To appear in “XIII Latin American Regional IAU Meeting (2010)” RevMexAA(SC) SUPERNOVA FEEDBACK AND THE BEND OF THE TULLY-FISHER RELATION M. E. De Rossi,1,2,3 P. B. Tissera,1,2 and S. E. Pedrosa1,2 RESUMEN 1 Estudiamos el origen de la relaci´on de Tully-Fisher analizando simulaciones hidrodin´amicas en un universo 1 ΛCDM. Encontramos que las galaxias m´as pequen˜as exhiben masas estelares menores que aquellas predichas 0 por el ajuste lineal de las galaxias m´as masivas, en consistencia con las observaciones. En este modelo, estas 2 tendencias son generadas por la acci´on m´as eficiente de la retroalimentaci´on al medio por Supernovas en la n regulaci´on de la formacio´n estelar de las galaxias m´as pequen˜as. Sin introducir para´metros dependientes de a escala,elmodeloprediceuncambiodependienteenlarelaci´ondeTully-Fisherparaunavelocidadcaracter´ıstica J de ∼100 km s−1, en acuerdo con resultados observacionales y te´oricos previos. 2 1 ABSTRACT ] We have studied the origin of the Tully-Fisher relation by analysing hydrodynamical simulations in a ΛCDM O universe. We found that smaller galaxies exhibit lower stellar masses than those predicted by the linear fit to C high mass galaxies (fast rotators), consistently with observations. In this model, these trends are generated h. by the more efficient action of Supernova feedback in the regulation of the star formation in smaller galaxies. p Without introducing scale-dependent parameters,the model predicts that the Tully-Fisher relationbends at a - characteristic velocity of ∼100 km s−1, in agreement with previous observationaland theoretical findings. o r Key Words: cosmology: theory — galaxies: evolution — galaxies: formation t s a 1. INTRODUCTION bTFR by analysing hydrodynamical simulations in [ The Tully-Fisher relation is very important for a ΛCDM universe 4. A version of the chemical 1 constraining galaxy formation models because it code GADGET-3 including treatments for metal- v links two fundamental properties of galaxiessuch as dependent radiative cooling, stochastic star forma- 7 6 thestellar(sTFR)orbaryonic(bTFR)massandthe tionandSNfeedback(Scannapieco et al. 2006)was 3 depth of the potential well. employed. The simulated volume corresponds to a 2 In particular, there are observational evidences cubicboxofacomoving10Mpch−1sidelength. The 1. that the sTFR does not follow a linear trend for simulation has a mass resolution of 6×106M⊙h−1 0 the whole range of observed rotation velocities. Ac- and 9×105M⊙h−1 for the dark and gas phase, re- 1 cordingtotheresultsofMcGaugh et al. (2000),the spectively. 1 sTFR bends at ∼ 90 km s−1 in such a way that Simulated disc-like galaxies were identified fol- : v smallergalaxieshavelowerstellarmassesthanthose lowing the methods describe by De Rossi et al. Xi derivedfromtheextrapolationofthelinearfittofast (2010). The mean properties of galactic systems rotators. Moreover,McGaugh et al. (2010)havere- were estimated at the baryonic radius Rbar, defined r a ported that there is also a bend in the bTFR but at astheonewhichencloses83percentofthebaryonic a lower rotation velocity (∼20 km s−1). mass of the systems. We found that the tangential velocity of these systems constitutes a good repre- 2. SIMULATIONS sentation of their potential wellso that, for the sake In this work, we studied the role of Super- ofsimplicity, weusedthe circularvelocityestimated nova (SN) feedback on the shape of the sTFR and at R as the kinematical indicator for our study. bar 1InstitutodeAstronom´ıayF´ısicadelEspacio,CONICET- 3. RESULTS AND DISCUSSION UBA, CC 67, Suc 28 (1428), Ciudad Aut´onoma de Buenos Aires, Argentina ([email protected], alternative address: The massive-end of the simulated sTFR can be [email protected]). 2Consejo Nacional de Investigaciones Cient´ıficas y fittedbyapower-lawoftheformlog(M∗/M⊙h−1)= T´ecnicas,CONICET,CiudadAut´onomadeBuenosAires,Ar- (3.68±0.09)×log(V/ 100 km s−1) +(9.42±0.26), gentina. 3FacultaddeCienciasExactasyNaturales,Universidadde 4Ωm = 0.3, ΩΛ = 0.7, Ωb = 0.04 and H0 = BuenosAires,CiudadAut´onomadeBuenosAires,Argentina. 100h−1kms−1Mpc−1 withh=0.7 1 2 DE ROSSI, TISSERA, & PEDROSA in general good agreement with observations. How- self-regulated cycle of heating and cooling strongly ever, at rotation velocities below ∼100 km s−1, the influencing the star formation activity of these sys- sTFR becomes steep and the residuals of the linear tems. In the case of massive galaxies, the hot phase fitdepartsystematicallyfromzero,consistentlywith is established at a higher temperature and SN heat- the findings of McGaugh et al. (2000). To analyse ing cannotgenerate an efficient transitionof the gas the role of SN feedback on the origin of the bend from the cold to the hot phase. Meanwhile, the of the sTFR, we run a feedback-free simulation. We cold gas remains available for star formation. On foundthat,whenSNfeedbackissuppressedfromthe the other hand, the cooling times for these galaxies model, the sTFR describes a linear behaviour indi- getlongercomparedto the dynamicaltimesandthe cating the crucial role of SNe in the modulation of hot gas is able to remain in the hot phase during this relation. longer time-scales. Hence, SN feedback is not effi- WithrespecttothebTFR,boththeSNfeedback cientatregulatingthestarformationinlargergalax- model and the feedback-free run show a single slope ies. Interestingly, in this model, the transition from for the relation at least for the range of velocities the efficient to the inefficient cooling regime for the resolved by these simulations (40 km s−1 < V < hot-gasphaseoccursatthesamecharacteristicrota- 250 km s−1). These results are not in disagreement tionvelocity wherethe sTFRbends (∼100km s−1) withthefindingsofMcGaugh et al. (2010)because andisalsoinagreementwith previousobservational they reported a bend in the bTFR at ∼ 20 km s−1. (e.g. McGaugh et al. 2000) and theoretical works (Larson 1974; Dekel & Silk 1986). By analysing the simulations at higher redshifts (0 < z < 3), we obtained similar trends. We have 4. CONCLUSION also checked that our results are robust against nu- WestudiedtheTully-Fisherrelationbyperform- merical resolution. For further details, the reader is ing numerical simulations in a cosmological frame- referred to De Rossi et al. (2010). work. We obtained a sTFR and a bTFR in gen- Finally, we explore how the SN feedback model eral agreement with observations. The SN feedback is capable of reproducing these behaviours without model is able to reproduce the observedbend of the introducing scale-dependent parameters. To anal- sTFR at a characteristic velocity of ∼ 100 km s−1 yse the impact of galactic outflows in the simulated without introducing scale dependent parameters. galaxies,we definedthefractionfvir asthe ratiobe- b We found that this characteristic velocity separates tween the baryonic mass within the virial radius to two different regimes: slow-rotators develop a self- theoneinferredfromthe universalbaryonicfraction regulated cycle of efficient SN heating and radiative (Ω /Ω ). In these simulations, fvir is within the b m b cooling while, in fast-rotators, the cold an hot gas- range 0.2−0.8 showing that for the whole sample phases seem to be more disconnected. The reader is a significant percentage of the gas in blown away as referred to De Rossi et al. (2010) for more details a consequence of efficient galactic winds. The more about this work. prominent losses are obtained for smaller galaxies. MEDR and SEP thank the Organizing Commit- Moreover, by comparing the SN feedback model tee for its partial financial support to attend this with the feedback-free run, we found that SN winds meeting. We acknowledge support from the PICT generateanimportantdecreaseinthestarformation 32342 (2005) and PICT 245-Max Planck (2006) of activity of galaxies with the larger effects in slow- ANCyT (Argentina). Simulations were run in Fenix rotating systems, consistently with the bend of the and HOPE clusters at IAFE and Cecar cluster at sTFR. Indeed, our results suggest that the role of University of Buenos Aires. SN feedback on the regulation of the star formation strongly depends on the mass of the galaxies. In REFERENCES particular, we distinguish two distinct regimes for Dekel, A.& Silk,J. 1986, ApJ, 303, 39 the thermodynamical transitions of the gas phase. De Rossi, M. E., Tissera, P. B., & Pedrosa, S. E. 2010, In the case of smaller galaxies, the virial temper- A&A,519, 89 atures are lower and SN heating is more efficient Larson, R.B. 1974, MNRAS,169, 229 at promoting gas from a cold to a hot phase (see McGaugh, S.S.,Schombert,J.M., Bothun,G.D.,&de Scannapieco et al. 2006 for more details about the Blok, W. J. G. 2000, ApJ,533, L99 McGaugh,S.S.,Schombert,J.M.,deBlok,W.J.G.,& model). However,the coolingtimesofthesesystems Zagursky, M. J. 2010, ApJ, 708L, 14 are shorter than the dynamical times and the hot Scannapieco, C., Tissera, P.B., White, S.D.M. & gascanreturntothecoldphaseinshorttime-scales. Springel, V.2006, MNRAS,371, 1125 Therefore, for slow-rotators, SN feedback leads to a