Characterisation of a pure LaCl3-based detector for proton therapy application

University of Seville Master in Nuclear Physics Characterisation of a pure LaCl3-based detector for proton therapy application Georgina Xifra Goya 17th July 2023 Supervisors: Enrique Nácher González Marcos Martínez Roig University supervisor: Manuel García León In collaboration with: Grupo de espectroscopía gamma y de neutrones (IFIC-CSIC) Abstract Proton therapy is an increasingly used technique for cancer treatment worldwide. However, the uncertainty in proton range verification requires the use of wide safety margins in proton doses, limiting its full potential. This project aims to test the feasibility of using a pure LaCl3scintillator, in coaxial configuration with a proton beam. The goal is to determine the position of the Bragg peak with a substantial improvement in its spatial uncertainty by detecting the gamma-rays and neutrons emerging from the patients. The concept relies on the solid angle effect, where closer sources result in higher counting rates in the detector. Pure LaCl3scintillators, currently not available in the market, offer excellent discrimination between neutrons and gamma-rays, surpassing the limitations of current detectors. Moreover, they enable double real-time verification of the proton range. To assess the sensitivity of the pure LaCl3detector, several studies have been conducted aiming to measure the variation in the count rate in response to millimetre-scale displacements of the Bragg peak. Resumen La protonterapia es una técnica cada vez más utilizada para el tratamiento del cáncer. Sin embargo, la incertidumbre en la verificación del rango de los protones implica la necesidad de utilitzar ámplios márgenes de seguridad en las dosis, limitando su potencial total. Este proyecto tiene como objetivo comprobar la viabilidad del uso de un centelleador de LaCl3 puro, en configuración coaxial con un haz de protones. Se busca determinar la posición del pico de Bragg, con una mejora sustancial en su incertidumbre espacial, mediante la detección de los rayos gamma y neutrones emitidos por los pacientes. El concepto se basa en el efecto del ángulo sólido, donde las fuentes más cercanas resultan en mayores tasas de conteo en el detector. Los centelladores de LaCl3puro, que no se encuentran actualmente en el mercado, ofrecen una excelente discriminación entre neutrones y rayos gamma, superando las limitaciones de los detectores actuales. Además, permiten la doble verificación en tiempo real del rango de protones. Para evaluar la sensibilidad del detector de LaCl3puro, se han llevado a cabo varios estudios donde se pretende medir la variación en la tasa de cuentas ante desplazamientos milimétricos del pico de Bragg. 1 Acknowledgements I would like to express my gratitude to my two thesis supervisors, Enrique Nácher González and Marcos Martínez Roig. Their invaluable mentorship, expertise, and unwavering support have been crucial to the successful completion of my thesis. I feel compelled to thank, once again, Phan Quoc Vuong and Hongjoo Kim from the Department of Physics at Kyungpook National University for letting us have the privilege to study a non commercial pure LaCl3scintillator to conduct my master thesis and develop the PRIDE project. I wish to extend my appreciation to my parents and brother for their encouragement in pursuing my dreams from a young age, as well as their continuous support throughout my bachelor’s and master’s studies. I would like to express my heartfelt gratitude to my friends for their ability to understand my limited availability and provide me with much-needed moments of disconnection when necessary. I would also like to thank my partner for his encouragement, understanding, and valuable insights during this journey. Thank you all for being there with me. 2 Contents 1 Introduction 4 2 Motivation and objectives 7 3 Methodology and content development 7 4 Testing the LaCl3with two different PMTs 9 4.1 Composition of the pure LaCl3scintillator .................. 9 4.2 Digitisermainfeatures ............................. 10 4.3 First test at the IFIC laboratory . . . . . . . . . . . . . . . . . . . . . . . . 12 4.3.1 Fine-tuning the digitiser parameters . . . . . . . . . . . . . . . . . . 13 4.4 HAMAMATSU PMT: R1924A-700 . . . . . . . . . . . . . . . . . . . . . . 16 4.4.1 Discussion of the chosen method to compute the test the short gates 17 4.4.2 Optimal voltage and gate parameters . . . . . . . . . . . . . . . . . 21 4.5 HAMAMATSU PMT: R6233-100 . . . . . . . . . . . . . . . . . . . . . . . 22 5 Measurements of LaCl3response with 10 MeV proton beam on CsI target 24 5.1 Simulation.................................... 25 5.1.1 Simulation framework . . . . . . . . . . . . . . . . . . . . . . . . . 26 5.1.2 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 5.1.3 Simulation results adding a neutron moderator . . . . . . . . . . . . 34 5.2 Protonbeamresults .............................. 37 5.2.1 PSD discrimination in measurements of 5 minutes . . . . . . . . . . 38 5.2.2 Variation of the neutron flux with polyethylene . . . . . . . . . . . 44 6 Conclusions 48 7 Next-steps 49 8 References 50 3 1 Introduction In recent years, proton therapy has emerged as a promising technique for cancer treatment, offering precise targeting of tumours while minimising damage to surrounding healthy tissues. It is possible to accurately control the radiation dose deposition inside the patient’s body thanks to the inherent physical properties of the protons. Nonetheless, precise range verification is crucial to maximise the potential of proton therapy. It involves verifying the depth of proton penetration in tissue to guarantee both treatment effectiveness and patient security. Therefore, the ability to verify the range of protons in real-time plays a vital role. Proton therapy is a technique employed for the treatment of cancerous tissue by using proton beams instead of X-rays, distinguishing it from conventional radiotherapy. In order to fully leverage the benefits of proton therapy over traditional X-ray therapy, precise positioning of the Bragg peak within the entire tumour being treated is crucial [1]. The use of charged particles allows to deposit the maximum amount of energy at the end of the trajectory of the particles, which is known as the Bragg peak. The position of this peak depends on the beam’s initial energy. One major advantage of protons is having a sharp Bragg peak and a distal falloff near their maximum penetration depth. This enables the precise targeting of the maximum dose at a specific location, resulting in an impact on the cancerous tissue while minimising the effect on the surrounding healthy tissue. Fig. 1 shows the dose profile with the penetration depth of protons and photons in water into the tumour region. Fig. 1. Comparison between deposited dose as a function of the penetration depth for different initial beam energy for photons and protons. Image taken from reference [2]. 4 Traditional X-ray radiotherapy has a higher chance of damaging the healthy tissue surrounding the affected area since they deposit the maximum dose at shallower depths. As they penetrate further, the dose decreases gradually until the X-rays lose all their energy or, ultimately, exit the body. Because of having a far higher selectivity than traditional radiotherapy, proton therapy may be a very good alternative for treating tumours in anatomically sensitive regions like the brain or spinal cord. To ensure the safety and success of the treatment, it is essential to continuously monitor the therapy by verifying the position of the Bragg peak. This can be achieved by measuring, in real-time, the spatial distribution of the dose. No accepted standard or commercial technique is currently used in all proton treatment facilities for proton range verification. Hence, when devising the treatment plan, prudent safety margins are implemented for the proton dosage. The main goal of this work is to test a detector, to be used in coaxial configuration with the proton beam, to infer the position of the Bragg peak from the detection of γ-rays and neutrons emerging from the patient. The method was proposed theoretically in [3], for the particular case of γ-rays, and it is based on the solid angle effect: the closer the emitting source (in this case the Bragg peak), the higher the counting rate in the detector. We want to follow the same approach but making use of the simultaneous detection of γ-rays and neutrons. Therefore, we need a scintillating crystal, in coaxial configuration, with a good Pulse-Shape Discrimination (PSD) of γ-rays and neutrons, also known as neutron-γ discrimination. CLYC scintillators (Cs2LiYCl6), are widely used and commercialised due to their advantageous features. First of all, their have a good neutron-γdiscrimination due to their PSD capability. CLYC detectors also have other notable features such as high efficiency for γ-rays and neutrons, excellent resolution, compact size and low sensitivity to temperature and humidity. Although CLYC scintillators offer advantages, they cannot discriminate between the reaction products within the crystal. The detection of fast neutrons, as well as the discrimination of (n,p) and (n,α) reactions in the crystal, are valuable features exhibited by other scintillators as further explained. Nowadays, La-halide scintillators, are commercially accessible. They have proven to be highly effective in the field of radiation measurement, especially in low energy nuclear spectroscopy experiments. Since Lanthanum has a high atomic number (Z=57), combined with a high density of the crystal, such as the one of LaCl3,ρ= 3.85 g/cm3, could permit to detect high-energy γ-rays. Nonetheless, La-halide detectors have a self-activity which reduces the detector sensitivity [4], and it can be activated in presence of a radioactive 5 background. Currently, LaCl3can only be commercially found doped with cerium (LaCl3:Ce). This scintillator shares similar characteristics with CLYC. Their properties have been thoroughly studied in [4]. Ce-doped lanthanum scintillators offer an outstanding trade-off between efficiency, energy resolution (3.3% FWHM at 662 keV) and light yield (50 photons/keV). These values are the ones corresponding to commercially optimised cerium doping levels [5]. With their sub-nanosecond time resolution, LaCl3:Ce detectors have the potential to serve as a viable alternative to HPGe detectors for γ-ray studies, as stated in [6]. The exceptional time-of-flight measurement, enabled by the sub-nanosecond time resolution, is typically utilised for discriminating between neutrons and γ-rays and for rejecting background signals. The concentration of cerium also plays an important role regarding PSD. At higher concentrations of cerium, the energy resolution improves to the expense of the PSD discrimination. When decreasing the concentration, the PSD discrimination gets better. Due to the improvement in PSD with the decrease in cerium doping, we could wonder if using pure LaCl3, that is, eliminating the doping, could yield better results. However, pure LaCl3crystals are not commercially available yet. Pure LaCl3offers one of the finest neutron-γPSD capabilities of any of them, including the CLYC and the Ce-doped LaCl3scintillators, according to [5]. This is achieved at the cost of getting reduced values of light yield (34 photons/keV), energy resolution (4.7% FWHM at 662 keV) and a slower decay than Ce-doped ones [7]. Apart from obtaining and excellent PSD discrimination, we could also do neutron spectroscopy using pure LaCl3crystals, since this scintillator is able to discriminate between proton and α-particles. It is viable because fast neutrons react with 35Cl isotope, the richest concentration of isotope present in the crystal, through two reactions with a significant cross-section:35Cl(n,p)35S and 35Cl(n,α)32P [8]. Up-to-date, several solutions have been proposed for real-time range verification problems. We want to propose pure LaCl3crystals in coaxial configuration for proton range verification, due to their neutron-γseparation, the optimum energy resolution, and the time response [5]. Proton range verification is the key aspect to reduce the margins of uncertainty introduced in proton therapy treatments, such as the uncertainty on the exact position of the Bragg peak. 6 use a neutron source. In our case 252Cf, with an activity of approximately 200 Bq. Fig. 5. PSD spectrum where we can see clearly the separation between the two peaks in the PSD ratio. In this plot the PSD is defined as PSD=QP/(QP+ QD). Image taken from [5]. Our main objective is to optimise the separation between the two peaks in the PSD ratio, as shown in Fig. 5. To do so, we defined the Figure of Merit (FOM) to measure the separation between two peaks in the PSD projection on the x-axis. We define it as the difference between the mean values of the two adjusted Gaussian functions representing the each peak, the alpha peak and the gamma peak, divided by the sum of their respective peak widths (FWHM): FOM = |Mα−Mγ| FWHMα+ FWHMγ .(2) As we can see, FOM is defined positive. In order to consider a detector’s PSD value acceptable to discriminate two particle types, it is typically required a FOM α/γ value higher than 1.5 [5]. The optimal FOM values achieved using LaCl3are usually around 2.5. 4.3.1 Fine-tuning the digitiser parameters Fine-tuning the CoMPASS configuration parameters is essential in order to achieve a high FOM value and therefore, a good PSD discrimination. The configuration parameters, depicted in Fig. 4, have a crucial role to play in this process. 13 Input Discriminator QDC Record length 14975 ns Mode leading edge Gain 160 fC/(LSBxVpp) 40 fC/(LSBxVpp) Pre-trigger 192 ns Threshold 100 LSB 25 LSB Gate 10000 ns Polarisation Neg. Trigger holdoff 1024 ns Short gate 200 ns N samples baseline 256 CFD delay 4 ns Pre-gate 100 ns Fixed baseline value 0CFD fraction 75% Charge pedestal 1024 LSB DC offset 20% Input Smooth Disabled Vpp 0.5 V Tab. 1. Parameters selected for the R1924A-700 PMT with a HV of -900 V. In blue, the parameters that change when exchanging the PMT for the 6233-100 model. The parameters that change with the HV are the long and short gates and with each PMT, the gain. Tab. 1 presents the optimised values for these parameters, categorised into input, discriminator, and QDC (Charge-to-Digital Converter). Additionally, Fig. 6 illustrates some of the gathered parameters. In the input category, there are parameters regarding the input of the signal. Initially, the record length refers to the fixed length within the visually defined acquisition window. The pre-trigger refers to a specific portion of a recorded signal that is captured before the triggering event. The polarisation must be the one that the PMT supports, in this case, negative. We have chosen to calculate the baseline dynamically, adjusting the number of samples at which the baseline is recalculated with the N samples baseline parameter. Thus, the fixed baseline value is equal to zero. The DC offset serves for the adjustment of the baseline voltage level of the captured signal. It is useful when there is a significant DC component, covering the full width of the pulse of the waveform within the dynamic range. It is expressed as the percentage of the full-scale range. Vpp voltage from peak-to-peak, quantifies the amplitude or voltage range of the captured signal. 14 Fig. 6. Diagram that represents the DPP-PSD parameters previously explained for one acquisition window. Picture taken from the User Manual UM5960 of CoMPASS, "Multiparametric DAQ Software for Physics Applications". In the discriminator section, there is firstly the choice of the triggering acquisition of pulses mode. We have chosen leading edge. There is also the threshold and the trigger holdoff, which is the period of time after the trigger that the system inhibits triggering again. The parameters regarding the constant fraction discrimination (CFD) do not affect when choosing leading edge. Finally, the input smooth is to reduce the fluctuations in the signal, if necessary, in order to enhance the visualisation of the signal profile. In the charge-to-digital converter (QDC) section, there are parameters regarding the charge integration gates. The gain allows to reescale the signal charge, a uselful feature to adjust the signal to the full-scale range. The gain units are fC/(LSBxVpp), where fC are femtocoulombs, LSB (Least Significant Bit) can be converted to volts using 1 LSB= Vpp/2Nbit conversion. The pre-gate is a parameter that sets the starting position of the gate and short gate before the trigger signal. The short and long gates values are the ones that refer to the charge integration. The charge pedestral is set by default and it is useful for energies close to zero. 15 To analyse the effect of varying the short gate value on the FOM value, we need to plot the PSD ratio spectrum. Ideally, we should observe distinct peaks corresponding to γ-rays and α-particles. In an optimal scenario, we might even observe an additional peak corresponding to protons. Using the 252Cf weak source and considering our current counting rate, we will determine the optimal values by adjusting the peaks of α-particles and γ-rays and calculating the FOM value. In the following, I will show this analysis and FOM optimisation for each particular PMT. 4.4 HAMAMATSU PMT: R1924A-700 In order to select the desired data, we can generate a plot of PSD vs Energy in the ROOT framework and then apply cuts on the data points corresponding to α-particles and γ-rays. There are two primary approaches to accomplish this: TCutG and energy cuts. On the one hand, the TCutG method allows us to define a region of interest on the plot by graphically selecting a list of points. This region can encompass the specific range of energies associated with the α-particles and γ-rays, effectively isolating them from the rest of the data. On the other hand, the energy cuts method is more straight-forward because it involves setting vertical boundaries on the energy axis to isolate and visualise the γ-rays and α-particles distinctly, filtering out noise or data below a certain energy value. Both methods offer different ways to select and analyse the desired data points, enabling us to focus on the α-particles and γ-rays in our study. It is important to mention that calibration was not applied in this case because any modifications to the gate values (prompt and delayed) and the HV directly affect the calibration. Therefore, calibration should be performed once the exploration of optimal parameters to achieve the best FOM is completed. We have calibrated the gate parameters using three different high voltage supply values (-900 V, -1000 V, and -1100 V) for the PMT. We selected these values based on the PMT specifications, considering the maximum voltage allowed (-1200 V) between the anode and cathode of the PMT. We observed the PMT signals in the "Waves" mode of the digitalizer, which functions as an oscilloscope, for each high voltage setting. During this analysis, we determined that the valid ranges for integrating the signal’s descent are between a short gate of 200 ns, which precisely aligns with the point before the tail begins, and 500 ns. To explore the impact of different short gate values, we tested four options: 200 ns, 300 ns, 400 ns, and 500 ns. As the voltage increases, both the integrated charge and the total length of the signal 16 increase, thereby affecting the delayed gate. Consequently, the optimal gate parameters may vary for each voltage value. In the "Waves" mode, we were able to adjust the delayed gate values for each voltage, as demonstrated in Tab. 2. HV (V) Short gate (ns) Delayed gate (ns) -900 200 300 400 500 12000 -1000 200 300 400 500 14000 -1100 200 300 400 500 16000 Tab. 2. Summary of the 12 configurations tested for the R1924A-700 PMT. 4.4.1 Discussion of the chosen method to compute the test the short gates To illustrate the findings obtained from the methods discussed earlier, we will consider the example of HV of -900 V, a short gate of 200 ns and a delayed one of 12000 ns. 4.4.1.1 First method: graphical cuts using TCutG For this method, we plot the PSD ratio in front of the energy measured for each event. Fig. 7. PSD in front of the EnergyPSD in front of the Energy for a HV of -900 V and a short gate of 200 ns. Three graphical cuts were implemented, corresponding to the gamma, proton and alpha regions. Using the TCutG functionality, we delimited the regions in which we know that αparticles, proton and high-energy γ-rays are, as shown in Fig. 7. The violet region represents the (n, α) reactions in the crystal, while the green region corresponds to the 17 (n,p) reaction, both products of the reaction of neutrons with 35Cl. The high-energy gamma region is delimited in red. To calculate the FOM value, we have depicted the projection of the the three graphical cuts delimited in Fig. 7, including and excluding the proton one. We can see that the alpha and gamma PSD value can be well distinguished visually without protons. We can observe that the delimited region for protons is closely adjacent to that of α-particles, making it challenging to distinguish them clearly in the PSD spectrum. Even though, we can fit the alpha and gamma peak with two Gaussian functions in both cases, obtaining a similar fitting values and thus, a similar FOM value. Using the optimised parameters that adjust best the alpha and gamma peaks, we can calculate the FOM and its corresponding error. We can use the Fit Panel tool incorporated in ROOT software to fit the data. As we can see in Fig. 8, when not including the proton graphical cut, the region between the alpha peak and the gamma one flattens, and the two peaks are perfectly discriminated using PSD. When including the three graphical cuts, a bump appears in the region between the alpha and gamma peaks, linking both curves. The proton peak is not clearly resolved, and therefore, with these parameters (HV = -900 V, short gate = 200 ns), we do not have a good α-proton or γ-proton discrimination. Fig. 8. Comparison between the PSD spectrum with (blue) and without (green) considering the proton region. The red line corresponds to the fitting of the alpha and gamma peaks. Nevertheless, including or excluding the proton cut does not affect the fitting of the 18 gamma and alpha peaks. This leads us to consider implementing a cut at low energies only, simplifying the analysis and facilitating the determination of parameters. Since the charge integral depends upon the integration gate and the HV supply values, the positions of graphical cuts also vary for each case. 4.4.1.2 Second method: vertical cuts To compute the FOM value, we need to repeat the previous procedure exchanging complex graphical cuts by a simple vertical cut. As we can see in Fig. 9, there is a vertical red line in E=1000 channel. We have set this lower limit because for low-energy γ-rays, there is a higher number of counts (yellow) than in the other part of the spectrum, where high-energy γ-rays and alpha regions (greenblue). This value coincide with the start of the alpha region, delimited in violet in Fig. 7 in 4.4.1.1, First method. Fig. 9. PSD in front of the Energy for a HV of -900 V and a short gate of 200 ns. A vertical cut was implemented in E=1000 channel. After applying the cut, we have to make a projection of the PSD ratio in the x-axis, representing the number of events as a function of the PSD. By doing so, we can adjust the two peaks that appear, the alpha one with a lower PSD ratio and the gamma one, with the higher value as done in Fig. 10. 19 Fig. 10. PSD spectrum for a HV of -900 V and a short gate of 200 ns. We can see that there are two peaks: the small peak around 0.885 PSD, which corresponds to the α-particles and the higher one, at 0.925 PSD consisting of the γ-rays. For a better visualisation, we can set the y-axis to a logarithmic scale after fitting Fig. 11. PSD spectrum for HV of -900 V and a short gate of 200 ns in logarithmic scale in the y-axis. We can see that there are two peaks: the small peak around 0.885 PSD, which corresponds to the α-particles and the higher one, at 0.925 PSD consisting of the γ-rays. In the previous figure, Fig. 11, we can see that there is a bump between the two peaks, 20 which we can infer corresponds to the protons. In Fig. 9, we observe a banana-shaped region below the gamma peaks, very close, which indicates the presence of protons. It is evident in this way why there is not enough clear separation. Once we have adjusted the two Gaussian functions to the PSD graph, we need to use the fitting parameters to compute the FOM α/γ value and its error. Type of cut Including the proton region FOM α/γ Graphical No 2.26 ±0.03 Graphical Yes 2.22 ±0.03 Vertical Yes 2.14 ±0.03 Tab. 3. Comparison between the FOM α/γ values obtained with graphical cuts, including and excluding protons, and with the vertical low energies cut for a high voltage of -900 V, a short gate value of 200 ns and a delayed gate of 12000 ns. Comparing the FOMs considering those three from Tab. 3, provides a comprehensive evaluation of the discrimination strategies employed. This analysis enables us to determine that the optimum FOM value is obtained through a more refined selection of regions, with graphical cuts in the PSD vs Energy plots where particles are present. However, we have opted to analyse the various potential short gate values with the vertical cut method, due to having less uncertainty in the number of counts included in the cut region. 4.4.2 Optimal voltage and gate parameters As we have just explained the procedure followed to obtain the FOM value step-by-step, the FOM values calculated for the 12 configurations under study are listed in the Tab. 4. The measurements have been taken in for two hours, obtaining enough statistics. HV= -900 V HV= -1000 V HV= -1100 V SG (ns) FOM α/γ SG (ns) FOM α/γ SG (ns) FOM α/γ 200 2.14 ±0.03 200 2.30 ±0.11 200 2.28 ±0.09 300 2.15 ±0.10 300 2.28 ±0.10 300 2.36 ±0.09 400 2.33 ±0.08 400 2.55 ±0.10 400 2.49 ±0.11 500 2.14 ±0.09 500 2.51 ±0.09 500 2.26 ±0.04 LG = 12000 ns LG = 14000 ns LG = 16000 ns Tab. 4. FOM α/γ values of the 12 configuration parameters for the R1924A-200 PMT. From Tab. 4, we can see that the best FOM α/γ value is the one corresponding to a HV of -1000 V and a short gate of 400 ns. Since it is the best possible configuration regarding the FOM value compatible with this PMT, it would be the best choice of parameters 21 for a further analysis in the second CMAM experience. But another important feature must be taken into account: the possible and useful resolution of the proton peak. Using a higher voltage supply value, we can see that the FOM value corresponding to a short gate of 400 ns has a very similar value to the ones previously discussed. Moreover, it can discriminate the proton peak from the other two, as can be seen in Fig. 12. As stated before, the more favourable scenario where we can discriminate between three types of particles (namely α-particles, γ-rays and protons) is a HV of -1100 V and a short gate of 400 ns. In this particular case, we have obtained that the FOM value for α/proton is 1.33 ±0.14, while for γ/proton, the FOM value is 1.55 ±0.19. Fig. 12. PSD spectrum for HV of -1100 V and a short gate of 400 ns in logarithmic scale in the y-axis. The fitted peaks correspond to α-particles, protons and γ-rays, respectively. 4.5 HAMAMATSU PMT: R6233-100 We aimed to compare measurements using two different PMTs while keeping the measurement conditions consistent. Due to the availability of PMTs for simultaneous experiments, we used the R6233-100 model for the initial measurements in the coaxial mode. For the ongoing measurements at CMAM, we are employing the R1924A-700 model. The different features of each PMT are gathered in Tab. 5. 22 Fig. 17. Total energy deposition and its four main contributions: proton and 35S contributions from the 35Cl(n, p)35Snuclear reaction, and α-particle and 32Pform the 35Cl(n, α)32Preaction. In order to examine the results, we will investigate several target distances. Specifically, positions of 160 mm, 170 mm, and 180 mm. One of the key findings of interest is the total deposited energy and its individual contributions. As becomes evident from Fig. 17, the proton and α-particle contributions exhibit a broader energy distribution that spans entirely the shown energy spectrum than 35Sand 32P. This disparity arises from the energy and momentum conservation. The residual nuclei have a greater mass compared to the protons or α-particles, resulting in less kinetic energy during the reaction and in consequence, their energy deposition is lower. 29 (a) Total number of events within the energy deposition range with proton and α-particles contributions. (b) Total number of events within the energy deposition range with 35S and 32P contributions. Fig. 18. Comparison of the total deposited energy with its main contributions for the beam-detector distances of 160 mm and 180 mm. In (a), including light particles and in (b), the residual nuclei. The solid lines in both graphs correspond to a distance of 160 mm, while the dashed lines represent a distance of 180 mm. This comparison allows for an analysis of the different particle contributions at these extreme distances. As depicted in Fig. 18, increasing the distance between the detector and the beam results 30 in a decrease in the total number of counts. This is primarily due to the solid angle effect: the closer the source and the detector are, the greater the solid angle and, consequently, the higher number of detected particles. The profiles of the various curves remain largely similar, though with a lower count rate. (a) Hits detected in x direction. (b) Hits detected in y direction. Fig. 19. Hits detected in the x and y direction for three source-detector distances: 160 mm, 170 mm and 180 mm. 31 Fig. 20. Hits detected in the z direction for three source-detector distances: 160 mm, 170 mm and 180 mm. Based on Fig. 19, it is evident that the number of hits in the x and y directions exhibit symmetry. It is important to note that the z-axis aligns with the beam direction, as depicted in Fig. 15. When the solid angle of the detectors decreases, indicating an increased distance between the beam and the detector, the number of hits along both the x and y directions decreases, as expected. We have also plotted the hits in the z-direction, Fig. 20, revealing that the majority of hits occur towards the rear side of the crystal, where the diameter is bigger. This is primarily because the larger diameter provides a greater surface area for detection, resulting in a higher number of hits. The same effect caused by the solid angle, which was previously explained, also applies to the variation in distance. To provide a concise overview, we have represented plots illustrating the number of detected events as a function of distance. These plots allow us to observe the overall trend for each of the products’ contribution. In Fig. 21, we can see that the number of events detected in the LaCl3detector decrease when increasing the distance between the detector and the beam. This is essentially due to the solid angle effect. Consequently, a lower number of counts is detected by the detector, which is in line with our expectations. 32 (a) Total LaCl3events as a function of the distance source-detector. (b) Main contributions to total LaCl3events as a function of the distance source-detector. Fig. 21. The top graph, (a), displays the total number of events within our distance range, while the bottom graph, (b), illustrates the main individual contributions. We have separated them into two graphs because the number of events differs significantly. Although all data points include y-error bars, they are negligible in size. In Fig. 21b, we can see that the products of the two reactions,35Cl(n,p)35S and 35Cl(n,α)32P. It is evident that the primary contributors to the number of events are protons, followed by 35S, both originating from the same reaction. The observed phenomenon can be attributed to the disparity in cross sections and energy threshold at which reaction channels open, as illustrated in Fig. 16. The 35Cl(n,p)35S reaction has a higher cross section and a 33 lower energy threshold at which the channel opens, for neutron incoming energy around 1 MeV. In contrast, the 35Cl(n,α)32P reaction opens for neutrons with incoming energy around 3 MeV. These two factors are the ones contributing to this phenomenon. This means that this particular reaction exhibits a higher number of events and, consequently, a greater probability. When comparing the other two products, αand 32P, their contributions appear to be quite similar. 5.1.3 Simulation results adding a neutron moderator Additionally, we conducted a simulation using a 5 cm thick (z-direction) polyethylene block of 8 cm in height (y-direction) and 6 cm in width (x-direction) to observe its impact on the neutron counting rate at three varying distances, as we can see in Fig. 22. Neutrons undergo thermalisation through elastic collisions with hydrogen atoms in polyethylene, resulting in the production of thermal neutrons. Upon reaching our detector with reduced energy, we anticipate a decrease in the number of counts in the α-channel due to the Q-value of the (n,α) reaction in LaCl3. Additionally, the proton channel is also diminished. This opens the possibility to use polyethylene around our crystal to thermalise and therefore reduce the contribution of neutrons coming from other sources than the beam. We need to consider that the LaCl3scintillator is sensitive to fast neutrons. In proton therapy facilities, the scattered neutrons in the walls often contribute significantly to the background. To mitigate this, we can reduce the background in our setup by shielding the detector with polyethylene. Fig. 22. A block of polyethylene with a thickness of 5 cm is shown as part of the detector setup in this illustration. Because we have not depicted the PMT device, the block appears to be large in comparison to the scintillating crystal. We can see in the right picture that the separation between the scintillator and the polyethylene block is 1 cm. 34 (a) Total number of events for each distance with and without polyethylene. (b) Main contributions of the total number of events in the polyethylene case. Fig. 23. The top graph, (a), displays the total number of events within our distance range with and without polyethylene, while the bottom graph, (b) illustrates their individual contributions for measures with (solid lines) and without (dashed lines) a block of polyethylene. We have separated them into two graphs because the number of events differs significantly. Although all data points include y-error bars, they are almost negligible in size. We have simulated the distances studied with polyethylene in the CMAM facility: 165 mm, 170 mm and 175 mm. As we can see in Fig. 23, the global trend of the plots of the total number of events and its contributions for each explored distance is exactly the 35 same with and without polyethylene. We can see that the number of events registered is higher in the non-polyethylene case than the one with polyethylene. Fig. 24. Deposited energy contributions of protons and α-particles at a source-detector distance of 165 mm with and without polyethylene. To be understood as an example for all simulated distance values. Regarding the energy deposition, we can see in Fig. 24 that the number of events for each energy deposition is lower when using a polyethylene block for protons and α-particles. This is due to the fact that these particles are the products of the neutron reactions (n,p) and (n,α). When neutrons are moderated, less neutrons interact with the 35Cl present in the LaCl3scintillator, and thus the number of products of this reactions are less. We have also computed the difference in the number of counts of the polyethylene and the non-polyethylene cases for three distances, as shown in Tab. 7. More specifically, we have computed the total and each of the four main contributions, obtaining a substantial difference in the proton contribution. Distances (mm) LaCl3Proton 35S Alpha 32P 165 409926 30081 22294 17721 17219 170 384124 28292 20927 16613 16133 175 361224 26830 19796 15634 15170 Tab. 7. Difference in event detection between the polyethylene and non-polyethylene scenarios. The variations in the number of counts and the main contributions have been documented. These observations were made at three distinct distances separating the detector from the beam. 36 5.2 Proton beam results The setup consists of a millimetre-precise platform that holds the CsI target, and allows an automated and controlled movement of its position with respect to the proton beam exit and the detector position. In Fig. 25, we can see a picture of the experimental setup in CMAM. Simultaneous measurements were conducted with other detectors to optimise beam time utilisation. Fig. 25. A photograph capturing the experimental setup was taken during the simultaneous measurements conducted with other detectors. We have designed a sketch, depicted in Fig. 26, to clarify the movement of the operational motor that controls the motion of the target, allowing us to observe the initial "home" configuration and the maximum allowed distance. We can see that the range of movement is 50 mm, but we will restrict the distance between the detector and the target to the 160 mm to 180 mm range. We have created two separate sets of positions: even positions and odd ones, in stages of 2 mm in order to measurement for the entire position range in steps of 1 mm. It is important to measurement the intensity of the beam inside and outside the vacuum chamber. For that purpose, we have used two conductive cups called Faraday cups, to collect charged particles. The amount of ions or electrons that strike the cup can be estimated from the produced current by measuring it. Unfortunately this could only be done in between measurements (runs of beam on target), but not during the measurements themselves. Furthermore, each time we had to move the Faraday cups in and out and this implies imprecision in the positioning and therefore the current measurements, as we 37 Fig. 26. CMAM setup diagram. will see later. 5.2.1 PSD discrimination in measurements of 5 minutes Fig. 27. PSD as a function of the ADC channel with the graphical cuts drawn for a beam-detector distance of 161 mm. To test the detector’s sensitivity to changes in the Bragg peak position, we conducted two sets of measurements. Firstly, we measured the set of even positions, ranging from 180 mm to 160 mm in increments of 2 mm. Subsequently, we repeated the measurements backwards for the set of odd positions, ranging from 179 mm to 161 mm. 38 Fig. 34. PSD as a function of ADC channel for a background measurement with polyethylene in a distance of 175 mm. Initially, a background measurement was conducted at a beam-detector distance of 175 mm, with a polyethylene block placed in front of the detector. The plot depicting the PSD versus ADC channel at the 175 mm position is presented in Fig. 34. It is evident that the neutron region exhibits very low counts. Comparing it with Fig. 35a, we observe that the low-energy gamma region also exhibits minimal counts in contrast to Fig. 35a. Nevertheless, the high-energy gamma region contains the majority of the counts. Consequently, we can attribute the bulk of high-energy gamma rays to background noise, including the contribution from the polyethylene block. Fig. 35a illustrates the graphical divisions made to define the areas corresponding to highenergy gammas (pink), low-energy gammas (red), and neutrons (green). The neutrons region encompasses the α-particles and proton products generated by the (n,p) and (n,α) reactions. Drawing a comparison between the results obtained with and without a block of polyethylene at the same position proves to be useful. Taking the PSD vs ADC channel plot at a distance of 170 mm as an illustrative example, we examine the case with polyethylene depicted in Fig. 35a, and the case without polyethylene shown in Fig. 35b. On visual inspection, no significant difference is discernible. However, to precisely determine the variance in counts, particularly for neutrons that undergo moderation with polyethylene, it is beneficial to compute the integral for each distinct graphical region. The discrepancy in counts has been compiled in Tab. 10. 45 (a) At 170 mm, with polyethylene. (b) At 170 mm, without polyethylene. Fig. 35. PSD as a function of the ADC channel for a 5-minute measurement conducted at a distance of 170 mm, both with and without a 5 cm thick polyethylene block. 46 Normalised number of counts Type Position (mm) High-energy gammas Low-energy gammas Neutrons Background with polyethylene 175 170 ±13 38818 ±197 30 ±5 With Polyethylene 165 9156 ±39 294481 ±226 6283 ±33 170 9457 ±40 305361 ±230 6497 ±33 175 9345 ±39 290326 ±219 6244 ±32 Without Polyethylene 165 24755 ±64 286070 ±225 6917 ±34 170 25183 ±66 273349 ±223 7370 ±35 175 24233 ±64 263740 ±220 7049 ±35 Tab. 10. Numerical values representing the normalised number of counts obtained for measurements conducted with and without a polyethylene moderator at various beamdetector distances of 165 mm, 170 mm and 175 mm. Fig. 36. Counts normalised with intensity due to intensity fluctuations as a function of the distance target-detector for the neutron region. We have represented the polyethylene and non polyethylene cases. In Fig. 36, it is evident that neither the polyethylene nor the non polyethylene case exhibit a clear tendency. As stated before, this is due to the intensity fluctuations. Upon comparing the two cases, it becomes clear that the presence of the polyethylene moderator reduces considerably the number of counts since it slows down the neutrons and our crystal is especially efficient to detect fast neutrons. This is in good agreement to the Monte Carlo simulations of the previous section. 47 6 Conclusions In conclusion, our study involved optimising the signal integration lengths and other acquisition configuration parameters to prepare the LaCl3detector to conduct CMAM measurements. We compared the figures of merit (FOM) obtained for the different PMTs, namely the R1924A-700 and R6233-100, using the same data acquisition parameters. Based on our analysis, we determined that the R1924A-700 was the most suitable PMT for conducting the measurements. However, due to concurrent experiments, we only had the R6233-100 available, which we used for our measurements. Furthermore, we performed a simulation using a neutron beam and a LaCl3crystal detector, without simulating the reaction of a proton beam impacting a CsI target or the processing of the PMT. The simulation results aligned with our expectations, demonstrating that the LaCl3detector is capable of discriminating a 1 mm difference in distance. The agreement was observed across various energy ranges, including low-energy and highenergy gammas as well as neutrons. Notably, the number of counts detected increased with proximity to the neutron source, highlighting the influence of the solid angle effect. Based on the experimental results of the CMAM measurements, conclusive results could not be obtained due to the unsuitability of the setup. Initially, we expected our results to align with the simulations; however, certain factors directly affected the outcomes. Specifically, the instability in beam intensity during the measurements and the need to turn the beam on and off to reposition the target significantly impacted the results. Beyond the instability itself, the fact that we did not have a direct measurement of this beam current during the measurements, only before and after each run, and they were always different. In a second experiment, beyond the scope of this master thesis, we could demonstrate that the system works very nicely when the beam is stable and we integrate the current on target. Our analysis indicates that the behaviour of low-energy gammas and neutrons mirrors the fluctuations in beam intensity, while high-energy gammas demonstrate an opposite fluctuation pattern. By segregating the measurements into two distinct sets, odd distance values plus the 170 mm distance, and even distance values without 170 mm, we still observe a fluctuating pattern, even though with reduced intensity fluctuations. Nevertheless, it is important to note that despite our efforts, we still do not observe a consistent overall trend and the detector still does not display sensitivity to a 1 mm variation in distance. In addition to the aforementioned effects of beam current instability on the measurements, we also observed that the presence of a polyethylene block further influenced the results. A comparative analysis conducted at identical distances, comparing measurements with and without the polyethylene block, highlighted the effectiveness of the moderator in 48 slowing down fast neutrons generated by the proton beam’s interaction with the target. This implies that incorporating a moderator surrounding the detector in all directions except forward, serves as shielding for the neutrons scattered in in the walls of the hall that would normally contribute to the background. In conclusion, our study indicates that the LaCl3detector, when used in an intensityvariable facility, does not yield reliable results for neutron-gamma discrimination with 1 mm increments. The analysis revealed limitations and challenges associated with the intensity fluctuations and other factors that directly affected the detector’s performance. 7 Next-steps We are currently conducting a second test in CMAM using the PMT R1924A-700 and a241Am/9Be source to investigate PSD discrimination in the LaCl3scintillator. In this second phase of the project, we could differentiate between protons, alphas, and gammas at a high voltage (HV) of -1100 V, as explained in the previously study. The 241Am emits alphas that are captured by 9Be, resulting in the emission of neutrons through the 9Be(α, n)12C reaction. We were able to continuously monitor the beam intensity using a current integrator connected to the target, which addressed the intensity fluctuations encountered in the previous CMAM experiment. The results of this second measurement are promising. Although it was beyond the scope of this thesis, I had the opportunity to participate in this experiment. To replicate the conditions of proton range verification in proton therapy, a third phase of the project would involve studying the variation of the position of the Bragg peak in a proton therapy facility. It is crucial to achieve better control over intensity fluctuations and overcome issues related to stopping and restarting the neutron emission in order to test the scintillator under real-life conditions. This would help determine its potential application in proton therapy. 49 8 References [1] R. P. Johnson, “Review of medical radiography and tomography with proton beams,” Reports on progress in physics 81 no. 1, (2017) 016701. [2] V. Giacometti, “Modelling and improvement of proton computed tomography,”. [3] F. Hueso-González and T. Bortfeld, “Compact method for proton range verification based on coaxial prompt gamma-ray monitoring: A theoretical study,” IEEE transactions on radiation and plasma medical sciences 4no. 2, (2019) 170–183. [4] D. Chattopadhyay, S. Sharma, and M. S. Sarkar, “Elimination of the effect of internal activity in lacl3: Ce scintillator,” Journal of Instrumentation 16 no. 06, (2021) P06025. [5] P. Q. Vuong, H. Kim, N. T. Luan, and S. Kim, “Neutron spectroscopy using pure LaCl3crystal and the dependence of pulse shape discrimination on ce-doped concentrations,” Nuclear Engineering and Technology 53 no. 11, (2021) 3784–3789. [6] F. Crespi, F. Camera, et al., “Alpha–gamma discrimination by pulse shape in labr3: Ce and lacl3: Ce,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 602 no. 2, (2009) 520–524. [7] C. Hoel, L. Sobotka, K. Shah, and J. Glodo, “Pulse-shape discrimination of la halide scintillators,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 540 no. 1, (2005) 205–208. [8] A. Pankratenko, T. Kormilitsyn, Y. A. Kashuck, S. Y. Obudovsky, D. Fridrikhsen, A. Shevelev, E. Khilkevitch, M. Iliasova, N. Bakharev, and O. Skrekel, “Analysis of the lacl3 (ce) scintillator response function to fast neutrons,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 1052 (2023) 168282. [9] J. A. Briz, A. Nerio, et al., “Proton radiographs using position-sensitive silicon detectors and high-resolution scintillators,” IEEE Transactions on Nuclear Science 69 no. 4, (2022) 696–702. [10] S. Agostinelli, J. Allison, et al., “Geant4—a simulation toolkit,” Nuclear instruments and methods in physics research section A: Accelerators, Spectrometers, Detectors and Associated Equipment 506 no. 3, (2003) 250–303. 50 [11] https://www.nndc.bnl.gov/nudat3/. [12] L. Farabee, “The removal of alpha-contamination from lanthanum by ion exchange,” US AEC, TID-7591 (1960) 78–80. [13] B. D. Milbrath, R. C. Runkle, T. W. Hossbach, W. R. Kaye, E. A. Lepel, B. S. McDonald, and L. E. Smith, “Characterization of alpha contamination in lanthanum trichloride scintillators using coincidence measurements,” Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 547 no. 2-3, (2005) 504–510. [14] https://www.caen.it/products/dpp-psd/. 51