PTF 10fqs: A LUMINOUS RED NOVA IN THE SPIRAL GALAXY MESSIER 99

We triggered our Target-of-Opportunity (TOO) program on the 8 m Gemini-South telescope. On 2010 April 18.227, the Gemini Observatory staff observed PTF 10fqs with the Gemini Multi-Object Spectrograph (GMOS; Hook et al. 2004). The parameters for the observations were: R400 grating, order-blocking filter GG455_G039, and a 10 slit. Two 10 minute integrations centered on 6700 and 6800 Å were obtained. The two observations allowed for coverage of the gap between the chips. The package gemini gmos working in the iraf framework was used to reduce the data. The spectrum is shown in Figure 2.

Zoom In Zoom Out Reset image size

The most prominent emission feature is an intermediate width (13 Å, 600 km s⁻¹)²⁰ Hα line consistent with the recession velocity of the galaxy (2400 km s⁻¹; see below). Hβ was not detected. From this spectrum alone, we concluded that PTF 10fqs is in M99 and the intermediate line width made it unlikely to be a supernova. PTF 10fqs appeared to be a transient in the "gap," and we initiated extensive multi-band follow-up observations.

We continued to monitor the spectral evolution with the Marcario Low-Resolution Spectrograph (LRS; Hill et al. 1998) on the Hobby–Eberly Telescope (HET).²¹ We used the G1 grating, with a 2'' slit and a GG385 order-blocking filter, providing resolution R = λ/Δλ ≈ 360 over 4200–9200 Å. Data were reduced using the onedspec package in the iraf environment, with cosmic-ray rejection via the la_cosmic package (van Dokkum 2001), and with spectrophotometric corrections applied using standard-star observations (specifically, BD332642).

On May 15, we also obtained relatively higher resolution spectroscopic observations and relatively better blue coverage with the Low-Resolution Imaging Spectrograph (Oke et al. 1995) on the Keck I telescope. First, we used the 831/8200 grating centered on 7905 Å to get higher resolution spectra of the calcium lines. On the blue side, we used the 300/5000 grism to cover Ca H + K lines. For higher resolution covering the Balmer lines, we used the 600/7500 grating (centered on 7201 Å ) in conjunction with the 600/4000 grism.

The log of spectroscopic observations is given in Table 1. The spectral evolution is shown in Figure 3.

Zoom In Zoom Out Reset image size

Observations with the robotic Palomar 60 inch telescope (Cenko et al. 2006) on April 20.4 confirmed that PTF 10fqs was rising (r = 19.4 ± 0.1 mag) and red (g − r = 1.0 mag). We show the photometric evolution in gri bands in Figure 4 and Table 2. On April 27.2, the light curve peaked at r = 18.9 ± 0.1 mag corresponding to M_r = −12.3 (correcting for foreground Galactic extinction of E(B − V) = 0.039; Schlegel et al. 1998). Aperture photometry was done after image subtraction using a custom modification of the CPM algorithm, mkdifflc (Gal-Yam et al. 2004). Template images for subtraction and reference magnitudes for zero-point computation were taken from the Sloan Digital Sky Survey (Abazajian et al. 2009).

Zoom In Zoom Out Reset image size

Near-infrared images were obtained with the Peters Automated Infrared Imaging Telescope (PAIRITEL; Bloom et al. 2006), and reduced by an automated reduction pipeline. We lack sufficiently deep template images, which are free of light from PTF 10fqs, to perform reliable image subtraction. Thus, we measure the flux from the source in a small circular aperture, removing the sky with a nearby background region, and adopt a systematic error of 0.2 mag in the J and H bands and 0.3 mag in the K_s band. The values reported in Table 2 have been calibrated against the Two Micron All Sky Survey (2MASS) system (Cohen et al. 2003).

We observed PTF 10fqs with the EVLA on April 20.19–20.26 at central frequencies of 4.96 GHz and 8.46 GHz. We added together two adjacent 128 MHz subbands with full polarization to maximize continuum sensitivity. Amplitude and bandpass calibration was achieved using a single observation of J1331+3030, and phase calibration was carried out every 10 minutes by switching between the target field and the point source J1239+0730. The visibility data were calibrated and imaged in the AIPS package following standard practice.

A radio point source was not detected at the position of the transient (Figure 5). After removing extended emission from the host galaxy, the 3σ limits for a point source are 93 μJy and 63 μJy at 4.96 GHz and 8.46 GHz, respectively (Table 3). At the distance of M99, this corresponds to L_ν < 2.1 × 10²⁵ erg s⁻¹ Hz⁻¹. Comparing with the compilation in Chevalier et al. (2006), this upper limit is at the level of the faintest Type II-P (SN 2004dj; Beswick et al. 2005) and Type Ic (SN 2002ap; Berger et al. 2002) supernovae. As noted by Berger et al. (2009), the nearby NGC 300-OT was also not detected in the radio to deeper luminosity limits.

We observed PTF 10fqs with Galaxy Evolution Explorer (GALEX; Martin et al. 2005) on two consecutive orbits starting at 2010 April 24.387 (total exposure of 2846 s). All images were reduced and co-added using the standard GALEX pipeline and calibration (Morrissey et al. 2007).

To create a reference image, we co-added 22 images of M99 prior to 2005 April 2 (total exposure of 18571 s). Next, we subtracted the reference image from observations of PTF 10fqs (see Figure 6). No source is detected (Table 3). We find a 3σ upper limit of NUV 22.7 AB mag in an aperture consistent with a GALEX point source (75 × 75).

To constrain the pre-explosion counterpart, we measured the limiting magnitude at the position of PTF 10fqs in the co-added reference image. The faintest detected object consistent with being a point source within the galaxy had NUV = 20.1 AB mag. The 3σ limit based on measuring the sky rms is NUV > 21.8 AB mag.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We observed PTF 10fqs with Swift/XRT on April 20.466 for 2507.3 s and April 22.024 for 2623.5 s. No source is detected to a 3σ limiting count rate (assuming an 18'' radius) of 4.6 × 10⁻⁴ counts s⁻¹. Assuming a power-law model with a photon index of two, this corresponds to a flux limit of 1.6 × 10⁻¹⁴ erg cm⁻² s⁻¹.

A query to the Hubble Legacy archive returned HST images of M99 in the F606W (2001), F336W (2009), and F814W (2009) filters. We multidrizzled these data (PI: Regan; Proposal ID 11966) and registered our Gemini/GMOS acquisition image with the HST/WFPC2 images. Unfortunately, PTF 10fqs is just off the edge of the chip for the F606W filter image.

The total 1σ registration error, added in quadrature, was 0.59 pixels. Their sources of error are as follows: centroiding error (0.17 in x, 0.30 in y), registration error between the Gemini image and the HST/F814W image (0.19 in x, 0.44 in y), and registration error between the HST/F814W image and the HST/F336W image (0.04 in x, 0.02 in y). Hence, in Figure 7, we plot a 5σ radius of 3 pixels or 027.

Zoom In Zoom Out Reset image size

No source is detected at the location of PTF 10fqs. To estimate the limiting magnitude, we ran SExtractor and performed photometry following Holtzman et al. (1995). We find 3σ limiting Vega magnitudes of I>26.9 and U>26 in the 1800 s and 6600 s exposures, respectively.

M99 was part of the sample of the SIRTF Nearby Galaxies Survey (SINGS) galaxies (Kennicutt et al. 2003). This program undertook IRAC and MIPS imaging in 2004–2005. No point source is detected at the location of PTF 10fqs (see Figure 8). We downloaded IRAC images from the final data release of SINGS and MIPS images from the standard Spitzer pipeline. Computed upper limits (see Table 4) assume a 2 pixel aperture radius and sky rms based on a 20 × 20 pixel box at the location.

The 0.76 m Katzman Automatic Imaging Telescope (KAIT²²; Li et al. 2000; Filippenko et al. 2001) had extensively imaged M99 in the past decade—113 images in the period 1999–2010. We stacked the images in each season and find no point source at the location of PTF 10fqs. Limiting magnitudes for each season are summarized in Table 5.

DeepSky²³ (Nugent 2009) also has imaging at the position of this field over the interval 2006–2008. No point source is detected in a yearly sum of these images (see Table 5).

We fit a blackbody spectrum to the optical and near-infrared fluxes of PTF 10fqs without taking into account any local extinction. The best fit gives a lower limit on the temperature of ∼3900 K.

We combined the four spectra obtained with HET (between +5 days and +17 days). The most prominent (narrow) features in the spectra of PTF 10fqs are Hα, [Ca ii], the Ca ii near-IR triplet, Na i D, and Hβ. The measured line fluxes and equivalent widths are summarized in Table 6. The Hα FWHM is ≈930 km s⁻¹ (taking into account the instrumental resolution).

The Ca ii near-IR triplet is of particular interest. The HET spectra appear to show a flux excess longward of 8300 Å beyond that expected from a simple, low-order polynomial fit to the continuum. Together with a possible broad flux deficit near 8300 Å, the overall effect suggests a P-Cygni profile. If we fit three Gaussians, the Ca ii near-IR triplet features are broader than the [Ca ii] doublet, and quite likely even broader than the narrow component of the Hα profile. There is a surplus of flux at 8600 Å, which falls right between the 8498.02, 8542.09 Å pair and the more isolated 8662.14 Å line, such as one would expect from an underlying broad feature.

We test this hypothesis further with SYNOW (Jeffery & Branch 1990) modeling. We do not get a good fit to the overall shape of the spectrum with an extinguished blackbody of any temperature (assuming standard dust). To fit the red end of the spectrum, we need high temperature and extinction (consistent with the strong Na i D absorption). We find that in addition to narrow emission from Ca ii IR, there is also a likely underlying broad component (see Figure 9). The width (FWHM) of this feature is ≈10,000 km s⁻¹.

A caveat to this interpretation is that a similar broad feature is not seen in the Hα profile. However, as noted below (Section 6.2), reinspection of the spectra of related transients shows possible evidence of a similar broad feature. Thus, we cautiously accept the interpretation that in addition to the low-velocity outflow seen in Hα, there is a higher velocity outflow in this explosion.

The light curves of all four transients (PTF 10fqs, SN 2008S, NGC 300-OT, and M85-OT) were red and evolved slowly for the first couple of months. PTF 10fqs had a well-sampled rise (Figure 4)—it rose by 1.1 mag in the r band in 10.8 days. After maximum, PTF 10fqs declined slowly in the r band by 1 mag in 68 days. Subsequently, it evolved more rapidly, declining by the next 1.3 mag in 16 days. PTF 10fqs had g − r = 1.0 at peak and declined relatively faster in the g band (1 mag in 40 days) than the r-band. In comparison, SN 2008S declined by 1 mag in 51 days in the R band and 44 days in the V band. The epoch of maximum light is uncertain for NGC 300-OT due to lack of observations and is constrained to be anywhere between 2008 April 24 and May 15 (Bond et al. 2009). If we assume it to be April 27, the evolution is that R band and I band are similar to that for PTF 10fqs (Figure 4).

The spectral evolution of SN 2008S (Botticella et al. 2009) and NGC 300-OT (Berger et al. 2009) were very well studied as they were in very nearby galaxies. We took this opportunity to reanalyze the spectrum of M85-OT reported by Kulkarni et al. (2007)²⁴.

Armed thus, we compare and contrast the spectral features of these four transients (see Figure 10).

• 1.  
  The Hα profile of SN 2008S showed a narrow component (unshocked circumstellar material (CSM; ≈250 km s⁻¹), an intermediate component (shocked material between the ejecta and the CSM; ≈1000 km s⁻¹), and a broad component (underlying ejecta emission; ≈3000 km s⁻¹). NGC 300-OT exhibited narrow (560 km s⁻¹) and intermediate-width components (1100 km s⁻¹). M85-OT only had a narrow component (350 km s⁻¹). PTF 10fqs shows an intermediate-width component (930 km s⁻¹) in the Hα emission line.
• 2.  
  SN 2008S had an Hα/Hβ ratio that evolved from 4 to 10. NGC 300-OT had a ratio of 6, while M85-OT showed a ratio of 3.5. PTF 10fqs has a ratio of 6.5. All events show flux ratios higher than 3.1 (the expectation from Case B recombination). This may be evidence for collisional excitation (Drake & Ulrich 1980).
• 3.  
  PTF 10fqs, NGC 300-OT, and SN 2008S exhibit three calcium features: Ca ii H&K in absorption, [Ca ii] and Ca ii near-IR triplet in emission. A reanalysis of M85-OT shows Ca ii H&K, as well as lower signal-to-noise ratio detections of both [Ca ii] and Ca ii IR. Smith et al. (2009) show a similarity between the spectra of SN2008S and a Galactic hypergiant (IRC+10420) and suggest that strong [Ca ii] is due to destruction of dust grains.
• 4.  
  As noted earlier (see also Figure 9), there is evidence for a broad feature around 8600 Å in the spectrum of PTF 10fqs. Motivated by this finding, we reinspected the spectra of previous transients and find that a similar broad feature may also be present in the spectra of M85-OT and NGC 300-OT.
• 5.  
  Narrow Fe ii lines are visible in NGC 300-OT and SN 2008S. Reanalysis of M85-OT spectra possibly shows Fe ii (74) and Fe ii (40, 92).
• 6.  
  For SN 2008S, Na i D evolves from strong absorption at early times to emission at very late times. This suggests a very dense CSM. O i λ7774 is also in emission at late times. For NGC 300-OT, Na i D has a much lower equivalent width at early times, but it also evolves from absorption to emission. Neither Na i D nor O i are seen in M85-OT, but there is possibly K i in emission. PTF 10fqs has an equivalent width of Na i D of 6.4, higher than SN 2008S (2.3–4.4) and NGC 300-OT (1.0–2.1). The equivalent width of Na i D is too high to apply a standard correlation to estimate extinction.

We plot the upper limits on the pre-explosion counterpart for PTF 10fqs in Figure 11. The most constraining limits are in the optical. Following the Geneva stellar evolution tracks (Lejeune & Schaerer 2001) for unenshrouded stars, the luminosity limit of M_I>−4.3 corresponds to a progenitor mass <4 M_☉. If there was extinction of, say 1.5 mag, this would change the limit to <7 M_☉. None of SN 2008S, NGC 300-OT, M85-OT, and PTF 10fqs have an optical counterpart in deep, pre-explosion optical images. The limits in all cases are deep enough to at least rule out red supergiants and blue supergiants.

Zoom In Zoom Out Reset image size

For both SN 2008S and NGC 300-OT, an extremely red and luminous mid-infrared pre-explosion counterpart is seen (Prieto et al. 2008; Thompson et al. 2009). Recently, Khan et al. (2010) show that such progenitors are as rare as one per galaxy (and possibly associated with a very short-lived phase of many massive stars). Thus, both of these transients can be reasonably associated with massive stars. Unfortunately, the large distance to M85 and M99 means that the pre-explosion Spitzer limits on M85-OT and PTF 10fqs are not deep enough by a factor of few to constrain their progenitors to similar depths (see Figure 12).

M85-OT is located in the lenticular galaxy M85 (also in the Virgo Cluster). Fortunately, this galaxy was observed with HST for the ACS Virgo Cluster Survey as well as for a GO program. The transient is not associated with any star-forming region and the absolute magnitude of the progenitor is fainter than M_g ≈ −4 (<7 M_☉ not correcting for extinction; Ofek et al. 2008). Thus, a massive-star origin is quite unlikely.

Zoom In Zoom Out Reset image size

In contrast, SN 2008S, NGC 300-OT, and PTF 10fqs occurred in star-forming galaxies. It may be worth noting here that three supernovae (all of the core-collapse variety) have previously been discovered in the host galaxy of PTF 10fqs. It is perhaps of some significance that eight supernovae (six core collapse, two unclassified) were discovered in NGC 6946 in addition to SN 2008S. Only one supernova (of Type Ia) has been discovered in NGC 300. Small-number statistics and discovery bias (incompleteness from variety of different searches) notwithstanding, we make the suggestion that galaxies with a high supernova rate preferentially produce luminous red novae. If this suggestion is correct, then it would be worth the effort to systematically maintain close vigilance on the nearest galaxies having large supernova rates.

Kulkarni et al. (2007) suggested that V838 Mon, V4332 Sgr, and M31 RV may also be luminous, red novae. We note here that the two Galactic sources are located in star-forming regions. Specifically, V838 Mon is in a young (25 Myr) star cluster and may even have a B3 companion (Afşar & Bond 2007). V4332 Sgr (Martini et al. 1999) is located toward the inner Galaxy (in Sagittarius). On the other hand, M31 RV is located in the bulge of M31. HST observations (undertaken with WFPC2 in parallel mode) taken about a decade ago show that the immediate environs of M31-RV are typical bulge-population stars (Bond & Siegel 2006). No unusual remnant star is seen at the astrometric position of M31 RV, nor any evidence of a light echo (consistent with the absence of dense circumstellar or interstellar gas that is essential to form echoes). Separately, there is no evidence for any luminous outbursts in this area in the period 1942–1993 (Boschi & Munari 2004). Thus, M31 RV appears to have been a cataclysmic event in the bulge of M31.

Zoom In Zoom Out Reset image size