LEAPS-INNOV D2.4 - Detector prototype performances tests report

Deliverable no. D2.4 Page 1 of 13 Project information Project full title LEAPS pilot to foster open innovation for accelerator - based light sources in Europe Project acronym LEAPS - INNOV Grant agreement no. 101004728 Instrument Research and Innovation Action (RIA) Duration 01/0 4 /202 1 – 3 0 /0 9 /202 5 Website https://www.leaps - innov.eu/ Deliverable information Deliverable no. D2. 4 Deliverable t itle Detector prototype performances tests report Deliverable responsible ESRF, (SOLEIL and DIAMOND) Related Work - Package/Task WP2 (XAFS - DET) / Task 2. 8 Type (e.g. Report; other) Report Author(s) F.J. Iguaz , N. Goyal , S.Cha tterji & E. N. Gimenez Dissemination level Public Document Version V1 Date 0 7 /07 /2025 Download page Document information Version no. Date Author(s) Comment V1 15/08/2025 FJ. Iguaz , N. Goyal , S.Chatterji & E.N.Gimenez Deliverable no. D2.4 Page 2 of 13 [Table of Contents] 1. Introduction 2. Characterization tests at laboratory 3. Surface scans with a collimated X-ray source 4. Beam test at ESRF BM05 beamline 5. Further tests: risetime distribution and noise spectrum 6. Conclusions and future steps 7. References 1. Introduction This document summarizes the characterization tests of XAFS-DET prototype [1-5] in 2025, as deliverable of task 2.8. Preliminary tests were performed during the detector assembly and reported in deliverable D2.3 [6]. During the tests, the detector was optimized to reduce the noise level: an RCfilter was included at high voltage (HV) line, the HV line was rerouted to a side flange and the optimum electronics box configuration was found (semi-closed with a copper shield). Tests were mainly performed at ESRF with contributions of other partners (DIAMOND, ESRF, INFN-Frascati, MAX-IV, SOLEIL). Regular fortnightly meetings took place months beforehand with all partners of the collaboration to achieve the deliverable. This document is divided in four parts: the characterization tests in the laboratory using radioactive sources, described in section 2; surface scans with a collimated X-ray source, in section 3; a beam test at ESRF BM05 beamline, in section 4; and a noise study at SOLEIL detector group lab, in section 5. The document is closed with some conclusions and future steps in section 6. 2. Characterization tests in the laboratory The first spectroscopic tests of XAFS-DET detector were performed at ESRF detector group lab, with the participation of partners from DIAMOND, ESRF and SOLEIL. Tests were made before and after: rerouting the HV line, the installation of RC-filter at HV line and for different electronic box configurations. Only results for the final detector are shown. Some of these results were presented at the Vienna Conference on Instrumentation 2025 and have been published in [7]. Two different Digital Pulse Processors (DPP) were used: DANTE with a limitation on the number of channels but with features that allowed the selection of the peaking time; and XSPRESS4 with 10 channels and the features for charge-sharing and crosstalk correction. The first tests were performed with the DANTE DPP. The seven central pixels of XAFS-DET were connected to DANTE which uses a trapezoidal filter. Two radioactive sources (55Fe and 241Am) were used to study the lower X-ray energy range (the main line is at 5.9 keV for 55Fe) and hard X-ray energy range (the main line is at 59.5 keV for 241Am). Each source was situated in front of the beryllium Deliverable no. D2.4 Page 3 of 13 entrance window at a distance so that the input count rate (ICR) was less than 10 kcps for all pixels. Two examples of energy spectra are shown in Figure 1. Figure 1: Energy spectra of 55Fe (left) and 241Am sources (right). The legend highlights the energy resolution (eV FWHM) of K-lines of Manganese (left) and different X-ray and gamma lines of Am-241. Images published in [7]. Energy spectra were acquired for peaking time (PT) values between 0.1 and 10 µs to generate the dependence of energy resolution with PT, shown in Figure 2. The best energy resolution values were of 285 eV (FWHM) at 5.9 keV and 445 eV at 59.5 keV at a 6 µs PT. The observed energy resolution is far from specifications: 180 eV at 5.9 keV and 350-400 eV at 59.5 keV at a PT of 1 µs. This deviation is under study and different reasons have already been identified. A noise model has been created from the dependence of energy resolution with peaking time, which points out to a higher pixel capacitance and 1/f voltage noise than the estimated values during design. Figure 2: Dependence of energy resolution at 5.9 keV (left) and at 59.5 keV (right) with the peaking time (PT). The target resolution at a peaking time value of 1 µs is indicated in green and the best current value at 6 µs is indicated in red. Images published in [7]. The tests then were performed with Xspress4 to evaluate the effect on the signal of the charge-sharing by applying the charge-sharing correction algorithm. The 7 central pixels of XAFS-DET prototype were connected to 7 channels of Xspress4, which was calibrated using a 55Fe source, as shown in Figure 3 Deliverable no. D2.4 Page 4 of 13 (left). There is a parameter in Xspress4 called “running average length” which tunes the shaping time of Xspress4. After optimal setting of “running average length” from 512 samples to 1025 samples (which corresponds with an increased shaping time), an energy resolution better than 265 eV for 5 out of 7 channels was achieved. A comparison of energy resolution of the seven channels using default and optimal running average length is shown in Table 1 below. Channel Number Energy resolution (Default running average length) Energy resolution (Optimal running average length) 0 403 eV 319 eV 1 324 eV 252 eV 2 292 eV 263 eV 3 312 eV 284 eV 4 349 eV 262 eV 5 313 eV 256 eV 6 307 eV 247 eV Table 1. Comparison of the energy Resolution of the seven central pixels using default and optimal running average length. Xspress4 firmware was updated to implement crosstalk and charge sharing correction. The detector was tested for charge sharing correction. The charge sharing algorithm cancels time-coincident events and hence improves the low energy tail as it is shown in Figure 3 (right), where the green line represents the 55Fe energy spectrum taken with chare-shared correction, while the red line represents the energy spectrum without the charge-sharing correction. The reduced low-energy tail of 5.9 keV events improves the signal-to-background ratio, a factor 4 in agreement with Xspress4 tests with other monolithic detectors [8, 9]. Crosstalk correction is more effective at higher count rates, since these tests were done using a low count rate 55Fe, crosstalk correction have not been tested so far. There is a plan to use crosstalk correction with beam at high count rates which should improve the Full Width at Half Maxima (FWHM) of the spectrum. Deliverable no. D2.4 Page 5 of 13 Figure 3, left: Photo of the XAFS-DET detector connected to Xspress4 DPP and calibrated by 55Fe source. Figure 3, right: Energy spectrum of 55Fe source before (red line) and the charge-sharing correction (green line) applied by Xpsress4 DPP firmware. 3. Surface scan with a collimated X-ray source XAFS-DET prototype was tested on 7-12th April 2025 at ESRF detector group lab, with the participation of members from ESRF and SOLEIL. The detector, operated at a bias voltage of +70 V, was installed on a motorized table (see Figure 4) facing the copper X-ray source. The source was collimated by two pairs of vertical and horizonal slits (16 mm2 area) and a titanium pin hole (100 µm diameter). X-ray flux could be attenuated by four copper filters, with thickness of 50, 100, 150 and 200 µm. Four detector backend board channels were connected to the four channels of a XIA-Mercury DSP, whose outputs were connected to four channels of a counting board. Mercury spectra were calibrated with a 55Fe source and a Single Channel Analyzer (SCA) centered at 8 keV was defined for each channel, so that the Mercury generates a TTL signal for the counting board each time an X-ray source event is detected. Figure 4: Schema and photo of the surface scan, described in detail in the text. Four surface scans, covering three or four pixels, were made with steps of 50 µm and an exposure time of 1 sec, as shown in Figure 5. In each table position, only the number of counts at the 8 keV SCA were acquired. The four scans were combined in a global one. Pixel shapes matched with the molybdenum collimator hole shapes, which meant that the collimator was well aligned with the germanium sensor. During these tests, four defects (in pixels 1, 5, 8 and 9) were observed. When the X-ray beam impinged Deliverable no. D2.4 Page 6 of 13 at one defect, the sensor leakage current increased, shortening the reset period to a level that made the detector blind. For instance, when the X-ray beam impinged at the defect of pixel 9 at low X-ray flux of 60 cps), the leakage current was 90 fA. When the ICR increased to 4.6 kcps, the detector was blind and the leakage current was 90 pA, i.e., 1000 times higher. Figure 5: The four surface scans made with a collimated X-ray source and the linear combination. The four identified defects are highlighted by a red arrow. In two scans, only three pixels were scanned but a fourth pixel was connected to XIA-Mercury DPP. In those cases, the extra pixel saw some counts: up to 150 cps when the other three pixels were scanned, but no counts if the collimator edge was scanned, as illustrated in Figure 6. These counts are produced by crosstalk either at germanium sensor or at front-end board. The crosstalk ratio, defined as the counting ration between the extra pixel and the three others, has been quantified in 0.024-0.045% with these two scans. These values are lower than the maximum estimated crosstalk at front-end board (0.2%). Figure 6: Surface scan of pixels 1, 3 and 9 (top-left image), and the response of pixel 6 (bottom-left). The crosstalk ratios between scanned pixels and pixel 6 are shown on the right. Deliverable no. D2.4 Page 7 of 13 Finally, three pixels (1, 8 and 9) were consecutively connected to the oscilloscope, and their surface was scanned (100 µm step). For each point, signal waveforms were acquired for 2 seconds, the risetime distribution was generated and the mean value was calculated. This procedure allowed to generate the surface distribution of mean risetime, shown in Figure 7. The mean value is 80 ns, in agreement with SOLEIL result at +70 V (77 ns, see Figure 14 at Section 5). The distribution is quite uniform, except for some border effect where the risetime increases about 10 ns. Figure 7: Surface distribution of the mean risetime for pixels 1, 8 and 9. 4. Beam test at ESRF BM05 beamline XAFS-DET prototype was tested on 9-11th May 2025 at ESRF BM05 beamline, with the participation of members from DIAMOND, ESRF, INFN-Frascati, MAX-IV and SOLEIL. The detector (see Figure 8) was installed on a motorized table (in horizontal and vertical axis) facing an X-ray beam, collimated down to 35 µm by horizontal and vertical slits. Beam size was visualized by a Basler camera, situated just after the slits. The X-ray beam intensity was adjusted to 30 keV by aluminium filters. The seven central pixels of the detector were connected to a XIA-DXP-XMAP digital pulse processor, which generated an energy spectrum for each scan position (exposure time of 1 sec). Figure 8: Photo of XAFS-DET detector during the beam test at ESRF BM05 beamline. Deliverable no. D2.4 Page 8 of 13 The germanium sensor surface was scanned at four different X-ray (20, 30, 40 and 50 keV) with spatial resolutions of 25 µm (for 20 and 40 keV) and 250 µm (for 30 and 50 keV), as shown in Figure 9. The defects previously observed for pixels 1, 5, 8 and 9 in X-ray generator scan were visible. Apart from that, four extra defects were visible at higher X-ray beam energies (one in pixels 1 and 6 and two in pixel 5). Tests were performed to understand the nature and effect on the detector’s performance of these defects. Figure 9: Surface scans at four different X-ray energies (20, 30, 40 and 50 keV). The new four defects are highlighted by a red arrow. Two defects from pixels 5 and 9 were scanned for different bias voltages (+70 V, +150 V and +250 V), as shown in Figure 10. The defect dimensions shrank at high bias voltages. For instance, the width of pixel 5 defect reduced from 0.624 mm at +70 V, to 0.425 mm at +150 V and to 0.324 mm at +250 V. This observation has been explained by a physical model, which points out to a better charge carrier collection at high bias voltage indicating that the two defects are situated at a depth of 0.5-0.7 mm from the entrance window. The presence of eight defects in the LPP germanium sensor was communicated to the manufacturer, which are investigating this from the manufacturer perspective. Figure 10: Surface scans of two defects in pixels 5 and 9 and for different bias voltages (+70 V, +150 V and +250 V). The defect dimensions are indicated for each image. Deliverable no. D2.4 Page 9 of 13 The surface distribution of energy resolution at 40 keV was also evaluated and it is shown in Figure 11. The distribution is quite uniform for each pixel except around defect edges. There is a big dispersion in energy resolution observed between pixels (from 1.24 keV for pixel 4 up to 3.15 keV for pixel 6) due to the noise conditions being worse in the beamline than in the laboratory as optimal operation of the electronics box was to be open. Figure 11: Energy resolution across the scanned area for all pixels and for 40 keV X-ray beam. The mean value for each pixel is highlighted. The deadtime was evaluated from the relation between the input count rate (ICR) and output count rate (OCR) by varying photon flux for two XIA-XMAP peaking time values (6 µs and 1 µs) and for the three TETRA gains (high, medium and low). Photon flux was varied by the slit opening, as illustrated in Figure 12 for TETRA high gain. OCR peaks were at 70-80 kcps (for 6 µs) and 150 kcps (for 1 µs), while deadtime values were 13.3-14.4 µs (for 6 µs) and 2.4-3.3 µs (for 1 µs). These values showed that XAFSDET prototype is not ready to be operated in a high-flux X-ray environment.