Deciphering the hydrogen shuttle-mediated kinetic memory of Ge-related defects in UV-grade fused quartz via in situ EPR with 3D imaging Authors: Xiaobing Lou, Fushan Geng, Kejiao Shi, Bingwen Hu Published: 2027 ABSTRACT This study elucidates the photo-induced defect dynamics in standard UV-grade fused quartz (JGS2) by synergizing in situ electron paramagnetic resonance (EPR) spectroscopy with 3D volumetric imaging. We identify a hydrogen shuttle mechanism driving the interconversion between Ge-E′ and Ge-E′H centers, which triggers a significant kinetic memory effect dramatically increasing the generation rate of Ge-E′ by 20-fold in subsequent UV cycles. 3D EPR imaging, effectively mapping the internal defect distribution, demonstrates that defect activation is precisely confined to the UV-illuminated volumes. Long-term tracking confirms that while Ge-E′ is unstable, Ge-E′H remains highly persistent for several months. A potential energy landscape is proposed to unify these spatiotemporal observations, providing a robust thermodynamic framework for evaluating the radiation resilience and long-term stability of high-purity silica in advanced optical applications. Introduction Vitreous silica (SiO2) is a cornerstone material in modern optics, telecommunications, and laser technologies due to its exceptional transparency and thermomechanical stability [1]. However, its optical performance in harsh environments, such as under ultraviolet (UV) or ionizing radiation, is strictly governed by the presence of point defects and trace impurities [2,3]. Among these, germanium (Ge), which naturally occurs as a substitutional impurity in quartz or is intentionally doped into optical fiber cores, plays a paradoxical role [4,5]. While Ge-related oxygen deficient centers (GeODCs, primarily in the form of ≡Ge–Ge≡ or ≡Ge–Si≡) are essential for the photosensitivity required to write Fiber Bragg Gratings (FBGs), they are also the primary precursors for radiation-induced attenuation and optical degradation [6,7]. Under UV excitation, these GeODCs undergo photochemical conversion into paramagnetic color centers, predominantly the Ge-E′ (≡Ge·) centers, which are structurally represented as a three-coordinated germanium atom with an unpaired electron localized in an sp3 hybrid orbital pointing into an oxygen vacancy [8,9]. These paramagnetic sites interact strongly with hydrogen species commonly hosted within silica matrices (such as standard fused quartz or H2-loaded optical fibers) [10]. A hallmark of this interaction is the well-known 119 G hyperfine doublet observed in X-band electron paramagnetic resonance (EPR) spectra. Seminal EPR studies by Vitko and subsequent photoluminescence correlations by Agnello et al. have unequivocally assigned this doublet to a hydrogen-compensated substitutional germanium center (≡Ge· ··· H), named Ge-E′H, which was also conventionally designated as the H(II) center previously [11,12]. However, despite the clear structural assignment of these centers, their dynamic interconversion pathways, particularly the long-term evolutionary behavior under repeated irradiation, remain elusive. The rapid proliferation of high-power UV lasers and space-grade optical sensors has reignited the urgency to understand the non-linear dynamic evolution of silica-based materials [13,14]. Contemporary studies reveal that the degradation of silica optics is not a simple linear accumulation of defects but a complex process involving transient species and dark reactions [15]. For instance, recent investigations on Ge-doped fibers suggest that mobile hydrogen shuttling between precursor sites significantly accelerates defect conversion rates [16]. Crucially, the material's response to radiation depends heavily on its prior exposure history, yet the underlying thermodynamic landscape governing this history-dependent behavior is not yet fully understood. This knowledge gap is particularly critical for JGS2 quartz, where the high concentration of H species and germanium impurities creates a complex environment for hydrogen-mediated defect transitions. Despite its critical engineering implications, the real-time kinetics of this post-irradiation hydrogen trapping remains poorly resolved, primarily because most prior EPR investigations have relied on ex-situ, steady-state measurements. To bridge this gap, we report a spatiotemporal in situ EPR study. By synergizing time-resolved kinetics analysis, 3D volumetric imaging, and long-term evolutionary tracking, we unambiguously map the hydrogen shuttle mechanism and decode the physical origins of the kinetic memory effect in JGS2 quartz. Our real-time monitoring reveals a fascinating two-stage kinetic behavior. During continuous UV irradiation, the EPR spectrum is dominated by the generation of the central Ge-E′ signal. Subsequently, upon ceasing the UV exposure, a pronounced dark adaptation process unfolds, characterized by the rapid decay of the Ge-E′ color centers and the simultaneous, explosive growth of the Ge-E'H signal. This transformation can be continuously reversed by reapplying UV irradiation, which triggers the fast regeneration of Ge-E′ centers and a concomitant suppression of the Ge-E′H signal, reflecting a highly dynamic and reversible interconversion between these two defect states. By quantifying the reaction rates of this in situ process, we establish compelling evidence for the kinetic memory effect. These findings not only elucidate the microphysical origins of post-irradiation instability in silica but also offer critical kinetic parameters for modeling the long-term reliability of Ge-doped optical devices in hydrogen-rich environments. Experimental Materials preparation and characterization The JGS1, JGS2, and JGS3 quartz were ordered from Pangu Quartz in China. The size of the quartz was determined to be 7 × 7 × 100 mm, which is suitable for X-band EPR cavity, simultaneously maintaining maximal EPR signals. The quartz cuboids were surface polished to ensure uniform light exposure. The quartz cuboids were cleaned with ethanol and deionized water followed by drying at room temperature. Room-temperature UV–visible (UV–Vis) absorption spectra were recorded on a Shimadzu UV-2700 spectrophotometer. The measurements were performed in transmission mode over a wavelength range of 200–800 nm with a scanning step of 0.5 nm. Air was utilized as the reference blank to subtract the background absorption. Fourier transform infrared spectroscopy (FTIR) analyses were performed at a Nexus670 infrared spectrometer (Nicolet) over a wave number range of 500–4000 cm−1. In situ UV-irradiation EPR measurements To capture the real-time structural evolution and the underlying kinetics of color centers, in situ UV-irradiation EPR experiments were performed using an ultraviolet mercury lamp with filtered wavelength focused on 240–400 nm and a Bruker ELEXSYS E580 EPR spectrometer at X-band equipped with a SHQE resonator. Unless otherwise stated, the UV light with the maximum power of 200 W was focused and directed into the EPR cavity from bottom side through a high-purity fiber optic bundle to guarantee thorough illumination of the quartz cuboids. The common parameters for in situ EPR spectra were: microwave frequency of ca. 9.55 GHz, microwave power of 4.743 mW (15 dB), modulation amplitude of 1 G, modulation frequency of 100 kHz, conversion time of 10 ms, receiver gain of 60 dB, sweep time of 30 s with one scan each spectrum, no sampling interval, sweep width of 300 G, and 3000 data points. The common parameters for in situ EPR imaging (EPRI) were: gradient field of 10 G/cm, 40 mm field of view, modulation amplitude of 2 G, modulation frequency of 100 kHz, conversion time of 20 ms, receiver gain of 60 dB, sweep time of 40 s with one scan each spectrum, no sampling interval. In the in situ EPR measurements, a modulation amplitude of 1 G was consistently applied to optimize the signal-to-noise ratio and, crucially, to balance the detection efficiency for both the narrow central Ge-E′ signal (linewidth: 0.7 G) and the Ge-E′H doublet (linewidth: 4 G) within the same experimental sequence. While this exceeds the intrinsic linewidth of the central signal, quantitative accuracy was maintained through spectral fitting that incorporates instrumental broadening. For EPR imaging, the modulation was further increased to 2 G to prioritize sensitivity and to deliberately broaden the signal into a single, simplified envelope. This approach facilitates a more robust deconvolution during image reconstruction, as a single-peaked kernel minimizes numerical artifacts and noise amplification, thereby ensuring the spatial fidelity of the reconstructed defect maps. In situ EPR and EPRI data were analyzed using homemade MATLAB scripts. The thousands of EPR spectra continuously collected were segmented at the consistent position to obtain the separate peaks, which were then fitted to obtain the parameters including the line width, g factor, hyperfine coupling constant (A), and double integral intensity. The detailed processing procedure of EPRI data is as follows: The recorded projection spectra were first deconvoluted from the reference spectrum, and a Gaussian apodization was used to improve the signal-to-noise ratio. Then the two- or three-dimensional images were obtained by a filtered back projection. The EPRI images obtained are projected images showing the sum of all defect signals along a particular direction. Results and discussion The ultraviolet absorption characteristics of the metal-doped quartz are primarily governed by electronic transitions localized at structural defects which introduce discrete energy levels within the wide band gap of the silica matrix. To investigate the ultraviolet (UV) optical properties of the JGS2 quartz, its UV–Vis spectroscopy was performed, compared with two other common quartz (JGS1 and JGS3) used in optical devices as plotted in Fig. 1a. In sharp contrast to negligible and weak ultraviolet absorption of JGS1 and JGS3, the JGS2 quartz shows strong ultraviolet absorption. The peak centered at 243 nm can be ascribed to the UV excitation of GeODCs. As shown in Fig. 1b, FTIR was also conducted to evaluate the content of hydroxyl. The absorption strength at 3670 cm-1 displays the –OH content ranking order: JGS1>JGS2>JGS3. These infrared-active hydroxyl groups (-OH) primarily originate from structural silanols (≡Si–OH), metal-associated hydroxyls (such as ≡Al–OH and ≡Ge-OH), and interstitial molecular water. Concurrently, it is noteworthy that infrared-inactive species, including interstitial molecular hydrogen (H2) and bonded hydrides (≡Si–H and ≡Ge–H), coexist within the network. The hydrogen-bearing configurations play a decisive role in modulating the generation and transformation of radiation-induced color centers [17]. Under UV irradiation, they undergo indirect activation or localized photolysis to liberate mobile hydrogen species (comprising transient, highly reactive H0 radicals and mobile interstitial H2 molecules). The subsequent long-range migration of these liberated species proceeds via the thermally activated hopping of transient H0 radicals along the intrinsic μ-oxo oxygen backbone of the silica matrix. Alternatively, they transport through the continuous interstitial channels of the glass network via the slow molecular diffusion of the reconstructed H2 molecules [18].Fig. 1(a) UV–Vis absorption spectra of the three quartz samples. (b) FTIR spectra of the three quartz samples. (c) Schematic illustration of the in situ UV-irradiation EPR setup coupled with UV fiber optic bundle from the bottom. (d) The EPR spectra of the three quartz samples before UV irradiation and after irradiation for 0.5 h. (a) UV–Vis absorption spectra of the three quartz samples. (b) FTIR spectra of the three quartz samples. (c) Schematic illustration of the in situ UV-irradiation EPR setup coupled with UV fiber optic bundle from the bottom. (d) The EPR spectra of the three quartz samples before UV irradiation and after irradiation for 0.5 h. UV-induced energy absorption in JGS2 quartz can trigger the dissociation of labile bonds at precursor sites (such as Ge–H and GeODCs), driving local electronic redistribution and the migration of hydrogen species to form stable or metastable paramagnetic defects [16]. To understand this process, the in situ UV-irradiation EPR was conducted using the configuration as shown in Fig. 1c and Fig. S1. The as-prepared quartz cuboid was directly placed in the cavity of EPR tube. A high-purity fiber optic bundle was used to connect the ultraviolet mercury lamp and the EPR tube. Initially, before the UV irradiation, no peaks could be found in the EPR spectra of the three quartz. After continuous irradiation for 0.5 h, there is still no distinct EPR response in JGS1 and JGS3 quartz. However, for JGS2, a strong center peak and two weak peaks are observed as shown in Fig. 1d and Fig. S2. According to the previous literature, the strong peak can be assigned to the signal of Ge-E′, while the two weak peaks are the classical 119 G doublet, which is structurally represented as a paramagnetic Ge-E′ center elastically coupled with a trapped hydrogen node (Ge-E′H) [10,12]. To capture evolution dynamics of the UV-induced Ge color centers, continuous in situ experiment was conducted as plotted in Fig. 2a. The EPR spectra were collected every half minute. The UV light was turned on and off three times during the whole test. During the UV on, the central Ge-E′ signal keeps increasing, exhibiting an exponential build-up. Fig. 2b displays the EPR spectrum and the molecular model of the Ge-E′ center recorded after prolonged UV exposure. The signal can be characterized by an anisotropic g tensor of g∥ = 2.0018 and g⊥ = 2.0008, with a linewidth of 0.7 G. The high quality of the least-squares fit indicates that the Ge-E′ center is the dominant paramagnetic species under the current irradiation conditions. Meanwhile, a weak 119 G doublet is also generated during UV irradiation, indicating concomitant generation of Ge-E′H. After turning off the UV, the Ge-E′ signal displays an exponential decay, while the Ge-E′H doublet exhibits significant growth. Fig. 2c shows the classic Ge-E′H signal acquired 3.3 h after ceasing UV irradiation. The signal is characterized by a prominent 119 G hyperfine doublet resulting from the interaction between the Ge-E′ unpaired electron and a nearby hydrogen nucleus (1H, I = 1/2) as illustrated in the inset of Fig. 2c. The EPR spectra can be accurately reproduced by a numerical simulation using g = [2.0004, 1.9988, 1.9925] and hyperfine coupling constants A = [122.2, 117.2, 121.9] G, with an average Gaussian/Lorentzian hybrid linewidth of 4.2 G.Fig. 2(a) 3D diagrams of in situ UV-irradiation EPR spectra of JGS2. (b) The magnifying EPR signal of Ge-E′ after UV irradiation for 3.3 h, inset is the molecular model of the Ge-E′. (c) The magnifying EPR signal of Ge-E′H after UV off for 3.3 h, inset is the molecular model of the Ge-E′H. (d) EPR power saturation curve (EPR signal intensity as a function of the microwave power). (a) 3D diagrams of in situ UV-irradiation EPR spectra of JGS2. (b) The magnifying EPR signal of Ge-E′ after UV irradiation for 3.3 h, inset is the molecular model of the Ge-E′. (c) The magnifying EPR signal of Ge-E′H after UV off for 3.3 h, inset is the molecular model of the Ge-E′H. (d) EPR power saturation curve (EPR signal intensity as a function of the microwave power). To further distinguish the local environments and spin dynamics of the two observed paramagnetic species, EPR microwave power saturation analyses were conducted under their steady state as shown in Fig. 2d. In the low-power regime, both signals exhibit a linear increase. Crucially, both centers begin to deviate from linearity and reach their saturation thresholds at approximately the same microwave power (P ≈ 6 mW). However, as the microwave power increases beyond the saturation point, the intensity of the central Ge-E′ signal undergoes a pronounced decay, while the Ge-E′H signal plateaus, maintaining a relatively constant intensity without significant attenuation up to the maximum measured power. This distinct post-saturation behavior highlights a fundamental difference in their line-broadening mechanisms. The continuous decay of the Ge-E′ signal is characteristic of a more homogeneously broadened spin system, where excessive microwave power leads to population equalization between spin states. Conversely, the plateauing behavior of the Ge-E′H doublet is highly indicative of an inhomogeneously broadened line. We propose that the strong hyperfine coupling with the nearby hydrogen atom in the Ge-E′H complex introduces a distribution of local magnetic environments, which fundamentally alters the spin-spin relaxation (T2) pathways and prevents the macroscopic signal from decaying sharply under high microwave power. It should also be stated that the microwave power used in the in situ EPR experiment is 4.743 mW, which ensures appropriate signal intensity and is below the saturation point, allowing for an accurate quantitative assessment of the defect evolution kinetics. The striking in situ dynamic evolution of the central Ge-E′ signal and the 119 G Ge-E′H doublet is shown in Fig. 3. As depicted in the 2D contour map (Fig. 3a), the UV irradiation induces the gradual accumulation of the Ge-E′ center, while the cessation of UV triggers an immediate redistribution of paramagnetic species. The rapid decay of the Ge-E′ centers is accompanied by a simultaneous, explosive growth of the Ge-E′H signal. This dark reaction likely suggests that the Ge-E′ centers act as precursors that capture mobile hydrogen species to form the more thermally stable Ge-E′H complex upon removal of the light source. The excellent reversibility observed over multiple UV on/off cycles further underscores a highly dynamic photo-switching mechanism, governed by the detrapping and re-trapping of hydrogen species by Ge-E′ within the silica network.Fig. 3(a) In situ 2D EPR contour map capturing the dynamic evolution of paramagnetic species in JGS2 quartz during multiple UV irradiation cycles. (b) Time-resolved double integral intensities of the central Ge-E′ center (red) and the 119 G Ge-E′H doublet (black, scaled by 1/5). (a) In situ 2D EPR contour map capturing the dynamic evolution of paramagnetic species in JGS2 quartz during multiple UV irradiation cycles. (b) Time-resolved double integral intensities of the central Ge-E′ center (red) and the 119 G Ge-E′H doublet (black, scaled by 1/5). Quantitatively, the kinetic traces in Fig. 3b reveal a striking 7.5-fold increase (based on the original, unscaled data) in the absolute spin population of the Ge-E′H compared to its Ge-E′ counterpart during irradiation. This massive amplification suggests that the Ge-E′ signal detected under UV represents only a limited, transient subset of the total photo-activated defect pool. The fact that the Ge-E′ linewidth (0.7 G) remains constant rules out relaxation broadening and confirms that its low intensity is indeed a population-limited effect, dominated by a highly dynamic equilibrium between photo activation and hydrogen-mediated recombination. Upon ceasing the UV exposure, we suggest that the hydrogen species act as a spin trapping agent that uncovers the true extent of the activated defect reservoir. These mobile hydrogen species effectively capture and anchor a much larger population of non-paramagnetic activated sites, potentially consisting of EPR-silent three-coordinate Ge sites or unstable GeODC fragments, that were generated during the light-on phase. By rescuing these precursors into the long-lived Ge-E′H state, the system provides a more faithful representation of the total defect-generating capacity of the JGS2 quartz. To gain deeper insight into the photo-switching dynamics, the temporal evolution curves (Fig. 3b) were quantitatively analyzed. As summarized in Table 1, while most kinetic evolutions follow bi-exponential behavior, a single-exponential model is sufficient to describe the initial generation of Ge-E′ and the suppression of Ge-E′H, respectively. Specifically, the decay and growth processes were fitted using y=y0+∑Aiexp-kit and y=y0+∑Ai1-exp-kit, respectively, where Ai represents the amplitude of the i-th kinetic component. The excellent agreement between the experimental data and theoretical fits, evidenced by high coefficients of determination, provides a rigorous validation of the proposed kinetic regimes. The relative importance of each pathway was quantified by the fractional amplitude, αi=Ai/∑Ai, providing a direct measure of the contribution from fast and slow species.Table 1Kinetic parameters (rate constants and fractional contributions) obtained from the temporal evolution fits of the Ge-E′ and Ge-E′H EPR signals across consecutive in situ UV-on and UV-off cycles for JGS2 quartz. The data were fitted using single- or bi-exponential models. The values in parentheses represent the relative fractional amplitude of each kinetic component. All fitting curves exhibit high reliability, with the coefficient of determination (R2) no less than 0.95.Table 1Rates1st UV on1st UV off2nd UV on2nd UV off3rd UV on3rd UV offGe-E′ke1 (h−1)1.185.83(54.6%)23.6(47.5%)5.24(66.5%)22.8(64.5%)8.84(56.7%)ke2 (h−1)N/A0.624(45.4%)0.229(52.5%)0.396(33.5%)1.56(35.5%)0.919(43.3%)R20.9990.9980.9860.9970.9950.999Ge-E′Hkh1 (h−1)1.935.99(53.4%)24.55.27(67.6%)31.4(97.0%)9.23(58.3%)kh2 (h−1)N/A0.628(46.6%)N/A0.339(32.4%)2.87(3.0%)0.908(41.7%)R20.9500.9950.9580.9941.0000.997 Table 1Kinetic parameters (rate constants and fractional contributions) obtained from the temporal evolution fits of the Ge-E′ and Ge-E′H EPR signals across consecutive in situ UV-on and UV-off cycles for JGS2 quartz. The data were fitted using single- or bi-exponential models. The values in parentheses represent the relative fractional amplitude of each kinetic component. All fitting curves exhibit high reliability, with the coefficient of determination (R2) no less than 0.95.Table 1Rates1st UV on1st UV off2nd UV on2nd UV off3rd UV on3rd UV offGe-E′ke1 (h−1)1.185.83(54.6%)23.6(47.5%)5.24(66.5%)22.8(64.5%)8.84(56.7%)ke2 (h−1)N/A0.624(45.4%)0.229(52.5%)0.396(33.5%)1.56(35.5%)0.919(43.3%)R20.9990.9980.9860.9970.9950.999Ge-E′Hkh1 (h−1)1.935.99(53.4%)24.55.27(67.6%)31.4(97.0%)9.23(58.3%)kh2 (h−1)N/A0.628(46.6%)N/A0.339(32.4%)2.87(3.0%)0.908(41.7%)R20.9500.9950.9580.9941.0000.997 Kinetic parameters (rate constants and fractional contributions) obtained from the temporal evolution fits of the Ge-E′ and Ge-E′H EPR signals across consecutive in situ UV-on and UV-off cycles for JGS2 quartz. The data were fitted using single- or bi-exponential models. The values in parentheses represent the relative fractional amplitude of each kinetic component. All fitting curves exhibit high reliability, with the coefficient of determination (R2) no less than 0.95. The most striking feature of the kinetic data is the perfect mathematical symmetry between the decay of the Ge-E′ center and the concurrent growth of the Ge-E′H doublet during the dark phases (UV off). For instance, in the 1st UV-off cycle, the Ge-E′ signal decays with a fast rate constant ke1=5.83h-1 and a slow constant ke2=0.624h-1. Remarkably, these values are nearly identical to the growth rate constants of the Ge-E′H signal (kh1=5.99h-1 and kh2=0.628h-1, respectively). This highly synchronized mirror effect is consistently maintained across all subsequent cycles. Furthermore, their fractional amplitudes are also strictly correlated (e.g., ∼54% for the fast component in both species). This rigorous quantitative parity provides unequivocal proof that the disappearance of the bare Ge-E′ center is strictly coupled to the formation of the Ge-E′H complex, effectively locking in the proposed direct hydrogen-trapping mechanism. The necessity of a bi-exponential model to describe the dark adaptation kinetics indicates that the hydrogen capture is not a homogeneous process, but rather involves distinct spatial pathways. We attribute the dominant fast component (k1, accounting for ∼54%–68% of the total amplitude) to the short-range recombination. In this scenario, the hydrogen species liberated by UV irradiation remain in close proximity to the activated Ge sites, allowing for rapid re-trapping upon light removal. Conversely, the slow component (k2, accounting for ∼32%–46%) reflects a long-range diffusion process. Here, the diffusing hydrogen species are at a distant site or have migrated deeper into the silica network during irradiation, and thus require significantly more time to migrate back and encounter the active Ge trap sites. This bi-exponential nature provides a kinetic mapping of the structural disorder and the heterogeneous distribution of hydrogen in the JGS2 quartz matrix. Complementary in situ tests at a reduced power of 50 W (Fig. S3, Table S1, and Note S1) further confirm the robustness of the observed mechanism. Although the defect generation rate is slower under lower photon flux, the characteristic hydrogen shuttle behavior upon UV cessation remains consistently reproducible. Crucially, this prolonged, hours-long dark adaptation kinetic behavior strongly points to a diffusion-controlled process rather than a fast, barrierless electrostatic protonation pathway. According to established defect physics, both the Ge-E′ and Ge-E′H centers are inherently electrically neutral paramagnetic centers rather than ionic species. During dark adaptation, the diffusing hydrogen migrates as neutral complexes—either via the slow molecular diffusion of interstitial H2 or through a thermally activated hopping mechanism of transiently localized species (e.g., ≡Si–O(··· H0)–Si) across the intrinsic oxygen backbone. Although free neutral atomic hydrogen is highly transient at room temperature and thus eludes steady-state EPR detection, its continuous, trap-limited migration and localized capture at uncompensated ≡Ge· dangling bonds govern the slow, long-term evolution of the Ge-E′H complex. Interestingly, Table 1 also reveals a profound kinetic memory effect upon cycling. During the initial state (1st UV on), the generation of Ge-E′ proceeds slowly via a single-exponential process (ke1=1.18h-1). However, in the subsequent light cycles (2nd and 3rd UV on), the photo-activation rate undergoes an explosive, 20-fold acceleration (ke1>22h-1). This drastic kinetic memory effect suggests that cleaving the freshly formed Ge-E′H complex is far more energetically efficient than activating the intrinsic precursors (e.g., GeODCs) in the silica glass. Once the initial cycle is completed, the system enters a highly responsive photo-switching state, where the rapid interconversion between Ge-E′ and Ge-E′H becomes the dominant pathway, while the slow generation of new Ge-E′ centers continues to contribute to the overall defect population, evidenced by the presence of a small ke2 in subsequent cycles. This indicates that the UV-induced activation of GeODCs is a slow, ongoing process that will not be fully exhausted during several-hour exposure. While the central Ge-E′ signal continues to accumulate across successive cycles, the 119 G doublet reaches a nearly identical steady-state intensity during each irradiation phase. This indicates that the population of Ge-E′H under UV is primarily constrained by the hydrogen-detaching capability of the incident flux rather than the total size of the expanding defect pool. Such a behavior reflects a robust dynamic equilibrium where the highly efficient photo-induced dissociation of Ge-E′H maintains its concentration at a constant ceiling level, regardless of the overall increase in photo-activated precursors. While the in situ kinetic study provides profound insights into the temporal evolution of Ge-related centers, understanding their spatial heterogeneity is equally crucial for elucidating the underlying radiation-induced mechanisms. To this end, we performed EPRI to map the macroscopic distribution of these defects within the JGS2 quartz matrix. Fig. 4a and b presents the 2D ZY-plane contour mappings of the Ge-E′ center (UV-on) and the Ge-E′H complex (UV-off), respectively. Although the current spatial resolution is inherently limited by the intensity and natural linewidth of the signals, these mappings successfully confirm that the photo-induced defect generation and the subsequent dark conversion occur globally along the illuminated volume of the sample cuboid. To demonstrate the feasibility of spatially controlled defect distribution, the UV illumination was introduced from the front, crossing a physical grating. The corresponding EPR image successfully resolves the spatially modulated distribution of the Ge-E′ centers, directly mirroring the periodic intensity of the incident UV light as shown in Fig. 4c. This proves that the spatial resolution of our EPR imaging setup is sufficient to resolve macroscopic patterns engraved into the silica matrix. Furthermore, volumetric reconstruction (Fig. 4d), its slice map (Fig. S4) and orthogonal projections (Fig. S5) provide a preliminary 3D visualization of the defect profile confined to the UV-illuminated volumes within the silica matrix. These 3D representations confirm that the UV-induced activation is not merely a surface effect but extends throughout the sample volume. Such volumetric mapping underscores the potential of EPRI as a non-destructive diagnostic tool for characterizing the homogeneity of internal defects in large-scale optical components without the need for physical sectioning.Fig. 42D spatial mapping in the ZY plane of (a) the central Ge-E′ center under UV irradiation, and (b) the Ge-E′H signal in the dark. (c) 2D ZY-plane imaging of the Ge-E′ center generated under spatially modulated UV illumination from the front using a grating. (d) Reconstructed 3D EPR image showing the volumetric distribution of the Ge-E′ centers. 2D spatial mapping in the ZY plane of (a) the central Ge-E′ center under UV irradiation, and (b) the Ge-E′H signal in the dark. (c) 2D ZY-plane imaging of the Ge-E′ center generated under spatially modulated UV illumination from the front using a grating. (d) Reconstructed 3D EPR image showing the volumetric distribution of the Ge-E′ centers. To fully understand the evolution of the Ge-E′ and Ge-E′H centers, their long-term stability and thermal reversibility were investigated (Fig. 5a). Following the initial UV irradiation, the sample was stored in the dark at room temperature for up to 3 months. The Ge-E′ center signal vanished completely within 10 days, underscoring its transient nature as it is progressively consumed by diffusion-limited hydrogen capture. Conversely, the continuous persistence of the 119 G doublet over months confirms that the Ge-E′H complex resides in a relatively stable intermediate state under ambient conditions. Over prolonged aging, the gradual weakening of this 119 G doublet signal indicates a slow thermodynamic relaxation, wherein the deeply-trapped metastable Ge-E′H complexes undergo structural reorganization via parallel relaxation pathways. On one hand, the localized hydrogen can be deeply trapped to form EPR-silent covalent Ge–H bonds, resulting in stable hydrogen-passivated configurations. On the other hand, the hydrogen can entirely detrap and escape from the defect sites, leaving uncompensated dangling bonds that structurally relax and recombine into diamagnetic network precursors, specifically oxygen deficiency centers (GeODCs). This structural healing process can be drastically accelerated by thermal annealing at 600 °C, which provides sufficient thermal energy to completely eliminate all paramagnetic signals within 1 h, thereby restoring the glass network to its initial ground state [20].Fig. 5(a) The EPR spectra of the long-term dark evolution and thermal bleaching of the UV-irradiated JGS2 quartz. (b) Schematic potential energy diagram illustrating the proposed mechanisms for the evolution of the Ge-related paramagnetic defects. (a) The EPR spectra of the long-term dark evolution and thermal bleaching of the UV-irradiated JGS2 quartz. (b) Schematic potential energy diagram illustrating the proposed mechanisms for the evolution of the Ge-related paramagnetic defects. Based on the combined kinetic and spectral data, a comprehensive potential energy diagram is proposed in Fig. 5b to describe the UV-induced defect dynamics of JGS2. The initial UV irradiation excites the system from the diamagnetic ground state (Ge–H and GeODC) to a high-energy excited state characterized by the central Ge-E′ signal and highly mobile, transient hydrogen species (including transient H0 fragments). Upon removing the UV source, the system relaxes. However, rather than returning directly to the ground state, these liberated hydrogen species undergo long-range transport via either thermally activated hopping across the oxygen backbone or interstitial molecular H2 diffusion. Subsequently, they are gradually captured by the active precursor sites, causing the system to drop into a local-energy-minimum metastable state (Ge-E′H). This energy landscape elegantly explains the profound memory effect and the 20-fold kinetic acceleration of Ge-E′ generation during subsequent UV illumination. Activating the system from the metastable state (H detrapping) requires significantly less energy than cleaving intrinsic bonds from the ground state. Finally, the prolonged aging or high-temperature annealing allows the system to overcome the activation barrier, driving the defect bleaching and covalent network reconstruction to eventually restore the diamagnetic ground state. It is worth noting that the generalizability of the reported phenomenon is strictly confined by both the excitation energy threshold and the precursor states of the material. First, photon energies below the UV range fall short of overcoming the electronic activation barriers required for defect photolysis or charge detrapping. Second, unlike high-energy ionizing radiation that can damage the intrinsic lattice indiscriminately, UV-induced color center generation fundamentally relies on the presence of pre-existing defect precursors, while the dynamic memory effect depends on an abundant structural hydrogen reservoir to sustain the hydrogen shuttle carriers. This dual requirement explains why the observed phenomenon is uniquely manifested in JGS2. Although JGS1 possesses a high hydrogen content, its ultra-high chemical purity results in a deficiency of the necessary foreign impurity defects (e.g., Ge substitutes) to serve as stabilization traps. Conversely, JGS3 is inherently precluded due to its severe deficiency of structural hydrogen. Consequently, neither JGS1 nor JGS3 can sustain the reported kinetic and thermodynamic defect evolution. Conclusions This work presents a unified thermodynamic model of defect dynamics in UV-grade fused quartz, established via in situ EPR spectroscopy and 3D imaging. We identify the hydrogen shuttle as a key mechanism for the Ge-E′/Ge-E′H transition, which fundamentally triggers the observed kinetic memory effect through metastable precursor accumulation. Our proposed potential energy landscape unifies these spatiotemporal observations, offering strategic insights into the long-term stability and radiation resilience of silica. While the current framework primarily focuses on Ge-related defect complexes, further work is warranted to explore its generalizability to intrinsic silica defects under diverse irradiation regimes. These findings provide critical theoretical guidelines for the development of advanced optical and dosimetric systems operating in extreme environments. CRediT authorship contribution statement Xiaobing Lou: Writing – review & editing, Writing – original draft, Visualization, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Fushan Geng: Visualization, Investigation. Kejiao Shi: Formal analysis, Data curation. Bingwen Hu: Writing – review & editing, Supervision, Resources, Investigation, Funding acquisition, Conceptualization. Declaration of competing interest The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Bingwen Hu is an editorial board member of Magnetic Resonance Letters but was not involved in the editorial review or the decision to publish this article. Other authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.