1 INTRODUCTION
Type I b and c supernovae are core-collapse supernovae (SNe) that show no hydrogen, and no hydrogen or helium, respectively, in their optical spectraFootnote 1 , and refer to them collectively as ‘Type I b/c’. I will also mention on occasion Type IIb supernovae, which are ones that show hydrogen lines early on that subsequently disappear. Type I b/c as well as IIb SNe are thought to arise from the core collapse of a massive star which has lost most or all of its envelope either due to stellar winds or due to mass-transfer to a binary companion prior to the SN explosion, and for this reason are sometimes known as ‘stripped envelope’ SNe (Clocchiatti et al. Reference Clocchiatti, Wheeler, Benetti and Frueh1996). They are of especial interest because they have been associated with gamma-ray bursts (GRB)Footnote 2 . For a review of the radio observations of GRBs, see Granot & van der Horst (2013).
In this review, I will discuss in particular the results of very long baseline interferometry (VLBI) radio observations of Type I b/c SNe. The primary reason for VLBI observations is that the unmatched resolution of VLBI provides almost the only way of obtaining spatially resolved information about the sources. VLBI can reach angular resolutions of 0.1 milli-arcsec, which corresponds to a distance of ~1 light-month at 50 Mpc. VLBI observations therefore potentially allow us to measure the size, shape, and speed of the expanding fireballs.
Generally in supernovae, the radio emission is synchrotron emission, which arises from the combination of magnetic fields and relativistic particles, both of which are produced by the shocks generated by the SN. A particular advantage of radio observations in this field is that radio emission traces the fastest ejecta, which are difficult or impossible to see optically. The explosive ejection of material in a SN generally produces a forward shock where the ejecta impact upon their surroundings, as well as a reverse shock which is driven back into the ejecta. In the well-studied case of the Type IIb SN 1993J, Bartel et al. (Reference Bartel, Bietenholz, Rupen and Dwarkadas2007) give arguments that the outer bounds of the radio-emission region are closely associated with this forward shock location.
It was generally thought that the most common progenitors of Type I b/c SNe were Wolf–Rayet stars (Van Dyk, Li, & Filippenko Reference Van Dyk, Li and Filippenko2003a; Woosley, Heger, & Weaver Reference Woosley, Heger and Weaver2002). However, recently evidence has emerged that suggests that lower mass He stars make up a substantial fraction, perhaps the majority, of Type I b/c progenitors (Eldridge et al. Reference Eldridge, Fraser, Smartt, Maund and Crockett2013; Smith et al. Reference Smith, Li, Filippenko and Chornock2011). The progenitors of both classes of SNe are thought to be similar, although there is evidence suggesting that the Type Ic progenitors may be somewhat more massive than the Type Ib ones (Kuncarayakti et al. Reference Kuncarayakti2013). However, given that to date there has been no definitive detection of a Type I b/c progenitor, such evidence must remain somewhat inconclusive [although note that Van Dyk, Li, & Filippenko (Reference Van Dyk, Li and Filippenko2003b) and Cao et al. (Reference Cao2013) report possible detections of both Type Ib and Ic progenitors].
Type I b/c SNe can exhibit very large ejection velocities, in some cases relativistic. SNe where the optical spectrum shows broad lines implying particularly large velocities (
$\mbox{\raisebox {-0.3em}{$\stackrel{\textstyle >}{\sim }$}}20 000$
km s−1) are often classed as ‘broad lined’. Type II SNe, by contrast, have not generally been seen to have high ejection velocities, with the exception of ones of Type IIb, such as SN 2003bg (Hamuy et al. Reference Hamuy2009) and SN 2011dh (Marion et al. Reference Marion2011), which do show high ejection velocities. Only 5–10% of all Type Ic SNe have broad lines.
Type I b/c SNe are particularly interesting because they have been associated with long-duration GRBs. So far, all the GRBs that have been reliably associated with a spectroscopically confirmed SN have been associated with one of Type I b/c. In the cases where the spectrum of the SN allowed a more precise classification, the SNe were of Type Ic with broad lines. Although many, perhaps all GRBs show a supernova-like ‘bump’ in their light curves, only in relatively nearby examples can the accompanying SN be identified more conclusively in the spectrum (Bersier Reference Bersier2012). Note, however, that the great majority of Type I b/c SNe are not accompanied by GRBs. In fact, even of the subset of broad-lined Type Ic SNe, only ~20% are associated with a GRB event.
The most popular model for long-duration GRBs is the so-called collapsar model, (Woosley Reference Woosley1993; MacFadyen, Woosley, & Heger Reference MacFadyen, Woosley and Heger2001), in which a core-collapse supernova produces or triggers a central engine (accreting, rapidly spinning compact object). The central engine drives a highly relativistic jet, while the more spherical SN explosion is powered by neutrinos. An observable GRB is produced when the jet is oriented close to the line of sight, and the strong Doppler boosting causes the observed strong gamma-ray emission. Such events also produce emission at lower frequencies, called the afterglow. In particular, radio afterglows are often produced. The radio emission is produced later in the evolution of a GRB, once the velocities have become sub-relativistic and the emission isotropised. Unlike the gamma-ray emission, therefore, the radio emission is not strongly beamed.
So it seems that both the stripped-envelope nature of the supernova and the presence of particularly high ejection velocities are required, but not sufficient for a SN to produce a GRB. The tentative picture that emerges is that a small subset of Type I b/c SNe are characterised by relativistic ejection velocities, and thus have broad-lined optical spectra. Some of these produce bright, cosmological (‘normal’) GRBs, which emit >1049.5 erg in gamma rays, whereas more of them produce low-luminosity bursts like GRB 0908425, which emit <1048.5 erg in gamma rays. In the case of the bright GRBs, the ejection is highly collimated (opening angles of ~0.1 radian), and is probably due to a relatively long-lived (at least tens of seconds) jet that emerges from the collapsing star. This jet occurs in conjunction with a more isotropic SN event, but in the case of an observed GRB where the jet is presumably oriented near our line of sight, the emission produced by the jet is usually brighter than that produced by the supernova.
In a bright GRB, the high-energy emission is thought to be produced by shocks internal to the jet. The emission at longer wavelengths, termed the ‘afterglow’, and in particular the radio emission, is thought to be produced by the external shocks as the jet interacts with the surrounding medium.
In the low-luminosity GRBs, the ejection is likely less collimated (opening angle
$\mbox{\raisebox {-0.3em}{$\stackrel{\textstyle >}{\sim }$}}$
1 radian), and the high-energy emission is probably associated with the supernova shock breakout (e.g. Kulkarni et al. Reference Kulkarni1998; Nakar & Sari Reference Nakar and Sari2012), and the jet, if present, does not emerge from the stellar surface, or is relatively weak compared to the more isotropic supernova shock breakout. Note that probably the majority of GRBs are of the low-luminosity variety, and only a minority of GRBs are ‘normal’, but the latter are far more easily detected (Liang et al. Reference Liang, Zhang, Virgili and Dai2007). Soderberg et al. (Reference Soderberg2006b) showed that the volumetric rate of low-luminosity GRBs is comparable to that of broad-lined Type I b/c SNe.
At present, it is not well understood why these high ejection velocities occur in some SNe, and what causes the collimation. Clearly information as to the size, shape, and speed of the expanding ejecta, such as can be obtained with VLBI, can provide important observational constraints in attempting to unravel these mysteries.
However, SNe that are sufficiently bright to be observable with VLBI occur only rarely. Only a fraction of SNe are ever detected in the radio, and of those, only for ones that are relatively nearby is VLBI imaging useful. For example, after 90 d, a SN expanding with a normal non-relativistic speed of 20000 km s−1 could be resolved with 22-GHz VLBI (resolution ~0.2 mas with a global array) out to a distance of ~5 Mpc, while one expanding with an apparent speed of c could be resolved out to ~80 Mpc. Given a model of the morphology of the radio emission, geometric models can be fit directly to data in the visibility (Fourier-transform) plane, and angular sizes determined with accuracies which depend on the signal-to-noise, but can be up to 20× higher than the nominal imaging resolution. Such angular size estimates, however, are dependent on the geometrical model assumed for the emission, which is not well known, in particular for relativistic events. This model uncertainty will usually dominate the uncertainty in the size estimate. Nonetheless, in most cases, angular size estimates can be made with a fractional accuracy of better than 50%, leading to estimates of the apparent expansion velocity of similar accuracy provided the distance and explosion date are well constrained (see Bartel et al. Reference Bartel2002; Bietenholz, Soderberg, & Bartel Reference Bietenholz, Soderberg and Bartel2009, for more detailed discussions of this process).
In the case of relativistic ejecta, it is possible that the apparent expansion speed becomes superluminal. Indeed, in the case of GRB 030329, associated with SN 2003dh, VLBI observations showing that the radio-emitting region was expanding with an apparent speed of 3–5 c during the first few months after the GRB (Taylor et al. Reference Taylor, Frail, Berger and Kulkarni2004, Reference Taylor, Momjian, Pihlström, Ghosh and Salter2005; Pihlström et al. Reference Pihlström, Taylor, Granot and Doeleman2007) constitute the most compelling evidence for relativistic expansion in GRB events.
However, given the small fraction of SNe that are sufficiently radio bright that such imaging is possible and near enough that it is useful, the number that has so far been observed is small. Furthermore, since the radio emission from Type I b/c SNe usually fades over a period of months,Footnote 3 the period over which any one SN can be observed will be limited. In this review, I will summarise the constraints that have been so far obtained through such observations.
Note that broadband radio observations of SNe which cannot be resolved by VLBI, provide a second, important albeit less direct constraint on the size of the forward shock region. The broadband spectral energy distributions of SNe typically are synchrotron self-absorbed below some turnover frequency, which is usually in the range of radio observation (0.1–100 GHz). For a given radio luminosity, the frequency at which synchrotron self-absorption becomes important depends only on the size of the emitting region, and therefore the size can be estimated if the spectral peak can be identified (provided the distance is known). This method is typically useful for measuring the size of the radio-emission region early on in the evolution of the SN (see, e.g. Chevalier & Fransson Reference Chevalier and Fransson2006). It is, however, dependent on several assumptions, including that of equipartition between relativistic electrons and magnetic field in the post-shock region. In the case of the Type IIb SN 2011dh, an estimate of size and expansion of the forward shock was obtained in this fashion, with the estimates subsequently confirmed by VLBI measurements (Bietenholz et al. Reference Bietenholz2012).
Table 1 gives an overview of the Type I b/c SNe which have been observed with VLBI to date. VLBI observations were also obtained of two further SNe, which were discovered in the radio, and are thus of uncertain type: SN 2008iz in M82, which was likely of Type II, but could have been of Type I b/c also (Brunthaler et al. Reference Brunthaler2010), and the unusual SN 1986J (see e.g. Bietenholz, Bartel, & Rupen Reference Bietenholz, Bartel and Rupen2010a), which is generally classed as Type IIn, but for which a possible Type Ib classification has also been suggested (Leibundgut et al. Reference Leibundgut, Kirshner, Pinto, Rupen, Smith, Gunn and Schneider1991). As can be seen, only a small number of SNe have been observed with VLBI, and in the remainder of this paper, I will discuss most of these relatively rare events in more detail.
Table 1 Type I b/c supernovae observed with VLBI.
a The peak spectral luminosity of the supernova at 8.4 GHz.
b ‘BL’ indicates a supernova classified as broad-lined due to the presence of broad absorption lines in the early optical spectrum.
2 SN 2001em AND OTHER ‘OFF AXIS’ AFTERGLOWS
A consequence of the idea that GRBs are produced by highly collimated jets oriented near the line of sight is that for each observed GRB, there should be numerous similar events but with the jets not oriented near the line of sight. These events, which I will call ‘off-axis’ events, and which are also known as orphan afterglows, would not produce observable gamma-rays. However, since the radio emission mostly arises once the source becomes mildly relativistic, so that its radio emission is more isotropic, the off-axis events should not be significantly more difficult to detect than on-axis ones in the radio. Radio wavelengths were shown to be good for detecting such off-axis events (e.g. Paczyński Reference Paczyński2001; Granot & Loeb Reference Granot and Loeb2003), and in a constant density medium, an off-axis jet is expected to peak between 1 and 3 yr after the event (van Eerten, Zhang, & MacFadyen Reference van Eerten, Zhang and MacFadyen2010; Granot & Loeb Reference Granot and Loeb2003). Note, however, that more recent results show that for a wind-stratified medium (with density ∝r −2), and using a reasonable range for parameters such as the energy of the explosion, the circumstellar density and the efficiency of magnetic field generation and particle acceleration at the shock, a wide variety of lightcurves can be produced, many of which are much fainter than the canonical models, and some of which overlap with the expected radio lightcurves of non-relativistic SNe (Bietenholz et al., Reference Bietenholz, De Colle, Granot, Bartel and Soderberg2013). It is therefore likely that only a small fraction of such off-axis relativistic jets will be observably bright in the radio.
Nonetheless, it was suggested that SN 2001em, at a distance of ~80 Mpc, and spectroscopically classified as a Type I b/c supernova, might possibly harbour such an off-axis GRB jet on the basis of its unusually high X-ray and radio luminosities, with the latter reaching a maximum only ~1000 d after the explosion (Granot & Ramirez-Ruiz Reference Granot and Ramirez-Ruiz2004, see also Paczyński Reference Paczyński2001). Several groups obtained VLBI observations of SN 2001em (Stockdale et al. Reference Stockdale2005; Bietenholz & Bartel Reference Bietenholz and Bartel2005; Paragi et al. Reference Paragi, Garrett, Paczyński, Kouveliotou, Szomoru, Reynolds, Parsley and Ghosh2005). The most constraining of these results was that of Bietenholz & Bartel (Reference Bietenholz and Bartel2005), obtained at t ~ 3.2 yr, where here and hereafter I will use t to indicate the time interval since shock breakout. The VLBI observations were not consistent with the relativistic expansion expected of a GRB jet, but rather suggested an average expansion velocity of 20000+7000 −12000 km s−1, typical of what is found in ordinary non-relativistic SNe.
Subsequent VLBI observations (Bietenholz & Bartel Reference Bietenholz and Bartel2007; Schinzel et al. Reference Schinzel, Taylor, Stockdale, Granot and Ramirez-Ruiz2009) have further confirmed the non-relativistic expansion, with the most recent 3σ limit on the expansion speed being 9 000 km s−1, and have also placed non-relativistic limits on the proper motion (Schinzel et al. Reference Schinzel, Taylor, Stockdale, Granot and Ramirez-Ruiz2009). Furthermore, broad Hα lines have appeared in SN 2001em's optical spectrum, prompting a re-classification to Type IIn (Soderberg, Gal-Yam, & Kulkarni Reference Soderberg, Gal-Yam and Kulkarni2004). An alternate model for SN 2001em, not involving a GRB event or any relativistic ejection, was proposed by Chugai & Chevalier (Reference Chugai and Chevalier2006), which involves the interaction of normal, non-relativistic, SN ejecta with a massive and dense circumstellar shell, produced by mass loss of the progenitor, and which seems to be able to account for the observational data.
So far, no off-axis events have been detected in blind surveys (e.g., Levinson et al. Reference Levinson, Ofek, Waxman and Gal-Yam2002; Gal-Yam et al. Reference Gal-Yam, Ofek and Poznanski2006). Note, however, that blind surveys of a sensitivity sufficient to ensure detection of off-axis events at the predicted brightness levels are still prohibitively expensive in observing time.
However, since GRBs arise in conjunction with supernovae, of which many are detected optically, the strategy of looking for off-axis GRB events accompanying known Type I b/c SNe suggested itself. So far, two different searchesFootnote 4 for late-time radio emission from Type I b/c SN did not yield any unambiguous detections: Soderberg et al. (Reference Soderberg, Nakar, Berger and Kulkarni2006a) carried out a search for late-time radio emission from Type I b/c SNe, observed 68 SNe at late times with the VLA. A further set of radio observations of 59 Type I b/c SNe was carried out by Bietenholz et al. (Reference Bersten, Tanaka, Tominaga, Benvenuto and Nomoto2013), who also did not detect any late-time radio emission due to off-axis GRB jets, and concluded that fewer than <2% of all Type I b/c SNe, and <30% of the broad-lined ones, are associated with jet comparable to those seen in bright GRB afterglows. However, they also showed that for more realistic parameters for the circumstellar density as well as the efficiency of particle acceleration and field generation, the radio emission from off-axis jets could be several orders of magnitude lower than previously predicted, and that therefore only a small minority of off-axis jets would likely produce observable radio emission, and that consequently the non-detection of such radio emission so far is readily explained even if a large fraction of SNe Type I b/c do harbour relativistic jets with energies comparable to that of GRB jets.
In Bietenholz et al.'s survey for late-time radio emission, only a single supernova, SN 2003gk, at ~45 Mpc, was reliably detected, with an 8.4-GHz spectral luminosity of ~6 × 1027 erg s−1 at t ≃ 6 yr. However, VLBI follow-up observations showed that SN 2003gk's average (over t ≃ 8 yr) expansion speed was ~10 000 km s−1, ruling out a relativistic jet, and suggesting that also in this case, interaction with the dense shell in the circumstellar medium was powering the radio emission (Bietenholz et al. Reference Bietenholz, De Colle, Granot, Bartel and Soderberg2013).
A further supernova, SN 2007bg, was also suggested to possibly harbour an off-axis jet on the basis of its radio emission (Prieto, Watson, & Stanek Reference Prieto, Watson and Stanek2009). No VLBI observations of it were undertaken due to its relatively large distance of 152 Mpc, however, also in this case, the radio (and X-ray) emission is due to interaction of non-relativistic ejecta with dense CSM resulting from episodic mass-loss from the progenitor rather than to any relativistic ejection (Salas et al. Reference Salas, Bauer, Stockdale and Prieto2013, see also Soderberg Reference Soderberg2009).
3 SN 2008D: JET OR SHOCK BREAKOUT?
SN 2008D was first discovered as the X-ray flash (XRF) 080109 with the accompanying supernova being discovered soon after in the optical (Soderberg et al. Reference Soderberg2008). No gamma-ray emission was seen even though the source was in the field-of-view of the Burst Alert Telescope. It is at a distance of ~28 Mpc, and was, as mentioned above, originally classified as of Type Ic with broad lines, but subsequently re-classified as Type Ib when narrow He lines appeared (e.g. Modjaz et al. Reference Modjaz2009).
The origin of the X-ray flash is the subject of some debate. Some authors suggested that it is of supernova-shock origin (Soderberg et al. Reference Soderberg2008; Chevalier & Fransson Reference Chevalier and Fransson2008), while others suggested that it was caused by a mildly relativistic jet, which penetrated the envelope of the progenitor (Mazzali et al. Reference Mazzali2008; Li Reference Li2008; Xu, Zou, & Fan Reference Xu, Zou and Fan2008).
Radio emission was detected from SN 2008 shortly after shock breakout. Several sets of VLBI observations were obtained. VLBI Observations were obtained with various arrays including telescopes of the NRAO Very Long Baseline Array and the European VLBI Network (EVN) as well as Arecibo, at frequencies of 22, 8.4, and 5.0 GHz, and at epochs t = 28, 30, 69, and 133 d (Bietenholz et al. Reference Bietenholz, Soderberg and Bartel2009; van der Horst et al. Reference van der Horst2011). Although the source was not definitively resolved at any of these epochs or frequencies, 3σ upper limits on the apparent expansion velocity, βapp, of 0.71 c could be setFootnote 5 from the observations at t = 133 d (Bietenholz et al. Reference Bietenholz, Soderberg and Bartel2009). I show a VLBI image of SN 2008D in Figure 1 (left). van der Horst et al. (Reference van der Horst2011) obtained similar limits on the βapp. These limits on βapp ruled out a long-lived, highly relativistic outflow, but are still compatible with a rapidly decelerating, mildly relativistic jet. Recent simulations by Bersten et al. (Reference Bersten, Tanaka, Tominaga, Benvenuto and Nomoto2013) suggest that SN 2008D's bolometric light curve cannot easily be reproduced without the presence of ~0.01 M⊙ of 56Ni-rich material in the outer layers of the ejecta, which they attribute to the action of such a jet.
Figure 1. VLBI images of SN 2008D and SN 2009bb. For both images, the lowest contour is drawn at 3× the image rms background brightness, the greyscale is labelled in μJy beam−1, the full width at half maximum of the convolving beam is indicated at lower left and north is up and East is to the left. Left: an image of SN 2008D at 5 GHz on 2008 May 21. The contours are drawn at −21, 21, 30, 40, 50, 70, and 90% of the peak brightness of 296 μJy beam−1, with the 50% contour being emphasised (from Bietenholz et al. Reference Bietenholz, Soderberg and Bartel2009). Right: an image of SN 2009bb at 8.4 GHz on 2009 June 11. The contours are at −57, 57, 70, 80, 90, and 95% of the peak brightness of 613 μJy beam−1 (see Bietenholz et al. Reference Bietenholz2010b, for details). I do not consider the extensions to the south of the supernova (visible in the greyscale, but below the 3σ contour) to be real.
4 SN 2009bb: RELATIVISTIC EJECTA?
The nearby SN for which there was the best evidence of relativistic expansion was SN 2009bb, detected by the Chilean Automatic SN Search Program (CHASE, Pignata et al. Reference Pignata2009), at a distance of ~40 Mpc in the nearby spiral galaxy NGC 3278. The shock breakout date was unusually well constrained to be March 19±1 UT. SN 2009bb showed broad lines, implying a photospheric velocity of ~25 000 km s−1, had a moderate X-ray luminosity, and was quickly found to be radio bright (Soderberg et al. Reference Soderberg2010b). Its peak 8.4-GHz spectral luminosity was ~5 × 1028 erg s−1 Hz−1, larger than that observed for any other SN I b/c at a similar time after shock breakout although only slightly higher than that of SN 1998bw.
Angular sizes derived from fitting a synchrotron self-absorption model to the broadband radio SED suggested a mean shock velocities of 0.85±0.02 c, assuming equipartition of energy between electrons and magnetic fields (Soderberg et al. Reference Soderberg2010b).
VLBI observations were undertaken at t = 85 d. Unfortunately, the declination of SN 2009bb was −40°, and the paucity of VLBI telescopes in the Southern hemisphere generally renders VLBI observations difficult, and furthermore the South African telescope at Hartebeesthoek, which provides the long baselines crucial for high resolution, was out of commission due to a bearing failure. The observations were obtained at a frequency of 8.4 GHz, when the total flux density of the SN at that frequency was 2.5 mJy, and I show the VLBI image in Figure 1 (right). Only an upper limit on the angular size could be determined, which was 0.64 mas (3σ; corresponding to a shock velocity of <1.74c; see Bietenholz et al. Reference Bietenholz2010b for details). The VLBI observations are therefore consistent with, but do not require, moderately relativistic expansion for SN 2009bb.
Chakraborti & Ray (Reference Chakraborti and Ray2011) interpreted SN 2009bb as having a baryon-loaded relativistic blast wave, launched by a central-engine-driven explosion, and predicted that it would be resolvable by VLBI by t ≃ 200 d. However, by that time, the flux density was already too low (
$\mbox{\raisebox {-0.3em}{$\stackrel{\textstyle <}{\sim }$}}1$
mJy at 8.4 GHz) to warrant further VLBI observations.
In the case of another nearby Type Ic SN, SN 2007gr, 5-GHz VLBI observations at t ≃ 25 and 85 d (Paragi et al. Reference Paragi2010) were first interpreted to suggest relativistic expansion. Relativistic expansion was unexpected since the radio luminosity of SN 2007gr was low, with the 8.4-GHz spectral luminosity peaking at ~1026 erg s−1 Hz−1, almost 1000 times lower than that of SN 1998bw. A large angular size and therefore large expansion velocity was suggested because the VLBI observations recovered only a fraction of the total flux density measured at WSRT, suggesting that a significant fraction of the flux density was on scales larger than those measured by VLBI, and therefore setting a lower limit on the size of the source. However, a re-examination by Soderberg et al. (Reference Soderberg, Brunthaler, Nakar, Chevalier and Bietenholz2010a) showed that prolonged relativistic expansion was not likely in this case. In particular, the difference between the total flux density as measured with the Westerbork Synthesis Radio Telescope and that observed even at the shortest VLBI baselines, if interpreted as being due to a resolved source, suggested a source much larger than expected even for credible relativistic expansion. Soderberg et al. (Reference Soderberg, Brunthaler, Nakar, Chevalier and Bietenholz2010a) suggested a more plausible explanation of a somewhat overestimated WSRT total flux density and moderate coherence loss in the VLBI observations. Maeda (Reference Maeda2013) also find that the radio light curves of SN 2007gr suggest non-relativistic expansion. Nonetheless, a relativistic ejection that is rapidly (<1 week) decelerated is still compatible with, although by no means required by, the VLBI measurements.
A further supernova, SN 2003L, (92 Mpc), although not showing any broad absorption lines, had a very high radio luminosity, comparable to that of SN 1998bw, which suggested that there might be relativistic ejecta present. It was observed with the NRAO VLBA at t = 65 d and the results suggested an expansion velocity of only 0.2c (Soderberg et al. Reference Soderberg, Kulkarni, Berger, Chevalier, Frail, Fox and Walker2005).
5 CONCLUSIONS
Type I b/c SNe are particularly interesting because they can have relativistic ejecta and because they have been associated with GRBs. However, the nature of the relativistic ejection and the relationship between Type I b/c SNe and GRBs are far from clear. Among the many open questions are: What causes some SNe to have a relativistic shock breakout? What causes some SNe to have a comparatively long-lived relativistic jet, which can produce a GRB? Where are the off-axis GRB jets? What is the relationship between GRBs and less extreme supernova? VLBI observations are virtually the only way to obtain spatially resolved information about the explosions, and thus directly constrain the geometry and expansion speeds of the ejection. Unfortunately, SNe that are close and radio bright enough to be accessible to VLBI are rare, to date only eight Type I b/c SNe have been observed with VLBI (see Table 1).
In none of this small sample observed so far have the VLBI observations produced conclusive evidence of relativistic expansion, or allowed the shape of the emission region to be determined. The VLBI observations have, however, ruled out long-lived relativistic jets such as are thought to produce GRBs in several cases (SN 2001em, SN 2003gk). The case of short-lived, more isotropic relativistic ejection is more difficult. Since the events are short-lived, only in the case of very nearby events does the fireball expand to a size resolvable by VLBI before it becomes too faint to observe. The evidence in these cases is consequently less conclusive, with the VLBI observations obtained so far being compatible with, but not requiring, the mildly relativistic ejecta suggested by observations at other wavelengths in the cases of SN 2008D and SN 2009bb.
Given the importance of direct observational determination of the size, speed, and shape of the expanding shock front, VLBI observations of any SNe that are sufficiently near and radio bright should be undertaken. With the increasing sensitivity due to both increasing bandwidth, such as the 2 Gbit s‒1 recording GHz bandwidth now available at the NRAO VLBAFootnote 6 (to be upgraded to 4 Gbit s‒1 in future), and the incorporation of new and refurbished telescopes into global VLBI networks (see van Langevelde Reference van Langevelde2013; Gaylard Reference Gaylard2013), individual SNe can be observed for longer times, and fainter ones are becoming accessible to VLBI observations. Although the resolution at any particular frequency is limited by the length of the baselines and therefore limited by the Earth's diameter for ground-based telescopes, higher sensitivity also allows observations at higher frequencies with a concomitant increase in angular resolution. Higher resolution could also be obtained by using space-based VLBI antennas such as RadioAstron (Kardashev et al. Reference Kardashev2013), however, due to the limited sensitivity of RadioAstron only exceptionally radio-bright SNe could be studied in this manner. The development of e-VLBI (see, e.g. van Langevelde Reference van Langevelde2009) offers quick response times, which could be crucial for resolving relatively nearby events early on in their evolution, with the first e-VLBI observations of a SN being those of the Type I b/c SN 2001em of Paragi et al. (Reference Paragi, Garrett, Paczyński, Kouveliotou, Szomoru, Reynolds, Parsley and Ghosh2005). The availability of a VLBI array, which operates full time, such as the NRAO VLBA, rather than only in sessions during part of the year, is also of importance due to the often short-lived nature of these events. It will be important also to follow up SNe not detected in the optical, since the examples of SN 2008iz (~4 Mpc) and SN 1986J (~10 Mpc) show that even relatively nearby SNe can go undetected in optical observations.
ACKNOWLEDGEMENTS
The research at York University was supported by the National Sciences and Engineering Research Council of Canada. I have made use of NASA's Astrophysics Data System Bibliographic Services. I thank N. Bartel, J. Granot, Z. Paragi and A. Gal-Yam for useful comments on the manuscript.
