Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 5, OCTOBER 2004 1979 Design and Performance of 0.18-m CMOS Charge Preamplifiers for APD-Based PET Scanners J.-F. Pratte, S. Robert, G. De Geronimo, P. O’Connor, S. Stoll, C. M. Pepin, R. Fontaine, and R. Lecomte Abstract—The CMOS 0.18- m technology was investigated for two analog front-end projects: the low-power budget rat-head mounted miniature rat conscious animal PET (RatCAP) scanner, and the high-performance, low-noise, high-rate PET/CT application. The first VLSI prototypes consisted of 1- and 5-mW charge sensitive preamplifiers (CSP) based on a modified cascode telescopic architecture. Characterization of the rise time, linearity, dynamic range, equivalent noise charge (ENC), timing resolution and energy resolution are reported and discussed. When connected to an APD-LSO detector, time resolutions of 2.49 and 1.56 ns full-width half-maximum (FWHM) were achieved by the 1- and 5-mW CSPs, respectively. Both CSPs make it possible to achieve performance characteristics that are adequate for PET imaging. Experimental results indicate that the CMOS 0.18- m technology is suitable for both the low-power and the high-performance PET front-end applications. Index Terms—Avalanche photodiode (APD), complementary metal–oxide–semiconductor (CMOS), charge preamplifier, computed tomography, front-end electronic, positron emission tomography, Rat conscious animal positron emission tomography (RatCAP). I. INTRODUCTION R ECENT studies on deep sub-micron CMOS technologies have triggered interest for applications where highly integrated analog front-end electronics is required [1]. In particular, the low power and high integration of the CMOS 0.18- m technology make it an attractive choice for high-resolution APD-based imaging detection systems like small animal Positron Emission Tomography (PET) scanners. The scaling of the power supply down to 1.8 V reduces the available dynamic range, but the gain in signal-to-noise ratio at a given dissipated power makes this choice attractive compared to older technologies [2]. In this paper, the experimental results obtained with two Charge Sensitive Preamplifiers (CSP) for PET imaging implemented in CMOS 0.18 m are reported. Two specific applications are considered: a low-power, 1-mW CSP for Manuscript received November 23, 2003; revised June 3, 2004. This work was supported in part by grants from the Canadian Microelectronics Corporation (CMC), Le Fonds québécois de la recherche sur la nature et les technologies (FQRNT) and the Natural Science and Engineering Research Council of Canada (NSERC). Studentship awards to S. Robert and J.-F. Pratte from the Institut des Matériaux et Systèmes Intelligents (IMSI) of Sherbrooke and scholarships from La Fondation Jacques et Suzanne Lagassé were greatly appreciated. This work is also supported under a grant from the DOE Office of Biological and Environmental Research and DOE Contract DE-AC02-98CH10886. J.-F. Pratte, G. De Geronimo, P. O’Connor, and S. Stoll are with Brookhaven National Laboratory, Upton, NY 11973-5000 USA (e-mail: [email protected]). S. Robert, C. M. Pepin, R. Fontaine, and R. Lecomte are with the Université de Sherbrooke, Sherbrooke, QC J1K 2R2, Canada. Digital Object Identifier 10.1109/TNS.2004.836120 TABLE I LSO/APD DETECTOR MODULE SPECIFICATIONS the low-power budget RatCAP [3], [4], and a high-performance, 5-mW CSP for a high-resolution PET/CT scanner with high-density multi-crystals (phoswich) detector modules [5]. II. CHARGE SENSITIVE PREAMPLIFIERS A. Design Initial specifications for the CSPs were based on the detectors planned for the RatCAP scanner (see Table I), which is a Hamamatsu S8550 4 8 APD-array coupled to 2 mm 2 mm 5 mm LSO crystals [6], [7]. In this application, the APD is DC coupled to the CSP input requiring a feedback MOSFET that provides a pathway for the APD leakage current. The feedback MOSFET is designed to operate above threshold for better white noise performance and in saturation for better compensation performance [8]. The targeted preamplifier gain of 3.3 mV/fC, selected to provide a maximum output swing of 100 mV, corresponds approximately to an input signal of 190 000 photoelectrons. The APDbased detector modules for the RatCAP are expected to be used at a gain of 50, producing approximately 115 000 photoelectrons for 511 keV photons in LSO. A higher signal is expected from the PET/CT detectors at 511 keV, but an excellent signal-tonoise ratio must also be achieved for the detection of low-energy X-rays. Fig. 1 shows the modified telescopic cascode topology of the CSP. Transistor M2 is used as the main current source to produce the desired transconductance in the input transistor M1. The CSPs were implemented in TSMC CMOS 0.18- m technology, having six metal layers and one poly layer, through the Canadian Microelectronics Corporation. B. Noise Optimization The N-channel MOSFET input device of the CSP has been optimized with respect to the technology parameters and the detector characteristics at its operating point (capacitance, leakage 0018-9499/04$20.00 © 2004 IEEE Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. 1980 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 5, OCTOBER 2004 Fig. 1. CSP architecture. The transistor M2 is used as the main current source to achieve the desired transconductance. current and gain), using the EKV transistor model [9], to minimize the equivalent noise charge (ENC) [10] TABLE II SHAPING FORM FACTORS (1) (2) (3) (4) with (5) where , and are the series, parallel and flicker ENC contribution, the Boltzmann’s constant, the the form factors of temperature in kelvin, is the total capacitance at the the shaping function [10], input, the shaping function peaking time, the coefficient of thermal noise for the input MOSFET, the source transconthe parasitic resistance in ductance of the input MOSFET, the NMOS flicker series with each transistor electrode, the gate oxide capacitance per unit area, noise coefficient, and the width and length of the input NMOS, the the APD gain, a coefficient which APD leakage current, the transcondepends on the region of operation, and ductance of the reset transistor. The size of the input device was optimized for the derivative of a third order semi-gaussian shaping function [11] with a peaking time of 70 ns and power consumption of 1 and 5 mW. Table II presents the shaping form factor for a CR-RC, CR -RC and the derivative of a 3rd order semi-Gaussian shaping function [11]. All other transistor dimensions of the CSP are optimized for minimum white series and noise contribution on the overall ENC, mainly set by the input device. III. EXPERIMENTAL RESULTS A test printed circuit board (PCB) developed at Brookhaven National Laboratory (BNL) was used for characterization. It includes an 1.8-V power and ground plane, a low-noise 1.8-V voltage regulator and an on-board Analog Devices AD8009 buffer used for both 50 ohms adaptation with the shaping 5 at the CSP output. All amplifiers and voltage gain of required high voltage hardware and connectors for the BNL and Sherbrooke APD detector modules were implemented on the PCB. Charges were injected with a test pulse applied to an on-board characterized 1-pF capacitor for each CSP. The chip was directly wire bonded to the PCB. The BNL detector module is based on the Hamamatsu S8550 4 8 APD-array coupled to a matched array of 2 mm 2 mm 5 mm LSO crystals. The Sherbrooke detector module is based on a former PerkinElmer BGO/APD module [12] retrofitted with 3 mm 5 mm 25 mm LSO scintillators. It has a leakage current of 45 nA and a capacitance of 22 pF when biased at 350 V. A. Electronic Characterization The rise time, linearity and the dynamic range of the CSPs were measured using a custom nuclear instrumentation module (NIM) pulser and a Tektronix TDS 784A digital oscilloscope for input capacitance of 0 to 22 pF. Figs. 2 and 3 present the measured linearity for the 1- and 5-mW CSPs. A gain of 2.7 mV/fC was deduced for both prototypes. The input capacitance had no influence on the dynamic range and the linearity. Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. PRATTE et al.: DESIGN AND PERFORMANCE OF 0.18- m CMOS CHARGE PREAMPLIFIERS FOR APD-BASED PET SCANNERS Fig. 2. Gain linearity for 1-mW CSP over operating range. Fig. 3. Gain linearity for 5-mW CSP over operating range. The rise time for the 5- and 1-mW CSP was 7 and 9 ns, respectively, without any capacitor added at their input ( pF). With a 10-pF capacitor at the input, a rise time of 13 and 18 ns was measured respectively. B. ENC Measurements Gain and noise measurements for the 1-mW CSP were performed for shaping times from 10 ns to 10 s with a custom NIM bipolar CR -RC shaper. The custom NIM pulser was used to inject the charges through the 1-pF capacitor. The output amplitude and peaking time of the chain was measured using a Tektronix TDS 784A digital oscilloscope and the noise using a Rhode & Schwartz TRUE-RMS meter. For the 5-mW CSP measurements, a combination of the Ortec 579 Timing Filter Amplifier and the Ortec 673 Spectroscopy Amplifier with unipolar shaping were used, for the same shaping time range. A BNC BL-2 Pulse Generator was used to inject the charges through the 1-pF capacitor. The output amplitude and peaking time of the chain was measured using a Tektronix TDS 784A digital oscilloscope and the noise using a Fluke 8920A TRUE-RMS. Figs. 4 and 5 present the measured ENC as a function of the input capacitance for the two devices. The discontinuities in the ENC curves in Fig. 5 are due to the use of two different shaping amplifiers, as mentioned above. The evaluated ENC for 1981 Fig. 4. ENC measurements for the 1-mW CSP. Fig. 5. ENC measurements for the 5-mW CSP. The discontinuities are due to the use of two different shapers (Ortec 579 timing filter amplifier and the Ortec 673 spectroscopy amplifier) in order to cover a peaking time from 10 ns to 10 s. the 1-mW CSP connected to the APD/LSO based detector is reported in Fig. 6. A minimum ENC of 1117 electrons-rms at 100 ns was measured. The noise parameters of the 1-mW CSP were fitted on that curve, using (1)–(5) and the theory exposed is the in [10] and [13]. They are reported in Table III where is the input referred noise-voltage flicker noise coefficient, spectral density, is the series and parallel noise corner time is the equivalent stray capacitance seen at the constant, CSP input node (including the 1-pF charge injection capacitor and 0.3-pF feedback capacitor). Fig. 7 is the ENC curve for the 10-pF capacitor, used to fit the noise parameters of the 5-mW device, also presented in Table III. C. Timing Measurements The coincidence timing resolution of both CSPs was measured using standard nuclear spectroscopy techniques [14]. The electronic timing resolution was measured using a Phillips Scientific 704 leading edge discriminator (LED) connected directly at the output of the CSP without filter, providing the stop signal to a time digitizer Phillips Scientific 7186. A LeCroy 9210 pulser, providing a signal amplitude corresponding to 511 keV in an APD/LSO detector, was connected Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. 1982 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 5, OCTOBER 2004 Fig. 6. ENC data fit with APD for the 1-mW CSP. A minimum ENC of 1117 electrons-rms at 100 ns was measured. Fig. 8. Electronic timing resolution for the 1-mW CSP. The Hamamatsu APD/LSO detector was connected at the input and biased at a gain of 50. TABLE III EXTRACTED NOISE PARAMETERS Fig. 9. Electronic timing resolution for the 5-mW CSP. The Hamamatsu APD/LSO detector was connected at the input and biased at a gain of 50. Fig. 7. ENC data fit with 10 pF for the 5-mW CSP. at the test input of the CSP. The trigger of the LeCroy pulser, through another channel of the LED, was used as the start signal in the time digitizer. The Hamamatsu APD/LSO detector was connected at the input and biased at a gain of 50 for the evaluation of both devices. The 1-mW CSP has an electronic timing resolution of 0.56-ns FWHM, as reported in Fig. 8. The 5-mW CSP achieved a slightly better electronic timing resolution of 0.4-ns FWHM, as presented in Fig. 9. For the detector time resolution, the CSPs were connected to a CR-RC Ortec 474 Timing Filter Amplifier and an Ortec 934 constant fraction discriminator (CFD). The 1-mW CSP measurement was performed with the Hamamatsu APD/LSO detector in coincidence with a fast PMT/BaF detector, using a Na source. For the measurement with the 5-mW CSP, a PerkinElmer APD/LSO detector in coincidence with a PMT/Plastic (Nuclear Enterprise NE-102A) detector and an F source were used. Coincidence timing resolutions of 2.49-ns FWHM and 1.56-ns FWHM were obtained for the 1and 5-mW devices respectively, as shown in Figs. 10 and 11. D. Energy Resolution The energy resolution of the 1-mW CSP was evaluated with the Hamamatsu detector connected to the input. The output of the preamplifier was fed to a CR-RC Ortec 579 Timing Filter Amplifier, with 50-ns integration and derivation time constants. The best energy resolution was measured with the APD biased at 380 V, which led to 17.4% FWHM, as presented in Fig. 12. For the 5-mW device connected to the PerkinElmer detector, using a Ortec shaper 452 with 250-ns integration and derivation time constants, an optimum energy resolution of 11.7% which was fairly insensitive to the APD bias was measured, as reported in Fig. 13. Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. PRATTE et al.: DESIGN AND PERFORMANCE OF 0.18- m CMOS CHARGE PREAMPLIFIERS FOR APD-BASED PET SCANNERS Fig. 10. Timing resolution spectrum for the 1-mW CSP. 1983 Fig. 13. Energy spectrum for the 5-mW CSP showing the electronic noise contribution measured with a pulser. expected. This can be explained by added parasitic capacitance and process variation. The 1- and 5-mW devices are linear over the designed dynamic range, suitable for the Hamamatsu detector. However, in order to use the 5-mW device with the PerkinElmer APD/LSO detector, which has a greater light output ( 10,000 electronhole pairs per MeV created in the APD before gain) and is used at a gain of 100, a larger dynamic range is required for the 5-mW CSP. Hence the gain of the 5-mW CSP shall have to be lowered in order to allow a greater maximum input charge. The measured rise times fit accordingly with the simulations. The 5-mW device shows a faster rise time than the 1-mW, due to its higher transconductance. Fig. 11. Timing resolution spectrum for the 5-mW CSP. The time calibration was obtained by adding an 8-ns delay to the stop channel. Fig. 12. Energy spectrum for the 1-mW CSP connected to the Hamamatsu APD/LSO detector biased at 380 V. IV. DISCUSSION A. Electronics The two devices show excellent linearity. The 2.7 mV/fC measured gain of the two prototypes is lower than the desired 3.3 mV/fC from the initial design specification, which means that the actual feedback capacitors are bigger than B. Equivalent Noise Charge As expected, the 5-mW CSP has a lower ENC as a function of the input capacitance compared to the 1-mW device. We notice that the measured minimum ENC is higher than expected and shifted to a higher peaking time. For instance, the minimum ENC of the 1-mW CSP measured with the detector biased at a gain of 50 is 1117 electrons-RMS at 100 ns, instead of 726 electrons-RMS at 70 ns as designed. This can be explained by an underestimation of the flicker noise coefficient during the design, which resulted in a higher contribution from the noise, as seen by the flatness of the curves. The main noise is the input transistor M1, operating besource of tween weak and moderate inversion for both CSPs. Tsividis noise et al. [15] suggest that the provided manufacturer models are aimed at digital design and usually do not appropriately characterize MOSFET devices operating in weak and moderate inversion regions. Test transistors were included on chip, and preliminary characterization of those confirms our assumption. In order to get the best noise performance, a review of the CSPs design will be necessary, using the appropriate noise conflicker noise coefficient. Considering the high tribution of the N-channel MOSFET input device, the use of a P-channel MOSFET looks appealing in theory. However, the requirement for a current return path between the input MOSFET and the detector high voltage power supply imposes significant design constraints [16]. Recent studies of deep sub-micron Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. 1984 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 51, NO. 5, OCTOBER 2004 TABLE IV FRONT-END FOR PET IMAGING CMOS technology indicate that the flicker noise contribution drops as the transistor gate length is slightly increased above the minimum dimension [17]. This aspect is currently under study for the CMOS 0.18- m technology and, depending on the conclusion, the adoption of a longer gate length could be considered. C. Timing Resolution The electronic timing resolution of the 5-mW CSP is slightly better than the one of the 1-mW device. This is consistent with the theory, as the 5-mW device benefits from a lower noise contribution and a faster rise time due to its higher transconductance, leading to a smaller time jitter in the LED. The coincidence timing resolution of the prototypes is in accordance with the previous results, where the 5-mW device presents better performance. The difference between the achieved coincidence timing resolutions is somehow bigger than the difference between the electronic timing resolutions because the PerkinElmer detector has a larger output signal and better energy resolution than the Hamamatsu detector array due to their different characteristics (ex: scintillator size, coupling, APD gain operating point). In both cases, the achieved coincidence timing resolutions using a CFD allows the use of a small coincidence timing window, which is particularly important to limit the number of random coincidences in PET imaging system. D. Energy Resolution In scintillator/APD-based detectors, the contribution of the CSP to the deterioration of the energy resolution is negligible, as it is mainly the stochastic nature of the scintillation process and the APD excess noise which determine the energy resolution [18]. The results obtained with the Hamamatsu-based detector module on the 1-mW CSP and with the PerkinElmer LSO/APD detector modules on the 5-mW CSP concur with previous measurements performed in our laboratory and elsewhere, and confirms the negligible contribution of the CSP to the energy resolution. E. Charge Sensitive Preamplifiers for PET Table IV compares the key performance parameters of CMOS charge preamplifiers previously developed for PET imaging with the 1- and 5-mW CSP investigated in this paper. As expected, the 5-mW CSP achieves better input referred noise-voltage and timing resolution compared to the 1-mW CSP, due to the higher transconductance of the input transistor M1. Still, the 1-mW device reaches nearly similar performance at a fraction of the power required by previous CSP designs. The 1-mW CSP fulfills the stringent power requirements of the RatCAP where a total power budget of only 3.9 mW per channel is allowed [3], [4]. V. CONCLUSIONS Two charge sensitive preamplifiers were designed in a 0.18- m CMOS technology and optimized for LSO/APD-based detectors for PET imaging. The performance of the low-power and high-performance devices, which is comparable to that obtained with previously developed charge sensitive preamplifiers, clearly indicates their suitability for PET front-end systems. Finally, it has been demonstrated that the CMOS 0.18- m technology is suitable for the realization of low power, low noise analog front-end electronics. ACKNOWLEDGMENT The authors would like to thank V. Radeka, J. Triolo, S. Rescia, R. Machnowski, C. Woody, A. Kandasamy, and the RatCAP group from BNL and D. Rouleau, R. Bernier, and L. Hubert from Université de Sherbrooke for support, advice, and their precious time, and J. Sondeen who provided excellent Magic tech-file support. Finally, the authors would like to acknowledge the assistance of the Canadian Microelectronics Corporation for providing tools and technical support for the chip fabrication. This work is part of a joint collaboration between Brookhaven National Laboratory and Université de Sherbrooke in Canada. REFERENCES [1] M. Manghisoni, L. Rattia, V. Reb, and V. Speziali, “Deep submicron CMOS transistors for low-noise front-end systems,” in 2001 Nuclear Science Symp. Medical Imaging Conf. Record CD, 2001. [2] P. O’Connor and G. De Geronimo, “Prospects for charge sensitive amplifiers in scaled CMOS,” Nucl. Instrum. Methods, vol. A480, no. 2, pp. 713–726, 2002. [3] P. Vaska, D. J. Schlyer, C. L. Woody, S. P. Stoll, V. Radeka, and N. Volkow, “Imaging the unanesthetized rat brain with PET: A feasibility study,” in 2001 Nuclear Science Symp. Medical Imaging Conf. Record, vol. 3, pp. 1569–1571. Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply. PRATTE et al.: DESIGN AND PERFORMANCE OF 0.18- m CMOS CHARGE PREAMPLIFIERS FOR APD-BASED PET SCANNERS [4] J.-F. Pratte, G. De Geronimo, S. Junnarkar, P. O’Connor, B. Yu, S. Robert, V. Radeka, C. Woody, S. Stoll, P. Vaska, R. Lecomte, and R. Fontaine, “Front-end electronics for the RatCAP mobile animal PET scanner,” in 2003 Nuclear Science Symposium and Medical Imaging Conference Record CD, 2003. [5] R. Fontaine, J.-B. Michaud, F. Bélanger, J.-D. Leroux, J.-F. Pratte, S. Robert, and R. Lecomte, “Architecture of a dual-modality, high-resolution, fully digital positron emission tomography/computed tomography (PET/CT) scanner for small animal imaging,” in 2003 IEEE/NPSS Real Time Conference CD, 2003. [6] P. Vaska, C. L. Woody, D. J. Schlyer, S. Shokouhi, S. P. Stoll, J.-F. Pratte, P. O’Connor, S. S. Junnarkar, S. Rescia, B. Yu, M. Purschke, A. Kandasamy, A. Villanueva, A. Kriplani, V. Radeka, N. Volkow, R. Lecomte, and R. Fontaine, “RatCAP: Miniaturized head-mounted PET for conscious rodent brain imaging,” in 2003 Nuclear Science Symp. Medical Imaging Conf. Record CD, 2003. [7] A. Kriplani, S. P. Stoll, D. J. Schlyer, S. Shokouhi, P. Vaska, A. Villanueva Jr, and C. L. Woody, “Light output measurements of LSO single crystals and 4 8 arrays: Comparison of experiment with Monte Carlo simulations,” in 2003 Nuclear Science Symp. Medical Imaging Conf. Record CD, 2003. [8] G. De Geronimo and P. O’Connor, “A CMOS detector leakage current self-adaptable continuous reset system: Theoretical analysis,” Nucl. Instrum. Methods, vol. A421, pp. 322–333, 1998. [9] C. C. Enz, F. Krummenacher, and E. A. Vittoz, “An analytical MOS transistor model valid in all regions of operation and dedicated to lowvoltage and low-current applications,” Anal. Integ. Circuits Signal Processing, vol. 8, pp. 83–114, 1995. [10] V. Radeka, “Low-noise techniques in detectors,” Ann. Rev. Nucl. Part. Sci., vol. 38, pp. 217–277, 1988. [11] S. Ohkawa, M. Yoshizawa, and K. Husimi, “Direct synthesis of the Gaussian filter for nuclear pulse amplifiers,” Nucl. Instrum. Methods, vol. 138, pp. 85–92, 1976. [12] A. W. Lightstone, R. J. McIntyre, R. Lecomte, and D. Schmitt, “A bismuth germanate-avalanche photodiode module designed for use in high resolution positron emission tomography,” IEEE Trans. Nucl. Sci., vol. NS-33, pp. 456–459, Feb. 1986. 2 1985 [13] R. Lecomte, C. M. Pepin, M. D. Lepage, J.-F. Pratte, and D. M. Binkley, “Performance analysis of phoswich/APD detectors and low noise CMOS preamplifiers for high resolution PET systems,” IEEE Trans. Nucl. Sci., vol. 48, pp. 650–655, Feb. 2001. [14] C. Carrier and R. Lecomte, “Timing performance of scintillators read out by silicon avalanche photodiodes,” Nucl. Instrum. Methods, vol. A299, pp. 115–118, 1990. [15] Y. P. Tsividis and K. Suyama, “MOSFET modeling for analog circuit CAD: Problems and prospects,” IEEE J. Solid-State Circuits, vol. 29, pp. 210–216, Jan. 1994. [16] P. O’Connor, J.-F. Pratte, and G. D. Geronimo, “Low-noise charge amplifiers in submicron CMOS,” presented at the 5th Int. Workshop on Front-End Electronics, Snowmass, CO, 2003. [17] P. F. Manfredi and V. Re, “Trends in the design of front-end systems for room temperature solid state detectors,” in 2003 Nuclear Science Symp. Medical Imaging Conf. Record CD, 2003. [18] R. Lecomte, C. Pepin, D. Rouleau, A. Saoudi, M. S. Andreaco, M. Casey, R. Nutt, H. Dautet, and P. P. Webb, “Investigation of GSO, LSO and YSO scintillators using reverse avalanche photodiodes,” IEEE Trans. Nucl. Sci., vol. 45, pp. 478–482, Feb. 1998. [19] D. Schmitt, R. Lecomte, M. Lapointe, C. Martel, C. Carrier, B. Karuta, and F. Duval, “Ultra-low noise charge sensitive preamplifier for scintillation detection with avalanche photodiodes in PET applications,” IEEE Trans. Nucl. Sci., vol. NS-34, pp. 91–96, Jan. 1987. [20] D. Binkley, J. Rochelle, M. Paulus, and M. Casey, “A low-noise, wideband, integrated CMOS transimpedance preamplifier for photodiode applications,” IEEE Trans. Nucl. Sci., vol. 39, pp. 747–752, Aug. 1992. [21] M. Paulus, J. Rochelle, M. Andreaco, and D. Binkley, “Low-noise, wideband CMOS charge sensitive preamplifier for use with APD/LSO PET detectors,” IEEE Trans. Nucl. Sci., vol. 43, pp. 1666–1671, June 1996. [22] D. M. Binkley, B. S. Puckett, M. E. Casey, R. Lecomte, and A. Saoudi, “A power efficient, low-noise, wideband, integrated CMOS preamplifier for LSO/APD PET systems,” IEEE Trans. Nucl. Sci., vol. 47, pp. 810–817, May 2000. Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on August 19, 2009 at 16:11 from IEEE Xplore. Restrictions apply.