6.6. ABSOLUTE DETECTION EFFICIENCY 87
(a) DSSD (b) CdTe
Figure 6.23: The absolute photo-peak efficiency of each DSSD and CdTe detector.
gamma-ray are already absorbed in upper layers. The experimental and simulation values are consistent within 10 % systematic error induced by uncertainties of the radioactive sources, except 511 keV incident gamma-ray. For 511 keV gamma-ray, the experimental detection efficiency is 20 % less than simulation value. One reason is that we do not take account of the broadening of 511 keV line in simulation due to the initial momentum of an electron before the pair annihilation, which is typically a few keV. When we expand energy window into 511 ± 10 keV, the experimental detection efficiency agrees with simulation within 10 % systematic error.
6.6.2 Compton mode
The absolute detection efficiency of Compton mode is investigated with 57Co, 133Ba, and
22Na radioactive sources. The data was obtained by the coincidence trigger between the DSSD and the CdTe detectors. As shown in Table. 6.2, 133Ba and 22Na sources were placed at (x,y,z) = (0,0,150), and 57Co was placed at (0,-60,30). Thus, the gamma-rays of 57Co (122.1 keV, 136.0 keV) are incident from the direction of 63.4 ◦. Although the detection of such low energy gamma-rays is difficult due to the small energy deposit in the DSSD, it is still possible to detect events with wide scattering angle, where, the energy deposits in DSSD become large enough to be detected with the DSSD energy threshold of 15 keV.
An absolute normalization is calculated with given intensity of the source and dead-time correction. We applied a secondary selection for the events where the calculated ARM is included within±10 ◦ range. In addition, we selected Grade 1 Compton events which has no charge sharing for both the DSSD and CdTe detectors.
Figure 6.24 shows the experimental Compton mode spectra of57Co and133Ba together with the reproduced spectra by simulation. In the case of 57Co, the both the shape and the area of spectrum seem to be well reproduced. For the higher energy lines from133Ba, the experimental spectrum shows slightly longer tailing structure than the simulated spectrum, as a result, the area of each line seems to become less than simulation. But this does not mean the disagreement of absolute efficiency between experiment and simulation.
Indeed, the ratio of integrated counts over 266–386 keV energy range including four lines of133Ba becomes a consistent value, exp(266–386 keV) / sim(266–386 keV) = 0.96, when taking account for 10 % uncertainty of source intensity. Although more precise study of the spectral shape would be needed, the absolute detection efficiency is mostly consistent with the simulation.
The energy dependence of detection efficiency is summarized in Fig. 6.25. We inte-grated the counts of each line from –10 keV to +5 keV energy window centered on the line energy. The error bar of each point is 10 % error induced by uncertainty of source intensity. The experimental value agrees well with simulation. The absolute detection efficiency is 10−3–10−2 % over 250–511 keV energy window.
6.6. ABSOLUTE DETECTION EFFICIENCY 89
(a)
(b)
Figure 6.24: Comparison of experimental absolute Compton mode spectrum with simulated one.(a)57Co, (b)133Ba
Figure 6.25: The energy dependance of detection efficiency.
6.6.3 Field of view
In the above section, we demonstrated the consistency of absolute detection efficiency between experiment and simulation for incident gamma-ray from vertical direction. By using 356 keV and 511 keV gamma-ray point sources, we investigated the detection efficiency as a function of incident direction. The point source was placed on the plane at z = 60 mm (See 6.2) and shifted to the y-direction with 20 mm pitch. The same criteria of event selection is applied, which used to study the absolute detection efficiency in above section 6.6.2. We normalized the efficiency to unity at the center position of (x, y, z) = (0,0,60), thus the uncertainty of the source intensity is canceled. The 1-σ statistical error is applied to each point for both experiment and simulation.
The experimental result is summarized in Fig. 6.26 together with the simulation result. The relative efficiency at y =100 mm corresponding to 60 ◦ direction is 36 % for 356 keV and 17 % for 511 keV. The efficiency curves almost agree with simulation.
For the very large incident direction, the experimental result is 10–20 % larger than simulation, witch is not explained only by applying the statistical error. The origin of this inconsistency is under investigation.
Considering the results obtained from this and above sections, it is possible to say that we reproduced the absolute detection efficiency of the Compton camera within 10 % systematic error induced by the source intensity except very large incident angle.
6.6. ABSOLUTE DETECTION EFFICIENCY 91
Figure 6.26: The field of view (left ) 356 keV, (right) 511 keV.
6.6.4 E ff ect of a clustering
Until now, we handled only Grade 1 Compton events. The detection efficiency can be improved by using Grade 2 Compton events, which includes the charge sharing events.
Figure 6.27 shows the Compton mode 133Ba spectrum in the two cases of Grade 1 and 2. No secondary selection is applied. The fraction of the Grade 2 event to the Grade 1 events is 10–15 %. This is consistent with estimation from the hit pattern of the DSSD and CdTe detector. Thus, the detection efficiency is improved about 10 % by joining Grade 2 events into analysis. But, it should be noted that the energy resolution of Grade 2 events is degraded in comparison Grade 1 events because the electronic noise of each channel is also summed when clustering the energy deposits.
Figure 6.27: The spectrum of Grade 1 selection and Grade 2 selection