Performance Characteristics of CZT Detectors for PET Imaging Applications
In this work, we developed an analytical model to study the charge collection and temporal behavior, which is expected to assist to optimize a number of design parameters including anode/cathode electrode pitch, steering electrode pitch and voltage bias.
Two common electrode configurations (pixilated and cross strip) for CZT detectors are shown in Fig. 1. When a 511 keV gamma ray hits the detector, a number of electron-holes are generated and travel within the internal electric field E. As stated by Shockley–Ramo theorem , the induced charge q(t) builds up as the electron-hole pairs moves from position r1 to r2, which can be written as q(t)=q0(Φω(r2)-Φω(r1))，where q0 is the elementary charge of electron, Φω is the weighting potential. To model the temporal dependency that impacts the time resolution, the drift time te,h of Δx distance is te,h= Δx/μe,hE, where μe,hare charge carries mobility.
For validation, we assemble two CZT detectors (Redlen, Canada) as shown in Fig. 1. For the cross-strip one, the cathodes orient orthogonally to the anodes, to provide spatial information in two dimensions. The detectors were biased at 400 V. After preamplifier and shaping amplifier circuits, the outputs were connected to a high-speed free running ADC (CAEN, sampling rate 500MS/s, dynamic range 1V) to measure energy spectrum.
The simulations and experimental results are illustrated in Fig. 2. The weighing potential has a pattern as shown in Fig. 2c, which is low near the cathode and rises rapidly when close to the anode. Such pattern is associated with the small pixel effect and single polarity sensing techniques, which affects both charge collection and temporal response of CZT detectors, as the function of different a/L ratio (pixel size to detector thickness). As a/L ratio decreases, the weighting potential shows a more abrupt change towards the anode region, which implies that the incomplete charge collection can be mitigated .
The normalized charge collection, q(t)/q0, as a function of normalized time t/Th (Th is the maximum hole transit time through detector thickness L) for three heights is shown in Fig. 2d, where the sharp rising presents the electrons’ motion, and the flat trend presents the holes’ motion. A1 point has the best charge collection effective and the rising time is about 0.12Th; Points A2 and A3 are farther away from cathode due to poor hole mobility, so a longer collect time are needed (0.68Th and 0.82Th respectively), which means for CZT non-uniform responses are occurred in different depths, which can be used to obtain DOI information. While, for cross section of CZT (A2, B2, C2), the charge signal are more uniform as shown in Fig. 2e.
The energy resolution is ~2.98±0.26% FWHM for 511 keV as shown in Fig. 2f, which is superior compared the typical value of 15-20% of a scintillation and PMT model. This result is based on a collimated beam of 511 keV gamma ray irradiating the whole pixel entering from the cathode side (i.e., interaction at all heights are included). In our future work, we will translate the beam along the Z direction to validate the detector performances again the analytical modeling.
 Hao Peng and Craig S Levin. “Design study of a high-resolution breast-dedicated PET system built system built from cadmium zinc telluride detectors”, Phys. Med. Biol., 55, pp. 2761-2788, 2010.
 D Palubiak, MM El-Desouki, O Marinov, MJ Deen, Q Fang, “High-speed, single-photon avalanche-photodiode imager for biomedical applications”, IEEE Sensors Journal, 11, pp. 2401-2412, 2011.
 S. Ramo, “Currents Induced by Electron Motion” Proc. IRE. 27, pp. 584-585, 1939.