4.3. Morphology of stellar deaths
The evolution of stars is determined primarily by their mass, because evolutionary timescales are set by the rate of consumption of the nuclear fuel. Figure 13 shows the evolutionary tracks of 1 M, 5 M, and 25 M stellar models in the luminosity-effective temperature plane. It is clear that certain phases of stellar evolution are characterized by considerable mass ejection, either via relatively gentle processes such as radiation pressure on atomic lines or on dust, or through stellar explosions. A process of the former type produces the objects known as planetary nebulae - shells of ejected matter from giant stars that are ionized by the radiation emitted by the hot ( 30,000 K) central core. The explosions known as Type II supernovae are the result of the dynamical collapse of the iron cores of massive stars, and the ejected nebulae form the objects known as supernova remnants.
Figure 13. Evolutionary tracks of 1 M, 5 M, and 25 M stars in the Luminosity-Effective Temperature (Hertzspring-Russell) diagram. Thick segments mark long, nuclear burning, evolutionary phases (from Iben 1985; adapted from Prialnik 2000.
One of the most striking phenomena revealed by HST observations of stellar deaths and of stars in their late stages of evolution is the fact that axisymmetric nebulae are extremely common. This observation remains true even after recognizing that the morphologies are blurred by projection effects, and by the lines used to obtain the image (i.e., the [O III] image may look different than the [N II] image). Furthermore, axisymmetry is found not only in planetary nebulae (PNe), that are formed by intermediate-mass stars (1-8 M), but also around Luminous Blue Variables (LBVs), that are extremely massive stars (~ 100 M), and supernovae (SNe), that represent the death of massive stars (8-30 M).
Eta Carinae ( Car), for example, belongs to the small class of very massive stars known as Luminous Blue Variables. These stars are believed to represent a rapid evolutionary phase, in which the stars experience severe mass loss (losing many solar masses in ~ 104 years), sometimes via giant outbursts (e.g., Davidson and Humphreys 1997). In the 1840s Car suffered such an outburst, increasing in brightness by several magnitudes and ejecting a large amount of material. The star stabilized around 1870 (except for a minor eruption between 1887 and 1895). The precise cause for this giant outburst is still unknown, but it has produced a spectacular, bipolar nebula commonly referred to as the Homunculus (Fig. 14). In addition to the Homunculus, the HST images revealed for the first time the presence of a ragged ejecta disk around the "waist" of the hourglass structure, composed, at least partly, of radial streaks. The total mass in the equatorial disk has been estimated (based on conventional gas to dust ratios) to be at least 0.1-0.2 M, but this could underestimate the mass significantly.
Figure 14. The Luminous Blue Variable star Eta Carinae, HST/WFPC2, 1994. Credit: NASA and J. Hester (Arizona State University). http://hubblesite.org/newscenter/archive/1994/09/
Recent spectroscopic observations of Car by the Space Telescope Imaging Spectrograph (STIS) on board HST showed that the spectrum could be fitted well using a model with a mass-loss rate of ~ 10-3 M yr-1. The minimum mass of the system is currently estimated to be 120 M (Hillier et al. 2001).
On the basis of photometric and radial velocity variations (in particular the disappearance of high-excitation lines like He I, Fe III) and x-ray observations, it appears that Car is a binary system, with an orbital period of 5.52 years (Damineli, Conti, and Lopes 1997, Damineli et al. 2000, Ishibashi et al. 1999). The mass of the secondary is not known, but it may be less the 30 M (Hillier et al. 2001).
Eta Carinae is an extremely enigmatic object on many fronts. However, the morphology of its ejected nebula becomes particularly intriguing when we realize that HST observations reveal almost identical nebular morphologies in objects of very different mass and evolutionary history. One of the best known of these is the supernova 1987A (SN 1987A) in the Large Magellanic Clouds (see, for example, McCray 2003). The supernova was first observed in February 1987 (and hence 3 years prior to the launch of HST) and was immediately classified as a Type II Supernova (SN II; representing the collapse of the iron core of a massive star) by virtue of its strong hydrogen lines (coming from the hydrogen-rich envelope). For the first time, the detection of neutrino events (formed copiously as matter rapidly neutronizes) directly confirmed the association between SNe II and core collapses of massive stars. The exploding star itself, SK -69°202, had actually been observed prior to the explosion to be a B3 blue supergiant, with a luminosity of L 1.1 × 105 L. Since the HST launch, SN 1987A has become a prime target for the telescope, being the nearest supernova in modern times. The HST observations have revealed a remarkable system of circumstellar rings surrounding the bright center (Fig. 15; Burrows et al. 1995, Pun 1997). These rings reflect the morphology of material ejected by the supernova progenitor a couple of tens of thousands of years before the explosion (Burrows et al. 1995). This can be inferred from the fact that the central ring is expanding at about 10 km s-1 (Crotts and Heathcote 1991) and it currently has a radius of about 6.3 × 1017 cm. While a full explanation for the formation of the rings is still lacking, there is very little doubt that what we are observing is a bipolar structure, in which the inner ring marks the narrow "waist," while the larger rings are somehow "painted" on the bipolar lobes, or possibly mark their edges. The entire structure is thus very similar to the one observed in Car.
Bipolar structures have not been restricted only to massive stars, however. Some planetary nebulae and symbiotic nebulae exhibit morphologies that are almost identical to those of Car and SN 1987A. Some of the best examples are probably My Cn18 (Fig. 16; the "hourglass" nebula, usually classified as a planetary nebula), and the "Southern Crab" (Fig. 17), now recognized as a symbiotic nebula (Corradi et al. 2001).
Figure 16. Planetary Nebula MyCn18, The
Hourglass Nebula, HST/WFPC2.
Credit: NASA, R. Sahai, J. Trauger (JPL) and the WFPC2
Figure 17. Nebula surrounding the symbiotic star system He2-104, the Southern Crab Nebula, HST/WFPC2, May 1999. Credit: NASA and R. Corradi (Instituto de Astrofisica de Canarias, Tenerife, Spain), M. Livio (Space Telescope Science Institute), U. Munari (Osservatorio Astronomico di Padova-Asiago, Italy), H. Schwarz (Nordic Optical Telescope, Canarias, Spain). http://hubblesite.org/newscenter/archive/1999/32/
Planetary nebulae represent the late stages in the lives of stars of about 1-8 M. At that phase, the stars eject their outer envelopes exposing the hot cores (T 30, 000 K), which, in turn, ionize the nebulae causing them to fluoresce. Symbiotic nebulae have at their centers symbiotic binary systems, consisting typically of a red supergiant and a white dwarf that provides the ionizing radiation.
The main question that arises, therefore, is: What is (are) the mechanism(s) that is (are) capable of producing such bipolar morphologies in stars of different masses, and different ages and evolutionary histories? The two main mechanisms that have been proposed are: (1) Interacting winds in the presence of an equatorial to polar density contrast, and (2) Magnetic tension of a toroidal field (see Balick and Frank 2002, for a more extensive discussion).
Let us first examine the interacting winds model. The original "interacting winds" model for planetary nebulae (Kwok 1982, Kahn 1982) suggested that old, intermediate-mass stars, in the phase of evolution known as the asymptotic giant branch (AGB), first emit a slow (~ 20 km s-1; of the order of the escape velocity from the AGB star's surface) wind, followed by a fast wind (~ 1000 km s-1), once the hot and compact nucleus (the AGB star's core) is exposed. The fast wind catches up with the slowly moving material and shocks it. Balick (1987) proposed that when the interacting winds are allowed to operate in the presence of a density contrast between the equator and the pole, a variety of axially symmetric morphologies can be obtained. The idea is that, for reasons that will be explored below, the slow wind contains a non-spherical density distribution, with material being denser around the equator than in the polar direction. Consequently, the (spherically symmetric) fast wind can penetrate more easily at the poles, forming an axisymmetric nebula. Numerical simulations have shown that when a range of density contrasts (between the equatorial and polar directions) is used, and, in addition, the nebular inclination with respect to the line of sight is taken into consideration, most of the observed morphologies can be reproduced (Soker and Livio 1989, Icke et al. 1992, Frank et al. 1993, Mellema 1995, Dwarkadas et al. 1996).
The second class of models involves the action of a magnetic field.
The toroidal field component in an outflow from a star is given by (Parks 1991)
where RS, BS are the stellar radius and the surface magnetic field, Vrot is the equatorial rotational velocity and VW() is the wind terminal velocity. Consequently, the ratio of the toroidal to radial component increases like (r/RS) at large distances, leading potentially to an axisymmetric configuration, by the fact that magnetic stresses can slow down the flow in the equatorial direction (while not interfering with the polar direction). The key physical parameters determining the obtained morphology are the stellar rotation rate, the ionizing radiation, and the stellar magnetic field. These can be expressed by: rot / crit, F*, and (Chevalier and Luo 1994, Garcia-Segura et al. 1999). Here crit is the critical (Keplerian) angular velocity, crit = (GMS / RS3)1/2, F* is the average flux of Ly photons (of order 1045-1047 s-1 for typical PNe nuclei), and is the ratio of the magnetic energy density to the kinetic energy density in the wind (e.g., Begelman and Li 1992)
Let us now examine the effects of each one of these parameters.
For the rotation to have a significant effect and produce a bipolar morphology, the star needs to rotate at a significant fraction of its breakup speed (rot / crit 0.5). Under these conditions, conservation of angular momentum results in a significant focusing of the wind towards the equatorial plane, leading to an equatorially compressed outflow (e.g., Bjorkman and Cassinelli 1993, Owocki et al. 1994, Livio 1994). The concomitant equator to pole density contrast leads to bipolar morphologies.
Ionization does not, in itself, produce a bipolar morphology. Rather, ionization fronts tend to excite instabilities (similar to the Rayleigh-Taylor instability or the instability discussed by Vishniac 1983), which, in turn, produce finger-shaped structures and dense "knots." Observations of a number of relatively nearby planetary nebulae, and in particular of the "Helix" nebula, reveal that such dense knots are probably very common (O'Dell et al. 2003), Speck et al. 2002)).
Figure 18. A schematic distribution of the planetary nebulae morphology in the rotation (-magnetic energy density () plane (see text; adapted from Garcia-Segura et al. 1999.
The toroidal magnetic field is carried by the fast wind and it can, in principle, produce a bipolar morphology even if the slow wind is spherically symmetric. Basically, as the magnetic energy density of the shocked wind becomes larger than the thermal energy density, the flow becomes (due to the increasing importance of the toroidal component) bipolar, with an increased collimation as the value of is increased. Numerical simulations (Garcia-Segura et al. 1999) show that qualitatively, the morphologies obtained for different combinations of rot / crit and are as shown in Fig. 18. The minimum field required to produce magnetic shaping was found to be of order (Chevalier and Luo 1994)
where VFW is the velocity of the fast wind.
From eq. 13 we see that if the star rotates too slowly, the minimum required field may be unattainable. Similarly, if the star does not rotate at a significant fraction of its breakup speed, an equatorially compressed outflow is not formed. The question of: What is the mechanism that produces highly bipolar outflows? can therefore be reduced to: What causes the star to rotate close to breakup? or: What can produce a strong density contrast between the equatorial and polar directions in the slow wind?
One obvious possibility is: binary companions!
Companions to the central star can act in several ways to aid in the formation of bipolar morphologies: (1) For binaries that were initially relatively close (separation less than ~ 1000 R), so that the primary could fill its Roche lobe (the critical potential surface beyond which mass transfer onto the companion occurs) during the asymptotic giant branch (AGB) phase, an unstable mass transfer ensues. As a result, the companion and the AGB star's core start spiralling-in inside a common envelope (see, e.g., a review by Iben and Livio 1993). This has two effects. First, the envelope of the primary can be spun-up to angular velocities of the order of
where MC is the companion's mass, Menv is the mass of the giant's envelope, MS Kg2 RS2 is the star's moment of inertia and a is the initial separation between the giant and the companion. Equation 14 shows that even brown dwarf (sub-stellar) companions can bring the envelope close to critical rotation. Second, since the envelope mass is ejected (due to orbital energy deposition) primarily close to the orbital plane (because angular momentum is also deposited into the envelope), the common envelope phase can generate an equator-to-pole density contrast. Hydrodynamic simulations of common envelope evolution reveal that during the late stages about 80% of the mass is ejected within 30° of the binary orbital plane (e.g., Terman, Taam, and Hernquist 1994, 1995, Rasio and Livio 1996). One can expect that due to cooling, the mass will sink even more toward the orbital plane at later times.
Another possibility for the presence of a higher equatorial density, in principle at least, that does not involve binary companions, is the inner rim of the protostellar disk. If the outer part of the protostellar disk survives till late stages in the stellar life (which may not be difficult in the case of massive, short-lived stars), then the fast wind could interact with the inner rim of this disk (Pringle 1989). In a few planetary nebulae (for example, the "Red Rectangle" and the "Egg Nebula," Bond et al. 1996, Thompson et al. 1997, respectively). HST observations reveal the presence of relatively large disks, similar to the ones observed in young stellar objects.
In at least some symbiotic nebulae (e.g., M2-9), the bipolar morphology may reflect the action of the white dwarf companion. The white dwarf accretes from the wind of the AGB star, an accretion disk is formed, and the disk powers a mildly collimated fast wind, which in turn produces the bipolar morphology (Soker and Rappaport 2001, Livio and Soker 2001).
The conclusion from this discussion is that several mechanisms are capable, in principle at least, to produce the observed bipolar morphologies. Different mechanisms may be operating in different systems. In some cases, we can look forward to the future and expect more definitive answers to emerge. For example, in SN 1987A, the supernova blast wave will eventually hit the entire inner ring, and the luminosity that will be generated by this interaction will illuminate the entire SN vicinity (McCray 2003). A few brightening spots, where the blast wave has already hit protrusions on the ring, have been observed by HST (Fig. 19; see also Panagia 2002). Generally, a transmitted shock at normal incidence is expected to propagate into the ring with a speed of Vring (no / nring)1/2 Vblast (where no, nring are the number densities of the circumstellar matter and the ring, respectively, and Vblast is the blast wave velocity). For SN 1987A, Vblast ~ 4000 km s-1, no ~ 150 cm-3, nring ~ 104 cm-3 giving Vring ~ 500 km s-1. Eventually, the H flux from the ring is expected to be more than 30 times higher than today (FH 3 × 10-12 erg cm-2 s-1), and even brighter in UV lines (Luo, McCray, and Slavin 1994). This ionizing flux will turn the circumstellar matter into an emission nebula, thus revealing its distribution and velocity field, and hopefully allowing for a reconstruction of the mass-loss history of the system. In all of this, HST will provide a front seat view.
Figure 19. An HST-WFPC2 image of the circumstellar ring around supernova 1987A, obtained in May 2002 with narrow band filters that include H 6563 Å and [NII] 6584 Å emission line radiation. We can identify a number of bright spots that correspond to gaseous protuberances on the inner side of the ring that are currently being hit by the supernova ejecta, thus becoming hot and luminous. Because of the narrow width of the filters employed to gather this image, the ring and the spots appear bright since they are mostly radiating in emission lines. On the other hand, both the supernova itself (a wide patch at the center of the ellipse) and a star (projected by chance near spot #6) are not visible in this image because most of their energy is radiated in the form of a rather smooth continuum. Credits: Nino Panagia (ESA/STScI), on behalf of the SINS Collaboration.
The collapses of very massive stars are believed to produce stellar-mass black holes. In some cases, these collapses produce the most spectacular explosions in the universe since the "big bang" - Gamma Ray Bursts. The Hubble Space Telescope has played an important role in the attempts to understand the nature of these dramatic explosions. However, much more massive black holes are also produced in the universe quite commonly, and the questions of how they form and evolve have intrigued astronomers for decades. The initial answers had to come from large collections of stars like galaxies and stellar clusters.
Given the unprecedented resolution that HST provides, very crowded fields, like the centers of galaxies, were obvious targets. In particular, the early suggestive evidence that galactic centers harbor supermassive black holes (see Kormendy and Richstone 1995, for a review), virtually ensured that centers of galaxies will be observed extensively with HST. Indeed, these observations were carried out and proved to be extremely fruitful.