The spectrum is not a standard cooling flow
The figure below shows the spectrum of the central 2 arcmin radius
region of the Perseus cluster subtracted by the spectrum from an
annular region from 4 to 6 arcmin. It shows that the cool component
is concentrated towards the center of the cluster and has a spectrum
that is strongly peaked at just above 1 keV.
The model indicated by the solid line is an example unabsorbed standard
cooling flow. The shoulder at ~800 eV is largely due to Fe XVII. We see
no strong evidence for this ionization stage.
Including extra absorption on the cooling flow does help to produce a
model approximately the same shape as the observed spectrum however the
detailed fit is not good. If the low temperature cut-off in the cooling
flow is allowed to be a free parameter then it is required to be > 1 keV.
At 90% statistical confidence the extra absorption is inconsistent with
being at zero redshift however, given the calibration uncertainties around
the instrument oxygen edge, we do not consider this a strong result.