Supernovae

Detailed radio observations of extragalactic supernovae are critical to obtaining valuable information about the nature and evolutionary phase of the progenitor star in the period of a few hundred to several tens-of-thousands of years before explosion. Additionally, radio observations of old supernovae (>20 years) provide important clues to the evolution of supernovae into supernova remnants, a gap of almost 300 years (SN$\sim$1680, Cas A, to SN 1923A) in our current knowledge. Finally, new empirical relations indicate that it may be possible to use some types of radio supernovae as distance yardsticks, to give an independent measure of the distance scale of the Universe. However, the study of radio supernovae is limited by the sensitivity and resolution of current radio telescope arrays. Therefore, it is necessary to have more sensitive arrays, such as the SKA, to advance radio supernova studies and our understanding of supernovae, their progenitors, and the connection to supernova remnants.

Supernovae (SNe) play a vital role in galactic evolution, through explosive nucleosynthesis and chemical enrichment, energy input into the interstellar medium, their stellar remnants (i.e., neutron stars, pulsars, and black holes), and the production of cosmic rays. SNe are also being utilized as powerful cosmological probes, both through their intrinsic luminosities and expansion rates. A primary goal of supernova research is an understanding of progenitor stars and explosion mechanisms for the different SNe types. Unfortunately, little is left of the progenitor star after explosion, and only the progenitors of three (SNe 1987A, 1978K, and 1993J) out of more than 1350 extragalactic SNe have been directly identified in pre-explosion images. Without direct information about the progenitors, thorough examination of the environments of SNe can provide useful constraints on the ages and masses of the progenitor stars.

SNe come in three basic types (e.g., Filippenko 1997): Ia, Ib/c, and II. Both SNe Ia and SNe Ib/c lack hydrogen lines in their optical spectra, whereas SNe II all show hydrogen in their optical spectra with varying strengths and profiles (Schlegel 1996). SNe Ib and SNe Ic subclasses do not show the deep Si II absorption trough near 6150 Å that characterizes SNe Ia, and SNe Ib show moderately strong He I lines, while SNe Ic do not.

These spectral differences are theoretically explained by differences in progenitors. SNe Ia are currently thought to arise from the total disruption of white dwarf stars, which accrete matter from a binary companion. In contrast, SNe II, SNe Ib, and SNe Ic are likely the explosions of massive stars. SNe II presumably result from the core collapse of massive hydrogen-rich supergiant stars with masses $8 \mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$ }}}\hbox{$<$ }}}M(M_{{\odot}}) \mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$ }}}\hbox{$<$ }}}40$. On the other hand, SNe Ib/c are believed to arise from a massive progenitor which has lost all of its hydrogen envelope prior to explosion (e.g., Porter & Filippenko 1987). One candidate progenitor for SNe Ib/c is exploding Wolf-Rayet stars (which evolve from stars with $M \mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$ }}}\hbox{$>$ } }}40\ M_{{\odot}}$; e.g., Conti et al. 1983: Humphreys, Nichols, & Massey 1985). An alternative candidate is exploding, relatively less-massive helium stars in interacting binary systems (Uomoto 1986; Podsiadlowski et al. 1992).

Possible variant of normal SNe II are the ``Type IIn,'' (Schlegel 1990) and the ``Type IIb'' (Filippenko 1988), which both show unusual optical characteristics. SNe IIn show the normal broad Balmer line profiles, but with the narrow peak sitting atop the broad base. The narrow component presumably arises from interaction with a dense ( $n \mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$ }}}\hbox{$>$ } }}10^7$ cm^-3) circumstellar medium (CSM) surrounding the SN. SNe IIb look optically like normal SNe II at early times, but evolve to more closely resemble SNe Ib at late times.

Radio Supernovae

Radio emitting supernovae (RSNe) have been extensively searched for since at least 1970 (e.g., de Bruyn 1973; Allen et al. 1976), but, due to low resolution, background confusion, and sensitivity limitations, only with the Very Large Array (VLA) was the first example found which could be studied in detail at multiple radio frequencies (SN 1979C; Weiler et al. 1981; see also Weiler et al. 1986, 1991, 1992a; Montes et al. 1998b). So far, about 22 RSNe have been detected since then with the VLA, with about 17 objects which have been extensively studied, including the SN II 1980K (Weiler et al. 1986, 1992b; Montes et al. 1998a), the SNe Ib/c 1983N (Weiler et al. 1986) and SN 1990B (Van Dyk et al. 1993b), the SNe IIn 1986J (Weiler, Panagia, & Sramek 1990) and SN 1988Z (Van Dyk et al. 1993a), and the SN IIb 1993J (Van Dyk et al. 1994). The SNe IIn are unusual not only in the optical, but also in the radio, in being exceptionally powerful radio sources ( $\sim 10^{28}$ erg s^-1 Hz^-1, or several thousand times the luminosity of Cas A, at 6 cm).

 

 

Fig. 2.25 shows several examples of detailed RSNe light curves. Fig. 2.26 shows an example of a particularly well-measured RSN, the SN IIb 1993J, which is only 3.6 Mpc distant, in M81.

Analysis of the radio emission provides vital insight into the interaction of the SN shock with preexisting circumstellar matter lost by the progenitor star or progenitor system, and, therefore, into the nature of presupernova evolution.

All RSNe appear to have the common properties of 1) nonthermal synchrotron emission with high brightness temperature; 2) a decrease in absorption with time, resulting in a smooth, rapid turn-on first at shorter wavelengths and later at longer wavelengths; 3) a power-law decline of the flux density with time at each wavelength after maximum flux density (optical depth $\approx$1) is reached at that wavelength; and 4) a final, asymptotic approach of spectral index $\alpha$ to an optically thin, non-thermal, constant negative value (Weiler et al. 1986, 1990).

The observed RSNe can in general be represented by the Chevalier (1982a, b) ``mini-shell'' model, which involves the acceleration of relativistic electrons and enhanced magnetic field necessary for synchrotron emission, arising from the SN shock interacting with a relatively high-density CSM, which has been ionized and heated by the initial SN UV/X-ray flash. The CSM is presumed to be the pre-SN mass loss in the late stages of the progenitor's evolution. The rapid rise in radio flux density results from the shock overtaking progressively more of the ionized wind matter, leaving less of it along the line-of-sight to absorb the emission from the shock region. The slow decline in flux density at each wavelength after the peak is then due to the SN shock expanding into generally lower density regions of the now-optically thin CSM.

This model has been parameterized by Weiler et al. (1986, 1990) as:

$S = K_1 (\nu/5GHz)^{\alpha} (t-t_0)^{\beta} e^{-\tau} \left(\frac{1-e^{-\tau^{\prime}}}{\tau^{\prime}}\right)\ \mbox{mJy, }$

$\tau = K_2 (\nu/5GHz)^{-2.1} (t-t_0)^{\delta} \ , \mbox{ and}$

$\tau^{\prime} = K_3 (\nu/5GHz)^{-2.1} (t-t_0)^{\delta^{\prime}}$.

K₁, K₂, and K₃ then formally correspond to the unabsorbed flux density (K₁), uniform (K₂) and non-uniform (K₃) optical depths, respectively, at 5 GHz one day after the explosion date t₀. The term $e^{-{\tau}}$ describes the attenuation of a local medium with optical depth $\tau$ that uniformly covers the emitting source (``uniform external absorption''), and the $(1-e^{-\tau'}) \tau'^{-1}$ term describes the attenuation produced by an inhomogeneous medium with optical depths distributed between 0 and $\tau^\prime$ (``clumpy absorption''). All absorbing media are assumed to be purely thermal, ionized hydrogen with opacity $\propto \nu^{-2.1}$. The parameters $\delta$ and $\delta'$ describe the time dependence of the optical depths for the local uniform and non-uniform media, respectively.

Normally $0 > \delta > \delta^\prime$, so that $\tau^\prime$ is the dominant opacity when $(t-t_0) \le (K_3/K_2)^{1/(\delta-\delta^{\prime})} \mbox{ days}$. At later times, the dominant opacity is $\tau$ until the CSM becomes optically thin and the radio emission is described by its characteristic power law decline with index $\beta$. In both Fig. 2.25 and 2.26 we show the model fits to the observed data.

As more radio information has become available, some interesting variations in this model have appeared, including clumpiness in the CSM, variations in mass-loss rates (and, thus, stellar evolution phase in the last few thousand years before explosion), and possibly synchrotron self-absorption (SSA) in the earliest phases of the SN evolution (Chevalier 1998) - the only example of SSA known outside of compact galactic nuclei and quasars.

However, even with the considerable improvement in VLA sensitivity over the past 20 years, the field of RSN studies is still very much sensitivity limited. More than 100 nearby SN events have been observed in the radio, with only a detection rate of 1/4, and we have only been able to develop even partial, multi-frequency radio light curves for fewer than half of those detected.

With more than 1300 SNe which have been discovered optically since the first modern SN discovery, SN 1885A (S Andromeda) in M31, there is insufficient radio sensitivity, even with the VLA, to have a chance of detecting even a small fraction of them. Such sensitivity limitations restrict the scope of most RSN studies to distances smaller than the Virgo cluster, a cosmologically insignificant distance.

Furthermore, because of sensitivity limitations, the statistics of radio emission from different types of optical SNe is very poor, with only 6 examples of SNe Ib/c and no examples of SNe Ia ever detected. Even the generally radio-brighter Type II SNe have only a dozen detections and fewer than half of that number have well measured, multi-frequency radio light curves.

New Observations Possible with the SKA

Radio Emission from Type Ia SNe

With the SKA our RSNe studies would enter a new era. We would be able to monitor RSNe at a practical threshold up to distances ten times further than is currently possible. As a result, statistics for both the Type II and Type Ib/c RSNe would substantially increase. At the very least, we would establish detections for a far larger sample of objects.

An aspect where the SKA would greatly advance RSNe studies and our general understanding of SNe is in the possible detection of radio emission from Type Ia SNe. These are the luminous objects currently serving as powerful cosmological probes out to $z \sim 1$ and providing interesting constraints on $\Omega_{\rm M}$ and $\Omega_{\rm\Lambda}$ (Riess et al. 1998; Perlmutter et al. 1998). Yet, we still have no idea what stars are giving rise to Type Ia SNe, although we suspect theoretically that they involve the deflagration or detonation of a white dwarf in a mass-transfer binary system.

Some scenarios for Type Ia SNe progenitor systems (see, e.g., Branch et al. 1995) would result in a CSM around the SN, which could produce faint radio emission, currently below the sensitivity limit for the VLA. The level of the SN shock/CSM interaction for Type Ia SNe, and its implication on the nature of the progenitor system, awaits a more sensitive radio array for these SNe to reach out to at least Virgo Cluster distances.

RSN Distance Determinations

Weiler et al. (1998) have presented evidence that the radio emission from SNe may have quantifiable properties which allow for distance determinations. Type Ib and Ic RSNe, based on a statistically very small sample of only four objects, may be approximate radio ``standard candles,'' with 6 cm peak luminosities of $\sim 19.9 \times 10^{26}$ and $\sim 6.5 \times 10^{26}$ erg s^-1 Hz^-1, respectively. Type II RSNe, based on a small sample of twelve objects, appear to obey a relation $L_{\rm 6\ cm\ peak} \simeq 5.5 \times 10^{23}\ (t_{\rm 6\ cm\ peak} - t_0)^{1.4}$ erg s^-1 Hz^-1 (with time in days) (Fig. 2.27). Thus, measurement of the radio turn-on time ( $t_{\rm 6\ cm\ peak} - t_0$) and peak flux density $S_{\rm 6\ cm\ peak}$ can yield a luminosity estimate and therefore a distance.

 

The reality of the ``standard radio candle'' hypothesis for Type Ib and Type Ic RSNe may be tested simply through the study of more objects, but since Type Ib/c SNe occur relatively infrequently nearby, higher sensitivity is needed to study more objects at larger distances. While the same is true for Type II SNe, some examples of the class are bright enough that $\sim$1 per year can presently be detected in the radio to slowly increase the available statistics. For the fainter Type II SNe, however, there exists a large gap in our knowledge between the very faint, somewhat oddball SN 1987A ($\sim$ 3 x 10²³ erg s^-1 Hz^-1 at 6 cm peak; which could only be detected in the radio because it was extremely nearby in the LMC), and the faintest of the normal RSNe, such as SN 1980K, which can be observed to the VLA sensitivity and are more than two orders-of-magnitude brighter at 6 cm peak.

Additionally, the SKA holds the possibilities for detection of the very luminous RSNe IIn at large distances. Fig. 2.28 illustrates that, at a sensitivity level of 150 nanoJy (10$\sigma$ in 8 hours), one can detect the brightest of RSNe, such as the Type IIn SN 1988Z and SN 1986J, at the z=3. Even more normal Type II RSNe, such as SNe 1979C and 1980K can be studied at cosmologically interesting distances ($z \sim 1$).

If we can extend our horizons to observe SNe up to a redshift of $\sim$1 we will fill in the gaps in our knowledge and improve the statistics for RSNe of all types, such that RSNe may eventually provide a powerful and independent technique for investigating the long-standing problem of distance estimates in astrophysics.

 

The SN-SNR Connection

Old SNe, such as SN 1968D in NGC 6946, SN 1970G in M101, and SN 1923A in M83, provide a connection to young supernova remnants (SNRs). Currently, a large gap in time exists between the oldest RSNe, such as these, and the youngest radio SNRs such as Cas A (SN $\sim$1680). (See Fig. 2.29.) Bridging this gap and understanding the connection between SNe into SNRs is vital for our understanding of the evolution of SNe, their interaction with the CSM, and their energy and chemical input into the ISM - with the resulting influence on star formation and galaxy evolution. The SKA would potentially allow detection of decades-old SNe which may still be radio emitters, but are currently well below the sensitivity limit of the VLA.

 

Summary and Conclusions

The present limit to our studies of RSNe are intimately tied to the present limitations of the VLA, such that: 1) a realistic study limit is $\sim$1 mJy peak flux density; 2) we can only detect normal SNe to $\sim$ Virgo Cluster distances ($\sim$20 Mpc); 3) we can only study very luminous RSNe to $\sim$100 Mpc; 4) there is significant delay from observing to mapping; and, 5) no realistic SN search modes are possible.

The current radio detection limit is to optical magnitude m$_{\rm v}$ $\sim$ 12 for normal Type II SNe. $\sim$1300 SNe are known, with only $\sim$25 radio detections. $\sim$150 SNe are discovered each year, with only $\sim$1-2 radio detections. The SKA could extend RSN detections to m$_{\rm v}$ $\sim$ 19, such that $\sim$50 radio detections per year would become possible. The SKA could provide better SN statistics not limited by absorption and dust, and, as a result, could discover ``hidden'' SNe. The SKA could therefore provide better galaxy SN rates, which would provide improved chemical and dynamical galaxy evolution modeling.

Radio data are vital for understanding the nature of SNe progenitor stars and stellar system, by probing the pre-SN mass loss in the late stages of the progenitor's stellar evolution. As a result, radio data place important constraints on the SNe progenitor properties and masses. Improved radio data could bridge the SN-SNR time gap.

With the SKA, normal SNe would be radio detectable to $z \sim 1$. Bright SNe would be radio detectable to $z \sim 3$. Radio distance estimates would then be available from the radio peak luminosity vs. turn-on time relation. H₀ determinations could be made independent of optical limitations. Estimation of other cosmological parameters, such as q₀ and $\Omega$, might be possible.

The current VLA is severely sensitivity limited for SN studies. The lack of on-line mapping at the VLA precludes RSN searches. The SKA would improve SN environment/progenitor studies, would improve SN statistics, would lead to improvements in our understanding of galactic chemical and dynamical evolution, and would provide independent distance and cosmological parameter estimates.