Determination of homogeneity of the top surface deadlayer in an old HPGe detector

A collimated source of 241Am was scanned over the endcap of a 21 year old coaxial HPGe-detector that had spent about 75% of its life at room temperature (and the remaining time at 77 K). The detector response was recorded and used as a measure of the relative thickness of the top deadlayer. This thickness was not homogeneous and was thicker near to the outer surface of the crystal compared to the centre, which could be a result of increased diffusion of Li atoms during times the detector was kept at room temperature. The results were compared with two newer HPGe-detectors that proved to have homogeneous top deadlayers. ABSTRACT A collimated source of 241 Am was scanned over the endcap of a 21 year old coaxial HPGe-detector that had spent about 75% of its life at room temperature (and the remaining time at 77 K). The detector response was recorded and used as a measure of the relative thickness of the top deadlayer. This thickness was not homogeneous and was thicker near to the outer 20 surface of the crystal compared to the centre, which could be a result of increased diffusion of Li atoms during times the detector was kept at room temperature. The results were compared with two newer HPGe-detectors that proved to have homogeneous top deadlayers.


Submission Files Included in this PDF
To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE Homepage, then click 'Download zip file'.

Re. Submission of an article for Applied Radiation and Isotopes
Dear Editor, On behalf of all the co-authors, I hereby submit the an article entitled "Determination of homogeneity of the top surface deadlayer in an old HPGe detector"

INTRODUCTION
In recent years it has become increasingly important to better understand the shape and 35 thickness of deadlayers in HPGe-detectors, see e.g. Huy et al. (2007) and Aguayo et al. (2013). One reason is that Monte Carlo simulations using computer models of detectors are today often used for calculating efficiency transfer factors, coincidence summing correction and even absolute full energy peak efficiencies. The thicknesses of deadlayers in germanium detectors are necessary to know quite well to be able to create good computer models. It is 40 therefore a topic that has given rise to much discussion. At e.g. conferences organized by the ICRM (International Committee for Radionuclide Metrology) it has been discussed that a better name for deadlayer could be zombielayer as it is not excluded that for certain energies, an interaction of a gamma-ray inside a deadlayer can result in a pulse. Manufacturers have generally a good understanding of the thickness of the contact structures they produce, but 45 this is not necessarily exactly the same as the thickness of the deadlayer. Furthermore, one cannot assume (like is done in most computer models) that deadlayers are perfectly straight and parallel with crystal surfaces. To make things more complicated one has to take into account that the Li-atoms that are introduced for creating an n + contact structure can migrate by interstitial diffusion. This process will be enhanced drastically if the Ge-crystal is not 50 constantly kept cold at liquid nitrogen temperature (77 K). This paper describes the analyses of surface scans of the top deadlayer of a 21 year old HPGedetector (Detector-1), that was kept cold only 25% of its life-time. The same instrumentation as was used by Andreotti et al. (2014) for another HPGe-detector was employed for the 55 scanning. The paper also describes a measurement of a 13 year old detector (DET28) that has been kept cold for 98.7% of its life-time as well as a detector (Ge-8) with thin deadlayer that was also kept cold more or less all the time.

The HPGe-detectors
Three detectors were used in this study. The first detector, which was the main detector under 65 investigation, is operated by University of Hasselt and will be referred to as Detector-1. The second detector was also used by Andreotti at al. (2014) and is used here as reference to check the scanning system but also in order to provide information on the evolution of deadlayer characteristics with time for this detector, which has the name Ge-8 and is operated by JRC-Geel. The third detector is named DET28 and is operated by SCK•CEN in Mol, Belgium. It is 70 a standard coaxial HPGe-detector with a deadlayer thickness somewhat below 1 mm. The characteristics of the three detectors are given in Table 1. 2.2 The scanning system The scanning system, named Brady, was developed for the GERDA project (The GERDA 80 Collaboration, 2017) and was used for scanning the specially designed Ge-crystals used in the GERDA Phase II experiment. It is described in detail by Andreotti et al. (2013 and2014) and is now available for scanning other Ge-detectors. It is especially important to study the deadlayer variation of older detectors. A point source of 241 Am (5 MBq) was inserted in a 30 mm thick copper-collimator, which has a 1 mm diameter hole. It was carefully positioned 85 2 mm above the endcap of each HPGe-detector in the study. The source was moved in steps of 0.1 mm (0.2 mm for DET28) in an almost straight line across the endcap surface. Due to small imperfections in the mechanics, the line was somewhat bent and resulted in an 8% offset (i.e. bending-radius of 42 cm) compared to a perfectly straight line. Three line-scans covering 76.1 mm each (the endcap diameter), and oriented in three different directions, each 90 rotated by 60, were performed on Detector-1. Two line-scans covering 77 mm each (slightly less than the endcap diameter of 82.5 mm), oriented in two directions differing by 20, were performed on Ge-8. Two line-scans covering 76.1 mm each (the endcap diameter), separated by 60 were performed on DET28. At each step, data was collected for 3 minutes resulting in a 38 hour acquisition time for each line-scan (19 hours for DET28). A fourth measurement 95 was performed by scanning DET28 with a 0.4 mm (28 mm diameter) tantalum-disc placed at the centre of the detector window. The aim was to determine the lateral resolution of the scanning system. The measurement set-up is located in the underground laboratory HADES (Andreotti et al., 2011). The main reason for this was to minimise the exposure to cosmic rays for the GERDA crystals but it is also advantageous when detectors from HADES and other 100 underground laboratories need to be scanned.

Monte Carlo simulations
The Monte Carlo simulations of Detector-1 and Ge-8 were performed using the EGSnrc code 105 with the add-on code "hpge3" (Lutter et al., 2017) for simulating HPGe-detectors and gammaray emitting radionuclides. The EGS package (and formerly EGS4) has been used at JRC-Geel since 1998 to simulate the response of HPGe-detectors (Gasparro et al., 2008). The Monte Carlo simulations of DET28 were performed using the EFFTRAN software (Vidmar, 2005). A crucial parameter when setting up computer models of detectors is the deadlayer 110 thickness. In the present work the thickness of the top deadlayer as a function of radial position was tested (for Detector-1) for a number of different configurations to see which configuration best replicated the scanning-results. For DET28 the simulation was solely used to determine the relative thickness variations of the deadlayer.  Figure 1 shows the results of the three line-scans performed on Detector-1. The y-axis shows 120 the raw-data, which is reported in counts per 3 minutes (the acquisition time per data point). Figure 2 shows the same as Figure 1 but for Ge-8. Figure 3 shows the same as Figure 1 but for DET28. The rising and tailing slopes of the scans provide some information. The lateral distance of the 12%-88% count-rate at the rising and tailing slopes, was 3.05 ± 0.16 mm for the three line-scans of Detector-1. The uncertainty is the standard-deviation of the six slopes 125 in the three plots in Figure 1. This value is a convolution of (i) the resolving power of the system, (ii) the side deadlayer thickness including imperfections and variations (iii) charge collection inefficiencies in the corner and (iv) the possible small tilting of the crystal. The same value for Ge-8 was 1.01±0.03 mm and for DET28 1.23±0.03 mm. 130 The flat profile of the scan of Ge-8 indicates that very change of the deadlayer has taken place and as the scan-result agrees with the previous scan made 2 years earlier (Andreotti et al., 2014), we conclude that the scanning system is giving the same response as during this previous study. Since the n+ contact of Ge-8, presumably, does not contain Li-atoms one can assume that t should remain stable and is therefore useful as reference. 135 Figure 4 shows a scan (using DET28) over an edge of a 0.4 mm thick piece of tantalum. The slopes (12%-88%) at the edges of the Ta are 0.78±0.01 mm. The same result exactly was 140 obtained by fitting a cumulative distribution function. This value is a direct measure of the resolving power (FWHM) of the system 1 . Subtracting in quadrature the FWHM of the system-resolution (0.78 mm) results in side edges of 2.95 mm, 0.64 mm and 0.95 mm, respectively for Detector-1, Ge-8 and DET28. This value is a convolution of the effects of the thickness of the side deadlayer, the alignment of the crystal and charge collection 145 inefficiencies in the corner.

Lateral resolution of the scanning system
A radiograph of Ge-8 reveals that there is no discernible tilting of the crystal so she sharp slopes of Ge-8 are indicative of a side deadlayer (at least the top 10 mm of the crystal) very close to the nominal value of 0.6 mm.
1 After finalising the scanning of the detectors described in this paper, the lateral resolution of the scanning system was improved to 0.706±0.015 mm by using a slightly smaller aperture.

Computer models based on the scans
The obtained line scans can be used to build good computer models of the detectors. The top deadlayer is one of the most important parameters in a computer model of a HPGe-detector. 155 As the line-scans of detector Ge-8 and DET28 were perfectly straight it was trivial to introduce straight top deadlayers in the computer models. Also in the corners, a straight top deadlayer combined with a straight side deadlayer gave a good result. For Detector-1 the situation was completely different. An iterative procedure (mainly trial and error) was used to find a model of the deadlayer that well replicated the experimental results. The resulting 160 model of the top deadlayer of Detector-1 is shown in Figure 5. The simulation results that this model generates (for 241 Am) are shown in Figure 6 together with the absolute and relative difference between the model and the measurements results. It was not possible to generate a good response unless making the model quite detailed towards the corner with steps of uneven lengths.

DISCUSSION AND CONCLUSIONS
The "reference-detector" (Ge-8) has a micro-metre-sized top deadlayer which is probably not created using Li-diffusion. Therefore is the scan of DET28 important as it proves that a 170 detector with a thick, i.e. mm-sized, deadlayer can have a flat profile when kept cold.
The general feature of the top deadlayer profile of Detector-1 is that the top deadlayer is about a factor 1.5 thicker at the rim compared to near the centre. Furthermore, there is a clear "dip" in the response at the centre, which indicates a slightly thicker deadlayer at the centre of the 175 crystal compared to 5 mm away from the centre. A tentative explanation of the shape is that Li-diffusion follows the crystal orientation and that diffusion from the side deadlayers contribute to the top deadlayer thickness near the rim. The model in Fig. 5 can generate results that replicate the scanning results very well. It is, however, possible to continue to adjust the model so that the difference between the model and the measurement is further 180 minimised but this makes little sense as it is still a model and not exactly reality. The impact of introducing the inhomogeneous deadlayer of Detector-1 compared to using a model with a straight deadlayer profile can be exemplified by a few cases of common types of samples and radionuclides given in Table 2. 185 One must bear in mind that although a Monte Carlo simulation using an advanced computer model can replicate the obtained measurement results rather well, it is still only a model. Reality can be different and one will have to study many different geometrical configurations of sources and at a wide range of energies before making strong claims about the true configuration. Still, for calculating correction factors for efficiency transfer and true summing 190 a computer model (even a crude one) is very useful. Future studies should focus on the effect of different energies on deadlayer structure. As the interaction probabilities of photoelectric effect, Compton scattering and pair-production changes with gamma-ray energy, it is likely that they also affect the response of the detector. 195 HPGe-detectors that are well taken care of can operate nicely for several decades also if kept uncooled from time to time. Therefore it is important to understand and study the extent by which deadlayers in Li-diffused contacts change and how it affects the quantitative results obtained by a specific detector.