Transport properties and performance of CdZnTe strip detectors 0. Tousignant, L.A. Hamel, J.F. Courville, Groupe de recherche en physique et technologie des couches minces (GCM), University of Montreal, Montreal, H3C 3J7, Canada. J.R. Macri, M. Mayer, M.L. McConnell, J.M. Ryan Space Science Center, University of New Hampshire, Durham, NH 03824, USA. Abst~uct We report on recent tests and computer s of a submillimeter pitch CdZnTe strip detector under study as a prototype imaging spectrometer. This paper presents new measurements and new analyses of previous measurements of the CdZnTe transport properties for this prototype and resolves previously reported quantitative discrepancies. Measurements of charge signals produced by arays are used to determine the transport properties. These are then used in the model to predict detection efficiencies and spectroscopic performance for yrays that are compared with the data. Goo I. INTRODUCTION energy resolution, spectral uniformity and submillimeter spatial resolution have been demonstrated with CdZnTe strip detectors [I, 2, 4, 6, 7, 81. Recent laser scans accross our strip detector showed a potential for position resolution much smaller than the strip pitch, possibly as good as 10 pm [8]. Strip detectors, however, must rely on efficient transport of both electrons and holes for coincident anode and cathode signal measurement. The poor and widely varying hole transport properties [l, 2, 7,9, 10,171 of currently available materials result in nonuniform response to photons interacting at different regions of the detector [2]. This can affect detection efficencies and must be considered when designing strip detector systems for specific applications. An appropriate model for the charge transport and signal generation processes in CdZnTe detectors is an essential tool for understanding the detector performance characteristics and can be used to help specify optimum detector geometries and signal processing electronics for each application [2, 3, 7, 8, 9, 10, 12, 13, 14, 151. Our model is intended to be an endtoend tool for simulating all detection and measurement processes from the radiation interaction down to the electronic chain. Use of such a model will help minimize the number of expensive and consuming hardware prototypes required in any development effort. Achieving agreement between s and measurements is a significant developmental milestone and a major goal of our present work. This paper presents measurements and s from a CdZnTe strip detector. Orthogonal anode and cathode strips have the same geometry: strip pitch is 375 pm, with 225 pm wide gold contact strips and 150 pm gaps (spaces) between the electrodes. The detector is 1.5 mm thick and is operated at 200 Volts. In a first step, single event aray signals at the preamplifier output of several anode and cathode strips are simultaneously recorded by digital storage oscilloscopes. Analysis of these provides a measurement of the electron and hole transport parameters. These parameters are then used to model the response of the detector to yrays to predict pulse height spectra which are compared to yray measurements. The aray measurements are identical or similar to those recently presented 181 but this new analysis with an improved model yields better transport parameters, especially for holes for which deep trapping, in addition to multiple trappingdetrapping, is now considered. 0780342585/98/$10.00 0 1998 IEEE 556
11. CHARGE TRANSPORT PROPERTIES A. Charge transport model 0"" model for charge transport and signal generation has already been described, along with comparisons to experimental signals [14, 15, 7,2,8]. The model successfully reproduced signals from CdTe detectors [15, 161. But in CdZnTe, a real quantitative agreement was still lacking, due to the poor hole transport properties of CZT [8]. Multiple trappingdetrapping on a single trap state, though adequate for CdTe and for electrons in CZT, does not adequately describe hole transport in CZT. To resolve this difficulty, two trap states are now used to model holes in CZT, one with trapping and detrapping, the second one for deep trapping only. The model thus now includes 3 parameters for electrons (mobility, trapping coefficient, detrapping coefficient) and 4 parameters for holes (mobility, shallow trapping coefficient, detrapping coefficient from shallow trap, deep trapping coefficient). B. Signals induced by alphas Electron transport is studied with arays incident on the cathode side by simultaneously recording electron transit signals on several anodes and cathodes. Events produced by alphas incident in the center of a particular cathode strip X15 are selected by requiring a large signal on X15 and equally small transient (zero net charge) signals on both neighbouring strips X14 and X16 (Figure la). Other electron signals due to alphas centered on anode Y8 are selected in a similar fashion (Figure lb). The signal for the anode Y8 is used to extract the electron transport parameters. These are presented in table 1. With these parameters, an excellent agreement is obtained with the data, not only for the signal on anode Y8 but, at the same tim.e, for signals measured on other electrodes for the same event. Note that the simulated signals also include an effective integration of 20 ns to account for the electronics rise. This limited rise being not much shorter than the features in the measured signals, these trapping and detrapping constants may not be very accurate. Nevertheless, the agreement with the model is satisfactory, at least on the scale of interest for radiation detection. These parameters yield an effective electron drift mobility, peff = pt/(tt + Td) of about 690 cm2v'sl, where.rt and Td are the trapping and detrapping s respectively. z 6 0.8 U B : CI o 0.6. Y8... experimental data 8.n c) E 0.4 cr W. 0......".."..'.. Z... k 5 z... Y7 0 P:... 2 ' h 0.2 a 3 0.. ~ ' ' '0:l' ' ' '012' ' ' '013' ' ' '0!4' ' ' '0!5' X14...... b) (ps) experimental data... ' ' ' 'Ofl' ' ' '012' ' ' '013' ' ' '0!4' ' ' '015' @) Fig. 1. Measured and simulated electron signals on two neighboring electrodes due to alphas incident on the cathode plane. a) signals on anodes Y8 and Y7 for alphas incident on anode Y8. b) signals on cathodes X15 and X14 for alphas incident on cathode X14. Hole transit signals produced by alphas were also measured. Figure 2 shows such an anode signal for the alpha interacting in its center. It is smaller than corresponding electron signals. 557
~ TABLE I TRANSPORT PARAMETERS EXTRACTED FROM THE (YRAY MEASUREMENTS. Electrons Mobility 1 1000 1 ( cm2vl s' ) Trapping Detrapping Deep trapping Effective mobility ( cm2vl s' ) pr product (cm2v') n c.* c.' E 55ns Holes 4.2 ps 25ns 2 P= 4 PS 690 4.7 I 2.8 x 105 I 0.61...,,..,...,..,, I...................... 0.3 e. 0.2 M m % 0.1 z x a.... experimental data I,.I,.I.I.I.I.IIIII.I..II...I,.,,... 0 2 4 6 8 10 12 14 16 18 M tp) Fig. 2. Measured and simulated hole signal induced on an anode by alphas incident on its center. In this geometry, holes receding from the anode induce 55% of the charge by moving only 15% of the detector thickness, indicating that most holes are trapped before reaching the cathode. No detrapping is seen during the measurement which can only be accounted for by deep trapping. The excellent agreement between the measured and simulated signals indicates that the main features of hole transport are well reproduced with i the parameters of table 1. This is a clear improvement over the previous model that involved only one trap state for holes [8]. The effective mobility of 4.7 cm2v'sl is much smaller than the usually reported hole mobility for CdZnTe. But when using a mobility of 50 cm2v's' and a life of 5 ps [9], the collection efficiency of our CdZnTe detector was overestimated [2]. Also, the present parameters yield a pr product of 2.8~10~~ cm2v' which is consistent with other measurements where values of a few s cm2v' were reported [16]. 111. 7RAY DETECTION EFFICIENCY AND SPECTROSCOPY SING the transport parameters extracted U from the aray signals, the detector response to 241Am and 57C0 yrays was calculated as a function of the position of interaction in the detector. The modeling process is detailed in reference [2]. A GEANT provides energy deposition spectra as a function of depth for each radioactive source situated at the cathode side. These are folded with the anode and cathode signals calculated by the model over a grid of 150 points in the detector. The includes the transport of both holes and electrons produced in the interaction, the weighting fields of every electrode involved, and the response function of the electronics used during the Tray measurements. Figure 3 shows relative anode (fig. 3a) and cathode (fig. 3b) signal amplitudes of the fast signal outputs used for the trigger, calculated with the parameters of table l. Figures 3c) and 3d) present the same signal amplitudes, but calculated with a hole mobility of 50 cm2v's'. The 15 kev threshold on all fast anode and cathode channels corresponds to an amplitude of 0.05 for the 241Am photons. A coincidence between the anode and cathode discriminators is required. With the parameters given in table 1, the calculated efficiency for 241Am, situated on the cathode side, is 80% compared with a measured efficiency of 89%. Since those photons are absorbed in a short distance from the cathode, this value is mainly determined by the electron collection efficiency. Such a good agreement is an indication 558
that the limited trapping associated with electrons is quite realistic. The coincidence detection efficiency (for an absorbed 7ray) is measured to be 69% for 57C0 compared to a calculated value of 62%. In this case, holes play a more important role in the signal generation. The efficiencies (measured and calculated) are lower for 57C0 than for 241Am because, though more charges are generated, these are created deeper in the detector, where the fast cathode signals are more likely to be below the threshold. The agreement between measured and calculated efficiencies is better than previously reported [2] with /.hh = 15 cm2vls'. that electron transport is well reproduced by our s. Figure 4c) shows the spectrum for cathode X15 while figure 4d) is the spectrum for the sum of amplitudes on X15 and X16. Cathode spectra are far from being as good as for anodes. Summing neighbouring cathodes gives some improvement. Here also the agreement is satisfactory, indicating that our parameters for holes are reasonable, though they could still be improved. Figures 4c) and d) also present cathode spectra simulated with ph = 50 cm2v'sl. These clearly do not fit the data. Fast Anode Signals Fast Anode Signals Fast Cathode Signals Fast Cathode Simals 25 25 20 20 15 15 10 10 5 5 0 0 0 50 100 150 0 50 100 150 energy (kev) energy (kev) 25 5 2 Q 07 20 20 &06 Do 0.6 i. 05 i. os 15 15 04 2 0.4 5 03 U 03 v 01 P 0.z 10 10 8 ni 8 01 a n a n P 5 5 c m 3 0 0 energy (kev) energy (kev) Fig. 3. Amplitude of the fast anode and cathode signals calculated : a) and b) with the parameters of table 1; c) and d) with an increased hole mobility of 50 cm2vls'. Figure 4 compares calculated (shaded histograms) and measured (dark lines) pulse height distributions for 57C0 photons incident on the cathode side. The source being collimated, a gaussian illumination distribution is considered with t~ = 30pm and its centroid near the edge of anode Y8 and the edge of cathode X15 (as deduced from the 7ray scans). Figures 4a and b show the spectra for anodes Y8 and Y7, respectively. Y8 being closer to the illuminated spot, sees most of the induced charge. The agreement is excellent showing Fig. 4. Measured (dark lines) and simulated (shaded histograms) pulse height distributions for 57C0 yrays incident on the cathode side close to anode Y8 and cathode X15 : a) anode Y8; b) anode Y7; c) cathode X15; d) amplitude sum from cathodes X14 and X15. On c) and d), the light histogram show simulated spectra with ph = 50 cm' Vlsl. IV. CONCLUSIONS RANSPORT parameters extracted from a T ray measurements have been used to predict counting efficiencies and anode and cathode pulse height distributions for our CdZnTe strip detector. The agreement between data and the is found to be very good. It should be noted that 559
the parameters of table 1 were extracted from a single a signal and that those signals exhibit large amplitude fluctuations. Also, there are significant fluctuations with the position over the detector surface. The present parameters must then be considered as representative of this specific spot of our detector. Nevertheless, the agreement is encouraging and shows that it is possible to provide realistic s that can be used to predict detector performance without the need to build too many prototypes. The s and measurements confirm that the performance of CdZnTe strip detectors is quite dependent on holes transport properties. The fact that these are quite poor sets limitations of the usability of strip detectors for some applications. Clearly, electrononly devices that could provide at the same good energy resolution and good 2D position resolutions are needed [ 171. ACKNOWLEDGEMENTS This work is supported by NASA s High Energy Astrophysics Gamma Ray Astronomy Research and Analysis program and by the Natural Sciences and Engineering Research Council of Canada. REFERENCES J.M. Ryan et al., Large Area SubMillimeter Resolution CdZnTe Strip Detector for Astronomy, SPIE Cod. Proc. San Diego, 2518, p. 292, 1995. M. Mayer et al., Performance and Simulation of CdZnTe Strip Detectors as Submillimeter Resolution Imaging Gamma Radiation Spectrometers, IEEE Trans. Nucl. Sci. 44, p. 922, 1997. H.B. Barber et al., CdZnTe arrays for nuclear medicine imaging, SPIE Conf. Proc. Denver CO, 2859, p. 26, 1996. J.R. Macri et al., Progress in the development of large area submillimeter resolution CdZnTe strip detectors, SPIE Conf. Proc. Denver CO, 2859, p. 29, 1996. J.L. Matteson et al., CdZnTe strip detectors for highenergy Xray astronomy, SPIE Conf. Proc. Denver CO, 2859, p. 58, 1996. P. Kurczynski et al., CZT strip detectors for imaging and spectroscopy: collimated beam and ASIC readout experiments, IEEE Nuclear Science Symposium Conference Record, Anaheim CA, p. 671, 1996. L.A. Hamel et al., Signal Generation in CZT Strip Detectors, IEEE Trans. Nucl. Sci., 43, p. 1422, 1996. 0. Tousignant et al., Progress in the study of CdZnTe strip detectors, SPIE Cod. Proc. San Diego CA, 3115, 1997. P.N. Luke, Unipolar charge sensing with coplanar electrodes Applications to semiconductor detectors, IEEE aans. Nucl. Sci., 42, p. 207, 1995. Z. He et al., Position sensitive single carrier CdZnTe detectors, IEEE Nuclear Science Symposium Conference Record, Anaheim CA, p. 331, 1996. F. P. Doty, Carrier mobilities and lifes in CdTe and CdZnTe, in Properties of Narrow Gap Cadmiumbased Compounds, P. Capper ed., in Electronic Materials Information Service Data Reviews Series, 10, p. 540, 1994. J.A. Heanue, J.K. Brown and B.H. Hasegawa, TWOdimensional modelling of compound semiconductor strip detectors, IEEE Nuclear Science Symposium Conference Record, Anaheim CA, p. 548, 1996. J.D. Eskin et al., The effect of pixel geometry on spatial and spectral resolution in a CdZnTe imaging array, IEEE Nuclear Science Symposium Conference Record, San Francisco CA, p. 544, 1995. L.A. Hamel, Signal induced in semiconductor detectors with a linear field profile in the presence of trapping and detrapping, submitted to Nucl. Instr. and Meth. A. L.A. Hamel and S. Paquet, Charge transport and signal generation in CdTe pixel detectors, Nucl. Instr. and Meth. A, 280, p. 238, 1996. R.B. James, T.E. Schlesinger, J.C Lund and M. Schieber, Cdl,Zn,Te Spectrometers for Gamma and XRays Applications in Semiconductors for Room Temperatue Nuclear Detector Applications, T.E. Schlesinger, R.B. James editors, Semiconductors and Semimetals, vol. 43, Academic Press, San Diego, p. 344, 1995. L.A. Hamel et al., An imaging CdZnTe strip detector with orthogonal anodes, to be presented at the MRS Fall Meeting, Boston MA, December 1997., I, 560