A systematic comparison of galaxy cluster temperatures measured with NuSTAR and Chandra

Temperature measurements of galaxy clusters are used to determine their masses, which in turn are used to determine cosmological parameters. However, systematic differences between the temperatures measured by different telescopes imply a significant source of systematic uncertainty on such mass estimates. We perform the first systematic comparison between cluster temperatures measured with Chandra and NuSTAR. This provides a useful contribution to the effort of cross-calibrating cluster temperatures due to the harder response of NuSTAR compared with most other observatories. We measure average temperatures for eight clusters observed with NuSTAR and Chandra. We fit the NuSTAR spectra in a hard (3-10 keV) energy band, and the Chandra spectra in both the hard and a broad (0.6-9 keV) band. We fit a power-law cross-calibration model to the resulting temperatures. At a Chandra temperature of 10 keV, the average NuSTAR temperature was $(10.5\pm 3.7)$ and $(15.7\pm 4.6){{\ \rm per\ cent}}$, lower than Chandra for the broad- and hard-band fits, respectively. We explored the impact of systematics from background modelling and multiphase temperature structure of the clusters, and found that these did not affect our results. Our sample are primarily merging clusters with complex thermal structures so are not ideal calibration targets. However, given the harder response of NuSTAR, it would be expected to measure a higher average temperature than Chandra for a non-isothermal cluster, so we interpret our measurement as a lower limit on the difference in temperatures between NuSTAR and Chandra.