thanpreviously thought, would give rise to transients from old stellarpopulation at these halos (Perets 2014). Since the underlying stellarpopulations can only be seen through deep imaging and/or stacking,transients arising from the halo stellar population in such galaxieswould appear to be located far from the stellar populations of theirhost galaxies, while, in fact, they formed in-situ from stars in thedifficult-to-observe underlying population.Here we follow this suggestion and study the detailed distri-butions of Ca-rich SNe and SGRBs, while dividing them betweenearly and late type galaxies, which posses very different underlyingstellar halos. As we discuss in the following, we find that the largeoffsets observed for both sGRBs and Ca-rich SNe can indeed benaturally explained without invoking any (or large) natal kicks, andtheir progenitors could have been formed in-situ.The paper is structured as follows. We first discuss the distri-bution of the stellar masses in early vs. late type galaxies (section2), we then analyze the offsets distributions of short Ca-rich SNeand sGRBs (section 3). We discuss our results and their implica- tions for Ca-rich SNe progenitors and the natal kicks given to sGRBprogenitors in section 4, and summarize (section 5).
Traditionally, only small fractions of the stellar mass in galaxieswere directly observed at large offsets (10-100 kpcs; a region whichwe term here the galaxy halo) from their centers. The halo of theMilky-way galaxy, for example, contains only 1-2 percents of thestellar mass in the Galaxy. For this reason, identification of transientsat large offsets from the centers of their host galaxies was typicallysuggested to originate from some type of velocity kicks given totheir progenitors, as such stellar progenitors were thought unlikelyto exist at such large offsets (especially when such large offsets arefrequent for some specific type of transients). In particular, largekick velocities could allow them to migrate to large distances fromtheir original birth place in the inner parts of the galaxy and/or theirdisk components, where most of the stars were expected to exist.However, the distribution of stellar mass in galaxy halos, farfrom the nucleus is difficult to measure given the low surface bright-ness in these regions. The direct measurement of the stellar massin galaxy halos for any single galaxy is notoriously difficult andcould significantly underestimate the stellar mass in the halo. Infact, galaxy formation simulations have suggested that many galaxyhalos should in fact contain significant fractions of the stellar mass(e.g Sanderson et al. 2018, and references therein).Only in the last few years did the stellar mass in the halos ofgalaxies were measured for different types and different masses ofgalaxies, over large statistical samples, allowing to infer the halos’stellar masses. Such measurements were done through the use ofstacking of observations of many galaxies belonging to the sametype and mass range (D’Souza et al. 2014), or through deep imagingof single galaxies using the Hyper Suprime-Cam (Huang et al.2018). These studies have shown that early type galaxies (ellipticalsand S0 galaxies; more centrally concentrated galaxies) have 30-70percents of their stellar mass in the halo (with a monotonically risingtrend with galaxy mass), while late type, spiral galaxies have only0-25 percents of their stellar mass in the halo (consistent with thecase of the Milky-way).The stellar population in early type galaxies is old, typicallyat least 9-10 Gyrs old (Gallazzi et al. 2005, e.g.), and no star for-mation is typically observed in galaxy halos . Taken together, onewould expect transients arising from young stellar populations tobe observed in the disks of spiral galaxies, and not in galaxy ha-los, and thereby generally have small (<10 kpc) offsets. In contrast,transients arising from old stellar progenitors should be observedwith comparable numbers in the halo and the central parts of early-type galaxies, and mostly in the disks and central parts of late-typegalaxies (since only a small fraction of the old stellar populationresides in the halo).As we discuss in what follows, these recent developments inunderstanding the distribution of the stellar mass in galaxies havefar reaching implications for the interpretations of the offsets’ distri-bution of transients, and its ramifications regarding the progenitorsof such transients, and the physical processes involved in their evo-lution. In particular, observations of large offsets for some types oftransients such as sGRBs and Ca-rich faint supernovae should notimply the need for some natal velocity kicks for their progenitors,but rather that such large-offset transients arise from old stellar pop-ulations, and were not likely to receive large natal kick and then
MNRAS000
MNRAS000 , 1–8 (0000) a-rich SNe & sGRBs ofsets distribution migrate to their observed position. As we show below, the correla-tion between the type of host galaxies and the offset distribution ofthe transients corroborates this. In the following we study the offset distribution of the two type oftransients, Ca-rich SNe and sGRBs. The samples are summarizedin Tables 1 and 2. Their cumulative distributions, divided betweenearly and late type galaxies are shown in Fig. 1. Descriptions of thetwo samples are summarized below.
Our current sample of Ca-rich SNe offsets is based on the samplefrom our original characterization study (Perets et al. 2010) basedon SNe from the LOSS and CCCP surveys, and the additionalCa-rich SNe discovered by the PTF/iPTF (Kasliwal et al. 2012;Lunnan et al. 2017; De et al. 2018), PESSTO (Valenti et al. 2014)and ZTF (De et al. 2020; Jacobson-Galán et al. 2020b) surveys.The SNe offsets are taken from these studies, and summarized inTable 1; the cumulative offset distributions divided between earlyand late type galaxies can be found in Fig. 1. The error bars onthe offsets of these SNe are typically not given but are comparableto the least significant digit. We do not consider one Ca-rich SN,PTF12bho suggested to explode in the intracluster medium (sincea large fraction of the stellar mass in galaxy clusters reside in theintracluster medium (e.g. Gonzalez et al. 2007; Da Rocha et al.2008, and references therein); the existence on one such SN inour sample could be expected). For the majority of the SNe, thehost galaxy is the closest galaxy and can be well identified. In twocases (PTF11kmb and PTF11bij) we followed the suggested hostand offset identification in the discovery papers, although severalgalaxies were found nearby, and the exact identification could bedebated. These might even relate to the intracluster regions wherethey were found. Indeed, ∼ −
20% of type Ia SNe in clusters arefound in the intracluster medium far from any host galaxy (Sandet al. 2011).We note that various SNe with only partial similarities to faint,type Ib, Ca-rich SNe characteristics were discussed in the literature(see e.g. De et al. 2020; Jacobson-Galán et al. 2020a for recentoverviews, and references therein); these are not considered in oursample of bona-fide Ca-rich SNe).
The offsets of sGRBs have been extensively studied by variousauthors. We have compiled an up-to-date list of sGRBs with pub-lished offsets and galaxy types from (Fong et al. 2017; Gompertzet al. 2020; Paterson et al. 2020; O’Connor et al. 2020). We limitthe sample to those bursts with well determined offsets (i.e. error inoffset determination of (cid:46)
50% and excluding bursts with ambiguityin the host galaxy determination). Our sample consists of 8 sGRBsin early type galaxies and 15 sGRBs in late type galaxies and islisted in table 2.In figure 1 we plot the cumulative offset distribution of sGRBsin early and late type galaxies (in contrast with the Ca-rich sam-ple, some of the error-bars on the offsets are not negligible, andwe therefore also show the cumulative distribution which includethe error-bars in Fig. A1; the overall distribution is, however, is
Table 1.
Sample of Ca-rich SNe used in this work.
Supernova offset (kpc) Galaxy type2000ds 3.77 Early2001co 7.07 Late2003H 8.73 Late2003dg 1.66 Late2003dr 2.65 Late2005E 24.27 Early2005cz 2.12 Early2007ke 16.71 Early2010et 37.64 Early2012hn 6.73 EarlyPTF11bij 34.42 EarlyPTF11kmb 150.05 Early2016hgs 5.9 Late2018ckd 19.08 Early2018lqo 15.46 Early2018lqu 26.70 Early2018gwo 8.56 Early2018kjy 6.35 Early2019ehk 1.8 Late2019hty 8.73 Early2019pxu 17.56 Late qualitatively the same. Large offsets occur predominantly in sGRBsassociated with early type galaxies. Indeed, already with the limitedsample size available at the present time, the offset distributions inearly and late type galaxies can be ruled out as being drawn from thesame population at a level of 90% confidence. This is inconsistentwith the hypothesis that the offsets are dominated by strong kicks, inwhich case, the observed offsets (and especially for the large offsetsystems) will become independent of the galaxy type.The bottom panel of figure A1 shows the same distributions,but now with the offsets normalized to the hosts’ effective stellarradii. With this normalization, the two distributions become sta-tistically consistent with each other, supporting the idea that thespatial extent of star formation is the main component controllingthe observed offsets. We caution however that although the hostnormalized offset is a big step towards accounting for the galaxies’underlying stellar mass distributions, it is not in itself complete, as,especially at larger offsets, the underling stellar mass can only berevealed by ultra deep / stacked images, as discussed above.
Ca-rich SNe are faint (> ∼ ∼
23 kpc) from its host galaxy center. At the time, all typeIb SNe were thought to arise from the core-collapse of a massivestar. We therefore initially thought that SN 2005E progenitor wasa massive star that formed in the galaxy nucleus (where gas andstar formation might exist even in an early type galaxy), and waslater ejected as a hypervelocity star following an interaction withthe central massive black hole. It then propagated in the galaxy andexploded in its observed remote location in the galaxy halo. Al-though potentially possible, our analysis showed that the likelihood
MNRAS , 1–8 (0000)
Perets & Benamini
Figure 1.
The cumulative offsets distributions of sGRBs and Ca-rich su-pernovae in early and late type galaxies. The offsets distributions of bothSGRBs and Ca-rich SNe, differs between the early type host galaxies andthe late type host galaxies. The offsets distribution is far more extendedin the former, compared with latter, consistent with the far more extendeddistribution of the halo stellar mass in early type galaxies, compared withlate type galaxies, and consistent with the observed stellar masses in thesetypes of galaxies (vertical lines, from D’Souza et al. 2014). These resultssuggest an in-situ formation for the progenitors of these transients, withoutthe need of any natal velocity kicks, nor the existence of globular clusters.. of observing an event from such a scenario was highly unlikely(Perets et al. 2010).After collecting and characterizing a sample of such SNe, wefound many of them to explode in old stellar environments, andsome at large offsets similar to SN 2005E. Together with our find-ing of inferred low ejecta-mass, low energy, low velocities and low Ni mass, as well as potentially large abundances of intermediateelements (and the lack of any star-forming regions close-by) weconcluded that Ca-rich SNe likely constitute a novel type of SNexplosion arising from a thermonuclear He-rich explosion on a rel-atively old white dwarf (Perets et al. 2010). We therefore suggestedthat this scenario well explains the properties of these SNe, and theexistence of such type Ib SNe in old stellar environments such asearly type galaxies and galaxy halos (Perets et al. 2010; Waldmanet al. 2011; Perets et al. 2011; Kasliwal et al. 2012; Lyman et al.2013; Perets 2014).The majority of Ca-rich SNe were found in old stellar pop-ulations in early type galaxies (Perets et al. 2010; De et al. 2020,e.g.) it was therefore expected that the locations and offsets of suchSNe in their host galaxies would follow the old stellar populationsin these galaxies. However, many of the SNe both in our original(Perets et al. 2010) sample and additional Ca-rich SNe identifiedlater on (Kasliwal et al. 2012; De et al. 2020) were found at largeoffsets from their host galaxies (see Table 1 and Fig. 1), where verylittle, if any, stellar population was thought to exist.As we first briefly suggested in Perets (2014) and now explorein detail in this study, the large offsets are in fact, a natural outcomefrom the existence of extended stellar halos of (mostly) early typegalaxies (D’Souza et al. 2014; Huang et al. 2018), as discussed insection 2.As can be seen in Fig. 1, The offsets distribution in early typeand late galaxies show a different behaviour, with most of the SNe
Table 2.
Sample of sGRBs used in this work.
GRB Type log ( 𝑀 ∗ / 𝑀 (cid:12) ) offset (kpc) norm. offset050509B early 11 . ± .
03 63 . ± . . ± . . ± .
07 3 . ± .
027 1 . ± . . ± .
05 2 . ± .
079 0 . ± . . ± .
64 2 . ± .
19 0 . ± . ∼ . ±
19 -061006 late 10 . ± .
23 1 . ± .
24 0 . ± . ∼ . . ∼ . . ± .
87 4 . ± . ∼ . . ± .
14 1 . ± . ∼ . . ± .
71 9 . ± . ∼ . . ± .
24 3 . ± . . ± . . ± .
19 10 . ± . . ± .
25 0 . ± . ∼ . . ± .
89 1 . ± . ∼ . . ± .
15 15 . ± . ∼ . . ± .
33 0 . ± . . ± . . ± . ∼ . . ± . ∼ . . ± .
17 1 . ± . . ± .
12 7 . ± .
07 0 . ± . . ± .
03 2 . ± .
001 0 . ± . . ± .
16 5 . ± .
38 -200522A late 9 . ± .
02 1 . ± .
27 4 ± . in late galaxies residing in the central (<10 kpc) parts, while mostof the SNe in early-type galaxies residing in the halo (>10 kpc),nicely consistent with the stellar mass fractions for late and earlytype galaxies, respectively, as found in observations (D’Souza et al.2014; Huang et al. 2018). In fact, in retrospect, the large offsetdistribution of these SNe, and its difference in early vs. late typegalaxies (first shown by us here in Fig. 1) anticipated these findings.The perplexing findings of large offsets instigated several stud-ies suggesting that the remote locations of such transients could beexplained either by large velocity kicks given to their progenitorseither because their progenitors were NSs receiving a natal-kick atbirth.Yuan et al. (2013) and later Sell et al. (2015) and Shen et al.(2019) suggested that the extended distribution of the locationsof Ca-rich is consistent with the distribution of globular clustersdistribution. However, photometric searches for globular clustersclose to the positions of known Ca-rich transients have generallybeen unsuccessful (Lyman et al. 2014). Furthermore, the relativelyhigh inferred rate of Ca-rich SNe (5-15% of the Ia SNe rate; Peretset al. 2010; De et al. 2020) could be difficult to explain with aglobular cluster origin, given the low mass fraction in globularclusters.Others suggested a large velocity kick is imparted to the pro-genitors of Ca-rich SNe (Lyman et al. 2014; Foley 2015; Lymanet al. 2016). Such scenarios are difficult to reconcile with the sug-gested WD merger/He-accretion origins of Ca-rich SNe (Perets et al.2010; Shen et al. 2010; Waldman et al. 2011; Dessart & Hillier 2015; MNRAS000
27 4 ± . in late galaxies residing in the central (<10 kpc) parts, while mostof the SNe in early-type galaxies residing in the halo (>10 kpc),nicely consistent with the stellar mass fractions for late and earlytype galaxies, respectively, as found in observations (D’Souza et al.2014; Huang et al. 2018). In fact, in retrospect, the large offsetdistribution of these SNe, and its difference in early vs. late typegalaxies (first shown by us here in Fig. 1) anticipated these findings.The perplexing findings of large offsets instigated several stud-ies suggesting that the remote locations of such transients could beexplained either by large velocity kicks given to their progenitorseither because their progenitors were NSs receiving a natal-kick atbirth.Yuan et al. (2013) and later Sell et al. (2015) and Shen et al.(2019) suggested that the extended distribution of the locationsof Ca-rich is consistent with the distribution of globular clustersdistribution. However, photometric searches for globular clustersclose to the positions of known Ca-rich transients have generallybeen unsuccessful (Lyman et al. 2014). Furthermore, the relativelyhigh inferred rate of Ca-rich SNe (5-15% of the Ia SNe rate; Peretset al. 2010; De et al. 2020) could be difficult to explain with aglobular cluster origin, given the low mass fraction in globularclusters.Others suggested a large velocity kick is imparted to the pro-genitors of Ca-rich SNe (Lyman et al. 2014; Foley 2015; Lymanet al. 2016). Such scenarios are difficult to reconcile with the sug-gested WD merger/He-accretion origins of Ca-rich SNe (Perets et al.2010; Shen et al. 2010; Waldman et al. 2011; Dessart & Hillier 2015; MNRAS000 , 1–8 (0000) a-rich SNe & sGRBs ofsets distribution Perets et al. 2019). They are also inconsistent with the expected ratesand delay time distribution NS-WD mergers origins (Toonen et al.2018; Zenati et al. 2019), suggested as progenitors for Ca-rich SNe(Metzger 2012), and with the ejection rates of hypervlocity stars bymassive black holes (Perets et al. 2010). Furthermore, as we showedhere, the offsets in early type galaxies appear to extend farther thanin late type galaxies, raising an extra challenge to a kick originwhich relies on the micro-physics of the progenitor rather than onthe galaxy macro-physics.The results shown here therefore reconcile the difficulties andinconsistencies inferred form the large sample of Ca-rich SNe withlarge offsets. We find the large apparent offsets of Ca-rich SNefrom the underlying stellar distribution is just an artifact of thelow-surface stellar brightness in galaxy halos, which is difficult toresolve with regular imaging, and requires the use of stacked or deepimaging. We conclude that Ca-rich SNe mostly arise from an oldstellar population, explaining the large fraction of early type galaxieshosts. The high frequency of early type hosts and the large extendedstellar halos of such galaxies explain the large offset distributionamong Ca-rich SNe.We note that in early type galaxies the fraction of Ca-rich SNein the central 2-3 kpc appears to be lower than that of sGRBs insuch galaxies. This might only reflect the low-statistics, but it isimportant to note that one should expect lower detection rates in thedenser inner regions of galaxies, especially for very subluminousSNe such as Ca-rich SNe, as already suggested by us (Perets 2014),i.e. we expect significant Shaw’s selection bias (Shaw 1979) forsuch SNe.Finally, early type galaxies have higher metallicities than late-type galaxies (Li et al. 2018, e.g). We therefore raise the possibilitythat Ca-rich SNe progenitors originate from high metallicity en-vironments, rather than low-metallicities as suggested by us andothers in the past (Yuan et al. 2013; Perets 2014; Shen et al. 2019).However, detailed study of the host metallicities for Ca-rich SNehas yet to be done.
The binary neutron star (BNS) merger origin for sGRBs has beensuggested decades ago (Goodman 1986; Eichler et al. 1989). Itreceived a conclusive proof with the detection of GRB 170817, aGRB accompanying a GW trigger from a BNS merger (Abbott et al.2017). Since BNSs receive kicks when the second star in the binarycollapses to a neutron star, they could potentially travel significantlybefore merging and producing a sGRB. An estimation of these kicks,merger delay times and offset distances can be obtained from theGalactic population of BNSs.Shortly after the discovery of the double pulsar J0737-3039(Burgay et al. 2003), the orbital parameters of this system led Piran &Shaviv (2004, 2005) to suggest that the younger pulsar in the system,pulsar B, must have formed with very small mass ejection and withvirtually no kick velocity. The system had a very small eccentricityand it was located in the Galactic plane. The small eccentricitysuggested a small mass ejection, unless a strongly fine-tuned kickwas given to the system. However, the latter would have led to a largeCM velocity and the location of the system in the Galactic planeindicated that at least the vertical (out of the plane) component ofthe velocity of the system was low. The very small proper motionpredicted by Piran & Shaviv (2004, 2005) was confirmed within ayear by pulsar timing observations (Kramer et al. 2006). Dall’Ossoet al. (2014) refined the original estimates, using the observed propermotion and other parameters of the system. They confirmed the earlier expectations and have shown that in this system, the olderpulsar (A) also most likely formed in this way.Later on, Beniamini & Piran (2016) have shown that out of 10BNS systems observed in the Galaxy that don’t reside in globularclusters, between 6 and 7 must have been formed by collapses involv-ing the release of a small amount of mass Δ 𝑀 (cid:46) . 𝑀 (cid:12) and impart-ing only a very weak kick 𝑣 𝑘 (cid:46) Δ 𝑣 cm (cid:46) − .The other BNS systems require more typical core-collapses with anejected mass of one to a few solar masses and kicks of a few hundredkm/sec. Since the publication of Beniamini & Piran (2016), fournew BNS systems have been reported: J1913+1102, J1757-1854,J1411-2551, J1946+2052 (Lazarus et al. 2016; Martinez et al. 2017;Cameron et al. 2018; Stovall et al. 2018). Three out of these fourhave small eccentricities 𝑒 = . , . , .
17, suggesting a weakcollapse origin and only one has 𝑒 = .
6. This ratio is in a niceagreement with the earlier expectations and suggests that the re-sult is not dominated by a statistical fluctuation. Furthermore, theseconclusions were independently reproduced by Tauris et al. (2017).Overall, approximately 2 / 𝑧 typ ∼ Δ 𝑣 cm 𝑃 𝑧 / 𝜋 ∼ .
08 kpc, where 𝑃 𝑧 ∼
50 Myr is thetime-scale for 𝑧 oscillations in the potential of the Galaxy. This ismore than two orders of magnitude lower than the offsets measuredfor some sGRBs in galaxies with similar stellar mass to that ofthe Milky Way (see table 2). Therefore, if BNSs are born closeenough to the Galactic plane, that their motion is dominated bythe Galactic potential, the expected distribution of kicks does notnaturally account for large off-sets. To demonstrate the latter pointwe plot in figure 2 the vertical offset distribution of BNS systems atthe time of merger, as resulting from a Monte Carlo simulation, inwhich the birth conditions of the binaries are informed by constraintsfrom the Galactic BNS population and the systems’ motion in theGalaxy’s potential is then calculated with galpy (Bovy 2015) usingthe MWPotential2014
Galactic potential (full details of the MonteCarlo calculation are given in appendix B). The results re-affirm theOoM estimate for 𝑧 typ above, and the tension with sGRB offsets.Of course, if BNS systems can be formed far away from thecenters of their host galaxies, large off-sets can be naturally ex-plained with no need for invoking kicks . In such a situation, theoffsets of BNS mergers should correlate with the spatial extent ofstar formation in their host galaxies. This too is demonstrated in the http://github.com/jobovy/galpy We note, however, that at large offsets, the galactic potentials can becomesufficiently weak such that the BNS can travel unimpeded, even with a verysmall change to its center of mass. The BNS will then cover a distance ≈ 𝑡 mrg Δ 𝑣 cm before merging (where 𝑡 mrg is the delay time between BNSformation and merger). Recently, (Beniamini & Piran 2019) have shown thatthe lifetime of the pulsars in Galactic BNS systems are sufficiently short,that the delay times may be directly inferred from the currently observedtime until merger in those systems. The result is a rather steep delay timedistribution in which at least 40 −
60% (10 − , 1–8 (0000) Perets & Benamini
Figure 2.
Results of a Monte Carlo simulation calculating the vertical off-sets, | 𝑧 | off , of BNS systems from the Galactic plane at the time of merger(taking into account constraints on kicks and time to merger, consistent withconstraints from the Galactic population, see §B for details). We show resultswith the initial vertical offsets, | 𝑧 | ini , either in the Galactic plane (yellow) orfollowing the disc thickness (purple). The results demonstrate that the offsetsinduced by kicks alone are typically (cid:46) . results of our Monte Carlo calculation for the Milky Way presentedin figure 2. This also naturally leads to a systematic difference be-tween the offsets of BNS mergers in early vs late time galaxies asseen in observations (see section 3). Ca-rich SNe and sGRBs are two groups of transients, of which mem-bers show an extended spatial distributions in their host galaxies. Inparticular, a large fraction of these transients are found at large (>10kpc) from the nuclei of their galaxy hosts, and apparently far fromany observed underlying stellar population, which could host theirprogenitors. These perplexing findings motivated various solutions.These include models in which the progenitors of such transientsreceived large velocity kicks allowing them to migrate to remotepositions where they exploded; or that the progenitors originatedin globular clusters, the distribution of which extend far into thehalo. However, such solutions are difficult to reconcile with otherproperties of the transients and their suggested progenitors, as wediscussed above.In this study we analyzed the distributions of the distanceoffsets of Ca-rich supernovae and sGRBs from their host galaxies.We pointed out that stacked-images and deep-imaging of early andlate type galaxies show different stellar mass distribution, and thatearly type galaxies have very extended stellar mass, with a largefraction, typically the majority of the stellar mass resides in thehalo beyond 10 kpc. Therefore, in contrast to previous studies, wedivided the offsets distributions between early and late type galaxies.We showed that the offsets in early type galaxies extend farther thanthat in late-type galaxies. Moreover, the fractions of transients in thehalo (>10 kpc) vs. the central parts (<10 kpc) are consistent with theobserved distinctly different halo to central parts stellar mass ratiosin early type and late type galaxies (D’Souza et al. 2014; Huanget al. 2018).We also studied the expected offset distribution for progenitors that receive velocity kicks and show that no large kicks are requiredin order to explain the observed offsets.We conclude that the progenitors of both Ca-rich SNe andsGRBs do not require large velocity kicks, nor do their progenitorneed to form in globular clusters. Rather, their offset distributionsare generally consistent with the underlying stellar mass distribu-tion of their host galaxies. The appearance of large offsets is thenconsistent with old stellar progenitors for significant fractions ofthese transients.Finally, we note that since stacked/ultra-deep imaging showthat early-type galaxies are more extended than late-type galax-ies, any study of other transients’ offset distribution, such as FRBsshould account for the host galaxy-type. In fact, the recent findingof FRBs at large offsets (Ravi et al. 2019), does not require theirprogenitors to have had natal kicks, and we expect FRBs with largeoffsets to be mostly identified in early type galaxies.
DATA AVAILABILITY
The simulations underlying this article will be shared on reasonablerequest to the corresponding author.
ACKNOWLEDGEMENTS
HBP acknowledges support for this project from the EuropeanUnion’s Horizon 2020 research and innovation program under grantagreement No 865932-ERC-SNeX. The research of PB was fundedby the Gordon and Betty Moore Foundation.
REFERENCES
Abbott B. P., et al., 2017, ApJ, 848, L12Beniamini P., Piran T., 2016, MNRAS, 456, 4089Beniamini P., Piran T., 2019, MNRAS, 487, 4847Beniamini P., Hotokezaka K., Piran T., 2016, ApJ, 829, L13Bloom J. S., Kulkarni S. R., Djorgovski S. G., 2002, AJ, 123, 1111Bovy J., 2015, ApJS, 216, 29Burgay M., et al., 2003, Nature, 426, 531Cameron A. D., et al., 2018, MNRAS, 475, L57D’Souza R., Kauffman G., Wang J., Vegetti S., 2014, MNRAS, 443, 1433Da Rocha C., Ziegler B. L., Mendes de Oliveira C., 2008, MNRAS, 388,1433Dall’Osso S., Piran T., Shaviv N., 2014, MNRAS, 438, 1005De K., et al., 2018, ApJ, 866, 72De K., et al., 2020, ApJ, 905, 58Dessart L., Hillier D. J., 2015, MNRAS, 447, 1370Eichler D., Livio M., Piran T., Schramm D. N., 1989, Nature, 340, 126Foley R. J., 2015, MNRAS, 452, 2463Fong W., Berger E., 2013, ApJ, 776, 18Fong W., et al., 2017, ApJ, 848, L23Gallazzi A., Charlot S., Brinchmann J., White S. D. M., Tremonti C. A.,2005, MNRAS, 362, 41Gompertz B. P., Levan A. J., Tanvir N. R., 2020, ApJ, 895, 58Gonzalez A. H., Zaritsky D., Zabludoff A. I., 2007, ApJ, 666, 147Goodman J., 1986, ApJ, 308, L47Huang S., et al., 2018, MNRAS, 480, 521Jacobson-Galán W. V., et al., 2020a, arXiv e-prints, p. arXiv:2010.15863Jacobson-Galán W. V., et al., 2020b, ApJ, 898, 166Kasliwal M. M., et al., 2012, ApJ, 755, 161Kramer M., et al., 2006, Science, 314, 97Lazarus P., et al., 2016, ApJ, 831, 150Li H., et al., 2018, MNRAS, 476, 1765 MNRAS000
Abbott B. P., et al., 2017, ApJ, 848, L12Beniamini P., Piran T., 2016, MNRAS, 456, 4089Beniamini P., Piran T., 2019, MNRAS, 487, 4847Beniamini P., Hotokezaka K., Piran T., 2016, ApJ, 829, L13Bloom J. S., Kulkarni S. R., Djorgovski S. G., 2002, AJ, 123, 1111Bovy J., 2015, ApJS, 216, 29Burgay M., et al., 2003, Nature, 426, 531Cameron A. D., et al., 2018, MNRAS, 475, L57D’Souza R., Kauffman G., Wang J., Vegetti S., 2014, MNRAS, 443, 1433Da Rocha C., Ziegler B. L., Mendes de Oliveira C., 2008, MNRAS, 388,1433Dall’Osso S., Piran T., Shaviv N., 2014, MNRAS, 438, 1005De K., et al., 2018, ApJ, 866, 72De K., et al., 2020, ApJ, 905, 58Dessart L., Hillier D. J., 2015, MNRAS, 447, 1370Eichler D., Livio M., Piran T., Schramm D. N., 1989, Nature, 340, 126Foley R. J., 2015, MNRAS, 452, 2463Fong W., Berger E., 2013, ApJ, 776, 18Fong W., et al., 2017, ApJ, 848, L23Gallazzi A., Charlot S., Brinchmann J., White S. D. M., Tremonti C. A.,2005, MNRAS, 362, 41Gompertz B. P., Levan A. J., Tanvir N. R., 2020, ApJ, 895, 58Gonzalez A. H., Zaritsky D., Zabludoff A. I., 2007, ApJ, 666, 147Goodman J., 1986, ApJ, 308, L47Huang S., et al., 2018, MNRAS, 480, 521Jacobson-Galán W. V., et al., 2020a, arXiv e-prints, p. arXiv:2010.15863Jacobson-Galán W. V., et al., 2020b, ApJ, 898, 166Kasliwal M. M., et al., 2012, ApJ, 755, 161Kramer M., et al., 2006, Science, 314, 97Lazarus P., et al., 2016, ApJ, 831, 150Li H., et al., 2018, MNRAS, 476, 1765 MNRAS000 , 1–8 (0000) a-rich SNe & sGRBs ofsets distribution Lunnan R., et al., 2017, ApJ, 836, 60Lyman J. D., James P. A., Perets H. B., Anderson J. P., Gal-Yam A., MazzaliP., Percival S. M., 2013, MNRAS, 434, 527Lyman J. D., Levan A. J., Church R. P., Davies M. B., Tanvir N. R., 2014,MNRAS, 444, 2157Lyman J. D., Levan A. J., James P. A., Angus C. R., Church R. P., DaviesM. B., Tanvir N. R., 2016, MNRAS, 458, 1768Maoz D., Mannucci F., Nelemans G., 2014, ARA&A, 52, 107Martinez J. G., et al., 2017, ApJ, 851, L29Metzger B. D., 2012, MNRAS, 419, 827O’Connor B., et al., 2020, arXiv e-prints, p. arXiv:2012.00026Paterson K., et al., 2020, ApJ, 898, L32Perets H. B., 2014, arXiv e-prints, p. arXiv:1407.2254Perets H. B., et al., 2010, Nature, 465, 322Perets H. B., Gal-yam A., Crockett R. M., Anderson J. P., James P. A.,Sullivan M., Neill J. D., Leonard D. C., 2011, ApJ, 728, L36Perets H. B., Zenati Y., Toonen S., Bobrick A., 2019, arXiv e-prints, p.arXiv:1910.07532Piran T., Shaviv N. J., 2004, ArXiv Astrophysics e-prints,Piran T., Shaviv N. J., 2005, Physical Review Letters, 94, 051102Ravi V., et al., 2019, Nature, 572, 352Sand D. J., et al., 2011, ApJ, 729, 142Sanderson R. E., et al., 2018, ApJ, 869, 12Sell P. H., Maccarone T. J., Kotak R., Knigge C., Sand D. J., 2015, MNRAS,450, 4198Shaw R. L., 1979, A&A, 76, 188Shen K. J., Kasen D., Weinberg N. N., Bildsten L., Scannapieco E., 2010,ApJ, 715, 767Shen K. J., Quataert E., Pakmor R., 2019, ApJ, 887, 180Stovall K., et al., 2018, ApJ, 854, L22Tauris T. M., et al., 2017, ApJ, 846, 170Toonen S., Perets H. B., Igoshev A. P., Michaely E., Zenati Y., 2018, A&A,619, A53Valenti S., et al., 2014, MNRAS, 437, 1519Waldman R., Sauer D., Livne E., Perets H., Glasner A., Mazzali P., TruranJ. W., Gal-Yam A., 2011, ApJ, 738, 21Yuan F., Kobayashi C., Schmidt B. P., Podsiadlowski P., Sim S. A., ScalzoR. A., 2013, MNRAS, 432, 1680Zenati Y., Perets H. B., Toonen S., 2019, MNRAS, 486, 1805
APPENDIX A: HOST NORMALIZED SGRB OFFSETS
We plot in Fig. A1 the cumulative distribution of sGRB offsetsseparating different galaxy types and considering also the host nor-malized offsets (see §3.2 for details).
APPENDIX B: MONTE CARLO SIMULATION OF BNSMERGER VERTICAL OFFSETS
We apply a Monte Carlo calculation to estimate the distribution ofvertical offsets of BNS mergers above the Galactic plane. We be-gin by drawing the birth locations of the binaries. The systems areassumed to either (i) be born in the Galactic plane (this allows to iso-late the effect of kicks from the initial height distribution) or (ii) bedrawn from a distribution 𝑑 P / 𝑑𝑧 ∝ exp (− 𝑟 / ℎ 𝑧 ) with ℎ 𝑧 = . MWPotential2014
Galactic potential (Bovy2015)). In either case, their Galactic radius is drawn from a distri-bution 𝑑 P / 𝑑𝑟 ∝ exp (− 𝑟 / ℎ 𝑟 ) with ℎ 𝑟 = . MWPotential2014 ). The systems are initially rotating in the Galac-tic with their tangential velocity set by the
MWPotential2014 ro-tation curve. Next, we randomize the kicks and mass ejection inthe superonova leading to the the birth of the second NS in thebinary. Following Beniamini & Piran (2016) we take log-normaldistributions with median values of Δ 𝑀 , 𝑣 k , respectively and with Figure A1.
Top: cumulative distribution of sGRB offsets. Results are sep-arated to early-type galaxies (blue) and late-type (red). Bottom: Same asabove, but for the offsets normalized by the respective effective stellar radiiof each galaxy. widths 𝜎 Δ 𝑀 / Δ 𝑀 = 𝜎 𝑣 k , / 𝑣 k , = .
5. As per Beniamini & Piran(2016), 2/3 of the systems surviving the supernova (and mergingwithin less than a Hubble time) have weak collapses and smallmass ejection, Δ 𝑀 = . 𝑀 (cid:12) , 𝑣 k , = − while the other1/3 have values more typical of standard core-collapse supernovae, Δ 𝑀 = 𝑀 (cid:12) , 𝑣 k , = − . We also include an additionalamount of ejecta, Δ 𝑀 𝜈 = . 𝑀 (cid:12) due to neutrino emission (Be-niamini et al. 2016). The direction of the kicks is assumed to beuncorrelated with the orientation of the binary’s orbit. For the lat-ter, we also assume the orbit to be circular before the supernovatakes place with the initial separation of the systems following a PLdistribution 𝑑𝑁 / 𝑑𝑎 ∝ 𝑎 − . above 𝑎 min = . × cm. The latterdistribution is consistent with the observed Galactic binaries andreproduces their observed delay time distribution between secondNS formation, and BNS merger (Beniamini & Piran 2019). Withthese ingredients in place, we self-consistently calculate the result-ing orbit in each system, the time between its formation and mergerand its center of mass velocity (see Beniamini et al. 2016 for moredetails). We then propagate the motion of the systems in the Galac-tic potential using galpy (Bovy 2015) until the time of merger. Werepeat this process 10 times and record the distribution of heightsof merging binaries above the Galactic plane. MNRAS , 1–8 (0000)
Perets & Benamini
This paper has been typeset from a TEX/L A TEX file prepared by the author. MNRAS000