A large, long-lived, slowly-expanding superbubble across the Perseus arm Authors: Bingqiu Chen, Guangxing Li, Haibo Yuan, Maosheng Xiang, Jixuan Zhou, Pinjian Chen, Martin Krause, Ashley Coombs Published: 2025 ABSTRACT AbstractStellar feedback is a crucial mechanism in galactic evolution, as demonstrated by the widespread bubbles observed with JWST. In this study, we combine data from Gaia and LAMOST to obtain a sample of young O-B2 stars with full three-dimensional velocity information. Focusing on the largest known superbubble in the Milky Way, we identify groups of O-B2 stars at its periphery, exhibiting a transverse velocity of 25.8 km s−1 and an expansion velocity of 6.2 km s−1. Using these velocities, we calculate a crossing time tcross ≈ 20 Myr and an expansion timescale texpansion ≈ 80 Myr. We estimate a survival timescale tsurvival ≈ 250 Myr and a supernova interval tSN≈0.1 Myr. Together with the Galactic shear timescale tshear ≈ 30 Myr, these values satisfy tSN 19,000 K. Source data are provided as a Source Data file. A significant portion of the sample consisted of late-type giants. The lack of extinction corrections led to some overlap between early B-type main-sequence stars and redder giants. To further refine the selection, we calculated spectral line indices from the LAMOST spectra for these stars. We computed indices such as CaII K, Hδ, CN, Ca4227, G4300, Hγ, Fe4383, He4388, Fe4531, Hβ, Ca4455, He4471, Fe4531, He4542, Fe4668, Hβ, Fe5015, Mg1, Mg2, Mg b, Fe5270, Fe5335, Fe5406, and Fe570954,55. To derive spectral line indices, we first identify the line of interest and select two continuum regions on either side of it. A linear fit is then applied to the flux values in these regions to establish the pseudo-continuum level. Finally, the line index (i.e., equivalent width) is computed by integrating the flux difference between the observed spectrum and the pseudo-continuum across the line’s wavelength range. Using machine learning techniques, specifically a random forest classifier, we identified OB stars among the candidates. Cross-referencing with the Simbad database, we selected 7873 stars with identified spectral types, including 3529 OB-type stars and 4344 stars of other types. We then trained a random forest classifier with spectral indices as inputs and spectral types as outputs. The distinct spectral line features of OB stars allowed the classifier to achieve an accuracy of 97%. Applying this classifier to the remaining candidates, we identified a final sample of 19,600 OB-type stars. The primary objective of our study is to isolate very young (with ages t < 20 Myr) O- and early B-type stars (O-B2) to trace the superbubble structure of the Milky Way. For this purpose, we calculated atmospheric parameters for the 19,600 OB stars identified through spectral line indices using the HOTPAYNE pipeline56. As shown in the right panel of Fig. 4, the majority of the selected OB stars are main-sequence stars with temperatures exceeding 10,000 K. We specifically focused on stars earlier than B3 in spectral type, with effective temperatures Teff > 19,000 K. This rigorous selection process resulted in a final sample of 1608 unique O-B2 stars. Three-dimensional position and velocity To determine the radial velocities of the refined O-B2 star sample, we analyzed their spectra obtained from the LAMOST telescope. Using theoretical template spectra for OB stars provided by TLUSTY57, we performed spectral fitting in the blue wavelength range (4000–5400 Å). This allowed us to accurately derive the line-of-sight velocities for each OB star. To estimate the uncertainty in radial velocities, we used a Monte Carlo (MC) method based on the spectral uncertainties reported by LAMOST. Specifically, we added normally distributed noise to the flux values based on their respective errors, recalculated the radial velocities 100 times, and derived the standard deviation of the results as the radial velocity uncertainty. By combining the distances calculated from Gaia DR3 parallaxes, the radial velocities obtained from LAMOST, and the positional and proper motion data from Gaia DR3, we derived the three-dimensional positions (X, Y, Z) and velocities (Vx, Vy, Vz) for every star in the sample. In this coordinate system, the X-axis points towards the Galactic center, the Y-axis aligns with the direction of Galactic rotation, and the Z-axis points towards the North Galactic Pole. Similarly, the U-axis is directed towards the Galactic center, the V-axis follows the Galactic rotation, and the W-axis extends towards the North Galactic Pole. For this study, we adopted the following reference values for the solar neighborhood: the Sun’s position is (X⊙, Y⊙, Z⊙) = (−8.34, 0, 0.025) kpc29,58, with local standard of rest (LSR) velocities of (U⊙, V⊙, W⊙) = (11.69, 10.16, 7.67) km s−1, Oort constants (A, B) = (16.31, − 11.99) km s−1 kpc−1, and the Sun’s rotational velocity VLSR = 231 km s−1 59. Using these values, we calculated the peculiar velocities for each star by subtracting the LSR motion, Galactic rotation, and shear effects10. The spatial distribution of the selected O-B2 stars in the Galactic X–Y plane is shown in the left panel of Fig. 5. Due to the observational footprint of the LAMOST survey, most of these stars are situated in the outer Galactic disk. The majority of the young O-B2 stars exhibit a clustered distribution, while those at larger distances appear more dispersed, likely due to greater uncertainties in the distance estimates. Among OB associations with well-determined distances within 3 kpc, the spatial arrangement of stars clearly reflects the spiral arm structure21. In particular, we observe distinct concentrations of OB stars aligned with the Orion and Perseus arms of the Milky Way. Distribution of O-B2 stars in the XY plane. Left panel: spatial distribution of the selected O-B2 stars in the Galactic XY plane. The Sun, located at (X, Y) = (−8.34, 0) kpc, is marked by a green pentagram. The background grayscale represents the dust distribution from ref. 30. Solid blue lines delineate the best-fitting spiral arm models for the Local and Perseus Arms from ref. 21, with dashed lines showing the 0.5 kpc range around the arm models. Right panel: peculiar velocity vectors of O–B2 stars associated with the Perseus Arm and located within 100 pc of the Galactic midplane. Stars near the Giant Oval Cavity are shown as red arrows, while other stars are shown as blue arrows. Some stars with exceptionally high velocities are not shown for clarity. The black arrow at the lower right represents a reference velocity of 20 km s−1. Source data are provided as a Source Data file. O-B2 stars associated with the Perseus arm We identified a subset of O- and early B-type stars associated with the Perseus arm of the Milky Way. This selection was based on the Perseus arm model described by ref. 21. Specifically, we selected stars located within 0.5 kpc of the Perseus arm in the Galactic X–Y plane. This criterion resulted in a sample of 369 O-B2 stars. The spatial distribution of these stars, as shown in the left panel of Fig. 5, supports their association with the Perseus arm. A detailed catalog of these stars is available online at the following URL: https://paperdata.china-vo.org/diskec/obsamp/ob2perseus.fits. This catalog offers comprehensive data for each star, including celestial coordinates, the signal-to-noise ratio of the LAMOST spectra, distance estimates, parallax values, proper motions, radial velocities, and associated uncertainties. The right panel of Fig. 5 displays the peculiar velocity vectors of stars associated with the Perseus arm and located within 100 pc of the Galactic midplane, in the X–Y plane. These velocity patterns provide insight into the local kinematics of OB associations along the arm. A clear contrast emerges: stars near the Giant Oval Cavity—particularly those in the Cassiopeia and Auriga Peninsula regions (marked with red arrows)—exhibit coherent radial expansion away from the cavity center. In comparison, stars in other parts of the Perseus arm, such as Auriga, Gemini, and Outer Orion (blue arrows), display more irregular motions without any systematic radial expansion. Uncertainties of the expansion and peculiar velocities To estimate the uncertainties of the expansion and peculiar velocities of the OB stellar groups associated with the Giant Oval Cavity, we employed an MC approach based on the uncertainties of the three-dimensional positions and velocities of the individual O-B2 stars. In this method, we perturbed the measured positions and velocities by adding random noise consistent with their observational errors and recalculated the expansion and peculiar velocities across 300 realizations. The standard deviations of these resultant velocities were adopted as the uncertainty estimates: ~2 km s−1 for the expansion velocity and 3 km s−1 for the peculiar velocity. Smoothing of vertical position and velocity profiles To derive smoothed profiles of the vertical positions (Z) and vertical velocities (VZ) as functions of X′, we utilized the smoothfit Python package, which applies constrained optimization techniques for smooth curve fitting. Specifically, we performed one-dimensional smoothing spline fitting using smoothfit.fit1d. The fitting was conducted over the relevant range using 1000 evenly spaced evaluation points, linear basis functions (degree = 1) to ensure piecewise continuity, and a regularization parameter λ = 0.1 to balance between fitting accuracy and smoothness. To assess the uncertainties of the smoothed profiles, we implemented 1000 nonparametric bootstrap resamplings of the original dataset. Each resampled dataset was smoothed using the same method, and the 1σ confidence intervals were computed from the pointwise standard deviations of the ensemble of smoothed curves. Number of supernova events To estimate the number of supernovae needed to create and maintain a superbubble in the Galactic disk, we modeled the bubble as an oblate spheroid with an in-plane radius of R ≈ 500 pc and a vertical height of h ≈ 100 pc, reflecting the disk’s thin structure. The volume V of the bubble is calculated as V=43πR2h, yielding V ≈ 3.1 × 1063 cm3. Our energy budget includes three components: gravitational binding energy (Ebind), turbulent pressure energy (Eturb), and radiative cooling losses (Eloss). First, Ebind is derived from the standard relation Ebind=35G(ΣgπR2)2R. Assuming a gas surface density of Σg = 10 M⊙ pc−2, resulting in Ebind ≈ 6.3 × 1051 erg. Second, Eturb is computed as Eturb=EturbV. We have Eturb=12ρ0δv2, with number density n0 = 1 cm−3 and velocity dispersion δv = 10 km s−1, giving Eturb ≈ 2.5 × 1051 erg. Third, Eloss is calculated as Eloss=ne2Λ(T)Vtexp, using electron density ne = 0.01 cm−3 (for dilute hot gas at T ~ 106 K), cooling function Λ(T) = 10−22 erg cm3 s−1, and expansion timescale texp=80Myr, yielding Eloss ≈ 7.8 × 1052 erg. The total energy requirement is Etotal ≈ 8.7 × 1052 erg. Assuming each supernova releases ESN≈1051erg with a coupling efficiency of 20%, this leads to a required number of supernovae as NSN≈435. This supernova number aligns with the expected output from a stellar population with a mass of M* ~ 105 M⊙, based on a standard Salpeter initial mass function (IMF) that assumes an average of one supernova per 150 M⊙ of formed stars37. The Kennicutt–Schmidt (KS) relation45 estimates a star formation rate surface density of: ΣSFR≈2.5×10−4Σg1M⊙pc−21.4M⊙yr−1kpc−2. For Σg = 10 M⊙ pc−2 over a period of 80 Myr, this results in a stellar mass of M* ~ 4 × 105 M⊙, corresponding to approximately 2600 supernovae. This value is slightly higher but remains consistent with our estimate, considering uncertainties in supernova feedback efficiency and the stochastic nature of IMF sampling. Supplementary information Transparent Peer Review file Source data Source Data Acknowledgements This work is partially supported by the National Natural Science Foundation of China 12173034 (B.Q.C.) and 12322304 (B.Q.C.). We acknowledge the science research grants from the China Manned Space Project with no. CMS-CSST-2021-A09 (B.Q.C., H.B.Y.), CMS-CSST-2021-A08 (B.Q.C., H.B.Y.), and CMS-CSST-2021-B03 (B.Q.C., G.X.L.). Guoshoujing Telescope (the LAMOST) is a National Major Scientific Project built by the Chinese Academy of Sciences. Funding for the project has been provided by the National Development and Reform Commission. LAMOST is operated and managed by the National Astronomical Observatories, Chinese Academy of Sciences. This work presents results from the European Space Agency (ESA) space mission Gaia. Gaia data are being processed by the Gaia Data Processing and Analysis Consortium (DPAC). Funding for the DPAC is provided by national institutions, in particular the institutions participating in the Gaia MultiLateral Agreement (MLA). The Gaia mission website is https://www.cosmos.esa.int/gaia. The Gaia archive website is https://archives.esac.esa.int/gaia. Author contributions B.C. developed the initial concept, conducted the analysis using LAMOST and Gaia data. B.C. and G.L. led the interpretation of the observational results, aided by H.Y., M.X., J.Z., and P.C., who focused on the OB stellar atmospheric parameters and velocity determinations. M.K. and A.C. performed hydrodynamic simulations and co-developed the evolutionary model. All authors contributed to manuscript writing and revisions. Peer review Peer review information Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Data availability The Gaia DR3 is publicly accessible through the Gaia Archive. Spectroscopic data from the LAMOST DR9 are available on the official LAMOST data portal. A catalog containing detailed information on 369 O-B2 type stars, used in this study, can be accessed at the following URL: https://paperdata.china-vo.org/diskec/obsamp/ob2perseus.fits. Source data are provided with this paper. Code availability The analysis in this work was carried out using several open-source Python packages, including Astropy, smoothfit, emcee, galpy, scipy, and scikit-learn. The visualization and data exploration tool TOPCAT was also employed. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Bingqiu Chen, Email: bchen@ynu.edu.cn. Guangxing Li, Email: gxli@ynu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-025-65591-5. References Associated Data Supplementary Materials Transparent Peer Review file Source Data Data Availability Statement The Gaia DR3 is publicly accessible through the Gaia Archive. Spectroscopic data from the LAMOST DR9 are available on the official LAMOST data portal. A catalog containing detailed information on 369 O-B2 type stars, used in this study, can be accessed at the following URL: https://paperdata.china-vo.org/diskec/obsamp/ob2perseus.fits. Source data are provided with this paper. The analysis in this work was carried out using several open-source Python packages, including Astropy, smoothfit, emcee, galpy, scipy, and scikit-learn. The visualization and data exploration tool TOPCAT was also employed.