L109 the Radio Supernebula in Ngc 5253 Jean L. Turner Sara

Home , NGC 5253

THE RADIO SUPERNEBULA IN NGC 5253 Jean L. Turner Department of Physics and Astronomy, UCLA, Los Angeles, CA 90095-1562; [email protected] Sara C. Beck Department of Physics and Astronomy, Tel Aviv University, Ramat Aviv, Israel; [email protected] and Paul T. P. Ho Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138; [email protected] Received 1999 December 30; accepted 2000 February 14; published 2000 March 9

ABSTRACT We present 1.3 cm and 2 cm subarcsecond resolution VLA images of the dwarf galaxy NGC 5253. Within ∼ the central starburst, we detect high-brightness [Tb (2 cm) 100–12,000 K] radio continuum sources. These appear to be very dense, “compact” H ii regions. The dominant radio source is a nebula ∼1–2 pc in size, requiring several thousand O stars within the volume to maintain its ionization. This nebula has no obvious optical counterpart. The number of ionizing photons we find for this cluster is nearly 2 orders of magnitude larger than indicated by Ha fluxes, and the deduced stellar content accounts for a significant fraction of the total infrared luminosity of the galaxy. This cluster is a strong candidate for a globular cluster in the process of formation, perhaps the youngest globular cluster known. Subject headings: galaxies: dwarf — galaxies: individual (NGC 5253) — galaxies: peculiar — galaxies: starburst — galaxies: star clusters — radio continuum: galaxies

1. INTRODUCTION 2. OBSERVATIONS NGC 5253 is a dwarf I0/S0 galaxy at a distance of 4 Mpc The 2 and 1.3 cm observations were made with the A con- ∼ # undergoing a starburst of infrared luminosity L IR 1.8 figuration of the VLA on 1998 October 13. Absolute flux cal- 9 10 L, (Beck et al. 1996). The starburst in NGC 5253 is very ibration was based on observations of 3C 286 and is good to young, displaying the optical (Gorjian 1996; Calzetti et al. better than 5%. Phase center was at a =13h39m55.84,s d = 1997) and infrared (Rieke, Lebofsky, & Walker 1988) signa- Ϫ31Њ38Ј29Љ.1 (J2000). The naturally weighted synthesized tures of a starburst no more than a few million years old. Its beams are0Љ.22 # 0Љ.08 (FWHM) at P.A. = Ϫ8Њ for the 1.3 cm radio continuum emission is unique among galaxies for being map and0Љ.33 # 0Љ.12 at P.A.=1Љ.9 for the 2 cm map. The almost entirely thermal (Beck et al. 1996), which suggests that north-south elongation of the beam is caused by the foreshort- the star formation has not persisted for the time it takes to ening of the array, due to the southerly declination of the source. produce a significant number of supernovae. The two bands have different sensitivities to lower spatial fre- Radio observations at 6, 3.6, and 2 cm made with the NRAO1 quencies, so we cannot directly compare total fluxes or compute Very Large Array with 1Љ–2Љ resolution indicated that the radio spectral indices from the maps. Because of the lack of short free-free emission in NGC 5253 is partly optically thick at spacings, the 2 and 1.3 cm maps are sensitive only to structures centimeter wavelengths (Turner, Ho, & Beck 1998). Classical Շ2Љ–4Љ in the east-west direction and Շ4Љ–8Љ north-south. Galactic H ii regions such as Orion are optically thick at wave- These observations were designed to determine brightness lengths around 1 m; the high densities needed for a source to temperatures and the structure and nature of compact radio be optically thick at 2 cm are seen in the Galaxy only in the sources rather than total fluxes, which are available from lower youngest ultracompact H ii regions, of diameter ∼0.01 pc (Hab- resolution observations (Turner et al. 1998). There was an in- ing & Israel 1979; Wood & Churchwell 1989). Thermal sources sufficient signal-to-noise ratio for self-calibration without sig- of this size cannot be detected in other galaxies (Turner & Ho nificant loss of spatial resolution, and so the measured source 1994). The thermal radio continuum emission in NGC 5253 is sizes may be broadened by seeing at the fractional arcsecond 1–2 orders of magnitude stronger than predicted by the ob- level, with a more pronounced effect expected at 1.3 cm. We served Ha emission (Calzetti et al. 1997), which implies that calculate from comparison with lower resolution maps (Turner the extinctions are substantial, another indicator of youth. This et al. 1998) that our 2 cm maps resolve out ∼50% of the total is very likely the location of the youngest part of the starburst, emission within the central20 # 40 , although they appear which is optically invisible. to recover all of the 2 cm emission in the inner6 # 6 (Beck To minimize the possibility of confusion by sources with et al. 1996). The images were deconvolved from the dirty different optical depths or synchrotron emission in the same beams using the CLEAN algorithm. The rms noise levels in beam, we reobserved the radio core of NGC 5253 at 2 cm with the continuum maps are 0.09 and 0.11 mJy beamϪ1 for 2 and the A configuration of the VLA, which gives a synthesized 1.3 cm, respectively. beam of 0Љ.1–0Љ.3 in size. We also extended the spectrum to 1.3 cm. It is on these observations that we report in this Letter. 3. THE SUPERNEBULA AND ITS ENVIRONS

1 The National Radio Astronomy Observatory is a facility of the National In Figure 1 we display the 1.3 and 2 cm VLA maps of the Science Foundation operated under cooperative agreement by Associated Uni- central 3Љ (50 pc) of the starburst in NGC 5253. The maps versities, Inc. show a dominant radio continuum source at both 2 and L109 L110 SUPERNEBULA IN NGC 5253 Vol. 532

# 2n/2,n=00 , 1, 2), times 0.3 mJy beamϪ1 (3 j). The beam size is Љ.33ע Fig. 1.—Left: 2 cm VLA map of NGC 5253. The map is contoured at levels of 0Љ.12, P.A. = 2 . The peak ﬂux density is 11.8 mJy beamϪ1. The size of the main source, which is located at a =13h39m55.964,s d = Ϫ31Њ38Ј24Љ.38, is 0Љ.1 # 0Љ.04, or 2 pc # 1 pc. Right: 1.3 cm map. The contour levels are as for the left panel (2.5 j for this map). The beam size is0Љ.22 # 0Љ.08 , P.A. = Ϫ8 . The peak ﬂux density is 9.1 mJy beamϪ1. The location of the central source is the same as for the 2 cm source; the eastern secondary peak is at a =13h39m55.98,s d = Ϫ31Њ38Ј24Љ.4.

1.3 cm. Weaker sources appear to the north and east of the 13h39m55.964,s d = Ϫ31Њ38Ј24Љ.38, in agreement with lower res- main radio nebula. The sources appear elongated in the north- olution images (Turner et al. 1998). The absolute positions are south direction, which is mostly due to the north-south elon- known to the ∼0Љ.01 accuracy of the phase calibrator and gation of the synthesized beam. In Figure 2, the 1.3 cm map relative positions to 0Љ.002–0Љ.005, based on the signal-to- is superposed on the Ha image of Calzetti et al. (1997). The noise ratio. The 1.3 cm map has a secondary peak at a = dominant radio source does not have an obvious counterpart 13h39m55.982,s d = Ϫ31Њ38Ј24Љ.37. This peak is not evident at Շ in the Ha image. Given the high (Av 35 mag) extinction 2 cm. There is also either a separate source or an extension of (Calzetti et al. 1997) estimated for the region, it is unlikely the main source northward by ∼1Љ, which is apparent in both that the Ha traces these high-density nebulae at all. maps but stronger at 2 cm. The position of the dominant radio source is a = The main source is slightly resolved. The size of the source, deconvolved from the beam assuming a Gaussian source, is or about , 7 ע0Љ.02(FWHM) at P.A.=27 ע 0Љ.10 # 0Љ.05 2pc# 1 pc for an adopted distance of 4 Mpc (Saha et al. ,mJy beamϪ1 1 ע The peak flux fitted to the core is11 .(1995 with the uncertainty largely due to the difficulty of fitting the compact source properties within the lower level extended emission. The observed brightness temperature is therefore K, close to the value of electron temperature 4000 ע 12,000∽ determined for the visible H ii regions in NGC 5253, which is ∼11,000–12,000 K (Campbell, Terlevich, & Melnick 1986; Walsh & Roy 1989; Kobulnicky et al. 1997). This indicates that the free-free emission has a significant optical depth at 2 cm (t տ 1 ). The total flux estimate for the central 3Љ mapped region is ∼22 mJy at 2 cm, in agreement with lower resolution observations (Beck et al. 1996; Turner et al. 1998). At 1.3 cm, the peak flux density for the main source is mJy beamϪ1. The observed brightness temperature of 1 ע 9 this source is ∼4000 K, given the source size measured at 2 cm. Since the 1.3 cm map may be affected by seeing, the true peak flux density may actually be higher. It may also be that the spectrum is turning over and that the source is becoming optically thin at 1.3 cm. The total flux density of this source at 1.3 cm is 20 mJy. Since the 1.3 cm map is more affected Fig. 2.—Gray scale: Hubble Space Telescope Ha image of the central by undersampling than the 2 cm map, we cannot determine a starburst region of NGC 5253 (Calzetti et al. 1997). The white patch in the center of the Ha is 1.3 cm emission. The registration accuracy is limited by reliable spectral index. However, it is clear that the spectrum the Hubble Space Telescope image, estimated to be good to ∼1Љ. North is up, is approximately flat and that the turnover frequency of the and east is to the left. free-free emission is about 22 GHz. The flat spectrum appears No. 2, 2000 TURNER, BECK, & HO L111 to rule out supernova remnants or radio supernovae as the of the cluster exciting this nebula are characteristic of globular source of the emission. clusters and inferred for super–star clusters (SSCs), likely pre- The eastern source in the 1.3 cm image has a peak flux cursors to globular clusters, which are seen in many intense density of ∼1.4 mJy beamϪ1, nearly 3 times higher than the starbursts including NGC 5253 (Meurer et al. 1995). Most of 2 cm flux density of ∼0.5 mJy beamϪ1. The minimum bright- the SSCs seen in the optical and ultraviolet in NGC 5253 are ness temperature is 100 K. The source appears to have a spec- not associated with radio sources (Gorjian 1996). This is prob- trum that is still rising at 22 GHz. If a rising spectrum is an ably because the youngest clusters that are most likely to have indication of youth, as we argue below, then this may be the the densest gas and the bright radio nebulae are also most likely very youngest part of the young part of the starburst in NGC to be highly obscured. 5253. While the total number of stars in the cluster is difficult to determine to better than an order of magnitude based solely on

4. EXCITATION OF THE SUPERNEBULA NLyc, because of uncertainties in the IMF and mass cutoffs, the luminosity of the cluster can be more tightly constrained, since Since the compact source is at least partly optically thick, it arises in the same O stars that produce the ionizing photons. nebular quantities such as the Lyman continuum rate N that Lyc The total stellar luminosity L of a zero-age cluster of O3–G are derived directly from the radio fluxes assuming optically OB stars with a Salpeter IMF is (again using the O star properties thin emission may underestimate the true Lyman continuum Ϫ of Vacca et al. 1996)L /N ∼ (2–3) # 10 44 . If all the stellar rates. However, since the spectrum appears to be turning over OB Lyc luminosity is seen in the infrared, as is the usual assumption at about 1.3 cm, the emission is not very thick and the optically for such an obscured source, the total infrared luminosity for thin assumption at these high frequencies yields a reasonable ∼ # 9 thin the observed NLyc will be L OB (0.8–1.2) 10 L,. The ob- estimate of NLyc. The “optically thin” estimates are N=Lyc # 9 # 49 2 2 cm# 52 Ϫ1 served IRAS fluxes give a luminosity of 1.8 10 L,. The 9 10 (D/Mpc) S=mJy 3 10 s for the dominant Ϫ luminosity we obtain for this cluster is at least half of the source, andN=thin(2–3) # 10 51 s 1 for the eastern 1 cm Lyc luminosity of the entire galaxy. What are the possible expla- source. nations for the low L /N ratio that must hold for this cluster? N can also be estimated from the size and emission mea- IR Lyc Lyc We do not have a definite answer to this question; this is sure of optically thick H ii regions, sinceN ∝ nV2 for an Lyc e the first source of its type to be observed. The combination of ionization-bounded nebula, and the emission measure EM = an extremely intense radiation field and high gas density in a ∫ ndl2 and source area (hence volume) can be determined from e very compact volume may put this source outside the usual the radio data. The turnover frequency n , where the free-free t relation between gas, dust, and stellar population which is re- spectrum changes from optically thick to thin, is related to the Ϫ flected in the “normal” L /N (e.g., Genzel et al. 1982) ratio. emission measure via (EM/cm 6 pc) = 12.1(n /GHz)2.1T 1.35 (see IR Lyc te It could be that the IMF is exceedingly top heavy, although Mezger & Henderson 1967). Since the turnover frequency n t this does not appear to be the case in another large young of the central source appears to be ∼22 GHz, EM ∼ (2–3) # Ϫ cluster, R136 (Massey & Hunter 1998). There are possibly other 10 9 cm 6 pc. The size of the nebula is determined to be energy sinks within this nebula in addition to dust. Since the 2pc# 1 pc from the 2 cm map. If we assume the line-of- nebula is very young, it may be that transient mechanisms are sight dimension is similar, an rms electron density of 21/2∼ # 4 Ϫ3 involved. For example, dissociation of the surrounding molec- Ane S 4 10 cm obtains, which is high for an H ii region 8 ular clouds and shocks are possible sinks of energy. A 10 M, of this size. The implied mass of ionized gas in the nebula is ∼ 4 molecular cloud of H2 molecules would last 10 yr at the ∼4000 M,. current L , at 4.5 eV per dissociation. Dissociation of molec- Based on our estimate ofnV2 , we find the “optically thick” OB e ular clouds is thus not a viable long-term energy sink, but could Lyman continuum rate required to ionize the volume of high- Ϫ well be operating in the short term in an extremely young density gas isN thick∼ 4 # 10 52 s 1. This is about 70 times Lyc source. Other sources of energy dissipation include winds and higher than what would be predicted on the basis of Ha fluxes shocks, for which there is ample evidence (Martin & Kennicutt (Calzetti et al. 1997). However, our N is within a factor of Lyc 1995; Marlowe et al. 1995; Calzetti et al. 1999). Another al- 2 of the observed Brackett line flux (Kawara, Nishida, & Phil- ternative to additional energy sinks is that there may be ad- lips 1989) prediction (Beck et al. 1996). ditional ionization sources within the nebula that have signif- icantly less total luminosity for a given ionization than does 5. THE CLUSTER WITHIN THE SUPERNEBULA an O star, such as Wolf-Rayet stars or X-ray binaries. The best- We have estimated that the Lyman continuum rate for this known nonstellar ionization source is an active galactic nucleus nebula is ∼4 # 10 52 sϪ1. We have assumed an ionization- (AGN). We cannot with these observations rule out the pos- bounded nebula; if the nebula is density-bounded, NLyc will be sibility that the ionization source is a mini-AGN, although there higher. If we assume that the average OB star is an O7 star, is little spectroscopic evidence for such a source in NGC 5253 the number of O7 stars required to excite the nebula is ∼4000. (Lutz et al. 1996; Calzetti et al. 1999). However, if this is an These stars must be located within a ∼1–2 pc diameter region, AGN, it is in a radically different environment from the usual or else the ultraviolet excitation requirements are even more AGN host. extreme. For a Salpeter initial mass function (IMF; which seems to hold for R136 in the Large Magellanic Cloud [Massey & 6. CONCLUSIONS Hunter 1998]) with an upper mass cutoff of 90 M, (O3) and lower mass cutoff of 1 M,, we estimate from the OB star VLA maps at 2 and 1.3 cm, with the A array, reveal a bright Љ # Љ # properties of Vacca, Garmany, & Shull (1996) that this NLyc radio nebula0.1 0.05 or 2 pc 1 pc in size in the starburst would correspond to ∼2 # 105 stars and as many as ∼ 2 # dwarf galaxy NGC 5253. This source is partially optically thick 10 6 stars if the IMF extends down to M. The cluster mass is at 2 cm, indicating an emission measure of ∼3 # 109 cmϪ6 pc 5 4 Ϫ3 ∼5 # 10 M, for a zero-age cluster of O3–G stars and and an rms electron density of at least4 # 10 cm . The 6 ∼1 # 10 M, for O3–M. The size and deduced stellar mass Lyman continuum rate required to excite this source is equiv- L112 SUPERNEBULA IN NGC 5253 Vol. 532 alent to 4000 O7 stars. For a Salpeter IMF with upper and clusters and may mark the location of a young protoglobular lower mass cutoffs of 90 and 1 M,, respectively, this would cluster. imply a cluster of at least2 # 10 5 stars within the 2 pc region and more if the IMF extends below 1 M,. The cluster luminosity predicted by observed Lyman continuum rate is We are grateful to Daniela Calzetti for the use of her HST 9 ∼1.0 # 10 L,. This cluster accounts for at least half of the Ha image and to an anonymous referee for helpful comments. infrared luminosity of the galaxy as observed by IRAS.We This work was supported in part by NSF grant AST 94-17968 suggest that this source, which we call a radio supernebula, is to J. L. T. and by US-Israel Binational Science Foundation the very young and gas-rich nebular counterpart of super–star grant 89-0070 to S. C. B.

REFERENCES

Beck, S. C., Turner, J. L., Ho, P. T. P., Lacy, J. H., & Kelly, D. 1996, ApJ, Marlowe, A. T., Heckman, T. M., Wyse, R. F. G., & Schommer, R. 1995, ApJ, 457, 610 438, 563 Calzetti, D., Conselice, C. J., Gallagher, J. S. III, & Kinney, A. L. 1999, AJ, Martin, C. L., & Kennicutt, R. C., Jr. 1995, ApJ, 447, 171 118, 797 Massey, P., & Hunter, D. A. 1998, ApJ, 493, 180 Calzetti, D., Meurer, G. R., Bohlin, R. C., Garnett, D. R., Kinney, A. L., Meurer, G. R., Heckman, T. M., Leitherer, C., Kinney, A., Robert, C., & Leitherer, C., & Storchi-Bergmann, T. 1997, AJ, 114, 1834 Garnett, D. R. 1995, AJ, 110, 2665 Campbell, A., Terlevich, R., & Melnick, J. 1986, MNRAS, 223, 811 Mezger, P. G., & Henderson, A. P. 1967, ApJ, 147, 471 Genzel, R., Becklin, E. E., Wynn-Williams, C. G., Moran, J. M., Reid, M. J., Rieke, G., Lebofsky, M., & Walker, C. 1988, ApJ, 325, 679 Jaffe, D. T., & Downes, D. 1982, ApJ, 255, 527 Saha, A., Sandage, A., Labhardt, L., Schwengeler, H., Tammann, G. A., Gorjian, V. 1996, AJ, 112, 1886 Panagia, N., & Macchetto, F. D. 1995, ApJ, 438, 8 Habing, H. J., & Israel, F. P. 1979, ARA&A, 17, 345 Turner, J. L., & Ho, P. T. P. 1994, ApJ, 421, 122 Kawara, K., Nishida, M., & Phillips, M. M. 1989, ApJ, 337, 230 Turner, J. L., Ho, P. T. P., & Beck, S. C. 1998, AJ, 116, 1212 Kobulnicky, H. A., Skillman, E. D., Roy, J.-R., Walsh, J. R., & Rosa, M. R. Vacca, W. D., Garmany, C. C., & Shull, J. M. 1996, ApJ, 460, 914 1997, ApJ, 477, 679 Walsh, J. R., & Roy, J.-R. 1989, MNRAS, 239, 297 Lutz, D., et al. 1996, A&A, 315, L137 Wood, D. O. S., & Churchwell, E. 1989, ApJS, 69, 831