HXD-PIN Detectors

7750

p+ n+

Al

SiO2 p+ p+

Guard Ring

Al n+

2490

2000 Polyimide

10750 bonding wire

Silicone rubber

Ceramics case Silicone rubber

Figure 3.14: A photograph (left) and a schematic picture (right) of a HXD-PIN diode

Figure 3.15: A PIN spectrum of²⁴¹Am obtained on the ground. The temper-ature is−20^◦C

Figure 3.16: Angular transmission function of the fine collimator calculated at azimuth angle of 0^◦. The 0^◦ azimuth azimuth angle is defined along the positive X-axis in Figure 3.11.

Chapter 4

In-Orbit Calibrations &

Performances of the HXD-PIN Detectors

In this chapter, we describe in-orbit calibrations and obtained performance of the HXD-PIN detectors. The in-orbit calibrations of the HXD-PIN diodes were carried out in three steps.

Firstly, we established the absolute energy scale, evaluated the energy resolution (§4.1), and, optimize the lower energy threshold (§4.2), for the 64 PINs individually. Secondly, we optimized event selection criteria in order to minimize the residual non X-ray background (§4.3). Finally, the response matrices of individual PINs were constructed, considering quantum efficiencies and effective areas (§4.4). Another important topic, background estimation is described in §4.5.

4.1 Energy Scale

Calibrations of gains and energy scale and measurements of energy resolution are essential in constructing the instrumental response function, which in turn is needed to reconstruct incident spectra from the observed pulse-height distributions. Before the launch, the energy scales of the 64 PINs were precisely measured using the standard gamma-ray sources, within ∼1% accuracy (Takahashi et al. 2007). These energy scales are not expected to change significantly after the launch, because neither the charge collection efficiency of the PIN diodes nor the capacitance of the charge sensitive amplifiers is sensitive to the environmental changes. Nevertheless, the energy scale is so important that it should be accurately reconfirmed using the actual data.

In order to determine the energy scale, line features in the spectrum are useful. Since the line emissions can be rarely observed from celestial objects in the bandpass of HXD-PIN, we used X-ray line emissions generated within the detector. Figure 4.1 illustrates the events used for the calibration. When an X-ray photon is absorbed in GSO, a fluorescent

BGO Well

GSO PIN

X-rays

Gd-K

Absorption

BGO Well

GSO PIN

X-rays

Bi-K Absorption

Figure 4.1: Illustration of an event in which Gd-K X-ray is detected with a PIN diode (left) and in which Bi-K is detected.

X-ray photon from gadlinium (Gd-K_α: 42.7 keV) sometimes escape from the GSO and is absorbed in the PIN detector with some probability. In the same way, fluorescent X-rays from bismuth of the BGO shield (Bi-Kα: 76.2 keV) are able to be detected with PIN diodes. We can extract these events by selecting PIN events coincident with signals from GSO or BGO.

After appropriate event selections, we obtained spectra of Gd-K lines and Bi-K lines for each PIN diode. Figure 4.2 and Figure 4.3 show the example spectra of Gd-K lines and Bi-K lines, respectively. The lines of Gd-K lines were fitted with four Gaussians which represent K_α1 (43.0 keV), K_α2 (42.3 keV), K_β1 + K_β3 (48.7 keV), and K_β2 (50.0 keV).

The Gaussian centroid energies were constrained to obey these line-energy ratios, and the Gaussian widths were tied together but left as a free parameter. The Bi-K lines were fitted with two Gaussians of Kα1 (77.1 keV) and Kα2 (74.8 keV). Again, the centroid energies were constrained and the widths of the Gaussian were tied, as was done with the Gd-K spectra. The best-fit models are indicated both in Figure 4.2 and in Figure 4.3 with the red curve and each component of the model is also shown with the dashed curves.

In addition to the two fluorescent lines, another anchor point is needed at low energies to accurately fix the energy scales. For this purpose, “pedestal channel”, which is defined as the peak channel of noise spectrum obtained by the random triggers from scintillators, is used (see Figure 4.4 for the actual spectrum). Although the energy deposits to the relevant PIN diode are considered in this case essentially zero, the channel becomes non-zero, because the peak-hold circuit before ADC latches the noise peak during each trigger gate of a widths ∼ 10 µs. Therefore, the pedestal channel of each PIN is thought to be proportional to its energy resolution. Based on ground measurements, the typical energy

Figure 4.2: An energy spectrum of a PIN diode, in which the coincident events of PIN and GSO are accumulated over a half year since the launch. The two peaks correspond to K_α and K_β lines of gadolinium.

Figure 4.3: A spectrum of Bi K_α line obtained from a PIN diode.

Figure 4.4: A spectrum of the pedestalpeak obtained from a PIN detector.

Finally, a spline curve is derived over an energy range of 2–77 keV for each PIN diode using the three calibration points, Gd-K, Bi-K, and the pedestal channel. In Figure 4.5, the obtained spline function is plotted with the calibration data points for a representative PIN diode. The in-orbit energy scales coincides well with the pre-launch ones measured in an energy range of 20–50 keV, while some PINs show significant nonlinearities above 50 keV. Thus, the accuracy of this energy scale is estimated to be ∼ 1%. As mentioned above the gains are stable at least for half a year. The energy scales determined here are used for all data analyses in this thesis.

The long-term stability of the energy scale of PIN detectors is monitored by using the Gd-K lines with its counting rate of∼1.2×10⁻⁴ counts s⁻¹. Figure 4.6 shows the compar-ison of the Gd-K peak channels obtained between the two periods of August–September 2005 and February–April 2006. As shown in the figure, the Gd-K peak positions in the PIN spectra have stayed constant, within one ADC channel (which corresponds to

∼0.4 keV), at least for half a year since the launch.

The fluorescent X-ray peaks also work as good indicators of energy resolution in-orbit derived from the Gd-K peaks. Figure 4.7 shows the plot of in-orbit energy resolutions of PIN diodes against the resolutions obtained from on-ground test. The typical in-orbit energy resolution is∼4 keV in FWHM, which is roughly consistent with those measured before launch. A slight increase of ∼ 0.3 keV from pre-launch resolutions are probably due to a difference in the electrical noise conditions.