CALCIUM-RICH GAP TRANSIENTS IN THE REMOTE OUTSKIRTS OF GALAXIES

We undertook deep imaging in the g-band and R-band filters at the position of PTF 09dav with the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) on the Keck I telescope on 2010 May 15.603 and July 9.584 (UT dates are used throughout this paper), 9–11 months after explosion. We registered these images with a Palomar 60 inch (P60) image of PTF 09dav. No source is detected to a 3σ limiting magnitude of R = 26.2 mag (Table 1 and Figure 3). This constrains any satellite, dwarf host to be fainter than M_R = −9.8 mag.

Zoom In Zoom Out Reset image size

We undertook imaging in the K' band with Laser Guide Star Adaptive Optics (Wizinowich et al. 2006; van Dam et al. 2006) on the Keck II telescope with the Near Infrared Camera 2 (NIRC2). On 2010 June 17.568, we obtained 10 images of 10 s co-added integrations. The zero point was derived relative to the Two Micron All Sky Survey catalog (Skrutskie et al. 2006). No source was detected to a 3σ limit of K' = 21.1 mag.

On 2009 November 11, only three months after maximum light, a spectrum taken with LRIS on Keck I revealed that PTF 09dav had become nebular. The timescale to become nebular was surprising as it was faster than that of typical supernovae by a factor of a few. Furthermore, only two emission features are seen—Hα and [Ca ii] λλ 7291, 7324 (Figure 4). The width, redshift, and flux of the lines are summarized in Table 2. The absence of the Ca ii near-infrared (IR) triplet (λλ 8498, 8542, 8662) is indicative of a low circumstellar density. No Hα was seen in the photospheric spectra presented by Sullivan et al. (2011). The presence of Hα emission is indicative of a photoionized circumburst medium, suggesting possible interaction with stellar wind. Neither oxygen (prominent in normal SN CC) nor iron lines (prominent in SNe Ia) are seen in the nebular phase. These characteristics of the nebular spectrum were unprecedented and no match could be found in supernova libraries.

Zoom In Zoom Out Reset image size

On 2010 May 31.241, PTF discovered a new transient, PTF 10iuv, at α(J2000) = 17^h16^m5427 and δ(J2000) = +31°33'517. It was initially found at R = 21.2 mag, and we monitored its brightness with the P60 in the Bgriz filters for three months (Cenko et al. 2006). Late-time photometric observations were taken with the Large Format Camera (LFC) on the Palomar 200 inch telescope and LRIS on the Keck I telescope.

Data were reduced following standard procedures and aperture photometry was performed. Photometric calibration was done relative to photometry of field stars from the Sloan Digital Sky Survey (Abazajian et al. 2009). A common set of calibration stars was chosen for the P48, P60, LFC, and LRIS data for consistency. Conversion from ugriz to the B band was done following Jordi et al. (2006).

The light curve is summarized in Figure 5 and Table 3. PTF 10iuv peaked on June 10 with r = 19.0 mag. In r band, it rapidly rose by 2 mag in 10 days, followed by a rapid decline at the rate of 1 mag in 12 days. After a month, PTF 10iuv evolved slowly at the rate of 0.02 mag day⁻¹ for three months followed by 0.005 mag day⁻¹. The color was neither extremely red nor blue. It evolved from g − i ≈ 0.4 near maximum to g − i ≈ 0.7 one month later.

Zoom In Zoom Out Reset image size

On June 7, we obtained a classification spectrum using ISIS on the William Herschel Telescope (WHT). Subsequently, we continued to monitor the evolution with the Keck I (+LRIS) and Keck II (+DEIMOS) telescopes until the spectra became completely nebular (Figure 6).

Zoom In Zoom Out Reset image size

The spectra evolved to show prominent helium features, resembling Type Ib spectra. The calcium lines become stronger with time, especially [Ca ii] relative to O i. We identify major lines both pre-maximum and post-maximum using SYNOW (Figure 7).

Zoom In Zoom Out Reset image size

As with PTF 09dav, the spectrum of PTF 10iuv became nebular in only three months. Both showed prominent calcium emission (Table 2 and Figure 4). Unlike PTF 09dav, PTF 10iuv showed [O i] emission and did not exhibit Hα emission in the nebular phase. Note that subluminous Type II SN also show a lower ratio of oxygen to calcium relative to normal SN CC (Pastorello et al. 2004), but they also show prominent hydrogen emission.

We observed PTF 10iuv with the Expanded Very Large Array twice, once on 2010 August 25.06 and again on 2011 May 12.20. We observed in the X band (8.46 GHz) and added together two adjacent 128 MHz subbands with full polarization to maximize continuum sensitivity. Amplitude and bandpass calibration was achieved using the archival flux value of J1721+3542. Phase calibration was carried out every 10 minute by switching between the target field and the point source J1721+3542. The visibility data were calibrated and imaged in the AIPS package following standard practice.

The transient was not detected with a 3σ upper limit of 189 μJy during the first epoch and 96 μJy during the second epoch. This corresponds to L_ν < 1.0 × 10²⁷ erg s⁻¹ Hz⁻¹.

To summarize, the five distinguishing characteristics of our peculiar new class of transients are (1) peak luminosity intermediate between that of novae and most supernovae, (2) faster photometric evolution (rise and decline) than that of normal supernovae, (3) photospheric velocities comparable to those of normal supernovae, (4) early evolution to the nebular phase, and (5) nebular spectra dominated by Calcium emission.

Based on available observations, the objects that satisfy the defining characteristics of this class of "calcium-rich gap" transients are SN 2005E, SN 2007ke, PTF 09dav, PTF 10iuv, and PTF 11bij. Their light curves are shown in Figure 8 and nebular spectra in Figure 4. Observed properties of these events are summarized in Table 4.

Our choice of these five characteristics is motivated by choosing the maximum set of characteristics that gives the largest number of related transients. Evidently, if we add characteristics to this list, the family size becomes smaller and we may overlook a unifying explanation. If we remove characteristics from this list, the family becomes larger but we may inadvertently introduce a diversity of progenitor channels. We discuss examples of each case below.

If we require an additional sixth property of a photospheric spectrum with prominent helium lines, only PTF 10iuv and SN 2005E would be members of the family. It is not clear whether the photospheric spectra of PTF 09dav exhibit helium (see discussion in Sullivan et al. 2011). The earliest spectra of SN 2007ke and PTF 11bij are not until +19 day and +21 day after maximum light. These two spectra already show the nebular features of prominent [Ca ii] emission common to the class, but we cannot confirm helium. Therefore, we allow a photospheric diversity but require a nebular uniformity in the class.

If in addition to calcium, we also require oxygen emission in the nebular spectrum, only PTF 10iuv, SN 2005E, and PTF 11bij would be members of this class. PTF 09dav does not show oxygen emission and SN 2007ke does not have spectra at a sufficiently late epoch to test this.

On the other hand, if we require the absence of hydrogen emission from this class, we would exclude PTF 09dav. It is the only member of this class that showed hydrogen, albeit only at late times.

It is pertinent here to discuss the case of SN 2008ha (Valenti et al. 2009; Foley et al. 2009, 2010), a peculiar transient found in an irregular galaxy. This transient satisfies four out of the five properties of this class. The only difference is that it showed extremely low photospheric velocities of ≈2000 km s⁻¹. The photospheric spectra of SN 2008ha have been compared to those of the SN 2002cx family of supernovae, and it has been suggested that SN 2008ha is a lower velocity and lower luminosity analog of this family. The SN 2002cx family has starkly different late-time spectra dominated by permitted iron lines (it has been argued that these spectra are not nebular even at +400 day, e.g., Jha et al. 2006; Sahu et al. 2008) and no evidence of calcium-richness. Therefore, the 2002cx family is likely unrelated to the class of transients discussed here.

The membership of SN 2008ha in either the SN 2002cx family or in the "calcium-rich gap" class of transients is unclear. On grounds that the extremely low velocities may be diagnostic of a different explosion mechanism, we tentatively exclude SN 2008ha as a member of the class of "calcium-rich gap" transients.

Originally, Filippenko et al. (2003) suggested that there was a class defined by only one property: prominence of calcium relative to supernovae of Type Ib. This property alone would suggest that SN 2000ds, SN 2003dg, SN 2003dr, and SN 2003H are also members of this class. As discussed in Section 6, these events were not monitored photometrically. Additionally, SN 2005cz (Kawabata et al. 2010; Perets et al. 2011), which also has a similar nebular spectrum but an extremely sparse light curve, may also become a member of this class (albeit the broad Hα emission; Kawabata et al. 2010). Another possible member of this class is a supernova offset from the galaxy UGC11021. It was discovered by MASTER (ATEL #3610) and showed some evidence of calcium-richness (ATEL #3615); neither a light curve nor a nebular spectrum are available for this transient.

Without a light curve, the ejecta mass in this classification would remain ambiguous. As discussed below, low ejecta mass is a crucial clue to distinguish this class from variants of normal SNe CC as well as SNe Ia. We note here that even if light curves were available and consistent with the properties of this class, it would not alleviate the challenge posed to many progenitor scenarios by the strong preference for remote locations. Specifically, SN 2005cz is in an old environment (Perets et al. 2011), SN 2000ds is in an elliptical galaxy, and SN 2003dr is offset from an edge-on galaxy.

In Sullivan et al. (2011), we discussed that although the photospheric spectra of PTF 09dav share some similarities to those of SNe Ia, there are several differences as follows. (1) PTF 09dav does not obey the Phillips relation (Phillips 1993). There is a break in the slope of the relation for subluminous SN Ia, but this is obeyed by all SN 1991bg-like supernovae including the least luminous member, SN 2007ax (Kasliwal et al. 2008). (2) The photospheric spectra of PTF 09dav show scandium and strontium (elements usually seen only in spectra of SN CC). (3) The nebular spectrum of PTF 09dav does not show any Fe-peak elements, which is inconsistent with nebular spectra of all SN Ia (Figure 4).

The photospheric spectra of PTF 10iuv around maximum light resemble those of SNe Ib. However, a distinguishing characteristic is that a month after maximum, calcium emission starts to become prominent. We checked spectra of 26 SNe Ib from PTF for calcium-richness. In Figure 11, we show seven spectra of SNe Ib. While the calcium near-IR triplet becomes prominent in all of the SN Ib as they evolve, the [Ca ii] doublet is especially prominent in PTF 10iuv. PTF 10vnv and PTF 10inj also show [Ca ii], but the relative flux ratio of [Ca ii] to O i is much lower than that seen in PTF 10iuv. The case of PTF 10hcw is less clear, as [Ca ii] is more prominent than O i but the supernova set soon after discovery and the photometric coverage stops before it reached maximum. In the nebular phase, the flux ratio of [Ca ii] to [O i] is also much lower in SNe Ib relative to PTF 10iuv (Figure 4). Furthermore, the light curves of SNe Ib are typically more luminous by about 2 mag and evolve more slowly by at least a factor of two.

Zoom In Zoom Out Reset image size

The light curve of PTF 10iuv is very well sampled. The rising portion of the light curve can constrain the ejecta mass and the late-time decay can constrain the radioactive mass. Therefore, we can test the hypothesis of whether this explosion is radioactively powered by ⁵⁶Ni as in SN Ia.

For supernovae with photospheric velocity, v, and rise time, t_r, the ejecta mass is M_ej∝v t²_r (Arnett 1982). Therefore, assuming the same opacity, we can derive an ejecta mass by scaling relative to a normal SN Ia with parameters 1.4 M_☉, 11,000 km s⁻¹, and 17.4 days. PTF 10iuv has a rise time of 12 days and an average photospheric velocity of 7600 km s⁻¹; hence, it has an ejecta mass of 0.46 M_☉. PTF 09dav has the same rise time but a lower velocity of 6000 km s⁻¹, giving an ejecta mass of 0.36 M_☉ (Sullivan et al. 2011). There are no data to constrain the rise time of SN 2005E. Assuming it was also 12 days and using its photospheric velocity of 11,000 km s⁻¹, we get an ejecta mass of 0.67 M_☉. Note that this is a factor of two higher than the total ejecta mass estimated by Perets et al. (2010) based on the nebular spectrum. This may not be surprising given that the nebular spectrum analysis may not have accounted for all of the helium. Moreover, higher ejecta mass for SN 2005E was expected based on light-curve modeling as well (Waldman et al. 2011).

For PTF 10iuv, given the peak luminosity of 4.6 × 10⁴¹ erg s⁻¹ and a 12 day rise, we can derive a ⁵⁶Ni mass of 0.016 M_☉. However, we find that the late-time photometry is brighter than what is expected from radioactive decay of ⁵⁶Ni (Figure 12). A possible caveat here is that there is some light from an underlying dwarf host galaxy. The last photometric point is fainter than the depth of our pre-explosion co-add. It is possible that there is some contamination in the light curve due to the flux of such a host. Therefore, we plot the dotted line in Figure 12, which subtracts the last two data points from the rest of the light curve. We are continuing to obtain even later time imaging to further test this caveat.

Zoom In Zoom Out Reset image size

Should further observations rule out host contamination, this may suggest either a radioactive species other than ⁵⁶Ni or an additional source powering the light curve. For example, a larger contribution by ⁴⁴Ti, which has a long half-life of 60 yr, may seem a promising candidate to explain the late-time slow evolution. However, if all of the high luminosity at late times were attributed to this element, it would require too large a quantity of ⁴⁴Ti (∼2 M_☉). Modeling of different combinations of other additional radioactive species (⁴⁴Cr, ⁴⁴Ti, and ⁵²Fe) in the context of helium-shell detonations on small (Waldman et al. 2011) and large CO white dwarf cores (Shen et al. 2010) has been undertaken. These efforts were limited by the missing light-curve data in the rising and very late phase of SN 2005E. The well-sampled light curve of PTF 10iuv should be able to better constrain these models.

Using the flux ratio between [Ca ii] and the Ca ii near-IR triplet, we can constrain the density given a temperature. For PTF 10iuv, the nebular spectrum at +87 day gives a ratio of 0.86 and the spectrum at +115 day gives a ratio of 1.8. For PTF 09dav, the nebular spectrum at +94 day gives a ratio of >2.2 (assumed width of 300 Å, 3σ limit is 3 × 10⁻¹⁸ erg cm⁻² s⁻¹ Å⁻¹). For SN 2005E, the ratio at +65 day is 0.55 (Perets et al. 2010). SYNOW fits to late-time spectra suggest a temperature roughly in the range 4500–5000 K. Following Figure 2 of Ferland & Persson (1989), assuming a temperature around 4500 K, the electron density is on the order of 10⁹ cm⁻³ and decreases by a factor of a few in a couple of months.

Next, we estimate the mass of the dominant species in the ejecta using the nebular spectrum.

We estimate the oxygen mass based on the luminosity of the [O i] line in the nebular phase. We can assume the high-density limit holds (>10⁶ cm⁻³) and estimate the oxygen mass as

$$\begin{eqnarray*} &&M_{O} = 10^{8}f_{[\rm O\,\mathsc{i}]} D_{\rm Mpc}^{2} e^{(2.28/T_4)} {M}_{\odot }, \end{eqnarray*}$$

following Uomoto (1986). Assuming a temperature of 4500 K, we get 0.025 M_☉ of oxygen for PTF 10iuv and 0.037 M_☉ for SN 2005E. Note that the oxygen mass is consistent with the mass derived by Perets et al. (2010). A cautionary note here is that this calculation is extremely sensitive to temperature; a difference of only 500 K in temperature changes this estimate by a factor of two.

Two months after maximum brightness, the luminosity in [Ca ii] nebular emission for SN 2005E was 2 × 10³⁹ erg s⁻¹ and the derived ejecta mass was 0.135 M_☉ (Perets et al. 2010). The [Ca ii] nebular luminosity is a factor of 2.5 smaller for PTF 10iuv and factor of 2.6 larger for PTF 09dav relative to SN 2005E. Under similar conditions, this may be representative of the range in calcium mass for these events.

The white dwarf scenario has tremendous appeal to explain "calcium-rich gap" transients for two reasons. First, the remote galactic locations, intracluster environments, and constraints on star formation of the observed members of this family point to an old population. Second, the low ejecta masses derived from the light curves and nebular spectra are in the range of what is normally derived from white dwarf explosions and not massive star explosions.

The family of transients presented here is not the standard thermonuclear explosion of a white dwarf as seen in SNe Ia. There are two major differences. First, none of these transients obey the Phillips (1993) relation. The inferred ejected masses are smaller than in SNe Ia. Second, Fe-group elements seen in all SNe Ia are absent in the nebular spectra of these explosions.

Next, we discuss several alternate explosions of white dwarfs that have been discussed in the literature to explain intermediate-luminosity and fast-evolving transients. First, we consider a ".Ia" explosion following the final helium flash in an ultracompact white-dwarf–white-dwarf binary (Bildsten et al. 2007; Shen et al. 2010). The observed rise time of "calcium-rich gap" transients is too slow for a ".Ia" explosion. The observed ejecta masses are too large for a shell detonation (Waldman et al. 2011) and suggest that the core also participated in the explosion.

Another possibility is deflagration of a sub-Chandrashekhar-mass white dwarf (Woosley & Weaver 1994), as it is also expected to have lower luminosity than an SN Ia. However, the late-time light curve and absence of Fe-peak elements in the nebular spectra are inconsistent with this model.

Woosley & Kasen (2011) derive several models for explosions of sub-Chandrashekhar white dwarfs. However, the peak absolute magnitude and rise-time range observed for "calcium-rich gap" transients are not predicted by any of these models.

Another proposed theoretical model is accretion-induced collapse (AIC) of a rapidly rotating white dwarf into a neutron star (Metzger et al. 2009). However, AIC predicts a spectrum dominated by intermediate-mass elements, much higher velocities (0.1 c), and much more rapid rise (one day) and decline (four to five days) than what is observed for Ca-rich gap transients.

The presence of helium in SN 2005E and PTF 10iuv can be explained for some white dwarf models, but the presence of hydrogen in PTF 09dav poses a major challenge to all white dwarf explosions. One possible scenario to explain late-time hydrogen in a white dwarf explosion is as follows: consider a binary where mass transfer is from a hydrogen-rich companion star onto a white dwarf. This accretion initially proceeded at a low rate, resulting in a series of nova eruptions prior to the sub-Chandrashekhar explosion or helium-shell detonation. The photons from the final explosion would eventually reach one of the previously ejected nova shells and the interaction would give Hα emission.

Quantitatively, if the shell was photoionized, the distance to this nova shell would be the speed of light multiplied by 95 days, or 2.5 × 10¹⁷ cm. Given the velocity of 1000 km s⁻¹ to traverse this distance, the nova eruption would then have occurred 78 yr prior to the supernova. Given the absence of hydrogen in early-time spectra, we consider instead collisional ionization when the shock front itself reaches the shell. The distance to the shell would then be 8.2 × 10¹⁵ cm and the nova shell would have to be ejected ∼ 950 days before the supernova. The mass of hydrogen needed to sustain the observed luminosity for an hour is 4/n  ×  10⁻⁴ M_☉, where n is the number of times the same hydrogen atom gets excited in one hour.

Yet another scenario was recently proposed by Metzger (2012) that can potentially explain a trace amount of hydrogen—tidal disruption of a white dwarf by a neutron star or stellar mass black hole. However, several other observables (e.g., velocities) are inconsistent with this model.

While the presence of five transients in remote locations is a strong clue in favor of the white dwarf scenario, the absence of any events in disks of galaxies is quite problematic. Studies of white dwarf binaries have shown that the number of white dwarfs in halos relative to that in the disk is nearly a third, larger than expected from the ratio of the overall mass or light (Ruiter et al. 2007). However, that is not sufficient to explain why there is a strong preference for remote locations.

If the progenitor is indeed a white dwarf, then there must be a characteristic of white dwarfs which can be attributed to location and affects the explosion signature. For example, the age of the white dwarf (e.g., if the timescale for evolution in a multiple star system were longer), the composition of the white dwarf (e.g., if helium white dwarfs were preferentially found in remote locations), and environmental characteristics (e.g., if metallicity/density affected the explosion or progenitor evolution timescale). One scenario that remains to be tested with detailed modeling is that at lower metallicity, even relatively lower mass stars can evolve to become white dwarfs in a Hubble time. Perhaps, explosions of these relatively lower mass white dwarfs are subluminous, fast evolving, and calcium-rich.

Given that none of the above white dwarf models easily explain all of the observed clues of "calcium-rich gap" transients, we briefly consider the massive star option. There are some explosion properties which are reminiscent of SNe CC: similarity of nebular spectra albeit with enhanced calcium, similarity of photospheric spectra to SNe Ib in the case of PTF 10iuv and SN 2005E, and the presence of hydrogen in the case of PTF 09dav.

8.2.1. Massive Star in the Outskirts?

8.2.2. Fallback Supernova?