Rings of Metal-Rich Ejecta in Puppis A: Hints of a SN Interaction in a Binary?

Parviz Ghavamian (Towson University), Ivo Seitenzahl (UNSW Canberra), M. A. Dopita (ANU), Frédéric Vogt (ESO), Ashley Ruiter (UNSW), Janette Suherli (ESO)

M. A. Dopita (1946-2018) Puppis A is Nested Within the Vela SNR

Ha [O III] 3º

Metal-Rich ejecta

XMM 0.3-0.7 keV (O VII+OVIII) Chandra 0.7-1.0 keV (Ne, Mg, Si) Chandra 1.0-8.0 keV (S, Ar) Vela: D = 250 pc ROSAT HRI (Aschenbach 1995) (Cha et al. 1995) Puppis A is a Middle-Aged SNR

neutron star (CCO) J0822-4300 -1 Metal-Rich ejecta (v ~ 1600 km s ) (v ~ - 1600 km s-1) (Becker & Winkler 2012) (Katsuda et al. 2010)

Bright Eastern Cloud (CSM’ISM interaction) (Katsuda et al. 2010) (XMM/Chandra) 0.5 - 0.7 keV (O VII+OVIII Kα) 0.7 - 1.2 keV (Ne Kα) 1.2 - 5.0 keV

• Age ~ 3700 yrs (optical; Winkler & Kirshner 1988) - 4400 yrs (X-rays) • Strong interaction w/H I and CO clouds on eastern side (Dubner 1988; Reynoso et al 1995; Blair et al. 1995) • R ~ 25ʹ (~ 16 pc assuming D = 2.2 kpc) • Si-rich (Hwang et al. 2008) and O/Ne/Mg-rich X-ray ejecta (Katsuda et al. 2008; 2010)

• Progenitor mass ~ 25 M⊙ inferred from abundance ratios (Katsuda et al. 2010) Imagery:P. F. Winkler P. F. Winkler (CTIO (CTIO Schmidt) Schmidt)

Ha elevated [NII] / Hα [O III] Baade & Minkowski (1954) Dopita et al 1977 O-rich radiative (relic RGB wind?) shocks (W+K 85, 88) v ~ 100 km s-1 Vela SNR shocks

Vela SNR shocks Imagery:P. F. Winkler P. F. Winkler (CTIO (CTIO Schmidt) Schmidt)

Ha elevated [NII] / Hα [O III] Baade & Minkowski (1954) Dopita et al 1977 O-rich radiative (relic RGB wind?) shocks (W+K 85, 88) v ~ 100 km s-1 Vela SNR shocks

Vela SNR shocks Imagery:P. F. Winkler P. F. Winkler (CTIO (CTIO Schmidt) Schmidt)

Ha elevated [NII] / Hα proper motion vectors over [O III] Baade & Minkowski (1954) Dopita1000 et al yrs 1977 (Winkler 1988) O-rich radiative (relic RGB wind?) shocks (W+K 85, 88) v ~ 100 km s-1 Vela SNR shocks

Vela SNR shocks Imagery:P. F. Winkler P. F. Winkler (CTIO (CTIO Schmidt) Schmidt)

Vela SNR shocks

‘the Swirl’ (Winkler et al. 1989) Imagery:P. F. Winkler P. F. Winkler (CTIO (CTIO Schmidt) Schmidt)

0.95ʹ × 0.92ʹ Vela SNR (0.6shocks pc × 0.58 pc)

1.35ʹ × 1.46ʹ NS ‘the Swirl’ (0.86 pc × 0.93 pc) (Winkler et al. 1989) WiFeS Fields in Puppis A

29 Fields observed (2×1800 s per field) ‣ B3000 (blue channel) + R3000 (red channel) ‣ Coverage 3500-9500 Å ‣ 100 km s-1 resolution ‣ Surface brightness limit ~ 10-17 ergs cm-2 s-1 arcsec-2 ‣ 10 nights (2016/2017) Stepping Through the Red Channel Datacube of the Swirl Stepping Through the Red Channel Datacube of the Swirl Background Subraction is Tricky for Some Lines

Background/ejecta line blending is a challenge in the Swirl data:

• Some ejecta Hα (vr ~-780 km s-1) blueshifted to background [N II] 6548

• Some ejecta [N II] 6583 (vr ~-1350 km s-1) blueshifted to background [N II] 6548 … this necessitates careful background subtraction in many locations

• Automated line-profile fitting to 1065 spectra the swirl in [N II] 6548 (vr ~ -1350 km s-1)

[N II] [N II] Hα [N II]/Hα ~ 10 (!) (ejecta) (ejecta) (sky) n(N)/n(H) ~ 150× solar [N II] [N II] [N II] (ejecta) (sky) (ejecta) SKY [N II] SUBTRACTION (sky) Hα (ejecta) Hα (ejecta) Another Example

Ejecta [S II] 6731 (vr ~ -780 km s-1) blueshifted to background [S II] 6716

the swirl in [S II] 6716 (vr ~ -1350 km s-1)

[S II] (6716 sky)+(6731 ejecta)

SKY [S II] SUBTRACTION (ejecta) [S II] [S II] (6716 ejecta) [S II] (ejecta) (sky 6731)

[S II] 6716/6731 ~ 1.1 ne ~ 500-1000 cm-3 [N II] Radial Velocities of the Swirl from WiFeS

-920 km s-1

- 1120 km s-1

- 1350 km s-1

dot radii ∝ line FWHM [S II] Radial Velocities of the Swirl from WiFeS

- 620 km s-1

- 1350 km s-1

dot radii ∝ line FWHM [O III] Radial Velocities of the Swirl from WiFeS

- 400 km s-1

- 750 km s-1

dot radii ∝ line FWHM Why the Rings Are So Unusual

‣ Their velocities seem quantized. Are they even associated with Puppis A? Expansion speeds, central location in Puppis A suggest yes

‣ Swirl shows only blue shifted emission, consistent with a binary explosion picture (secondary located between us and the primary during SN)

‣ Dynamical timescale of knots: typical radiatively shocked clump ~ 1017 cm (limited by seeing), gives survival time � ~ 100-150 yrs

… Implies: what we’re seeing has been recently shocked! \

‣ Rings similar in abundances to FMFs in Cas A (Fesen et al. 1987), though probably slower than the main body of ejecta The Astrophysical Journal,792:66(15pp),2014September1 Hirai, Sawai, & Yamada

(a): (d):

The AstrophysicalNearly Journal,792:66(15pp),2014September1 70% of all Massive Star SystemsHirai, are Sawai, &Multiple! Yamada 4 6 (a): t=1.22x10 s (d): t=1.14x10 s Numerical Simulation of a 10 M⊙ Binary Interaction

The Astrophysical Journal,792:66(15pp),2014September1 Hirai, Sawai, & Yamada ejecta ‘shadow’

(a): (b):(d): (e):

The Astrophysical Journal,792:66(15pp),2014September1 Hirai, Sawai, & Yamada

4 6 (a): t=1.22x10 s (d): t=1.14x10 s

5 6 (b): t=5.43x10 s (e): t=3.61x10 s

(b): (c):(e): (f):

The Astrophysical Journal,792:66(15pp),2014September1 Hirai, Sawai, & Yamada

4 5 6 6 (a): t=1.22x10 s (b): t=5.43x10(d): t=1.14x10s s (e): t=3.61x10 s

5 6 (c): t=1.78x10 s (f): t=7.95x10 s Hirai et al. (2014)

(c): (f): Figure 6. Density profiles at Step 2. The time elapsed since SN explosion is indicated for each panel. The radius of each circle is 8 1013 cm. The color bar is in log scale. × (A color version of this figure is available in the online journal.) 5 (f): 6 5 (c): t=1.78x10 s 6 t=7.95x10 s (b): t=5.43x10 s (e): t=3.61x10 s with the same physical setup but with different spatial resolu- We hence believe that the resolution employed in this study is tions. The number of grid points was changed separately for r appropriate. and θ directions. They are summarized in Table 1. 4 4 102km/s The dynamics2×10 km/s of these test runs are compared× with the 3.2.4. Separation Dependence Figure 7. Velocity vectorsbase at Step model 2. Note in that Figure red arrows10.Theflowpatternshardlyshow on the right panels have lengths amplified by a factor ofIn 50 the compared base with model, blue arrows the binary on the left separation panels. is assumed to be (A color version of this figure is available in the online journal.) any difference for the models with different grids in the r a amin.Thisisthesmallestpossiblevalueandismost direction. On the other hand, as the grid becomes coarser in likely= to be too small. We have hence performed the same these effects but thethe lastθ onedirection, will tend the to shock decrease waves the amount are slightly of broadened,and the while total removedsimulation mass is for smaller. several As models the forward varying shock the separation as a = the removed mass. Thecomplex smaller flow curvature patterns of particularly the SNE may behind lead the companionreaches the star opposite(1 side.1, 1. of2,..., the companion2.5) amin starto see (panel the (c) dependence of of removed are changed even qualitatively. The removed mass, which is × to smaller deflections by the bow shock and contribute to some Figures 11 and 12),mass much13 on smaller separation. amounts As the of binary matter becomes become wider, the density of Figure 6. Density profiles at Step 2. The time elapsedour mainsince SN concern explosion in is thisindicated paper, for each is hardly panel. The affected, radius of however. each circle is 8 10 cm. The color bar is in log scale. direct stripping. unbound near the× surface.SNE at the impact against the companion star is decreased; the Its values are listed on the table4 for each model. Note that the 2 SNE is more extended specially and the collision lasts longer; (A color version of this figureFor is available other models in the online with journal.) wider separations,2 forward×10 km/s and reverse The removed4×10 km/s masses are calculated in the same way as in shocks are formednumber when theof tracer SNE particles hits the is surface unchanged of the in theseSection experimental3.2.2 and aresince shown the as solid a function angle of of the time companion in Figure 13 starfor is also decreased, the Figure 7.5Velocity vectors at Step 2. Note that red arrows on6 the right panels have lengths amplified by a factor of 50 compared with blue arrows on the left panels. companion(c): (panelt=1.78x10 (a)computations. ofs Figures Although11 and 12 there)aswasthecase(f): ist=7.95x10 a clear trends thatsome the removed representativetotal models. energy Other contained models have in the similar portion features. of SNE that collides with with the same physicalfor the setup base but model. with different Themass forward decreases spatial shock resolu- as has the smaller numberWe pressure henceof(A grid color believeand versionpoints ofthat increases,Although this the figure resolution is theavailable the removed employedin thethe online masses companion journal.) in thistendstudy star to asymptotic is is reduced values proportionately; within the curvature of tions. The number of grid points was changedvariation separately is small, for1%,r andappropriate. does not affect our conclusion. the dense shell in the SNE becomes less pronounced. All of velocity,sincetheincidentejectaenergydensityissmaller.Since∼ afewmonthsinallmodels,thosewithwiderseparationssettle and θ directions. They are summarizedthese in effects Table 1 but. the last one will tend to decrease the amount of and the total removed mass is smaller. As the forward shock it takes the forward shock longer to sweep the entire star, the 3.2.4.at earlier Separation times. Dependence It will also take more time until the companion The dynamics ofhemisphere these test of runs thethe companion are removed compared mass. star that with The faces smaller the the curvature primary star of the and SNE mayregains lead mechanicalreaches7 and the thermalopposite equilibrium side of the companion in these models. star (panel (c) of base model in Figureis shock-heated10.Theflowpatternshardlyshow earlierto smaller has deflections a highly extended by the bow envelope shockIn compared and the contribute base model, to someFor the binarythe modelsFigures separation with11 and the is12 assumed widest), much separations, to smaller be amounts the heated of matter but become any difference for the models with different grids in the r to the opposite hemispheredirect stripping. that is heated later, makinga a themin.Thisisthesmallestpossiblevalueandismost star still bound matterunbound reaches near the outersurface. boundary before the forward = direction. On the othertake hand, a very as asymmetric the gridFor other becomes shape models (panel coarser with (b) wider ofin Figures separations,likely11 and to forward be12). too and small. reverseshock We reaches haveThe hencethe other removed performed side masses of the the are companion same calculated star. in the This same may way as in the θ direction, the shockSince waves the shock areshocks slightly heating are broadened, is formed significantly when while theweaker SNEsimulation in hits the the wide for surface severallead of models the to errors varyingSection in the3.2.2 the estimation separationand are shown of as thea as a unbound function of mass, time insince Figure 13 for complex flow patterns particularly behindcompanion the companion(panel (a) of star Figures 11 and 12)aswasthecase some representative models. Other= models have similar features. binaries, only the initially shocked regions become(1.1, 1 unbound.2,...,2.5) theamin heatingto see has the not dependence finished, and of removed our Bernoulli criterion may are changed even qualitatively. Thefor removed the base mass,model. which The forward is shockmass has on smaller separation. pressure× As the and binaryAlthough becomes the wider, removed the density masses of tend to asymptotic values within our main concern in this paper, isvelocity,sincetheincidentejectaenergydensityissmaller.Since hardly affected, however. SNE at the impact against the companionafewmonthsinallmodels,thosewithwiderseparationssettle star is decreased; the it takes the forward shock longer to sweep the entire star, the at earlier times. It will also take more time until the companion Its values are listed on the table for each model. Note that the SNE is more extended8 specially and the collision lasts longer; number of tracer particles is unchangedhemisphere in these of theexperimental companion star that faces the primary star and regains mechanical and thermal equilibrium in these models. is4 shock-heated earlier has a highly extendedsince the2 envelope solid angle compared of the companionFor star the is models also decreased, with the widest the separations, the heated but computations. Although there is2× a10 clearkm/s trend that the removed total4×10 energykm/s contained in the portion of SNE that collides with mass decreases as the number ofto grid the points opposite increases, hemisphere the that is heated later, making the star still bound matter reaches the outer boundary before the forward Figure 7. Velocity vectors at Step 2. Notetake that a red very arrows asymmetric on the right shape panels (panel havethe lengths (b) companion of amplified Figures by star11 a factorand is reduced12 of). 50 compared proportionately;shock with reaches blue arrows the the other curvatureon the side left panels. of the companion star. This may variation is small, 1%, and does not affect our conclusion. the dense shell in the SNE becomes less pronounced. All of ∼ Since the(A color shock version heating of this is figure significantly is available weaker in the online in the journal.) wide lead to errors in the estimation of the unbound mass, since binaries, only the initially shocked regions become unbound the heating has not finished, and our Bernoulli criterion may 7 these effects but the last one will tend to decrease the amount of and the total removed mass is smaller. As the forward shock the removed mass. The smaller curvature of the SNE may lead reaches the opposite side8 of the companion star (panel (c) of to smaller deflections by the bow shock and contribute to some Figures 11 and 12), much smaller amounts of matter become direct stripping. unbound near the surface. For other models with wider separations, forward and reverse The removed masses are calculated in the same way as in shocks are formed when the SNE hits the surface of the Section 3.2.2 and are shown as a function of time in Figure 13 for companion (panel (a) of Figures 11 and 12)aswasthecase some representative models. Other models have similar features. for the base model. The forward shock has smaller pressure and Although the removed masses tend to asymptotic values within velocity,sincetheincidentejectaenergydensityissmaller.Since afewmonthsinallmodels,thosewithwiderseparationssettle it takes the forward shock longer to sweep the entire star, the at earlier times. It will also take more time until the companion hemisphere of the companion star that faces the primary star and regains mechanical and thermal equilibrium in these models. is shock-heated earlier has a highly extended envelope compared For the models with the widest separations, the heated but to the opposite hemisphere that is heated later, making the star still bound matter reaches the outer boundary before the forward take a very asymmetric shape (panel (b) of Figures 11 and 12). shock reaches the other side of the companion star. This may Since the shock heating is significantly weaker in the wide lead to errors in the estimation of the unbound mass, since binaries, only the initially shocked regions become unbound the heating has not finished, and our Bernoulli criterion may

8 One Possible Scenario: A Massive Binary System One Possible Scenario: A Massive Binary System

slow ejecta ‘shadow’

fast (main body) of ejecta One Possible Scenario: A Massive Binary System

slow ejecta ‘shadow’

fast (main body) of ejecta reverse shock Conclusions

‣ A conical ejecta pattern may produce slow lateral expansion of rings if they are approaching us

(… may be why Winkler et al (1989) get such a short dynamical age for rings (~ 800 yrs), so may make 2nd supernova unnecessary)

‣ Assuming v ~ 1350 km/s for outer N-rich ring, t = 3700 - 4400 year lifetime of the SNR gives a Swirl distance ~ v t = 5 - 6 pc from center of Puppis A (assuming undecelerated ejecta)

… This is well inside Puppis A (R ~ 16 pc)

‣ Weak H emission within the rings and O-rich ejecta suggests a star that lost most of its H mass before the explosion, and maybe some mixing between layers (SN Type IIb/L: Chevalier 2005) (SN 1993J-type)

‣ Pre-SN wind explanation for Swirl? (how to get v ~ 1350 km s-1?)

‣ Could the H have been donated to secondary star?

‣ Faint Ca, Ar, Ni, He also detected in WiFeS data of Swirl: ongoing analysis

‣ Proper motions for the ring complex must be measured Optical/X-Ray Comparison of Swirl