Photoabsorption of X-rays below the Xenon K-binding energy at 34.5 keV,
occurs in the L or lower order shells. In this case only a single cloud is
produced for each X-ray event, because the probability of atomic
relaxation via the Auger effect is much higher than via fluorescence
emission and in any case the mean penetration depth of the L fluorescence
photon, in 5 atmospheres Xenon, is less than 1 mm. For each X-ray event
with 
keV only a single VUV burst will be produced and
then detected by the HPGSPC.
For incident X-ray energies above 34.5 keV, the probability that
photoabsorption occurs in the K shell is 86% while the probability
that atom relaxes via a 
(29.7 keV) or a 
(33.8 keV) fluorescence photon is as high as 87%. Depending on
geometry, filling pressure and energy of the original X-ray event
the fluorescence photon can be reabsorbed in the gas cell,
generating a second localized electron cloud at a different point
respect to the primary photoelectron cloud. The two electron clouds
from the double interaction (from now on referred as residual and
fluorescence) will enter the Scintillation Region with a time
difference of 
, limited by the maximum drift time in
the Drift Region. In this case, a double VUV light burst associated
to a single X-ray event will be produced and detected by the
HPGSPC. The detection of a 
or 
photon will
``gate" the partial photoabsorption.
X-ray events above 34.5 keV can, however, interact with the L
shell or interact with the K shell and the atom relaxes via an Auger
electron. In these cases only a single cloud is produced. Finally,
for X-ray events that give rise to a 
or 
fluorescence photon a single event can still be detected if the two
clouds are not spatially resolved or when the fluorescence photon
escapes the detector leaving only the residual energy deposit of 
, where 
is the binding energy and 
is
the energy of the incident X-ray.
It has been suggested by many authors that the detection of the double events may provide a unique signature for a true X-ray event and, in this sense, can be used to discriminate against a non X-ray background event (Manzo et al. 1980; Ramsey et al. 1990; Dangendorf et al. 1989).
The ``Fluorescence Gated" technique can also be used to improve the energy resolution above the Xenon K shell (Taylor et al. 1981). Since the energy of the fluorescence photon is exactly known, only the statistical fluctuations relative to the clouds of primaries (residuals) will affect the energy determination.
Following the physics of photoabsorption three different scientific modes have been defined for the HPGSPC and implemented during the on ground data analysis:
In SE mode only X-ray photons detected as a single VUV light burst are collected. The SE mode can be used in the HPGSPC whole range and is the only mode for X-ray events with energy below 34.6 keV. In FG mode only incident X-rays producing double correlated events are accumulated. The FG mode is used only above 34.6 keV. In the ``All" mode both ``Single" and ``Double correlated" events are accumulated. Defined in the entire energy band of the HPGSPC, the "All" mode maximizes the detection efficiency of the Instrument.
The HPGSPC energy resolution for a well collimated ``on axis" X-ray beam is determined in the first approximation by the statistical fluctuations in the number of electrons produced in the absorption region and the statistical fluctuations in the number of VUV photons detected by the PMTs (i.e. the photon counting statistics). In the case of the HPGSPC, considering a Fano factor of 0.15 (Anderson et al. 1979) and the VUV light yield as reported in Nguyen et al. (1980), a theoretical value for the energy resolution of 2.5% at 60 keV is obtained ( in ``All" mode).
When the HPGSPC is illuminated over the full geometrical aperture
(full area illumination) the energy resolution 
(where 
is the measured energy i.e. the sum of seven PMTs
signals) degrades and only slightly depends on the energy. This is
due to the dependence of the measured energy 
on the (
)
position at which the primary electron cloud enters the
Scintillation Region. Indeed, it is the effective solid angle
subtended by the seven PMTs (and then the VUV scintillation light
collected) that strongly depends on the (
) scintillation
position. Solid angle can vary from the center of the detector to
the walls by up to about 20% (Giarrusso et al.
1989). The measured energy 
, however, can be
corrected to the ``on axis" value and the energy resolution can be
restored just by multiplying the measured energy 
by the
ratio of the total effective solid angle ``on axis" to
the total effective solid angle subtended by the PMTs at the (
) point:
with
In the previous equation 
is the total solid angle
subtended by the seven PMTs at the (
) position and 
is
the solid angle subtended by the PMT i. Although different methods have
been derived to determine a statistically optimum estimation of the
scintillation position for an array of PMTs operating in an Anger camera
configuration, they require difficult to implement on-board hardware, or
a long processing time (Giarrusso et al. 1995). The method
implemented for onboard position reconstruction and energy correction
processing makes use of the signal of the central PMT and the highest
signal of the lateral PMTs, to determine for each event two energy
independent coefficients which are defined as:
and
where 
is the signal detected by the central PMT and
is
the highest signal among the lateral PMTs. Because there is an univoque
correspondence between the (
) position in a given sector and the
(
, 
) space, once the 
and 
coefficients are
obtained, the 
correction factor is extracted from a look-up table
that is experimentally determined and implemented onboard as a memory
matrix of 
values. Each value in the table is addressed by the
and 
coefficients measured for the event. To make the
determination of the 
value faster, the look-up table is built
using the isocoefficients 
. Thus the look-up table is defined
constant map. More details on the position reconstruction and
energy correction of the event can be found in (Giarrusso et al.
1995; Giarrusso et al. 1989).