arXiv:astro-ph/0401431 v1 21 Jan 2004 ar X iv :a st ro -p h/ 04 01 43 1 v1 21 J an 2 00 4 Energy Crisis in the Superbubble DEM L 192 (N 51D) Randall L. Cooper1,2, Mart́ın A. Guerrero1,3, You-Hua Chu1, C.-H. Rosie Chen1, Bryan C. Dunne1 Astronomy Department, University of Illinois, 1002 West Green Street, Urbana, IL 61801 ABSTRACT Superbubbles surrounding OB associations provide ideal laboratories to study the stellar energy feedback problem because the stellar energy input can be es- timated from the observed stellar content of the OB associations and the inter- stellar thermal and kinetic energies of superbubbles are well-defined and easy to observe. We have used DEM L 192, also known as N 51D, to carry out a detailed case study of the energy budget in a superbubble, and we find that the expected amount of stellar mechanical energy injected into the ISM, (18 ± 5) × 1051 ergs, exceeds the amount of thermal and kinetic energies stored in the superbubble, (6 ± 2) × 1051 ergs. Clearly, a significant fraction of the stellar mechanical energy must have been converted to other forms of energy. The X-ray spectrum of the diffuse emission from DEM L 192 requires a power-law component to explain the featureless emission at 1.0–3.0 keV. The origin of this power-law component is unclear, but it may be responsible for the discrepancy between the stellar energy input and the observed interstellar energy in DEM L 192. Subject headings: ISM: bubbles — ISM: individual (DEM L 192) – Stars: winds, outflows – Supernovae: general – X-ray: ISM 1. Introduction Massive stars inject large amounts of mechanical energy into the ambient interstellar medium (ISM) via fast stellar winds and supernova explosions. The stellar energy feedback 1Astronomy Department, University of Illinois, 1002 W. Green Street, Urbana, IL 61801; rcooper1@astro.uiuc.edu, mar@astro.uiuc.edu, chu@astro.uiuc.edu c-chen@astro.uiuc.edu, car- olan@astro.uiuc.edu 2Now at Department of Astronomy, Harvard University, rcooper@cfa.harvard.edu 3Now at Instituto de Astrof́ısica de Andalućıa (CSIC), Spain. – 2 – alters the physical conditions of the ISM and consequently may regulate the formation of future-generation stars. To gain a better understanding of the physical structure and evolu- tion of the ISM in a galaxy, it is necessary to investigate empirically the conversion efficiency of stellar mechanical energies into interstellar kinetic and thermal energies. Massive stars usually form in groups such as OB associations, and their fast stellar winds and supernova ejecta collectively sweep up the ambient ISM to form superbubbles (Mac Low & McCray 1988). Superbubbles provide excellent laboratories to study the effects of stellar energy feedback on the ISM because the kinetic and thermal energies in a superbubble are well-defined and can be measured with little ambiguity. Superbubbles in the Large Magellanic Cloud (LMC) at 50 kpc are particularly well suited for such studies, as their stellar content and ISM can be observed in great detail with little foreground extinction. We have chosen the superbubble DEM L 192 (Davies, Elliott, & Meaburn 1976) in the LMC for a detailed study of the interaction between its stars and the ambient ISM. DEM L 192, also known as N 51D (Henize 1956), hosts two OB associations, LH 51 and LH 54 (Lucke & Hodge 1970). The stellar content of these two OB associations and the dy- namics of the superbubble have been analyzed in detail by Oey (1996a) and Oey & Smedley (1998). They conclude that LH 51 and LH 54 are both young, ∼ 3 × 106 yr old. The diffuse X-ray emission from the hot interior of DEM L 192 was detected by Einstein and ROSAT observations (Chu & Mac Low 1990; Dunne, Points, & Chu 2001), but high-quality spectra of the X-ray emission have been provided only by XMM-Newton observations. The emission line features in the X-ray spectra suggest an overabundance in oxygen and neon which is used to argue for the occurrence of supernova explosions (Bomans et al. 2002). This paper reports our investigation of the energetics of the superbubble DEM L 192. In §2 we analyze archival XMM-Newton observations of DEM L 192 to determine the physical properties of the hot gas in the superbubble interior. In §3 we determine the thermal energy of the hot superbubble interior and the kinetic energies of the ionized and neutral gas in the superbubble shell. In §4 we compute the stellar wind energy and supernova explosion energy expected from the known stellar content of LH 51 and LH 54. Finally, a discussion of the overall energy budget of the DEM L 192 superbubble and our conclusions are given in §5. 2. XMM-Newton Observations of Hot Gas in DEM L 192 DEM L 192 was observed with the XMM-Newton Observatory on 2001 October 31 and November 1 using the EPIC/MOS1, EPIC/MOS2, and EPIC/pn CCD cameras (Obs. ID: 7194). The two EPIC/MOS cameras were operated in the Prime Full-Window Mode for a – 3 – useful exposure time of 31.6 ks, and the EPIC/pn camera was operated in the Extended Prime Full-Window Mode for a useful exposure time of 24.6 ks. The thin filter was used for all observations. We retrieved the XMM-Newton pipeline products from the XMM-Newton Science Archive (XSA)4 and further processed the data using the XMM-Newton Science Analysis Software (SAS ver. 5.3.3) and the calibration files from the Calibration Access Layer available on 2002 December 13. The event files were screened to eliminate bad events (e.g., events due to charged particles) so that only events with CCD patterns 0-12 (similar to ASCA grades 0-4) were selected for the EPIC/MOS observations, and events with CCD pattern 0 (single pixel events) were selected for the EPIC/pn observation. We also screened the event files to remove periods of high background. To assess the background rate, we binned the counts over 50 s time intervals for each instrument in the 10-12 keV energy range which is domi- nated by the background. Only for a short period of time did the high background, with count rates ≥0.3 counts s−1 for the EPIC/MOS or ≥0.6 counts s−1 for the EPIC/pn, affect the observations; events from this high-background period were discarded. The resulting net exposure times are 30.6 ks and 23.2 ks for the EPIC/MOS and EPIC/pn observations, respectively. 2.1. Spatial Analysis We extracted images from the EPIC/pn and EPIC/MOS observations in the 0.25–2.5 keV energy band, which includes most of the diffuse X-ray emission from DEM L 192. These images were then combined to obtain a higher S/N ratio image and to reduce the null exposure in the gaps between CCDs (the satellite was not dithered during the observations). The resulting raw EPIC image is presented in Fig. 1a. This image was further adaptively smoothed using Gaussian profiles with FWHM ranging from 2′′ to 15′′ and divided by a normalized exposure map to remove the vignetting of the telescope and sensitivity variations of the instruments. A grey-scale presentation of this adaptively smoothed EPIC image is shown in Fig. 1b. For comparisons, an Hα image of the same region is displayed in Fig. 1c, and the X-ray contours extracted from the smoothed EPIC image are overplotted on the Hα image in Fig. 1d. Diffuse X-ray emission from the interior of DEM L 192 is clearly detected. The brightest X-ray emission region is coincident with the OB association LH 54, peaking particularly at 4The XMM-Newton Science Archive is supported by ESA and can be accessed at http://xmm.vilspa.esa.es/xsa. – 4 – the bright Wolf-Rayet star HD 36402. The X-ray peaks coincident with bright stars may be dominated by stellar emission; therefore, emission from these peaks is excluded from our analysis of diffuse X-rays. The diffuse X-ray emission shows a rough limb-brightened morphology in the main body of the superbubble, with the brightest diffuse X-ray emission projected interior to the brightest Hα emission regions of the shell. Faint diffuse X-ray emis- sion is detected to the north and southwest of the main body; however, these faint X-ray extensions do not necessarily imply blowouts, as they are bounded by faint Hα filaments ex- ternally. The superbubble DEM L 192 must be in an inhomogeneous ISM, and the variations in the Hα surface brightness reflect the density variations in the ambient ISM. The rough correspondence between the diffuse X-ray surface brightness and the nebular Hα surface brightness suggests that a mixing of the dense, cooler, superbubble shell gas into the hot interior has occurred. 2.2. Spectral Analysis The X-ray spectrum of DEM L 192 is extracted from only the event file of the EPIC/pn camera because its sensitivity is greater than that of the EPIC/MOS cameras. The source aperture, marked in Fig. 1a, includes a large elliptical region that covers the main body of DEM L 192 and the north extension and a smaller square region that covers the southwest extension; these two regions together encompass all diffuse X-ray emission within the super- bubble. The background contribution was assessed using an aperture exterior to DEM L 192 with similar area to the source aperture. All point sources were excised from the source and background apertures. A circular region that includes most stars in LH 54 is also excluded from the source aperture to minimize the stellar contribution to diffuse emission. After the background subtraction, a total of ∼16,000 counts are detected in the source aperture. Some residuals of the background subtraction are noticeable at the energy of the Al Kα fluoresce line at ∼1.5 keV. The energy range 1.45-1.55 keV, thus, has been omitted in the subsequent spectral analysis. The background-subtracted EPIC/pn spectrum of DEM L 192, shown in Fig. 2, peaks at 0.5–0.6 keV, with a bright shoulder at ∼0.7 keV and a weaker peak at 0.9–1.0 keV. Above 1.0 keV, the spectrum is featureless with emission detected up to 3.5 keV. The overall spectral shape indicates thermal plasma emission, with spectral features corresponding to the He-like triplet of O vii at ∼0.57 keV, the H-like O viii line at ∼0.65 keV, and the He-like triplet of Ne ix at ∼0.92 keV. Therefore, we have adopted the MEKAL optically- thin plasma emission model (Kaastra & Mewe 1993; Liedahl, Osterheld, & Goldstein 1995) with 1/3 Z⊙ abundances as appropriate for the LMC and the absorption cross-sections from – 5 – Balucińska-Church & McCammon (1992). The spectral fits are carried out using XSPEC ver. 11.2.0 by folding the model spectrum through the EPIC/pn response matrices and comparing the modeled spectrum to the observed spectrum in the 0.3–4.5 keV energy range. The χ2 statistics are used to determine the best-fit model. The best-fit MEKAL model with 1/3 Z⊙ abundances is very unsatisfactory (reduced χ 2 = 7.2), failing to fit the spectrum at 1–3 keV and showing large residuals at the Ne ix lines at ∼0.92 keV. Adding a second thermal component at a higher temperature can improve the spectral fits significantly (reduced χ2 = 2.2), but the temperature of the second component is pegged at the upper limit allowed by XSPEC, i.e., kT = 80 keV. A thermal component at such a high temperature is not expected in the ISM; furthermore, it is dominated by bremsstrahlung emission of which the spectral shape is featureless and similar to that of a power law. Therefore, we have added a power-law component, instead of a second thermal component, and the reduced χ2 is lowered to 2.2 for the best-fit model. The large residuals at the Ne ix lines remain, and they can be minimized only if the Ne abundance is allowed to vary. By increasing the Ne abundance by a factor of 3 (i.e., solar Ne abundance and 1/3 solar for the other elements), we are able to obtain a best-fit model with a reduced χ2 of 1.4, as shown in Fig. 2. This confirms the enhanced Ne abundance suggested by Bomans et al. (2002); however, we do not find compelling evidence for an enhanced O abundance, which has also been suggested by Bomans et al. (2002). In our best-fit model for the diffuse X-rays from DEM L 192, the thermal component dominates the emission below 1 keV with a temperature kT of 0.21±0.01 keV, an absorption column density of (3.4±1.7)×1020 H-atom cm−2, and a normalization factor5 A = (1.6±0.4)× 10−3 cm−5. The power law component dominates the emission from 1 to 3 keV and has a photon index Γ of 1.3 ± 0.2. The observed flux of diffuse X-ray emission from DEM L 192 is 1.1 × 10−12 ergs cm−2 s−1 and the X-ray luminosity is ∼ 4 × 1035 ergs s−1 in the 0.3–3.0 keV band. The thermal emission accounts for ∼70% of the observed luminosity (but ∼85% of the total number of detected photons). 3. Energy Content of DEM L 192 For a superbubble like DEM L 192, we expect the stellar mechanical energy injected into the ISM to be stored in the thermal energy of its hot interior and the kinetic energy of the dense, swept-up shell. As the ionization front might be trapped in the swept-up shell, both 5The normalization factor A is defined to be A=1 × 10−14 ∫ N 2 e dV /4πd2, where d is the distance, Ne is the electron number density, and V is the volume. – 6 – ionized and neutral gas in the superbubble shell must be considered. Below we examine separately the thermal energy of the hot interior and the kinetic energies of the ionized shell gas and associated H i gas. 3.1. Thermal Energy of the Superbubble Interior The thermal energy of the hot gas in a superbubble is Eth = (3/2)kT Nǫ1V , where T is the plasma temperature, N is the total particle number density, ǫ1 is the volume filling factor of the hot gas, and V is the volume interior to the superbubble. Assuming a canonical helium to hydrogen number density ratio of NHe/NH = 0.1, the total particle number density, approximated by the sum of electron density, hydrogen density, and helium density, is N ∼ Ne + NH + NHe ∼ 1.92Ne. For a given volume filling factor ǫ1, the rms Ne can be derived from the normalization factor A of the best-fit model: Ne = 2 × 10 7d(πA/ǫ1V ) 1/2. For a distance of d = 50 kpc, a normalization factor of A = (1.4 ± 0.4) × 10−3 cm−5, and a volume V = 4.4 × 1061 cm3 (approximated by a 120 × 120 × 200 pc3 ellipsoid), we find Ne = (0.031 ± 0.004) ǫ −1/2 1 cm −3. The total mass of the hot gas in the superbubble is 1.3 × 103 ǫ 1/2 1 M⊙, and the total thermal energy of the hot gas is Eth = (1.3 ± 0.3) × 10 51 ǫ 1/2 1 ergs. The filling factor ǫ1 is most likely between 0.5 and 1.0. For ǫ1 = 0.5, Ne = 0.044±0.004 cm−3, the hot gas mass is 920 M⊙, and Eth = (9 ± 2) × 10 50 ergs. The cooling timescale of the hot gas tcool is ∼ (3/2)NkT /Λ(T ), where Λ(T ) is the cooling function for gas at temperature T in units of ergs cm−3 s−1. For a plasma temperature of a few × 106 K, Λ(T )/N 2H = 7 × 10 −24 ergs cm3 s−1 for 1/3 Z⊙ abundances (Dalgarno & McCray 1972). With NH ∼ 0.83Ne, N ∼ 1.92Ne, and kT = 0.21 ± 0.01 keV, the cooling timescale tcool is ∼ 200 Myr. Since the cooling timescale is much larger than the age of the OB associations LH 51 and LH 54, ∼ 3 Myr, only a negligible amount of thermal energy has been radiated away. 3.2. Kinetic Energy of the Ionized Superbubble Shell The mass of the ionized superbubble shell can be derived from the Hα luminosity, LHα. The Hα flux from DEM L 192 is fHα = (860 ± 80) × 10 −12 ergs s−1 cm−2 (Kennicutt & Hodge 1986). For a distance of d = 50 kpc, LHα = 4πd 2fHα = (2.6±0.3)×10 38 ergs s−1. In addition, for ionized gas at temperatures of ∼104 K, LHα = 3.56×10 −25NeNpV ǫ2 ergs s −1, where Ne and Np are the particle number densities of electrons and protons in units of cm −3, respectively, V is the volume of the region in units of cm3, and ǫ2 is the volume filling factor. Again, we – 7 – approximate the superbubble volume by a 120 × 120 × 200 pc3 ellipsoid, and the volume is V = 4.4×1061 cm3. Helium is most likely singly ionized, thus Ne = 1.1Np = (4.5±0.5) ǫ −1/2 2 cm−3. The total mass of the superbubble shell is M = 1.27NeV ǫ2mH, where mH is the mass of a hydrogen atom. Using the rms Ne derived from LHα, we find the superbubble shell mass to be M = (4.2 ± 0.5) × 1038 ǫ 1/2 2 g = (2.1 ± 0.2) × 10 5 ǫ 1/2 2 M⊙. The expansion velocity of the DEM L 192 superbubble has been measured in the Hα line to be vexp ∼ 35 km s −1 (Meaburn & Terrett 1980). Thus, the kinetic energy in the ionized superbubble shell is Ekin = 1 2 Mv2exp = (2.6 ± 0.3) × 1051ǫ 1/2 2 ergs. If the thickness of the ionized superbubble shell is 1/10 of the radius, then the filling factor of shell gas is ǫ2 = 0.27, the ionized shell mass is M ∼ 1 × 10 5 M⊙, and the ionized shell kinetic energy is Ekin ∼ 1 × 10 51 ergs. 3.3. Kinetic Energy of Associated H i Gas The kinetic energy of the neutral gas in the swept-up shell can be derived from obser- vations of the 21-cm line emission from the H I gas. Such observations of DEM L 192 have been extracted from H I surveys of the LMC made with the Australia Telescope Compact Array (ATCA) and the Parkes Observatory 64-m telescope, where single-dish observations from the Parkes telescope have been combined with the ATCA data to “fill-in” low spatial frequencies not observed by the interferometer. The ATCA and Parkes observations are detailed in Kim et al. (1998) and Kim et al. (2003), respectively. The kinematics of the H I gas can be determined from the position-velocity plots in Fig. 3. The H I gas shows a broad velocity profile centered at the systemic velocity of Vhel ∼ 300 km s −1 and with a width of ∼70 km s−1, corresponding to twice the expansion velocity measured in Hα emission, but no clear expanding shell structure can be identified. The H I channel maps (Dunne, Chu, & Staveley-Smith 2004) and interstellar absorption observations of stars in DEM L 192 (Wakker et al. 2004) indicate that the H I velocity structure is rather complex and some velocity components may not be physically associated with the superbubble. A total 21-cm line flux density of 470 Jy is measured from the combined radio observa- tions within the dotted polygon outlined in the H I column density map in Fig. 3. Given the LMC distance of 50 kpc, we estimate a total H I mass of 3 × 105 M⊙ from this flux density. The total kinetic energy of the H I gas can be approximated by 3/2 of the second moment of the integrated velocity profile, assuming that the motion is isotropic. This estimated kinetic energy is ∼ 3 × 1051 ergs. We can also determine the H I kinetic energy using the total H I mass and the expansion velocity of the superbubble shell; the resulting kinetic energy is 3.7 × 1051 ergs, compatible with the H I kinetic energy estimated from the second moment – 8 – of the velocity profile. Note that these H I mass and kinetic energy estimates should be considered as upper limits, as some of the H I gas may not be physically associated with the superbubble. 4. Stellar Energy Injected into DEM L 192 Stellar mechanical energy is injected into the ambient ISM via fast stellar winds and supernova explosions. Below we determine the contributions of these two types of stellar energy feedback. 4.1. Stellar Wind Energy To determine the stellar wind energy injected into DEM L 192, we consider the 50 most luminous stars in the OB associations LH 51 and LH 54 given by Oey & Smedley (1998). The spectroscopically determined spectral types of these stars are matched with the theoretical models of Schaerer & de Koter (1997) to assess the stellar mass M, mass loss rate Ṁ , and stellar wind terminal velocity v∞. Since ZLMC ∼ 1/3 Z⊙, we apply the necessary corrections to the listed theoretical values of Ṁ and v∞ for low metallicity stars as described in Schaerer & de Koter (1997). From these data, we estimate the stellar wind luminosity by Lw = (1/2)Ṁv 2 ∞ . The metallicity-corrected parameters and mechanical luminosities of the 50 stars are listed in Table 1 in the same order as given in Table 1 of Oey & Smedley (1998). We calculate the total mass loss rate to be 4.4 × 10−5 M⊙ yr −1 and the total stellar wind luminosity to be 5.5 × 1037 ergs s−1. Thus, for an age of ∼ 3 Myr (Oey & Smedley 1998), the total stellar mass loss is ∼ 100 M⊙, and the total wind energy input is about (5 ± 1) × 1051 ergs. 4.2. Supernova Energy To estimate the number of stars that have already exploded as supernovae, we use the combined mass function of LH 51 and LH 54. UBV photometry of 456 stars in LH 51 and LH 54 is available from Oey (1996a). The mass of each main sequence star is determined by comparing its location on a color-magnitude diagram (CMD) to the evolutionary tracks of stars with different masses. As the reddening is not known for individual stars, we use the extinction-free magnitude W = V − DM − (B − V )AV /E(B − V ) and reddening-free color Q = (U −B)−(B −V )E(U −B)/E(B −V ), where DM = 18.5 is the distance modulus of the – 9 – LMC, E(U − B)/E(B − V ) = 0.72, and AV /E(B − V ) = 3.1. Fig. 4 shows the reddening- free CMD of LH 51 and LH 54, where the evolutionary tracks of main sequence stars with Z = 0.4 Z⊙ provided by Lejeune & Schaerer (2001) are plotted for initial masses of 5, 7, 10, 12, 15, 20, 25, 40, and 60 M⊙. Stars in each mass bin are counted. We assume a Salpeter initial mass function for LH 51 and LH 54: f (M) = KM−2.35, where K is a constant that can be determined from the star counts. The number of stars with masses between Mmin and Mmax is N = ∫ Mmax Mmin f (M) dM. For a Mmin of 5–7 M⊙ and a Mmax of 20–25 M⊙, we find K = 3040 ± 460. This constant is insensitive to the choice of Mmax but rather sensitive to the choice of Mmin; the error bar reflects the uncertainty caused by different Mmin. The number of stars with masses greater than 25 M⊙ is ∫ 100M⊙ 25M⊙ f (M) dM = 25 ± 4. From Table 1, we find 12 stars with masses greater than 25 M⊙, so the number of stars that have exploded in supernovae is (25 ± 4) − 12 = 13 ± 4. We have chosen to count main-sequence stars in the 5–25 M⊙ range because the evolutionary tracks of such stars are well constrained by the empirically derived relationship between stellar masses and spectral types and are not subject to the uncertainties for the most massive stars (e.g., Aloisi, Tosi, & Greggio 1999). We have compared the evolutionary tracks of stars in the 5–25 M⊙ range between the Geneva models (Schaerer, Meynet, Maeder, & Schaller 1993) and the Padua models (Fagotto et al. 1994), and find that they are indeed similar. Assuming that each supernova releases about 1051 ergs of explosion energy, the total supernova energy input is approximately (1.3 ± 0.4) × 1052 ergs. 5. Energy Budget of the Superbubble DEM L 192 The energies stored in the superbubble DEM L 192 and the stellar energy input are summarized in Table 2. The values and error bars of the thermal and kinetic energies take into account the most probable range of filling factors. The stellar energy input includes contributions from fast stellar winds and supernovae. The superbubble energy includes the thermal energy of its hot interior and the kinetic energy of the superbubble shell.6 It is immediately clear that the expected amount of stellar energy injected into the ISM is ∼ 3 times as large as that observed in the superbubble. Previously, observed dynamics of bubbles blown by single stars or superbubbles have been compared with pressure-driven bubble models by Weaver et al. (1977) or variations of 6The thermal energy of the ionized superbubble shell is not included in the energy budget discussion because the superbubble shell is photoionized and the thermal energy is provided by the UV radiation of the OB associations. – 10 – such models, and it has been commonly concluded that the observed stellar wind strength is too high for the observed bubble dynamics or that the bubble/superbubble shell expands too slowly for the given stellar wind strengths (e.g., Garćıa-Segura, Langer, & Mac Low 1996; Oey 1996b; Nazé et al. 2001). These results are equivalent to what we have found in DEM L 192, i.e., the stellar mechanical energy injected into the ISM exceeds the energies storied in the superbubble. More specifically, according to the Weaver et al. (1977) model, the 35 km s−1 expansion velocity and 60 pc radius of DEM L 192 indicate a dynamic age of 1 Myr, which is shorter than the age of the OB associations LH 51 and LH 54. The mass of the ionized superbubble shell consists of the material swept up from the superbubble interior, thus the volume and mass of the H II gas given in §3.2 imply an ambient density of ∼ 3 H-atom cm−3. The wind mechanical luminosity required by Weaver et al. (1977) for DEM L 192 is 1.5 × 1038 ergs s−1, and the total energy required during the 1 Myr dynamic time is ∼ 5 × 1051 ergs, which is much smaller than the total stellar energy expected from LH 51 and LH 54. Evidently there is an energy crisis: the observed interstellar energies cannot account for all the stellar energy injected into the ISM. Several mechanisms may drain energies from a superbubble. The most obvious mechanism is a superbubble blowout or breakout, in which the interior hot gas spews out, directly reducing the thermal energy and pressure of the superbubble interior. However, no blowouts are obvious in DEM L 192. As shown in Fig. 1, the extended diffuse X-ray emission from DEM L 192 is completely bounded by outer, curved Hα filaments, in sharp contrast to the blowout of the superbubble DEM L 152, also known as N44, in which the diffuse X-ray emission extends beyond open, streamer-like Hα filaments (Chu et al. 1993). Another mechanism that may drain the internal energy of a superbubble is the evapo- ration of dense cooler gas into the hot interior. In addition to the dense ionized superbubble shell, DEM L 192 contains dense cloudlets with ionized surface layers, as revealed in high- resolution HST WFPC2 Hα images (Chen et al. 2000). The main energy loss would occur at the interface layer between the 106 K hot gas and the 104 K cool gas because the cooling rate is the highest for temperatures of a few ×105 K. We may estimate the energy loss due to cooling from interface layers. Assuming that the interface layers are in pressure equilibrium with the hot interior gas, their electron number density would be 0.3–0.4 cm−3. The total volume of the interface layers is nevertheless unknown because no interface layer has ever been spatially resolved observationally and even the existence of heat conduction and mass evaporation may be questionable (Shelton 2003). The spatial extent of an interface layer can be estimated from the relative locations of diffuse X-ray emission and the optical shell of a bubble or superbubble. The highest-quality X-ray observations of wind-blown bubbles are the Chandra observation of NGC 6888 and the XMM-Newton observations of S308 (Gruendl, – 11 – Guerrero, & Chu 2004; Chu et al. 2003). While NGC 6888 shows no appreciable separation between the outer edge of the diffuse X-ray emission and the outer edge of the optical shell, S308 shows a gap of 90–200′′ (or 0.5 to 1.7 pc), which contains the dense ionized gas shell and the interface layer. The interface layer must be less than 0.5 pc in thickness. Adopt- ing this upper limit of the interface layer thickness for DEM L 192, the volume occupied by the interface layer is no more than 2.5% of the total volume. Using a cooling function Λ(105K)/N 2H ∼ 3 × 10 −22 ergs cm3 s−1 appropriate for the LMC metallicity (Dalgarno & McCray 1972), the current cooling rate would be 4 × 1037 ergs s−1. If the superbubble has been evolving with the radius ∝ tη, where t is the age and η is a constant (e.g., η = 0.6 for the Weaver et al. model), then the volume has been growing ∝ t3η, and the total energy radiated away in 3 Myr by the interface layer would be 4 × 1051(3η + 1)−1 ergs. If η = 0.6, the total radiative loss is ∼ 1.5 × 1051 ergs. This generous upper limit for radiative loss is non-negligible but not sufficiently significant in the energy budget. Therefore, radiative loss in the interface layer between dense cold gas and the hot superbubble interior cannot be the cause of the energy discrepancy. Where did the stellar energy go if it is not converted into the thermal and kinetic energies of the ISM? The interaction between a fast moving plasma, being it tenuous stellar winds or dense supernova ejecta, and the ambient ISM is a complex magnetohydrodynamic problem (Spitzer 1978). It is perhaps an oversimplification to assume that supernova ejecta and fast stellar winds are thermalized in the hot interior of a superbubble. In fact, the XMM-Newton X-ray observations of DEM L 192 reveal a possible mechanism that may play an important role in the energy budget. Our analysis of the X-ray spectrum of the diffuse emission from DEM L 192 indicates that not all emission originates from a thermal plasma; a power-law component is needed to explain the emission in the 1.0–3.0 keV range. This power-law component cannot originate from unresolved point sources, as the smoothed X-ray image in the 1.1–3.5 keV band displayed in Fig. 5 shows that this hard X-ray emission is diffuse. It is possible that this power-law component originates from the interactions of supernova ejecta and fast stellar winds with the hot interior of the superbubble, but the exact origin is unclear. It is unlikely that this power-law component represents a synchrotron X-ray emission similar to those seen in young supernova remnants such as SN 1006 (Koyama et al. 1995) and Cas A (Vink & Laming 2003) and the superbubble 30 Dor C (Bamba et al. 2004) for two reasons. First, the X-ray photon index of DEM L 192, Γ = 1.3 ± 0.2, is lower than those observed in the other objects, Γ ∼ 2. Second, the synchrotron X-ray emissions from young supernova remnants and 30 Dor C show limb-brightened morphologies because the relativistic particles are accelerated in the shocks, but the power-law X-ray emission from DEM L 192 does not show enhancement along the edge (see Fig. 5). Observations of more superbubbles are needed to determine whether a power-law component is commonly needed to model the spectra of – 12 – their diffuse X-ray emission. Future theoretical models of superbubbles need to solve the plasma interaction problem realistically instead of making simplifying assumptions about shocks and converting mechanical energy into only kinetic and thermal energies, so that true progress in the understanding of bubbles and superbubbles can be made. YHC acknowledges the support of NASA grant NAG 5-13076. We thank the referee for prompt reading and providing useful suggestions to improve this paper. We also thank Yaël Nazé for her critical reading of the paper. – 13 – REFERENCES Aloisi, A., Tosi, M., & Greggio, L. 1999, AJ, 118, 302 Balucińska-Church, M., & McCammon, D. 1992, ApJ, 400, 699 Bamba, A., Ueno, M., Nakajima, H., Koyama, K. 2004, ApJ, in press (astro-ph/0310713) Bomans, D. J., Rossa, J., Weis, K., & Dennerl, K. 2002, in IAU Symposium, No. 212, ed. K. A. van der Hucht, A. Herrero, & C. Esteban (San Francisco: ASP), 637 Chen, C.-H. R., Chu, Y.-H., Gruendl, R. A., & Points, S. D. 2000, AJ, 119, 1317 Chu, Y.-H., Guerrero, M. A., Gruendl, R. A., Garćıa-Segura, G., & Wendker, H. J. 2003, ApJ, 599, 1189 Chu, Y.-H., & Mac Low, M.-M. 1990, ApJ, 365, 510 Chu, Y.-H., Mac Low, M. M., Garćıa-Segura, G., Wakker, B., & Kennicutt, R. C. 1993, ApJ, 414, 213 Dalgarno, A., & McCray, R. A. 1972, ARA&A, 10, 375 Davies, R. D., Elliott, K. H., & Meaburn, J. 1976, MmRAS, 81, 89 Dunne, B. C., Chu, Y.-H., & Staveley-Smith, L. 2004, in preparation Dunne, B. C., Points, S. D., & Chu, Y. 2001, ApJS, 136, 119 Fagotto, F., Bressan, A., Bertelli, G., & Chiosi, C. 1994, A&AS, 105, 39 Garćıa-Segura, G., Langer, N., & Mac Low, M.-M. 1996, A&A, 316, 133 Gruendl, R. A., Guerrero, M. A., & Chu, Y.-H. 2004, in preparation Henize, K. G. 1956, ApJS, 2, 315 Kaastra, J. S., & Mewe, R. 1993, Legacy, 3, 16, HEASARC, NASA Kennicutt, R. C., Jr., & Hodge, P. W. 1986, ApJ, 306, 130 Kim, S., Staveley-Smith, L., Dopita, M.A., Freeman, K.C., Sault, R.J., Kesteven, M.J., & McConnell, D. 1998, ApJ, 503, 674 Kim, S., et al. 2003, ApJS, 148, 473 – 14 – Koyama, K., Petre, R., Gotthelf, E. V., Hwang, U., Matsuura, M., Ozaki, M., & Holt, S. S. 1995, Nature, 378, 255 Lejeune, T., & Schaerer, D. 2001, A&A, 366, 538 Liedahl, D. A., Osterheld, A. L., & Goldstein, W. H. 1995, ApJ, 438, L115 Lucke, P. B., & Hodge, P. W. 1970, AJ, 75, 171 Mac Low, M., & McCray, R. 1988, ApJ, 324, 776 Meaburn, J., & Terrett, D. L. 1980, A&A, 89, 126 Nazé, Y., Chu, Y.-H., Points, S. D., Danforth, C. W., Rosado, M., & Chen, C.-H. R. 2001, AJ, 122, 921 Oey, M. S. 1996a, ApJS, 104, 71 Oey, M. S. 1996b, ApJ, 467, 666 Oey, M. S., & Smedley, S. A. 1998, AJ, 116, 1263 Salpeter, E. E. 1955, ApJ, 121, 161 Schaerer, D., & de Koter, A. 1997, A&A, 322, 598 Schaerer, D., Meynet, G., Maeder, A., & Schaller, G. 1993, A&AS, 98, 523 Shelton, R. L. 2003, ApJ, 589, 261 Spitzer, L., Jr. 1978, Physical Processes in the Interstellar Medium, (Wiley-Interscience) Vink, J. & Laming, J. M. 2003, ApJ, 584, 758 Wakker, B. P., et al. 2004, in preparation Weaver, R., McCray, R., Castor, J., Shapiro, P., & Moore, R. 1977, ApJ, 218, 377 This preprint was prepared with the AAS LATEX macros v5.0. – 15 – Fig. 1.— (a) XMM-Newton EPIC image of DEM L 192. The spectrum of the diffuse X-ray emission is extracted from the source aperture consisting of elliptical and square regions marked. The point sources, marked by small circles, and stellar emission from the OB association LH 54, marked by a larger circle, are excluded from the source aperture. (b) Adaptively smoothed and vignetting-corrected X-ray image. The contour levels are at 9, 15, 22.5, and 30 σ above the background. (c) Hα image of DEM L 192. (d) Hα image overplotted with X-ray contours. – 16 – Fig. 2.— XMM-Newton EPIC/pn spectrum of the diffuse X-ray emission from DEM L 192. The best-fit model is overplotted in the solid curve. The thermal plasma component and the power-law component are individually plotted in dashed curves. The bottom panel plots the residuals of the fit. – 17 – Fig. 3.— Hα image (left) and H I 21-cm line map (right) of DEM L 192. Both images are overlaid with H I 21-cm line emission contours at levels of 50%, 65%, 77%, 85%, and 95% of the peak. The dotted polygon plotted over the H I map represents the extent of the superbubble and the region over which the H I mass is extracted. We have produced position- velocity plots for two 3′-wide slices through the superbubble along the right ascension and declination axes. These slices are shown above and to the right, respectively, of the H I image accompanied with greyscale wedges in arbitrary units. The velocities are heliocentric. The area covered by each slice is indicated by the dashed lines on the Hα image. – 18 – Fig. 4.— Reddening-free color-magnitude diagram of stars in LH 51 and LH 54. The photo- metric data of these 456 stars are from Oey (1996a). See the text for the definition of the reddening-free magnitude W and color Q. The evolutionary tracks of stars are from Lejeune & Schaerer (2001). – 19 – Fig. 5.— XMM-Newton pn image of DEM L 192 in the 1.1–3.5 keV energy band. The image has been adaptively smoothed with Gaussians of FWHM of 2′′ to 15′′. The emission in the 1.1–3.5 keV range is dominated by the power-law component. The contours are at 3σ and 6σ above the background. It is evident that the diffuse emission in this energy range is widespread. The source regions are marked for comparison with Fig. 1a. – 20 – Table 1. Stellar Data M log Ṁ v∞ log Lw Star ID Spectral Type (M⊙) (M⊙ yr −1) (km s−1) (ergs s−1) L54SA-1b O8 Iaf 51 -5.170 1699 36.79 L54SA-2 O8 III((f)) 55 -5.491 2064 36.64 L54S-2 O9 Ib(f) 51 -5.170 1699 36.79 L51N-1 O8.5 III((f)) 55 -5.491 2064 36.64 L54S-1a O7: III(f) 55 -5.491 2064 36.64 L54S-4 O4 III(f*) 96 -5.105 2445 37.17 L54S-5 O8 III((f)) 55 -5.491 2064 36.64 L54S-3a O6.5: V 32 -6.527 2422 35.54 L54S-3b B2: II: 19 -6.623 1561 35.26 L54S-1b B0.5: III: 19 -6.623 1561 35.26 L54N-2 O6.5 V 32 -6.527 2422 35.54 L54N-3 B1 IIIe 19 -6.623 1561 35.26 L54N-4 O9 V 20 -7.096 2205 35.09 L54S-8 B0.5 Ib 34 -5.529 1438 36.29 L51N-3 O9 V 20 -7.096 2205 35.09 L54S-9 B0 III 19 -6.623 1561 35.26 L54N-6 O9 V 20 -7.096 2205 35.09 L54S-10 early B:: +neb 19 -6.873 1909 35.19 L54N-7 B0.5:: V +neb 19 -6.873 1909 35.19 L51S-2 O6 V((f))e 39 -6.409 2332 35.83 L54SA-4 B0.5 III 19 -6.623 1561 35.26 L54S-12 B0 V # 19 -6.873 1909 35.19 L54S-13 B0.5 Ve 19 -6.873 1909 35.19 L54S-14 O9.5 V 20 -7.096 2205 35.09 L54S-15 O8 Ve 24 -6.835 2244 35.37 L54S-16 O9.5 V # 20 -7.096 2205 35.09 L51S-3 O8 V 24 -6.835 2244 35.37 L51S-4 B0.5 V 19 -6.873 1909 35.19 L54S-18 O9 V 20 -7.096 2205 35.09 L54S-19 B0.5: V # 19 -6.873 1909 35.19 – 21 – Table 1—Continued M log Ṁ v∞ log Lw Star ID Spectral Type (M⊙) (M⊙ yr −1) (km s−1) (ergs s−1) L51S-5 B0.5 V 19 -6.873 1909 35.19 L51N-9 B1: V +neb 19 -6.873 1909 35.19 L51N-10 B0 V 19 -6.873 1909 35.19 L54S-20 O9.5 V 20 -7.096 2205 35.09 L54SA-5 B1:: V +neb 19 -6.873 1909 35.19 L51S-7 O9.5 V 20 -7.096 2205 35.09 L54SA-1a WC 5 · · · · · · · · · · · · L51N-12 B0.5 V +neb 19 -6.873 1909 35.19 L54S-22 O9.5-B0 V 19 -6.873 1909 35.19 L54S-24 O9.5 V +neb 20 -7.096 2205 35.09 L54N-14 O9.5 V 20 -7.096 2205 35.09 L54S-25 B1.5 III +neb 19 -6.873 1909 35.19 L54N-16 O9.5 V 20 -7.096 2205 35.09 L54S-26 B1:: V +neb 19 -6.873 1909 35.19 L54N-18 O6: Ve 39 -6.409 2332 35.83 L51S-9 B0:: V 19 -6.873 1909 35.19 L54SA-7 B1.5 V 19 -6.873 1909 35.19 L54N-19 O9.5 V 20 -7.096 2205 35.09 L54N-21 B2.5 V 19 -6.873 1909 35.19 L54S-36 B1.5:: Ve 19 -6.873 1909 35.19 L54N-25 B2:: V +neb 19 -6.873 1909 35.19 L54N-27 O9.5-early Be 19 -6.873 1909 35.19 L51S-14 B1:: V 19 -6.873 1909 35.19 L54N-33 B1:: V 19 -6.873 1909 35.19 Note. — # indicates spectral binary candidate. See Table 1 of Oey & Smedley (1998) – 22 – Table 2. Energy Budget of the Superbubble DEM L 192 Energies Amount (×1051 ergs) Eth of the hot gas in superbubble interior 1.1 ± 0.5 Ekin of the ionized (H II) superbubble shell 1.5 ± 0.5 Ekin of the neutral (H I) superbubble shell 3.2 ± 0.5 Total energy observed in the superbubble 6 ± 2 Stellar wind energy input in 3 Myr 5 ± 1 Supernova energy input 13 ± 4 Total stellar energy input 18 ± 5