Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 Accounting for incomplete charge collection in Monte Carlo simulations of the efficiency of well-type Ge-detectors F. Herna! ndez*, F. El-Daoushy Department of Physics, Uppsala University, Box 530, Uppsala 75121, Sweden Received 30 July 2002; received in revised form 22 November 2002; accepted 25 November 2002 Abstract The influence of the process of incomplete charge collection on the performance properties of HPGe well-type detectors is studied. A simplified sub-routine, that takes into account the total collection time of charges within the Gecrystal, has been developed and coupled to the Monte Carlo photopeak efficiency. Only the manufacturer’s data of the detector’s geometry and the detector’s operating voltage was used in the sub-routine. The corrected photopeak efficiency simulations, as based on this new sub-routine, showed much better agreement with the experimental data, especially for energies above 40 keV. The ratio between the experimental and simulated efficiency at 1460 keV were, for example, improved from 0.73 to 0.96 after performing these corrections on our largest well-type Ge-detector. r 2002 Elsevier Science B.V. All rights reserved. PACS: 29 Keywords: Monte carlo simulations; HPGe detectors; Efficiency; Incomplete charge collection 1. Introduction Environmental studies often involve the measurement of very small amounts of material. This is partly due to sampling restrictions and partly because the sampled material is generally shared by multiple users that require different type of measurements. Well-type Ge-detectors, with their high efficiency, is the type of detectors most often used for the measurement of small samples (e.g. Refs. [1–2]). *Corresponding author. Tel.: +46 -(0 1 8)-4713564; fax: +46-(0 1 8)-4713524. E-mail address: francisco.hernandez@fysik.uu.se (F. Hern!andez). The calculation of absolute activities of natural and artificial radionuclides by gamma spectrometry requires reliable and accurate determination of the detector’s photopeak efficiencies. Although the experimental determination of the efficiency of Ge-detectors is still the most accurate approach [1,3], this often requires extensive and delicate laboratory work, both in term of source preparations and measuring time. A complete evaluation of the photopeak efficiency of Ge detectors using only Monte Carlo simulations has been evolving since the seventies (e.g. Refs. [4–6]). Different approaches and codes can be found in literature [2,5,6–15] where various sample-to-detector geometries and photon energy ranges are treated. However, efficiency results are 0168-9002/03/$ - see front matter r 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0168-9002(02)02080-6 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 341 mostly given for planar and coaxial type detectors since these are the most commonly used types of detectors. Scarce data are available for well-type Ge-detectors [2,11,15]. The Monte Carlo code used to simulate the efficiency of the different Ge-detectors, in this paper, was the well-established MCNP Monte Carlo code [16]. Simulations of the performance of the planar-type Ge-detectors, available in our laboratory, showed that this code yields good agreement between the simulated and experimental efficiency [17]. However, this code failed to reproduce the experimental data of two well-type Ge-detectors available also in our laboratory. The largest differences between the experimental and the simulated data were observed for the largest of the two well-type Ge-detectors (Fig. 1). The discrepancies commonly existing in literature between experimental and simulated data of the efficiency calibration of Ge-detectors have been attributed to the uncertain knowledge of the geometric characteristics of the detectors. These discrepancies have been dealt with by manipulating the dimensions of the detector’s crystal and the contact layers [e.g. Refs [2,14]). In this paper, we show that incomplete charge collection is a more likely cause of the discrepancies in the simulation of the efficiency of well-type Ge-detectors than inaccurate detector specifications supplied by manufactures. Our approach consists in two steps: we use a collimated source to demonstrate that the efficiency in different sections of the well-type crystal is not what one would expect from simple geometric considerations, and we show that Monte Carlo results improve dramatically when incomplete charge collection is accounted for with the method we developed. 2. Method 2.1. Measurements Four ultra low-level gamma spectrometry systems consisting of seven different detectors are available at our laboratory [1,17]. Two of these detectors, that are discussed here, are HPGe well- Fig. 1. Description of the physical dimensions, contact layers and operating voltages of the two well-type Ge-detectors as well as the physical dimensions of the samples. (a) The smallest welltype Ge-detector. (b) The largest well-type Ge-detector. The Figures are not drawn to scale. 342 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 type detectors. Fig. 1(a) and (b) specify the physical dimensions and operating voltages of these two detectors. A total of five different radioactive standards were used for the preparation of numerous calibration sources for the experimental calibration of the two well-type Gedetectors. These radioactive standards included mixed sources containig the well-known single gamma ray emitters, single Pb-210, single Am-241, single K-40 and Ra-226 + daughters. Radionuclides such as Ce-139, Y-88 and Co-60 were discarded from the final data-sets because of coincidence summing. Multiple runs were performed for each sample-to-detector geometry to reduce the random uncertainties related to the preparation of the different calibration sources. Further description of the Ge-detection systems and the experimental calibration of the efficiencies is given elsewhere [1,15,18]. A collimated spherical source with radioactive material at its centre was used to measure the efficiency response of the largest well-type Gedetector at two different energies, 59.5 and 186 keV. This source allowed scanning the active parts of the Ge-crystal by moving inside the well cavity of the detector. The source could be placed at different positions on the z-axis, which is passing through the centre of the well cavity. It could, also, be rotated around the z-axis [0–2p] and around the radial-axis [0–p] of the horizontal plane which is perpendicular to the z-axis. These movements allowed directing a collimated beam, from the pin hole of the source, into different active zones within the Ge-crystal. Ultra-pure lead (1 mBq/g), electrolytic copper and styrene were used for shielding and sealing the source. The spherical source had an inner and outer radius of 1.5 and 6.5 mm, respectively. The radioactive material was concentrated in a low absorption spherical solid matrix and placed in the inner central part of the source. The pin hole opening of the source provided the collimated photons with a radius of 1 mm. Several plastic frames of different lengths allowed holding the source in stable positions at various heights inside the well cavity. The photon emission rate of the collimated source was determined for the two different photon energies using three different HPGe planar detectors. Description of these detectors is given elsewhere [15]. The source was always placed on the symmetry axis of the planar-type detectors with the collimated beam perpendicular to the detectors’ windows. No measurements were performed at other angles of photon incidence. Measurements at various distances ranging from 8 to 20 mm, from the hole of the source to the surface of two of the planar Ge-crystals, were performed to study the influence of the source-todetector distance on the measured count rate. Calculation of the source’s expected count rate in the well-type detectors involved minor corrections to consider the absorption of photons in the Al windows of these detectors. Two of the planartype Ge-detectors available in our laboratory constitute a sandwich Ge-system. This system allowed calculating the percentage of the 186 keV photons that escaped isotropically the shield of the source due to its small dimensions and the relative long range of these photons. All the subsequent measurements of the 186 keV photons count rate were corrected for this small percentage (o 2%). 2.2. Simulations The simulations of the efficiency were done, using MCNP [16], for both well-type Ge-detectors, and for energies ranging from 20 to 1460 keV. Only the dimensions given by the detector’s manufacturer were used in the simulation. The simulation results comprise the average of several runs with variable numbers of initial photons ranging from 105 to 1010. The largest number of the initial photons was used for the simulations involving photon energies corresponding to the lowest efficiency in order to improve the statistical uncertainty. Simulations of the collimated source were also made for the largest well-type Gedetector to allow comparisons with the experimental data. A simplified diagram of the collimated source-to-detector geometry is shown in Fig. 2. These simulations were done for different beam angles around the radial-axis [0–p] and different distances from the bottom of the well cavity. ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 343 Fig. 2. Simplified geometry used in the simulation of the efficiency of the collimated source. In this case, the source is placed on the symmetry axis at a distance of 47 mm from the base of the detector. Theta (y) represents the rotation angle of the collimated beam around a radial axis situated in the horizontal plane. This horizontal plane is perpendicular to the symmetry axis z. 2.3. Modelling the charge collection process In order to model the charge collection process, it was necessary to calculate the electric field inside the active regions of the different detectors. Analytical models exist for the calculation of the field inside planar and true-coaxial Ge-detectors [19]. However, the complicated geometry of the well-type Ge-detector had to be solved numerically. The FlexPDE software [20], a programme designed for solving differential equations for given boundary conditions, was used to perform the calculations of the electric field for the planar, true-coaxial and well-type Ge-detectors. The results were compared to the available analytical solutions. The good agreement of the magnitude of the electric field between the analytical and simulated data indicated the reliability of the used software. Fig. 3 shows a simplified version of the cross-section of the calculated electric field for the largest well-type detector. A more detailed and full colour version of this figure is available elsewhere [17, http://www.fysik.uu.se/isotopgeo/ research/compmaterial/]. The boundary conditions used in the calculations of the field corresponded to the operating voltages given by the detectors’ specification sheet. The voltage drop in the detector’s load resistor, of the bias circuit, was neglected since it was found to be very small. The variations in the uniformity of the electric field, i.e. the strength and the direction, are more significant in the well-type Ge-detector compared to the planar-type Ge-detector. Strong variations in the strength and direction of the field were observed at the bottom corners of the well cavity, and the bottom corners of the Ge-crystal. The lowest 344 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 Fig. 3. Cross-section of the largest well-type Ge-detector with a simplified version of the electric field distribution within the active detection region. The arrows indicate the approximate field direction. values of the field are located at the bottom corners of the Ge-crystal of the well-type detector. 3D maps of the drift velocities of charges due to the electric field for various detectors were calculated using the empirical expression [21]: mEðx; y; zÞ Vdrift ðx; y; zÞ ¼ h b i1=b 1 þ Eðx; y; z=e ð1Þ where E(x,y,z) is the magnitude of the electric field at position (x,y,z), m is the mobility of the charge at low electric field, and e as well as b are constants. Values for e and b for both electrons and holes are available elsewhere [21]. A computer code was built to simulate the motion of the different charges inside the Ge crystals, and to simulate the output pulse of the detector as a function of time. This code was, also, used to calculate the total collection time of all the charges as a function of their initial position. The initial positions for the electron–hole pairs were fed from the outcome of the Monte Carlo simulations of the photopeak efficiency. Only the photons that transferred all the energy inside the detector, i.e. that belonged to the photopeak, were considered. Secondary photons, which deposited energy in several locations inside the detector, were initially treated as independent events and their contributions were recombined and compiled in the final analysis of the data. Trapping of the charges inside the detector due to impurities was not considered in our model. This process is expected to have little influence in high purity Ge detectors and should only be considered if heavy radiation damage is suspected. Fig. 4(a) displays a cross-section of the largest well-type Gedetector with the total collection time distribution as a function of position. Fig. 4(b) shows the development of the output pulse versus time due to the drift of electrons and holes for 9 different initial positions as marked in Fig. 4(a). The magnitude of the total charge collection time agrees well with the data given by Ref. [22]. These ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 345 Fig. 4. (a) Cross-section of the largest well-type Ge-detector showing the total charge collection time at the recommended operating voltage of 3000 V. (b) Calculated output pulse for 9 given positions in the largest well-type Ge-detector as marked in Fig. 3(a). The pulses are due to the electron’s and the hole’s motion in the detector. 346 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 results were, also, tested for the other well-type and planar-type Ge-detectors. Though the corrections of the efficiency simulation for the incomplete charge collection should have been based on the initial slope of the pulses, this proved to be time consuming and inconvenient. Multi-parameterisation was required to add each of the charge functions produced by the motion of both the electrons and holes to their respective collection contacts. In addition, no information is available from the manufacturer about the discrimination settings of the different preamplifiers. It was, therefore, considered to be more advantageous to base the corrections on the total charge collection time, where a cut-off time would be used as a discriminating limit for the rejection of pulses. There were two main reasons for this. Firstly, this parameter (the total collection time) depends solely on the detector’s operating voltage and the geometrical data provided by the manufacturer, which eases the extension of the model to other well-type Ge-detectors. Secondly, there exists a good correlation between the regions with high probability for shallow initial slope of the pulses and the longest collection times as observed in Figs. 4(a) and (b). This model saves considerable computing time. However, it has some drawbacks, especially when applied to photon energies below 40 keV. In general, pulses with shallow initial slope would give the longest collection times. However, pulses originating from low energy photons (detected very close to the surface of the active volume) may have rather steep initial slopes and long total collection times. This is due to the fast collection of the holes in the contact nearest to the surface of the Ge-crystal facing the source, and the drift of electrons all the way to the other side of the depletion layer (see pulse from position 6). The collection cut-off time in the model was set equal to the total collection time required for an electron–hole pair produced at position 9 (or equivalent for the other well-type Ge-detector). This time was observed to be compatible with the collection times calculated for the planar detectors and with the experimental results obtained from the scanning of the largest welltype Ge-detector. 3. Results and discussion Fig. 5 displays the experimental results and the MCNP simulations of the relative efficiency of the largest well-type Ge-detector at two different energies, 59.5 and 186 keV. This data corresponds to the case where the collimated source is being centred on the symmetry axis at a distance of 47 mm from the bottom of the detector and rotating around the radial-axis in the horizontal plane perpendicular to the bottom of the detector. Zero degrees correspond to the collimated beam perpendicular to the bottom of the detector. Normalisation and conversion of the data to relative efficiencies was done by comparing the different counting rates measured in the well-type detector at different incident angles to the counting rate measured in one of the planar-type detectors (2.5 cm thick) at zero incidence angle. Therefore, 100% relative efficiency corresponds to the same measured counting rate. The counting rates measured by the well-type Ge-detector were corrected for the different thicknesses of the Al window seen by the photons. Comparison of the counting rates between the detectors was possible since the thickness of the planar-type detector is almost equal to that of the side of the well-type Ge-detector. Because of the geometrical aspect of the field distribution, the planar-type detector was assumed to have complete charge collection. The simulated data was included in Fig. 5 to indicate what the expected efficiency would have been if this detector had complete charge collection. The error bars given in Fig. 5 represent one-sigma uncertainties. The uncertainty in the angle was estimated to be below 7%. The high attenuation coefficient of the 59.5 keV photons in germanium limits their range in the Ge-crystal. The mean free path of these photons is of the order of 1 mm. This means that the efficiency at this energy is independent of the direction of the collimated beam as long as the beam hits the depletion layer. In the case of 186 keV photons, the mean free path is of the order of 1 cm and thus the efficiency depends on the thickness of the detector as well. The larger the active Ge-volume ‘seen’ by the collimated beam, the higher the efficiency. Irregularities in the uniformity of the P+ contact layer in ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 347 Fig. 5. Scanning of the efficiency of the largest well-type Ge-detector using a collimated beam directed, at different angles y, towards the surface of the Ge-crystal. y=0 corresponds to the beam perpendicular to the base of the detector. The collimated source is centred on the axis of symmetry at a distance of 47 mm from the base of the detector. Two different photon energies are given. Information about the normalisation of the data is given in the text. the detector should not make substantial contributions to the efficiency measurements at any of these energies. The biggest differences between experimental and simulated data appear at the extreme angles for both energies, i.e. at 01 and 1351. The difference at 1351 can partly be explained by the escape of photons from the top of the detector. This is most likely due to the spread of the collimated beam with distance from the pin hole. For the 59.5 keV photons the relative efficiency should have been the same as for any of the other angles. However a lower efficiency was, also, measured. For the 186 keV photons, 01 would have corresponded to the position where the detector size, as seen by the collimated beam, is largest. However, the efficiency was much lower than expected. These results give clear evidence that the whole germanium crystal is not equally efficient in detecting the photons or translating the deposited energies to electric pulses. As shown in Fig. 4(a), the longest charge collection times are calculated for photon interactions that occur at the bottom of the well cavity (which correspond to an angle of 01) and the bottom of the Ge-crystal (0–301) due to ‘irregularities’ in the field. Due to the field characteristics, a larger number of pulses would be rejected in these zones explaining the lower efficiency that is seen experimentally. An estimation of the differences between the experimental and the simulated data for a cylindrical source, as based on the measurements described above, gave values of 10–20% for the 186 keV photons and 1–5% for the 59.5 keV photons. The experimental results of the collimated beam at different distances from the base of the detector showed the same relation between the 186 keV photons and the volume of Ge-crystal ‘seen’ by the beam. The efficiency of the collimated beam at different angles around the symmetry axis did not indicate variations in magnitude. Calculations of the charge collection cut-off times, as described in the ‘modelling of the charge collection process’ section, provided values of 250 ns for the largest well-type detector and 190 ns for the smallest well-type detector. These values can also be estimated directly from Figs. 4(a) and 6. The smallest well-type Ge-detector showed 348 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 a faster and more uniform collection time map (Fig. 6). This was not only due to the smaller dimensions of this detector but also because it runs at a higher operating voltage. The smaller dimensions of this detector give the better field homogeneity while the dimensions and the operating voltage determine the field strength. Figs. 7(a) and 8(a) show the experimental and MCNP simulated (with and without correction) efficiency for the two well-type Ge-detectors. The fitting of the experimental data was done using multiple polynomial functions of the logarithm of the efficiency and the logarithm of the energy. The calculation of the uncertainties of the simulated efficiency where based on the following assumptions: (1) the uncertainties in the dimensions of the detector are small and (2) the uncertainty in the charge collection cut-off time is 10 ns for both well-type Ge-detectors. Analysis of the efficiency corrections for small variations in the cut-off time showed that intermediate energies (80–800 keV) are not especially sensitive. However, low and high energies showed variations in magnitude of up to 10%. These variations were more substantial in the smaller well-type detector probably due to the smaller collection time differences across the whole detector. It can be observed from Figs. 7(a), (b), 8(a) and (b) that a better agreement between the experimental and the simulated data is obtained for both detectors after charge collection correction. Above 40–50 keV, the statistical agreement between the experimental and simulated data after these corrections is, in fact, rather good. Further improvements are, however, still needed at energies below 40 keV. The improvements are more dramatic for the largest well-type Ge-detector than for the smallest one. The ratios between the experimental and simulated data displayed in Figs. 7(b) and 8(b) increase, for example, from 0.73 to 0.96 after correction for the largest well-type detector at 1460 keV, and from 0.85 to 0.92 for the smallest well-type detector at the same photon energy. Well-type Ge-detectors are difficult to use because of summing effects arising from low energy photons. Lining of these detectors with a high-Z absorber has been recommended by some authors to prevent coincidence summing. In those cases the presented method would work perfectly. A more detailed model considering the rising slope of the output pulse would, however, be required to improve the data at energies below 40 keV. Furthermore, experimental analysis of the dimensions of the detector, the end-cup window Fig. 6. Cross-section of the smallest well-type Ge-detector showing the total charge collection time at the recommended operation voltage of 3500 V. ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 349 The largest Well-type Ge-detector Photopeak Efficiency 1 0.1 Experimental data Experimental data fit Original MCNP simulation Corrected MCNP simulation 10 (a) 100 1000 Energy (keV) The largest well-type Ge-detector Ratio Experimental / Simulated Efficiencies 1.2 1.0 0.8 0.6 Uncorrected simulation 0.4 Corrected Simulation 0.2 10 (b) 100 1000 Energy (keV) Fig. 7. (a) Experimental and simulated efficiency of the largest well-type Ge-detector for a cylindrical source (radius = 6.5 mm and height = 40 mm). (b) Ratio between the experimental (or fitted curve of the experimental data) and simulated efficiency before and after charge collection correction. 350 ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 The smallest well-type Ge-detector Photopeak Efficiency 1 0.1 Experimental data Original MCNP simulation Corrected MCNP simulation Experimental data fit 0.01 10 100 (a) 1000 Energy (keV) The smallest well-type Ge-detector Ratio Experimental / Simulated Efficiencies 1.2 1.0 0.8 0.6 Uncorrected simulation Corrected Simulation 0.4 0.2 10 (b) 100 Energy (keV) 1000 Fig. 8. (a) Experimental and simulated efficiency of the smallest well-type Ge-detector for a cylindrical source (radius = 4 mm and height = 40 mm). (b) Ratio between the experimental (or fitted curve of the experimental data) and simulated efficiency before and after charge collection correction. and the dead contact layers is also required to unfold the different uncertainty contributions due to the input parameters in the simulations at such low photon energies. Currently, semi-empirical Monte Carlo approaches still provide a more accurate (within 5%) alternative for the simulation of the performance of well-type Ge-detectors at energies below 40 keV [15]. ! F. Hernandez, F. El-Daoushy / Nuclear Instruments and Methods in Physics Research A 498 (2003) 340–351 4. Conclusions Corrections for the incomplete collection of charges in Ge-crystals are necessary for improving the simulation of the efficiency of well-type Gedetectors. The scanning experiment of the efficiency of Ge-detectors presented here shows clear differences in the efficiency in different parts of the Ge-crystal. The lower part of the Ge-crystal shows the largest discrepancies in efficiency, most likely due to the lack of uniformity in direction and strength of the electric field. A simple sub-routine that simulates the total collection time of charges produced by the interaction of photons in the Gecrystal can be used to correct the Monte Carlo efficiency simulation. Good statistical agreement is obtained between the experimental and simulated efficiencies after charge collection correction for both well-type Ge-detectors and for photon energies above 40 keV. A more detailed model considering the initial slope of the produced pulse from the detector is required to correct the efficiency simulation at energies below 40 keV. Further experiments, using collimated sources, are necessary for detailed studies of the different parts of other Ge-crystals. Acknowledgements We would like to thank Cecilia Johansson for her introductory lesson in MCNP Monte Carlo Coding and for helping us to set the first geometry. We would, also, like to thank the referee for his positive criticisms. References [1] F. El-Daoushy, R. Garcia-Tenorio, Nucl. Instr. and Meth. A 356 (1995) 376. 351 [2] M. Laborie, G. Le Petit, D. Abt, M. Girard, Nucl. Instr. and Meth. A 479 (2002) 618. [3] O. Sima, Nucl. Instr. and Meth. A 450 (2000) 98. [4] Ch. Meixner, Nucl. Instr. and Meth. 119 (1974) 521. [5] S. Jiang, J. Liang, J. Chou, U. Lin, W. Yeh, Nucl. Instr. and Meth. A 413 (1998) 281. [6] O. Sima, D. Arnold, Appl. Radiat. Isot 56 (2002) 71. [7] F. Bronson, L. Wang, Canberra, Validation of the MCNP Monte Carlo code for germanium detector gamma efficiency calibrations, presented at waste management ’96, Tucson, AZ, USA, February 28, 1996. [8] G. Haase, D. Tait, A. Wiechen, Nucl. Instr. and Meth. A 329 (1993) 483. [9] S. Ashrafi, A. Likar, T. Vidmar, Nucl. Instr. and Meth. A 438 (1999) 421. [10] H. Baas, P. Bode, M. Blaauw, Radiat. Phys. Chem. 61 (2001) 637. [11] T. Wang, I. Hou, C. Tseng, Nucl. Instr. and Meth. A 425 (1999) 504. ! [12] J. Rodenas, A. Martinavarro, V. Rius, Nucl. Instr. and Meth. A 450 (2000) 88. [13] I. Ewa, D. Bodizs, Sz. Czifrus, Zs. Molnar, Appl. Radiat. Isot. 55 (2001) 103. [14] M. L!epy, T. Altzitzoglou, D. Arnold, F. Bronson, R. Capote, M. D!ecombaz, F. De Corte, R. Edelmaier, E. Herrera, S. Klemola, M. Korun, M. Kralil, H. Neder, J. Plagnard, S. Pomm, J. De Sanoit, et al., Appl. Radiat. Isot. 55 (2001) 493. [15] F. Hernandez, F. El-Daoushy, Nucl. Instr. and Meth. A 484 (2002) 625. [16] J. Briesmeister (Ed.), MCNP—A general Monte Carlo Nparticle transport code, version 4B, LA-12625-M, Los Alamos National Laboratory, Los Alamos, NM, March 1997. [17] F. Hern!andez, Optimization of environmental gamma spectrometry using MC methods, Comprehensive summaries of Uppsala dissertations from the Faculty of Science and Techonology 751, Acta Universitatis Uppsaliensis, Uppsala 2002, ISBN: 91-554-5399-6. [18] F. El-Daoushy, F. Hernandez, The Analyst 127 (2002) 981. [19] G.F. Knoll. Radiation detection and measurements, 2nd Edition, John Wiley, New York, 1989. [20] FlexPDE — PDE solutions Inc. P.O. Box 4217, Antioch, CA 94531–4217. http://www.pdesolutions.com. [21] T.W. Raudorf, R.H. Pehl, Nucl. Instr. and Meth. A 255 (1987) 538. [22] EG&G ORTEC. Catalog—Modular pulse-processing electronics, semiconductor radiation detectors, Printedin, USA, November 1994.