Line detections in photospheric radius expansion bursts from 4U 1820-303

Context. The Neutron star Interior Composition ExploreR (NICER) is the instrument of choice for the spectral analysis of type I X-ray bursts, as it provides high throughput at the X-ray CCD resolution down to 0.3 keV. Aims. Triggered by the detection of absorption and emission lines in the first four photospheric radius expansion (PRE) bursts detected by NICER, we wish to test the dependence of the absorption line energies on the inferred blackbody radius because it was reported that the absorption line energies were positively correlated with the inferred blackbody radius. This was tentatively explained by a combination of a weaker gravitational redshift and higher blueshifts in a burst with a larger blackbody radius. Methods. We thus reanalysed these four bursts and analysed another eight bursts from 4U 1820-303, for which we report evidence for PRE. We first followed the spectral evolution of the burst on the shortest possible timescales (tenth of a second). We adopted two parallel continuum descriptions to characterise the photospheric expansion and line evolution. Using the accretion-enhanced model, in which the burst emission is modelled as the sum of a blackbody and a component describing the persistent emission recorded prior to the burst and multiplied by a constant ( f a ), we inferred maximum equivalent blackbody radii up to ∼900 km. The peak bolometric (0.1–20 keV) luminosity reached between 4 and 7 × 10 38 erg s −1 (and even higher when absorption from a putative photoionised absorber is accounted for) in our sample of bursts. This exceeds the Eddington luminosity of a helium accretor. In individual bursts, we detected absorption lines and assessed their significance through extensive Monte Carlo simulations. To characterise the spectral lines, we used dedicated plasma codes available within SPEX with a phenomenological continuum. A deep search throughout the temperature–velocity parameter space was run to explore Doppler shifts and minimise the chance of becoming stuck in local minima. Results. We detected several significant (> 99.9 % significance) absorption lines, including the 2.97 keV line that was previously reported. We do not confirm the correlation between the line energies and the inferred blackbody radius, but for some bursts with larger radii, up to four lines are reported, and the line strength is higher. From the modelling of the feature lines, a photoionised or collisionally ionised slightly redshifted (almost rest-frame) gas in emission is suggested in most cases. In particular for the burst presenting the greatest PRE, a combination of photoionisation plasma in emission and absorption is preferred, however.