The first X-ray survey of Galactic Luminous Blue Variables

Aims: The X-ray emission of massive stars has been studied when these objects are in their main-sequence phase, as well as in their Wolf-Rayet phase. However, the X-ray properties of the transitional Luminous Blue Variable (LBV) phase remain unknown. Methods: Using a dedicated but limited XMM survey as well as archival XMM and Chandra observations, we performed the first X-ray survey of LBVs: about half of the known LBVs or candidate LBVs are studied. Results: Apart from the well known X-ray sources eta Car and Cyg OB2 #12, four additional LBVs are detected in this survey, though some doubt remains on the association with the X-ray source for two of these. For the other LBVs, upper limits on the flux were derived, down to $\log[L_{\rm X}/L_{\rm BOL}]-9.4$ for PCyg. This variety in the strength of the X-ray emission is discussed, with particular emphasis on the potential influence of binarity.

💡 Research Summary

The paper presents the first systematic X‑ray survey of Galactic Luminous Blue Variables (LBVs), a transitional evolutionary phase of massive stars that bridges the main‑sequence O‑type stage and the Wolf‑Rayet (WR) stage. While X‑ray emission from O‑type and WR stars has been extensively studied, the X‑ray properties of LBVs have remained largely unexplored. To fill this gap, the authors compiled a comprehensive list of LBVs and candidate LBVs (cLBVs) from the catalogs of Clark et al. (2005) and van Genderen (2001), adding recent candidates identified in Spitzer‑based studies. The final sample comprises 67 objects (12 confirmed LBVs and 55 candidates). Approximately half of this list (31 objects) have been observed with the two most sensitive modern X‑ray observatories, XMM‑Newton and Chandra, either through a dedicated XMM‑Newton program (ObsID 0600030) or via archival data. The selection criteria required the target to lie within 15′ of the XMM‑Newton field centre or 10′ of the Chandra field centre and to have a net exposure longer than 5 ks.

Data reduction followed standard pipelines: XMM‑Newton EPIC data were processed with SAS v10.0, filtered for good patterns and background flares, and images were created in the 0.5–8 keV band. Source detection employed the edetectchain algorithm with a likelihood threshold corresponding to a 4σ detection (log L ≥ 10). Sensitivity maps were generated (log L = 2.3) and converted to flux limits using WebPIMMS, assuming an optically thin thermal plasma with kT = 0.6 keV and an absorption column derived from the colour excess via N_H = 5.8 × 10²¹ E(B–V) cm⁻². The authors note that varying the plasma temperature between 0.3 and 1.0 keV or the column density by an order of magnitude changes the predicted count rates by only ~10–20 %, well within the overall systematic uncertainties.

Chandra data were re‑processed with CIAO 4.2 and CALDB 4.3. Source regions were circular with a 5″ radius, and background was taken from annuli (5″–15″) or nearby circular regions when nearby companions would contaminate the background. Special care was taken for crowded fields near the Galactic centre, where only Chandra’s superior angular resolution could separate sources.

The survey confirms the two well‑known X‑ray bright LBVs, η Carinae and Cyg OB2 #12, both exhibiting X‑ray luminosities of 10³⁴–10³⁵ erg s⁻¹. In addition, four new LBVs are detected in X‑rays. Two of these have secure positional coincidences; the other two have ambiguous associations, requiring higher‑resolution follow‑up. For the remaining 27 LBVs/cLBVs, only upper limits could be derived. The most stringent limit is for P Cyg, with log(L_X/L_BOL) = −9.4, considerably below the typical value for O‑type stars (≈ −7). Upper limits for the other non‑detections range between −7.5 and −9.5, depending on distance, extinction, and exposure depth.

A central theme of the discussion is the role of binarity. The two X‑ray bright objects are known massive binaries where colliding stellar winds generate hot plasma (kT ≈ several keV) and thus strong, hard X‑ray emission. The newly detected sources may also be binaries, but the evidence is not yet conclusive. In contrast, the non‑detected LBVs either lack a close massive companion or have binary separations too large to produce efficient wind collisions. Moreover, the dense circumstellar nebulae that often surround LBVs can absorb soft X‑rays, further reducing detectability. The authors argue that the observed diversity in X‑ray luminosities can be largely explained by the presence or absence of colliding‑wind binaries, combined with varying amounts of circumstellar absorption.

The paper concludes that LBVs are generally X‑ray faint unless they are part of a massive binary system. This finding adds an important piece to the massive‑star evolutionary puzzle: the X‑ray output can serve as an indirect tracer of binarity during the LBV phase, which in turn influences mass‑loss histories and the transition to the WR stage. The authors recommend deeper, high‑resolution X‑ray observations (e.g., Chandra ACIS‑S/HETG) and complementary spectroscopic monitoring to confirm binarity and to quantify wind‑collision parameters. Such multi‑wavelength campaigns will refine models of mass loss, wind interaction, and the ultimate fate of the most massive stars.