The non-thermal superbubble in IC 10: the generation of cosmic ray electrons caught in the act
Volker Heesen, Elias Brinks, Martin G. H. Krause, Jeremy J. Harwood, Urvashi Rau, Michael P. Rupen, Deidre A. Hunter, Krzysztof T. Chyzy, Ged Kitchener
aa r X i v : . [ a s t r o - ph . GA ] N ov Mon. Not. R. Astron. Soc. , 000–000 (0000) Printed 13 August 2018 (MN L A TEX style file v2.2)
The non-thermal superbubble in IC 10: the generation of cosmic rayelectrons caught in the act
Volker Heesen, ⋆ Elias Brinks, Martin G. H. Krause, , Jeremy J. Harwood, † Urvashi Rau, Michael P. Rupen, Deidre A. Hunter, Krzysztof T. Chy˙zy and Ged Kitchener School of Physics and Astronomy, University of Southampton, Southampton SO17 1BJ, UK Centre for Astrophysics Research, University of Hertfordshire, Hatfield AL10 9AB, UK Excellence Cluster Universe, Technische Universit¨at M¨unchen, Boltzmannstrasse 2, D-85748 Garching, Germany Max-Planck-Institut f¨ur extraterrestrische Physik, Giessenbachstr. 1, D-85741 Garching, Germany NRAO, P.V.D. Science Operations Center, National Radio Astronomy Observatory, 1003 Lopezville Road, Socorro, NM 87801, USA Lowell Observatory, 1400 West Mars Hill Road, Flagsta ff , AZ 86001, USA Obserwatorium Astronomiczne Uniwersytetu Jagiello´nskiego, ul. Orla 171, 30-244 Krak´ow, Poland
Accepted 2014 October 15. Received 2014 October 15; in original form 2014 September 22
ABSTRACT
Superbubbles are crucial for stellar feedback, with supposedly high (of the order of 10 percent) thermalization rates. We combined multiband radio continuum observations from theVery Large Array (VLA) with E ff elsberg data to study the non-thermal superbubble (NSB) inIC 10, a starburst dwarf irregular galaxy in the Local Group. Thermal emission was subtractedusing a combination of Balmer H α and VLA 32 GHz continuum maps. The bubble’s non-thermal spectrum between 1 . . ± µ G, and measuring the radiation energy density from
Spitzer
MIPS maps as 5 ± × − erg cm − , we determine, based on the spectral curvature, a spectralage of the bubble of 1 . ± . I data cube showsan expanding H I hole with 100 pc diameter and a dynamical age of 3 . ± . M >
23 M ⊙ ). The resultsare consistent with the expected evolution for a superbubble with a few massive stars, wherea very energetic event like a Type Ic supernova / hypernova has taken place about 1 Myr ago.We discuss alternatives to this interpretation. Key words: radiation mechanisms: non-thermal – cosmic rays – galaxies: individual: IC 10– galaxies: irregular – galaxies: starburst – radio continuum: galaxies.
Stellar feedback is a fundamental process that regulates the for-mation and evolution of galaxies. Supernovae (SNe) inject en-ergy into the interstellar medium (ISM), heating the gas to X-ray emitting temperatures and accelerating cosmic rays via shockwaves. Galactic winds, hybridly driven by the hot gas and cos-mic rays, remove mass and angular momentum (Everett et al.2008; Strickland & Heckman 2009; Dorfi & Breitschwerdt 2012;Hanasz et al. 2013; Salem & Bryan 2014). Cosmological simula-tions without stellar feedback not only predict wrong global massestimates, but mass concentrations towards the centres of galaxies ⋆ E-mail: [email protected] † Now at ASTRON, Postbus 2, 7990 AA Dwingeloo, the Netherlands. that are too high, leading to rotation curves that are steeper thanobserved (Scannapieco 2013). The most abundant type of galaxiesin the local Universe, dwarf galaxies, are particularly a ff ected byoutflows: their weak gravitational potentials make them suscepti-ble to outflows and winds (Tremonti et al. 2004). In the paradigmof a Λ CDM Universe, the removal of baryons in the least massivedark matter haloes may resolve the long standing ‘missing satel-lites’ problem (Moore et al. 1999). The loss of baryonic matterand associated angular momentum at early stages in their forma-tion and evolution can a ff ect the distribution of the non-baryonicmatter as well, rendering the inner part of the rotation curves lesssteep (Governato et al. 2010; Oh et al. 2011a,b). Furthermore, out-flows and winds in dwarf galaxies may be behind the magnetizationof the early Universe (e.g. Pakmor, Marinacci & Springel 2014;Siejkowski et al. 2014). c (cid:13) V. Heesen et al.
Massive stars are the agents of stellar feedback and they man-ifest themselves by carving bubbles – cavities of tenuous, hot gas– into the ISM. They usually form in groups, so that their bubblesstart to overlap when expanding and subsequently merge, forminglarger structures in excess of 100 pc, so-called superbubbles. Thewind of massive stars, especially during their Wolf–Rayet (WR)phase, powers the early expansion of the bubble. Subsequent SNecreate strong shocks in the bubble interior that are responsible forthe thermal X-ray and the non-thermal synchrotron emitting gas(Krause et al. 2014). Stellar feedback in the form of SNe is moree ffi cient for clustered SNe than for randomly distributed ones assubsequent SNe explode in the tenuous gas of the bubble and theirshock waves are not su ff ering from strong radiative cooling. Hence,the thermalization fraction of clustered SNe is higher (Krause et al.2013).An intriguing example of SN feedback is presented bywhat has become known as the non-thermal superbubble (NSB;Yang & Skillman 1993) in the nearby dwarf irregular galaxyIC 10, a member of the Local Group at a distance of 0 . = . >
23 M ⊙ (Silverman & Filippenko 2008), which forms together with themassive WR star [MAC92] 17A a highly variable luminous X-ray binary, known as IC 10 X-1 (J2000 .
0, RA 00 h m s . . ◦ ′ . ′′
95; Bauer & Brandt 2004; Barnard et al. 2014). Ithas been speculated that a core collapse of the IC 10 X-1 progeni-tor in a ‘hypernova’ could be responsible for the NSB, rather thana series of SNe (Lozinskaya & Moiseev 2007).In this Letter, we present multiband radio continuum observa-tions with the NRAO Karl G. Jansky Very Large Array (VLA)to study the non-thermal radio continuum spectrum of the NSB.This project follows on, and extends some preliminary results pre-sented in Heesen et al. (2011). The data cover the frequency rangebetween 1 . We observed IC 10 with the VLA (project ID: AH1006). Obser-vations were taken in D-array in 2010 August and September at L band (1 . . C band (4 . . . . X band(7 . . Ka band (27–28 and 37–38 GHz) with ≈ L , C and X band with ≈ + L -band B-array data observed with the historical VLAin 1986 September (ID: AS0266) from Yang & Skillman (1993).We followed standard data reduction procedures, using theCommon Astronomy Software Applications package ( CASA ), de-veloped by NRAO, and utilizing the flux scale by Perley & Butler(2013). We self-calibrated the L -, C - and X -band data with two The National Radio Astronomy Observatory is a facility of the NationalScience Foundation operated under cooperative agreement by AssociatedUniversities, Inc. rounds of phase-only antenna-based gain corrections, using im-ages from the C-array data as a model. In C and X bands, weself-calibrated in phase and amplitude, adding in the D-array data(self-calibrated in phase), checking that the amplitudes did notchange by more than 1–2 per cent. For the imaging we used CASA ’s implementation of the Multi-Scale Multi-Frequency Syn-thesis (MS–MFS) algorithm (Rau & Cornwell 2011), which simul-taneously solves for spatial and spectral structure during wide-bandimage reconstruction. A radio spectral index image was producedby MS–MFS as well, which we used to refine the self-calibrationmodel. A post-deconvolution wide-band primary beam correctionwas applied to remove the e ff ect of the frequency-dependent pri-mary beam. For the spectral analysis, we imaged subsets (‘spectralwindows’) of data with either 128 or 256 MHz bandwidth, vary-ing Briggs’ ‘robust’ parameter as function of frequency to achievea synthesized beam of a similar angular size. All data were con-volved with a Gaussian kernel in AIPS to an identical resolution of5 . ≈
30 m(elevation dependent), there is a limit to the largest angular scalethat can be observed, resulting broadly in flux densities that are toolow compared with single-dish measurements; this is known as the‘missing zero-spacing flux’. Our VLA flux density at 1 . . . ff elsbergtelescope (Chy˙zy et al. 2003, 2011), of 343, 277 and 156 mJy, canbe fitted with a constant spectral index of − .
41. We can interpolatethem to estimate the missing zero-spacing flux in each spectral win-dow. We found that at frequencies of 4–6 GHz, 10–20 per cent ofthe flux density was missed by the VLA, which increased to 30–40per cent at frequencies of 7–9 GHz. In order to correct for this, wemerged the VLA and E ff elsberg data using IMERG in AIPS . We usedthe VLA 1 . . ff els-berg map. We hence interpolated the 1 . . ff elsberg maps at an angular resolution of 78 arcsec, assuming aconstant but spatially resolved spectral index, to have a template ofthe large-scale emission in each spectral window. We merged thedata of each spectral window with the appropriate template, mak-ing sure that the integrated flux densities of several regions agreedto within 5–10 per cent and the discrepancy between the total inte-grated flux densities was less than 4 per cent. This was achieved byadjusting the ‘uvrange’ parameter in IMERG , which prescribes theangular scale at which the single-dish image is scaled to interfero-metric image; we used values within the range of 0 . . λ . In Fig. 1(a), we present a 6 . ff elsberg observations at 3 . I map from LITTLE THINGS(Hunter et al. 2012). The NSB is centred on RA 00 h m s . . ◦ ′ ′′ , which is 5 arcsec south-west of IC 10 X-1, andhas a diameter of 54 arcsec or 184 pc. We created a map ofthe thermal (free–free) emission from the Balmer H α emissionmap of Hunter & Elmegreen (2004) following standard conver-sion (e.g., Deeg, Duric & Brinks 1997, equation 3, T = K), AIPS, the Astronomical Image Processing Software, is free softwareavailable from the NRAO. c (cid:13) , 000–000 he non-thermal superbubble in IC 10
28 26 24 220.4 0.6 0.8 1.00.1 mJy/beam 0.2 0.33015 D E C L I NA T I O N ( J2000 )
16 4559 17 45 (b)(c) D E C L I NA T I O N ( J2000 ) (a) Jy / b ea m * k m / s D E C L I NA T I O N ( J2000 )
00 20 40 00 20 30 00 20 20RIGHT ASCENSION (J2000) 200300200 pc59 1859 1759 1659 15
HI Integrated Intensity
Young Stellar Clusters
Non−thermal 6.2 GHzThermal 6.2 GHz(H−alpha + 32 GHz)
Position of IC10 X−1Slice for PV−Diagram
00 20 34 + 6.2 GHz Radio ( )
32 30
Figure 1. (a) Integrated H I emission line intensity as grey-scale at 5 . × . = ◦ ) resolution of an approximately 1 kpc region to the south-eastof the centre of IC 10. Contours show the 6 . µ Jy beam − , i.e. the superposition of thermal and non-thermalemission. The white line corresponds to the slice used to extract the PV -diagram at an angle of − ◦ , centred on the H I hole (see the text for details). Greenplus signs show the position of stellar clusters (Hunter 2001). (b) Non-thermal radio continuum at 6 . α and 32 GHz emission. The dashed line indicates the 80 per cent attenuation level of the primary beam at 32 GHz. In panels (a)–(c), themagenta star indicates the position of IC 10 X-1 and the angular resolution of the radio data is 3 . where we corrected for foreground absorption using E ( B − V ) = .
75 mag (Burstein & Heiles 1984). This map was combined withour 32 GHz map of the south-eastern starburst region, which we useas an extinction free measurement of the thermal radio continuumemission (Fig. 1 c). A comparison between the two maps showedagreement to within 10–20 per cent in areas outside of the compactH II regions ( I th < . − ), indicating that our estimate ofthe optical foreground absorption is accurate.The main fraction of thermal radio continuum is located inthe H II regions, north-west of the NSB. Whereas the NSB isprominent in the non-thermal radio continuum, there has thus farbeen little other evidence reported in the literature that the NSBconstitutes a cavity in the ISM. Wilcots & Miller (1998) find anH I hole at its position, but do not report any signs of expan-sion. There is weak, di ff use H α emission from ionized hydrogenand an increased line width, corresponding to a thermal veloc-ity dispersion of 35 km s − , but nothing to suggest an expand-ing shell (Thurow & Wilcots 2005). Using the natural weightedH I data cube from LITTLE THINGS (FWHM = . × . = ◦ ; Hunter et al. 2012), we have created a position–velocitydiagram of the NSB and its surroundings, presented in Fig. 2. Wefind a cavity of little prominence, centred on RA 00 h m s . . ◦ ′ . ′′
9, which is 4 . I hole with a diameter of 100 pc and an extent in velocityspace of 30 km s − , or consists of two smaller H I holes with diam-eters of 76 pc and extents in velocity space of 18 km s − each. Fora single hole the expansion velocity is 15 km s − , leading to an es- HI bubble
Figure 2.
Position–velocity ( PV ) diagram of the NSB and its surroundings,from the LITTLE THINGS H I data cube. The position of the slice is shownin Fig. 1(a). South-east is to the left, north-west to the right. timate of the bubble’s dynamical age of τ dyn = . ± . BRATS ; Harwood et al. 2013).The spectrum of the NSB presented in Fig. 3 (flux densities aretabulated in Table 1) shows a conspicuous curvature, which canbe well fitted by a Ja ff e–Perola (JP; Ja ff e & Perola 1973) model,shown as red solid line, with an injection spectral index of α inj = c (cid:13) , 000–000 V. Heesen et al. log10 (Frequency [Hz])9.3 9.5 9.6 9.7 9.8 9.9 l og ( F l u x d e n s i t y [ Jy ]) −1.8−1.9−2.0 Injection index: 0.60B−Field: 5.60E−09 T Spectral Age: 0.62 MyrReduced Chi−Squared: 1.26 −1.4−1.5−1.6−1.7 9.2 9.4
Figure 3.
Non-thermal spectrum of the NSB between 1 . . ff e–Perola model fit to the data and the solid blackline is a linear fit to data points > . Table 1.
Non-thermal flux densities of the NSB. ν (GHz) S ν (mJy) ν (GHz) S ν (mJy) ν (GHz) S ν (mJy)1 .
52 40 . .
32 15 . .
59 10 . .
55 17 . .
45 15 . .
72 10 . .
68 17 . .
95 11 . .
85 10 . .
81 16 . .
08 11 . .
01 10 . .
93 16 . .
21 11 . .
27 10 . .
06 16 . .
33 11 . .
53 10 . .
19 15 . .
46 11 . .
78 9 . . ± .
1. The JP model describes the evolution of radio continuumemission from a cosmic ray electron (CRe) population within aconstant magnetic field strength following a single-injection. Thereexist variations to the JP model, such as the KP (Kardashev 1962)and Tribble (Tribble 1993) model. Our data cannot di ff erentiatebetween them as any di ff erences are only notable close to thebreak frequency. Assuming energy equipartition and using the re-vised equipartition formula by Beck & Krause (2005), we find atotal magnetic field strength of 44 ± µ G ( U B = . ± . × − erg cm − ). The total infrared luminosity from Spitzer
MIPS24–160 µ m maps from Dale & Helou (2002) lead to a radiation en-ergy density of U rad = U star + U IR = ± × − erg cm − , wherethe contribution from stellar light is taken as U star = . × U IR asmeasured in the solar neighbourhood (Draine 2011).The spatially resolved distribution of the spectral age is shownin Fig. 4, where we applied a S / N-cuto ff of 5 in each pixel. Thereis an east–west gradient, where the age in the eastern part is τ = . h χ i = .
6) than for the integrated spectrum ( χ = . τ spec = . ± . χ = .
5, far inferior to the JP modelfit. Secondly, a power-law fit to data points > . . .
59 18 0017 45301500301516 4515 4500 D E C L I NA T I O N ( J2000 )
00 20 36 34 32 30 28 26 24 22 20 18RIGHT ASCENSION (J2000) 1.0 Myr0.80.60.40.20.0
Position of IC10 X−1
Spectral age + 1.5 GHz NT Radio
Figure 4.
Spectral age of the cosmic ray electrons in the NSB and its en-vironment. The angular resolution is 5 . . × µ Jy beam − andthe yellow star the position of IC 10 X-1. We first review the parameters derived for the IC 10 NSB: the inte-grated current cosmic ray energy in the NSB is 1 × erg, wherewe modelled the bubble as a sphere and used the assumption ofenergy equipartition ( U CR = U B ), injected approximately 1 Myrago. Following the calculations of Bagetakos et al. (2011), we canderive the energy required to create the H I hole as 0 . × erg,with the upper value appropriate if two holes were formed, wherethe ambient density of the neutral, atomic gas is 0 . . − (in-cluding helium). The contribution from turbulent gas within theNSB traced in H α (Thurow & Wilcots 2005) is probably not sig-nificant when taking into account that the filling factor for emissionline gas is probably low.We can compare our findings with 3D simulations byKrause et al. (2013, 2014). They predict that superbubbles reachdiameters of the order of 100 pc even before the first SN. Each SNthen first heats the bubble, accelerates the shell, and then dissipatesthe injected energy entirely at the leading radiative shock wave, andvia radiative cooling in mixing regions at the location of the shell,on a time-scale of a few 10 yr. The shell slows down accordingly,resulting in a discrepancy between spectral and dynamical age. Therather low shell velocity of the IC 10 NSB (high-velocity superbub-bles have a few times faster shells, compare e.g. Oey 1996) is in-deed expected if the last embedded SN exploded about 1 Myr ago,as suggested by the non-thermal emission. The CRe would havebeen accelerated as the accompanying shock wave traversed thebubble. Using the method of Bagetakos et al. (2011) on the afore-mentioned 3D simulations at a similar time, we find 10 erg asminimum energy to create the cavity, in agreement with the upperlimit from the observations. The energy found in cosmic rays ishowever surprisingly large. Assuming an acceleration e ffi ciency of10 per cent (e.g. Rieger, de O˜na-Wilhelmi & Aharonian 2013), atleast 10 erg would have to have been released.Could this have happened in a single explosion? Highly en-ergetic SNe are thought to be related to long duration gamma-raybursts (e.g. Mazzali et al. 2014, and references therein). The asso-ciated Type Ic SNe have energies of up to a few times 10 erg,adequate to account for our observations. We note that a higher c (cid:13) , 000–000 he non-thermal superbubble in IC 10 energy than the standard 10 erg would also better explain thehigh shell velocities in some high-velocity superbubbles (Oey1996; Krause & Diehl 2014). It is noteworthy that the NSB is cen-tred to within 16 pc on IC 10 X-1, suggesting an association.This system contains at least one massive star, [MAC92] 14A,which has a mass larger than 17 M ⊙ and more likely 35 M ⊙ (Silverman & Filippenko 2008), also a possible progenitor for aType Ic SN. Alternatively, multiple SNe may have exploded inthe past 1 Myr. We cannot rule this out from the spectral ageinganalysis, because a constant CRe injection rate since approximately1 Myr would still lead to a spectral downturn, caused by the old-est CRe. However, the position of the stellar clusters (Fig. 1 a) andthe distribution of the thermal radio continuum emission and hencethat of massive stars (Fig. 1 c), argues against this scenario, be-cause there is no spatial correlation. It is, however, conceivable thata less massive SN has exploded more recently, o ff set from IC 10X-1, which could explain the east–west gradient in the spectral agedistribution.Another way to explain the presence of non-thermal particleswould be perhaps the energy release from IC 10 X-1. It is a debatedpossibility that the X-ray emission of black hole binaries partiallyoriginates from a jet in addition to that of the more conventional X-ray corona (Grinberg et al. 2014). If the current X-ray luminosityof IC 10 X-1, 10 erg s − (Barnard et al. 2014), comes exclusivelyfrom the jet, an outburst length of 1 Myr would be su ffi cient toexplain the cosmic ray energy, assuming a 10 per cent acceleratione ffi ciency. One would then have to explain why this channel wasso active in the past and by now has ceased almost entirely with nocompact radio source observed in the vicinity of IC 10 X-1. In this Letter, we have presented a multiband radio continuum studyof the NSB in the nearby starburst dwarf irregular galaxy IC 10.Conventional wisdom tells us that dwarf galaxies are weak in non-thermal (synchrotron) emission, being easily subjected to outflowsand winds and not likely able to retain cosmic rays. IC 10 is no ex-ception, it has a large thermal fraction of 50 per cent at 6 GHz and isunderluminous in terms of its radio continuum emission comparedto its star-formation rate (Heesen et al. 2011, 2014). However, highspatial resolution observations (10–20 pc) show complex cosmicray and magnetic field distributions. The NSB stands out as thebrightest non-thermal structure and its spectrum shows a conspic-uous downturn towards higher frequencies, something that to datehas rarely been observed in any nearby galaxy.We fit a JP spectral model to the data, which describes the ra-dio continuum emission of an ageing population of CRe in a con-stant magnetic field. Estimating the magnetic field from equiparti-tion and the radiation energy density from
Spitzer
MIPS maps, wefind a spectral age of τ spec = . ± . τ dyn = . ± .
3, measured from the ex-pansion speed of its corresponding ‘H I hole’. Our results suggestthat the NSB was generated by the wind of the progenitor of IC 10X-1, a massive stellar mass black hole, during its main-sequencelife and WR phase. Considering alternative explanations, we findthat most likely a single energetic explosion of the progenitor ofIC 10 X-1 released & erg, accelerating the non-thermal par-ticles and the shell at the same time. The latter than slowed downvia interaction with the ambient medium to its current velocity of15 km s − . We are observing the NSB in the early stages of its evolution, of what may become over the next few 10–50 Myr a su-perbubble of a few hundred parsec size visible as a large H I hole. ACKNOWLEDEGEMENTS
VH acknowledges support from the Science and Technology Facili-ties Council (STFC) under grant ST / J001600 /
1. MK acknowledgessupport by the DFG cluster of excellence ‘Origin and Structure ofthe Universe’ and by the ISSI project ‘Massive star clusters acrossthe Hubble time’. JJH wishes to thank the University of Hertford-shire for generous financial support and STFC for a STEP award.We thank our referee, Biman Nath, for a constructive and thought-ful report.
REFERENCES
Bagetakos I., Brinks E., Walter F., de Blok W. J. G., Usero A., Leroy A. K.,Rich J. W., Kennicutt, R. C., Jr. 2011, AJ, 141, 23Barnard R., Steiner J. F., Prestwich A. F., Stevens I. R., Clark J. S., KolbU. C., 2014, ApJ, 792, 131Bauer F. E., Brandt W. N., 2004, ApJ, 601, L67Beck R., Krause M., 2005, Astron. Nachr., 326, 414Burstein D., Heiles C., 1984, ApJS, 54, 33Chy˙zy K. T., Knapik J., Bomans D. J., Klein U., Beck R., Soida M., Ur-banik M., 2003, A&A, 405, 513Chy˙zy K. T., We˙zgowiec M., Beck R., Bomans D. J., 2011, A&A, 529,A94Dale D. A., Helou G., 2002, ApJ, 576, 159Deeg H.-J., Duric N., Brinks E., 1997, A&A, 323, 323Dorfi E. A., Breitschwerdt D., 2012, A&A, 540, A77Draine B. T., 2011, Physics of the Interstellar and Intergalactic Medium,Princeton University Press, Princeton, NJEverett J. E., Zweibel E. G., Benjamin R. A., McCammon D., Rocks L.,Gallagher, III J. S., 2008, ApJ, 674, 258Governato F. et al., 2010, Nature, 463, 203Grinberg V. et al., 2014, A&A, 565, A1Hanasz M., Lesch H., Naab T., Gawryszczak A., Kowalik K., W´olta´nskiD., 2013, ApJ, 777, L38Harwood J. J., Hardcastle M. J., Croston J. H., Goodger J. L., 2013, MN-RAS, 435, 3353Heesen V., Brinks E., Leroy A. K., Heald G., Braun R., Bigiel F., Beck R.,2014, AJ, 147, 103Heesen V., Rau U., Rupen M. P., Brinks E., Hunter D. A., 2011, ApJ, 739,L23Hunter D. A., 2001, ApJ, 559, 225Hunter D. A., Elmegreen B. G., 2004, AJ, 128, 2170Hunter D. A. et al., 2012, AJ, 144, 134Ja ff e W. J., Perola G. C., 1973, A&A, 26, 423Kardashev N. S., 1962, SvA, 6, 317Krause M., Diehl R., B¨ohringer H., Freyberg M., Lubos D., 2014, A&A,566, A94Krause M., Fierlinger K., Diehl R., Burkert A., Voss R., Ziegler U., 2013,A&A, 550, A49Krause M. G. H., Diehl R., 2014, ApJ, 794, L21Lozinskaya T. A., Moiseev A. V., 2007, MNRAS, 381, L26Massey P., Holmes S., 2002, ApJ, 580, L35Mazzali P. A., McFadyen A. I., Woosley S. E., Pian E., Tanaka M., 2014,MNRAS, 443, 67Moore B., Ghigna S., Governato F., Lake G., Quinn T., Stadel J., Tozzi P.,1999, ApJ, 524, L19Oey M. S., 1996, ApJ, 467, 666Oh S.-H., de Blok W. J. G., Brinks E., Walter F., Kennicutt, R. C., Jr.2011a, AJ, 141, 193Oh S.-H., Brook C., Governato F., Brinks E., Mayer L., de Blok W. J. G.,Brooks A., Walter F., 2011b, AJ, 142, 24c (cid:13) , 000–000 V. Heesen et al.
Pakmor R., Marinacci F., Springel V., 2014, ApJ, 783, L20Perley R. A., Butler B. J., 2013, ApJS, 204, 19Rau U., Cornwell T. J., 2011, A&A, 532, A71Rieger F. M., de O˜na-Wilhelmi E., Aharonian F. A., 2013, Frontiers Phys.,8, 714Salem M., Bryan G. L., 2014, MNRAS, 437, 3312Scannapieco C., 2013, Astron. Nachr., 334, 499Siejkowski H., Otmianowska-Mazur K., Soida M., Bomans D. J., HanaszM., 2014, A&A, 562, A136Silverman J. M., Filippenko A. V., 2008, ApJ, 678, L17Strickland D. K., Heckman T. M., 2009, ApJ, 697, 2030Thurow J. C., Wilcots E. M., 2005, AJ, 129, 745Tremonti C. A. et al., 2004, ApJ, 613, 898Tribble P. C., 1993, MNRAS, 261, 57Wilcots E. M., Miller B. W., 1998, AJ, 116, 2363Yang H., Skillman E. D., 1993, AJ, 106, 1448This paper has been typeset from a TEX / L A TEXfile prepared by the au-thor. c (cid:13)000