Emulating the Effects of Pile-Up on X-ray Spectra

Pile-up in an X-ray CCD detector occurs when two or more X-ray photons striking the detector within the same readout time are interpreted as a single event. In instruments such as the ACIS detector on the Chandra X-ray Observatory, the effects of pile-up can include the registration of events as events of higher energies, a decrease in event count rate, and the migration of events to “grades” of poorer quality. Pile-up distorts the X-ray spectrum of bright objects and prevents us from making accurate estimates of the spectral parameters of X-ray sources, such as black holes. We use a neural network to emulate the effects of pile-up on spectra from different astrophysical sources. We construct training data sets using both simulated data generated with MARX (https://space.mit.edu/cxc/marx/) and observed data from Chandra/ACIS and Chandra/HETGS+ACIS-S (High Energy Transmission Grating Spectrometer + ACIS-S). Using these training data sets, we emulate the distortion of different source spectra into piled-up spectra. We discuss the potential applications of our work to the characterization of bright spectral states of black hole binaries and to improved analysis of archival ACIS data. We also discuss the prospects for generalizing our method to state-of-the-art and upcoming X-ray telescopes. This work is supported by NASA ADAP grant 80NSSC18K0425.