Type Ia supernovae are characterized by the absence of hydrogen and helium and the presence of processed material, mostly calcium, silicon and sulphur, in their spectra during the peak phase. At late phases the spectrum is dominated by emission lines of iron-group elements.
SNe Ia further exhibit a distinct light curve shape, extreme luminosity, absence of any appreciable amount of circumstellar material, and a lack of detectable polarization.
2.1. Observational material
Recent years have seen a major increase in reliable SN Ia data at all wavelengths. The large collection of optical data from the Calán/Tololo Supernova Survey (Hamuy et al. 1995, Hamuy et al. 1996c) has superseded the older, inhomogeneous catalogs (Barbon et al. 1973a, Leibundgut et al. 1991a). The Calán/Tololo survey has delivered 29 SNe Ia with at least one classifying spectrum and light curves in BVI. At the same time, many nearby supernovae have been observed with modern methods (reference lists can be found in Filippenko 1997a, Branch 1998, Contardo et al. 2000). Of particular interest are also the new observations of SN 1997br (Li et al. 1999), SN 1997cn (Turatto et al. 1998) and SN 1998bu (Suntzeff et al. 1999, Jha et al. 1999, Hernandez et al. 2000). A catalog of SN Ia observations has been published by Riess et al. (1999a) summarizing the data collection of 22 SNe Ia in BVRI.
Currently there are a number of supernova searches under way which will produce more densely sampled light curves and spectroscopic evolutions. Data are collected in a systematic way at several places. A large program of supernova observations has been ongoing at Asiago (Barbon et al. 1993) for many years. Bright supernovae are regularly observed in BVRI and optical spectroscopy. The group at the Center for Astrophysics is collecting data on most bright new supernovae. In addition to the published data (Riess et al. 1999a, Jha et al. 1999) several more SNe Ia have been observed in BVRI and occasionally in U. For a few objects also JHK light curves are being assembled. The robotic telescope at Lick Observatory (Richmond, Treffers, & Filippenko 1993) is now discovering new supernovae routinely. These objects are then followed in BVRI and with spectroscopy. The Supernova Cosmology Project (e.g. Perlmutter et al. 1997, 1999) has started to search for nearby SNe Ia to supplement the currently existing sample. In a massive search and follow-up program involving observatories around the globe many objects will be observed extensively in UBVRI and with optical spectroscopy in the next few years. The Mount Stromlo and Sidings Springs Abell Cluster Search has found around 50 supernovae (Reiss et al. 1998). The light curves from this search are a combination of very wide filters (the Macho VM and RM filter set, cf. Germany et al. 1999) and regular BVRI observations. Only a fraction of the objects has a spectroscopic classification. So far 6 SNe Ia have been reported (Reiss et al. 1998). A follow-on project to the Calán/Tololo Supernova Survey has been initiated recently, which concentrates on finding bright SNe Ia and follow them in UBVRIJHK and spectroscopy. Infrared photometry and spectroscopy has regularly been obtained at UKIRT and the IRTF (Spyromilio et al. 1992, Meikle et al. 1996, Bowers et al. 1997, Meikle & Hernandez 1999, Hernandez et al. 2000).
Many observations are also obtained by amateur astronomers. Collaborative efforts are undertaken by the International Supernova Network 4, the Variable Star Network 5 (VSNET), and through the Astronomy Section of the Rochester Academy of Sciences 6. The collaboration between professional and amateur astronomers is becoming a very valuable extension of supernova research. The combination of observations from all sources has played an important role in some of the recent research projects (e.g. Riess et al. 1999b).
2.1.1. SN Ia Rates
The frequency of supernovae carries important information about their parent population and the physical process which drives the explosions. Although the relative importance of this measurement has been recognized, it has been very difficult to derive good numbers for the supernova rates (for reviews see van den Bergh & Tammann 1991, Tammann 1994, Strom 1995). The main problem is the extreme rarity of supernovae which makes statistically significant samples very difficult to come by. The problem worsens when the rates are split into many supernova and galaxy subtypes leaving very few objects per sampling bin. The searches have further to consider the time for which supernovae could have been discovered. The control time is an important parameter which is typically difficult to calculate as it depends on light curve shape, absolute luminosity, and the absorption by dust in the local supernova environment (Tammann 1994, Strom 1995). Supernova rates are normally expressed as the number of supernovae per century per blue unit (i.e. solar) luminosity (typically per 1010 LB). The latter reflects the belief that supernovae are linked to the stellar population which dominates the galaxy light. The attempt to measure the SN Ia rate per H luminosity, which traces older stellar populations, seems more reasonable (van den Bergh 1990), if we believe that SNe Ia come from long-lived progenitor systems. Unfortunately, there are no good H luminosities available for large samples of galaxies. Since the total luminosity depends on the galaxy distances the supernova rates depend on H02.
Two complementary efforts can be distinguished. One approach is to collect as many supernovae as possible and then define the galaxy sample from which they emerged (Tammann et al. 1994). This is hampered by the fact that galaxies without supernova enter the sample only according to some selection criteria (e.g. contained in a given volume). The other approach is to only include supernovae which have been detected in a pre-defined galaxy sample (Cappellaro et al. 1997). This restricts the number of supernovae significantly. A new approach, which is tailored for more distant searches, is to define a galaxy luminosity function and use this information (as a function of redshift) to determine the rates (Pain et al. 1996, Reiss 1999). This may be the most efficient way to treat this problem avoiding massive redshift surveys which cover all galaxies accessible in the search area and may well require prohibitive amounts of observing time.
Other problems affecting SN rates are the internal extinction in the host galaxy. This extinction depends on the local environment and may differ considerably for the various supernova types. The extinction of course depends on the parent population of the supernovae. SNe Ia presumably originate in an older population where extinction should be less of a problem (but see below and section 3.1 for doubts on the universality of this assumption). Unless the age and initial mass function of the supernova parent population is the same as that of the dominant stellar population, the assumption that the B luminosity may be a good comparison is questionable.
The rates of SNe Ia are very low with about 1 event every 500 to 600 years for a galaxy with 1010 LB and a Hubble constant of 65 km s-1 Mpc-1 (Cappellaro et al. 1997, Reiss 1999). There is a dependence on galaxy type which shows that SNe Ia are observed less often in early type (elliptical and S0) galaxies than in spirals. This goes against the claim that they emerge from very old stellar populations. A further puzzling fact questioning the old paradigm is that there seems to be some preference of SNe Ia in star forming galaxies to lie in or near spiral arms (Bartunov et al. 1997) or at least in their vicinity (McMillan & Ciardullo 1996) which would make them of intermediate age (> 0.5 Gyr). The rate in the field and in galaxy clusters seems to vary very little (Reiss 1999).
The rates of distant (z > 0.3) supernovae have been derived only for two very small samples (Pain et al. 1996, Reiss 1999). For the distant supernovae the difficulties in calculating rates are exacerbated by the small number statistics of spectroscopically classified objects. An additional factor for the distant searches are the exact galaxy luminosities and an approach over some general luminosity functions is required (Reiss 1999). The restriction to a given filter passband (mostly the B band) becomes questionable as there are significant color changes for distant galaxies. It is hence not too surprising that the two estimates are not concordant. Larger supernova samples at high redshift have already been observed and new rates will become available soon.
2.1.2. Light curves
Light curves form one of the main information sources for all supernovae. Typically they are observed in the broad-band optical filters following the Bessell (1990) system, which combines the older Johnson (Johnson & Harris 1954) and Cousins (1980, 1981) filter passbands. The optical UBVRI bands have been observed for bright, nearby supernovae. Observations in the near infrared JHK filters have been obtained for a few supernovae only (Elias et al. 1981, 1985, Frogel et al. 1987, Meikle 2000).
Figure 1 displays the characteristic shape of SNe Ia in the various filters (Suntzeff et al. 1999, Jha et al. 1999, Hernandez et al. 2000). Observers usually use the B maximum as the zero-point for the light curves. We will follow this practice here as well.
Figure 1. Optical light and near-infrared light curves of SN 1998bu. The symbols are for different data sets (circles: Suntzeff et al. (1999); squares: Jha et al. (1999); hexagons: Hernandez et al. (2000)).
The light curves have been investigated in detail over the last decade. After the early assumption of a single time evolution (Minkowski 1964, Leibundgut 1988, Leibundgut et al. 1991a) clear differences emerged when new objects were observed in more detail. A striking example of the differences has been demonstrated by Suntzeff (1996) with the R and I light curves. Earlier indications of deviant objects had been ignored (Phillips et al. 1987, Frogel et al. 1987, Leibundgut 1988).
The exact, objective description of optical light curves has become an industry (Hamuy et al. 1996d, Riess et al. 1996a, Vacca & Leibundgut 1996, Perlmutter et al. 1997). However, no overall agreement has emerged yet.
Rise Times SNe Ia rise to maximum very fast. Only in very lucky occasions have early observations been recorded. One such case has been the occurrence of a second SN Ia within 100 days in the same galaxy (SN 1980N and SN 1981D - Hamuy et al. 1991). The new supernova could be detected on the deep photographic plates obtained to follow the late light curve of the first object. Thus the earliest observations were obtained 15.3 days before B maximum light was reached. Other observations this early were reported for a small number of SNe Ia (SN 1971G: -17 days, Barbon et al. 1973b; SN 1962A: -16 days, Zwicky & Barbon 1967; SN 1979B: -16 days, Barbon et al. 1982; SN 1999cl: -16 days, Krisciunas et al. 2000).
Densely sampled supernova searches provide the best chance to obtain very early observations. The current Lick Observatory Supernova Search has already discovered several objects very early (SN 1994ae; -13 days, Riess et al. 1999a).
A systematic search for early supernova data has been conducted by Riess et al. (1999b). The earliest reported observations are for SN 1990N (-17.9 days) and SN 1998bu (-16.7 days). It is thus clear that SNe Ia rise to B maximum in more than 18 days. The rise is very steep with about half a magnitude per day brightness increase until about 10 days before maximum (SN 1990N; Riess et al. 1999b).
The pre-maximum light curve is often approximated with a t2 function (Riess et al. 1999b, Aldering et al. 2000) assuming an expanding fireball with a very slowly changing temperature. The fit to the data demonstrates the suitability of such an assumption. Riess et al. find a rise time of about -19.5 days for SNe Ia. The rise time determined by Vacca & Leibundgut (1996) and Contardo et al. (2000) were based on a less extended data set and a different functional form.
Maximum phase The maximum phase starts about 5 days before the peak in the B filter. At this time a SN Ia has most likely reached its maximum brightness in the near-IR filters JHK (Meikle 2000). We currently have IR observations for only one SN Ia at these early phases, SN 1998bu (Meikle & Hernandez 1999, Hernandez et al. 2000). A dip in the light curve about 10 days after B maximum had been observed in JHK for other SNe Ia (Elias et al. 1985, Meikle 2000). SN 1986G possibly had the IR maxima observed just a few days before the B maximum (Frogel et al. 1987).
Although there is quite a range in relative epochs of maximum in the different filters it is clear that in most cases SNe Ia reach maximum earlier in I than in B (Contardo et al. 2000). The one object which clearly deviates is SN 1991bg. It reached I maximum about 6 days after the B maximum (Contardo et al. 2000). This is in striking contrast with the other object reported to be in a similar class, SN 1997cn, which reached maximum in all filters within a couple of days (Turatto et al. 1998).
The peak phase can be approximated fairly well by Gaussian curves (Vacca & Leibundgut 1996, Contardo et al. 2000, Pinto & Eastman 2000) or second-order polynomials (Hamuy et al. 1996d, Riess et al. 1999a).
The colors evolve very rapidly and non-monotonically around maximum. While they appear fairly constant during the pre-maximum phase, they change from blue (B - V - 0.1) at 10 days before to red (B - V 1.1) 30 days after maximum. Other colors evolve similarly, although not as strongly (V - R, R - I, Ford et al. 1993). A very strong color evolution can be seen in J - H (from -0.2 to 1.3), while the H - K changes only mildly (from 0.2 to -0.2; Elias et al. 1985, Meikle 2000), the only color where SNe Ia become bluer. In this color the difference between individual supernovae can be substantial (Meikle 2000). At maximum the typical absorption corrected B - V is about -0.07 ± 0.03. The V - I color is -0.32 ± 0.04 with a slight dependence on the light curve shape (Phillips et al. 1999).
After maximum the supernovae start to fade slowly and go into a decline at UV and blue (U and B) wavelengths. The redder wavelengths progressively show a decrease of the decline after about 20 days (V), to a shoulder (R) and a second maximum (IJHK). The epoch of the second maximum in I also correlates with other parameters, in particular the decline rate and the peak luminosity (Suntzeff 1996, Hamuy et al. 1996d, Riess et al. 1996a).
Second maximum The characterization of the decline is not easy and several methods have been proposed. Only the densely sampled and accurate photometry which became available in the last decade has allowed us to explore this part of the light curve more systematically.
A pronounced second maximum has been observed in the I and redder light curves (Ford et al. 1993, Suntzeff 1996, Lira et al. 1998, Meikle 2000). This has been a rather unexpected feature but had been pointed out already by Elias et al. (1981, 1985). The second maximum has not been characterized formally and its interpretation is still unclear (see section 4). The I light curve peaks between 21 days (SN 1994D) and 30 days (SN 1994ae) after the B maximum. The peaks, however, with quite some spread, are around 29, 25, and 21 days past B maximum for J, H, and K, respectively. The rise of the second maximum is very pronounced and amounts from dip to maximum to about 0.7 mag in J, 0.6 mag in H, and 0.4 mag in K (Elias et al. 1985, Leibundgut 1988, Meikle 2000). These values are based only on very few objects and any systematic differences could not be described. It is, however, striking to see how well the templates fit new data like SN 1998bu (Meikle 2000).
This second peak has been conspicuously absent in the I light curves of SN 1991bg (Filippenko et al. 1992b, Turatto et al. 1996) and SN 1997cn (Turatto et al. 1998), although a slight change in the B and V light curve decline rates during this phase has been reported (Leibundgut et al. 1993). The IR light curves of SN 1986G (Frogel et al. 1987) display only a plateau instead of a well formed peak.
Late declines After about 50 days the light curves settle onto a steady decline which is exponential in luminosity. The decline rates are the same for basically all SNe between 50 and ~ 120 days (Wells et al. 1994, Hamuy et al. 1996d, Lira et al. 1998). The B light curves decline by about 0.014 mag/day, the V by 0.028 mag/day, and I by 0.042 mag/day. Exceptions are SN 1986G (Phillips et al. 1987) and SN 1991bg (Turatto et al. 1996). They declined faster in B (0.019 and 0.020 mag/day, respectively). The decline in V is identical for all SNe Ia. In I SN 1991bg declined marginally slower than other SNe Ia (0.040 mag/day). The IR light curves have been observed for only a handful of objects to about 100 days past maximum (Meikle 2000). The decline rate is fairly constant for the few objects where it has been observed. It is 0.043 mag/day in J and 0.040 mag/day for H and K (Elias et al. 1985, Leibundgut 1988). These values are almost entirely based on SN 1972E, SN 1980N, and SN 1981B. Only SN 1980N, SN 1981B, and SN 1981D have been followed to about 380 days after maximum and show a more or less exponential decline out to the last observation (Elias & Frogel 1983). Data in this range are clearly missing and these epochs will be important to explore in the future.
Not many SNe Ia have been followed much further. At a phase of 150 days past B maximum a typical supernova is about 5 magnitudes below its peak brightness and many have disappeared into the glare of their host galaxy. The few objects which have been observed longer show a change of slope in the V, R, and I filters between 120 and 140 days (Fig. 2; Doggett & Branch 1985, Lira et al. 1998), when the decline slows to 0.014, 0.015, and 0.011, respectively. The decline rates after 140 days are identical for SN 1990N, SN 1992A, and SN 1991T (Suntzeff 1996, Lira et al. 1998). The B light curve maintains its previous slope also at these late phases (Minkowski 1964, Kirshner & Oke 1975, Suntzeff 1996, Lira et al. 1998).
Figure 2. Optical light curves between 100 and 200 days after maximum. Data of the following supernovae are plotted: SN 1992A (squares), SN 1994D (hexagons), SN 1991T (crosses), SN 1990N (circles), SN 1986G (triangles; V only), and SN 1989B (triangles; R and I only). The lines are fits to the data of SN 1992A.
A special case is SN 1991T which was observed out to over 1000 days. A flattening of the B, V, and R light curves after about 600 days was found (Schmidt et al. 1994) and has been observed until 2570 days after maximum so far (Sparks et al. 1999). The flattening can be explained by a light echo produced in a dust layer in front of the supernova.
Bolometric light curves Given the complex and wavelength-dependent nature of the opacity in SNe Ia it is clear that the brightness evolution in individual filter bands depends on these modulations. Physically more relevant is the total flux and its change with time. Bolometric light curves can provide exactly this. Of course, we can not construct fully bolometric light curves, but only sum over the observed flux. Since this includes the near-UV, optical and near-infrared wavelengths, such light curves are often referred to as UVOIR. We will refer to these light curves as bolometric in the following. Note that we are explicitly excluding the contributions by -rays. Since most of the flux emerges in the optical, at least during the first few weeks, the construction of bolometric light curves is possible (Suntzeff 1996, Vacca & Leibundgut 1996, Turatto et al. 1996, Contardo et al. 2000).
The contribution from the UV is expected to be less than 10% at maximum (Suntzeff 1996, Leibundgut 1996) and the IR should also not contribute significantly. The published bolometric light curves extend from about 10 days before maximum and span a little over 100 days. The most striking feature is the secondary shoulder which shows up between 20 and 40 days past maximum (Suntzeff 1996, Contardo et al. 2000) in all SNe Ia but SN 1991bg. We show here the bolometric light curve of SN 1998bu (Fig. 3; Contardo 2000). The secondary shoulder is visible about 30 days after maximum. The contribution of the near-IR passbands JHK is about 5% at peak as predicted and increases through the shoulder as the SN turns redder.
Figure 3. Bolometric light curve of SN 1998bu (Contardo 2000). The open symbols show the UBVRI integration while the filled squares display the bolometric light curve including the JHK bands. The line is the bolometric light curve derived from fitting the UBVRI filter curves individually before integration.
The peak phase of the bolometric light curve is slightly asymmetric with the rise from half the peak luminosity being slightly shorter than the decline to this brightness (Contardo et al. 2000). It takes from 7 to 11 days to double the luminosity before maximum and 10 to 15 days to halve it again. The rise to and fall from the maximum is slower for more luminous objects.
The secondary shoulder is visible in many objects, but may vary considerably in strength and duration. As with the filter passbands, the shoulder is occurring later for more slowly declining supernovae (as measured by the decline to half the luminosity) and hence the more luminous objects (Contardo 2000).
At late phases SNe Ia settle onto a decline which is very similar for all objects, with the exception of SN 1991bg. The decline rate between 50 and 80 days past maximum for the bolometric flux corresponds to 0.026 ± 0.002 mag/day, while SN 1991bg declined 0.030 mag/day at this phase.
One of the most important ingredients for any analysis of the energetics of SNe Ia is the maximum luminosity. It is also essential for the use of SNe Ia as distance indicators and the measurement of the Hubble constant (Branch & Tammann 1992, Branch 1998 and references therein). The best values are currently derived for the few nearby SNe Ia for which a distance can be determined by Cepheids. A mean value of MB = -19.5 ± 0.1 and MV = -19.5 ± 0.1 (error of the mean) for a set of 8 SNe Ia has been measured (Saha et al. 1999, Gibson et al. 2000). It has become custom to normalize all SNe Ia luminosities to a given decline rate (see section 3.1). Hence, slightly different averages can be found for analyses which make differing assumptions on absorption and perform such a normalization. A subset of five supernovae treated differently for absorption yields MB = -19.7 ± 0.1, MV = -19.6 ± 0.1 and MI = -19.3 ± 0.1 (Suntzeff et al. 1999), while another collaboration (Jha et al. 1999) found MV = -19.3 ± 0.2 after the decline rate correction, which amounts to mcorr = -0.26 globally, for four SNe Ia. The Suntzeff et al. and Jha et al. absolute magnitudes are hence the same and differ only marginally from the uncorrected values given in Saha et al. The discrepancy can be traced to the absorption corrections. Suntzeff et al. and Jha et al. apply a correction for the host galaxy absorption which is not done explicitly in Saha et al.
Apart from the systematic differences on the exact absolute value of the luminosity it is striking how small the overall scatter of the measurements is even before the light curve shape corrections are applied. The total range spans less than 0.5 magnitude in B and V (Saha et al. 1999, Gibson et al. 2000). It has to be noted that no Cepheid distance to a truly peculiar object, e.g. SN 1991bg or SN 1991T, has been measured so far. The data for NGC 4639 (SN 1991T) have been obtained and are being analyzed.
The bolometric luminosity of SNe Ia has been measured for only a handful of objects. The typical maximum luminosity these objects reach (see Table 1) is 1043 erg s-1 (Contardo et al. 2000). Faint events, like SN 1991bg, are, however, much less luminous (~ 2 × 1042 erg s-1), while the brightest objects reach > 2 × 1043 erg s-1 (SN 1991T).
For a recent, very complete, review on the optical spectra of supernovae of all types see Filippenko (1997a). SNe Ia are discussed extensively and readers are referred to this publication for optical spectra (and references). A large sample of infrared spectra is described in Meikle et al. (1996) and Bowers et al. (1997).
The evolution of a SN Ia spectrum is dominated by the changing influence of various emission and absorption lines. During the early phases until the late decline in the light curve begins the spectrum is dominated by P-Cygni lines of intermediate-mass elements. Most prominent is the Si II doublet ( 6347Å and 6371Å) with a prominent absorption of its P-Cygni profile around 6100Å and for a long time the defining feature of SNe Ia. Other prominent lines of SNe Ia near maximum light are Ca II ( 3934Å, 3968Å, and 8579Å), Si II (3858Å, 4130Å, 5051Å, and 5972Å), Mg II (4481Å), S II (5468Å and 5612Å, 5654Å), and O I (7773Å). The spectrum is scattered with low-ionization Ni, Fe, and Co lines which increase after the peak (e.g. Jeffery et al. 1992, Mazzali et al. 1993, 1995, 1997). Typical velocities observed in the lines are between 10000 and 15000 km s-1.
The spectrum below 3500Å is strongly suppressed by lines from iron-peak elements (Harkness 1991, Pauldrach et al. 1996). UV spectra have been obtained for only a few SNe Ia and only SN 1990N (Leibundgut et al. 1991b) and SN 1992A (Kirshner et al. 1993) have regular coverage. All IUE observations are available as a uniform sample (Cappellaro et al. 1995). The features in this part of the spectrum are not due to regular line formation, but are regions of suppressed line opacity (Pinto & Eastman 2000, see also section 4.2).
The near-IR spectral range is comparatively featureless. Lines of Si II (1.67µm), Ca II (1.15µm), Mg II (1.05µm) and iron-peak elements (between 1.5µm < < 1.7 µm and 2.2µm < < 2.6 µm) are observed (Wheeler et al. 1998). A debate on the possible identification of He I (1.083µm) started with the observations of SN 1994D (Meikle et al. 1996, Mazzali & Lucy 1998).
There is no appreciable polarization measured in broad-band photometry and spectra of SNe Ia (McCall et al. 1984, Spyromilio & Bailey 1993, Wang et al. 1996, 1997). With the exception of SN 1996X (Wang et al. 1997) polarized at about 0.2%, all SNe Ia have no detectable polarization in their spectra (Wang et al. 2000).
After the transition from an absorption spectrum, which is superposed on a pseudo-continuum, to a pure emission spectrum all lines can be attributed to forbidden Co and Fe transitions (Kirshner & Oke 1975, Spyromilio et al. 1992, Kuchner et al. 1994, Bowers et al. 1997, Mazzali et al. 1998, Wheeler et al. 1998). The nebular phase is dominated by the changing strength of these individual line multiplets.
Deviations Some SNe Ia have shown significant deviations from the above picture. Especially SN 1991T (Filippenko et al. 1992a, Phillips et al. 1992, Jeffery et al. 1992, Mazzali et al. 1995) and SN 1991bg (Filippenko et al. 1992b, Leibundgut et al. 1993, Turatto et al. 1996, Mazzali et al. 1997) have drawn attention to individual differences among SNe Ia.
SN 1991T developed the classic Si II and Ca II lines very late and also with diminished strength. Instead, its early spectrum was dominated by Fe III lines (Filippenko et al. 1992a, Ruiz-Lapuente et al. 1992). In the nebular phase SN 1991T was very similar to other SNe Ia (Leibundgut et al. 1993) suggesting similar excitation conditions and densities. The line widths did, however, indicate a higher expansion velocity (Spyromilio et al. 1992, Mazzali et al. 1998).
SN 1991bg on the other hand displayed an absorption trough near 4000Å which was attributed to Ti II ( 4395Å, 4444Å, and 4468Å) absorption (Filippenko et al. 1992b, Mazzali et al. 1997). The emergence of this line blend has been explained as a temperature effect (Nugent et al. 1995).
The stronger lines all show a clear velocity evolution with epoch, which differs significantly among individual supernovae (e.g. Branch et al. 1988, Leibundgut et al. 1993, Nugent et al. 1995, Patat et al. 1996). These measurements are not very reliable as they assume that the expansion velocity can be determined from the absorption trough of the P-Cygni line. Typical lines analyzed are the Si II and the Ca II doublets. High velocity carbon has been inferred from the earliest spectrum of SN 1990N blended with the Si II doublet (Fisher et al. 1997). The identification is based on the line profile, with the C II (6580Å) line formed in a detached shell. It is unclear, whether this is a regular feature of other SNe Ia as well or was special to SN 1990N.
Line strengths change among individual SNe Ia as well (Nugent et al. 1995). In particular, a range of Ca II and Si II line strengths has been found. At late phases the line widths also show differences (Mazzali et al. 1998).
2.2. Other wavelengths
There have been attempts to detect nearby SNe Ia in -rays with CGRO. These observations would measure the -rays from the nuclear decay. The COMPTEL observations of SN 1991T have yielded a possible detection (Morris et al. 1997, Diehl & Timmes 1998). Deep observations of SN 1998bu with CGRO have been obtained, but the first reports are negative. The COMPTEL upper limit clearly excludes the most luminous detonation models (Georgii et al. 2000). Also the next -ray observatory, INTEGRAL will only detect SNe Ia closer than about 10 to 15 Mpc depending on the explosion models (Timmes & Woosley 1997, Höflich et al. 1998a). Prospects for a direct calibration of he 56Ni mass hinge on chances for very nearby events.
No X-ray observations have been reported for SNe Ia. These supernovae are not expected to emit any significant radiation in this wavelength regime.
Radio observations of SNe Ia have been obtained, but no positive detection has been reported (Weiler et al. 1989, Eck et al. 1995). A total of 24 SNe Ia, including all nearby and bright objects, has been observed at radio wavelengths without a single detection (Panagia et al. 1999).
4 http://www.supernovae.net/isn.htm Back.
5 http://www.kusastro.kyoto-u.ac.jp/vsnet/SNe/SNe.html Back.
6 http://www.ggw.org/asras/snimages Back.