merger product for gw170817: bh or ns?

MergerproductforGW170817:BHorNS?

@ 厦⻔门 2018-05-23 �

He Gao（高鹤）

Department of Astronomy, Beijing Normal University 北京师范大学－天文系

第二届引力波天体物理学术研讨会�

Abbott et al. 2017，ApJL �

Multi-Messenger GW detector triggered； 1.7s later, Fermi／GBM Detect a Short GRB; 10 hours later, an optical IR counterpart detected; 9 days later，Chandra find a X-ray counterpart; 16 days later, VLA detected a radio counterpart.

Era of GW Astronomy has been Opened GW170817�

Figure 2. Timeline of the discovery of GW170817, GRB 170817A, SSS17a/AT 2017gfo, and the follow-up observations are shown by messenger and wavelengthrelative to the time tc of the gravitational-wave event. Two types of information are shown for each band/messenger. First, the shaded dashes represent the times wheninformation was reported in a GCN Circular. The names of the relevant instruments, facilities, or observing teams are collected at the beginning of the row. Second,representative observations (see Table 1) in each band are shown as solid circles with their areas approximately scaled by brightness; the solid lines indicate when thesource was detectable by at least one telescope. Magnification insets give a picture of the first detections in the gravitational-wave, gamma-ray, optical, X-ray, andradio bands. They are respectively illustrated by the combined spectrogram of the signals received by LIGO-Hanford and LIGO-Livingston (see Section 2.1), theFermi-GBM and INTEGRAL/SPI-ACS lightcurves matched in time resolution and phase (see Section 2.2), 1 5×1 5 postage stamps extracted from the initial sixobservations of SSS17a/AT 2017gfo and four early spectra taken with the SALT (at tc+1.2 days; Buckley et al. 2017; McCully et al. 2017b), ESO-NTT (attc+1.4 days; Smartt et al. 2017), the SOAR 4 m telescope (at tc+1.4 days; Nicholl et al. 2017d), and ESO-VLT-XShooter (at tc+2.4 days; Smartt et al. 2017) asdescribed in Section 2.3, and the first X-ray and radio detections of the same source by Chandra (see Section 3.3) and JVLA (see Section 3.4). In order to showrepresentative spectral energy distributions, each spectrum is normalized to its maximum and shifted arbitrarily along the linear y-axis (no absolute scale). The highbackground in the SALT spectrum below 4500Å prevents the identification of spectral features in this band (for details McCully et al. 2017b).

4

The Astrophysical Journal Letters, 848:L12 (59pp), 2017 October 20 Abbott et al.

GW Analysis：no answer

Total mass 2.74M¤

Gamma-ray burst：no answer

BH? �

Magnetar?�

Merger-novae ?

Kilpatrick et al. 2017, Nature�

c�

Red�

Blue�

Large Mej means Large Mass Ratio

4

FIG. 3.—

tative microphysical EoS, which are derived within different

frameworks, including nonrelativistic and relativistic nuclearenergy-density functionals, Brueckner-Hartree-Fock calcula-tions, and phenomenological models, such as the liquid-dropmodel (Bauswein et al. 2013, and reference therein for de-tails). In figure 3, we plot how the mass of the dynamicalejecta varying with the mass ratio for different EoS. Fromthe fitting of kilonova emission of GW 170817, we have5.2 < Mej < 13. The mass ratio is thus constrained in a tightrange of 0.40 - 0.51. Therefore, the mass of two NSs couldbe constrained to 0.86 M� < M1 < 0.95 M� and 1.87 M� <M2 < 2.26 M� respectively. M1 . 1 M�, M2 & 2M�

5. CONCLUSION AND DISCUSSION

This work is supported by the National Basic Research Pro-gram (973 Program) of China (Grant No. 2014CB845800),the National Natural Science Foundation of China underGrant No. 11722324, 11603003, 11633001 and 11690024,and the Strategic Priority Research Program of the ChineseAcademy of Sciences, Grant No. XDB23040100.

REFERENCES

Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2016,PhRvL.116f1102AAbbott, B. P., Abbott, R., Abbott, T. D., et al. 2016,PhRvL.116x1103AAbbott, B. P., Abbott, R., Abbott, T. D., et al. 2017,PhRvL.,118v1101AAbbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, PhRvL., 119v141101AAbbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, PhRvL., 119v161101AAbbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, ApJL., 848,13Baiotti, L., & Rezzolla, L. 2017, Reports on Progress in Physics, 80, 096901Barnes, J. and Kasen, D., 2013,ApJ,775,18BBauswein, A., Goriely, S., & Janka, H.-T. 2013, ApJ, 773, 78Berger, E. 2014, ARA&A, 52, 43Dietrich, T., & Ujevic, M. 2017, Classical and Quantum Gravity, 34, 105014Eichler, D., Livio, M., Piran, T., & Schramm, D. N. 1989, Nature, 340, 126Fernández, R., & Metzger, B. D. 2013, MNRAS, 435, 502Gao, H. and Ding, X. and Wu, X.-F. and Zhang, B. and Dai, Z.-G.,

2013,ApJ,771,86GGao, H., Zhang, B., Lü, H.-J. 2016, Phys. Rev. D, 93, 044065Goldstein, A., Veres P., Burns, E., et al. 2017, ApJL in pressHotokezaka, K., Kiuchi, K., Kyutoku, K., et al. 2013, Phys. Rev. D, 87,

024001Just, O., Bauswein, A., Pulpillo, R. A., Goriely, S., & Janka, H.-T. 2015,

MNRAS, 448, 541Kasen, D., & Bildsten, L. 2010, ApJ, 717, 245Kasen, D., Badnell, N. R., & Barnes, J. 2013, ApJ, 774, 25Kasen, D., Metzger, B. D., & Barnes, J. 2017, Nature, in pressKorobkin, O., Rosswog, S., Arcones, A., & Winteler, C. 2012, MNRAS, 426,

1940Kotera, K., Phinney, E. S., & Olinto, A. V. 2013, MNRAS, 432, 3228

Lasky, P. D., Haskell, B., Ravi, V., Howell, E. J., & Coward, D. M. 2014,Phys. Rev. D, 89, 047302

Lehner, L., Liebling, S. L., Palenzuela, C., et al. 2016, Classical and QuantumGravity, 33, 184002

Lei, W. H., Zhang, B. & Liang, E. W. 2013, ApJ, 756, 125Li, L.-X. and Paczynski, B., 1998,ApJ,507L,59LMetzger, B. D. and Martínez-Pinedo, G. and Darbha, S. and Quataert, E. and

Arcones, A. and Kasen, D. and Thomas, R. and Nugent, P. and Panov, I. V.and Zinner, N. T., 2010MNRAS.,406.2650M

Metzger, B. D., & Piro, A. L. 2014a, MNRAS, 439, 3916Metzger, B. D., & Fernández, R. 2014, MNRAS, 441, 3444Metzger, B. D. 2017, Living Reviews in Relativity, 20, 3Nakar, E., & Piran, T. 2011, Nature, 478, 82Narayan, R., Paczynski, B., & Piran, T. 1992, ApJ, 395, L83Rezzolla, L., Giacomazzo, B., Baiotti, L., et al. 2011, ApJ, 732, L6Rosswog, S., Piran, T., & Nakar, E. 2013, MNRAS, 430, 2585Savchenko, V., Ferrigno C., Kuulkers, E., et al. 2017, ApJL in pressSong, C.-Y., & Liu, T. 2017, arXiv:1710.00142Tanaka, M., & Hotokezaka, K. 2013, ApJ, 775, 113Timmes, F. X., Woosley, S. E., & Weaver, T. A. 1996, ApJ, 457, 834Wu, M.-R., Fernández, R., Martínez-Pinedo, G., & Metzger, B. D. 2016,

MNRAS, 463, 2323Yu, Y.-W. and Zhang, B. and Gao, H., 2013,ApJ,776L,40YZhang, B., 2013,ApJ,763L,22ZZhang, B.-B., Zhang, B., Sun, H., et al., 2017, Arxiv: 1710.05851

Gao, Cao, Ai, and Zhang 2017，ApJL �

Ma,Lei,Gaoetal.2018,ApJL

Black Hole Fallback Accretion Enhance merger-nova

The variation of photon luminosity Lph is shown by the bluesolid line in Figure 6, where Lph is calculated by

· ( )ò p=L R V u dR4 . 11R

R

zph phin

out

It is seen that Lph is in the range of ~ -10 10 erg s50 53 1 for˙- -m0.001 10, which is more than 10orders of magnitude

higher than the Eddington luminosity. The released photons aremainly in the gamma-ray band according to the thermalradiation of the inner disk with < <T10 K 10 K10 11 , as shownby Figure 2. The huge amount of gamma-ray photons escapefrom the optically thick disk through the vertical advectionprocess, which is much more efficient than the diffusionprocess. Such an extremely high photon radiation should haveobservational effects. We will have a discussion on that in thenext section.

4. Conclusions and Discussion

In this work, we have studied the structure and radiation ofhyper-accretion flows around stellar-mass black holes byconsideringthe role of the vertical advection process.Throughthe comparison of our results with the classic NDAFsolutions, we have shown that the density is higher, thetemperature is lower, and the vertical height is thinner in oursolutions. The physical reason is that a large fraction of photonscan escape from the optically thick disk through the verticaladvection process. As a consequence, the neutrino luminosityfrom the disk is decreased. Thus, even without calculating theneutrino annihilation luminosity, we can conclude that theannihilation mechanism cannot be responsible for the long-duration GRBs and X-ray flares.

We would point out that outflows are not taken intoconsideration in the present work. However, outflows arebelieved to generally exist in accretion flows. Recent MHDsimulations have shown that outflows exist both in opticallythin flows (Yuan et al. 2012a, 2012b) and optically thick flows(Jiang et al. 2014; Sadowski & Narayan 2015, 2016). From theobservational view, Wang et al. (2013) reveals that more than99% of the accreted mass escapes from the accretion flow byoutflows in our Galactic center. Based on the energy balanceargument, Gu (2015) shows that the outflow is inevitable forthe accretion flows where the radiative cooling is far below theviscous heating, no matter whetherthe flow is optically thin orthick. Thus, outflows may work as another process to help thetrapped photons to escape (Shen et al. 2015), and will also haveeffects on the structure and neutrino radiation of the accretionflow. Such a mechanism is not included in the current work.

Our calculations are based on the relation l=V cz s withl = 0.1. As mentioned in Section2, the value for λ is adoptedfollowing the simulation results for ˙ =M L c220 Edd

2 (Jianget al. 2014). In our case, the mass accretion rate is higher formore than 10 orders of magnitude. Then, a key question mayexist whether the vertical advection due to the magneticbuoyancy can also work for such hyper-accretion systems. Inour opinion, the radiation pressure is always dominant up to˙ :1 -M M0.1 s 1 or for the outer part of even higher accretionrates. Thus, such a mechanism seems to be an efficient process.On the other hand, even for the case in whichthe parameter λis significantly smaller than 0.1 in the hyper-accretion case,such as several orders of magnitude smaller, the released

gamma-ray photons may still be extremely super-Eddingtonand the potential application is significant.In this work, we have assumed a = 0.02 according to the

simulation results of Hirose et al. (2009). However, othersimulations may provide different values for α. As shown byYuan & Narayan (2014), such a value may be related to themagnitude of net magnetic flux in the simulations. The valuesof α may also have significant effects on the energy transport ofthe vertical advection. As Equations (2) and (6) imply, vR isproportional to α andQadv is proportional to vR and therefore α.Thus, we can expect that, for a larger value of α, the advectivecooling rate Qadv can significantly increase, and therefore thecooling rate due to the vertical advection Qz will decreaseaccording to the energy balance of Equation (4). Nevertheless,the luminosity related to the radial integration of Qz will still behyper-Eddington even though Qz may be lower than Qadv for alarge range of radii.Our results of extremely super-Eddington luminosity of

gamma-ray emission can also be generally applied to shortGRBs. It is commonly believed that short GRBs originate fromthe merger of compact objects, i.e., the black hole–neutron starbinary or the binary with double neutron stars. Moreimportantly, the merger is a significant source of gravitationalwave event. Thus, the released high-energy photons may havesignificant contribution to an electromagnetic counterpart forthe gravitational wave event, such as kilonovae. Kilonovaehave been widely studied in recent years (e.g., Li &Paczyński 1998; Metzger & Berger 2012; Yu et al. 2013;Gao et al. 2015; Kasen et al. 2015; Fernández et al. 2016; Jinet al. 2016; Kawaguchi et al. 2016; Metzger 2016). In our case,a new picture is shown by Figure 7. It is seen from this figurethateither the merger of a black hole and a neutron star or themerger of two neutron stars may result in a gravitational waveevent and a black hole hyper-accretion disk. According to ourstudy in this work, most gamma-ray photons escaping from thedirection perpendicular to the equatorial plane together withthe neutrino annihilation contribute to the thermal fireball, and

Figure 7. Illustration of the association of short GRBs, kilonovae, andgravitational wave events.

5

The Astrophysical Journal, 836:245 (6pp), 2017 February 20 Yi et al.Yi,Gu,Yuan,LiuandMu.2017,ApJ

GW170817: Magnetar as Merger Product Yu & Dai , 2017, ApJ�

Red：Magnetar powered

Blue：r-process

6

0 5 10 15 2039.5

40.0

40.5

41.0

41.5

42.0

Log10!t"days#

Log 10!L bol"ergs!1 #

FIG. 1: Bolometric light curve of a kilonova of hybrid energy sources. The thin dashed and dash-doted lines represent theheating power of radioactivity and NS spin-down, respectively. The thick dashed and dashed-dotted lines are bolometric lightcurves powered by the corresponding single energy source. The solid line is the result of the combination of the two energysources. A uniform intermediate opacity is taken.

!!! !! !! !

!

! !!

!

!!

!!

!

!!!

!

!!

!!!

!!

!!

!

!

!!

!!!!!!

!!!

!

!!! ! !

!

!

!

!

!

!

0 5 10 15 20 25

30

25

20

15

Time since GW 170817 " days

Apparentmagnitude!AB#

FIG. 2: UVOIR light curves of kilonova AT 2017gfo. From top to bottom: (K,Ks)−3, H−2, J−1, y−0, (z, z′)+1, (i, i′I)+2, (r′, r, R) + 3, V + 4, g + 5, (B, g′) + 6, (U,F336W,u) + 7. The solid lines are given by our hybrid energy source modeland the observational data (solid circles) or limits (triangles) are taken from Villar et al. (2017). The ejecta parameters areκ = 1.0 cm2g−1, Mej = 0.035M⊙, vmin = 0.12c, vmax = 0.35c, and δ = 1.5. The NS parameters are tsd = 1.6 × 105s andξLsd(0) = 2.0× 1041erg s−1 for α = 1 (upper thick lines) and tsd = 4.0× 105s and ξLsd(0) = 2.5× 1041erg s−1 for α = 2 (underthin lines). The dashed lines are the results by only considering the radioactive power.

6

0 5 10 15 2039.5

40.0

40.5

41.0

41.5

42.0

Log10!t"days#



!!! !! !! !

!

! !!

!

!!

!!

!

!!!

!

!!

!!!

!!

!!

!

!

!!

!!!!!!

!!!

!

!!! ! !

!

!

!

!

!

!

0 5 10 15 20 25

30

25

20

15

Time since GW 170817 " days

Apparentmagnitude!AB#


6

0 5 10 15 2039.5

40.0

40.5

41.0

41.5

42.0

Log10!t"days#



!!! !! !! !

!

! !!

!

!!

!!

!

!!!

!

!!

!!!

!!

!!

!

!

!!

!!!!!!

!!!

!

!!! ! !

!

!

!

!

!

!

0 5 10 15 20 25

30

25

20

15

Time since GW 170817 " daysApparentmagnitude!AB#


3

spin-down1, as an alternative to the widely-consideredmultiple-opacity model.Our calculations are implemented in the framework of

a simplified radiation-transfer model designed by Met-zger (see Section 4 of Reference [39]), which can wellapproximate the radiation transfer and reveal the influ-ence of mass distribution. For simplicity, a black-bodyspectrum is always adopted. The density distribution ofthe ejecta is taken as a power law as [80]

ρej(r, t) =(3 − δ)Mej

4πr3max

[

(

rmin

rmax

)3−δ

− 1

]−1(

r

rmax

)−δ

,

(3)where Mej is the total mass and rmax and rmin are theradii of the head and bottom of the ejecta, respectively.By considering that the injected energy is obviouslysmaller than the initial kinetic energy of the ejecta, its dy-namic evolution due to adiabatic expansion is neglected.Therefore, the radii can be calculated by rmax = vmaxtand rmin = vmint by introducing the maximum and min-imum velocities. The spin-down power carried by theNS wind consisting of electrons and positrons can evolvewith time as

Lsd(t) = Lsd(0)

(

1 +t

tsd

)−α

, (4)

where the spin-down timescale tsd can be determined ei-ther by magnetic dipole radiation (MDR) or by GW ra-diation, corresponding to a decay index of α = 2 or 1,respectively. In any case, the initial value of Lsd canalways be approximated by the luminosity of MDR as

Lsd(0) =B2

pR6s

6c3

(

2π

Pi

)4

= 9.6×1048R6s,6B

2p,15P

−4i,−3erg s

−1,

(5)where Rs,6 = Rs/106cm, Bp,15 = Bp/1015G, and Pi,−3 =Pi/1ms are the radius, the polar magnetic field strength,and the initial spin period of the NS, respectively, and c isthe speed of light. Finally, as discussed, in our calculationa uniform opacity is considered for the observed ejecta,the value of which is probably not much higher than afew because of the re-ionization of the ejecta by NS windemission.An example bolometric light curve of a kilonova is pre-

sented in Figure 1. While the radioactive power domi-nates the kilonova emission of the first few days, its lateemission would be gradually controlled by the spin-downpower of the remnant NS. The UVOIR light curves of

1 Due to the extra NS energy source, the luminosity of the ejectaemission could in principle vary in a wide range. Therefore, Yuet al. [53] suggested to name the emission of the merger ejectaby a “mergernova”, which can be divided into NS-powered onesand radioactivity-powered ones (i.e. kilonovae). Possible NS-powered mergernova candidates have been previously discussedby Fan et al. [75] and Gao et al. [76].

AT 2017gfo, which are taken from Villar et al. [63], aremodeled by our hybrid energy source scenario in Figure2. The deviation of the model from the data in someindividual filters could somewhat arise from the blackbody simplification of the spectra. The excess of thedata during the first day can be ascribed to the emis-sion from a cooling cocoon as suggested by Kasliwal etal. [16], where the cocoon is formed as a result of thepropagation through and breakout from the ejecta of aGRB jet. In any case, while the mass and velocity of theejecta are generally consistent with the results of pre-vious works [8, 15–17, 22, 63], an intermediate opacityκ = 1.0 cm2g−1 is revealed, which indicates that the pri-mary kilonova emission is neither “blue” nor “red”. Dueto the existence of the remnant NS, such a “purple” kilo-nova can be naturally emitted by the tidal tail ejectathat is lanthanide-rich but deeply ionized. In contrast tothe popular multiple-opacity model, an advantage of oursingle-opacity model is that the observation of a single“purple” kilonova component is insensitive to the open-ing angle of polar ejecta and the viewing angle of ob-servers, because the tidal tail ejecta is observable in alldirections.The NS parameters are degenerated and thus it is dif-

ficult to distinguish the braking mechanisms of the rem-nant NS by the present data. For GW radiation braking,the spin-down timescale can be expressed by

tsd,gw =5P 4

i c5

2048π4GIϵ2= 9.1× 105ϵ−2

−4I−145 P 4

i,−3s (6)

where G is the gravitational constant, I45 = I/1045g cm2

is the moment of inertia, and ϵ is the ellipticity of the NS.Combining Equations 5 and 6 with the parameter valuestsd = 1.6× 105s and ξLsd(0) = 2.0× 1041erg s−1 that areinferred from AT 2017gfo, we can obtain

Bp = 1.4× 1011ξ−1/2R−36 P 2

i,−3G, (7)

and

ϵ = 3.4× 10−4I−1/245 P 2

i,−3. (8)

The derived magnetic field is similar to those of canonicalGalactic pulsars. The parameter ξ, which represents theenergy fraction of the e±-wind of the NS that can be in-jected into the merger ejecta, could be much smaller thanunity, based on the following two reasons. (1) The over-whelming majority of the energy of the e±-wind couldbe collimated into a small cone that points to the GRBjet, and (2) a remarkable fraction of energy could be re-flected back into the wind when the wind emission en-counters with the bottom of the merger ejecta [54]. Fora small ξ, the polar filed strength presented in Equa-tion 7 can be somewhat increased, but cannot be veryhigh. Meanwhile, the relatively high ellipticity presentedin Equation 8 may imply that the internal and proba-bly toroidal magnetic fields of the NS is ultra-strong, ifthe ellipticity primarily induced by the magnetic fields as[81] ϵ ≈ 10−4(Bint/1016G)2. In any case, by considering

3

spin-down1, as an alternative to the widely-consideredmultiple-opacity model.Our calculations are implemented in the framework of

a simplified radiation-transfer model designed by Met-zger (see Section 4 of Reference [39]), which can wellapproximate the radiation transfer and reveal the influ-ence of mass distribution. For simplicity, a black-bodyspectrum is always adopted. The density distribution ofthe ejecta is taken as a power law as [80]

ρej(r, t) =(3 − δ)Mej

4πr3max

[

(

rmin

rmax

)3−δ

− 1

]−1(

r

rmax

)−δ

,

(3)where Mej is the total mass and rmax and rmin are theradii of the head and bottom of the ejecta, respectively.By considering that the injected energy is obviouslysmaller than the initial kinetic energy of the ejecta, its dy-namic evolution due to adiabatic expansion is neglected.Therefore, the radii can be calculated by rmax = vmaxtand rmin = vmint by introducing the maximum and min-imum velocities. The spin-down power carried by theNS wind consisting of electrons and positrons can evolvewith time as

Lsd(t) = Lsd(0)

(

1 +t

tsd

)−α

, (4)

where the spin-down timescale tsd can be determined ei-ther by magnetic dipole radiation (MDR) or by GW ra-diation, corresponding to a decay index of α = 2 or 1,respectively. In any case, the initial value of Lsd canalways be approximated by the luminosity of MDR as

Lsd(0) =B2

pR6s

6c3

(

2π

Pi

)4

= 9.6×1048R6s,6B

2p,15P

−4i,−3erg s

−1,

(5)where Rs,6 = Rs/106cm, Bp,15 = Bp/1015G, and Pi,−3 =Pi/1ms are the radius, the polar magnetic field strength,and the initial spin period of the NS, respectively, and c isthe speed of light. Finally, as discussed, in our calculationa uniform opacity is considered for the observed ejecta,the value of which is probably not much higher than afew because of the re-ionization of the ejecta by NS windemission.An example bolometric light curve of a kilonova is pre-

sented in Figure 1. While the radioactive power domi-nates the kilonova emission of the first few days, its lateemission would be gradually controlled by the spin-downpower of the remnant NS. The UVOIR light curves of

1 Due to the extra NS energy source, the luminosity of the ejectaemission could in principle vary in a wide range. Therefore, Yuet al. [53] suggested to name the emission of the merger ejectaby a “mergernova”, which can be divided into NS-powered onesand radioactivity-powered ones (i.e. kilonovae). Possible NS-powered mergernova candidates have been previously discussedby Fan et al. [75] and Gao et al. [76].

AT 2017gfo, which are taken from Villar et al. [63], aremodeled by our hybrid energy source scenario in Figure2. The deviation of the model from the data in someindividual filters could somewhat arise from the blackbody simplification of the spectra. The excess of thedata during the first day can be ascribed to the emis-sion from a cooling cocoon as suggested by Kasliwal etal. [16], where the cocoon is formed as a result of thepropagation through and breakout from the ejecta of aGRB jet. In any case, while the mass and velocity of theejecta are generally consistent with the results of pre-vious works [8, 15–17, 22, 63], an intermediate opacityκ = 1.0 cm2g−1 is revealed, which indicates that the pri-mary kilonova emission is neither “blue” nor “red”. Dueto the existence of the remnant NS, such a “purple” kilo-nova can be naturally emitted by the tidal tail ejectathat is lanthanide-rich but deeply ionized. In contrast tothe popular multiple-opacity model, an advantage of oursingle-opacity model is that the observation of a single“purple” kilonova component is insensitive to the open-ing angle of polar ejecta and the viewing angle of ob-servers, because the tidal tail ejecta is observable in alldirections.The NS parameters are degenerated and thus it is dif-

ficult to distinguish the braking mechanisms of the rem-nant NS by the present data. For GW radiation braking,the spin-down timescale can be expressed by

tsd,gw =5P 4

i c5

2048π4GIϵ2= 9.1× 105ϵ−2

−4I−145 P 4

i,−3s (6)

where G is the gravitational constant, I45 = I/1045g cm2

is the moment of inertia, and ϵ is the ellipticity of the NS.Combining Equations 5 and 6 with the parameter valuestsd = 1.6× 105s and ξLsd(0) = 2.0× 1041erg s−1 that areinferred from AT 2017gfo, we can obtain

Bp = 1.4× 1011ξ−1/2R−36 P 2

i,−3G, (7)

and

ϵ = 3.4× 10−4I−1/245 P 2

i,−3. (8)

The derived magnetic field is similar to those of canonicalGalactic pulsars. The parameter ξ, which represents theenergy fraction of the e±-wind of the NS that can be in-jected into the merger ejecta, could be much smaller thanunity, based on the following two reasons. (1) The over-whelming majority of the energy of the e±-wind couldbe collimated into a small cone that points to the GRBjet, and (2) a remarkable fraction of energy could be re-flected back into the wind when the wind emission en-counters with the bottom of the merger ejecta [54]. Fora small ξ, the polar filed strength presented in Equa-tion 7 can be somewhat increased, but cannot be veryhigh. Meanwhile, the relatively high ellipticity presentedin Equation 8 may imply that the internal and proba-bly toroidal magnetic fields of the NS is ultra-strong, ifthe ellipticity primarily induced by the magnetic fields as[81] ϵ ≈ 10−4(Bint/1016G)2. In any case, by considering

GW170817: Magnetar as Merger Product Li et al. arXiv:1804.06597

Blue and Red

Magnetar powered

max (1 )TOVM M Pβα= +

GW170817: Allowed Parameter Space of NS Ai, Gao, Dai et al. 2018, ApJ�

Total mass ! Spin Period�

NS should be with millisecond spin �

Post-merger GW signal ! NS Ellipticity�

No Constraint on NS Ellipticity�

Optical data �

Merger-Nova Luminosity �

Ejecta Velocity�

Radio data �

Gamma/X-ray data�

NS should be with small B field

GW170817: Allowed Parameter Space of NS Ai, Gao, Dai et al. 2018, ApJ�

Basic Assumption

Conclusion

�  Part of (or all) SGRBs are from NS-NS mergers �  Cosmological NS-NS systems have the same mass distribution as

the observed Galactic system �  Internal plateau marks the collapse of a magnetar to a BH.

�  Equation of State �  NS-NS merger product:40% BH，30% stable NS，30% NS->BH �  Initial spin period of the magnetar is around 1ms； �  High magnetic field：1015 G �  High X-ray radiation efficiency，>40% �  Large ellipticity, GW radiation dominates spin-down

Mmax = 2.37 M(1+1.58×10−10 P−2.84 )

Magnetar properties from SGRB data �Gao, Zhang and Lv, 2016 PRD �

Merger-novae: no answer Post merger GW detection is essential 2

�� -��

��-��

��-��

��-��

(�/��

)

��-�� (�� )��-�� (�� )��

��

��

�� (��)

FIG. 1. Schematics showing the detector design (left) and the resulting sensitivity (right). The blue curve shows a possible intermediate steptowards the target sensitivity in red. The target is around a factor of 30 below Advanced LIGO design at 3 kHz: a factor of 3 from 10dBsqueezing, 2 from high power, and 5 from detuned signal recycling with the optomechanical filter extending the improvement around thedetune frequency. Sensitivities of LIGO Voyager [29], Einstein Telescope [30], and Cosmic Explorer [31] are shown as references.

Parameters Values

Interf

erome

ter

arm length 4 kmarm cavity power 6.0 MW (1.5MW)test mass 200 kglaser wavelength 2000 nm (1064 nm)temperature 120 K (295 K)SRM transmission 3750 ppmSR detune 1.5 kHzinternal loss: ETM to SRM 800 ppm (2000 ppm)output loss 3% (5%)

Optom

echani

calfilt

er

oscillator mirror mass 5 mgmirror radius, thickness 1.4 mm, 0.35 mmloss angle of substrate, coating 1.0 ⇥ 10�9, 2.0 ⇥ 10�6

suspension quality factor 3.0 ⇥ 106

bare frequency, optical spring 10 Hz, 12 kHzcavity length 4.3 mcavity bandwidth 1.4 kHzbeam radius 0.52 mmresonating power 338 W (180 W)round-trip loss 5 ppm (10 ppm)

Optic

alspri

ng

laser wavelength 1064 nmphotodiode quantum e�ciency � 0.999cavity length 10 cmcavity bandwidth, detune 60 kHz, 0.9 MHzresonating power 680 Wround-trip loss 1 ppmtemperature 16 K

SQZfi

lter

squeezing (observed) 10 dBsqueezing angle 0 radfilter cavity 1 (bandwidth, detune) 4.66 Hz, �42.6 Hzfilter cavity 2 197 Hz, 3409 Hzfilter cavity 3 355 Hz, 1107 Hzfilter cavity 4 510 Hz, �1920 Hz

TABLE I. Parameters of the design. Values in the bracket are thoseused in producing the blue curve in Fig. 1.

at 2000 nm is equivalent to around 3.0 MW at 1064 nm.

Comparing to current advanced detectors, the key designdi↵erence comes from the configuration of the signal recy-cling cavity as shown in Fig. 1. We introduce an internalsignal recycling mirror (iSRM) which forms an impedancematched cavity with the input test mass (ITM) mirror. Theadvantage is that the GW signal is not a↵ected by the narrowbandwidth of the arm cavity. However, optical losses fromthe central beam splitter (BS) and also ITM substrate are res-onantly enhanced, which puts a hard bound on the ultimatedetector sensitivity [36]. The total round-trip loss between theSRM and the end test mass (ETM) mirror should be the orderof 1000 ppm to reach a sensitivity around 5⇥10�25/

pHz with

MW arm cavity power. This goal requires di↵raction loss peroptical surface to be less than 25 ppm for 1064 nm and 7 ppmfor 2000 nm wavelength. Also, reflectivity of the BS and ITManti-reflective coatings should be less than 10 ppm. TakingAdvanced LIGO as an example and with these parameterswhich are achievable using existing technology, we set ITMtransmission to 0.05 to limit the total loss, which however in-creases the optical load on BS by a factor of 3 compared to theAdvanced LIGO design. This number can be reduced in thefuture, once optical surfaces with higher quality are available.

In addition to iSRM, the optomechanical filter module isadded to the signal recycling cavity. This module compen-sates for the phase lag acquired by signal sidebands in thefree space interferometer, which results in a broadband reso-nance of the signal [25]. The key component of the system isa high quality mechanical oscillator embedded in an opticalcavity, which is pumped by an external filter laser with properfrequency and amplitude. We propose to implement the me-chanical oscillator by comprising a low loss quasi-monolithicsuspension [37] and a milligram-scale mirror. In order forthe phase compensation work properly in kHz regime, we in-crease the oscillator frequency to 12 kHz by creating an op-tical spring with an auxiliary laser and optical cavity, which

Miao et al. , 2017�

Higher luminosity merger-novae

Merger-nova：r−process or Magnetar �

BH: 2% <25 mag Magnetar: 46.2% <25mag

delay distribution models to construct several redshiftdistributions that correspond to the three delay models withthe best-fit parameters. First, we randomly generate 10,000compact star binary systems with a redshift distributiontracking the SFH following the model of Yüksel et al. (2008).Next, we randomly generate the merger delay timescales ofall these systems based on the three merger delay timescalemodels listed in Table 2. For each model, we derive thelookback time of SGRBs by subtracting the merger delaytime from the formation time, and we transfer the lookbacktime to redshift. We repeat the process 10,000 times (eachwith 10,000 events simulated). By averaging the results,we are able to derive the average redshift distribution ofthe simulated samples. We fit the derived redshift distribution(for all three merger delay models) using multiple-PLfunctions and derive an empirical expression of f(z) foreach model. The simulated results with best-fit empiricalmodels are shown in Figure 1. The distributions arenormalized to unity at the local universe (z= 0). Theempirical formulae of f(z) for the three merger delay modelsare as follows.

For the Gaussian delay model (Virgili et al. 2011b), one has

f z zz

z z

110.17

14.12

14.05

, 20

SGRBG 5.0

0.87

8.0 20.51

( ) ( )

( )

= + ++

++

++

hh

h h- - h

⎜ ⎟

⎜ ⎟ ⎜ ⎟

⎡⎣⎢

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎤⎦⎥

with η = −2, which is roughly a BPL with redshift breaks atz1 = 0.45, z2 = 2.0, andz3 = 3.0.For the lognormal delay model (Wanderman & Piran 2014),

one has

f z zz

z z

110.36

13.3

13.3

, 21

SGRBLN 5.7

1.3

9.5 24.51

( ) ( )

( )

= + ++

++

++

hh

h h- - h

⎜ ⎟

⎜ ⎟ ⎜ ⎟

⎡⎣⎢

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎤⎦⎥

with η = −2, which is roughly a BPL with redshift breaks atz1 = 0.35, z2 = 1.5, andz3 = 2.3.For the PL model (Wanderman & Piran 2014), one has

f z zz

z z

11

2.5

13.8

17.7

, 22

SGRBPL 1.9

1.2

4.4 111

( ) ( )

( )

= + ++

++

++

hh

h h

-

- - h

⎜ ⎟

⎜ ⎟ ⎜ ⎟

⎡⎣⎢

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎛⎝

⎞⎠

⎤⎦⎥

with η = −2.6. This model has a wider redshift distributioncompared to the first two models (owing to the wide range ofthe merger delay time). It is roughly a BPL with redshift breaksat z1 = 0.42, z2 = 3.4, andz3 = 11.3. The SGRB data do not

Table 2Best-fit Merger Delay Models of SGRBs with Respect to Star Formation History

Delay Model Formula Best-fit Parameters Reference

Gaussian (G) m d dexp 2tG 2 t,G

d,G 2

t,G2( ) ( )t t ps t= - t

s-⎜ ⎟⎛

⎝⎞⎠

tt,G = 2 Gyr, σt,G = 0.3 (1)

Lognormal (LN) m d dln exp 2 lntLN

ln ln

2 t,LNd,LN 2

t,LN2 ( )( ) ( )t t ps t= - t

s-⎜ ⎟⎛

⎝⎞⎠

tt,LN = 2.9 Gyr, σt,LN = 0.2 (2)

Power law (PL) g d dPLt( )t t t t= a- αt = 0.81 (2)

References. (1) Virgili et al. 2011b; (2) Wanderman & Piran 2014.

Figure 1. Redshift distribution derived from Monte Carlo simulations for short GRBs considering three delay time models with respect to star formation history:Gaussian (black), lognormal (blue), and PL (red). For each model, the result is derived from the average of 10,000 simulations, each with simulated 10,000 systems.Dots are the simulated results, and the curve is the empirical multiple-PL fits given in Equations (20)–(22).

6

The Astrophysical Journal, 812:33 (18pp), 2015 October 10 Sun, Zhang, & Li

Ai et al. 2018, in prep. �

r-process Merger-nova Candidate�Tanvir et al. Nature 2013�Yang et al. 2015, NatCo �Jin et al. 2016, NatCo �

Long-duration (42 s) g-ray bursts (GRBs) are believed tooriginate from Collapsars that involve death of massive starsand are expected to be accompanied by luminous super-

novae (SNe). GRB 060614 was a nearby burst with a duration of102 s at a redshift of 0.125(ref. 1). While it is classified as a longburst according to its duration, extensive searches did notfind any SNe-like emission down to limits hundreds of timesfainter2–4 than SN 1998bw, the archetypal hypernova thataccompanied long GRBs5. Moreover, the temporal lag and peakluminosity of GRB 060614 fell entirely within the short durationsubclass and the properties of the host galaxy distinguish it fromother long-duration GRB hosts. Thus, GRB 060614 did not fitinto the standard picture in which long-duration GRBs arise fromthe collapse of massive stars while short ones arise from compactbinary mergers. It was nicknamed the ‘long–short burst’ as itsorigin was unclear. Some speculated that it originated fromcompact binary merger and thus it is intrinsically a ‘short’GRB1,4,6–8. Others proposed that it was formed in a new type of aCollapsar which produces an energetic g-ray burst that is notaccompanied by an SNe2–4.

Two recent developments may shed a new light on the origin ofthis object. The first is the detection of a few very weak SNe (forexample, SN 2008ha9) with peak bolometric luminosities as lowas LB1041 erg s! 1. The second is the detection of an infraredbump, again with a LB1041 erg s! 1, in the late afterglow of theshort burst GRB 130603B10,11. This was interpreted as aLi-Paczynski macronova (also called kilonova)12–19—a near-infrared/optical transient powered by the radioactive decay ofheavy elements synthesized in the ejecta of a compact binarymerger. Motivated by these discoveries, we re-examined theafterglow data of this peculiar burst searching for a signalcharacteristic to one of these events.

The X-ray and UV/optical afterglow data of GRB 060614, wereextensively examined in the literature20,21 and found to followvery well the fireball afterglow model up to tB20 days22. TheJ-band has been disregarded because only upper limitsB19–20th mag with a sizeable scatter are available at t 42.7days, and these are too bright to significantly constrain evensupernovae as luminous as SN 1998bw23. In this work we focuson the optical emission. We have re-analysed all the late time(that is, t Z1.7 days) very large telescope (VLT) V, R and I -bandarchival data and the Hubble space telescope (HST) F606W andF814W archival data, including those reported in the literature3,4

and several unpublished data points. Details on data reductionare given in the Methods.

ResultsThe discovery of a significant F814W-band excess. Figure 1depicts the most complete late-time optical light curves(see Supplementary Table 1; the late VLT upper limits are notshown in Fig. 1) of this burst. The VLT V, R and I-band fluxesdecrease with time as pt! 2.30±0.03 (see Fig. 1, in which the VLTV/I band data have been calibrated to the F606W/F814W filtersof HST with proper k-corrections), consistent with that foundearlier3,20,21. However, the first HST F814W data point issignificantly above the same extrapolated power-law decline.The significance of the deviation is B6s (see the estimate in theMethods). No statistically significant excess is present in both theF606W and the R bands. The F814W-band excess is made mostforcibly by considering the colour evolution of the transient,defined as the difference between the magnitudes in each filter,which evolves from V–IE0.65 mag by the VLT (correspondinglyfor HST we have F606W–F814WE0.55 mag) at about t B1.7days to F606W–F814WE1.5 mag by HST at about 13.6 daysafter the trigger of the burst. With proper/minor extinction

corrections, the optical to X-ray spectrum energy distribution forGRB 060614 at the epoch of B1.9 days is nicely fitted by a singlepower law3,20,21 Fvpv! 0.8. In the standard external forwardshock afterglow model, the cooling frequency is expected to dropwith time as22 vcpt! 1/2. Thus, it cannot change the opticalspectrum in the time interval of 1.9–13.6 days. Hence, theremarkable colour change and the F814W-band excess ofB1 mag suggest a new component. Like in GRB 130603B thiscomponent was observed at one epoch only. After the subtractionof the power-law decay component, the flux of the excesscomponent decreased with time faster than t! 3.2 for t 413.6days. Note that an unexpected optical re-brightening was alsodetected in GRB080503, another ‘long–short’ burst24. However,unlike the excess component identified here, that re-brighteningwas achromatic in optical to X-ray bands and therefore likelyoriginated by a different process.

DiscussionShortly after the discovery of GRB 060614 it was speculated that itis powered by an ‘unusual’ core collapse of a massive star2,3.We turn now to explore whether the F814W-band excesscan be powered by a weak supernova. Figure 2 depicts the

20

22

24

26

28

30

32

34

Mag

nitu

des

(Veg

a)

F814WR+3

F606W+5SN2008ha F814W (ref. 26)

F814W-band excess

–1 0 1

–1 0 1

Res

idua

ls

–1 0 1

105 106

t (s)

Figure 1 | The afterglow emission of GRB 060614. The VLT and HSTobservation vega magnitudes including their 1s statistical errors of thephoton noise and the sky variance and the 3s upper limits (the downwardarrows) are adopted from Supplementary Table 1. The small amounts offoreground and host extinction have not been corrected. Note that the VLTV/I band data have been calibrated to the HST F606W/F814W filters withproper k-corrections (see Methods). The VLT data (the circles) arecanonical fireball afterglow emission while the HST F814W detection(marked in the square) at tB13.6 days is significantly in excess of the sameextrapolated power-law decline (see the residual), which is at odds with theafterglow model. The F814W-band light curve of SN 2008ha 27 expected atz¼0.125 is also presented for comparison. The dashed lines are macronovamodel light curves generated from numerical simulation 28 for the ejectafrom a black hole–neutron star merger. Error bars represent s.e.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8323

2 NATURE COMMUNICATIONS | 6:7323 | DOI: 10.1038/ncomms8323 | www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

Criteria

GRB060614�

GRB050907�

GRB130603B�

All nearby short GRBs have been explored. With late time optical/IR observation. Deviation from afterglow model. �

this field. The redshifts of the afterglow21 and the host galaxy22 wereboth found to be z 5 0.356.

Another proposed signature of the merger of two neutron stars or aneutron star and a black hole is the production of a kilonova (some-times also termed a ‘macronova’ or an ‘r-process supernova’) due tothe decay of radioactive species produced and initially ejected duringthe merger process—in other words, an event similar to a faint, short-lived supernova6–8. Detailed calculations suggest that the spectra ofsuch kilonova sources will be determined by the heavy r-process ionscreated in the neutron-rich material. Although these models10–13 arestill far from being fully realistic, a robust conclusion is that the opticalflux will be greatly diminished by line blanketing in the rapidly expan-ding ejecta, with the radiation emerging instead in the near-infrared(NIR) and being produced over a longer timescale than would other-wise be the case. This makes previous limits on early optical kilonovaemission unsurprising23. Specifically, the NIR light curves are expectedto have a broad peak, rising after a few days and lasting a week or morein the rest frame. The relatively modest redshift and intensive study ofGRB 130603B made it a prime candidate for searching for such a kilonova.

We imaged of the location of the burst with the NASA/ESA HubbleSpace Telescope (HST) at two epochs, the first ,9 d after the burst(epoch 1) and the second ,30 d after the burst (epoch 2). On each occa-sion, a single orbit integration was obtained in both the optical F606Wfilter (0.6mm) and the NIR F160W filter (1.6mm) (full details of the imag-ing and photometric analysis discussed here are given in Supplemen-tary Information). The HST images are shown in Fig. 1; the key result isseen in the difference frames (right-hand panels), which provide clearevidence for a compact transient source in the NIR in epoch 1 (we notethat this source was also identified24 as a candidate kilonova in indepen-dent analysis of our data on epoch 1) that seems to have disappeared byepoch 2 and is absent to the depth of the data in the optical.

At the position of the SGRB in the difference images, our photo-metric analysis gives a magnitude limit in the F606W filter ofR606,AB . 28.25 mag (2s upper limit) and a magnitude in the F160Wfilter of H160,AB 5 25.73 6 0.20 mag. In both cases, we fitted a modelpoint-spread function and estimated the errors from the variance ofthe flux at a large number of locations chosen to have a similar back-ground to that at the position of the SGRB. We note that some tran-sient emission may remain in the second NIR epoch; experimentingwith adding synthetic stars to the image leads us to conclude that anysuch late-time emission is likely to be less than ,25% of the level inepoch 1 if it is not to appear visually as a faint point source in epoch 2,however, that would still allow the NIR magnitude in epoch 1 to be upto ,0.3 mag brighter.

To assess the significance of this result, it is important to establishwhether any emission seen in the first HST epoch could have a con-tribution from the SGRB afterglow. A compilation of optical and NIRphotometry, gathered by a variety of ground-based telescopes in thefew days following the burst, is plotted in Fig. 2 along with our HSTresults. Although initially bright, the optical afterglow light curve dec-lines steeply after about ,10 h, requiring a late-time power-law decayrate of a < 2.7 (where F / t2a describes the flux). The NIR flux, on theother hand, is significantly in excess of the same extrapolated powerlaw. This point is made most forcibly by considering the colour evolu-tion of the transient, defined as the difference between the magnitudesin each filter, which evolves from R606 2 H160 < 1.7 6 0.15 mag at about14 h to greater than R606 2 H160 < 2.5 mag at about 9 d. It would bevery unusual, and in conflict with predictions of the standard external-shock theory25, for such a large colour change to be a consequence oflate-time afterglow behaviour. The most natural explanation is there-fore that the HST transient source is largely due to kilonova emission,and the brightness is in fact well within the range of recent modelsplotted in Fig. 2, thus supporting the proposition that kilonovae arelikely to be important sites of r-process element production. We notethat this phenomenon is strikingly reminiscent, in a qualitative sense,of the humps in the optical light curves of long-duration c-ray bursts

produced by underlying type Ic supernovae, although here the lumino-sity is considerably fainter and the emission is redder. The ubiquity andrange of properties of the late-time red transient emission in SGRBswill undoubtedly be tested by future observations.

The next generation of gravitational-wave detectors (Advanced LIGOand Advanced VIRGO) is expected ultimately to reach sensitivity levelsallowing them to detect neutron-star/neutron-star and neutron-star/black-hole inspirals out to distances of a few hundred megaparsecs26

(z < 0.05–0.1). However, no SGRB has been definitively found at anyredshift less than z 5 0.12 over the 8.5 yr of the Swift mission to date27.This suggests either that the rate of compact binary mergers is low,implying a correspondingly low expected rate of gravitational-wavetransient detections, or that most such mergers are not observed asbright SGRBs. The latter case could be understood if the beaming ofSGRBs was rather narrow, for example, and the intrinsic event rate was,as a result, two or three orders of magnitude higher than that observedby Swift. Although the evidence constraining SGRB jet opening anglesis limited at present28 (indeed, the light-curve break seen in GRB 130603Bmay be further evidence for such beaming), it is clear that an alterna-tive electromagnetic signature, particularly if approximately isotropic,

21

1 10

X-ray

F606W

F160W22

23

AB

mag

nitu

de

24

25

26

27

104 105

10–11

10–12

10–13

10–14

106

28

29

Time since GRB 130603B (s)

Time since GRB 130603B (d)

X-ray flux (erg s–1 cm

–2)

Figure 2 | Optical, NIR and X-ray light curves of GRB 130603B. Left axis,optical and NIR; right axis, X-ray. Upper limits are 2s and error bars are 1s. Theoptical data (g, r and i bands) have been interpolated to the F606W band andthe NIR data have been interpolated to the F160W band using an averagespectral energy distribution at ,0.6 d (Supplementary Information). HSTepoch-1 points are given by bold symbols. The optical afterglow decays steeplyafter the first ,0.3 d and is modelled here as a smoothly broken power law(dashed blue line). We note that the complete absence of late-time opticalemission also places a limit on any separate 56Ni-driven decay component. The0.3–10-keV X-ray data29 are also consistent with breaking to a similarly steepdecay (the dashed black line shows the optical light curve simply rescaled tomatch the X-ray points in this time frame), although the source had droppedbelow Swift sensitivity by ,48 h after the burst. The key conclusion from thisplot is that the source seen in the NIR requires an additional component abovethe extrapolation of the afterglow (red dashed line), assuming that it also decaysat the same rate. This excess NIR flux corresponds to a source with absolutemagnitude M(J)AB < 215.35 mag at ,7 d after the burst in the rest frame. Thisis consistent with the favoured range of kilonova behaviour from recentcalculations (despite their known significant uncertainties11–13), as illustrated bythe model11 lines (orange curves correspond to ejected masses of 1022 solarmasses (lower curve) and 1021 solar masses (upper curve), and these are addedto the afterglow decay curves to produce predictions for the total NIR emission,shown as solid red curves). The cyan curve shows that even the brightestpredicted r-process kilonova optical emission is negligible.

RESEARCH LETTER

5 4 8 | N A T U R E | V O L 5 0 0 | 2 9 A U G U S T 2 0 1 3

Macmillan Publishers Limited. All rights reserved©2013

Gao et al. , 2017, ApJ， 837，50 �

Magnetar Merger-nova Candidate �Gao et al. , 2015, ApJ, 807, 163�

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 061006SGRBs with extended emission or internal plateau, which may signify the presence of magnetars as the central engine.�With standard parameter values, the magnetar remnant scenario could well interpret the multi-band data of three bursts, including the extended emission and their late chromatic features in the optical and X-ray data.�

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 050724

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 070714B

Gao et al. , 2017, ApJ， 837，50 �

Magnetar Merger-nova Candidate �Gao et al. , 2015, ApJ, 807, 163�

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 061006SGRBs with extended emission or internal plateau, which may signify the presence of magnetars as the central engine.�With standard parameter values, the magnetar remnant scenario could well interpret the multi-band data of three bursts, including the extended emission and their late chromatic features in the optical and X-ray data.�

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 050724

101 102 103 104 105 106 10710−8

10−6

10−4

10−2

100

102

104

106

Time (s)

F ν(µ

Jy)

γ -ray

X-ray

Opt

GRB 070714B

Time (s)101 102 103 104 105 106 107

F ν(µ

Jy)

10-8

10-6

10-4

10-2

100

102

104

106

γ-ray

X-ray

Opt

GRB 050724

6

36

37

38

39

40

41

42

43

44

45

log 10(L peak)

Nova

Kilo-Nova

Super-Nova

Super-Luminous Super-Nova

050709

050724

060614

061006070714B

130603B

FIG. 2.— Peak luminosity for all claimed “kilo-novae" and magnetar-powered merger-novae.

merger-novae in the future.It is interesting to compare the properties of magnetar-

powered merger-novae and the r-process powered merger-novae claimed in the literature. In Figure 2, we present thepeak luminosities of all claimed cases, compared with the typ-ical luminosities of novae, supernovae, and super-luminoussupernovae. One can see that the three r-process poweredmerger-novae associated with GRB 050709, GRB 060614,and GRB 130603B indeed have peak luminosities about 1000times of that of a typical nova. The three magnetar-poweredmerger-novae claimed in this paper, on the other hand, are sys-tematically brighter by more than one order of magnitude, sothat the term “kilo-nova” cannot catch the properties of theseevents. The two populations are clearly separated from eachother. More late-time follow-up observations of short GRBsare needed to quantify the fraction of NS-NS mergers with amagnetar merger product.

To interpret the data of GRB 050724, we need a rela-tively large value of the initial spin period of the magnetarPi = 5 ms. In this case, the total energy budget of the mag-netar power is relatively small, ∼ 1051 ergs. This may bedue to gravitational wave radiation loss during the mergerprocess (Radice et al. 2016) or after the merger due to thelarge deformation of the magnetar (Fan et al. 2013b; Gao et al.2016; Lasky & Glampedakis 2016). Most recently, Fong et al.

(2016) studied the long-term radio behavior of GRB 050724with the Very Large Array, and placed a stringent limit ofEmax ≈ (2 − 5)× 1051 erg on the rotational energy of a sta-ble magnetar. This is consistent with our results. However,we notice that the limit in Fong et al. (2016) is placed by as-suming ϵB = 0.1, a relatively extreme value for ϵB in GRB af-terglow modeling (Kumar & Zhang 2015; Wang et al. 2015).We adopt ϵB = 0.001 to interpret the afterglow data of GRB050724, in this case, the constraint on the rotational energywould become much looser, e.g., Emax could be larger than1052 erg (Fong et al. 2016).

Taking into account GRB 080503, we now have 4 candi-dates of magnetar-powered merger-nova. Among the sample,GRB 080503 and GRB 050724 show late re-brightening fea-ture in the X-ray band, indicating a stable magnetar (at leaststable up to 105 s) as the central engine. For GRB 070714Band GRB 061006, the supra-massive NSs seem to have col-lapsed to black holes before their surrounding ejecta becometransparent (collapse before 105 s). Although the sample issmall, the ratio between stable magnetars and supra-massivemagnetars is roughly 1 : 1, which is consistent with the re-sults predicted in Gao et al. (2016), where a neutron star EoSwith a maximum mass close to a parameterization of Mmax =2.37M⊙(1 + 1.58× 10−10P−2.84) is adopted. A larger sampleof magnetar-powered merger-novae in the future could givemore stringent constraints on the EoS for neutron matter.

With the current sample, some simple statistics may be ob-tained. For instance, for the central magnetar, the values ofinitial spin period spans from 2 ms to 5 ms, and the dipolarmagnetic field of strength span from 5×1015 G to 1016 G. Themass of the ejecta material spans from 10−3 M⊙ to 10−2 M⊙.A larger sample in the future would increase the statisticsand shed light into the detailed properties of the binary NSmerger products, both the central magnetar and the surround-ing ejecta.

This work is supported by the National Basic Research Pro-gram (973 Program) of China (Grant No. 2014CB845800),and the National Natural Science Foundation of China un-der Grant No. 11543005. L. H. J. acknowledges supportby the Scientific Research Foundation of Guangxi University(Grant No. XGZ150299) and One-Hundred-Talents Programof Guangxi colleges.

REFERENCES

Abbott, B. P., Abbott, R., Adhikari, R., et al. 2009, Reports on Progress inPhysics, 72, 076901

Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2016b, Physical Review Letters,116, 241103

Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2016a, Physical Review Letters,116, 061102

Acernese, F., Alshourbagy, M., Amico, P., et al. 2008, Classical and QuantumGravity, 25, 114045

Barbier, L., Barthelmy, S. D., Cummings, J., et al. 2007, GRB CoordinatesNetwork, 6623, 1

Barnes, J., & Kasen, D. 2013, ApJ, 775, 18Barthelmy, S. D., Chincarini, G., Burrows, D. N., et al. 2005, Nature, 438,

994Bauswein, A., Goriely, S., & Janka, H.-T. 2013, ApJ, 773, 78Berger, E., Price, P. A., Cenko, S. B., et al. 2005, Nature, 438, 988Berger, E., Fox, D. B., Price, P. A., et al. 2007, ApJ, 664, 1000Berger, E. 2011, New A Rev., 55, 1Berger, E., Fong, W., & Chornock, R. 2013, ApJ, 774, L23Burenin, R., Pavlinsky, M., Sunyaev, R., et al. 2005, GRB Coordinates

Network, 3671, 1

Burrows, D. N., Grupe, D., Capalbi, M., et al. 2006, ApJ, 653, 468Campana, S., Tagliaferri, G., Lazzati, D., et al. 2006, A&A, 454, 113Cenko, S. B., Berger, E., Nakar, E., et al. 2008, arXiv:0802.0874Chu, Q., Howell, E. J., Rowlinson, A., et al. 2016, MNRAS, 459, 121Connaughton, V., Burns, E., Goldstein, A., et al. 2016, arXiv:1602.03920Covino, S., Antonelli, L. A., Romano, P., et al. 2005, GRB Coordinates

Network, 3665, 1Dai, Z. G., Wang, X. Y., Wu, X. F., & Zhang, B. 2006, Science, 311, 1127D’Avanzo, P., Covino, S., Antonelli, L. A., et al. 2005, GRB Coordinates

Network, 3690, 1De Pasquale, M., Schady, P., Kuin, N. P. M., et al. 2010, ApJ, 709, L146Dessart, L., Ott, C. D., Burrows, A., Rosswog, S., & Livne, E. 2009, ApJ,

690, 1681Eichler, D., Livio, M., Piran, T., & Schramm, D. N. 1989, Nature, 340, 126Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2009, MNRAS, 397, 1177Fan, Y.-Z., & Xu, D. 2006, MNRAS, 372, L19Fan, Y.-Z., Yu, Y.-W., Xu, D., et al. 2013a, ApJ, 779, L25Fan, Y.-Z., Wu, X.-F., & Wei, D.-M. 2013b, Phys. Rev. D, 88, 067304Fernández, R., & Metzger, B. D. 2013, MNRAS, 435, 502Fong, W.-f., Metzger, B. D., Berger, E., & Ozel, F. 2016, arXiv:1607.00416

Taking into account GRB 080503, we now have 4 candidates of magnetar-powered merger-nova. GRB 080503 and GRB 050724 indicating a stable NS; GRB 070714B and GRB 061006, indicating a supra-massive NS. 3 candidates of r-process powered merger-nova have been claimed. The ratio of BH, stable NS and supra-massive NS is roughly 1:1:1, which is consistent with previous results.�

Magnetar-powered merger-novae are systematically brighter. We propose to call r-process powered merger-nova.�

Gao et al. , 2017 ApJ, 837, 50 �

Merger-nova：r−process or Magnetar �

� Era of GW Astronomy has been Opened � Merger product for GW170817: BH or NS?�� More GW170817-like multi-messenger

evens to test the post merger product.�� Late time follow up observation of Short

GRB data is important.

Summary

Thanks for the attention！

merger product for gw170817: bh or ns?

Documents