THE DIVERSITY OF MASSIVE STAR OUTBURSTS. I. OBSERVATIONS OF SN2009ip, UGC 2773 OT2009-1, AND THEIR PROGENITORS

UGC 2773 was observed with HST/WFPC2 on 1999 August 14 (Program 8192; PI: Seitzer). The observations included two exposures of 600 s each with the F606W and F814W (roughly V and I) filters. NGC 7259 was observed with HST/WFPC2 on 1999 June 29 (Program 6359; PI: Stiavelli). Exposures of 200 and 400 s were obtained with the F606W filter.

To determine whether the progenitors of SN 2009ip and UGC 2773 OT2009-1 are detected in the archival HST observations, we performed differential astrometry using optical observations of the transients. Observations of UGC 2773 OT2009-1 were obtained with the Gemini Multi-Object Spectrograph (GMOS) on the Gemini-North 8 m telescope, and the astrometry was performed using 55 objects in common with the HST/WFPC2 images resulting in an astrometric rms of σ_{GB → HST} = 24 mas in each coordinate. Observations of SN 2009ip were obtained with the Inamori Magellan Areal Camera and Spectrograph (IMACS) on the Magellan/Baade 6.5 m telescope, and the astrometry was performed using 10 objects in common with the HST/WFPC2 image resulting in an astrometric rms of σ_{GB → HST} = 38 mas in each coordinate.

The positions of the two transients on the archival HST images are shown in Figure 1. In both cases, we find a clear coincidence with objects in the archival HST images. For SN 2009ip we find an offset of 24 ± 38 mas relative to the object in the WFPC2/F606W image, while for UGC 2773 OT2009-1 we find an offset of 32  ±  24 mas relative to the object in the WFPC2/F606W and F814W images.

Zoom In Zoom Out Reset image size

The measurements of the photometry for UGC 2773 OT2009-1 and nearby stars were done using HSTphot 1.1 (Dolphin 2000). HSTphot was run using a weighted PSF-fit, which is recommended for crowded fields, and a local sky determination, which is recommended for rapidly varying backgrounds. HSTphot performs the conversion from HST/WFPC2 flight magnitudes to the Bessel magnitude system. Our astrometry and photometry for the nominal progenitor of UGC 2773 OT2009-1 and nearby stars are listed in Table 1.

We performed photometry of the point source coincident with SN 2009ip using a 05 aperture and a zero point of 22.47 mag appropriate for the F606W filter. We further applied a correction of −0.29 mag to convert to the Vega system and applied a correction for the Galactic extinction of 0.05 mag. The resulting magnitude of the source is 21.84 ± 0.25 mag. For reasonable colors (see Section 3.1.1), this corresponds to M_V = −10.3  ±  0.3 mag. We note that there is a potential systematic offset of 0.5 mag in the absolute magnitude that corresponds to the assumed distance modulus (see Footnote 8).

The stars we have identified as the progenitors are the same as those reported by Miller et al. (2009), Berger & Foley (2009), and S10. Additionally, our photometry of the progenitors is consistent with that of S10.

We obtained optical photometry of UGC 2773 OT2009-1 with the Gemini Multi-Object Spectrograph (GMOS) on the Gemini-North 8 m telescope in the gri filters. We performed the photometry using IRAF/phot with the standard GMOS zero-points.¹⁰ Our results are presented in Table 2.

We obtained near-infrared (NIR) photometry of UGC 2773 OT2009-1 with FanCam, a 1024 × 1024 HAWAII-I HgCdTe imaging system on the University of Virginia's 31 inch telescope at Fan Mountain, just outside of Charlottesville, VA (Kanneganti et al. 2009). Each epoch consists of fifteen minutes of integration in JHK_s bands, which have detection limits at the 10σ level of 0.066, 0.098, and 0.156 mJy (or 18.5, 17.5, and 16.5 mag), respectively. Individual exposures are sky-background-limited and have an integration time of either 30 or 60 s. Flat-field frames are composed of dusk and dawn sky observations. We employed standard NIR data reduction techniques in IRAF.¹¹ Because of the relatively small galaxy size, it was possible to fit the entire galaxy in a single array quadrant. Empty quadrants were efficiently utilized as sky exposures. Data were taken with the galaxy placed in each quadrant and each quadrant was reduced separately. Ultimately, all reduced quadrants were co-added. We performed photometry with IRAF's PSF package. For magnitude calibration, the transient is compared to 2MASS reference stars located in the field of view. Table 2 lists our JHK_s photometry, which is similar to the single epoch JHK_s data from S10.

We obtained UV and optical observations of SN 2009ip with the Swift UV/optical telescope on 2009 September 10. The data were processed using standard routines within the HEASOFT package. Photometry of the transient in all filters, with the exception of UVW2, was performed using a 2'' aperture to avoid contamination from nearby objects. Aperture corrections to the standard 5'' aperture were determined using isolated stars; photometry of the source in the UVW2 filter was performed using a 5'' aperture.

We observed SN 2009ip and UGC 2773 OT2009-1 with the Swift X-ray Telescope on 2009 September 10 for a total exposure time of 9.0 and 4.2 ks, respectively. No X-ray counterpart is detected at the position of either source to a limit of F_X ≲ 2.8 × 10⁻¹⁴ and ≲ 1.1 × 10⁻¹³ erg s⁻¹ cm⁻², respectively (95% limit). In both cases we assume a power-law model with an electron index of −2, and account for the Galactic neutral hydrogen column. The corresponding limits on the luminosity are L_X ≲ 1.9 × 10³⁹ and ≲ 4.8 × 10³⁸ erg s⁻¹. These limits are comparable to the X-ray emission from SNe on a similar timescale (e.g., Soderberg et al. 2008).

We observed both events with the Very Large Array¹² (VLA) following their optical discovery to search for radio counterparts, under Rapid Response programs AS1001 and AS1002 (PI: Soderberg). Our radio observations were carried out at two frequencies, 8.46 and 22.5 GHz, on dates spanning 2009 September 7.36–16.51 in the C-array antenna configuration. All observations were taken in standard continuum observing mode with a bandwidth of 2 × 50 MHz. Phase referencing was performed with calibrators J0325+469 and J2213-254, and we used 3C38 (J0137+331) for flux calibration. Data were reduced using standard packages within the Astronomical Image Processing System (AIPS).

We detect no radio sources in positional coincidence with either object and derive upper limits summarized in Table 3. At 8.5 GHz, our upper limits¹³ correspond to L_ν < 1.3 × 10²⁶ erg s⁻¹ Hz⁻¹ and L_ν < 2.6 × 10²⁴ erg s⁻¹ Hz⁻¹ for SN 2009ip and UGC 2773 OT2009-1, respectively. These limits are less luminous than an extrapolation of the observed SN 1961V radio emission at t ≈ 10 yr, to a similarly early epoch as L_ν∝t^−1.75 (Stockdale et al. 2001). We note, however, that the SN 1961V radio emission may have reached maximum intensity significantly later than our observations of UGC 2773 OT2009-1 and SN 2009ip, similar to the radio evolution of SNe IIn which typically reach maximum light several years after the explosion (van Dyk et al. 1996).

A comparison of these radio upper limits for the outbursts to the observed properties of other core-collapse SNe places them among the least luminous events, 2–4 orders of magnitude less luminous than the most powerful SNe IIn, and 4–200 times higher than the early radio signal seen for SN 1987A (Ball et al. 1995). Through this simple comparison we emphasize that radio data alone cannot distinguish between massive star outbursts and catastrophic explosions.

We obtained low- and medium-resolution spectra of SN 2009ip and UGC 2773 OT2009-1 with the MagE spectrograph (Marshall et al. 2008) and LDSS3 spectrograph¹⁴ on the Magellan Clay 6.5 m telescope, the Blue Channel spectrograph (Schmidt et al. 1989) on the MMT 6.5 m telescope, and GMOS (Hook et al. 2004) on the Gemini-North 8 m telescope. A journal of our optical spectroscopic observations can be found in Table 4.

Standard CCD processing and spectrum extraction were accomplished with IRAF. The data were extracted using the optimal algorithm of Horne (1986). Low-order polynomial fits to calibration-lamp spectra were used to establish the wavelength scale, and small adjustments derived from night-sky lines in the object frames were applied. For the MagE spectra, the sky was subtracted from the images using the method described by Kelson (2003). The GMOS data were reduced using the Gemini IRAF package (for details, see Foley et al. 2006). We employed our own IDL routines to flux calibrate the data and remove telluric lines using the well-exposed continua of the spectrophotometric standards (Wade & Horne 1988; Foley et al. 2003, 2009).

Representative spectra of SN 2009ip and UGC 2773 OT2009-1 are presented in Figure 2. Both objects have similar blue continua, but the line features are very different. SN 2009ip has few line features besides strong H Balmer lines, Na D, and He i. Although UGC 2773 OT2009-1 has a strong Hα line, it is much narrower and weaker than that of SN 2009ip. UGC 2773 OT2009-1 also displays many additional narrow line features, including lines from intermediate mass and Fe-group elements. Blueward of ∼5500 Å, UGC 2773 OT2009-1 is dominated by a forest of Fe ii lines. Finally, UGC 2773 OT2009-1 has very strong Ca ii NIR triplet lines and [Ca ii] λλ7291, 7324 lines. These features are rarely seen in classical LBV outbursts (and [Ca ii] has never been seen in one), but were distinguishing features of SN 1999bw (Garnavich et al. 1999), SN 2008S (e.g., Smith et al. 2009), and NGC 300 OT2008-1 (e.g., Bond et al. 2009; Berger et al. 2009b).

Zoom In Zoom Out Reset image size

The spectra from 2009 September 21 (corresponding to days 34 and 24 for UGC 2773 OT2009-1 and SN 2009ip, respectively) were obtained on the same night as the LRIS spectra shown by S10.

Our late-time spectrum of SN 2009ip, taken 253 days after maximum is very similar to our 86 day spectrum. The key differences is that the equivalent widths of the Balmer lines have decreased by ∼50% and their widths have increased (see Section 3.3.1).

On 2009 September 9 (22 days after maximum), we obtained a 2400 s NIR spectrum of UGC 2773 OT2009-1 with TripleSpec, a medium resolution NIR spectrograph located at Apache Point Observatory. This spectrograph is one of three NIR, cross-dispersed spectrographs covering wavelengths from 1 to 2.4 μm simultaneously at a resolution of ∼3500 (Wilson et al. 2004; Herter et al. 2008). We collected eight, 300 s sky-background-limited exposures, for a total integration time of 2400 s. We extracted the spectrum with a modified version of the IDL-based SpexTool (Cushing et al. 2004). This tool removes any contribution from the underlying galactic arm by fitting the background with a 2nd order polynomial.

3.1.1. SN 2009ip

We use the absolute magnitude, M_V = −10.3 ± 0.3 mag, of the progenitor of SN 2009ip, along with an estimated range of V − I colors of −0.05 to 1.4 mag (representative of LBV colors spanning from O to F spectral types), to plot it as a line on a color–magnitude diagram (Figure 3). For this progenitor we find a Milky Way extinction of E(B − V) = 0.019 mag (A_V = 0.05 mag; Schlegel et al. 1998). We adopt a distance modulus of μ = 32.05 mag for NGC 7259 (see discussion in Section 1), and assume no additional host galaxy or circumstellar extinction.

Zoom In Zoom Out Reset image size

In Figure 3, we compare the color of the SN 2009ip progenitor to the non-rotating, standard mass-loss evolutionary tracks of the Geneva group (Schaller et al. 1992). From this plot we can place a lower initial mass limit of 60 M_☉ on the progenitor of SN 2009ip in the absence of a color estimate for this progenitor, the higher-mass evolutionary tracks all coincide with its estimated location on the color–magnitude diagram, precluding us from placing an upper limit on this initial mass estimate. Figure 3 assumes a solar metallicity for these tracks; however, we find that our progenitor mass prediction is consistent across the full range of metallicities accommodated by the Geneva evolutionary tracks (Z = 0.05 Z_☉ to Z = 2 Z_☉). It should be noted that an increased amount of extinction, from the host galaxy or circumstellar environment, could also effectively increase the estimated initial mass of this progenitor.

S10 estimated the initial mass for the progenitor of SN 2009ip to be 50–80 M_☉. Although the HST photometry from S10 is the same as that presented here, their assumed distance modulus is 0.50 mag smaller than our assumed value. They also make no color correction to transform the F606W measurements into V. Despite these differences, the two mass ranges are nearly identical.

3.1.2. UGC 2773 OT2009-1

3.2.1. SN 2009ip

We present the 24 and 86 day spectra of SN 2009ip in Figure 4. In the upper panel of Figure 4, the 24 day spectrum is compared to the 2 day spectrum of the LBV outburst SN 1997bs (Van Dyk et al. 2000). Both objects have blue continua, strong and narrow H Balmer lines, and He i and Fe ii emission lines. Unlike SN 1997bs, SN 2009ip has a particularly strong He i λ5876 line (with some possible contribution from Na D), and all H Balmer lines and He i λ5876 show strong absorption features with minima blueshifted by ∼3000 km s⁻¹ (see Section 3.3.1 for a detailed discussion of this high-velocity absorption).

Zoom In Zoom Out Reset image size

We have reduced our spectra with many different extraction regions and backgrounds, with the high-velocity absorption features are present in all reductions. The same feature is present in all spectra taken with the MMT on days 23 and 24, which were taken with different gratings and wavelength regions. It is also present on day 86, but with a different velocity. The absorption is present for all Balmer lines and He i λ5876. S10 concurrently detected the high-velocity absorption features, giving additional strength to the existence of the features.

The bottom panel of Figure 4 displays the spectra of SN 2009ip and SN IIn 1998S (after subtracting a 10,000 K blackbody spectrum from both) from 86 and 25 days, respectively. An inverse-variance weighted Gaussian filter (with a width of 1000 km s⁻¹) has been applied to the spectrum of SN 2009ip (Blondin et al. 2006). This filtering will smear out features with intrinsic widths less than 1000 km s⁻¹, but will appropriately smooth features on larger scales. The high-velocity absorption in the 86 day spectrum of SN 2009ip is at a higher velocity than at 24 days. At this epoch, the velocity of the fast-moving SN 2009ip ejecta are very similar to that of SN 1998S. Although the H Balmer emission lines are much stronger in SN 2009ip, most other features are similar in the two spectra. In particular, SN 2009ip shows the H Balmer, He i, Sc ii, and Fe ii features seen in SN 1998S. SN 2009ip is missing the strong absorption at 6250 Å that is attributed to Si ii in SN 1998S (Leonard et al. 2000). This feature may be the result of a significant amount of nuclear burning, and thus not present in the ejecta of SN 2009ip.

The high-velocity absorption feature is not present in our spectrum from 253 days, despite it having a comparable signal-to-noise ratio as the other spectra. It is likely that by this point the ejecta have become nebular, and similar to an SN II at late time, the P-Cygni absorptions are no longer visible.

3.2.2. UGC 2773 OT2009-1

As discussed in Section 2.5, UGC 2773 OT2009-1 has a spectrum with narrow Hα emission, [Ca ii] emission, and P-Cygni absorption from many intermediate-mass and Fe-group elements. Perhaps the most distinguishing feature compared to other massive star outbursts is the [Ca ii] emission. In Figure 5, we compare the 15 day spectrum of UGC 2773 OT2009-1 to spectra of the low-luminosity transients NGC 300 OT2008-1 (Berger et al. 2009b) and SN 2008S (Smith et al. 2009), as well as SN IIn 1994W (Chugai et al. 2004); all of these objects have [Ca ii] emission in their spectra.

Zoom In Zoom Out Reset image size

All spectra in Figure 5 are relatively similar. The continuum of each spectrum is well described by a blackbody spectrum, with all four objects having a similar temperature. Each object has a prominent Hα emission line, with UGC 2773 OT2009-1 having a narrower line than the other objects. Additionally, SN 1994W has a strong Hα absorption line blueward of its emission peak.

NGC 300 OT2008-1 and SN 2008S are very similar objects with massive (10–25 M_☉), dusty progenitors (Prieto 2008; Prieto et al. 2008; Berger et al. 2009b; Bond et al. 2009; Botticella et al. 2009). Smith et al. (2009) first noticed the similarities of the spectra of these objects with those of the yellow hypergiant IRC+10240 (e.g., Jones et al. 1993; Humphreys et al. 2002). Although UGC 2773 OT2009-1 shares some spectroscopic properties with these two transients and IRC+10240 (see S10 for additional discussion), the latter objects lack the forest of absorption lines in UGC 2773 OT2009-1. These lines are reminiscent of an F-type supergiant. The P-Cygni profiles of these lines and the hydrogen Balmer emission are very similar to S Dor (e.g., Massey 2000) or other LBVs (e.g., Humphreys & Davidson 1994) during a cool phase.

SN 1994W was very luminous at peak (M_V ≈ −19 mag), but generated at most 0.03 M_☉ of ⁵⁶Ni (Sollerman et al. 1998). Dessart et al. (2009) presented an alternative method of producing the photometric and spectroscopic properties of this object: the collision of two massive hydrogen shells ejected from the star with no core collapse. Spectra of SNe IIn are rather heterogeneous (see Figure 5 of Smith et al. 2010 for a comparison of various objects), and SN 1994W is relatively distinct for its narrow absorption features. Given the spectral similarity between UGC 2773 OT2009-1 and SN 1994W, the strict upper limit of ⁵⁶Ni mass in SN 1994W, and the alternative model of Dessart et al. (2009), one must further question if SN 1994W destroyed its progenitor star.

3.2.3. Contrasting SN 2009ip and UGC 2773 OT2009-1

In this section, we examine the line profiles of Hα and Ca lines. These features provide an indication of the kinematics of the emitting material. The narrow lines are a tracer of the pre-shock circumstellar material, while the high-velocity absorption features in the spectra of SN 2009ip probe the outburst ejecta.

3.3.1. Hα

In Figure 6, we present the Hα line profiles of UGC 2773 OT2009-1 and SN 2009ip. Three separate epochs are shown for each object. Both objects have asymmetric line profiles. There are absorption components at about −350 km s⁻¹ for UGC 2773 OT2009-1 and between −3000 and −6000 km s⁻¹ for SN 2009ip. The line profile of SN 2009ip has a different shape and is much broader than that of UGC 2773 OT2009-1. We have attempted to fit these line profiles, but because of the asymmetry of the profiles, we first fit only the redshifted portion of each line profile and then add an absorption component to reproduce the blueshifted profile.

Zoom In Zoom Out Reset image size

For the 15 day spectrum of UGC 2773 OT2009-1, we fit a Gaussian with FWHM = 780 km s⁻¹ to the red side of the feature. This value is twice that of the value found by S10 for a spectrum from day 22. Our spectrum is much lower resolution than that of the spectrum presented by S10, and we are not able to properly resolve the narrow emission in our spectrum. Therefore, the width of the line at this epoch is probably lower than our measured value. The 34 day spectrum of UGC 2773 OT2009-1 is contaminated by host-galaxy emission lines, making a fit to the inner regions of the line profile problematic. Ignoring this region, we were able to fit the redshifted portion of the line profile with a single Gaussian with FWHM = 590 km s⁻¹. The 96 day spectrum has lower resolution, but is successfully fit by a Gaussian profile with FWHM = 470 km s⁻¹.

To account for the asymmetric profile, we add an absorption component to the Gaussian line profiles. Fitting the full profile with two Gaussian functions, the emission component fit to the red side of the line and the absorption component added to fit the blue side of the line, we find absorption minima at −180, −110, and −80 km s⁻¹ for the 15, 34, and 96 day spectra, respectively. This is different from the value of the actual minimum (−350 km s⁻¹) since the relatively strong emission masks the true minimum. S10 uses the observed minimum (−350 km s⁻¹) as the wind outflow velocity for UGC 2773 OT2009-1, while we suggest that it is slightly slower. Regardless, their conclusion that the velocity is consistent with the winds of blue supergiants, LBVs, and classical LBV eruptions still stands even with this lower velocity.

The line profiles of the first two spectra (days 4 and 24) of SN 2009ip are well fit by Lorentzian profiles with FWHM = 780 km s⁻¹ and the third is best fit by a Lorentzian profile with FWHM = 890 km s⁻¹, which are larger than that found by S10, 550 km s⁻¹. A Lorentzian profile of 550 km s⁻¹ is not a particularly bad fit to our data, but we find that the larger velocities better represent the data. One can also see in Figure 8 of S10, that the 550 km s⁻¹ Lorentzian slightly underpredicts the true FWHM of the line, so the data appear to be consistent.

In the 24 day spectrum of SN 2009ip, we see an absorption feature with a minimum at a velocity of about −3000 km s⁻¹. (This high-velocity absorption is seen for all Balmer lines with varying instrument configurations and on two epochs; see Section 3.2.1.) This feature is well fit by including a Gaussian absorption component with a minimum at −2800 km s⁻¹. Adding a component with this velocity also improves the fit to the 4 day Hα profile slightly, but not in a significant way. The 86 day spectrum shows an even stronger high-velocity absorption component with the minimum of the absorption at a larger velocity of −4800 km s⁻¹. The blue wing of the absorption component, representing the fastest moving material, corresponds to a velocity of about −4500 and −7000 km s⁻¹ for the 24 and 86 day spectra, respectively.

These velocities are much larger than the wind speed of LBVs and are larger than the measured velocity for any LBV eruption with the exception of the 1843 eruption of η Car, which had some material expelled at 3000–6000 km s⁻¹ (Smith 2008). The velocities measured for SN 2009ip are similar to that of the ejecta of typical core-collapse SNe (such as SN 1998S; see Figure 4 and Section 3.2.1) and are somewhat similar to that of Wolf-Rayet winds (e.g., Abbott & Conti 1987). S10 concurrently discovered this high-velocity material and arrived at similar conclusions. We discuss the implications of these features in Section 4.2.

The 253 day spectrum of SN 2009ip does not contain an absorption component, which is probably the result of the ejecta becoming nebular. The emission component is very wide with an FWHM ≈ 1000 km s⁻¹. It cannot be well fit by only a Lorentzian or Gaussian profile, but a combination of the two provides an excellent fit. Our best fit consists of a Lorentzian profile with FWHM = 1060 km s⁻¹ and a Gaussian profile with FWHM = 2750 km s⁻¹; however, we caution that this fit is not strictly unique, and there may be an unresolved narrow component that affects the profile. Regardless, the high velocities of the emission component at late time further suggest fast-moving ejecta.

3.3.2. Permitted and Forbidden Ca ii

Only our first and last spectra of SN 2009ip cover the Ca ii NIR triplet, and no spectrum shows obvious [Ca ii] λλ7291, 7325 lines, similar to the spectra presented by S10. Furthermore, the Ca H&K lines are confused by the strong Balmer sequence in SN 2009ip. Because of these factors, it is difficult to evaluate the characteristics of the Ca ii behavior in this object (other than the absent [Ca ii] lines).

UGC 2773 OT2009-1, on the other hand, has strong Ca ii features. This can be seen in Figure 2. We examine the Ca H&K, [Ca ii] λλ7291, 7325, and Ca ii NIR triplet line profiles in Figure 7. The Ca H&K lines show a broad absorption extending from −1000 to +500 km s⁻¹ and a minimum at about −50 km s⁻¹ that does not appear to change significantly between the two epochs. Each component of the Ca ii NIR triplet shows a strong P-Cygni profile with a minimum at approximately −250 km s⁻¹, slightly larger than the minima of Ca H&K.

Zoom In Zoom Out Reset image size

The [Ca ii] λλ7291, 7325 lines are visible in all epochs of our spectroscopy. We confirm the additional line between this doublet seen by S10 and identify this as Fe ii λ7308. Both [Ca ii] lines have asymmetric profiles in all spectra; the peak is at zero velocity, but the emission extends further to the red than to the blue. The lines from all epochs have FWHMs of ∼400 km s⁻¹, which is about half the width of Hα (see Section 3.3.1), similar to that found for NGC 300 OT2008-1 (Berger et al. 2009b).

Using our available photometry and spectroscopy, we can examine the SED of both objects. We have only optical spectra of SN 2009ip, which limits our ability to examine multiple blackbody components for this object. A 10,000 K blackbody fits our optical spectra well, which is consistent with that found by S10.

Our single epoch of Swift photometry occurred during the dramatic fading of the light curve immediately following maximum brightness (S10). In Figure 8, the Swift photometry is combined with the unfiltered photometry (approximately R band) of S10 (with an uncertainty of 0.5 mag to account for the 16 hr difference in the epoch of the observations) during the minimum. We overplot the 23 day spectrum for comparison. The optical photometry is consistent with the optical spectrum and a 10,000 K blackbody (modulo emission lines). The UVW2 flux is also consistent with this blackbody, however, the UVM2 and UVW1 measurements fall well below this curve. Although this may be the result of line blanketing, these data are also consistent with a blackbody curve with a temperature as low as 8000 K. If we ignore the UVW2 measurement, the data can be fit by a 7000 K blackbody. Although our data suggest a possible change in the SED during the fading event, the lack of necessary comparison UV data from a different epoch prevent a clear indication of a change.

Zoom In Zoom Out Reset image size

Using the 15 day optical spectrum and 22 day NIR spectrum, we are able to examine the SED of UGC 2773 OT2009-1 over nearly a decade in wavelength. Between these dates, the light curve of UGC 2773 OT2009-1 was essentially constant, having the same magnitude (within 1σ; S10). Using the long wavelengths of the NIR spectrum, our data are sensitive to any relatively low temperature thermal components.

We fit a single blackbody to these data, ignoring regions with strong line features and simultaneously fitting the scaling between the optical and NIR spectra. Doing this results in a best-fit temperature of 6800 K. This single blackbody consistently underpredicts the flux at NIR wavelengths. As a result, we have also attempted to fix the spectrum with a double blackbody model. This model, which produces a much better fit, results with temperatures T₁ = 6950 K and T₂ = 2100 K. The full spectrum and associated fits are shown in Figure 9. Although a dust emission spectrum would slightly deviate from a blackbody, a blackbody is a reasonable approximation (e.g., Smith et al. 2008), although the blackbody temperature usually overestimates the true dust temperature by up to 20% (e.g., Nozawa et al. 2008).

Zoom In Zoom Out Reset image size

The NIR SED can also be reasonably fit by free–free emission, which would be produced if there is a significant stellar wind. However, the flux from this emission would require a sustained mass-loss rate of $\dot{M} \approx 10^{-3}$ M_☉ yr⁻¹, which is an order of magnitude larger than that of typical LBVs in a short-lived shell ejection phase (Humphreys & Davidson 1994). It is therefore unlikely that free–free emission is dominating the NIR emission.

S10 noted that UGC 2773 OT2009-1 had a (photometric) NIR excess, but could not distinguish between circumstellar extinction and dust emission. To test the former case, we attempted to fit the spectrum with a single blackbody, but with an additional extinction term. With R_V fixed to 3.1, this model did not fit the data well. The model was able to sufficiently reproduce the data if we allowed R_V < 1, which is unphysical. We therefore conclude that the NIR excess is likely due to dust emission.

Scaling our spectrum to our broadband photometry, we can calibrate the blackbody flux, which in turn constrains the ratio R/D, where R is the radius of the blackbody radiation and D is the distance to the object. Using D = 6 ± 0.5 Mpc, we find that the hot and cool blackbodies have radii of (1.50  ±  0.16) × 10¹⁴ cm and (4.3  ±  0.4) × 10¹⁴ cm (13.0  ±  1.1 AU and 29  ±  2 AU), respectively. The size of the cool emitting region is of the same order of magnitude of the size of the Homunculus nebula surrounding η Car, but about an order of magnitude smaller than the pre-outburst dust radii of NGC 300 OT2008-1 (330 AU; Prieto et al. 2009; Berger et al. 2009b) and SN 2008S (80-160 AU; Berger et al. 2009b; Prieto et al. 2009; Wesson et al. 2009).

Following the prescription outlined by Smith et al. (2008 and references therein), we can measure the mass of the emitting dust. Specifically,

$$\begin{equation} M_{d} = 1.08 \times 10^{-8} \rho \left(\frac{R_{d}}{10^{14}\; {\rm cm}} \right)^{2} \left(\frac{T_{d}}{1000\; {\rm K}} \right)^{-2} M_{\odot }, \end{equation} \tag{ 1 }$$

where M_d is the dust mass, R_d is the radius of the dust, T_d is the dust temperature, and ρ is the dust density. For the values obtained from the spectra, T_d = 2100 K and R_d = 4.3 × 10¹⁴ cm, and assuming a dust grain density of ρ = 2.25 g cm⁻³, we find a dust mass of M_d ≈ 1 × 10⁻⁸ M_☉. Since there could be a significant amount of dust emission at lower temperatures, this is a lower limit on the total dust mass; however, it is worth noting that this measurement is orders of magnitude less than the dust created in some SNe (e.g., Kotak et al. 2009 and references therein). We note depending on the dust composition, the dust temperature may differ from the blackbody temperature by hundreds of degrees.

The dust is very close to the star and its temperature is near the limit of grain survival. Additionally, the blackbody fits indicate that the cooler component is outside the hot component, which presumably comes from the outburst ejecta. Given these conditions, it is very likely that pre-existing circumstellar dust was heated and is emitting as it is being vaporized, rather than newly formed post-shock dust emitting as it cools.

UGC 2773 OT2009-1 increased its optical brightness by ∼1 mag during outburst and has a cool spectrum with many narrow absorption lines and [Ca ii] emission. It occurred near a star cluster containing stars with initial masses of ∼25 M_☉ and shows evidence for a very cool (T ≈ 2100 K) thermal component that is radiated by circumstellar dust. SN 2009ip had a more massive and relatively isolated progenitor. At peak, it had risen at least 4 mag over the previous year, and its spectrum was hot and dominated by H Balmer emission lines. After having a significant fading and rebrightening over three weeks, it developed high-velocity absorption lines. Its SED is consistent with no dust emission.

The high temperature and large increase in optical luminosity for SN 2009ip indicates that it was a true giant eruption akin to the 1843 eruption of η Car. UGC 2773 OT2009-1, on the other hand, has spectral characteristics similar to that of S Dor at maximum. Although UGC 2773 OT2009-1 brightened by ∼5 mag (including the current eruption) over the last 10 years (Section 2; see also S10), which is significantly larger than the largest normal variation of S Dor stars (∼3 mag; van Genderen 2001), its optical luminosity did not increase as much as typical giant eruptions. Part of the increase in optical luminosity is probably the result of dust destruction, which would increase the optical luminosity while keeping the bolometric luminosity constant and not affecting the spectral type. In this case, UGC 2773 OT2009-1 would be more similar to typical LBV variability, but the presence of circumstellar dust (which is destroyed during the event) makes the optical variability larger than that of other LBVs undergoing a similar event.

The outbursts of NGC 300 OT2008-1 and SN 2008S both had relatively low-temperature SEDs (Berger et al. 2009b), but neither had the forest of absorption lines found in UGC 2773 OT2009-1. All three objects had [Ca ii] emission, but as suggested by S10, this may be linked to the circumstellar environment, and particularly dust destruction, rather than the event. Our observations have shown that there is dust in the circumstellar environment of the progenitor, and that it was likely pre-existing dust that is in the process of being vaporized. Additionally, SN 1999bw had [Ca ii] emission (Garnavich et al. 1999) and had an IR excess consistent with dust emission at late times (Sugerman et al. 2004). All four massive star outbursts with observed [Ca ii] emission (SN 1999bw, SN 2008S, NGC 300 OT2008-1, and UGC 2773 OT2009-1) have evidence of circumstellar dust. We do note that SN 2000ch had a ∼7000 K spectrum and an infrared excess consistent with dust emission, but no strong [Ca ii] emission was detected in the relatively low signal-to-noise spectra presented by Wagner et al. (2004). It is therefore possible to have a cool object and circumstellar dust yet not have strong [Ca ii] emission. Although all objects with [Ca ii] emission show evidence for circumstellar dust, it is possible to have [Ca ii] emission without any connection to circumstellar dust, so the presence of this feature may not be perfectly correlated with the existence of circumstellar dust.

Although UGC 2773 OT2009-1, NGC 300 OT2008-1, and SN 2008S have circumstellar dust and similar temperatures, other than the narrow H Balmer and [Ca ii] (which we link to the presence of circumstellar dust) emission, the spectra and progenitors are not particularly similar. Particularly, NGC 300 OT2008-1 and SN 2008S had relatively featureless spectra and progenitors with initial masses of 10–25 M_☉, while UGC 2773 OT2009-1 had a spectrum dominated by narrow lines and a more massive progenitor (≳ 25 M_☉). Further contrasting these events, while NGC 300 OT2008-1 and SN 2008S had relatively fast declining light curve after maximum (Bond et al. 2009; Smith et al. 2009), UGC 2773 OT2009-1 has a relatively flat light curve near maximum (S10). Additional data are necessary to determine if the outburst mechanisms in these objects are similar.

Using SN 2009ip and UGC 2773 OT2009-1 as examples, there appears to be two distinct elements that determine the observational properties of massive star outbursts. The first is the temperature of the outburst, which may be related to the increase in luminosity, the instability that causes the eruption, the width of the emission lines, and possibly the energetics of the outburst and if there is an explosion (see Section 4.2). This directly determines the shape of the optical SED, the ionization, if there is a forest of absorption lines, and possibly the shape of the line profiles, if there is high-velocity absorption. The other characteristic is the amount of circumstellar dust, which may cause strong Ca emission (and particularly [Ca ii] emission) and will determine the shape of the SED at longer wavelengths. UGC 2773 OT2009-1 and SN 2009ip would occupy very different regions of the parameter space created by these two dimensions. NGC 300 OT2008-1 and SN 2008S would be close to UGC 2773 OT2009-1, while LBV giant eruptions such as SN 1997bs would be close to SN 2009ip. Alternatively, Kashi et al. (2010) suggested that NGC 300 OT2008-1 was more similar to an LBV giant eruption.

It remains to be seen if there are hot massive star outbursts with a large amount of circumstellar dust or if there are cool massive star outbursts with little circumstellar dust. η Car has 0.125 M_☉ of dust surrounding it (Smith et al. 2003); if it were to have another giant eruption today, would it be cool? SN 1999bw had [Ca ii] emission and displayed dust emission at late times, was it hot? An IR survey of recent massive star outbursts with good spectroscopic coverage may provide these answers. In the future, optical and NIR observations may be sufficient to determine these characteristics for other massive star outbursts.

The spectra of SN 2009ip have absorption attributed to high-velocity (up to ∼7000 km s⁻¹) material (see Section 3.3.1; S10). Contrary to what is expected from a single outburst or explosion, the velocity of the absorption feature increases with time. In a typical SN, the ejecta naturally follow a Hubble law with the highest velocity material being the most distant from the explosion site. Spectral lines have a blueshifted absorption due to the scattering processes in the photosphere of the SN. Low velocity material is hidden behind the photosphere, only to be revealed at later times. As the photosphere recedes, the highest-velocity material becomes optically thin, resulting in the blueshifted velocity of a spectral line to decrease with time. Since the absorbing material must be at just slightly larger radii than the photosphere, the high-velocity material must have been ejected during the eruption. (If the absorbing material were from a previous eruption, the ejecta from the more recent eruption would have had to be moving even faster.)

It is possible that the high-velocity absorption is a component of P-Cygni features from the ejected material. The Lorentzian profile slightly underestimates the emission flux in the 86 day spectrum (see Figure 6), which may be the result of P-Cygni emission contributing to the line. Since the high-velocity absorption is coming from the ejecta, the outburst of SN 2009ip must have been extremely energetic, expelling a large amount of material at very high velocities. However, for the velocity to increase with time, either the ejecta must not follow a Hubble expansion or the radius of the photosphere (in velocity space) must somehow increase with time.

In a single explosion, the ejecta naturally follow a Hubble law; however, multiple explosions can change the velocity profile of the ejecta. If two explosions occurred in short succession, one can produce the inverted velocity gradient seen in SN 2009ip. In this toy model, the photosphere would recede into the ejecta of the first explosion, but at some point the fastest-moving ejecta of the second explosion would overtake the photosphere, increasing the velocity. If there are no other explosions, the velocity of the absorption would decrease from there.

The photometric behavior of SN 2009ip is consistent with this picture. The first explosion would produce the fast rise to maximum. As noted by S10, the timescale of the fading is much shorter than the timescales for many physical processes such as dust extinction.¹⁵ This behavior is very similar to that of SN 2000ch, which brightened by 2.1 mag in 9 days to maximum, then immediately faded by 3.4 mag in 7 days, immediately followed by a 2.2 mag rise in 4 days, after which the magnitude stayed relatively constant (Wagner et al. 2004). Spectra of SN 2000ch taken during the fading and on its plateau show no strong evidence for high-velocity ejecta, but the spectra may not be of high enough quality to see these features.

S10 hypothesized that rapid fading may have been caused by an optically thick shell being ejected after the first outburst. If this process did occur in SN 2009ip, then there are several implications: (1) the velocity of the absorption should eventually decrease, (2) the interaction of the ejecta from the two explosions could be a significant source of X-ray and radio emission, and (3) the X-rays might excite certain elements producing high-excitation lines such as He ii in optical spectra. Our X-ray limit of L_X < 4.8 × 10³⁸ erg s⁻¹ taken during the minimum is not particularly constraining. We do not detect any He ii λ4686 emission in the 4 or 24 day spectra; however, there is a low significance detection of a line consistent with He ii λ4686 emission in the 86 day spectrum. Additional spectroscopy is necessary to determine the late-time velocity gradient.