State-of-the-art silicon carbide diode dosimeters for ultra-high dose-per-pulse radiation at FLASH radiotherapy

Objective. The successful implementation of FLASH radiotherapy in clinical settings, with typical dose rates >40 Gy s−1, requires accurate real-time dosimetry. Approach. Silicon carbide (SiC) p–n diode dosimeters designed for the stringent requirements of FLASH radiotherapy have been fabricated and characterized in an ultra-high pulse dose rate electron beam. The circular SiC PiN diodes were fabricated at IMB-CNM (CSIC) in 3 μm epitaxial 4H-SiC. Their characterization was performed in PTB’s ultra-high pulse dose rate reference electron beam. The SiC diode was operated without external bias voltage. The linearity of the diode response was investigated up to doses per pulse (DPP) of 11 Gy and pulse durations ranging from 3 to 0.5 μs. Percentage depth dose measurements were performed in ultra-high dose per pulse conditions. The effect of the total accumulated dose of 20 MeV electrons in the SiC diode sensitivity was evaluated. The temperature dependence of the response of the SiC diode was measured in the range 19 °C–38 °C. The temporal response of the diode was compared to the time-resolved beam current during each electron beam pulse. A diamond prototype detector (flashDiamond) and Alanine measurements were used for reference dosimetry. Main results. The SiC diode response was independent both of DPP and of pulse dose rate up to at least 11 Gy per pulse and 4 MGy s−1, respectively, with tolerable deviation for relative dosimetry (<3%). When measuring the percentage depth dose under ultra-high dose rate conditions, the SiC diode performed comparably well to the reference flashDiamond. The sensitivity reduction after 100 kGy accumulated dose was <2%. The SiC diode was able to follow the temporal structure of the 20 MeV electron beam even for irregular pulse estructures. The measured temperature coefficient was (–0.079 ± 0.005)%/°C. Significance. The results of this study demonstrate for the first time the suitability of silicon carbide diodes for relative dosimetry in ultra-high dose rate pulsed electron beams up to a DPP of 11 Gy per pulse.


Introduction
The recent development of FLASH radiotherapy has led to the challenge of developing adequate sensors for active dosimetry in ultra-high dose rate (UHDR) beam delivery.Especially in the case of FLASH electron beams the dose delivery can reach up to several Gy even in a single pulse with a few microsecond duration.The accurate dosimetry of this new UHDR radiotherapy modalities represents a key issue for its clinical translation (Schüller et al 2020, Romano et al 2022).
Solid-state dosimeters are widely used for active real-time dosimetry in radiotherapy.Silicon diodes in particular have been used for years as standards for relative dosimetry in conventional photon and charged particle radiotherapy.Some of the advantages of silicon diodes are their high sensitivity compared to ionization chambers (about 18 000 times higher for the same active volume), use without external voltage bias, mechanical robustness, and well-developed manufacturing technology that allows producing complex structures adapted for each application.However, silicon diodes are sensitive to light and temperature variations and exhibit pulse dose rate and average dose rate dependence at higher pulse dose rates and higher dose per pulse (DPP).In addition, their radiation hardness is only moderate showing a gradual sensitivity decay with accumulated dose (Jursinic 2013, Rosenfeld et al 2020).These drawbacks make them unsuitable for the strict requirements of dosimetry in high and ultra-high dose rate pulsed particle beams.
Diamond is a unique semiconductor material for radiation detectors due to its extreme radiation hardness, fast charge collection, and excellent noise performance.Its use in a variety of applications, including particle colliders and radiation dosimetry and microdosimetry, has become feasible following significant improvements in the quality of the material (Kagan and Trischuk 2018, Bassi et al 2021, Loto et al 2021).Marinelli et al have recently presented a chemical vapour deposition (CVD) single crystal diamond dosimeter optimized for ultrahigh dose per pulse conditions that has shown signal linearity up to a dose per pulse of at least 20 Gy and has been successfully used for commissioning of FLASH electron beams (Marinelli et al 2022, Verona Rinati et al 2022).However, diamond is still expensive to produce, has repeatability issues in production, and is not available in large wafers, which limits its potential for large-scale applications.
Silicon carbide (SiC) is another semiconductor with attractive physical properties for radiation detection.The polytype of SiC more interesting for radiation detector applications is 4H-SiC as it has the widest bandgap and a high, almost isotropic, electronic mobility (Tudisco et al 2018).SiC, like diamond, is a wide bandgap semiconductor that results in a significantly reduced rate of thermally generated charge carriers, reducing the detector leakage current and noise compared to silicon.This also makes it essentially insensitive to visible light and temperature variations.SiC has a higher displacement energy threshold than silicon and thus an expected higher radiation hardness (Sellin and Vaitkus 2006, Rafí et al 2018, 2020).The lower ionization energy for silicon yields a higher charge for the same deposited energy and thus a higher energy resolution for spectroscopic measurements.On the other hand, the lower SiC sensitivity per active volume and deposited energy makes it a better choice for applications where a large signal might saturate the semiconductor, like UHDR beam dosimetry.In addition, unlike silicon, SiC is both bio and hemocompatible and therefore is a very promising material for the realisation of advanced biomedical devices (Saddow 2022).The possibility of integrating very thin graphene layers into the SiC surface to act as electrical contacts instead of metals has recently been demonstrated (Ugobono et al 2022), which opens the door to fabricating devices that are functional at high temperatures, have very thin entrance windows, and are overall metal-free.
First studies of silicon carbide as an alternative to silicon for radiation dosimeters, neutron and particle detectors were already published 20 years ago (Dulloo et al 2003, Pini et al 2003, Moscatelli et al 2005, Sellin and Vaitkus 2006) but the lack of cost-effective, high-quality substrates meeting the specifications for radiation detectors limited their use to research purposes.However, in recent years, SiC wafer manufacturing innovations strongly driven by the automotive industry have pushed substrate quality close to that of silicon with precise control of doping, homogeneity, and defect concentration.Currently, SiC has a mature microelectronic technology that allows producing complex and reproducible structures and is available in large detector-grade wafers (200 mm diameter is an industry standard).However, at the moment SiC radiation detectors do not exist in the commercial market and the implementation of this technology has yet to be fully realized.
Table 1 shows the material parameters for Si, CVD diamond, and 4H-SiC more relevant for radiation detection (Nava et al 2008).The radiation sensitivity per unit volume has been derived from the ionization energy and density of each semiconductor.The effective atomic number for SiC has been calculated through the total electron stopping power for the nominal energy of 20 MeV used in this work (Taylor 2011).The diodes were processed on 100 mm diameter 4H-SiC wafers by using standard SiC fabrication steps developed for power devices (Godignon et al 2011) and adapted to the requirements of radiation detectors.The SiC substrate is a 3 μm thick sensitive layer made of lightly n-type epitaxial 4H-SiC doped with nitrogen with a nominal concentration of 2 × 10 14 cm −3 , on top of 350 μm thick low resistivity substrate made of 4H-SiC highly doped with nitrogen at a concentration of 10 19 cm −3 .The PiN diode anode was implemented through aluminium ion implantation followed by a high-temperature ramped annealing, forming a graded junction with the high doping at the interface to form a good enough ohmic contact.An 800 nm SiO 2 dielectric layer was created and selectively etched to form the top surface passivation layer.Diode metallization in the top surface (anode) was achieved with a thin metal stack (Ti/Al/Ti/Ni) to form the low contact resistance contact with the P-N junction followed by an Au layer in the periphery to facilitate probing and wire bonding.A second metal stack (Ti/Ni/Au) was formed on the back surface for the cathode contact.
The final fabricated wafers contain circular diodes with different diameters to allow for different sensitivities and spatial resolutions.For the dosimetry application, 1-2 mm diameter diodes were implemented.These dimensions were chosen to achieve sensitivities on the order of nC/Gy, considering the nominal active thickness of the diodes (3 μm) and the radiation sensitivity per unit volume of 4H-SiC (425 pC/(mGy•mm 3 )).In addition, the wafer includes smaller diodes for microdosimetry (10-30 μm diameter) set in different array configurations (single diodes, strip and pixel arrays).The wafer layout can be seen in the optical photograph of a wafer in figure 1.
The SiC devices used for this study are circular PiN diodes with 1 mm diameter.A top-view optical microscope picture of a fabricated diode and a schematic cross-section are shown in figure 1. the central diode is surrounded by a multi-guard ring structure to extract the surface current contributions from the diode current and to define the sensitive area.
After the electrical characterization, the SiC wafer was diced.The 1 mm diodes with the best electrical characteristics (breakdown voltage >100 V) were selected and were encapsulated by PTW Freiburg with their commercial microSilicon housing (Schönfeld et al 2019, Weber et al 2020, Akino et al 2021) to provide electrical connectivity and waterproofness to test it in a water phantom in the same way as established dosimeters.

Electrical characterisation
The electrical characterization of the diodes was done on wafer, on an electrically shielded MPI TS2000-SE probe station at 20 ± 1 °C.An HP4155b semiconductor parameter analyser was used to obtain the current-voltage (I-V ) characteristics.To evaluate the quality and yield of the fabrication, all single diodes with 1 mm diameter (60 per wafer) and 2 mm diameter (18 per wafer) were tested.These devices were chosen because they are distributed across the entire wafer and, as such, provide a suitable means to assess the yield.

Characterisation in electron beam
Diode characterization with UHDR electrons was performed in PTB's ultra-high dose per pulse reference electron beam (Bourgouin et al 2022) in a water phantom at the reference depth.To achieve a wide range of DPP, different source-to-surface distances (SSD), i.e. the distance between the beam exit window and the entrance window of the water phantom, were used: 90 cm and 50 cm.In addition, different aluminium scattering plates (1, 2, and 6 mm thick) were positioned at the linac exit window.The variation of the dose per pulse in each setup was achieved by modifying an upstream slit opening in the beam and the pulse duration (0.6-3 μs).Fluence per pulse was accounted by measuring the charge of each beam pulse using an integrating current transformer (ICT), from Bergoz Instrumentation, placed in the beam line.The ICT readout was calibrated against the absorbed dose at the reference depth in the water phantom through alanine measurements for the setup with SSD 90 cm and 6 mm aluminium scattering plate as well for the setup with SSD 70 cm without scattering plate, together with a measurement for the calibration of a flashDiamond prototype (serial number SN 7610) from Tor Vergata University and PTW (Verona Rinati et al 2022) at the same position as the alanine pellets directly before and after the alanine irradiation.For the other used combinations of SSD, pulse duration, and aluminium scattering plates thickness, the ICT readout was calibrated against the calibrated flashDiamond.
A 20 MeV monoenergetic electron beam was used for the characterization of the SiC prototype.The DPP ranged from 0.1 Gy per pulse to 11 Gy per pulse.The pulse durations were 2.9 μs, 1.6 μs, 1.5 μs and 0.5 μs and the pulse repetition frequency of the beam was fixed to 5 Hz.Measurements were performed in a PMMA water tank (in which the entrance window was replaced by 5 mm width PEEK window) with a motorized positioning system that was used to position the detector and for scanning the depth-dose distributions.
The determination of the reference depth was based on the percentage depth dose curves (PDD) measured with a flashDiamond prototype (SN 7610) (Verona Rinati et al 2022) and calculated according to the TRS-398 code of practice (IAEA 2000).
The detector readout was done with a calibrated Keithley 616 Electrometer.A capacitor of 33 nF was used at the input of the electrometer to reduce the voltage on the signal cable sufficiently to keep the operational amplifier of the electrometer in the linear range and to ensure that the voltage on the signal cable did not affect the diode even at the largest charge pulse from the detector.The capacitor is charged within the beam pulse duration, i.e. within a few μs.Subsequently, the capacitor will be discharged by the electrometer within 20-200 ms (depending on the input resistance of the measuring range used).The electrometer was operated in current mode, i.e. the current from the capacitor was measured continuously.The analogue output of the electrometer was recorded using a 16-bit analogue-to-digital converter with 200 kHz and analysed using custom software.From the measured current the total charge from the SiC diode was determined for each pulse and recorded together with the corresponding charge of the beam pulse determined by means of the ICT.For each DPP setting, 50 pulses were measured pulse-resolved to reduce the uncertainty of the mean value.A calibrated current source (Keithley 6430) was used to calibrate the electrometer with its read-out system.
For the reference dosimetry, the flashDiamond prototype detector (SN 7610) in combination with absolute alanine measurements was used, as described above.
Irradiations to evaluate the linearity response of the SiC prototype (serial number SN152384) were performed positioning the detector in the water tank at the reference depth of measurement in each setup and varying the dose per pulse.Two different linearity measurements were done: one with the SiC prototype without any significant pre-irradiation, and a second one after the radiation hardness test, with the SiC diode irradiated to a cumulative dose exceeding 100 kGy.In the second experiment, the response of one microSilicon diode (SN152259) was also measured in similar conditions.The dose calibration of all diodes was performed with the reference flashDiamond in the low dose per pulse range (<0.2 Gy).
PDD measurements were performed with the SiC diode and the flashDiamond in ultrahigh dose per pulse conditions.
Next, the SiC diode was irradiated in single steps of 10 kGy and its response was determined in each step in the range DPP < 0.2 Gy to assess the sensitivity variation with accumulated dose.
The temporal characteristics of the diode response were investigated.The time-resolved beam current during each electron beam pulse was measured by means of the ICT connected to a transient recorder (Spectrum M3i.4142).The instantaneous current from the SiC diode was obtained via a time-resolved measurement of the voltage drop across the 50 Ω resistor of the input of a preamplifier (FEMTO HVA-200M-40-B) that was directly connected to the output cable of the SiC detector.The output of the preamplifier was connected to the second input channel of the same transient recorder as the used for beam current measurement.
Finally, the temperature dependence of the response of the SiC diode was evaluated by removing some water from the phantom, refill by pouring hot water into it and allowing several minutes for the SiC diode to take on the increased temperature of the water.Then the signal of the diode was measured as a function of the DPP in the range 0.2-0.9Gy per pulse.
The flashDiamond and the microSilicon diode were operated without external bias voltage in accordance with their specifications.The prototype SiC diode was also operated without bias voltage.It had not been irradiated previous to this characterization.

Results
Figure 2 shows the current-voltage curves of six representative SiC diodes with 1 mm diameter measured at low bias near the turn-on voltage of 2.5 V and at high reverse bias.Breakdown voltages are higher than the maximum sourced voltage of 100 V and reverse currents are of the order of tenths of pA, close to the resolution limit of the measurement setup.The series resistances extracted from the I-V curves at high forward current (10 mA) are in the 20-25 Ω range where this value includes the diode series resistance in series with parasitic resistance contributions from the test setup.
In terms of fabrication yield it was found that 97% of the 1 mm diodes had breakdown voltages greater than 20 V, highly exceeding the requirements of non-bias operation, with 85% of them greater than 100 V.In the case of the 2 mm diodes, 94% had breakdown voltages greater than 100 V.
Regarding the characterization in the electron beam, figure 3 shows the SiC diode response from 1 to 11 Gy per pulse, well into the ultra-high dose per pulse regime, for different pulse durations of 0.5, 1.6 and 2.9 μs.Maximum pulse dose rates within the pulse are about 4 MGy s −1 .The diode signal shows independence both of DPP and of pulse dose rate with a relative deviation of <2% including the dose measurement uncertainty of the reference detector.The slope of the linear fit to the experimental data is 0.90 nC Gy −1 .
The SiC diode was used to measure percentage depth dose profiles (PDD) with UHDR electrons with different doses per pulse and pulse durations.These measurements are presented in figure 4 together with the measurement with the reference flashDiamond.All measured PDD curves are superimposed with local relative minimal differences between them, after maximum normalization, that can reach 3%.The effect of such differences in the PDD are below 1.3 mm in R50 and lower than 0.8 mm for the reference depth evaluation.
Next, a characterization of the effect of the radiation dose on the SiC diode sensitivity was performed.For this, the evolution of the signal response of the SiC detector as a function of the total dose of electrons was measured.As is shown in figure 5, the SiC diode sensitivity decreases with total irradiation dose.After an irradiation with 100 kGy of electrons, the signal measured is 98.4% of the initial, corresponding to a loss of response extracted from the linear fit to the data of 0.018%/kGy.
After the irradiation experiment, a further measurement of the response of the SiC diode as a function of the dose per pulse was performed and compared to one microSilicon diode under similar irradiation conditions.The responses of the two detectors are reported in figure 6.For 1.5 μs pulse duration, the microSilicon diode exhibits a saturated response from a DPP of about 700 mGy.This is consistent with the reported for other commercial silicon diodes irradiated in similar conditions, that start to show a saturated response at hundreds of mGy per pulse (tens of Gy/s) for μs pulse duration (Di Martino et al 2020, Konradsson et al 2020).In contrast, the SiC diode hardly shows deviation from linearity.However, after irradiation of the diode during the previous tests the sensitivity was only 0.75 nC Gy −1 , whereas it was 0.90 nC Gy −1 at the very first irradiation test after its production (figure 3).
The time-resolved response of the SiC detector is shown in figure 7 and compared to the signal recorded simultaneously by the ICT, for four different pulse structures.The pulses shown in figures 7(a) and (c) are standard pulses, while in (b) and (d) the HV of the linac has been set outside the optimum value in order to produce a pulse structure that differs from the rectangular pulse.The traces from the ICT and the diode are similar and the SiC detector is capable of tracking the pulse current effectively.The DPP calculated from the integrated signal of the SiC diode differs from the determined with the ICT by less than 5% in the four examples.Nevertheless, there are some overshoots in the diode signal that can be attributed to cable reflections and interference since neither the output of the diode nor its output cable are adapted to 50 Ω.It should also be mentioned that in the present setup the signal from the diode was superimposed with an electromagnetic interference pulse from the linac's modulator.This interference pulse was recorded when the linac was running and only the beam was switched off so that it could later be subtracted from the measurement signal with beam.
The temperature dependence of the SiC diode is presented in figure 8.The diode readings are normalized to the measurement at 19 °C.The sensitivity shows a slight linear decrease as the temperature increases within the measured range of 19 °C-38 °C.The sensitivity variation with temperature calculated from the linear fit is (-0.079 ± 0.005)%/°C.

Discussion
Under 20 MeV electron beams, 4H-SiC diodes with a 1 mm diameter and a 3 μm active thickness, operated without external bias, exhibit a pre-irradiation sensitivity of ∼1 nC Gy −1 .This response allows for measurements under conventional dose rates, enabling the use of SiC diodes for cross-calibration between conventional and UHDR modes.Furthermore, the high signal yield of SiC makes it possible to manufacture diodes at a millimetre scale, enabling precise spatial resolution measurements.
Deviations of response linearity of solid-state diode dosimeters operating at zero bias in pulsed radiation beams at high dose rates have been correlated with the presence of a series resistance in the diode that, with high instantaneous radiation-induced currents, produces a voltage drop opposite to the built-in potential that hinders charge collection (Kranzer et al 2022, Marinelli et al 2022).This series resistance is caused by the metalsemiconductor contact resistances, the resistance of the metal layers, the finite conductivity of the semiconductor material and other general resistances.In our investigation, a diode dosimeter with an extended  range of linear operation for high-dose pulsed radiation has been achieved by a combination of low radiationinduced instantaneous currents through reduced sensitivity by using the wide band gap semiconductor 4H-SiC, a small sensitive area, and a series resistance of tens of ohms through an optimized fabrication process.In addition, the charge collection efficiency at zero bias has been enhanced by implementing a graded P-N junction in the SiC that results in a large built-in voltage of ∼2.5 V.
Although recent studies have shown promising results of silicon carbide diodes with linear performance up to a few Gy per pulse (Romano et al 2023), this has been achieved only under the application of high bias voltages of hundreds of volts to ensure proper charge collection.In contrast, the SiC dosimeter presented in this paper is the first, to the authors' knowledge, that does not need any applied voltage to respond linearly with an acceptable deviation of <3% up to at least 10 Gy per pulse, as shown in figures 3 and 6.This is a much larger linear range than the observed with a commercially available silicon diode irradiated under the same conditions.Furthermore, as shown in figure 6, the linearity with DPP is maintained up to at least 6.5 Gy per pulse after the SiC diode has been irradiated to more than 100 kGy of 20 MeV electrons.
In terms of sensitivity to accumulated radiation dose, in the radiation test with 20 MeV electrons reported in figure 5 the SiC diode exhibits a decrease of the sensitivity of only 0.018%/kGy.This value is in line with the observed in SiC diode prototypes irradiated with 9 MeV electrons to comparable cumulative doses (Romano et al 2023).It has to be noted, however, that this measurement was performed on the same diode that had already been irradiated to tens of kGy in the dosimetry tests carried out before and does not represent the behaviour of the SiC device during first irradiations.In fact, during the initial tests of the diode, a stronger decrease of sensitivity with the accumulated dose was observed.This suggests a SiC behaviour similar to silicon, in which the creation of radiation-induced defects in the crystal is more pronounced at the start of the irradiation and tends to saturate after a certain amount of accumulated dose (Grusell and Rikner 1984, Bruzzi 2016, Bueno et al 2022).Therefore, a more systematic evaluation of the SiC response as a function of radiation dose is required.
The SiC performance with accumulated radiation dose can also be compared with the shown by commercially available silicon diodes: for example, the signal of PTW's microSilicon detector decreases 0.5% per kGy when irradiated with 10 MeV electrons, according to the manufacturer (PTW 2019).Furthermore, the sensitivity decrease measured with a Sun Nuclear EDGE diode is 0.4% per kGy of 10 MeV electrons  (Rahman et al 2023).Similar values can be extracted from the characterization of different silicon prototypes (Bruzzi 2016, Bueno et al 2022).The better dose response stability of the SiC can be explained by the larger displacement energy of the 4H-SiC crystal compared to silicon (table 1) that leads to lower creation of damage (Sellin andVaitkus 2006, Rafí et al 2020).On the other hand, CVD diamond is still the most robust material for extreme radiation environments with a reported stable response after an integrated dose of the order of 3 MGy (Kranzer et al 2022), although the material cost is much higher and its availability is limited.The intrinsic energy dependence of SiC detectors is expected to be smaller than that of current state of the art silicon devices.The differences observed in electron beam PDD in figure 4 exhibit a dependence attributable more to the dose per pulse and beam time structure than to the energy dependence of the SiC diode.Nevertheless, additional experiments performed with SiC diodes from this work in megavoltage photon beams (6 MV and 15 MV) showed an excess of signal up to 7% for output factors of 40 cm × 40 cm when normalized to 10 cm × 10 cm.These results are currently being carefully investigated to determine the effect of metallic layers and encapsulation on energy response in order to improve the design and performance of the current SiC detector prototypes.
Several studies have shown that SiC detectors can achieve time resolutions comparable to that of silicon detectors, with nanosecond time resolutions measured in both Schottky and PiN SiC diodes with applied external bias (De Napoli 2022).This makes SiC detectors suitable for applications that require fast and accurate detection of radiation pulses.In this work, the SiC diode operated without bias has been able to follow the temporal structure of the 20 MeV electron beam even for irregular pulse structures.Nevertheless the SiC temporal response shows some anomalous features that can be attributed to the non-optimal electrical connection (impedance mismatch) of the SiC prototype.Therefore, although it is shown that the diode is fast enough to follow large fluctuations within a linac beam pulse, its output would have to be matched and optimized to transmit the high-frequency signal without reflections and interference.
Regarding the temperature stability, the SiC diode shows a small reduction in sensitivity with increasing temperature, in contrast to the positive trend observed in silicon diodes.The sensitivity variation with temperature is -0.08%/°C.This is lower in absolute value than the measured in a variety of silicon diodes produced using different technologies, that are typically within the range of 0.2%-0.5%/°C(Grusell and Rikner 1986, Saini and Zhu 2002, Ozleyis Altunkok et al 2015, Akino et al 2020), with some exceptions with better stability like some unirradiated diodes (Grusell andRikner 1986, Saini andZhu 2002) and PTW's microSilicon (Akino et al 2020).It has to be noted that at the moment of this measurement, the SiC diode had already been exposed to a total dose exceeding 100 kGy of 20 MeV electrons.In that regard, while it has been demonstrated for silicon diodes that the temperature coefficient depends on the irradiation history of the diode (Grusell and Rikner 1993), a similar study for SiC diodes is yet to be done.
Finally, it is important to mention that the dosimetric characterization presented in this paper was done with just one SiC diode sample due to time limitations and might not be representative of the entire batch.A complete study of the radiation response across different diodes is yet to be performed.However, given that the SiC diodes were batch-produced on a wafer level using highly reproducible fabrication processes and production-grade SiC wafers, it is expected that they show a reproducibility on par with silicon diodes, both within and across wafers.Indeed, the good electrical characteristics of the SiC diodes, together with the high fabrication yield achieved, underscore the reliability of the manufacturing process and the high quality of the SiC wafers currently available.This suggest that it is possible to scale up the SiC detector technology to larger devices and complex diode arrays, opening the door to new applications in medical dosimetry and beyond.

Conclusions
Silicon carbide diodes are today a real alternative to silicon in a wide range of applications where accurate realtime relative dosimetry, fast response and long-term stability are required.In addition, the technological maturity of SiC also makes it a suitable candidate for large-area detectors where diamond is not a realistic solution and silicon is limited by its moderate radiation hardness, variation of response with dose per pulse and temperature sensitivity.
Future work includes the systematic characterization of the SiC diodes in a wide range of beam configurations with electrons and protons and the validation of the other detector structures fabricated, including diodes with lateral etching to improve the precision of the measured dose, and pixelated detectors to produce 2D dose maps.

Figure 1 .
Figure 1.(a) Schematic diode cross-section (not to scale).(b) Optical photograph of one of the fabricated 100 mm 4H-SiC wafers showing a multiplicity of test devices.(c) Optical microscope picture of the 1 mm diode.

Figure 2 .
Figure 2. Current-voltage characteristics of six representative SiC diodes with 1 mm diameter.

Figure 3 .
Figure 3. Upper panel: charge per pulse measured in the SiC diode as a function of the dose per pulse of 20 MeV electrons and linear fit to the experimental data.Lower panel: relative deviation of the diode signal from linearity.

Figure 4 .
Figure 4. Upper panel: percentage depth dose measurements by the SiC diode prototype and a PTW flashDiamond detector.The depth dose curves shown for SiC were obtained with different dose per pulse and pulse duration irradiations.The DPPs shown in the legend correspond to the maximum of the PDD curves.Lower panel: deviations of the SiC diode measurements with respect to the flashDiamond normalized to the maximum of the PDD.

Figure 5 .
Figure 5. SiC diode response in the range DPP <0.2 Gy as function of dose of 20 MeV electrons (the diode was pre-irradiated to tens of kGy).

Figure 6 .
Figure 6.Upper panel: signal response of the SiC diode and a microSilicon diode with 20 MeV electrons.The linear trend is shown by a straight line.Lower panel: relative deviation from linearity of detector reading (logarithmic x-axis for clarity).The uncertainty is estimated as ±2%.

Figure 7 .
Figure 7. Time-resolved signal of the SiC diode (red) compared with the beam current measured with the ICT (blue).

Figure 8 .
Figure 8. Temperature dependence of the SiC diode.

Table 1 .
Material properties of silicon, CVD diamond and 4H-SiC.
2.1.DevicesDevice design and fabrication were done at the Institute of Microelectronics of Barcelona (IMB-CNM, CSIC).