NuSTAR observation of GRO J1744-28 at low mass accretion rate

Context. Neutron stars in low-mass X-ray binaries are important systems for studying the physics of accretion onto compact objects. The system GRO J1744-28 is particularly interesting as it usually shows clear pulsations as well as X-ray bursts. Additionally, there are claims for a magnetic field of 5x10(11) G through the detection of a cyclotron resonant scattering feature (CRSF).Aims. We present the spectral analysis of GRO J1744-28 using similar to 29 ks of NuSTAR data taken in 2017 February at a low luminosity of 3.2x10(36) erg s(-1) (3-50 keV). Our goal is to study the variability of the source spectrum with pulse phase and to search for the claimed CRSF.Methods. The continuum spectrum was modeled with an absorbed power law with exponential cutoff, and an additional iron line component. We found no obvious indications for a CRSF, and therefore performed a detailed cyclotron line search using statistical methods. We performed this search on pulse phase-averaged spectra and on phase-resolved spectra.Results. GRO J1744-28 was observed in a low-luminosity state. The previously detected Type II X-ray bursts are absent. Clear pulsations at a period of 2.141124(9) Hz are detected. The pulse profile shows an indication of a secondary peak that was not seen at higher flux. The upper limit for the strength of a CRSF in the 3-20 keV band is 0.07 keV (90% CL), lower than the strength of the line found at higher luminosity.Conclusions. The detection of pulsations shows that the source did not enter the "propeller" regime, even though the source flux of 4.15x10(-10) erg cm(-2) s(-1) was almost one order of magnitude below the threshold for the propeller regime claimed in previous studies on this source. The transition into the propeller regime in GRO J1744-28 must therefore be below a luminosity of 3.2x10(36) erg s(-1) (3-50 keV), which implies a surface magnetic field less than or similar to 2.9x10(11) G and mass accretion rate less than or similar to 1.7x10(16) g s(-1). A change of the CRSF depth as function of luminosity is not unexpected and has been observed in other sources. This result possibly implies a change in emission geometry as function of mass accretion rate to reduce the depth of the line below our detection limit.