Energy resolution and transport properties of CdTe-Timepix-Assemblies

Journal of Instrumentation OPEN ACCESS Energy resolution and transport properties of CdTe-Timepix-Assemblies To cite this article: D Greiffenberg et al View the article online for updates and enhancements. Related content - Study of a GaAs:Cr-based Timepix detector using synchrotron facility P. Smolyanskiy, D. Kozhevnikov, O. Bakina et al. - X-ray absorption and charge transport in a pixellated CdTe detector with single photon processing readout E Fröjdh, B Norlin, G Thungström et al. - Imaging properties of small-pixel spectroscopic x-ray detectors based on cadmium telluride sensors Thomas Koenig, Julia Schulze, Marcus Zuber et al. Recent citations - Energy calibration of photon counting detectors using a single monochromatic source C. Feng et al - Laboratory based study of dynamical processes by 4D X-ray CT with subsecond temporal resolution D. Vavík et al - Characterization of a pixelated CdTe Timepix detector operated in ToT mode T. Billoud et al This content was downloaded from IP address 46.3.196.214 on 25/01/2018 at 18:56

PUBLISHED BY IOP PUBLISHING FOR SISSA 12 th INTERNATIONAL WORKSHOP ON RADIATION IMAGING DETECTORS, JULY 11 th 15 th 2010, ROBINSON COLLEGE, CAMBRIDGE U.K. RECEIVED: November 15, 2010 ACCEPTED: December 10, 2010 PUBLISHED: January 11, 2011 Energy resolution and transport properties of CdTe-Timepix-Assemblies D. Greiffenberg, a,1 A. Fauler, a A. Zwerger b and M. Fiederle a a Freiburg Materials Research Institute (FMF), Albert-Ludwigs-Universität Freiburg, Stefan-Meier-Strasse 21, D-79104 Freiburg, Germany b Physikalisches Institut, Albert-Ludwigs-Universität Freiburg, Hermann-Herder-Strasse 3, D-79104 Freiburg, Germany E-mail: Dominic.Greiffenberg@fmf.uni-freiburg.de ABSTRACT: CdTe is a promising material for the detection of γ- and X-rays as 1 mm thick CdTe sensors offer an absorption efficiency that is higher than 50 % for photon energies up to 120 kev. Therefore 1mm thick CdTe from Acrorad has been flip-chipped as the sensor material on a Timepix readout-electronics ASIC at the Freiburg Materials Research Center (FMF). The transport properties of the CdTe material are investigated by determining the µτ-product of the charge carriers by illuminating the non-collecting electrode with alpha particles. The method commonly used to determine the µτ-product for non-pixelated devices was adapted for the Medipix pixelated readout-electronics. The position of the alpha particles on the pixel matrix provides spatial information, which was used to create a mapping of the µτ-product. The energy resolutions and the positions of the noise edge of two CdTe-Timepix assemblies, one with a pixel pitch of 110 x 110 µm 2 and another one with a pixel pitch of 55 x 55 µm 2 were investigated by performing a threshold scan of a 241 Am source. A comparison of the energy spectra obtained with the two assemblies with different pixel pitches confirms the strong dependence of charge sharing on the energy spectra, as expected. KEYWORDS: Charge transport and multiplication in solid media; X-ray detectors 1 Corresponding author. c 2011 IOP Publishing Ltd and SISSA doi:10.1088/1748-0221/6/01/c01058

Contents 1 Introduction 1 2 Experimental setup 2 3 Results 3 3.1 Determination of the charge carrier transport properties 3 3.2 Energy spectra 5 4 Conclusions 7 1 Introduction CdTe is a promising material for the detection of γ- and X-rays with photon energies up to 120 kev. 1mm thick CdTe offers an absorption efficiency higher than 50 % within this energy regime due to its high atomic number (Z=48, 52). The sensor material was provided by Acrorad [1]. Further processing and flip-chipping with the pixelated readout electronics Timepix was done at the Freiburg Materials Research Center (FMF) [2, 3]. The assembly had a pixel pitch of 110 x 110 µm 2 and had intentionally ohmic contacts. The transport properties of the charge carriers in the CdTe sensor, namely the mobility-lifetime product (µτ-product), have a major influence on the charge collection efficiency (CCE) of the detectors [4]. The µτ-product was investigated by using alpha particles from a 241 Am source. The alpha particles are absorbed in the first 30 microns of the sensor and the charge carriers have to drift almost through the complete sensor to the pixel contacts. This method is commonly used for nonpixelated systems, where the induced charges are generating pulse amplitudes, which are measured for various applied bias voltages. As the slope of the Hecht relationship is relevant for the value of the µτ product, a proportional conversion of the number of charge carriers to pulse height is a necessary requirement. Finally by fitting the slope using a Hecht relationship, the µτ-product can be extracted [5]. The Hecht-relationship is shown in equation (1.1). Q = Q 0 [ µe τ e U bias d 2 ( 1 e x 0 d µeτeu bias ) + µ hτ h U bias d 2 ( 1 e (d x 0 )d µ h τ h U bias )] (1.1) Here Q refers to the pulse height, which is correlated with the induced charge. µ is the mobility and τ the lifetime of the charge carriers. The index e refers to electrons, while h refers to holes. U bias is the bias voltage on the sensor and d is the sensors thickness. After absorption of the alpha particles, the generated charge cloud will expand while drifting towards the collecting electrode. As the CdTe-Timepix assembly is a pixelated readout system, the charge carriers will influence charge in a circular shaped cluster expanding over several pixels. 1

Therefore the method described before had to be adapted, which was achieved by using the Timeover-Threshold (ToT) readout mode of the Timepix chip [6, 7]. This readout mode is measuring the time, which the signal of the charge sensitive amplifier (CSA) is above the energy threshold. The ToT is measured by counting the number of period lengths given by an internal frequency, being further referred to as ToT counts. In the case of the Timepix pixel electronics the pulse length is directly correlated to the pulse height, thus providing a measure for the pulse height in each pixel [7]. The overall charge created by an alpha particle will create pulse heights in several pixels which when summed up corresponds to the pulse height as would be induced in a non-pixelated readout system. Thus, a reconstruction of the overall pulse height is possible by summing the pulse heights in each pixel of the cluster. Moreover, the method allows spatially resolved µτ-mappings of the sensor, which then can be correlated with other properties of the assembly (e.g. quantum efficiency). The energy resolution and the energy level of the noise edge of the assembly was determined by performing an energy calibration. For this purpose, a scan of the energy threshold was performed, while the assembly was illuminated with a 241 Am source, where the alpha particles were shielded. More details of the method are given elsewhere [8, 9]. In order to compare the results with an CdTe assembly with a pixel pitch of 55 x 55 µm 2, the same procedure of obtaining an energy resolution was performed with such an assembly. 2 Experimental setup A Timepix chip from the Medipix2 collaboration was used as the readout electronics [10]. The Timepix chip consists of 256 x 256 = 65536 pixels with a pixel size of 55 x 55 µm 2 covering an active area of 1.98 cm 2. It is designed as hybrid readout electronics and is capable of collecting either electrons or holes. Thus different sensor materials like Si, GaAs and CdTe can be used. The Timepix chip also supports different readout modes like the Time-over-Threshold (ToT) mode, which measures the time, while the signal is over the energy threshold in units of the clock period of an external signal. The readout modes have been described in several publications [6, 7]. In the Timepix readout electronics, the charge sensitive amplifiers (CSA) are discharged by a current source which is controlled by the DAC IKrum. The shaped pulse decreases linearly, creating a triangular shaped pulse [6, 7]. The rise time of the CSA is constant over a wide range of induced charge. One consequence of this behavior is a direct correlation between the pulse length and the pulse height [7]. Therefore, the pulse length measured using the ToT mode is a good measure for the pulse height. For the determination of the µτ-product, a 1 mm thick CdTe sensor was flip-chipped to Timepix readout electronics. By connecting every 2 nd pixel, a pixel pitch was 110 x 110 µm 2 was achieved [3]. The system consisting of sensor material and readout electronics will be referred to as an assembly. The contacts were intentionally ohmic, thus the sensor acted as a photoresistor, allowing positive or negative voltage to be applied and the µτ-product to be measured for either electrons or holes. The clock frequency which determines the resolution of the ToT mode was set to 80 MHz, and the energy threshold was set a value of (15.3 ± 0.2) kev. Before these measurements, a threshold equalization with respect to the noise edge was performed to ensure an equal response of the pixels. 2

Alpha particles emitted by a 241 Am source with a mean energy of 5.48 MeV were used [11]. The 241 Am source was placed around 4 mm above the sensor surface. In order to reduce the interaction of the alpha particles with the ambient air, the measurements have been performed in a vacuum chamber with a residual pressure of 0.5 mbar. The interaction of the alpha particles with the sensor material happens close below the surface. Therefore, this method of measuring the µτproduct is adaptable for electrons as well as holes, as the contribution from the opposite charge is negligible. Thus, equation 1 can be reduced to one sort of charge carrier, while furthermore being adjusted with the fitting parameter U 0. Q(U) = Q 0 µ τ ( ( )) d 2 (U U d 2 µ τ (U U 0) 1 e 0 ) (2.1) The fitting parameter U 0 takes into account the effect of the energy threshold. Descriptively, it is the minimum voltage that needs to be applied in order to induce enough charge in a single pixel and the pulse height reaches above the energy threshold. For the measurement of the energy spectra the 241 Am source was used, but alpha particles were shielded. An energy calibration of the energy threshold (THL) was performed using the single photon counting mode, which registers the number of photons whose pulses were above the energy threshold. The energy calibration was done by scanning the THL, and therefore by measuring the integral spectrum. Finally, the 241 Am spectrum can be extracted by differentiation. More details on the procedure are described elsewhere [8, 9]. Two Timepix assemblies were used for the measurements. One assembly was the assembly described above with a pixel pitch 110 µm and quasi-ohmic contacts, the other assembly was a M-π-n diode structure described in publication [12] and has a pixel pitch of 55 µm In both cases, electrons were collected and the sensor voltage was set to 500 V. Prior to the measurements, a threshold equalization was performed with respect to the noise edge. Moreover, a moving average was applied to the spectra. In the case of the 110 x 110 µm 2 assembly, the spectrum was obtained by using the complete pixel matrix, while for the 55 x55 µm 2 assembly the analysis area was restricted to 15 x 15 pixels. 3 Results 3.1 Determination of the charge carrier transport properties For the measurement of the µτ-products, it was essential to obtain the position as well as the ToT counts of single clusters created by alpha particles, thus the acquisition time was set to 2 ms, to avoid overlapping of individual clusters. In a first approach, the mean number of pixels in a cluster was measured versus the applied bias voltage (figure 1). As a necessary first step, an algorithm was used to check each cluster, to see if the complete cluster was registered by the pixel matrix, or cut, e.g. by hitting the sensor edge or by hitting dead pixels. Figure 1 shows the mean cluster size for a voltage range between 50 to 500 V in case of electron collection. The maximum cluster size is around 35 pixels for the smallest bias voltage of 50 V. The cluster size decreases with increasing bias voltage due to the shorter drift time and therefore the shorter time for the charge cloud to diffuse. When fitting the data with a function describing 3

Figure 1. Number of pixels in a cluster versus the applied bias voltage. The saturation value given by fitting with a function describing an exponential decay results a minimal cluster size of (14.19±0.14) pixels, which corresponds to a diameter of (468±5)µm, presuming circular shaped clusters. an exponential decrease, a minimal cluster size of (14.19 ± 0.14) pixels is obtained. Presuming circular shaped clusters, that corresponds to a mean diameter of the clusters of (468 ± 5) µm. At the beginning of the measurements the mean µτ-product was determined. In order to investigate the functionality of the procedure, as a first step the spatial information of the position of the clusters as neglected. In case of electron collection, the bias voltage was varied between 10 and 70 V, while in case of hole collection it was varied between 25 and 200 V. The resulting values of the reconstructed pulse heights versus bias voltage for electron collection can be seen on the left side in figure 2. By using equation 2, a mean µτ-product for electrons of µ e τ e = (1.9 ± 0.6). 10 3 cm 2 /V was measured. This value is within the range of values reported by other authors for Acrorad material [1, 13 16]. The saturation value of the reconstructed pulse height in units of ToT counts resulting from the fit is Q 0 = (3753 ± 112). By using the energy calibration for the ToT counts, the energy corresponding to Q 0 is (4.41 ± 0.13) MeV. Thus, a charge collection efficiency for the electron collection is estimated to be (80.4 ± 2.4) %. The slope of the reconstructed pulse heights versus bias voltage, in case of hole collection, can be seen on the right side of figure 2. By fitting the Hecht relationship, a mean µτ-product of µ h τ h = (0.75 ± 0.25). 10 4 cm 2 /V is obtained. Again, this value is within the range reported by different authors [1, 13 16]. The saturation value was Q 0 = (1591 ± 204), which corresponds to an energy of (1.86 ± 0.24) MeV and results a charge collection efficiency of (34.0 ± 4.4) %. In order to exploit the additional spatial information provided by the position of the clusters, the 128 x 128 pixel matrix was divided into 16 x 16 sub-squares, so each square contained 8 x 8 pixels. An algorithm checked if the complete cluster lay within the square and then reconstructed the overall pulse height by summing up the ToT counts for each single pixel of the cluster. A spatial mapping of the µτ-product was obtained with a spatial resolution of 880 x 880 µm 2. These 4

Figure 2. (left) Measured pulse heights versus bias voltage in case of electron collection fitted with equation 2 describing the Hecht relationship. The values resulting from the fit are: µ e τ e = (1.9 ± 0.6). 10 3 cm 2 /V, Q 0 = (3753 ± 112), U 0 = (7.4 ± 1.7) V. (right) Measured pulse heights versus bias voltage in case of hole collection fitted with equation 2 describing the Hecht relationship. The values resulting from the fit are: µ h τ h = (0,75 ± 0,25). 10 4 cm 2 /V,Q 0 = (1591 ± 204), U 0 = (13,5 ± 7,3) V. measurements have only been performed for electron collection as this is the typical operation mode due to the better transport properties of the electrons. The resulting mapping of the values for the µτ-product for electrons is shown in figure 3. The mapping reveals slightly higher values for the transport properties in the lower right part of the sensor compared to the mean value at around (2.2 2.4). 10 3 cm 2 /V. The transport properties in the upper half of the sensor are slightly reduced to values of around (1.5 2.1). 10 3 cm 2 /V. As the charge collection efficiency is connected with the µτ-product, a flood image was taken in order to compare the count rate of the pixels with the transport properties in the respective areas. The flood images were created by using an X-ray source with an acceleration voltage of 20 kv to ensure an absorption close below the sensor surface. The bias voltage was set to 500 V and prior to the measurements a threshold equalization with respect to the noise edge was performed. The flood image is shown in figure 3 on the left side. There is a correlation visible between the counting behavior of the pixels and the transport properties of the charge carriers. 3.2 Energy spectra The energy spectra, obtained using a threshold scan of a 241 Am source, are shown in figure 4. On the left side of figure 4, the energy spectrum obtained with the assembly with quasi-ohmic contacts and a pixel pitch of 110 x 110 µm 2 is shown. The spectrum shown includes all pixels from the pixel matrix. The highest energetic peak with a photon energy of 59.5 kev at a mean THL value of (509.4 ± 0.2) can clearly be resolved. The count rate floor for THL values below can be attributed to events caused by charge sharing of 59.5 kev photons, which are only partially collected in a single pixel. There several peaks are visible for THL values below 350, which can be assigned to the respective photon energies emitted by the source or by fluorescence effects of the CdTe. For example, the self-absorption of the Te-fluorescence (27.4 kev) by the Cd K-edge located at 26.7 5

Figure 3. (left) Flood image with an X-ray tube. The acceleration voltage was 20 kv and the bias voltage was 500 V. Prior to the measurements a threshold equalization was performed with respect to the noise edge. (right) µτ-mapping for electrons with a bilinear filter applied. The two dots with higher values for the µτproduct in the µτ-mapping are arising from the fit, which is indicated by relatively high errors for these two values. Figure 4. (left) Spectrum of a 241 Am source obtained by performing a threshold scan with an assembly with quasi-ohmic contacts with a pixel pitch of 110 x 110 µm 2. Several peaks can be resolved. (right) Spectrum of a 241 Am source obtained by performing a threshold scan with an assembly with a M-π-n diode structure with a pixel pitch of 55 x 55 µm 2. No peaks can be resolved, only shoulders are visible. kev can be seen. After being absorbed by the Cd K-edge, photons with an energy of 23.1 kev originating from the Cd-fluorescence will contribute to the spectrum. Using the assignment of the THL values to its respective photon energies, an energy calibration was made, this exhibits a linear dependence (figure 5). The energy resolution of the 59.5 kev peak was determined exploiting the energy calibration, finally resulting a FWHM of the 59.5 kev peak of 5.6 %. Moreover, the position of the noise edge with respect to energy could be measured to be at (5.42 ± 0.14) kev. The THL position of the noise edge was extracted from the information provided by the threshold equalization. When performing the determination of the energetic position of the noise edge with a standard silicon assembly, a value of (4.0 ± 0.3) kev was measured. On the right side of figure 4, the energy spectrum of the 241 Am source is shown, using the a M-π-n diode structure on a Timepix readout electronics with a pixel pitch of 55 x 55 µm 2. Single 6

Figure 5. Energy calibration for the 110 110 µm 2 pixel pitch assembly wih quasi-ohmic contacts. peaks could not be resolved, as it was possible for the assembly with the 110 x 110 µm 2 pixel pitch. Two shoulders are visible, which can most probably be attributed to the 59.5 kev photons and the 23.1 kev Cd-fluorescence. The reason that the peaks cannot be resolved, is assumed to be due to charge sharing, as it was shown by Nillson et al. [17]. 4 Conclusions The transport properties of the charge carriers of a CdTe sensor from Acrorad on a Timepix readout electronics with a pixel pitch of 110 x 110 µm 2 have been investigated. The method is an adaption to pixelated readout systems, of the procedure to determine the µτ-product using alphaparticles. The results obtained are comparable to the results presented by other authors. Moreover, the spatial information was exploited to create a mapping of the µτ-product and a correlation between the counting behavior of the assembly and the distribution of the µτ-products in the corresponding areas. The energy resolution and the energetic position of the noise edge were determined by using a threshold scan of a 241 Am source. The FWHM of the 59.5 kev peak was 5.6 %, the position of the noise edge was at (5.42 ± 0.14) kev. Measurements with a silicon assembly resulted a noise edge energy of (4.0 ± 0.3) kev. Thus, with respect to the noise edge, the CdTe assemblies are comparable to the silicon assemblies. An energy spectrum taken with an assembly with a pixel pitch of 55 x 55 µm 2 revealed, that it s not possible to resolve single peaks, which is most probably due to charge sharing. 7

Acknowledgments The work was funded by the,,bundesministerium für Bildung und Forschung (Förderkennzeichen: 05 KS7VFA). The work was carried out within the Medipix collaboration. (www.cern.ch/medipix) The authors thank Prof. P. H. Butler from the University of Canterbury for proofreading this article. References [1] Acrorad, http://www.acrorad.co.jp. [2] M. Fiederle, H. Braml and A. Fauler, Development of flip-chip bonding technology for (Cd,Zn)Te, IEEE Trans. Nucl. Sci., 51 (2004) 1799. [3] A. Zwerger and M. Fiederle, Medipix2: processing and measurements of GaAs pixel detectors, Nucl. Instrum. Meth. A 576 (2007) 23. [4] Y. Du et al., Temporal response of CZT detectors under intense irradiation, IEEE Trans. Nucl. Sci. 50 (2003) 1031. [5] K. Hecht, Zum Mechanismus des lichtelektrischen Primärstromes in isolierenden Kristallen, Z. Physik A 77 (1932) 235. [6] X. Llopart et al., Timepix, a 65k programmable pixel readout chip for arrival time, energy and/or photon counting measurements, Nucl. Instrum. Meth. A 581 (2007) 485. [7] Jan Jakùbek, Precise energy calibration of pixel detector working in time-over-threshold mode, article in press in Nucl. Instrum. Meth. A (corrected proof). [8] M. Fiederle et al., Energy calibration measurements of MediPix2, Nucl. Instrum. Meth. A 591 (2008) 75. [9] Ch. Broennimann et al., The PILATUS 1M detector, J. Synchrotron Rad. 13 (2006) 120. [10] Medipix, http://www.cern.ch/medipix. [11] WWW Table of Radioactive Isotopes, Radiation Search, http://ie.lbl.gov/toi/radsearch.asp. [12] D. Greiffenberg et al., Preliminary characterisation of CdTe M-π-n diode structures, article in press in Nucl. Instrum. Meth. A (corrected proof). [13] G. Sato et al., Characterization of CdTe/CdZnTe detectors, IEEE Trans. Nucl. Sci. 49 (2002) 1258. [14] P.J. Sellin et al., Drift mobility and mobility-lifetime products in CdTe:Cl grown by the travelling heater method, IEEE Trans. Nucl. Sci. 52 (2005) 3074. [15] H. Shiraki et al., Improvement of the productivity in the growth of CdTe single crystal by THM for the new PET system, IEEE Nucl. Sci. Symp. Conf. Rec. (2007) 1783. [16] T. Takahasi et al., High resolution Schottky CdTe diode detector, IEEE Trans. Nucl. Sci. 49 (2002) 1297. [17] H.-E. Nilsson et al., Simulation of photon and charge transport in X-ray imaging semiconductor sensors, Nucl. Instrum. Meth. A 487 (2002) 151. 8