Optically stimulated luminescence detectors for dosimetry and LET measurements in light ion beams

Objective. This work investigates the use of Al2O3:C and Al2O3:C,Mg optically stimulated luminescence (OSL) detectors to determine both the dose and the radiation quality in light ion beams. The radiation quality is here expressed through either the linear energy transfer (LET) or the closely related metric Q eff, which depends on the particle’s speed and effective charge. The derived LET and Q eff values are applied to improve the dosimetry in light ion beams. Approach. OSL detectors were irradiated in mono-energetic 1H-, 4He-, 12C-, and 16O-ion beams. The OSL signal is associated with two emission bands that were separated using a pulsed stimulation technique and subjected to automatic corrections based on reference irradiations. Each emission band was investigated independently for dosimetry, and the ratio of the two emission intensities was parameterized as a function of fluence- and dose-averaged LET, as well as Q eff. The determined radiation quality was subsequently applied to correct the dose for ionization quenching. Main results. For both materials, the Q eff determinations in 1H- and 4He-ion beams are within 5 % of the Monte Carlo simulated values. Using the determined radiation quality metrics to correct the nonlinear (ionization quenched) detector response leads to doses within 2 % of the reference doses. Significance. Al2O3:C and Al2O3:C,Mg OSL detectors are applicable for dosimetry and radiation quality estimations in 1H- and 4He-ions. Only Al2O3:C,Mg shows promising results for dosimetry in 12C-ions. Across both materials and the investigated ions, the estimated Q eff values were less sensitive to the ion types than the estimated LET values were. The reduced uncertainties suggest new possibilities for simultaneously estimating the physical and biological dose in particle therapy with OSL detectors.


Introduction
The use of protons and heavier ions for radiotherapy treatments is increasing (Grau et al 2020), but the variation of the relative biological effectiveness (RBE) with the local ionization density remains a challenge (McNamara et al 2015, Hahn et al 2022. To predict the biological effect in ion beams, RBE-models are applied, where the linear energy transfer (LET) is a central parameter. For a more complete reporting of ion beam therapy, Report 93 (ICRU 2016) suggests to report LET-related quantities for treatment plans, e.g. dose-weighted LET distributions. For proton therapy, where a constant RBE is still clinical practice, a direct scaling of RBE with LET may be implemented in treatment planning systems. Methods such as LET painting have been developed to best handle and account for LET distributions for RBE calculations (Bassler et al 2010). Nevertheless, LET is defined for a single ion type and energy in Report 16 (ICRU 1970). For a mixed radiation field, various definitions of averaged LET have been developed.
Due to the lack of a general or reference LET detector, however, the measurement and experimental validation of such LET-optimized plans is challenging. Active detectors may be capable to resolve either the microdosimetric spectrum (Conte et al 2020) or lead to a derivation of the particle spectrum using e.g. timeresolved silicon detectors (Nabha et al 2022). Active detectors are nonetheless limited by their fluence saturation in raster scanning delivering systems, or by a physical size that hinders their use inside anthropomorphic phantoms. Volumetric detectors have been suggested for LET measurements (Maeyama et al 2022, Nielsen et al 2022, but they exhibit an ionization quenched response (Hoye et al 2017). Passive detectors like fluorescent nuclear track detectors are also capable of estimating both LET and dose (Klimpki et al 2016, Muñoz et al 2022, but are challenging to use for clinically relevant doses due to track overlapping. Alternatives for LET measurements with passive detectors are based on thermoluminescence detectors (TLDs) (Vana et al 1996) or, in recent years, optically stimulated luminescence detectors (OSLDs) (Sawakuchi et al 2010. Particularly, the Al 2 O 3 :C OSLD was shown to be able to simultaneously determine the LET and dose at a point of interest (Granville et al 2016). Further work demonstrated its capability for LET determinations in clinically relevant proton beams (Granville et al 2014a, 2014b, with indications of possible LET measurements in ions heavier than protons (Yukihara et al 2015). Also shown to be dose-rate independent, Al 2 O 3 :C OSLDs can be used for dosimetry and to derive the LET under ultra-high dose rate conditions (Christensen et al 2021). Finally, its high sensitivity and small size means that it could be utilized as a point-like object and fitted into phantoms to derive the radiation quality at the point of interest.
Another candidate for LET determinations is Al 2 O 3 :C,Mg, which is a material with similar properties to Al 2 O 3 :C (Rodriguez et al 2011). Al 2 O 3 :C,Mg was initially developed as fluorescent nuclear track detector (Akselrod et al 2006), but both its radiophotoluminescence and optically stimulated luminescence (OSL) properties can be correlated to the LET in light ions , De Saint-Hubert et al 2021. Al 2 O 3 :C and Al 2 O 3 :C,Mg are characterized by OSL emission bands in the blue and UV regions upon green light stimulation . Sawakuchi et al (2010) demonstrated that the ratio of the two emission bands correlates well with the doseor fluence-averaged LET in protons. Given an LET calibration, the ratio of the UV and blue emission bands can then be used to estimate the average LET in the OSLD, whereupon the estimated LET can be applied to correct the ionization quenched dose derived from the blue emission band (Granville et al 2016). Yukihara et al (2015) demonstrated that LET measurements of light ions using Al 2 O 3 :C OSLDs is problematic due to saturation effects, and that the response correlates poorly with dose-averaged LET.
A broader challenge in LET determination is caused by the varying track structure of ions, where the radial dose distribution spans several orders of magnitude in light ion beams. The distinct radial energy distributions of two different ions with the same LET will thus result in two different detector responses, i.e. a single LET value is insufficient to accurately predict the detector response when measured with an OSLD (Yukihara et al 2015) or TLD (Olko et al 2002). This motivates the search for other radiation quality metrics which are less sensitive to the specific ion type.
Within this study, we propose the use of Q eff to act as a radiation quality metric for ionization quenching corrections. Q eff is related to the LET and depends on the ion's effective charge and speed. Initially used for particle emulsions (Barkas 1963) and amorphous track structure models (Katz 1978), Q eff was recently demonstrated to correlate better with the RBE than LET (Lühr et al 2017, Kalholm et al 2022. Therefore, this study aims at investigating the use of Al 2 O 3 :C and Al 2 O 3 :C,Mg OSLDs under irradiation with mono-energetic 1 H-, 4 He-, 12 C-, and 16 O-ion beams for dosimetry, and Q eff , and LET determinations. To improve the accuracy of previous OSL studies in light ions (Yukihara et al 2015), the measured OSL signals were here corrected using automatic reference irradiations of each detector in line with Christensen et al (2022). Furthermore, we investigated whether Q eff provides a better prediction of the detector response than LET does, and, ultimately, improves the dosimetry.

Radiation quality metrics
The radiation quality Q eff is independent of the material density and given as a function of the relativistic speed β = v/c as is the effective charge of a particle with charge z and speed v (Barkas 1963). To calculate an averaged LET, a distinction is made between the LET calculated by weighting it by the dose (d-LET) or fluence (f-LET), where the latter also is referred to as track-averaged LET. The Monte Carlo scoring of the averaged values follows the implementation defined in Cortés-Giraldo and Carabe (2015) as method C. In this work, Q eff is exclusively used in its fluence-averaged definition. Q eff , f-LET, and d-LET are henceforth collectively referred to as the radiation quality metrics.

Optically stimulated luminescence detectors 2.2.1 Detector preparation
The Al 2 O 3 :C and Al 2 O 3 :C,Mg OSLDs consist of a (50 ± 3) μm Al 2 O 3 layer mixed with a binder on a 75 μm polyester substrate as described in Ahmed et al (2014). The detectors were cut to ø4.5 mm and optically bleached with a green LED prior to irradiation to erase previous signal caused by background irradiation. For each irradiation, the OSLDs were placed in custom made opaque polylactide (PLA) containers (5 cm × 5 cm × 0.3 mm), each accommodating five detectors of both OSL materials in a centered 2 cm × 2 cm grid. The thickness of the PLA cover shielding the OSLDs from external light was measured to be 1.0 mm with a density of 1.15 g cm −3 .

OSL readout and reference irradiation
The blue and UV emission bands of the Al 2 O 3 :C and Al 2 O 3 :C,Mg detectors were measured using a time-resolved OSL readout technique . The OSLDs were stimulated with 100 μs pulses and the two components were separated using the stimulation parameters detailed in Christensen et al (2022). Due to the build-up rate of the UV emission band, the OSLDs were read out between (14-19) days after the irradiation to minimize the variation of the OSL intensities during the course of the readouts. The fading rate of the blue emission band is approximately constant after a week . Both signals from the blue and UV emission bands were normalized to the signal 14 days after irradiation with fading or build-up corrections <1 %. Each OSLD was subject to the same automated readout sequence in a Risø reader (TL/OSL-DA-20, DTU Nutech, Denmark). The sequence is detailed in Christensen et al (2022) and consists of (i) a 300 s readout of an irradiated OSL using green light stimulation giving the integral signal S, (ii) a 30 s irradiation with the build-in 90 Sr/ 90 Y beta source corresponding to ≈1 Gy, followed by (iii) a second 300 s readout of the reference irradiation giving the integral signal S ref .
The OSL intensity after ∼250 s stimulation is generally 3-4 orders of magnitude lower than the initial OSL intensity. This allows for an estimation of the background OSL signal from the last 10 s readout, which subsequently is subtracted from the OSL curve. Here, the signal S describes the integral luminescence of either the blue or UV emission band from the irradiation in an ion beam, whereas the ratio S/S ref represents the ratio between the luminescence signal after the unknown dose and the luminescence signal after the irradiation using the built-in beta source, both signals defined and calculated in the same way. Although the quantity S/S ref is sensitive to the irradiation history of the OSLD, it normalizes the OSL signal to the detector size and sensitivity when all OSLDs are subject to the same dose.
Whilst this work models the OSL response based on the integral OSL signal, Sawakuchi et al (2010) demonstrated that the shape of the OSL curve also can be used to infer the radiation quality. Relating the shape of the OSL curve to the radiation quality has the advantage that a constant background signal is irrelevant. On the other hand, the OSL curve shape is more sensitive to data fluctuations at low doses than an integral OSL signal is (Sawakuchi et al 2010). The use of the OSL shape to determine the radiation quality is not pursued in this work.

Irradiations
All OSLDs used for this work were irradiated at ≈0.25 Gy at the Center for Proton Therapy (CPT) at the Paul Scherrer Institute (PSI) or the Heidelberg Ion-Beam Therapy Center (HIT). At each facility, the OSLDs were contained within a 2 cm × 2 cm grid and the dose at the OSLD position was measured with an ionization chamber with its effective measurement point coinciding with the OSLDs. For irradiations at both centers, the ionization chamber measurements were corrected by as described in the TRS-398 (IAEA 2000).
The effect of overlapping ion tracks may affect the OSL emission intensities but requires further studies. To exclude such an effect from this study, the dose was chosen to be sufficiently low for the effect of track overlap to be negligible (Granville 2015), while still providing a luminescence signal equivalent to at least 100 mGy for the heavily ionization quenched 16 O-ions.

PSI irradiations
The irradiations with protons at the CPT were performed with the OSLDs positioned below 2 cm solid water (RW3 Water Slab, PTW Freiburg, Germany) in (70, 100, 230) MeV beams. The dose to each OSLD measurement point was measured with an ionization chamber (Advanced Markus TM34045, PTW Freiburg, Germany).
Measurements at lower energies were obtained by varying the upstream solid water thickness in the 70 MeV field as detailed in table 1, while scaling the fluence to achieve a 0.25 Gy dose at the OSLD location.
The dose calibration used for all readouts were obtained by delivering the 230 MeV protons at different doses to the OSLDs placed below 2 cm solid water, as detailed in Christensen et al (2022).

HIT irradiations
The irradiations with 1 H-, 4 He-, 12 C-, and 16 O-ions at HIT were conducted with the OSLDs placed behind 0.5 cm of the same type of solid water slabs used at PSI. This thickness was used to ensure charge particle equilibrium, even for the highest energies, at the point of interest. The dose was measured for each ion and energy using an ionization chamber (Farmer 001714, PTW Freiburg, Germany). For protons, an energydependent radiation quality-correction factor was used, which is parameterized in terms of the residual energy (IAEA 2000). For 4 He-and 12 C-ions a constant radiation quality-correction was applied, which is based on calorimetry measurements (Holm et al 2021).
Each ion/energy combination is detailed in table 1. A 3.0 mm PMMA ripple filter was used during the measurements with the 4 He-, 12 C-, 16 O-ions (Weber and Kraft 1999). The ripple filter is used in clinical setups to increase the the Bragg peak's width and was included to facilitate the validation of the simulations by comparing the measurements to HIT's reference data (Parodi et al 2012).

Monte Carlo simulations
The experimental setups used at PSI and HIT were simulated using Monte Carlo particle transport methods to calculate the radiation quality parameters at the OSLD positions. The grid resolution in the Monte Carlo scoring with TOPAS was set to 0.5 mm to approximate the detector thickness. For the scoring of Q eff , f-LET, and d-LET, secondary particles heavier than the primary particle were excluded to avoid biasing the d-LET for 1 H-and 4 Heions. The kinetic energy of the primary particles was scored at the OSLD position. The LET was scored for water at density 1.0 g cm −3 . Delta-particle equilibrium was assumed, so unrestricted LET was used, i.e. secondary electrons in the LET averaging were excluded to avoid double counting their contribution. The software and Table 1. Al 2 O 3 :C and Al 2 O 3 :C,Mg OSLD irradiation conditions used for the measurements at PSI and HIT, as well as the Monte Carlo simulated derived quantities. The given water-equivalent depth includes the RW3 slabs and the PLA container in front of the OSLDs in the irradiation with particles of initial energy E initial . The Monte Carlo simulated energy in the OSLD (E OSLD ) only includes contributions from the primary particles. The Q eff defined in equation (1), as well as the fluenceand dose-averaged LETs, include the contributions from primary and secondary particles. The LET values are given for water.

Irradiation conditions
Monte Carlo simulated results H-ions at CPT and in 1 H-, 4 He-, 12 C-ions at HIT.

Q eff and LET calibrations
The ratios of the UV and blue emission bands from Al 2 O 3 :C and Al 2 O 3 :C,Mg were parameterized as a function of each of the Monte Carlo derived radiation quality metrics Q eff , f-LET, d-LET. To model the data and establish a relationship between the OSL response and each of the three metrics, empirical functions were fitted to the data. The three models, described through an arc-tan, an error-function, and a logistic function are defined as

Relative effectiveness
To describe the detector response in the different radiation qualities, the relative effectiveness is introduced to relate the relative response of the OSL detector to the reference ionization chamber. The relative detector efficiency is here defined using the iso-dose definition 3 Results and discussion

Monte Carlo simulations
The simulated depth-dose curves for 1 H-ions at PSI and 1 H-, 4 He-, and 12 C-ions at HIT are shown in figure 1, along with the ionization chamber measurements. The overall agreement between the measurements and simulations validates the use of the Monte Carlo models to simulate the Q eff and LET at the OSLD positions. Dose to water, Q eff , fluenceand dose-averaged LET in water; Excluding secondaries heavier than the primary particle. # Histories 10 8 primaries for each radiation quality

LET and Q eff distributions
A comparison between the simulated Q eff and LET distributions in He-ions at HIT is shown in figure 2. As a particle slows down, Q eff and f-LET are generally monotonically increasing functions along the central beam axis until the Bragg peak region. For d-LET, however, the heavy dose-weighted fragments released at high particle speeds results in a slightly lower d-LET gradient along the central axis. Along with the higher weight of heavy fragments, it indicates that it will be difficult to establish a unique relation between the OSLD response and d-LET. For each irradiation, the simulated values of the energy, Q eff , f-LET, and d-LET at the OSLD position are given in table 1.

OSL curves
Examples of OSL curves for Al 2 O 3 :C and Al 2 O 3 :C,Mg, where the blue and UV emissions are separated, are shown in figure 3 for 1 H-, 4 He-, and 12 C-ions measured at HIT. For each irradiation, between four and five OSLDs were read out, as shown in each sub-figure. The OSLDs may differ slightly in size and sensitivity, which is  evident from the OSL curves in figure 3(a). Nonetheless, the integral signal of each curve S is eventually normalized by the signal S ref from the reference irradiation in the reader, which cancels out the differences .

Detector efficiencies in ion beams
The relative detector efficiency defined in equation (5) is plotted in figure 4 for 1 H-, 4 He-, 12 C-, and 16 O-ions for both Al 2 O 3 :C and Al 2 O 3 :C,Mg as a function of Q eff . The relative detector efficiencies as a function of f-LET and d-LET are included in the supplementary materials figure A.1 for reference. As also observed in figure 3(e), the intensity of the UV emission for Al 2 O 3 :C increases for particles near the Q eff or LET range relevant to 4 He-ions, and exceeds the efficiency relative to fast 1 H-ions in figure 3(d). The UV efficiency of Al 2 O 3 :C,Mg exceeds unity but is lower than the increase of Al 2 O 3 :C. Indeed, the relative efficiency close to unity for the Al 2 O 3 :C,Mg UV emission band for 1 H-and 4 He-ions indicates its relevance for dosimetry in 4 He-ions with smaller correction factors for ionization quenching than e.g. for Al 2 O 3 :C.
Whilst the relative detector efficiency for the blue emission band in figure 4(a) is relatively similar for Al 2 O 3 :C and Al 2 O 3 :C,Mg ions, the relative efficiencies for the UV emission band between the two materials differ. Thus, the ratio of the UV/blue emission band as a function of Q eff or LET will differ for the two OSL materials Al 2 O 3 :C and Al 2 O 3 :C,Mg.
Interestingly, the relative detector efficiency of two different ions at the same value of Q eff is almost similar for each emission band, e.g. for the slow 4 He-ions and fast 12 C-ions in figure 4(b). This indicates that Q eff is less sensitive to the ion type than e.g. f-LET, where the relative detector efficiency for two different ions at the same LET value differs, as demonstrated in previous studies (Yukihara et al 2015, Christensen andAndersen 2018) or in supplementary material figure A.1.
Due to the non-monotonic behavior of the UV emission band in both materials, shown in figure 4(b), the relative detector efficiency as a function of Q eff or LET is modeled with cubic splines shown in figure 4 as solid lines.

Detector response
The ratio of the two emission bands as a function of Q eff is exemplified for Al 2 O 3 :C in figure 5. The figure highlights the two cases where the uncorrected UV/blue emission ratio is shown in figure 5(a). The referencecorrected UV/blue emission ratio in figure 5(b), where the latter is corrected using the reference irradiation signals S ref as described in section 2.2.2.
In each case, the three models in equations (2)-(4) are fitted to the data as given in the figure legend. Although the figure only shows the relationship for Al 2 O 3 :C as a function of Q eff , similar results were obtained for Al 2 O 3 :C,Mg and parameterized as a function of f-LET or d-LET.
The use of the reference irradiation S/S ref serves both to reduce the variation in the group of OSLDs irradiated in the same radiation quality , but also provides a different response between ion types. Henceforth, only the reference-corrected OSLD readouts are used for dosimetry and to estimate the radiation quality metric.  Furthermore, the use of the reference corrections in figure 5(b) appears to give a well-defined relationship for 1 H-and 4 He-ions across all energies, which indicates a possible common calibration for ions with charge Z 2. The latter could be particularly advantageous for dosimetry applications in clinical 4 He-ion beams, where the mixed fields are mostly composed of He-and H-ion species.

OSLD determination of the radiation quality
The following investigates the ability to determine the radiation quality for both OSLD materials. This prediction is a challenge due to the effect of track structures, which causes the OSL response to vary for two different ion types at the same radiation quality, and further complicated by the apparent saturation at high ionization densities as observed in figure 4. The section is divided into four parts: one for each of the two OSL materials, which then is subdivided depending on the investigated ions for (i) 1 H-and 4 He-ions, and for (ii) 1 H-, 4 He-, and 12 C-ions. Due to its saturation, 16 O-ions are omitted in the following analysis but included in the figures for reference. Although other functions than the ones described in equations (2)-(4) may describe the data slightly better, it is not expected to change the overall conclusions.

Al 2 O 3 :C response in 4 He-ions
To establish the best relationship between the OSLD response and the radiation quality metrics, the three models are fitted to the Al 2 O 3 :C data for 1 H-and 4 He-ions in figure 6. The 1 H-and 4 He-ion data are well-defined as a function of Q eff but splits slightly up for f-LET depending on the ion type, as seen in the residual plots. For d-LET, in particular, a unique calibration to each ion type is needed to model the data. Hence, d-LET cannot be determined in a mixed particle field with Al 2 O 3 :C, which is in agreement with the conclusions in Yukihara et al (2015).
As evident from figures 6(a)-(c), the UV/blue ratio of slow 1 H-and 4 He-ions is too different from that of fast 12 C-ions to be modeled with a single curve. Thus, the use of Al 2 O 3 :C to determine the radiation quality metrics in 12 C-ions is challenging, as the 1 H-and 4 He-fragments will deviate from the 12 C-ion calibration curve, and will 3.5.2 Radiation quality metrics in 4 He-ions for Al 2 O 3 :C To assess how well the Al 2 O 3 :C OSLDs can be used to determine each radiation quality metric in 4 He-ions, the empirical models fitted to the data in figure 6 are used to evaluate the radiation quality metrics as shown in figure 7.
The lower subfigures show the deviation from the OSLD-derived metric relative to the Monte Carlosimulated one. For each radiation quality metric, the standard deviation of the data for the empirical function providing the best agreement is shown with a shaded area. The OSLD determinations of Q eff using the f atan model gives a standard deviation of 5 %, whereas that of the f-LET determinations was 10 %. This indicates, that Al 2 O 3 :C OSLDs can be used to determine the radiation quality in 4 He-beams.

Al 2 O 3 :C,Mg response in 4 He-ions
Unlike the results for Al 2 O 3 :C, the UV/blue ratio for Al 2 O 3 :C,Mg is well-approximated with a single model across 1 H-, 4 He-, and 12 C-ions in figure 8 for Q eff . Again, the data are better described with Q eff than LET, where each ion type follows a unique relationship. This indicates that Al 2 O 3 :C,Mg can be applied for dosimetry in 12 C-ions.
Following from the fit in figure 8, also a 4 He-beam can be described through Q eff . For consistency, the results where the empirical models only are fitted to 1 H-and 4 He-ions for Al 2 O 3 :C,Mg, are shown in the supplementary material figure A.3.

Radiation quality metrics in 12 C-ions for Al 2 O 3 :C,Mg
The determination of the radiation quality metric with Al 2 O 3 :C,Mg for ions lighter than 16 O-ions is displayed in figure 9. As for Al 2 O 3 :C in figure 7, the lowest standard deviation between the OSLD derived and the Monte Carlo simulated values are found for Q eff . When averaged LET is used as the radiation quality metric, the data are splitting up depending on the ion type.
However, in each case, the addition of the 12 C-ions gives a poorer fit to the data, than when only 1 H-and 4 He-ions are considered. The slowest 12 C-ions, which have the highest ionization densities, deviate the most from the fits, which could indicate a trend towards saturation. The results for Al 2 O 3 :C,Mg for a 4 He-beam are included in the supplementary figure A.4 with the same conclusion. Figure 7. The OSLD derived radiation quality for the metrics Q eff , f-LET, and d-LET for Al 2 O 3 :C and the fits to 1 H-and 4 He-ions in figure 6 plotted as a function of the Monte Carlo simulated values. The shaded band illustrates the standard deviation of the data for the model that describes the data the best, in each case the f atan function in equation (2). Again, a single function is unable to describe the mixed ion data for d-LET.

OSLD measurements in light ions
The ability of the Al 2 O 3 :C and Al 2 O 3 :C,Mg OSLDs to determine the radiation quality is now used to correct the ionization density dependent response. The estimates of the radiation quality metrics are used to determine the relative detector efficiency, as shown in figure 4 for Q eff . The dose is subsequently calculated for the intensities of both the blue and UV emission bands and corrected by the determined relative efficiency for each band.
The OSLD estimates of the radiation quality metrics and quenching-corrected doses are compiled in figure 10 for both materials. The results are presented as the relative error between the OSLD determined value and the true value, where the latter is taken to be either the dose measured with an ionization chamber or the radiation quality determined through the Monte Carlo simulations 3.6.1 Dosimetry in 4 He-beams For both the Al 2 O 3 :C and Al 2 O 3 :C,Mg OSL measurements in 1 H-and 4 He-ions in figures 10(a), (b), the determinations of Q eff are closer to the reference values than the LET determinations. This in turn gives a better dose correction and ultimately an improved dosimetry when Q eff is used to predict the radiation quality.
The doses in the mono-energetic 1 H-and 4 He-ions were determined with an average relative error on a 1.5 % level for both materials when Q eff is used as the radiation quality metric. The relative error increases to 2% when f-LET or d-LET is used.
3.6.2 Dosimetry in 12 C-beams When all 1 H-, 4 He-, and 12 C-ions are considered, only Al 2 O 3 :C,Mg is found able to predict the radiation quality with a relative error lower than 10 % relative to the simulated values.
Across all ions, Q eff was found to be the best describing radiation quality metric with a 9 % average relative error. The relative detector efficiency for Al 2 O 3 :C,Mg for the UV emission band in figure 4(b) is closer to unity for 4 He-ions than that of Al 2 O 3 :C, which means that the ionization quenching correction factors for Al 2 O 3 :C,Mg are smaller and vary less than those of Al 2 O 3 :C. Hence, the dose corrections using the Al 2 O 3 :C,Mg UV emission band are insensitive to small deviations in the radiation quality determinations, and the determined doses are all of similar quality with a relative around 3%. The dose determinations relying on the Q eff determinations are closer to the reference values than the ones relying on the LET determinations.
As Q eff was demonstrated to correlate better with the RBE than LET does (Kalholm et al 2022), these results suggest that not only the physical dose in light ion beams can be estimated with OSLDs but also the