Discovery of luminous pulsed hard X-ray emission from anomalous X-ray pulsars 1RXS J1708-4009, 4U 0142+61, and 1E 2259-586 by Integral and RXTE

We investigated the time-averaged high-energy spectral characteristics of the persistent anomalous X-ray pulsars (AXPs) 1RXS J1708-4009, 4U 0142+61, 1E 2259+586, and 1E 1048.1-5937, using RXTE PCA (2-60 keV), RXTE HEXTE (15-250 keV), and INTEGRAL IBIS ISGRI (20-300 keV) data. We discovered hard spectral tails for energies above 10 keV in the total and pulsed spectra of AXPs 1RXS J1708-4009, 4U 0142+4009, and 1E 2259+586, but 1E 1048.1-5937 appeared to be too weak to be detected. Improved hard X-ray spectral information for AXP 1E 1841-045 is also presented. The pulsed and total spectra above 10 keV have power-law shapes, and there is so far no significant evidence for spectral breaks or bends up to similar to 150 keV. The pulsed spectra above 10 keV are exceptionally hard with indices for four AXPs in the range similar to 1.0 to 1.0. Below 10 keV these indices were in the range 2.0-4.3, indicating the very drastic spectral changes in a narrow energy interval around 10 keV. The best-fit power-law models to the total spectra between similar to 10 and 150 keV are significantly softer, with indices measured for 1E 1841-045, 1RXS J1708-4009, and 4U 0142+61 in the range 1.0-1.4. For the latter AXPs the pulsed fractions are consistent with 100% around 100 keV but are different at 10 keV: similar to 10% for 4U 0142+61, similar to 25% for 1E 1841-045, and consistent with 100% for 1RXS J1708-4009. The luminosities of these total and pulsed spectral tails (10-150 keV) largely exceed the total available spin-down powers by factors ranging from similar to 100 to similar to 600. We also reanalyzed archival CGRO COMPTEL (0.75-30 MeV) data: no AXP detections can be claimed, and the obtained upper limits indicate for 1RXS J1708-4009, 4U 0142+61, and 1E 1841-045 that strong breaks or bends must occur somewhere between similar to 150 and 750 keV. We discuss predictions from first attempts to model our hard X-ray and soft gamma-ray spectra in the context of the magnetar model.