On joint analysing XMM-NuSTAR spectra of active galactic nuclei

A recently released XMM-Newton technical note has revealed a significant calibration issue between NuSTAR and XMM-Newton EPIC, and provided an empirical correction to EPIC effective area. To quantify the bias caused by the calibration issue to joint analysis of XMM-NuSTAR spectra and verify the effectiveness of the correction, in this work we perform joint-fitting of NuSTAR and EPIC-pn spectra for a large sample of 104 observation pairs of 44 X-ray bright AGN. The spectra were extracted after requiring perfect simultaneity between XMM-Newton and NuSTAR exposures (merging GTIs from two missions) to avoid bias due to rapid spectral variability of AGN. Before the correction, the EPIC-pn spectra are systematically harder than corresponding NuSTAR spectra by $\Delta \Gamma \sim 0.1$, subsequently yielding significantly underestimated cutoff energy $E_{\rm cut}$ and the strength of reflection component R when performing joint-fitting. We confirm the correction is highly effective and can commendably erase the discrepancy in best-fit $\Gamma$, $E_{\rm cut}$ and R, and thus we urge the community to apply the correction when joint-fitting XMM-NuSTAR spectra. Besides, we show that as merging GTIs from two missions would cause severe loss of NuSTAR net exposure time, in many cases joint-fitting yields no advantage compared with utilizing NuSTAR data alone. We finally present a technical note on filtering periods of high background flares for XMM-Newton EPIC-pn exposures in the Small Window mode.