The discovery of photospheric absorption lines in XMM-Newton spectra of the X-ray bursting neutron star in
EXO 0748-676 by Cottam and collaborators allows us to constrain the neutron star mass-radius ratio from the measured gravitational redshift. A radius of R=9-12km for a plausible mass range of M=1.4-1.8M
☉ was derived by these authors. It has been claimed that the absorption features stem from gravitationally redshifted (z=0.35) n=2-3 lines of H- and He-like iron. We investigate this identification and search for alternatives. We compute LTE and non-LTE neutron-star model atmospheres and detailed synthetic spectra for a wide range of effective temperatures (T
eff=1-20MK) and different chemical compositions. We are unable to confirm the identification of the absorption features in the X-ray spectrum of
EXO 0748-676 as n=2-3 lines of H- and He-like iron (FeXXVI and FeXXV). These are subordinate lines that are predicted by our models to be too weak at any T
eff. It is more likely that the strongest feature is from the n=2-3 resonance transition in FeXXIV with a redshift of z=0.24. Adopting this value yields a larger neutron star radius, namely R=12-15km for the mass range M=1.4-1.8M
☉, favoring a stiff equation-of-state and excluding mass-radius relations based on exotic matter. Combined with an estimate of the stellar radius R>12.5km from the work of Oezel and collaborators (
2006Natur.441.1115O), the z=0.24 value provides a minimum neutron-star mass of M>1.48M
☉, instead of M>1.9M
☉, when assuming z=0.35. The current state of line identifications in the neutron star of
EXO 0748-676 must be regarded as highly uncertain. Our model atmospheres show that lines other than those previously thought must be associated with the observed absorption features.