Supernovae without host galaxies? The low surface brightness host of SN 2009Z
P.-C. Zinn, M. Stritzinger, J. Braithwaite, A. Gallazzi, P. Grunden, D. J. Bomans, N. I. Morrell, U. Bach
aa r X i v : . [ a s t r o - ph . C O ] N ov Astronomy&Astrophysicsmanuscript no. SN2009Z˙final c (cid:13)
ESO 2018October 13, 2018
Supernovae without host galaxies?
The low surface brightness host of SN 2009Z
P.-C. Zinn , , M. Stritzinger , , J. Braithwaite , A. Gallazzi , P. Grunden , D. J. Bomans , N. I. Morrell , and U. Bach Astronomical Institute, Ruhr-University Bochum, Universit¨atsstraße 150, 44801 Bochum, Germanye-mail: [email protected] CSIRO Astronomy & Space Science, PO Box 76, Epping, NSW, 1710, Australia The Oskar Klein Centre, Department of Astronomy, Stockholm University, AlbaNova, 10691 Stockholm, Sweden Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, 2100 Copenhagen Ø, Denmark Argelander Institut f¨ur Astronomie, Universit¨at Bonn, Auf dem H¨ugel 71, 53121 Bonn, Germany Las Campanas Observatory, Carnegie Observatories, Casilla 601, La Serena, Chile Max-Planck-Institute for Radio Astronomy, Auf dem H¨ugel 69, 53121 Bonn, GermanyReceived 04 / / / / ABSTRACT
Context.
A remarkable fraction of supernovae (SNe) have no obvious host galaxy. Two possible explanations are that (i) the hostgalaxy is simply not detected within the sensitivity of the available data or that (ii) the progenitor is a hypervelocity star that hasescaped its parent galaxy.
Aims.
We use the Type IIb SN 2009Z as a prototype of case (i), an example of how a very faint (here Low Surface Brightness; LSB)galaxy can be discovered via the observation of a seemingly host-less SN. By identifying and studying LSB galaxies that host SNerelated to the death of massive stars, we can place constraints on the stellar population and environment of LSB galaxies, which atpresent are poorly understood.
Methods.
We use archival ultraviolet (UV) and optical imaging, as well as an H I spectrum taken with the 100 m E ff elsberg RadioTelescope to measure various parameters of the host galaxy, in particular its redshift, stellar and H I mass, and metallicity. Results.
From the E ff elsberg spectrum, a redshift z = ± ± M ⊙ are computed. Thisredshift is consistent with that obtained from optical emission lines of SN 2009Z. Furthermore, a gas mass fraction of f g = . ± . ± . M ⊙ yr − . Based on the B -band luminosity we estimatean extinction-corrected metallicity following the calibration by Pilyugin (2001) of 12 + log (cid:16) OH (cid:17) = . ± . Conclusions.
The presence of a Type IIb SN in an LSB galaxy suggests, contrary to popular belief, that massive stars can be formedin this type of galaxies. Furthermore, our results imply that LSB galaxies undergo phases of small, local burst activity intermittentwith longer phases of inactivity, rather than a continuous but very low SFR. Discovering faint (LSB) galaxies via bright supernovaevents happening in them o ff ers an excellent opportunity to improve our understanding of the nature of LSB galaxies. Key words. supernovae: individual: SN 2009Z – galaxies: evolution – galaxies: stellar content – methods: observational
1. Introduction
The Sternberg Astronomical Institute (SAI) supernova (SN) cat-alog (Tsvetkov et al. 2004) lists over 5000 objects of which asurprising fraction have no obvious host galaxy. Two possiblescenarios discussed in the literature (e.g. Hayward et al. 2005)are:1. The host galaxies are simply not detected, given the sensitiv-ity of the available data;2. The progenitors of these SNe are hypervelocity stars ( v ≥
100 km s − , see Martin 2006) that have escaped the gravita-tional potential of their parent galaxy.In this paper we examine the case of the Type IIb SN 2009Z,which is an example of possibility 1. The nearest possible hostappeared initially to be the face-on spiral galaxy UGC 8939,whose core lies nearly 3 arcmin away, i.e. ∼
90 kpc on theplane of the sky, from the location of the SN. However, un-der close inspection of deep archival images, an irregular dwarfgalaxy, 2dFGRS N271Z016 (hereafter N271, also known as J140153.80-012035.5 in the Sloan Digital Sky Survey; SDSSAbazajian et al. 2009), has been identified as the true host. Thisgalaxy is at the edge of SDSS detection limits – the SDSS desig-nation was assigned to it prior to the supernova, but the detectionis at a very low confidence limit. We classify this galaxy belowas a low surface brightness (LSB) galaxy (for a concise reviewof this class of galaxies see for example Impey & Bothun 1997).This is interesting, since SNe are rarely found in LSB galaxies.As easily detectable point sources, SNe are a promisingtool for discovering very faint and / or LSB galaxies. In contrast,sensitivity-limited galaxy surveys yield an incomplete sampleof galaxies in which LSB galaxies are very likely to be under-represented. Consequently, the contribution of LSB galaxies toboth the total baryon density of the universe and their contri-bution to the galaxy number density are still uncertain. For ex-ample, Hayward et al. (2005) argued that LSB galaxies containonly a small fraction of the baryons and are therefore ‘cosmo-logically unimportant’, whereas Minchin et al. (2004) found thatLSB galaxies account for 62 ±
37% of gas-rich galaxies by num-ber.
Obviously to use SNe to find faint galaxies, a survey needsto be ‘non-targeted’ rather than looking at likely SN locations.Currently there are several wide-field, non-targeted SN surveysunderway (e.g. Pan-STARRS or the Palomar Transient Factory,Kaiser et al. 2002; Law et al. 2009) whose goal, amongst others,is to characterize SN events that occur in all types of galaxies.SN 2009Z in N271 can therefore be put into context with a num-ber of other recently studied core-collapse (CC) SNe, includingthose that are associated with a long-duration gamma-ray burst(GRB), that have occurred in faint dwarf galaxies. A number oflong GRBs have now been associated with broad-lined Type IcSNe, but it is clear from the statistics that not all broad-linedType Ic SNe produce a GRB, either on-axis or o ff -axis. LongGRBs are found preferentially in small irregular galaxies, andin the more luminous parts of their hosts, in this respect simi-lar to Type Ic SNe, but unlike Type II SNe (Fruchter et al. 2006;Kelly et al. 2008). Furthermore, Modjaz et al. (2008) found thathosts of Type Ic SNe associated with a GRB have lower metal-licity on average than those without any GRB. In general longGRBs are associated with low metallicity (Stanek et al. 2006),which may be related to the bias towards high redshift. From atheoretical point of view, is it thought that low metallicity some-how helps the progenitor to retain more angular momentum, viasuppression of wind, for instance. It is generally accepted thatrapid core rotation is essential to produce a GRB, as well as aprogenitor significantly above the ∼ M ⊙ CCSN threshold (seee.g. Woosley 2011, and refs. therein). In any case, the matterof SNe Ic with and without GRBs clearly demonstrates the needfor more thorough studies of SNe and their hosts, in particulara larger variety of hosts – previous SN surveys have targetedmainly bright, giant galaxies.Although all the work summarized above mainly focusseson Type Ic events and associated GRBs, other types of CCSNeare useful in shedding light on the stellar population of theirhost galaxies, since they also require high-mass progenitors. Forinstance, studies of the environments of regular and strippedCCSNe have been made on their metallicity (Anderson et al.2011; Modjaz et al. 2011; Leloudas et al. 2011) and the age ofthe stellar population (Leloudas et al. 2011).LSB galaxies, according to prevailing opinion, have a com-parable total H I mass to high surface brightness (HSB) galax-ies, but a lower surface density, too low for molecular clouds toform (for the surface density criterion see Kennicutt 1989). Thisleads to lower rates of star formation and metal production. Notsurprisingly, LSB galaxies have higher mass-to-light ratios thanHSB galaxies (de Blok & McGaugh 1996).In this paper we present our examination of the propertiesof N271, the LSB host galaxy of SN 2009Z, concentrating onits stellar population, gas mass fraction and the other proper-ties of its interstellar medium (ISM). Throughout this paper, weadopt a flat Λ CDM cosmology with H = . − Mpc − and Ω Λ =
2. Observations
SN 2009Z was discovered on 2.53 February 2009 UT bythe the Lick Observatory Supernova Search (Filippenko et al. 2001) with an unfiltered magnitude of 18.1. Soon afterwardsStritzinger & Morrell (2009) classified it as a Type IIb, spectro-scopically most similar to SN 1993J around maximum. Detailedoptical and near-IR observations were obtained by the CarnegieSupernova Project (Hamuy et al. 2006). An analysis of prelimi-nary light curves reveals a peak B -band maximum of 17.85 ± B -band magnitude of M B = − . ± . z = . α line to eventually detect H α emission from N271. Fitting aGaussian to the broad H α line leaves a small “cap” on top ofit (see right panel in Fig. 1). Since this “cap” exactly matchesthe redshift derived from the H I spectrum, we conclude that itis originating from N271 itself, adding to the H α emission ofSN 2009Z. We used archival photometric data on N271 from the SDSS andESO archives. The left panel of Fig. 2 shows a color SDSS imagecomposed of g − , r − and i -band images of N271. In addition, toenhance the accuracy of the spectral energy distribution (SED)measurement of N271 (see Sec. 3.2), ultraviolet (UV) imag-ing data was obtained from the GALEX (Milliard et al. 2001)database. To ensure that both GALEX and SDSS magnitudeswere comparable for the SED fitting process, the GALEX fluxdensities in both the far ultraviolet (FUV at about 1500 Å) andnear ultraviolet (NUV at about 2300 Å) bands were measuredusing the same aperture as for the computation of the photome-try from the SDSS images. The GALEX FUV image of N271 isshown in Fig. 3. A journal of the complete photometric data setused in this work is given in Table 1. Unfortunately, there is noinfrared (IR) data for N271, neither in the 2MASS survey (dueto a high flux limit) nor in the UKIDSS survey (which does notcover the location of N271 yet). Despite being detected in the2dF survey (Folkes et al. 1999), N271 has no spectrum availablewith a su ffi cient S / N to determine either its redshift or metallic-ity. As the SDSS images are neither sensitive enough nor pro-vide su ffi cient spatial resolution to perform morphological anal-yses for such faint galaxies as N271, it was necessary to ob-tain additional deep imaging, particularly to determine the scalelength of N271 and allow a precise measurement of its surfacebrightness profile. We therefore obtained from the ESO archivean R -band image taken on June 23, 2004 with EMMI mountedto the 3.6 m New Technology Telescope (NTT). With an inte-gration time of 300 s this image is much deeper than those fromthe SDSS archive (52 s with a 2.5 m mirror). The NTT imagealso benefits from excellent seeing conditions ( ∼ . ′′ B -band surface brightness of µ B = . ± .
13 mag arcsec − . For this calculation a scale length of h r = Fig. 1.
Left, an optical spectrum of SN 2009Z (blue line) obtained 10 days after maximum light with the 2.5 m du Pont Telescopeat Las Campanas Observatory is shown. The spectrum is de-redshifted using z = . α line of SN 2009Z is shown. A Gaussian fit (red line) to this line reveals the presence of a small amount ofH α emission “on top” of the broad supernova line. This additional H α flux is most likely to originate from the host galaxy N271. Table 1.
Photometric data points used in this work and itssources.
Band mag mag error Source
FUV a re-measured NUV a re-measured u c SDSS DR6 b g b r b i b z c SDSS DR6 ba Milliard et al. (2001), fluxes were re-extracted to ensure the sameaperture radius as for the optical data. b Adelman-McCarthy et al. (2008). c Because the u - and z -band SDSS images are of low signal-to-noise(S / N), no detection of N271 could be made down to the 2.5 σ level,hence these values are not used when fitting the SED. . task ellipse . Adopting a magnitude cut definition forLSB galaxies either of 23 mag arcsec − (Impey & Bothun 1997)or 22 mag arcsec − (McGaugh et al. 1995), N271 is clearly aLSB galaxy. IRAF is distributed by the National Optical AstronomyObservatories, which are operated by the Association of Universitiesfor Research in Astronomy, Inc., under cooperative agreement with theNational Science Foundation.
Fig. 3.
GALEX FUV image of N271. This image consists of ap-proximately the same region as the NTT R -band image shownFig. 2. As one can clearly see, N271 looks very flocculent in theFUV, in particular when looking at the east and west edges of thegalaxy whereas a central stripe has significantly less UV emis-sion. This supports the idea described above that LSB galaxieshave localized starforming regions in contrast to a starburst eventacross the entire galaxy. A H I spectrum of N271 was obtained with the 100 m E ff elsbergRadio Telescope. This spectrum was used to measure the redshift Fig. 2.
Left, color composite of SDSS g -, r -, and i -band images of N271 (located at the very center of the picture) and its surrounding.The circle indicates the E ff elsberg half-power beam width at 21 cm. Right, NTT R -band close-up of N271 (region of close-uphighlighted in the left panel by a white box) with two ticks marking the position of SN 2009Z).from the 21 cm line, as well as to determine the H I mass con-tent. Therefore a 20 MHz filter was used, spread over a frequencyregion from − , well enough to separate even small velocityand hence redshift di ff erences. During the reduction of the spec-trum, a binning (bin-width of six channels) was performed toincrease the S / N ratio.The FWHM beam width at this wavelength is 8 arcmin sowe expect lines from more than one galaxy: in fact we find twoclear H I emission lines (Fig. 4). UGC 8939, whose redshift isalready known ( z = + ff ecting the detection of UGC 8939.Therefore we identify the weaker of the two observed lines withN271.We stress that there is only little risk of confusing the H I sig-nals identified in Fig.4. The shape of the emission line belongingto UGC 8939 matches exactly the expactations for an H I line ofa face-on spiral, so a single line instead of a doulbe peaked pro-file which is typical only for edge-on spirals. The non-detectionsof the two galaxies MCG + + This range was chosen because it corresponds to the redshift ofUGC 8939, with which we assumed N271 is associated. the latter one being pretty much outside the E ff elsberg beam withan antenna response of only 0.2% at this distance from the point-ing center.A baseline subtraction and Gaussian fits to the two lines wereperformed in order to measure the redshifts and H I masses, fol-lowing the method of Roberts (1962). For N271, this yieldeda peak radial velocity of v = ± − relative to the restfrequency of neutral hydrogen , corresponding to a redshift of z = ± d L = . ± . Λ CDM cosmo-logical model adopted in Sec. 1. Therefore its error only reflectsthe measurement uncertainty of the redshift, errors due to a pe-culiar velocitiy that N271 may have or errors of the cosmologicalparameters were not taken into account.From the stronger emission line of UGC 8939, we measurea H I mass of 8 10 M ⊙ , typical of a Sb spiral galaxy. For N271we arrive at M H I = . ± .
12 10 M ⊙ , (1)putting N271 within the (upper part of the) range of H I massesof dwarf galaxies (see e.g. Zwaan et al. 2005). This, and its ab-solute B -band magnitude of M B = − .
22, confirm that N271is a LSB dwarf galaxy. This velocity is calculated relative to the Local Standard of Rest(LSR). Please note that for the estimation of a cosmological redshift nopeculiar velocity of N271 and no rotation correction for the MW weretaken into account.4.-C. Zinn et al.: Supernovae without host galaxies?
Fig. 4.
H I E ff elsbergspectrum. The blackline is the actual mea-surement, while thesmooth grey line rep-resents Gaussian fits ofthe two detected emis-sion lines after baselinesubtraction. The galaxyidentities are marked.Note that the flux isgiven in arbitrary units.
3. Inferred properties of N271
We now measure the stellar mass of N271 using the work ofBell et al. (2003) who connected the stellar mass of a galaxy tomeasurable parameters assuming a Salpeter (1955) initial massfunction (IMF). Based on the g - and r -band luminosities ofN271, which are less a ff ected by contemporary star formationthan the B -band (classically used for this estimation), we com-pute a stellar mass oflog( M ∗ / L r ) = . g − r ) − . → M ∗ = . ± .
77 10 M ⊙ . (2)From this a gas mass fraction f g = M g / ( M g + M ∗ ) can be derivedfollowing Schombert et al. (2001) of f g = + M ∗ η M H I ! − = . ± . , (3)where we have adopted η = . f g ≈ .
9, and the mean H I mass isaround three times lower than that in N271. Also, only one per-cent of the sample have a central surface brightness as low as24 mag arcsec − . In summary, N271 is a rather extreme LSBdwarf specimen. To characterize the stellar population of N271 we used the avail-able photometric data to fit SED templates of various galaxytypes, using the publicly available SED fitting code hyperz by Bolzonella et al. (2000). This package contains two setsof template SEDs, both of which we included in our fittingprocedure. Firstly, the observationally generated templates byColeman et al. (1980), CWW hereafter, include E / S0, Sbc, Scdand Im galaxies. Secondly, the synthetically generated templatesfrom the GISSEL library by Bruzual & Charlot (1993; evol forshort) include starburst, E, S0, Sa, Sb, Sc, Sd and Im galaxies, and allow an estimation of the age of the stellar population. Thefitting process is a χ minimization technique which makes useof the measured redshift of z fit = χ evol = χ CWW = ±
160 Myr was de-termined. This relatively young age (see e.g. Li & Han 2007,for a characterization how young stellar populations influencetheir parent galaxies) is in good agreement with the work byHaberzettl et al. (2008) who analyzed LSBs in the
Hubble deepfields, resulting in the main finding that the stellar populationsof LSBs tend to be younger than in comparable HSBs. Thiswould either imply that N271 is currently undergoing its firstmajor star formation event at all or at least has become activeagain after a longer phase of quiescence. This is also empha-sized by the high extinction A V = . m employed by the best-fitting model. Deriving the current star formation rate of N271using the FUV flux as measured using GALEX data ( L FUV = . ± .
05 10 erg s − Hz − ) as a star formation tracer, follow-ing the calibration of Kennicutt (1998) and Madau et al. (1996),one gets a star formation rate of: S FR
FUV = . ± . M ⊙ yr − , S FR
FUV , corrected = . ± . M ⊙ yr − . (4)The corrected SFR accounts for extinction in the galaxy itselfwhich is most important since UV wavelengths are extremelya ff ected by dust attenuation. To correct the FUV flux for thisextinction, we adopted A V = . m from the SED-fit, yielding anattenuation at 1516 Å (e ff ective wavelength of the GALEX FUVband) of A FUV = . m using the Calzetti et al. (2000) extinctionlaw. Note that FUV -derived SFRs are known to be notoriouslya ff ected by the amount of internal extinction and the reddeningcurve adopted for the computation of A FUV and that thereforethe value actually derived as SFR has to be treated with care(see. Sec. 3.4 for a closer discussion).Nevertheless, we come to the conclusion that a star forma-tion rate of a few 0.1 M ⊙ yr − supports the idea that N271 is cur-rently undergoing a major star formation episode, considering Fig. 5.
The two best-fit model SEDs computed with hyperz . Left:
The best-fitting model ( χ = right: The Im galaxy template ( χ = ffi cient employed by the model is A V = . m . The x error bars correspond to the FWHM of the GALEX and SDSS bandpasses.that it is a LSB dwarf galaxy. For comparison we note that nor-mal HSB spiral galaxies such as the Milky Way typically showstar formation rates of ∼ M ⊙ yr − , only a factor of a few higherthan that of N271, although it is about 100 times more massive. Given this relatively high SFR, one might wonder whether thecurrently ongoing star-formation event is the first one in the his-tory of N271 or whether there has been previous star-formationactivity. By estimating the metallicity of N271, we shall see thatthe latter of these two possibilities is much more likely. Due tothe lack of a su ffi ciently high S / N spectrum, we estimated themetallicity using the rough metallicity – luminosity relationship,as calibrated by Pilyugin (2001) for dwarf irregular galaxies.Based on N271’s absolute B -band magnitude calculated usingthe Fukugita et al. (1996) conversation equations between SDSSand Johnson / Kron-Cousins bandpasses, M B = − .
22, we ob-tain an oxygen-related gas phase metallicity of12 + log OH ! = . ± . , + log OH ! corrected = . ± . . (5)As before, we used an extinction of A V = . m as suggestedby the SED-fit and a Calzetti et al. (2000) extinction law tocompute an extinction-corrected B -band absolute magnitude of M B , corrected = − .
67. Despite the relatively large error associ-ated with the metallicity inferred from the coarse relation be-tween metallicity and luminosity, one has also to bear in mindthat N271 was demonstrated to be an extreme example of adwarf galaxy, hence may fall o ff the calibration by Pilyugin atan even larger fraction. But since the galaxy is quite faint sothat there are no optical spectra available from which a moreaccurate metallicity could be derived, we decided to adopt this This large error arises due to the large intrinsic scatter of the wellknown relationship between metallicity and luminosity, see for examplePilyugin (2001), their Fig. 2. value with the corresponding errors. With that, one cannot regardN271 to be a metal-poor galaxy, particularly if one compares itto other low-luminosity dwarf irregulars such as the Large andSmall Magellanic Clouds (12 + log (O / H) = .
50 and 8 .
09 re-spectively which also fall very well on the metallicity – lumi-nosity relation). This implies, even when only considering theuncorrected value in (5), that earlier SNe must have occurred inN271 to enrich the ISM with metals. Assuming a constant SFRthroughout the entire 360 Myr of age of the fitted stellar popu-lation of N271 we come up with a total mass of stars formed of M ∗ tot = . M ⊙ . This is in good agreement with the mass-to-light ratio determined stellar mass as well as with the stellarmass derived from a Bayesian approach as outlined in Sect. 3.4.All together, this underlines the picture that star formation inLSB galaxies occurs in small distinct bursts that are well sep-arated in time because the bulk of the stellar content of N271seems to have been produced during the current burst. To test the reliability of the galaxy’s parameters derived so far us-ing a variety of well-known scaling relations, we also performeda more sophisticated SED fit which follows a Bayesian ap-proach. We make use of a large library of model SEDs obtainedby convolving Bruzual & Charlot (2003) simple stellar popula-tions of di ff erent metallicities with Monte Carlo star formationhistories and dust attenuations. For dust attenuation we adoptin this case the Charlot & Fall (2000) two-component model,regulated by the total e ff ective optical depth τ V a ff ecting starsyounger than 10 yr and the fraction µ contributed by the ISM.As a result the dust attenuation curve is not constant in timeand is not a simple power law for composite stellar populations(as opposed to the Calzetti et al. (2000) attenuation law adoptedabove).To derive galaxy’s physical parameters such as stellar mass,dust attenuation, mean light-weighted age and SFR, we comparethe galaxy SED to all the SEDs in the model library and buildthe probability density function of each parameter. The advan-tage of this approach is that it provides a robust estimate of the Table 2.
Synopsis of SED-fitting derived parameters. classic classic BayesianSalpeter IMF Chabrier IMF Chabrier IMF A V [mag] 1 . ± .
15 0 . ± . A FUV [mag] 2 . ± . . ± . M ∗ [10 M ⊙ ] 2 . ± .
77 1 . ± .
04 1 . ± . S FR [ M ⊙ yr − ] 0 . ± .
34 0 . ± .
23 0 . ± . ±
160 960 ± uncertainties in the derived parameters coming both from ob-servational uncertainties and model degeneracies. It is howevermore sensitive to the adopted prior distribution of the model pa-rameters.The parameters derived both with the “classical” method andwith the Bayesian approach are summarized in Tab. 2. Note thatthe “classical” parameters were calculated adopting a Salpeter(1955) Initial Mass Function (IMF) whereas the Bayesian calcu-lation relies on a Chabrier (2003) IMF. While the choice betweenthe two IMFs does not a ff ect the color evolution, hence color-derived stellar population properties such as age and dust atten-uation, it a ff ects integrated quantities such as stellar mass andSFR. Based on Bruzual & Charlot (2003) models, we estimatethat M S alp ≈ . M Chab and
S FR
S alp ≈ . S FR
Chab and adoptthese conversions for comparison between the two approaches inTab 2. Both the stellar mass and the attenuation in the optical de-rived with the two methods agree very well within the combineduncertainties. The dust attenuation in the FUV is instead quitedi ff erent, most likely as a result of the di ff erent attenuation lawsadopted in the two cases. This a ff ects mostly the SFR estimate,which decreases by a factor of ∼ A FUV = . m instead of A FUV = . m to correct the FUV luminosity. We notethough that by adopting A FUV = . m the SFR estimated di-rectly from the UV luminosity using the Kennicutt (1998) for-mula agrees very well with the one derived by the Bayesian SEDfitting. Because of the age–dust degeneracy the di ff erence in dustattenuation is also somewhat reflected in the estimated stellarage, which is higher in the Bayesian approach.As a whole these results point to a low-mass galaxy with ayoung stellar population having ongoing star formation at a levelof at least 0 . M ⊙ yr − and a fair amount of dust attenuation.
4. Discussion
The progenitors of SNe IIb are believed to be massive sin-gle stars that have lost much of their hydrogen envelope(Woosley et al. 1993) or massive evolved stars in a binary system(e.g. Crockett et al. 2008; Thielemann et al. 1996; Woosley et al.1995; Shigeyama et al. 1990). In particular, the archetype ofsuch SNe, SN 1993J was identified to have had a massive bi-nary companion of 14 M ⊙ (Maund et al. 2004). Type IIb SNetherefore demonstrate the presence of massive stars, so the caseof SN 2009Z contradicts the long-held belief that LSB galaxiescontain only low-mass stars, but corresponds to the findings ofMattsson et al. (2007), that the initial mass functions (IMFs) ofLSB galaxies do extend to high mass.The extremely high gas mass fraction is a strong hintthat this type of galaxy is ine ffi cient in star formation (e.g.Schombert et al. 1990). Our star-formation rate estimate in N271of 0.44 M ⊙ yr − is larger than typical LSB galaxies, which lie inthe range 0.02 to 0.2 M ⊙ yr − (van den Hoek et al. 2000), andalmost comparable to that of normal HSB spirals. In addition its metallicity of 12 + log (cid:16) OH (cid:17) corrected = . ± .
70 is higher thantypical LSB dwarfs, but is normal for HSB galaxies of compara-ble mass (Pilyugin & Thuan 2007).A star formation history of LSB galaxies that includes theexistence of short (a few 100 Myr) bursts separated by longerquiescent periods is preferred by many authors (Schombert et al.2001; Boissier et al. 2003; Vallenari et al. 2005; Boissier et al.2008); these starbursts are too short-lived to transform the galaxyto HSB. In N271 this scenario looks very likely; the progeni-tor of this core-collapse SN presumably formed in the most re-cent starburst. Furthermore, it may be related to the finding byGrunden et al. (in prep.) that the ratio of core-collapse to ther-monuclear SNe is two times higher in LSB galaxies than in HSBgalaxies.SNe in dwarf galaxies have recently become a heavily dis-cussed topic. It is informative to compare this SN in a LSBdwarf to SNe in other, both LSB and non-LSB, dwarfs. Usingthe first compilation of 72 SNe from the Palomar TransientFactory (PTF), Arcavi et al. (2010) analyzed statistics of CCSNein dwarfs and giant galaxies. They found a significant excessof Type IIb events in dwarfs (defined as M r > −
18; N271 has M r = − M ⊙ limit for CCSNe (see Smartt et al.2009; Heger et al. 2003). The case of SN 2009Z then clearlydemonstrates that at least intermediate-mass star formation (asIIb event, the progenitor of SN 2009Z must have had at least30 M ⊙ as single star or 15 M ⊙ when member of a binary system)does happen in LSBs, too. However, we want to point out oncemore that because of the large error bars of our inferred metal-licity (see Sect. 3.3, this conclusion has to be treated with care.Although SN 2009Z was not a luminous event accordingto the definition of Neill et al. (2011), i.e. peak M V < − = SFR / M ∗ ) of their luminous-SN host galaxies and found that most of them were very bluedwarfs with low stellar masses and high sSFRs. The sSFR ofN271 is 1.71 10 − yr − , which as we can see from Fig. 6, is wellwithin the range of the hosts of these luminous SNe. In accor-dance with the authors cited above, Neill et al. invoke metallicityto explain the correlation between faint, blue dwarf hosts and lu-minous SN events. Specifically, they argue that at higher metal-licity, massive stars su ff er much greater wind mass loss and thatonly in metal-poor galaxies one should expect to find the verymassive ( > M ⊙ ) progenitors required to produce luminuousSNe (Neill et al. 2011).Surveys of large areas of sky (of the order of a few thou-sand square degrees) with search cadences of a few days are dis-covering large numbers of SNe in low-metallicity galaxies. ThePalomar Transient Factory now has a large sample of such SNe Fig. 6.
Specific star formation rate (sSFR) vs. stellar mass.The sample of hosts of luminous (M V < -21) supernovaeof Neill et al. (2011) (see their Fig. 3) is plotted togetherwith N271 and the LMC (values from Westerlund 1997 andHarris & Zaritsky 2009) and a sample of about 60,000 SDSS-GALEX galaxies from Wyder et al. (2007) relying on stellarmasses and star formation rates deduced by Kau ff mann et al.(2003) and Brinchmann et al. (2004). Clearly, N271 fits quitewell to the “extreme” hosts although SN 2009Z was not a veryluminous event.at low redshift and the relative rates are surprising (Arcavi et al.2010). The Pan-STARRS1 survey is searching for low-z SNe inthe 3Pi faint galaxy survey (Valenti 2010; Young et al. 2010),too, but has also has found high-z ultraluminous SNe at z = ff mann et al. (2003) to get metallicity information for thehosts of about 120 supernova events of all types. Their main find-ing that SN Ib / c seem to be more abundant in metal-rich galax-ies while SN II seem to occur more often in metal-poor ones,also supports the argumentation outlined in this paper. They alsomatchd their supernova sample to pure SDSS image data to goto fainter host galaxies. This resulted in the finding that lumi-nous supernovae tend to appear in faint hosts, as for instance thehypernova-like event SN 2007bg which happened in an extremedwarf of M B = − . ± . , one of the faintest SN hosts everobserved (Young et al. 2010).The findings concerning the star formation history of N271and its current stellar content could be related by taking into ac-count the work by Rosenbaum & Bomans (2004) who analyzedthe environment of LSB galaxies in the SDSS early data release.They find that LSBs are, unlike HSBs, often found in less denseenvironments or even in void structures. Therefore they undergofewer interactions with other galaxies, which are known to trig-ger star formation. Despite this very low luminosity in the B band, Young et al. (2010)determined the metallicity of the host to be 12 + log (cid:16) OH (cid:17) = . ± .
5. Conclusions
We have investigated the dwarf galaxy N271 which is thehost of the Type IIb SN 2009Z. It is a low surface brightness(LSB) galaxy with central surface brightness µ B = . ± .
13 mag arcsec − . Using a 21cm spectrum obtained with theE ff elsberg Radio Telescope we measured a redshift of z = . ± .
12 10 M ⊙ . Using SDSS g - and r -band magnitudes to estimate a mass-to-light ratio and thereforea stellar mass, we arrive at the rather high gas mass fraction of f g = . ± . . M ⊙ yr − , which is somewhat higher than typ-ical LSB values. This picture of N271 currently witnessing astarburst event is supported by its relatively high metallicity of12 + log (cid:16) OH (cid:17) corrected = . ± .
70, comparable to the Magellanicclouds, implying metal-enrichment from previous bursts. Suchdistinct bursts may be a common phenomenon in LSB dwarfgalaxies.We conclude that LSB galaxies do not represent a completelyalternative evolutionary path from HSB galaxies but rather are ina certain evolutionary state, as proposed e. g. by Haberzettl et al.(2008) and van den Hoek et al. (2000). The fact that there ishigh-mass star formation in LSBs (as shown by SN 2009Z)clearly demonstrates that the old picture of LSB galaxies be-ing trapped in a low metallicity / low star-formation “cage” mustbe revised. More likely is that these galaxies are going throughan LSB phase but at some time evolve into normal HSB galax-ies. Whether this LSB phase is long or short may depend on thestrength of the small bursts of star formation, more precisely ifone of those bursts is strong enough to permanently transformthe LSB into an HSB galaxy. Further exploration is needed toinvestigate this phase for HSB galaxies: Whether present-dayHSBs ever went through a significant LSB phase is not yet clear.Finally, further work is needed regarding the apparent correla-tion between SN type and host galaxy type. Acknowledgements.
We like to thank our referee, S. J. Smartt, for all the helpfulcomments, especially on contemporary literature from the “supernova side”, andfurther suggestions he made which substantially improved this paper.DARK is funded by the Danish NSF. We thank Norbert Langer for finan-cial support. Funding for the SDSS and SDSS-II has been provided by theAlfred P. Sloan Foundation, the Participating Institutions, the National ScienceFoundation, the U.S. Department of Energy, the National Aeronautics and SpaceAdministration, the Japanese Monbukagakusho, the Max Planck Society, andthe Higher Education Funding Council for England. The SDSS Web Site ishttp: // / .Based on observations made with ESO Telescopes at the La Silla or ParanalObservatories under programme ID 073.A-0503(A). References
Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ueros, M. A., et al. 2009, ApJS,182, 543Adelman-McCarthy, J. K., Ag¨ueros, M. A., Allam, S. S., et al. 2008, ApJS, 175,297Anderson, J. P., Habergham, S. M., & James, P. A. 2011, ArXiv e-printsArcavi, I., Gal-Yam, A., Kasliwal, M. M., et al. 2010, ApJ, 721, 777Barbon, R., Benetti, S., Cappellaro, E., et al. 1995, A&AS, 110, 513Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149, 289Boissier, S., Gil de Paz, A., Boselli, A., et al. 2008, ApJ, 681, 244Boissier, S., Monnier Ragaigne, D., van Driel, W., Balkowski, C., & Prantzos,N. 2003, Ap&SS, 284, 913Bolzonella, M., Miralles, J., & Pell´o, R. 2000, A&A, 363, 476Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151Bruzual, A. G. & Charlot, S. 1993, ApJ, 405, 538