To evaluate the radiological characteristics of STG, four smartwatch models were selected. According to market data from the first quarter of 2025, Apple held the largest share of the global smartwatch market (36%), followed by Samsung (10%), Huawei (7%), and Xiaomi (5%) (24). One representative model was selected from each manufacturer, and multiple identical models were used to evaluate the radiological characteristics. The analysis results of the material composition and physical specification of STG are shown in Table 1. The material composition was analyzed using inductively coupled plasma optical emission spectrometry and wavelength-dispersive X-ray fluorescence spectrometry at the Korea Polymer Testing & Research Institute. The radiological characteristics of STGs were not evaluated according to the molecular structure or material composition.

To apply STG in dose assessment, EPR spectrum should first be analyzed. According to previous studies (25–27), the EPR spectra of glass are influenced by various external factors. The BKS represents the intrinsic EPR spectrum obtained in the absence of external pretreatment or irradiation. The MIS corresponds to the EPR spectrum mechanically induced in STG, resulting from the mechanical stress of the samples. Additionally, LIS are generated by exposure to UV light, and RIS refer to signals induced by ionizing radiation. The BKS, LIS, and RIS were measured using powdered samples, while the MIS was measured using both bulk and powdered samples. The BKS was obtained by separating the STGs from smartwatches and performing annealing at 200 °C for 1 h using a dry oven (JEIO TECH, Republic of Korea) (26). Annealing was conducted to determine the BKS by removing potential non-specific signals induced during the manufacturing and transportation of the smartwatches. To evaluate the MIS, each STG sample was first measured in bulk form and then crushed into powder for repeated measurements. For accurate comparison, mass correction was applied to the MIS to minimize distortion caused by sample loss. The LIS and RIS were analyzed based on EPR spectral changes following UV light and gamma-ray stimulation. External stimulation of UV light and radiation was performed using a UV curing system (Skycares, Republic of Korea) and a BioBeam 8,000 irradiator (STS Steuerungstechnik & Strahlenschutz GmbH, Germany). The wavelength and fluence of the UV light were 365 nm and 2,905 kJ/m², respectively. Gamma irradiation was conducted using the BioBeam 8,000 irradiator at a dose of 20 Gy from a ¹³⁷Cs radioisotope source.

In retrospective dosimetry, it is often challenging to immediately assess the absorbed dose of exposed personal belongings after radiological incidents. Therefore, evaluating the time stability of the RIS is essential to ensure accurate dose assessment. The STG samples were irradiated with a radiation dose of 20 Gy, and their EPR spectra were monitored over 4 months. The samples were stored under controlled conditions at 23 ± 1 °C and 35 ± 2% relatively humidity, while being continuously exposed to visible light. The influence of visible light on RIS was assumed to be insignificant in this study.

Because EPR dosimetry relies on regression analysis derived from dose–response curves, it is necessary to evaluate the response of STG samples to different radiation doses. The EPR measurements for dose response were therefore conducted using STG samples irradiated with radiation doses of 0, 5, 10, 15, and 20 Gy.

To evaluate the variation in RIS caused by UV light exposure, 20 Gy-irradiated STG samples were assessed according to the amount of UV light exposure. The UV light consisted of approximately 95% UVA and 5% UVB, as UVC is completely absorbed by the stratospheric ozone layer in real situation (28). Using a UV light meter (LT Lutron, Republic of Korea) to measure natural sunlight, the intensity of UVB irradiation was found to be approximately one-hundredth that of UVA. Therefore, UVA was used as the representative component of actual UV light to evaluate the light effect on RIS. To assess the effect of UV light exposure time, it was assumed that the samples were exposed to an irradiance of 2.5 mW/cm², a daily sunlight exposure of 12 h, and the variation in RIS was compared with the time stability of RIS.

Smartwatches can be exposed to a variety of thermal conditions. Engin et al. (29) investigated the thermal effect on RIS under different temperature conditions showing that the RIS in window glass exhibited signal fading depending on the annealing and storage temperatures. STG samples were evaluated at various temperatures. To evaluate the thermal stability of RIS, the samples were subjected to thermal exposure in a dry oven at 50 °C for 1, 8, and 24 h. Additionally, to assess the effect of relatively high-temperature exposure, samples were exposed at 100 °C and 200 °C for 1 h.

Pretreatment is conducted to improve the accuracy of dose assessment and the efficiency of EPR measurement by reducing distortion of the EPR spectrum. In STG-EPR dosimetry, pretreatment involves chemical processing using a solution to remove surface impurities and adhesives, as well as preparing the sample in powdered form to improve measurement efficiency. Bulk samples were used to minimize sample loss during chemical treatment. Additionally, the samples were shaken between measurements, and repeated measurements were performed to reduce measurement variability caused by sample anisotropy. Chemical processing consisted of treating bulk STG fragments for 5 min with three cleaning solutions: distilled water, 99% ethanol (DUKSAN, Republic of Korea), and a commercial sticker remover (NABAKEM, Republic of Korea).

The minimum detectable dose (MDD) was determined according to the guidelines outlined in ISO 13304-2 (3). Establishing distinct criteria to differentiate between the BKS and RIS was crucial for accurately determining the MDD. An EPR signal was considered RIS when its intensity exceeded three times the standard deviation of the BKS, ensuring statistical significance (30).

A blind test was conducted under laboratory conditions to assess the potential of the preliminary STG-EPR dosimetry protocol designed based on the radiological characteristics of STG. Two smartwatches were irradiated with different radiation doses. To validate the protocol under a realistic post-accident scenario, the blind test was designed based on the assumption that the dose assessment would be conducted 5 d after the exposure and that the device was worn and used normally during this latency period. To simulate the environmental factors encountered during this period, the smartwatches were subjected to thermal exposure in a dry oven at 50 °C for 4 h and exposed to UV light using a UV curing system for 10 s after irradiation. Subsequently, the irradiated samples were stored under ambient conditions (relative humidity: 35 ± 2%, temperature of 23 ± 1 °C) for 5 d. STG samples were exposed to UV light at an intensity of 3,228 mW/cm², corresponding to a cumulative exposure of 968 kJ/m². This exposure energy is equivalent to sunlight UV exposure at 2.5 mW/cm² over a continuous duration of 0.98 d. After 5 days, the STGs were extracted from both devices and processed according to the established protocol. The evaluated doses were compared with the reference doses as follows:

where x represents the evaluated dose in the STG, X represents the reference dose, U_x is the expanded uncertainty of the evaluated dose in the STG, and U_X represents the expanded uncertainty of the reference dose. The E_n scores were calculated using Equation 1 based on ISO 13528(2022) (23).

The EPR spectra of the different types of STGs were obtained under the effects of annealing, mechanical stress, UV light exposure, and irradiation. The BKS was evaluated from powdered samples after annealing, and the results of the BKS in the STGs are shown in Figure 1. The EPR spectra obtained after annealing were determined as the BKS. STG-1, STG-2, and STG-4 exhibited identical EPR spectra before and after annealing. However, STG-3 showed spectral changes due to annealing. Consequently, the intrinsic signals remaining after thermal stabilization were used as the baseline for dose assessment. The subsequent evaluation of the radiological characteristics of the STGs was conducted using the annealed STG samples.

Figure 2 shows the results of STG samples crushed from bulk into powder form. MIS was not observed in the EPR spectra of the STGs. The slight EPR spectral differences observed in Figure 2 are attributed variations in the cavity Q-factor resulting from the difference in packing factor between the bulk and powdered samples (31). Additionally, no specific signals induced by mechanical stress were detected in the STGs.

The changes in EPR spectra were further evaluated following UV light exposure. Figure 3 shows the result of LIS in STGs. In STG-2, slight variations in the EPR spectrum were observed; however, after a few minutes, repeated measurements were consistent with the EPR spectrum of BKS. Consequently, LIS was not detected in STG-1, STG-3, or STG-4.

A clear distinction between BKS and RIS is essential for the practical application of STG in dose assessment. Figure 4 shows the EPR spectra obtained for the four types of STGs following irradiation with 20 Gy.

The g-factor of the STG samples were evaluated to be in the range of 1.998–2.008. The observed g-factors are characteristics of intrinsic radiation-induced defect centers in silicate glasses (32). The proportion of RIS in each STG sample was evaluated relative to the overall spectrum, including both BKS and RIS for the following smartwatch models: STG-1 (0.31%), STG-2 (86.02%), STG-3 (14.11%), and STG-4 (79.89%). STG-2, STG-3, and STG-4 exhibited distinct RIS, whereas STG-1 showed no detectable EPR spectra, indicating limited applicability for retrospective dosimetry. Therefore, STG-1 was excluded from subsequent analyses.

Figure 5 shows the time stability of RIS for three types of STG. The results indicated distinct differences in RIS time stability among the samples. The RIS intensity decreased by 89.47% in STG-2, 64.71% in STG-3, and 8.48% in STG-4 relative to their initial EPR intensities. While the time stability of RIS for each STG allows for time-dependent dose assessment, STG-4 is a promising candidate for retrospective dosimetry, particularly when the exact time of radiation exposure is unknown.

To evaluate the absorbed doses using STGs, analysis of the linear correlation between dose response and radiation dose is required. The dose–response curves of the different STGs are shown in Figure 6. STG-2 showed a clearly distinguishable RIS from the BKS; however, owing to its low RIS stability, its long-term applicability is limited. STG-3 demonstrated minimal differences between RIS and BKS in the EPR spectrum, whereas STG-4 exhibited a distinct linear correlation between irradiation dose and RIS intensity.

STG samples were exposed to UV light to evaluate the variation of RIS. Figure 7 shows the RIS variation under UV light and non-UV light exposure. The EPR intensity of RIS under UV light decreased gradually over 35 d for STG-2, 10 d for STG-3, and 100 d for STG-4. The RIS in STG-4 decreased by 87.29% and was evaluated as a residual RIS that was not eliminated by annealing or UV exposure. UV light induced a rapid decrease in RIS compared with the result of time stability of RIS. The attenuation of RIS induced by UV light could lead to underestimation in EPR dosimetry; therefore, the absorbed dose should be calibrated using the RIS fading curve corresponding to UV light exposure.

STG-1, STG-2, and STG-3 were excluded from the thermal effect evaluation because no RIS was observed, or the RIS decreased rapidly. STG-4 was used to assess the RIS stability in STGs under thermal exposure. Figure 8 shows the RIS variation with respect to exposure time and temperatures.

As shown in Figure 8A, the RIS variation at 50 °C decreased by 1.77, 0.36, and 2.54% after thermal exposure for 1, 8, and 24 h, respectively. Although RIS showed slight variation at 50 °C, the statistical variation was not significant. The intrinsic temperature of STG-4 during both operation and charging remained below 30 °C. For STG-4, the RIS variation at 50 °C, 100 °C, and 200 °C was −2.59, 2.64%, and −87.74%, respectively. RIS exhibited low variation below 100 °C, whereas a rapid reduction was observed at 200 °C. The thermal effect on RIS in glass is important for BKS estimation in retrospective dosimetry (29).

Additionally, to evaluate the BKS estimation using the thermal effect on RIS, the STG-4 samples were annealed at 200 °C. The signal intensity of STG-4 irradiated with 20 Gy, measured after annealing at 200 °C was approximately 42.73% higher than the initial BKS. The excess signal was attributed to the stable residual RIS that persisted even after thermal treatment. Consequently, a method for BKS estimation accounting for the residual RIS was applied to determine the original BKS of STG-4.

For STG-1, STG-2, and STG-3, it was difficult to accurately evaluate the pretreatment effect on RIS due to the absence or fading of RIS in STGs. Therefore, STG-4 was used to evaluate the pretreatment effect on RIS. STG-4 was soaked in distilled water, ethanol, and a commercial sticker remover for 5 min. Figure 9 shows the RIS in STG-4 after 5 min of pretreatment, with changes of −1.23, +2.96%, and −0.82%, respectively.

STG-4, identified as the most promising STG among the four types, was selected for MDD evaluation. The minimum irradiation dose required to distinguish BKS from RIS in STG-4 was approximately 0.50 Gy. Furthermore, the MDD for STG-4 was determined to be 0.93 Gy. The MDD of STG (EPR) is higher than that of retrospective dosimetry based on tooth (EPR), bone (EPR), resistor (TL), and glass (OSL) (1).

The radiological characteristics of the four types of STGs were evaluated for their applicability in assessing doses for individuals exposed during a radiological accident. Among the STGs, STG-4 was selected for the preliminary STG-EPR dosimetry protocol. The distinction between BKS and RIS was clear, and no MIS occurred. The influence of UV light on RIS can be estimated and calibrated, while the reduction of RIS at daily operating temperatures was minimal. In addition, STG-4 exhibited excellent dose response and linearity.

A schematic overview of the preliminary protocol for STG-EPR dosimetry is shown in Figure 10. The procedure consists of four main phases: sample collection, sample preparation, EPR measurement, and dose assessment. First, in the sample collection phase, the smartwatch is obtained from the exposed individual. During sample preparation, the STG is separated from the smartwatch, and the exterior is cleaned using distilled water, ethanol, and a sticker remover to remove adhesives and surface impurities. The cleaned STG is then crushed and sieved to obtain a uniform particle size of 100–1,000 μm. A sample mass of 100 mg is loaded into an EPR tube for analysis. The EPR measurement phase involves several critical steps. An initial measurement is performed to assess the signal intensity. To ensure accuracy, the protocol applies a correction for signal fading caused by UV light using the established RIS fading curve. A dose–response curve is then constructed using the additive dose method. Additionally, BKS estimation is conducted by annealing the sample at 200 °C for 1 h to measure the residual signal. Finally, in the dose assessment phase, the estimated BKS is subtracted from the total signal to isolate the net radiation-induced signal, and the absorbed dose of STG-4 is determined based on the calibration curve.

To ensure the reliability of the preliminary STG-EPR dosimetry protocol, measurement uncertainty was evaluated in accordance with the Guide to the Expression of Uncertainty in Measurement (GUM) (33).

The uncertainties for pretreatment, time stability of RIS, reproducibility, thermal stability of RIS, BKS estimation, and light (UV) effect were evaluated as Type A uncertainties, derived from the standard deviation of experimental measurements. Specifically, the pretreatment uncertainty was evaluated based on the RIS variation occurring during chemical cleaning and sample crushing. The time stability of RIS was assessed using the variation of RIS measured over 4 months following 20 Gy irradiation. Reproducibility was determined by evaluating the stability of the EPR spectrometer using STG-4 powdered samples. The uncertainty due to thermal stability of RIS was evaluated based on the RIS variation at 50 °C for 24 h. The BKS estimation uncertainty was derived from the variation in radiation dose calculated using the BKS estimation method. The light (UV) effect on RIS was determined from the repeated measurement results of RIS fading after UV exposure equivalent to 0.98 days.

The calibration curve uncertainty was evaluated as a combined uncertainty incorporating the reference irradiation (Type B), mass of the sample (Type A), and regression curve (Type A). The uncertainty for the mass of the sample was derived from mass variations during repeated measurements, while the regression curve uncertainty was evaluated based on the residual standard deviation of the regression analysis (34). The reference irradiation uncertainty was obtained from the calibration of the irradiation system.

The considered contributing factors included pretreatment of the sample, time stability of RIS, reproducibility, thermal stability of RIS, BKS estimation, the light (UV) effect on RIS, reference irradiation, mass of sample, and regression curve. The measurement uncertainty components and their contributions to the STG-EPR dosimetry protocol are summarized in Table 2.

As shown in Table 2, the most significant contribution to the overall uncertainty was from the light (UV) effect on RIS, which was evaluated as 8.97% and accounted for 57.90% of the combined uncertainty. The uncertainties of the calibration curve and BKS estimation were 5.73 and 4.30%, accounting for 23.65 and 13.31% of the combined uncertainty, respectively.

To assess the preliminary STG-EPR dosimetry protocol, a blind test was conducted using two smartwatches containing STG-4. As shown in Figure 10, the absorbed doses of the two samples were determined to be 4.35 ± 1.02 Gy and 11.04 ± 2.60 Gy, respectively. The corresponding reference doses were 4.88 ± 0.32 Gy and 12.18 ± 0.69 Gy. The absorbed doses of STG-4 were underestimated compared with the reference doses. The difference between reference and evaluated doses is considered to be predominantly attributable to the UV light effect on RIS and BKS estimation. Overall, the E_n scores for blind tests 1 and 2 were −0.49 and −0.42, respectively, which are within the acceptance criterion of ± 1 (Figure 11).