Proc. NatI. Acad. Sci. USA Vol. 84. pp. 6961-6964, October 1987 Astronomy On the possibility of dust condensation in the ejecta of 1987a (infrared/novae/pulsars) R. D. GEHRZ AND E. P. NEY Department of Astronomy, School of Physics and Astronomy, University of Minnesota, Minneapolis, MN 55455 Contributed by E. P. Ney, June 15, 1987

ABSTRACT We suggest that supernova 1987a may con- Existing data for dust formation in type II supernovae do dense dust of substantial visual optical thickness as do many not appear to cover the initial stages ofthe formation process; novae. The dust will act as a calorimeter of the photon they therefore yield little information on the effects of the of any central engine that is dominant at the time of early stages of dust formation on the development of the dust formation. Observations of novae suggest that dust visible light. Estimates of the dust shell optical depth offered formation may occur when the expanding ejecta reach a by Dwek et al. (6) depend heavily upon theoretical assump- temperature of 1000 K. The early luminosity of the supernova tions about the development of the central engine. The may be dominated by radioactivity that is unrelated to the fundamental deficiency in the data is due to the fact that dust central engine that determines the energy balance for the formation in these supernovae has occurred when they were long-term development of the supernova. We discuss the in the daylight sky and were not observed during critical possibility that a constant luminosity central power source such epochs. However, by analogy with novae, one might spec- as a pulsar dominates the luminosity of the supernova ejecta by ulate that 1979c had an optically thick shell and that the shell the time that dust can condense and argue that, if a shell mass in 1980k was optically thin. of more than a few tenths of one solar mass was ejected, Fig. 1 summarizes the state of our knowledge about dust emission from dust may be observable in the thermal infrared formation in novae and supernovae. The constancy of the spectral region. Maximum dust optical depth should occur by luminosity of the central engine in novae renders the distinc- late 1987 or early 1988. If the dust becomes optically thick, the tion between the optically thick (NQ Vul) and the optically visual light from the supernova may drop precipitously. The thin (V1668 Cyg) dust shells quite certain. While the infrared characteristics of an optically thick dust shell as a calorimeter energy distributions show that dust probably formed in the of the luminosity of the central engine are discussed and are ejecta of supernovae 1979c and 1980k (6-8) the shell optical related to previous observations of dust formation in type II depths and central engine at the time of dust supernovae. It is suggested that dust of several chemical formation remain uncertain. compositions may form at different epochs. Supernova 1987a as a Test Case for Dust Formation History of Dust Formation in Novae and Supernovae Supernova 1987a presents an unusual opportunity to study the dust formation process in supernovae because it is bright Many classical novae form optically thick dust shells in their enough to be observed using state of the art optical/infrared ejecta. Two recent examples are NQ Vul (1) and LW Ser (2). techniques (even in daylight) throughout the critical time The optically thick shells are a calorimeter of the period during which dust grains can nucleate and grow. constant luminosity phase of the nova remnant that is caused It appears plausible that a dust formation phase will attend by nuclear burning in material that relaxes onto the white a supernova event such as 1987a. We suggest that a pulsar dwarf following the eruption (3, 4). In some cases, such as could act as a constant luminosity central engine to power the V1668 Cyg, the dust shells that condense are optically thin dust emission by the time that dust formation occurs. In fact, (5). an optically thick dust shell could provide a measure of the The case for dust formation in type II supernovae, which luminosity of an embedded pulsar at wavelengths where dust result from the core collapse of massive , is less well is an effective absorber. It is difficult to estimate how the documented. Dwek et al. (6) have argued that dust formed in early luminosity of the supernova is related to the luminosity the ejecta of supernova 1980k. The development of the dust of the central engine that characterizes the dust formation in phase. The initial luminosity of the ejecta may be dominated formation phase this supernova was not documented by the decay of short-lived radioactive isotopes such as 56Ni because it could not be observed during the critical period of and 56Co (9). transition between the gas expansion and the dust formation Although some of the energy emitted by a pulsar in the phases. The dust shell of 1980k radiated 2 x 10-3 of the form of high-energy electrons and cosmic rays may; not outburst luminosity when it was first observed, and the visual couple efficiently to a dust shell, Ostriker and Gunn (10) have brightness of the supernova was large enough to suggest that argued that nearly all of the magnetic dipole radiation of the the shell was not optically thick. Merrill (27) showed that dust pulsar will be converted in a compact fireball to wavelengths condensation probably occurrred in supernova 1979c, but the that would efficiently couple to dust formed in the overlying available data do not elucidate the formation process. In this ejecta. We believe that only a few percent of the luminosity case, the dust shell luminosity was 1% of the initial outburst of a supernova will be in high-energy cosmic rays that cannot luminosity; the short-wavelength brightness in this case was couple to dust grains. In the cases of supernovae 1979c and low, suggesting that the shell was optically thick. 1980k (see Fig. 1), the luminosities reradiated by the dust were respectively 100 and 500 times lower than the initial The publication costs of this article were defrayed in part by page charge outburst luminosities. We examine below the case in which payment. This article must therefore be hereby marked "advertisement" the luminosity ofthe central engine can be efficiently coupled in accordance with 18 U.S.C. §1734 solely to indicate this fact. to the dust grains.

6961 Downloaded by guest on September 30, 2021 6962 Astronomy: Gehrz and Ney Proc. Natl. Acad. Sci. USA 84 (1987) z

~~~~~~~~DayC~~~~~~~~~j 52

F- 115 , l,,,e, ' , , , - 3: (C) (d) Z 10 Supernova 1979c 10 Supernova 1980k

10-1ArI22,19791 November 2,1980

10-19 -- -

0.5 1 2 4 810 0.5 1 2 4 810 WAVELENGTH IN MICRONS FIG. 1. Summary of known scenarios for dust formation in novae and supernovae. (a) Optically thick nova dust shell of NQ Vul in which the dust shell luminosity repeats the outburst luminosity (data from ref. 1). (b) Optically thin nova dust shell of V1668 Cyg where the shell luminosity was about 10% of the outburst luminosity (data from ref. 5). (c) Dust in supernova 1979c where the shell may have been optically thick 256 days after the outburst as inferred from the energy emitted at 3-4 Aun! compared to the short-wavelength luminosity at the same time (data from refs. 7 and 8). (d) Dust in supernova 1980k, which was optically thin with respect to the central engine as inferred from comparison of the infrared and short-wavelength luminosities 215 and 357 days after the outburst (data from refs. 6-8).

Physical Conditions Attending Dust Formation by the condensation of carbon on cohenite (Fe3C) seed nuclei. Supernova 1987a may condense dust of substantial optical Production of substantial amounts ofdust in the supernova thickness as do many novae if the dust grain nucleation and ejecta requires that the average shell gas density be higher growth proceed in a similar manner. The temporal develop- than some critical value Pc at the condensation point. If the ment of the optically thick case for novae is typified by NQ shell has a thickness AR = fR at time td, the shell density p Vul (1) as diagrammed in Fig. 2. Optically thick dust shells in is given by novae are black in the 2- to 23-gm infrared spectral region suggesting that the grain material is carbon (1, 2, 5). The time p= M/4rfR3 = M/4mjfV3t', [21 td for the dust shell to become optically thick is given by where M is the total mass lifted off in the ejection. From Eq. 1 and assuming thatf = 0.1, we find that the shell density is td = RV-1 = V-'[L14sTrr1 / [11 p = 1.2 x 10-4(M/Mi) 31/2 gcm-3, [3] where L is the luminosity of the central engine, V is the (L/Lo) outflow velocity ofthe ejecta in which the grains nucleate and which is seen to be a function of only the ejected mass and grow, T is the temperature of the dust shell when it reaches the luminosity of the central engine at the time of dust maximum optical depth, and R = [L/41roT71112 is the shell formation. Assuming that the same shell geometry applies to radius at maximum optical depth. Past observations of novae both novae and type II supernovae, we can estimate the that produce optically thick carbon shells have shown that T critical density Pc for optically thick dust shell production by is always 1000 K. This temperature is much lower than is examining the data for a typical optically thick nova eruption. predicted by standard theory for carbon condensation (11). In the case of LW Ser (2), M = 2 x 10-5 M® and L = 4 x 104 Lewis and Ney (12) have suggested that carbon condensation L®,D and Eq. 3 yields Pc = 3 x 10-16 gcm3. We therefore is suppressed above 1000 K and can be expedited at 1000 K predict from Eq. 3 that the ejecta of supernova 1987a, for Downloaded by guest on September 30, 2021 Astronomy: Gehrz and Ney Proc. Natl. Acad. Sci. USA 84 (1987) 6963 J ULI AN DAY 2443000 + A 20 MO , such as is believed to be the precursor of 1987a, 092 112 132 152 172 192 212 can eject as much as 90% (18 MO) of its processed material 6 | (13). Supernova ejecta may be highly enhanced compared to nova ejecta in the elements that contribute to dust production W for the material be 8 and the critical density condensible may sVISUAL reached for a much lower shell mass. We conclude that the Z 10 ejecta of 1987a will be dense enough at the condensation point to produce an optically thick dust shell. If an optically thick dust shell forms, the central source will be obscured resulting 12 in an abrupt drop in the visible light. The visible light drop , o may be moderated if radioactive decay, which will deposit a energy throughout the ejecta, is still a significant source ofthe photon of the at the time dust 2 - luminosity supernova of

-13 10 < 0-14mWaxS= 5Xi I) E (I)~~~~~~~~~~~~~~~~~~~~~~I H -14 T1 0 \ < 10 Outburst1 Z Luminosity T 10K 10C \ ?FIG.3. Predicted energy dis- \ / tribution of supernova 1987a for .-<15 _ \ / \ the case where an optically thick (h) dust shell at the 1000 K conden- sation temperature forms around a central engine that maintains the outburst luminosity (curve a) and for the case where a similar shell |/i X X forms around a central engine with 10 1% of the outburst luminosity WAVELENGTH IN MICRONS (curve b). Downloaded by guest on September 30, 2021 1v( 6964 Astronomy: Gehrz and Ney Proc. Natl. Acad. Sci. USA 84 (1987) grains may condense later. Dust condensates in novae have may explain in general why evolved supernova remnants included carbon grains (1, 2, 5) and silicon-rich materials (18, such as the Crab are relatively dust-free. On the other hand, 19). the observation that dust apparently did form at some level If the luminosity of the central engine is less, the dust in supernovae 1979c and 1980k demonstrates that protection should condense sooner. For example, the scenario dis- was afforded the dust production zone. cussed by McCray et al. (9) in which radioactivity dominates the early luminosity would be a case in point. Dust produc- We thank E. Dwek, J. S. Gallagher, M. Shull, S. Starrfield, and tion as early as late July 1987 could result if the low-density W. A. Stein for information and stimulating discussions. The authors high-velocity ejecta (14,000 km sec-1) are capable of sup- are supported by the U.S. Air Force, the National Science Foun- porting grain nucleation and growth. An optically thick dust dation, and the Graduate School of the University of Minnesota. shell with T = 1000 K and the outburst luminosity would have 1. Ney, E. P. & Hatfield, B. F. (1978) Astrophys. J. Lett. 219, an angular radius of about 23 milliarc seconds, and would L111-L115. have approximate infrared magnitudes of K = 0.9, L = -1.0, 2. Gehrz, R. D., Grasdalen, G. L., Hackwell, J. A. & Ney, E. P. M = -1.8, [10 Am] = -2.7, and [20 tsm] = -3.1 (see Fig. 3). (1980) Astrophys. J. 237, 855-865. If the shell can unambiguously be demonstrated to be op- 3. Nariai, K. (1974) Publ. Astron. Soc. Pac. 26, 57-64. tically thick, for example by the presence of a visible light 4. Gallagher, J. S. & Starrfield, S. G. (1976) Mon. Not. R. transition, then the physical size of the shell will be known Astron. Soc. 176, 53-61. explicitly because of the known distance to the large Mag- 5. Gehrz, R. D., Hackwell, J. A., Grasdalen, G. L., Ney, E. P., ellanic cloud. The outflow velocity of the ejecta participating Neugebauer, G. & Sellgren, K. (1980) Astrophys. J. 239, in the dust formation process can thus be well specified in this 570-580. 6. Dwek, E., A'Hearn, M. F., Becklin, E. E., Hamilton Brown, case. For an optically thick dust shell reemitting 0.01 of the R., Capps, R. W., Dinerstein, H. L., Gatley, I., Morrison, D., outburst luminosity, the infrared fluxes would be 5 magni- Telesco, C. M., Tokunaga, A., Werner, M. W. & Wynn- tudes fainter than those listed above, and the angular radius Williams, C. G. (1983) Astrophys. J. 274, 168-174. would be 2.3 milliarc seconds. Condensation of a thick dust 7. Barbon, R., Ciatti, F. & Rosino, L. (1982) Astron. Astrophys. shell should lead to an abrupt drop in the visual brightness 116, 35-42. accompanying the infrared increase, and the infrared lumi- 8. Barbon, R., Ciatti, F., Rosino, L., Ortilani, S. & Rafanelli, P. nosity will equal the luminosity of the central engine if the (1982) Astron. Astrophys. 116, 43-53. shell is optically thick. 9. McCray, R., Shull, J. M. & Sutherland, P. (1987) Astrophys. J. McCray et al. (9) have predicted that several light echoes Lett. 317, L73-L77. 10. Ostriker, J. P. & Gunn, J. E. (1971) Astrophys. J. Lett. 164, with power at 10 Am of about 1037 erg-s' may occur off L95-L104. preexisting dust grains. Light echoes from supernovae have 11. Clayton, D. D. (1979) Astrophys. Space Sci. 65, 179-189. also been discussed by Bode and Evans (20) and by Dwek 12. Lewis, J. & Ney, E. P. (1979) Astrophys. J. 234, 154-157. (21). The shell emission we describe here could exceed those 13. Clayton, D. D. (1983) Principles of Stellar Evolution and effects by more than an order of magnitude and might render Nucleosynthesis (Univ. of Chicago Press, Chicago), pp. the light echoes difficult to observe photometrically. The 543-545. crucial test oflight echoes versus dust condensation would be 14. Cropper, M., Bailey, J. A., Peacock, T. I. & Wickramasinghe, the angular size ofthe shell. Light echo shells are 10-30 times D. T. (1987) Int. Astron. Union Circ. 4319. larger than dust condensation shells. 15. Feast, M. W. (1987) Int. Astron. Union Circ. 4320. 16. Clayton, D. D. (1982) Q. J. R. Astron. Soc. 23, 174-212. Several novae have formed either optically thin dust shells 17. Lattimer, J. M., Schramm, D. N. & Grossman, L. (1978) (5, 18, 19) or no dust at all (22-24). Gallagher (23) has argued Astrophys. J. 219, 230-249. that early ionization of the ejecta can lower their opacity 18. Gehrz, R. D., Ney, E. P., Grasdalen, G. L., Hackwell, J. A. sufficiently that hard photons from the relaxing remnant can & Thronson, H. A., Jr. (1984) Astrophys. J. 281, 303-312. prevent the nucleation and growth of the grains. In the case 19. Gehrz, R. D., Grasdalen, G. L., Greenhouse, M., Hackwell, oftype II supernovae, energetic photons from the radioactive J. A., Hayward, T. & Bentley, A. F. (1986) Astrophys. J. Lett. decay phase could suppress early dust formation. An alter- 308, L63-L66. native possibility is that the shell densities ofthese novae fell 20. Bode, M. F. & Evans, A. (1980) Mon. Not. R. Astron. Soc. considerably below the criterion of Pc = 3 x 10-16 g9cm3 193, 21P-24P. established by Eq. 3. For example, we find p = 4 x 10-1' 21. Dwek, E. (1983) Astrophys. J. 274, 175-183. 3 22. Ennis, D., Becklin, E. E., Beckwith, S., Elias, J., Gatley, I., g-cm for V1500 Cygni (24). Matthews, K., Neugebauer, G. & Willner, S. P. (1977) None of the dust suppression scenarios envisioned above Astrophys. J. 214, 478-487. can be ruled out for supernova 1987a. In addition, it seems 23. Gallagher, J. S. (1977) Astron. J. 82, 209-215. clear that the grain growth processes require that the dust 24. Gallagher, J. S. & Ney, E. P. (1976) Astrophys. J. Lett. 204, production zone be protected from radioactive decay y rays L35-L39. and the high-energy radiation that would be characteristic of 25. Mitchell, R. M. & Evans, A. (1984) Mon. Not. R. Astron. Soc. pulsar emission. There is ample evidence that grains that 209, 945-954. form in nova shells are destroyed by the radiation field as the 26. Mitchell, R. M., Robinson, G., Hyland, A. R. & Neugebauer, shell becomes optically thin following the grain growth phase R. (1985) Mon. Not. R. Astron. Soc. 216, 1057-1071. (2, 25, 26). This effect may obtain in the ejecta of 1987a and 27. Merrill, K. M. (1980) Int. Astron. Union Circ. 3444. Downloaded by guest on September 30, 2021