I know that for many sources, not having > 77 keV data will have no significant impact on spectral fits, given that the PIN-Si diode cross-section for photon detection decreases above 90 keV, but I am curious as to the nature of the 77 keV cutoff.

I did not find this information in the HXD performance paper or in analysis meeting excerpts, for instance, (the only place I am aware of it being mentioned is Nakazawa-san's 20 Oct email describing pre-r1.2-r1 processing). I apologize if this information has been covered previously.

> What is the nature of the sharp 77 keV cutoff in
> pre-rev. 1.2 PIN source data, which is not
> present in rev 0.7 data? Does it have to do with
> the improvements to the energy scale?

Our answer is simply YES, it is caused by the change of gain definition.

Please read also section 3.1 of Kokubun et al. PASJ issue (also on the web http://www.astro.isas.jaxa.jp/suzaku/news/2006/1016/index.html.en) and on astro-ph yesterday.

What we found is that in some of the PINs, their gain do not look linear in the hard band, i.e. the peak position of 81 keV (133Ba, ground-cal) and 76 keV (Bi-K, in-orbit coincidence events) is slightly lower in PHA channel than those predicted by the the linear gain extrapolated from below ~40 keV. As mentioned in section 3.4 of Kokubun et al., this energy band is very important in generating the current version of the response. Although the nature of this effect is not yet clear, we hence decided to utilize a simple spline function to address the gain parameters. Then, the gain above the Bi-K line is not defined with this scheme, causing the cut off in the spectra. (It is possible but needs additional work.)

As you mentioned, the spectra above 77 keV is not important in most cases. What is more, it is NOT well calibrated (i.e. difficult to calibrate). Though we think it was the best way to obtain the most correct calibration, we are sorry for not noticing this apparent change from rev0.7 to rev1.2 enough. Note that the GSO data will be used in this energy band.

> I did not find this information in the HXD performance
> paper or in analysis meeting excerpts, for instance,
> (the only place I am aware of it being mentioned is
> Nakazawa-san's 20 Oct email describing pre-r1.2-r1
> processing). I apologize if this information has been
> covered previously.

2006-11-08 Accept　
2006-11-08 Forward to HXD team
2006-11-08 Done