Low-metallicity massive single stars with rotation. III. Source of ionization and C-IV emission in I Zw 18

TL;DR

The paper tests whether chemically homogeneous evolution (CHE) of fast-rotating, metal-poor massive stars can explain the strong He II ionization and CIV emission observed in the metal-poor dwarf galaxy I Zw 18. Using PoWR synthetic spectra tailored to IZw18’s composition and new CHE-driven WN/WO models, the authors perform population synthesis with a 10% CHE fraction and compare the predicted He II ionizing flux and CIV 1550 emission to observations, accounting for aperture effects. They find that WO stars produced via CHE can account for the UV CIV emission and, together with the broader hot-star population, the He II ionization, with the most consistent results arising when post-main-sequence CHE stars contribute significantly. The study implies that locally, metal-poor Pop II-like stars with CHE can mimic the ionizing photons expected from the first galaxies, carrying implications for JWST surveys and the progenitors of gravitational waves, and suggests a new O→WN→WO evolutionary sequence for such low-metallicity massive stars.

Abstract

Chemically homogeneously evolving stars have been proposed to account for several exotic phenomena, including gravitational-wave emissions and gamma-ray bursts. Here we study whether these stars can explain the metal-poor dwarf galaxy I Zwicky 18. We apply our synthetic spectral models from Paper II to (i) establish a classification sequence for these hot stars, (ii) predict the photonionizing flux and the strength of emission lines from a IZw18-like stellar population, and (iii) compare our predictions to available observations of this galaxy. Adding two new models computed with PoWR, we report (i) these stars to follow a unique sequence of classes: O->WN->WO (i.e. without ever being WC). From our population synthesis with standard assumptions, we predict that (ii) the source of the UV C-IV and other emission bumps is a couple dozen WO-type Wolf-Rayet stars (not WC as previously assumed) which are the result of chem. hom. evolution, while these, combined with the rest of the O-star population, account for the He-II ionizing flux and spectral hardness. Contrasting our results against published optical and UV data and accounting for different aperture sizes and spatial regions probed by the observations, we find that (iii) our models are highly consistent with them. Since our "massive Pop II stars" might just as well exist in early star-forming regions, our findings have implications for upcoming JWST surveys; and given that our results apply for binary populations too as long as the same fraction (10%) of the systems evolves chem. homogeneously, we conclude that the stellar progenitors of gravitational waves may very well exist today in IZw18.

Figures (16)

• Figure 1: Left: Hertzsprung--Russell diagram summarizing the findings of Kubatova:2019 (their Fig. 1 & Table 4) and our Sect. \ref{['sec:newmodel']}, showing that chemically homogeneous stellar evolution at I Zw 18 composition proceeds from class O via a shorter-lived class WN to a longer-lived WO phase -- i.e. without experiencing any long-term WC phase. Evolutionary models are taken from Szecsi:2015 (main-sequence) and Szecsi:2022ok, without correcting for the wind optical depth (see the last paragraph of Sect. \ref{['sec:newmodel']}). Initial masses are labelled, showing where the tracks start their evolution, which proceeds towards the hot side of the diagram (i.e. CHE). Colours show the central helium mass fraction, and dots represent every 10$^5$ years of evolution. Dashes mark equiradial lines with 1, 10, and 100 R$_{\odot}$ from left to right. The black star symbols represent the models for which PoWR synthetic spectra were computed in Kubatova:2019 and Sect. \ref{['sec:newmodel']}. From right to left (note that the evolution progresses from red to blue, contrarily to classical massive-star evolution): main-sequence models with surface helium mass fractions of 0.28, 0.5, 0.75, and 0.98, while the fifth symbol on the very left corresponds to the core-helium-burning phase (post-main-sequence, Sect. \ref{['sec:newmodel']}). Spectral classification was performed in Kubatova:2019; the results (for four different assumptions about mass loss and clumping) were presented in their Table 4 and Appendix A. Here the labels summarize the ranges in class found there, except for the WN and WO stars which are discussed in Sect. \ref{['sec:newmodel']} (see also Fig. \ref{['fig:131evol']}). Right: HR diagram summarizing our findings from Szecsi:2015. Stellar models computed with I Zw 18's chemical composition are shown following normal evolution (NE) towards the red supergiant branch with slow rotation ($\lesssim$ 300 km s$^{-1}$). With fast rotation, they are seen following CHE towards hot surface temperatures; that is, leftward from the ZAMS. This latter evolutionary path is what is elaborated upon in the left panel. For more details on how we construct a synthetic population out of both these kinds of models, see Sect \ref{['sec:population']}.
• Figure 2: Spectral energy distributions of our PoWR synthetic spectra of chemically homogeneously evolving massive stars published in Kubatova:2019 with the corresponding black body distributions overplotted for context. The models correspond to the second (MS) and last (pMS) phases on the HR diagram in Fig. \ref{['fig:classification']}; their main physical parameters and ionizing fluxes are listed in Table \ref{['tab:Q']}. Titles indicate the initial mass of the evolutionary models with the actual mass (i.e. the mass when the synthetic spectra were computed) in parenthesis. The three main ionizing continua (Lyman: $<$ 912 $\text{\normalfont\AA}$, HeI: $<$ 504 $\text{\normalfont\AA}$, HeII: $<$ 228 $\text{\normalfont\AA}$) are marked (note the X scale showing the frequency instead of the wavelength, in order to spread out the relevant high-energy parts of the distributions). Framed boxes present the number of ionizing photons predicted by both the PoWR models (unclumped, with nominal mass-loss rates) and the black body distributions, in units of log(s$^{-1}$); the values being close to each other is an interesting coincidence. Left column: Models in the main-sequence (MS) phase with surface helium mass fraction of Y$_{\rm S}$$=$ 0.5. Right column: Models in the post-main-sequence (pMS) phase (in the case of the 131 M$_{\odot}$ model, the late-pMS phase, see Sect. \ref{['sec:newmodel']} and Fig. \ref{['fig:newmodel']}). Note the order of magnitude difference in the Y scale between the left column and the right column.
• Figure 3: Time evolution of surface properties of the evolutionary model with M$_{\rm ini}$$=$ 131 M$_{\odot}$ after the terminal-age main sequence (TAMS). Reproduced from Szecsi:2016, adding information about our three post-main-sequence PoWR models scrutinized in Sect. \ref{['sec:newmodel']} and Fig \ref{['fig:newmodel']}. Black star-symbols mark their positions on the central helium mass fraction: Y$_{\rm C}$$=$ 0.8 (early), Y$_{\rm C}$$=$ 0.5 (mid), Y$_{\rm C}$$=$ 0.1 (late). The rudimentary classification of Szecsi:2016 is surpassed by the spectral-line-based classification scheme (see also Fig. \ref{['fig:newmodel']} showing all the optical emission lines), meaning that chemically homogeneously evolving stars at I Zw 18 metallicity do not spend significant time in the WC phase, proceeding from their main-sequence O-star phase to WN and then directly to WO.
• Figure 4: PoWR model spectra of individual stars during the post-main-sequence (core-helium burning) phase of the same evolutionary model (M$_{\rm ini}$$=$ 131 M$_{\odot}$, actual mass between 93$-$106 M$_{\odot}$). The model's approximate position is shown on the HR diagram in Fig. \ref{['fig:classification']} by the last point (pMS phase) at around log(T$_{\mathrm{eff}}$/K) $\sim$ 5.14 and log(L/L$_{\odot}$) $\sim$ 6.7. Top: Optical range. Bottom: UV range. Nominal mass loss rate and no clumping was assumed, to be consistent with previous work (see Kubatova:2019, and the discussion of the caveats in Sect. \ref{['sec:caveats']}). Y$_{\rm C}$ indicates core-helium abundance and thus the evolutionary progress during the post-MS. The earliest model (red straight line, Y$_{\rm C}$$=$ 0.8, i.e. just burned about 20% of its helium in the core) shows prominent emission only in helium, while the mid (blue straight line, Y$_{\rm C}$$=$ 0.5) and late (black straight line, Y$_{\rm C}$$=$ 0.1, corresponding to the bottom right panel of Fig. \ref{['fig:BB']}) ones develop carbon and oxygen lines too. (The Cl and S lines are modelling side-effects, we do not expect them to show up in observations.) However, the distinguishing emission line Ciii $\lambda$5696 $\text{\normalfont\AA}$ which serves as the basis of classifying a stars as WC Crowther:1998 is missing in all three models. So are the nitrogen emission lines that would categorize a star as late-type WN Crowther:1995Smith:1996Crowther:2011. If observed, therefore, these stars would be identified as early-type WN (i.e. WN2) and then during most of the core-helium burning, as WO -- more precisely, WO 2 evolving to WO 1.
• Figure 5: Top panel: Number of ionizing photons in the HeII continuum in all our spectral models from Kubatova:2019. Evolutionary stages are indicated by colours (ZAMS by yellow; two main sequence models with surface helium mass fractions Y$_{\rm S}$$=$ 0.5 and 0.75 by green and red, respectively; terminal-age-main-sequence model, TAMS, by blue; and post-main-sequence model, pMS, by purple). For every evolutionary stage, Q$_{\ion{He}{II}}$ from all four spectral models of Kubatova:2019 are shown (hence the spread in the coloured stripes). Yellow circles represent the HeII ionizing photon number computed from corresponding black body distributions (see Fig. \ref{['fig:BB']}). To account for a population of massive stars, we interpolate linearly between the logarithm of the three simulated masses. Models with nominal mass-loss rate and unclumped wind are used; interpolated values are shown with the solid black lines, small triangles marking the bin sizes. Above and below our highest and lowest mass models, the values of those models are applied (down to 9 M$_{\odot}$). Bottom panel: Same as the top panel, but for the line luminosities in CIV $\lambda$1550 $\text{\normalfont\AA}$ of the pMS model. The second trio of pink rectangles with higher values correspond to clumped wind models (see also Fig. \ref{['fig:Civ']}). Other line luminosities are treated the same way (see Sect. \ref{['sec:luminosity']}).