Where is the engine hiding its missing energy? Constraints from a deep X-ray non-detection of the Superluminous SN 2015bn

SN 2015bn is a nearby hydrogen-poor superluminous supernova (SLSN-I) that has been intensively observed in X-rays with the goal to detect the spin-down powered emission from a magnetar engine. The early-time UV/optical/infrared (UVOIR) data fit well to the magnetar model, but require leakage of energy at late times of $\lesssim 10^{43}$ erg s$^{-1}$, which is expected to be partially emitted in X-rays. Deep X-ray limits until $\sim$300 days after explosion revealed no X-ray emission. Here, we present the latest deep 0.3--10 keV X-ray limit at 805 days obtained with \textit{XMM-Newton}. We find $L_X < 10^{41}$ erg s$^{-1}$, with no direct evidence for central-engine powered emission. While the late-time optical data still follow the prediction of the magnetar model, the best-fit model to the bolometric light curve predicts that $\sim$97\% of the total input luminosity of the magnetar is escaping outside of the UVOIR bandpass at the time of observation. Our X-ray upper limit is $<$1.5\% of the input luminosity, strongly constraining the high-energy leakage, unless non-radiative losses are important. These deep X-ray observations identify a missing energy problem in SLSNe-I and we suggest future observations in hard X-rays and $\gamma$-rays for better constraints. Also, independent of the optical data, we constrain the parameter spaces of various X-ray emission scenarios, including ionization breakout by magnetar spin-down, shock interaction between the ejecta and external circumstellar medium, off-axis $\gamma$-ray burst afterglow, and black hole fallback accretion.