A 200-Second Quasi-Periodicity After the Tidal Disruption of a Star by a Dormant Black Hole

See allHide authors and affiliations

Science  24 Aug 2012:
Vol. 337, Issue 6097, pp. 949-951
DOI: 10.1126/science.1223940

You are currently viewing the figures only.

View Full Text

  1. Fig. 1

    XMM-Newton and Suzaku light curve of Sw J1644+57 together with the SWIFT-XRT 0.3-to-10-keV light curve for reference (gray) (37, 38). For 12 XMM-Newton observations (XMM #1 to 12), we extracted 0.2- to 10-keV source and background light curves from the Epic-PN camera, using 40–arc sec circular regions. After accounting for the flaring background, a substantial fraction of the data in XMM #2, 3, and 12 had to be excluded. For Suzaku, we used a box region of 250–arc sec to extract the source light-curve from the two front-illuminated cameras and 150–arc sec for the background. Every observation had a background level less than ∼5% of the total flux. For each observation, we created an energy spectrum to which we fitted a model consisting of an absorbed power-law and used this to obtain a conversion factor between the count-rate and fluxes in physical units. The average flux levels in each observation are shown in real time in the main axis (stars, where the 1-SD error bars are smaller than the symbol size), and the inset compares each observation with data points binned in 32-s intervals. The vertical lines separate the various observations. The right-hand axis gives the conversion to luminosity of the source assuming isotropic radiation, and the dashed curve shows a t−5/3 relation, with t being the time since 28 March 2011.

  2. Fig. 2

    (A) Power spectra for Suzaku and (B) for XMM #1, in the 2- to 10-keV energy range. The power spectra are normalized so that their integral gives the squared RMS fractional variability. The Poisson noise level expected from the data errors is shown as the dashed horizontal line. We checked that the peak is robust to a variety of frequency and time resolutions. The arrow in both panels mark the presence of a QPO with a centroid frequency of νsuzaku ~ 4.8 mHz. The solid curves enclose the range of the best fit to the underlying continuum as described in table S1 (model 2). The dotted curves show the 99.99 and 99.73% (3σ) threshold for significant detection.