Bright Type IIP Supernovae in Low-Metallicity Galaxies
Spencer Scott, Matt Nicholl, Peter Blanchard, Sebastian Gomez, Edo Berger
DDraft version December 18, 2018
Typeset using L A TEX modern style in AASTeX62
Bright Type IIP Supernovae in Low-Metallicity Galaxies
Spencer Scott, Matt Nicholl,
2, 3
Peter Blanchard, Sebastian Gomez, and Edo Berger Center for Astrophysics | Harvard & Smithsonian, 60 Garden Street, Cambridge, MA, 02140, USA Institute for Astronomy, University of Edinburgh, Royal Observatory, Blackford Hill, Edinburgh EH9 3HJ, UK Birmingham Institute for Gravitational Wave Astronomy and School of Physics and Astronomy, University ofBirmingham, Birmingham B15 2TT, UK
ABSTRACTWe present measurements of the pseudo-equivalent width of the Fe II λ M > −
17) dwarf host galaxies.The Fe II λ II features,with a probability of 10 − that the two samples are drawn from the same distribution. Since low-mass galaxies are expected to contain a lower fraction of metals, our findings are consistent witha metallicity dependence for Fe II λ Keywords: supernovae:general INTRODUCTIONMost stars with initial masses (cid:38) (cid:12) endtheir lives in supernova (SN) explosions whentheir iron cores collapse. The spectroscopicproperties of SNe depend on the distributionof elements in their progenitor stars, with themost common type, Type II SNe (or SNe II),being defined by strong and broad Balmer lines– thus these stars have retained their hydro-gen envelopes until the moment of explosion.Serendipitous detections of the progenitors inpre-explosion images have confirmed that SNeII result from red supergiants (Smartt 2009).Historically, SNe II have been divided intosub-classes based on the morphology of theirlight curves: either a ∼
100 day ‘plateau’ inluminosity (SN IIP), or a ‘linear’ decline (SN
Corresponding author: Spencer [email protected]
IIL). More detailed studies from a range ofauthors have revealed a richer picture, witha range of plateau durations, decline rates,and luminosities, and evidence for a more con-tinuous distribution of properties (e.g. Arcaviet al. 2012; Anderson et al. 2014; Faran et al.2014; Sanders et al. 2015; Gall et al. 2015;Gonz´alez-Gait´an et al. 2015; Pejcha & Prieto2015; Valenti et al. 2016; Rubin et al. 2016;Guti´errez et al. 2017). Along with progenitorradii and masses, and synthesised Ni mass,the presence of circumstellar material lost priorto explosion is likely responsible for some ofthis diversity. All of these factors can be influ-enced by the metallicity of the progenitor star.Therefore constraining the metallicity of SNe IIis key to fully understanding their properties.Dessart et al. (2014) showed using spec-tral models that the equivalent width of theFe II λ a r X i v : . [ a s t r o - ph . S R ] D ec Scott et al. metallicity resulting in an equivalent width oflarger absolute value at a given time from ex-plosion. This was confirmed observationallyby Anderson et al. (2016), who compared re-sults from a large sample of SN II spectra tothe metallicities indicated by their host galaxyspectra. As this sample was compiled from theliterature, it consisted primarily of SNe II inmassive host galaxies, which were all at metal-licities greater than that found in the LargeMagellanic Cloud.Taddia et al. (2016) presented an untargetedsample of SNe II from the Palomar Tran-sient Factory, spanning a wider range of hostenvironments. They found several events infaint galaxies, with lower Fe II λ Z ∼ . (cid:12) (Dessart et al. 2014).The photometric properties of SNe II alsoseem to be different at low-metallicity. Taddiaet al. (2016) noted a possible correlation sug-gesting lower-metallicity SNe II were more lu-minous. More recently, Guti´errez et al. (2018)conducted a detailed statistical study analysinga sample of SNe II selected in faint galaxies,comparing to the higher-metallicity samples ofAnderson et al. (2016) and Guti´errez et al.(2017). As well as confirming that the SNeII in low-luminosity (metal-poor) galaxies haveweaker Fe II λ II λ II λ ∼ METHODS2.1.
Sample of SNe II in faint galaxies
Our SNe were selected from the Pan-STARRS Survey for Transients (PSST; Huberet al. 2015). We prioritised these events forspectroscopic classification because they exhib-ited a contrast between the transient and hostgalaxy of (cid:38) pyraf . One spec-trum, of PS17aio, was taken from the extendedPublic ESO Spectroscopic Survey of TransientObjects (ePESSTO; Smartt et al. 2015). Red-shifts were determined from host galaxy emis-sion lines where possible; otherwise the red-shift of the best-matching spectrum from SNID(Blondin & Tonry 2007) was used. All spectrawill be made available via WISeREP (Yaron &Gal-Yam 2012) and the Open Supernova Cat-alog (Guillochon et al. 2017).2.2.
Pseudo-equivalent width measurements
The equivalent width of a spectral line is de-fined as W λ = (cid:90) (1 − F λ F ) dλ, (1)where F λ is the flux of the absorption feature,and F is the flux in the continuum. In thecase of SN spectra, the true continuum is of-ten difficult to place due to blending of manybroad absorption and P Cygni features. In-stead, it is useful to define a ‘pseudo’ equiva-lent width (pEW), where F is assumed to be right Type IIP SNe in Low-Metallicity Galaxies Figure 1.
Left: Fe II λ II pEW, and have been normalised and shifted vertically for clarity. The Fe II λ II line and continuum fit with a first-order polynomial used to measure the pEW.The upper and lower polynomials correspond to 1 σ contours around the continuum. the peak flux on either side of the absorptionline in question.For each spectrum, we measure the pEW ofthe Fe II λ ∼ − ) and the rest-frame wavelength of 5018 ˚A.We then fit a Gaussian profile to the line us-ing the least squares fit as implemented by scipy.curve fit , and integrated equation 1taking this Gaussian fit as F λ and the linearfit as F . Errors are estimated by adjustingthe continuum by ± σ , where σ is the stan-dard deviation of the flux in the regions used todefine this continuum. Figure 1 demonstratesthis method, and shows a close-up of the Fe II line for all SNe in our sample.Some of our spectra show no clear detection ofFe II absorption features. In this case, we placean upper limit on the pEW using the methodpresented in Leonard & Filippenko (2001) (seealso Graham et al. 2017), where the 3 σ upperlimit W λ (3 σ ) is defined by W λ (3 σ ) = 3∆ λ ∆ I (cid:18) W line ∆ λ × B (cid:19) / (2) where W line is the width of the absorption fea-ture, ∆ λ is the spectral resolution (rangingfrom 6-18 ˚A), and ∆ I is the 1 σ rms fluctua-tion of the normalized continuum. B is theratio of the spectral resolution to the pixelscale; B = 3 − W line = 5018 − A .2.3. Light curves
The strength of spectral lines in SNe evolvesover time as the ejecta expand and cool. Itis therefore important to determine the phaserelative to explosion at which our spectra wereobtained. We estimate this using the lightcurves from PSST. Photometry from PSST isprimarily in the r and w filters, and occasion-ally in i (all in the PanSTARRS system de-scribed by Tonry et al. 2012). To calculate ab-solute rest-frame r -band magnitudes, we calcu-lated K -corrections from our spectra using thes3 package presented by Inserra et al. (2018),using cross-filter corrections where necessary.Distances were computed assuming a Planckcosmology (Planck Collaboration et al. 2016).Photometry was also corrected for Galactic ex-tinction using the E(B-V) values estimated us-ing Schlafly & Finkbeiner (2011). However, nocorrection for internal host galaxy extinctionis applied at this point in the analysis, as thisparameter is more difficult to measure reliably.We will discuss the significance of this in sec-tion 4. Scott et al.
The light curves are shown in Figure 2, withepochs of spectroscopy marked. Andersonet al. (2014) defined three separate epochs ofSN II light curve evolution: decline from max-imum, plateau, and decline on the radioactivetail. We measure the slopes of the light curvesof the SNe in our sample to ensure that weare in the plateau regime. A linear fit to thelight curves of each SN in our sample yielded amean slope of 0.012 mag per day, with a stan-dard deviation of 0.016. This is similar to themean slope during the plateau (their ‘s ’ pa-rameter) reported by Anderson et al. (2014).This suggests that our spectra were indeed ob-tained during the plateau.We estimate the phase of the spectra, inrest frame days since explosion, and the corre-sponding uncertainty by assuming a simplifiedSN II light curve morphology with an instan-taneous rise followed by a 100 day plateau andthen a sharp drop. We determine the minimumpossible phase by assuming the first detectedpoint on the PSST light curve is the date ofthe explosion; in this case the time of the spec-trum is simply the rest-frame time since thefirst PSST data point. We determine the max-imum possible phase by assuming the last datapoint on the PSST light curve is just before theplateau phase ends; in this scenario the spec-trum would correspond to a phase of 100 daysminus the time from the spectrum until the endof the plateau. The estimated time is thus de-fined as the mean between the upper and lowerbounds, and the uncertainty is defined as thedifference between the mean and these upperand lower bounds. We use this range of datesalong with the light curve slopes to estimatethe magnitude at the middle of the plateau.2.4. Comparison sample
To understand how the properties of SNe IIin dwarf galaxies differ from those in massivegalaxies, we construct a comparison sample ofliterature SNe II using the Open SupernovaCatalog (OSC; Guillochon et al. 2017). Wefirst filtered supernovae by type (II or IIP), andthen required that each SN have at least 5 spec-tra available in the catalog, in order to selectonly well-studied events. These criteria nar-rowed the literature sample down to 24 SNe.We were specifically interested in spectra ob-tained during the middle of the plateau phase(i.e. around 50 days after light curve maxi-
Figure 2.
Light curves for all of our PSST SNe.Data have been converted to rest-frame r -band andshifted to match at t = 0. Ticks denote dates whenspectra were obtained. Time zero has been definedas the first data point for each light curve, and acorrection for cosmological time dilation applied.The light curves are normalised such that the firstpoint corresponds to 0 magnitudes. SNe PS17aio,PS17bsd, and PS17aki only have one photometricdata point, so each object is plotted only at t = 0. mum). We visually inspected all light curvesto ensure that the automatically-derived dateof maximum light from the OSC was accurate,and corrected this when necessary. The spec-trum that was closest to 50 days was then usedfor analysis.Many of the literature SNe were observedonly in the Johnson-Cousins photometricbands. For a fair comparison with our sample,we applied a cross-filter K -correction using S3(Inserra et al. 2018) to convert to rest-frame r -band in the PanSTARRS system. We againcorrected for Milky Way extinction only. Be-cause most of these SNe are relatively nearby,we used redshift-independent distances fromthe Nasa Extragalactic Database (NED). Inmost cases we used the median of the reporteddistance estimates. Some objects had a SN dis-tance from the expanding photosphere method(Schmidt et al. 1994; Poznanski et al. 2009;Rodr´ıguez et al. 2014; Tak´ats et al. 2015) thatdiffered significantly from the median distance,in which case we selected the SN distance.2.5. Statistical Tests
The parameters of interest in this study arethe pEW of Fe II λ right Type IIP SNe in Low-Metallicity Galaxies scipy.ks 2samp . The 2 sample KS test en-ables us to test the null hypothesis that the twosamples are drawn from the same distribution.In the case of the Fe II λ RESULTSFigure 3 shows a corner plot with the distri-butions of SN and host absolute magnitudesand Fe II λ Galaxy Properties
We first compare the photometric propertiesof the host galaxies. We find the mean M host in our sample is − . − . × − . We therefore confirm that ourSNe are in a significantly less luminous galaxypopulation than the control sample, and henceshould probe SN II properties at lower metal-licities.3.2. SN Spectroscopic Properties
Comparing the pEW of the Fe II λ II λ − . Combiningthis with our knowledge that the host galax-ies in our sample are significantly fainter andtherefore likely more metal-poor than the con-trol sample, this confirms that SNe II at lowmetallicity show weaker Fe II λ SN Photometric Properties
The mean M SN in our sample (estimatedat the t=50 days epoch following the methodin Section 2.3) is − .
5, with standard devi-ation 0.6. For the comparison sample, themean M SN is − . × − . While at first sightthe luminosity differences between the samplesis striking, there are important caveats due toselection effects and host galaxy extinction aswe discuss below.3.3.1. Possible selection effects in SN luminosity
At least some of the difference between thetwo samples is likely a consequence of our tar-geted search for SNe that outshine their hostgalaxies, which clearly favours brighter SNe.To quantify our selection effects, we make useof the Open Supernova Catalog. We down-loaded all SNe II in the catalog, and in thelower panels of Figure 3 we plot SN peak mag-nitudes against host magnitudes, indicatingany SNe that would have passed our selectioncuts (section 2.1). The median apparent mag-nitude of all SNe before data cuts is 18 . ± . . ± .
3. Applying thesame methods to absolute magnitudes insteadof apparent magnitudes, the medians beforeand after cuts are − . ± . − . ± . ∼ . M SN vs M host when correctedto an absolute magnitude scale. This assumesthat all SNe passing the cuts are equally likelyto be classified; a somewhat larger selection ef-fect could result if the likelihood of classifica- Scott et al. M S N Comparison SampleOur Sample M H o s t M SN M Host
Figure 3.
Top: Corner plot of Fe II λ M SN , and M host . The normalized histograms illustratethe differences in these properties measured in our sample and in the catalog sample. The trend found inpEW vs M host indicates that Fe II λ right Type IIP SNe in Low-Metallicity Galaxies The impact of host galaxy extinction
Another important factor to consider is dustextinction within the host galaxies. We havenot corrected the SNe in either the low-metallicity or comparison sample for internalextinction. More massive galaxies generally ex-hibit higher extinction. Garn & Best (2010)provide an empirical relationship between stel-lar mass and extinction, finding that an ex-tinction in H α (similar to the wavelength of r -band) of 1.5 magnitudes (i.e. the difference inbrightness between our two samples) is typicalof galaxies with stellar masses (cid:38) × M (cid:12) .Many of the literature SNe are indeed in galax-ies within this mass range.To test whether extinction alone can accountfor the sample luminosity differences, we esti-mate host galaxy extinctions, A r , from r -bandgalaxy luminosities using the relation of Garn& Best (2010) and the SDSS mass-to-light ra-tios from Kauffmann et al. (2003) ( M ∗ /L r ≈ A r = 0 . − . − . DISCUSSION4.1.
Metallicity
Our pEW results clearly support previousfindings from Dessart et al. (2014); Andersonet al. (2016); Taddia et al. (2016); Guti´errezet al. (2018). Our SNe are in faint galaxies(absolute M r > − (cid:46) . (cid:12) (though likelywith significant scatter). All the SNe II inthese galaxies had Fe II λ <
15 ˚A,
Figure 4.
Top: evolution of the pEW ofFe II λ (cid:12) , as solidlines and the PTF sample from Taddia et al. (2016)as crosses. well below typical values for SNe II in massivegalaxies. Thus we confirm that the pEW ofthe Fe II λ (cid:46) . (cid:12) , withseveral matching models at 0.1 Z (cid:12) . This is con-sistent with the metallicities estimated fromtheir host luminosities (Arcavi et al. 2010).The control sample of literature SNe are mainlyconsistent with Solar metallicity or greater.4.2. Luminosity
Interestingly, we find that the SNe in oursample may be more luminous on average thanthe literature sample, suggesting a possible re-lationship between metallicity and SN II lumi-nosity. This is clear in the M SN vs pEW panelof Figure 3. However, the significance of anysuch relation is highly sensitive to assumptionsabout extinction in the SN host galaxies.Several SNe in our sample are brighter thanany SNe in the control sample, at least indi-cating that luminous SNe II do occur at lowmetallicity. Taddia et al. (2016) also observedmore luminous SNe II at lower metallicity,finding that their low-metallicity events (blackand red points in Figure 4) were brighter by0.7 mag than the high-metallicity events (blue Scott et al. and green points). Thus the size of the effectthey measured is comparable to our findingshere.More recently, Guti´errez et al. (2018) founda correlation between the Fe II λ CONCLUSIONSIn this work, we presented a new sampleof 12 SNe II that occurred in low-luminosity( M (cid:38) −
17) host galaxies. We measured thepseudo-equivalent width of the Fe II λ II λ ∼ . Anderson, J. P., Gonz´alez-Gait´an, S., Hamuy, M.,et al. 2014, ApJ, 786, 67Anderson, J. P., Guti´errez, C. P., Dessart, L.,et al. 2016, A&A, 589, A110 Arcavi, I., Gal-Yam, A., Kasliwal, M. M., et al.2010, ApJ, 721, 777Arcavi, I., Gal-Yam, A., Cenko, S. B., et al. 2012,ApJL, 756, L30Blondin, S., & Tonry, J. L. 2007, ApJ, 666, 1024 right Type IIP SNe in Low-Metallicity Galaxies Dessart, L., Gutierrez, C. P., Hamuy, M., et al.2014, MNRAS, 440, 1856Dressler, A., Bigelow, B., Hare, T., et al. 2011,PASP, 123, 288Fabricant, D., Cheimets, P., Caldwell, N., &Geary, J. 1998, Publications of theAstronomical Society of the Pacific, 110, 79Faran, T., Poznanski, D., Filippenko, A. V., et al.2014, MNRAS, 442, 844Flewelling, H., Magnier, E., Chambers, K., et al.2016, arXiv preprint arXiv:1612.05243Gall, E. E. E., Polshaw, J., Kotak, R., et al. 2015,A&A, 582, A3Garn, T., & Best, P. N. 2010, Monthly Notices ofthe Royal Astronomical Society, 409, 421Gonz´alez-Gait´an, S., Tominaga, N., Molina, J.,et al. 2015, MNRAS, 451, 2212Graham, M. L., Kumar, S., Hosseinzadeh, G.,et al. 2017, Monthly Notices of the RoyalAstronomical Society, 472, 3437Guillochon, J., Parrent, J., Kelley, L. Z., &Margutti, R. 2017, The Astrophysical Journal,835, 64Guti´errez, C. P., Anderson, J. P., Hamuy, M.,et al. 2017, ApJ, 850, 90Guti´errez, C. P., Anderson, J. P., Sullivan, M.,et al. 2018, MNRAS, 479, 3232Huber, M., Chambers, K., Flewelling, H., et al.2015, ATel, 7153, 1Inserra, C., Smartt, S. J., Gall, E. E. E., et al.2018, MNRAS, 475, 1046Kauffmann, G., Heckman, T. M., White, S. D.,et al. 2003, Monthly Notices of the RoyalAstronomical Society, 341, 33Kewley, L. J., & Ellison, S. L. 2008, TheAstrophysical Journal, 681, 1183Leonard, D. C., & Filippenko, A. V. 2001, 113,920Lunnan, R., Chornock, R., Berger, E., et al. 2014,ApJ, 787, 138 Pejcha, O., & Prieto, J. L. 2015, ApJ, 806, 225Planck Collaboration, Ade, P. A. R., Aghanim,N., et al. 2016, A&A, 594, A13Poznanski, D., Butler, N., Filippenko, A. V.,et al. 2009, ApJ, 694, 1067Quimby, R. M., Kulkarni, S., Kasliwal, M. M.,et al. 2011, Nature, 474, 487Rodr´ıguez, ´O., Clocchiatti, A., & Hamuy, M.2014, AJ, 148, 107Rubin, A., Gal-Yam, A., De Cia, A., et al. 2016,ApJ, 820, 33Sanders, N. E., Soderberg, A. M., Gezari, S.,et al. 2015, ApJ, 799, 208Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ,737, 103Schmidt, B. P., Kirshner, R. P., Eastman, R. G.,et al. 1994, ApJ, 432, 42Schmidt, G. D., Weymann, R. J., & Foltz, C. B.1989, Publications of the Astronomical Societyof the Pacific, 101, 713Smartt, S. J. 2009, ARA&A, 47, 63Smartt, S. J., Valenti, S., Fraser, M., et al. 2015,A&A, 579, A40Stevenson, K. B., Bean, J. L., Seifahrt, A., et al.2016, The Astrophysical Journal, 817, 141Taddia, F., Moquist, P., Sollerman, J., et al.2016, A&A, 587, L7Tak´ats, K., Pignata, G., Pumo, M. L., et al. 2015,MNRAS, 450, 3137Tonry, J., Stubbs, C., Lykke, K., et al. 2012, ApJ,750, 99Tremonti, C. A., Heckman, T. M., Kauffmann,G., et al. 2004, The Astrophysical Journal, 613,898Valenti, S., Howell, D. A., Stritzinger, M. D.,et al. 2016, MNRAS, 459, 3939Yaron, O., & Gal-Yam, A. 2012, PASP, 124, 668 Scott et al.
Table 1.
A table of the measured properties of the SNe used in this study.Name z Fe II pEW M ∗ SN M ∗ host A † r,host PS16bbl 0.055 16 . . . − . − . . . . − . − . . . . − . − . . . . − . − . < . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . < . − . − . < . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . . . . − . − . ∗ Magnitudes listed are all in r-band, after correcting for distance, Milky Way extinction and K -corrections(and filter corrections where necessary). Internal host galaxy extinction has not been applied here. † Estimated extinction from the host galaxy in rr