Timing of the accreting millisecond pulsar IGR J17511-3057

Context. Timing analysis of accretion-powered millisecond pulsars (AMPs) is a powerful tool for probing the physics of compact objects. The recently discovered IGR J17511-3057 was the twelfth discovered of the 13 AMPs known. The Rossi XTE satellite provided an extensive coverage of the 25 days-long observation of the source outburst. Aims. Our goal is to investigate the complex interaction between the neutron star magnetic field and the accretion disk by determining the angular momentum exchange between them. The presence of a millisecond coherent flux modulation allows us to investigate this interaction from the study of pulse arrival times. To separate the neutron star proper spin frequency variations from other effects, a precise set of orbital ephemeris is mandatory. Methods. Using timing techniques, we analysed the pulse phase delays by fitting differential corrections to the orbital parameters. To remove the effects of pulse phase fluctuations, we applied the timing technique that had been already successfully applied to the case of an another AMP, XTE J1807-294. Results. We report a precise set of orbital ephemeris. We demonstrate that the companion star is a main-sequence star. We find pulse phase delay fluctuations on the first harmonic with a characteristic amplitude of about 0.05, similar to those also observed for the AMP XTE J1814-338. For the second time, an AMP shows a third harmonic detected during the entire outburst. The first harmonic phase delays exhibit a puzzling behaviour, while the second harmonic phase delays clearly spin-up. The third harmonic also shows a spin-up, although not highly significant (3 sigma c.l.). The presence of a fourth harmonic is also reported. If we assume that the second harmonic is a good tracer of the spin frequency of the neutron star, we infer a mean spin frequency derivative for this source of 1.65(18) x 10(-13) Hz s(-1). Conclusions. To interpret the pulse phase delays of the four harmonics, we apply the disk threading model, but obtain different and incompatible. M estimates for each harmonic. In particular, the phase delays of the first harmonic are heavily affected by phase noise, and consequently, on the basis of these data, it is not possible to derive a reliable estimate of. M. The second harmonic gives a. M consistent with the flux assuming that the source is at a distance of 6.3 kpc. The third harmonic gives a lower. M value, with respect to the first and second harmonic, and this would reduce the distance estimate to 3.6 kpc.