In recent work with high-resolution grating spectrometers (RGS) aboard XMM-Newton Pinto et al. (2016) have discovered that two bright and archetypal ultraluminous X-ray sources (ULXs) have strong relativistic winds in agreement with theoretical predictions of high accretion rates. It has been proposed that such winds can become optically thick enough to block and reprocess the disk X-ray photons almost entirely, making the source appear as a soft thermal emitter or ultraluminous supersoft Xray source (ULS). To test this hypothesis we have studied a ULX where the wind is strong enough to cause significant absorption of the hard X-ray continuum: NGC 55 ULX. The RGS spectrum of NGC 55 ULX shows a wealth of emission and absorption lines blueshifted by significant fractions of the light speed (0.01 − 0.20)c indicating the presence of a powerful wind. The wind has a complex dynamical structure with the ionization state increasing with the outflow velocity, which may indicate launching from different regions of the accretion disk. The comparison with other ULXs such as NGC 1313 X-1 and NGC 5408 X-1 suggests that NGC 55 ULX is being observed at higher inclination. The wind partly absorbs the source flux above 1 keV, generating a spectral drop similar to that observed in ULSs. The softening of the spectrum at lower (∼ Eddington) luminosities and the detection of a soft lag agree with the scenario of wind clumps crossing the line of sight, partly absorbing and reprocessing the hard X-rays from the innermost region.

Ultraluminous supersoft sources (ULSs) are defined by a thermal spectrum with colour temperatures ∼0.1 keV, bolometric luminosities ∼ a few 10 39 erg s −1 , and almost no emission above 1 keV. It has never been clear how they fit into the general scheme of accreting compact objects. To address this problem, we studied a sample of seven ULSs with extensive Chandra and XMM-Newton coverage. We find an anticorrelation between fitted temperatures and radii of the thermal emitter, and no correlation between bolometric luminosity and radius or temperature. We compare the physical parameters of ULSs with those of classical supersoft sources, thought to be surfacenuclear-burning white dwarfs, and of ultraluminous X-ray sources (ULXs), thought to be super-Eddington stellar-mass black holes. We argue that ULSs are the sub-class of ULXs seen through the densest wind, perhaps an extension of the soft-ultraluminous regime. We suggest that in ULSs, the massive disk outflow becomes effectively optically thick and forms a large photosphere, shrouding the inner regions from our view. Our model predicts that when the photosphere expands to 10 5 km and the temperature decreases below ≈50 eV, ULSs become brighter in the far-UV but undetectable in X-rays. Conversely, we find that harder emission components begin to appear in ULSs when the fitted size of the thermal emitter is smallest (interpreted as a shrinking of the photosphere). The observed short-term variability and absorption edges are also consistent with clumpy outflows. We suggest that the transition between ULXs (with a harder tail) and ULSs (with only a soft thermal component) occurs at blackbody temperatures of ≈150 eV.

Some ultraluminous X-ray sources (ULXs) are surrounded by collisionally ionized bubbles, larger and more energetic than supernova remnants: they are evidence of the powerful outflows associated with super-Eddington X-ray sources. We illustrate the most recent addition to this class: a huge (350 pc × 220 pc in diameter) bubble around a ULX in NGC 5585. We modelled the X-ray properties of the ULX (a broadened-disc source with LX ≈ 2–4 × 1039 erg s−1) from Chandra and XMM–Newton, and identified its likely optical counterpart in Hubble Space Telescope images. We used the Large Binocular Telescope to study the optical emission from the ionized bubble. We show that the line emission spectrum is indicative of collisional ionization. We refine the method for inferring the shock velocity from the width of the optical lines. We derive an average shock velocity ≈125 km s−1, which corresponds to a dynamical age of ∼600 000 yr for the bubble, and an average mechanical power Pw ∼ 1040 erg s−1; thus, the mechanical power is a few times higher than the current photon luminosity. With Very Large Array observations, we discovered and resolved a powerful radio bubble with the same size as the optical bubble, and a 1.4-GHz luminosity ∼1035 erg s−1, at the upper end of the luminosity range for this type of source. We explain why ULX bubbles tend to become more radio luminous as they expand while radio supernova remnants tend to fade.