Silicon carbide detectors for dosimetry and monitoring of ultra-high dose rate beams

FLASH radiotherapy, which employs ultra-high dose rate (UHDR) beams with a mean dose rate > 40 Gy/s and a total irradiation time < 200 ms to treat tumors, exhibits remarkable ability to spare healthy tissue while maintaining the same efficiency in treating tumors. However, UHDR presents challenges in dosimetry and beam monitoring, as the dosimeters recommended for conventional radiotherapy, i.e. the ionization chambers, show saturation at such high dose rates and dose delivered per pulse and hence cannot be employed for accurate dosimetry in the future clinical transition of FLASH radiotherapy. This implies the need to develop alternative techniques and dosimeters able to sustain the peculiar conditions of the UHDR beams. This study investigates the feasibility of using a new generation of Silicon Carbide (SiC) detectors for the measurement of the instantaneous dose rate of UHDR electron beams. An experimental investigation was conducted with the ElectronFLASH linac developed by the SIT Sordina company and able to accelerate 7 and 9 MeV electron pulsed beams at FLASH regimes. The signals produced in the SiC detectors were acquired and compared with the signals detected by the monitoring system currently mounted along the LINAC, i.e. two AC current transformers supplied by the Bergoz company. The main purpose of the experiment was to demonstrate the capability of the developed SiC detector to measure the single pulse duration and waveform with high time resolution and accuracy. The test was performed by using both 7 and 9 MeV electron beams and has shown promising results.


Introduction
For decades, conventional radiotherapy (CONV-RT), which utilizes high energy photons, has been the standard approach for cancer treatment using external radiation.However, this approach has not fully met the main objective of radiotherapy, as it can cause significant damage to healthy tissues around the tumour, leading to adverse side effects and long-term complications.For this reason, several approaches have been employed and are still being explored on the possible ways to further optimize radiotherapy.
In recent years, FLASH radiotherapy (FLASH-RT) has emerged as a promising technique for radiation therapy.FLASH-RT involves the delivery of ultra-high dose rate (UHDR) radiation in a single or a few fractions within a very short irradiation time, typically, less than 100-200 ms [1].The minimum average dose rate needed to trigger the FLASH effect is around 40 Gy/s (with instantaneous dose rate reaching up to 10 7 Gy/s in a μs pulse), which is significantly higher compared to the average dose rates of 0.5-10 Gy/min in CONV-RT.This drastic increase in dose rate enables the delivery of therapeutic doses (tens of Gy) in a very short irradiation time, several order of magnitude less than the time required for CONV-RT.Several experimental studies have demonstrated that irradiation using UHDR beams achieves the same lethal effect on the tumour while inducing a higher degree of sparing to the healthy tissues (compared to CONV-RT) [1][2][3].This phenomenon is known as the FLASH effect.What this implies is that FLASH-RT increases the therapeutic window allowing the possibility of escalating the radiation dose delivered to the tumor and increasing the tumor control probability while minimizing the potential risk of complications in normal tissue.The study of the mechanisms and factors for the FLASH effect is ongoing, including factors like oxygen depletion and oxygen concentration [4][5][6].FLASH-RT is still under investigation to explore its potential benefits and address the associated challenges.
Despite the benefits of FLASH-RT, there are considerable challenges associated with UHDR dosimetry and beam monitoring.The standard dosimeters recommended by the international dosimetric protocol for the clinical treatments [7], such as the ionization chambers, often suffer from non-negligible saturation effects, primarily due to ion recombination at very high dose delivered per pulse [8].This makes them unsuitable for accurate dosimetry in these extreme conditions.To address this issue, researchers are actively exploring alternative approaches, such as the possibility to modify the geometry or the composition and pressure of the filling gas of the existing ionization chambers to improve the ion collection efficiency [9][10][11], use dose-rate independent passive detectors like alanine and radiochromic film (RCF) [12] (which are not feasible from the clinical perspective due to their time-consumption -1 -and not able to provide an "on-line" information), and identifying new technologies that can overcome the limitations encountered with traditional active detectors.For instance, solid-state detectors, such as diamond [13] and Silicon Carbide (SiC [14]), have recently emerged as a promising solution.According to Marinelli et al. (2022) [13], a novel diamond Schottky diode detector prototype (Flash diamond) developed for FLASH-RT showed a linear charge response up to a dose per pulse of 26 Gy/pulse when irradiated with the FLASH electron beam.Also, in a study by Romano et al. (2023) [14], the performance of novel SiC detectors were investigated using UHDR electron beams.A similar SiC detector is used for the present work and will be described in detail in the next section.The SiC device demonstrated a linear charge response up to 2 Gy/pulse, in contrast to a Silicon diode that saturated before reaching 0.5 Gy/pulse under similar conditions [14].These findings suggest that these solid-state detectors hold potential as alternatives to ionization chambers or Silicon-based dosimeters for real time dosimetry.Furthermore, unlike continuous beams which can be described with average dose rate, pulsed beams require consideration of the instantaneous dose rate.As shown in figure 1 for a pulsed electron beam with pulse width (), pulse repetition frequency () and total irradiation time (), the average instantaneous dose rate is defined as the dose delivered per pulse divided by the pulse width (  /).On the other hand, the average dose rate is the product of the , and the dose delivered per pulse ( •   ).In FLASH-RT, a significantly high total dose is delivered within a noticeably short time, which for pulsed electron beams implies delivering extremely high instantaneous dose rates of order of MGy/s within a single pulse [15].These beam parameters and their combination can contribute to inducing the observed radiobiological FLASH effect and they need to be measured with a good accuracy to access the conditions that optimize the desired effect.Also, accurate monitoring of possible fluctuations in beam current at the level of the single electron pulse and, consequently, the intra-pulse instantaneous dose rate can be essential from the perspective of a full control of irradiation in the FLASH regime.For this purpose, novel Silicon Carbide (SiC) detectors, already tested under UHDR electron beams and proved to be dose-rate independent, have been used to detect the temporal structure of the single electron pulses accelerated with the ElectronFLASH linac at SIT Sordina srl company [16].The results and the comparison with the well-established monitoring system based on the AC-Current Transformer, will be presented in this contribution.

The SiC detector
SiC detectors are emerging as one of the promising alternatives for dosimetry and monitoring of UHDR beams.They exhibit high radiation hardness, fast response, high sensitivity, and dose rate independence [14,17].Table 1 provides a comparison of some properties between Silicon (Si), -2 -diamond, and 4H-SiC (a hexagonal crystal structure of SiC).As evident in the table, 4H-SiC represents a compelling compromise between outstanding characteristics and cost effectiveness.This detector consists of 50% Si and 50% diamond, combining the sensitivity of Si and the robustness of diamond.Ultra-thin SiC detectors of different active areas and thicknesses have been developed by the ST-Lab, a start-up company in Catania, in collaboration with the Catania division of INFN in the framework of the FRIDA (Flash Radiotherapy with hIgh Dose rate particle beAms) project financed by the CSN5 of the INFN.As shown in figure 2a, this detector structure is based on PiN junctions and the sensitive part comprises of a thin p + highly doped layer (0.3 μm thick) with acceptor concentration (N  ) of 1 × 10 19 cm −3 atop an n − lightly doped layer ranging from 200 nm to 100 μm thickness, with donor concentration (N  ) of 8 × 10 13 cm −3 .This sensitive part is situated on an n + thick substrate (370 μm thick) with N  of 5 × 10 18 cm −3 .The detector offers various active areas, ranging from 1 × 1 mm 2 to 10 × 10 mm 2 , with active thicknesses varying from 0.2 μm to 100 μm. Figure 2b shows the SiC sensor of 10 μm active thickness and 1×1 cm 2 active area.In a previous work [14], the first dosimetric characterization of the novel SiC detector was performed with UHDR electron beams.-3 -

The ElectronFLASH LINAC
The ElectronFLASH (EF) is a specialized linear accelerator (LINAC) developed by SIT Sordina, located in Aprilia, Italy.It is designed for research involving FLASH Ultra-High-Dose-rate/Ultra-High-Dose-per-pulse [18,19].The EF used for this work is equipped with a diode gun, allowing the selection of two beam current modes: one of very low current (Conventional mode) and the other of high current (FLASH mode) corresponding to a beam current exceeding 100 mA.The EF can accelerate electron beams at nominal energies of 7 MeV and 9 MeV, offers a variable pulse width between 0.5-4.5 μs, and operates at a repetition rate (PRF) ranging from 1 Hz to 245 Hz.The total irradiation time is determined by the number of pulses to be delivered and the PRF selected.The EF features two AC current transformers (ACCT), located immediately after the in-vacuum section of the accelerator, serving as its monitoring system.The ACCT is well characterized for beam monitoring and can accurately measure the current of the beam traversing even at UHDR [20].

Experimental setup
The experimental setup employed is illustrated in figure 3.As shown, one ACCT was mounted along the LINAC, just after the in-vacuum section of the accelerator, while the SiC detector was located at the end of the final PMMA applicator which is directly connected to the radiant head of the EF.The applicator serves the purpose of shaping the beam and obtaining a uniform transversal dose distribution on a circular shape at the irradiation point.For this experiment, a 40 mm diameter, 40 cm length applicator was used.The SiC detector was then irradiated at nominal energies of 7 and 9 MeV, with variations in the pulse width.In particular, a SiC detector 10 μm thick and about 3 mm 2 (1 mm radius) active area was positioned after the applicator (at a few cm of air from the end of the applicator).A Keithley 6517A electrometer was used to apply a bias voltage of 80 V to the SiC detector, while an 8 GS/s fast oscilloscope (RIGOL, 100 MHz) with a 50 Ω termination was employed to record the single pulse waveforms captured by both the SiC detector and the ACCT.The signal acquired from the oscilloscope provides a visual representation of the voltage's temporal variation during a single pulse for the two detectors and at both energies.

Results
To have a direct comparison between the signals arising from the ACCT and SiC detectors, the amplitudes of the signals were normalized.This normalization allows for the observation of any differences in pulse shape between the signals of the two detectors.In figure 4a and figure 4b, we can -4 -

JINST 19 C03064
observe the normalized pulse shapes obtained with the ACCT and SiC detectors for the same electron pulse.In these specific cases, both were recorded with a 4 μs pulse width at an energy of 9 MeV in figure 4a and 7 MeV in figure 4b, thus facilitating the comparison of the signals.The ACCT signal displays a nearly sharp step response, maintaining a relatively constant beam current throughout the entire pulse width.However, the SiC detector's signal shows a slight deviation, with a particularly noticeable reduction in beam current at the beginning of the pulse, occurring within the initial 1.5 μs.In other words, an instantaneous dose rate lower than the nominal dose rate is observable in the SiC signal within the first 1.5 μs.The area under the signal produced was evaluated for the ACCT and the SiC detectors, resulting in a percentage difference of 7.36% for 9 MeV and 7.26% for 7 MeV between the two signals.The relative closeness in this difference implies consistency in the dissimilarity between the signals, regardless of the energy of the electron beam, suggesting that the discrepancy could be related to the particular setup employed, i.e. as for instance the different position of the ACCT and the SiC detectors along the beam line, other than some physical discrepancy of the detectors.Nevertheless, further investigation is required to fully understand the causes of these differences.The actual pulse width of the signals recorded by the two detectors were obtained by calculating the time interval between the point at 50% rise and the one at 50% fall.The values obtained are consistent with the nominal values.For example, setting a pulse duration of 4 μs in the control panel of the accelerator, the SiC detector recorded a pulse width of 3.78 μs, while the ACCT recorded 3.96 μs.This difference in pulse width (around 4% in this case) is due to the discrepancies between the two signals.Furthermore, the fall time (20-80%) of the signals was analysed, revealing a fall time of around a hundred nanoseconds for both detectors.Notably, the SiC showed a slightly shorter fall time compared to the ACCT.For the 4 μs pulse at 9 MeV energy, the fall time was 129 ns for SiC and 164 ns for the ACCT.Similarly, for the 4 μs pulse at 7 MeV, the fall time was 110 ns for SiC and 151 ns for the ACCT.This suggests that the SiC detector exhibits a faster decay response compared to the ACCT.

Conclusion
One of the crucial beam parameters to monitor for a deep study of the FLASH effect especially with electron UHDR beams is certainly the instantaneous dose rate and its possible variation within the single electron pulse.The use of a dosimeter able to also detect in real time the small and fast variations of the beam current with high temporal resolution, is certainly something required both for a realistic investigation of the FLASH effect and for its future applicability in the clinical routine.
In order to access the SiC detector's capability to measure in real-time the intra-pulse variation of the beam current, the single pulse current waveforms of UHDR low energy electron beams accelerated with the ElectronFLASH LINAC were acquired using the SiC detector and compared to the wellcharacterized ACCT detector.The comparison revealed discrepancies in the time structure of the single electron pulse for the first 1.5 μs of the pulse duration which could be ascribed to the different position of the detectors along the beam line.On the other side, the consistent percentage discrepancies (around 7%) of the integrals of the signal acquired with the SiC and the ACCT for the two electron energies, i.e. 7 and 9 MeV, suggest that such difference is independent on the particle energy.It could be inferred that the discrepancy may be easily corrected for the same pulse width by using a constant correction factor to account for the different position of the ACCT with respect to the SiC, i.e. irradiation point.The inability to validate these discrepancies by moving the detectors stemmed from the integration of the ACCT inside the accelerator as a monitoring system, restricting its free movement.Therefore, to gain more insights, plans are being made to conduct a comparison of the two detectors positioned at the same location along the beamline.This strategy will allow us to know how much the observed discrepancies can be attributed to the detectors placement and how much it is linked to the detector's performance.
Despite these observed discrepancies, the SiC detector was able to measure pulse widths that closely aligned with the nominal values.This is a significant result and shows the SiC detectors potential in the context of UHDR dosimetry.
Therefore, our preliminary study has shed the light on using the SiC detector for UHDR dosimetry and has provided room for further study.As we continue our research, we look forward to understanding the factors influencing the detectors responses, with the main aim of harnessing the full potential of the SiC detector for UHDR dosimetry.

Figure 1 .
Figure 1.Schematic of the time structure of a pulsed electron beam.

Figure 2 .
Figure 2. The SiC detector (a) showing the schematic.It depicts a SiC sensor of 10 μm active thickness and 370 μm substrate.(b) Picture of the SiC detector.

Figure 3 .
Figure 3.The schematic of the experimental setup.

Figure 4 .
Figure 4. (a) Comparison of the signals from the SiC detector and the normalized signal from the ACCT at 9 MeV and 4 μs pulse width.(b) The normalized signals at 7 MeV and 4 μs pulse width.