X-ray stellar feedback in low-metallicity starbursts: Insights from the template starburst galaxy ESO 338-IG04 and its halo

TL;DR

This study analyzes deep Chandra and XMM-Newton data of the metal-poor starburst ESO 338-IG04 to characterize its X-ray source population and halo. It identifies five ULXs (with ULX1 highly variable) and a two-temperature diffuse halo, yielding $L_X$ values indicating strong X-ray activity per unit star formation. Photoionization modeling shows that ULX1 can produce a nebular He II $\lambda4686$ luminosity of about $1.9\times10^{39}$ erg s$^{-1}$, comparable to the observed total, suggesting X-ray binaries—especially a dominant ULX—significantly contribute to the hard ionizing radiation in this system. Together, the results imply that X-ray binaries play a meaningful role in the feedback, halo energetics, and He II excitation of low-metallicity starbursts, with halo properties more tightly linked to the SFR than metallicity.

Abstract

The X-ray output of low-metallicity starburst galaxies is a key component of stellar feedback, tracing the processes responsible for gas ionization and chemical enrichment. The integrated X-ray luminosity ($L_X$) from high-mass X-ray binaries in star-forming galaxies scales with both the star formation rate (SFR) and host-galaxy metallicity $Z$. Due to the inverse correlation between $L_X/\mathrm{SFR}$ and $Z$, the contribution of X-ray binaries to the ionizing photon budget is enhanced in metal-poor systems and may ionize He II in the surrounding interstellar medium, powering nebular He II $\lambda4686$ emission. The blue compact dwarf galaxy ESO 338-IG04 (ESO 338-4) provides a nearby laboratory for studying stellar feedback in a low-metallicity starburst, combining vigorous recent star formation, low metallicity ($12+\log(\mathrm{O/H})\approx7.9$), and a rich population of massive stellar clusters. We characterize the X-ray emission of ESO 338-4 and its halo using new deep Chandra and XMM-Newton observations. We analyze X-ray spectra, light curves, and images to constrain the nature of its X-ray sources. We identify five ultra-luminous X-ray sources (ULXs) and diffuse hot gas surrounding the galaxy. Two ULXs are spatially associated with stellar clusters. The total X-ray luminosity exceeds $10^{41}\,\mathrm{erg\,s^{-1}}$. The brightest source, ULX1, shows variability on day timescales and lacks a stellar-cluster counterpart. Photoionization modeling shows that X-ray sources significantly impact the ionizing photon budget; models with ULX1 as the ionizing source predict nebular He II $\lambda4686$ luminosities of $\sim10^{39}\,\mathrm{erg\,s^{-1}}$.

Figures (10)

• Figure 1: Merged intensity image of the 16 Chandra observations analyzed in this work (the log of the observations is summarized in Appendix \ref{['app:log']}). The observations have been taken with the ACIS-I array, and the image is exposure corrected. We highlight the boundaries of the count distribution of each source at a 3$\sigma$ level through white ellipses. Magenta ellipses present the source locations along their 1$\sigma$ uncertainty in position. ULX1-4 and X6 were detected by wavdetect while ULX5 was manually added as described in the text.
• Figure 2: Intensity image of the combined XMM-Newton EPIC pn and MOS1/2 exposures in the $0.2$--$12$ keV band. The image has been adaptively smoothed and background corrected following standard SAS procedures and is presented with the gist_ncar colormap. We highlight with black ellipses the Chandra 3$\sigma$ contours presented in Fig. \ref{['fig:Chandra_img']}. Since XMM-Newton does not resolve the individual sources, we depict them through a blended "central region" represented by the larger black ellipse. To highlight the different scale between the Chandra and XMM-Newton images, we note with a white box the extent of Fig. \ref{['fig:Chandra_img']}. For purposes of spectral fitting in Sect. \ref{['sec:spec_anal']} we describe the "galactic halo" by the region between the black ellipse and the magenta circle. Beyond the halo, soft, diffuse structures are visible, extending significantly beyond the extent of the galaxy. A second "extended halo" is defined by the region between the magenta and blue circles. Its properties are in agreement with the "galactic halo" and are discussed in Appendix \ref{['app:extended_halo']}. The D25 ellipse HyperLedaDistance2014 is noted with dashed black to facilitate comparison with the optical extent of the galaxy. The two bright loci in the NW and NE of the galaxy are background sources not associated with ESO 338-4.
• Figure 3: Upper panel: Light curve of ULX1 from the 16 Chandra observations obtained in 2023. The curve is constructed from the total count rate in the $0.5$--$7$ keV range. Lower panel: Hardness ratio of each Chandra observation between the $0.5$--$3$ keV and the $3$--$7$ keV range.
• Figure 4: Upper panel: XMM-Newton EPIC pn, MOS1, and MOS2 spectra of ESO 338-4 (black, red, and green curves, respectively) with 3$\sigma$ errors. The best-fit model (described in Sect. \ref{['sec:spec_anal_XMM']}) of the galaxy is shown with solid lines. The individual best-fit model components are shown with dashed lines. The model parameters are given in Table \ref{['tab:Spec_pam']}. Lower panel: residuals between the data and the best-fit model.
• Figure 5: Same as Fig. \ref{['fig:XMM_spec_center']} for the galactic halo of ESO 338-4 defined in Fig. \ref{['fig:XMM_img']}. The best-fit model is described in Sect. \ref{['sec:spec_anal_XMM']} and the model parameters are given in Table \ref{['tab:Spec_pam']}.