Some variability of OSL sensitivity in magmatic and metamorphic quartz was observed and linked to magma-crystallisation processes and/or pressure-temperature conditions during metamorphism (Lukas et al., 2007; Sawakuchi et al., 2011a, b; Jeong and Choi, 2012; Zular et al., 2015; Guralnik et al., 2015). Higher TL-OSL sensitivity observed in quartz from bedrock formed at high temperatures probably results from crystallization temperature on the production of intrinsic or extrinsic defects in quartz (Sawakuchi et al., 2011b). Higher crystallization temperature favours both key processes at the heart of the luminescence phenomenon in quartz by increasing the occurrence of Si- and O-vacancies (Preusser et al., 2009) and the substitution of Si⁴⁺ cations by Al³⁺ and Ti⁴⁺ (Dennen et al., 1970; Götze et al., 2001; Wark and Watson, 2006). Laboratory experiments also report increasing luminescence sensitivity in quartz with heating (Bøtter-Jensen et al., 1995; David and Sunta, 1981; Koul, 2006; Polymeris et al., 2007; Poolton et al., 2000; Rink et al., 1993).

Reported values of TL-OSL sensitivity in quartz range from two to five orders of magnitude between crystalline rocks (magmatic and metamorphic) and sedimentary rocks (Alexanderson, 2022; Mineli et al., 2021; Sawakuchi et al., 2011a, 2020; Souza et al., 2023). Indeed, both sensitisation of quartz luminescence and growing contribution of a fast component signal are supposed to be linked with sediment transport distance and the number of burial-exposure cycles (Pietsch et al., 2008; Fitzsimmons, 2011; Rhodes and Pownall, 1994; Preusser et al., 2006; Lukas et al., 2007; Gray et al., 2019), as documented by laboratory experiments with repetitive irradiation-bleaching treatments (McKeever et al., 1996; Li, 2002; Moska and Murray, 2006; Koul and Chougaonkar, 2007). Thus, low sensitivity of quartz grains is commonly attributed to minimal transport distance and/or young sedimentary history. However, the downstream increase in quartz TL-OSL sensitivity is not always documented in natural environments (Sawakuchi et al., 2018; Capaldi et al., 2022; Magyar et al., 2024; Parida et al., 2025) and the role of transport in quartz OSL sensitization remains source of discussion. By contrast with bright TL-OSL signals measured in highly reworked sediments, primary source bedrocks contain dim quartz grains characterised by low OSL sensitivities and reduced fast component, making them very difficult to use as only few grains provide a luminescence signal (Sawakuchi et al., 2011a; Jeong and Choi, 2012; Guralnik et al., 2015; Mineli et al., 2021).

It may be also indicative of different intrinsic concentration of holes at certain light-sensitive recombination centres and types of impurities and defects in the mineral structure, which control the main luminescence traps and centres (Preusser et al., 2006, 2009).

ESR spectra linked to Al and Ti paramagnetic centres traditionally measured in quartz aliquots for dating studies (Ikeya, 1993) document large variabilities related to the saturation value or the Ti / Al intensities ratios. Interestingly, geographical provenance of quartz, likely in relation with the regional bedrock geological nature, seems playing a key role here with variable configurations: high intensities for both centres (e.g. USA, Indonesia, Toyoda and Ikeya, 1994; Fig. A1) or high intensity for one centre only, including (very) high Al versus weak (or even absent) Ti intensity in the Pyrenees foreland (Delmas et al., 2018) and Chinese lœss (Toyoda, 2015), or the opposite (e.g. Philippines; Ingicco et al., 2024). Moreover, the very few existing studies on primary source bedrocks mainly focused on the non-radiosensitive E1' centre of quartz (i.e. O-vacancy electron centre). As it may act as a non-radiative centre competing with the recombination process (Poolton et al., 2000), it is not used for dating applications (e.g. Jani et al., 1983; Toyoda and Ikeya, 1991, 1994; Ono et al., 1998; Toyoda et al., 2016). However, Duttine et al. (2002) and Kotova et al. (2008) highlighted some differences between the relative intensities of Ge, Al and Ti paramagnetic centres in different bedrocks ([GeO₄ / Na⁺]°, [GeO₄ / Li⁺]°, [AlO₄ / h]° [TiO₄ / H⁺]°, [TiO₄ / Li⁺]°, [TiO₄ / Na⁺]°). Moreover, Shimada et al.'s (2013) study on the E1' centres in relation to signal intensity of Al and Ti-Li in quartz from various Japanese bedrocks and sedimentary deposits showed distinctions between volcanic and granitic sources, thereby suggesting that ESR signals in quartz may be used as a source tracer. In addition, intensity ratios of Al, Ti-Li or Ti-H centres in quartz grains from various river sediments can be related to provenance in the catchment (Tissoux et al., 2015; Shimada et al., 2016). Whilst the influence of grain-size on the ESR response of quartz was thoroughly studied (Timar-Gabor, 2018), comparable results can only be obtained using samples of similar grain-size fractions (Shimada et al., 2016).

The metamorphic basement covers ca. 20 % of the drainage area. It is composed of sillimanite paragneiss from the so-called Monotonous Gneiss Unit (von Eller, 1961). The latter formed during the Carboniferous from late Cambrian-early Ordovician sediments (Skrzypek et al., 2012) under a medium pressure and temperature conditions prograde metamorphism (see Table 1, z1d and z1e) followed by a near isothermal decompression associated with rapid exhumation (Latouche et al., 1992; Rey et al., 1992; Schulmann, 2002). Spatial variations occur within the Monotonous gneiss facies linked to the gneiss protolith and later evolution involving fracturing, magmatic and hydrothermal activity. In particular, gneissic units to the South of the Bilstein fault zone are affected by an increasing degree of partial melting materialised by the transition to migmatitic domains to the south of the watershed (z1e, Fig. 1b; Fluck et al., 1991; von Eller, 1961).

Four different Carboniferous leucogranites cover ca. 50 % of the drainage area and intrude the metamorphic basement along major faults, such as the Bilstein fault zone (BFZ) in the Strengbach catchment (Fig. 1b; Fluck et al., 1991; Skrzypek et al., 2014). They match one of the latest magmatic events during the Vosges Massif's formation which occurred during a phase of N-S extensional subsidence of the orogen ca. 10–15 Ma after the metamorphic basement's rapid exhumation (Skrzypek et al., 2014; Tabaud et al., 2014). The so-called Thannenkirch granite (γ1b) first intruded the basement (Kratinova et al., 2007). It is a porphyroid, biotite-rich anatectic monzogranite that formed from the melting of the surrounding gneiss (von Eller, 1961). The so-called Brézouard (γ1c), Bilstein (γ1d) and Turckheim (γ1e) granites subsequently formed from melting of deeper crustal material (Fluck et al., 1991; von Eller, 1961). The Brézouard granite was second to intrude the basement, followed by the syntectonic Bilstein granite (Kratinova et al., 2007). The Bilstein granite formed while the BFZ was active and deformed following a ca. N70 senestral motion (von Eller, 1961). Such deformation can also be found in the southern part of the Brézouard pluton and within the Thannenkirch pluton (Fluck et al., 1991; Kratinova et al., 2007). Within the Thannenkirch pluton, the so-called Verreries facies hosts granito-gneissic features as the anatectic fusion remained in an early stage (von Eller, 1961). The Turckheim granite only marginally intrudes the Monotonous Gneiss Unit to the south of the Strengbach catchment. While its relative formation timing remains unknown, it locally evidences granito-gneissic characteristics (Blanat et al., 1972).

Triassic Buntsandstein units (mostly sandstones) are deposited on top of crystalline rock formations, following the intrusion of various leucogranites during the Carboniferous and a long-lasting erosional episode materialised by a ca. 50 Ma hiatus in the geological record. Three preserved units cover ca. 25 % of the Strengbach's drainage area (Fig. 1b). The lower one is the Vosgian basal unit (t1a) made of clayey ferruginous sandstone (lower Buntsandstein). The intermediate and main one (t1–2V and 2P) matches undifferentiated Vosgian sandstone and conglomerate (lower to upper Buntsandstein). The upper one (t2–3 and t3c) is made of Voltzia and shell-rich sandstone (upper Buntsandstein to lower Muchelkalk). Whereas fluvial depositional environments are traditionally attributed to the two lower units, the upper one is related to a deltaic environment (Gall, 2006). The subsidence of the Upper Rhine Graben (mostly Cenozoic) led to the formation of major N-S oriented faults that crosscut the westernmost part of the Strengbach catchment (von Eller, 1961).

The Strengbach catchment's small size (ca. 40 km²) and easy access to source material both enabled systematic sampling of all quartz-bearing rock formations. Altogether, fifteen bedrock samples were collected throughout the catchment (Fig. 1b, Table 1, Fig. B1). Three of them were collected from the Monotonous gneiss unit: one in each facies that are laminated (Z1d, #14) and migmatized (z1e, #02) paragneisses (i.e. “Gneiss”) along with one metamorphic quartz vein sample (#01; i.e. “Vein”). Nine granitic samples form the bulk of our dataset as this lithology roughly constitutes the half of the drainage area: Three samples were collected in the Bilstein granite (γ1d, #9A, 9B and 10) close to a shear zone (i.e. “Deformed granite”). Six granites were collected from intrusions (γ1b and γ1c #15, 17, 20, 45; i.e. “Granite”) and (γ1e, #32, 33; i.e. “Altered granite”). The remaining three samples (#29, 30, 44) were collected from the Triassic Buntsandstein formations, i.e. one sample per sedimentary unit (t1a, t1–2V and 2P and t2–3 and t3c; i.e. “Sandstone”). Table 1 summarises the sampled rock formations with labels, ages and a short description of lithology and conditions of formation.

Bedrock samples were prepared and processed at the OSL laboratory of the Centre de Recherche et d'Enseignement des Géosciences de l'Environnement (CEREGE, Aix en Provence, France). For each bedrock sample, between ∼ 1 and 4 kg of rocks were gently crushed and sieved to isolate the 180–250 µm grain-size fraction, followed by a first chemical bath with 37 % HCl to remove potential surface carbonates, followed by magnetic separation with a Frantz magnetic separator (two steps at 10°–0.5A and 10°–1.5A) to remove magnetic and/or heavy minerals. Following this separation, ∼ 50 to 70 g of the remaining material undergone chemical treatment for a week with a slight stirring in a solution of 1/3 of HCl and 2/3 of H₂SiF₆ to remove a large part of the feldspar grains. Density separation using lithium heteropolytungstate (LST) heavy liquid at 2.62 g cm⁻³ density was performed to further isolate quartz from feldspar. If feldspar grains were still detected under the binocular magnifier, we performed a second magnetic separation with addition of magnetite powder, followed by a rinsing step in a solution of 10 % of HCl. No HF treatment was performed for preserving as much as possible the quartz-grain structure and surface for future analyses such as trace-element distribution with Laser Induced Breakdown Spectroscopy (LIBS). This step was replaced by a second treatment for a week in a solution of 1/3 of HCl and 2/3 of H₂SiF₆ to remove any remaining feldspar inclusions. Finally, the prepared samples were separated into two batches to be measured by ESR and OSL. In parallel, 100 µm thick sections of each rock were created for analysis by Laser ablation inductively coupled mass spectrometry (LA-ICPMS).

Aliquots of ∼ 100–300 quartz grains were mounted in stainless discs using silicon oil, and a first evaluation of feldspar contamination was performed using infrared (IR) stimulation at 60 °C before further analysis. Luminescence signals were measured on stainless plates using a Lexsygsmart reader equipped with a beta radiation source delivering a dose rate of ∼ 0.124 Gy s⁻¹, with blue (458 nm, maximum power of 100 mW cm⁻²) and infrared (850 nm, maximum power of 200 mW cm⁻²) LEDs for stimulations and filters (Hoya-U340 + Delta-BP 365/50) for light detection in the ultraviolet band (380 nm). For all TL steps, the heating rate was 5 °C s⁻¹.

Luminescence measurements for OSL at 125 °C and TL sensitivities were carried out following the protocols modified from Sawakuchi et al. (2020) or Mineli et al. (2021), which are presented in Table 2. For each bedrock sample, three aliquots were measured and subsequently weighted for mass normalisation.

The aliquots were first bleached for 300 s at 90 mW cm⁻²and 125 °C before any regeneration dose. Then, a regeneration dose of 50 Gy was given to all aliquots to allow, along with mass normalisation, uniform inter-comparisons of the measured luminescence signals among samples. In a first experiment, the TL peak of quartz at 110 °C (Murray and Roberts, 1998) is investigated to evaluate its regenerated sensitivity and possible correlation with the OSL sensitivity. The TL peak sensitivity is calculated by integrating the TL signal between 75 and 125 °C (TL_110 °C), normalized by the regeneration dose and the aliquot mass (photon cts Gy⁻¹ mg⁻¹). After IR stimulation at 60 °C to confirm the quartz purity of the aliquot, the OSL sensitivity is measured using blue light stimulation at 125 °C (BOSL125°C) and corresponds to the counts integrated over the first second of light emission, minus the normalised background from the last 10 s of the OSL decay curve (Fig. C1). In a second experiment (treatment B in Table 2), we evaluate the quartz OSL sensitivity after a second similar irradiation dose but measured at room temperature (BOSLf_25 °C), which corresponds to the integral of the first second of light emission, minus the late background. Both signals are also normalized by the regeneration dose and the aliquot mass (photon cts Gy⁻¹ mg⁻¹).

Measurements by ESR spectrometry imply the separation of all samples into two aliquots weighing ∼ 100 mg each. One aliquot of each sample was exposed to light (SOL2 solar simulator – HONLE) for 2000 h (Toyoda et al., 2000) to achieve maximum bleaching of Al and Ti ESR centres. Both bleached and natural (non-bleached) aliquots were then measured in an EMX Brucker ESR spectrometer located in the laboratory of Histoire Naturelle de l'Homme Préhistorique (Paris, France) using following parameters: 5 mW microwave power; temperature of 90 °K; 1024 point resolution; 20 mT sweep width; 100 kHz modulation frequency; 0.1 mT modulation amplitude; 40 ms conversion time – 20 ms time constant. The Al, Ti-Li and Ti-H centres were measured simultaneously following Toyoda and Falguères's (2003) recommendations for the Al centre and those of Toyoda et al. (2000) for Ti centres. ESR intensities were measured for all centres and provided following information: (i) the total intensity of the Al centre from the natural aliquot (Al), (ii) the optically non-bleachable intensity of the Al centre (DAT) corresponding to the intensity of the bleached aliquot; (iii) the OBAT: Optically bleachable intensity of the Al centre (OBAT) obtained by (Al) – (DAT), (iv) the bleaching percentage of the Al centre (Bl %) obtained by ((Al - DAT) / Al) × 100, the Ti-Li intensity (Ti-Li) and the Ti-H intensity (Ti-H). We reasonably assume that both Al and Ti centres are in saturation before sample processing and measurements. Indeed, according to single exponential dose response curves, the saturation dose of the Al centre is reported in literature from 3000 to 6000 Gy, and that of the Ti centre was between 2700 and 4700 Gy for quartz (Rink et al., 2007; Duval, 2012). Fitted with a single saturating plus linear function (Duval 2012) the saturation dose of the Al centre range between 1900 and 6500 Gy). In our samples, annual doses were calculated from laboratory gamma-ray spectrometry measurements on rocks sampled in the Strenghbach basin (Ortec HPGe spectrometer, MNHN, France). Average annual doses on granites are of the order of 4000 µGy yr⁻¹ (4 Gy/1000 years). Because source bedrocks are older than 250 Ma, the dose received is estimated at around min. 1 000 000 Gy. For gneiss, annual doses are a about 3000–3500 µGy yr⁻¹, and doses received are therefore higher than 1 000 000 Gy. For sandstones, the calculated annual dose is around 1000 µGy yr⁻¹. Considering the Triassic age of the deposits, the total dose received is higher than of 300 000 Gy. This implies that all of our samples, regardless of their own saturation value, received a radiation dose greater than this value.

LA-ICPMS analyses were performed in the laboratories of the French Geological Survey (BRGM, Orléans, France) on an Agilent 8900 Triple Quadrupole ICP-MS coupled to a 193 nm Teledyne CETAC Technologies Photon-Machine Excite Excimer laser equipped with a HelEx 2 laser-ablation cell. The laser was run at a shot frequency of 8 Hz, energy was set to 6 mJ, and a spot size of 50 µm was used. The acquisition of each spot lasted 60 s with 20 s of background before the start of the 30-second-long ablation. To prevent quartz from breaking under the laser beam, analyses were carried out on thick sections made from whole rock samples from bedrocks. A total of fifteen samples were analysed for trace elements in quartz (including Al, Li, and Ti) with between 28 and 60 spots on the various quartz grains present on the thick section. On each section, the quartz grains present were located and targeted beforehand by microscopy. All quartz grain types of sufficient size and comparable to those extracted for ESR and OSL were analysed. No matrix or other mineral were targeted by the laser. Element concentrations were obtained from the raw data using the software Glitter. Nist 612 was used as external standard (Pearce et al., 1997) and a concentration of 99.9 % SiO₂ was used as internal standard for quantification. Data are based on a background-substracted-signal. Signal smaller than three times the background noise standard deviation was excluded, and detection limits were calculated based on these criteria for each analysis. Commonly, the highest calculated detection limit of each analysed element is selected and used to further exclude any data lower than these values. This is done so that the dataset is consistent and comparable. However, in our case, this implies the exclusion of an important number of otherwise reliable data (i.e. all Li concentrations of sample STR22-33). We decided instead to keep values for all the analyses that were above their associated limit of detection and to provide the maximum Limit Of Detection (LOD) for each sample as an indicator. Values below or close to the maximum LOD of each sample should be considered as indicative values rather than exact values. The median concentration for each considered element was calculated for each sample to allow comparisons with ESR and OSL data (i.e. aliquots with several hundreds of grains). The estimation error is defined as the half of the interquartile range. Tables with detection limits and median values per sample are provided in the supplementary information (Table D1) .

i.

sensitivity values related to BOSL_125 °C measurements range from ∼ 10 to ∼ 250 cts Gy⁻¹ mg⁻¹. This variability together with an OSL sensitivity gradient allow distinguishing two categories of bedrocks including eight and seven samples, respectively. On the one hand, low sensitivities ranging from 10 to 60 cts Gy⁻¹ mg⁻¹ are found in the quartz vein (z1d; #01), both gneisses (z1d–e; #02, 14) and the syntectonic Bilstein granite (γ1d; #9A–B, 10) along with the gneissic Turckheim granite (γ1f; #32, 33) (Table 3, Fig. C1). On the other hand, both remaining granites (γ1b; #15, 17 and γ1c; #20, 45) along with all Triassic sandstones (t1a, t1–2 and t2–3; #29,30, 44) present much brighter signals comprised between 100 and 250 cts Gy⁻¹ mg⁻¹. The highest sensitivity is observed in the lower Triassic sandstone (#29), which also shows a higher variability in measured aliquots.

Optically unbleachable (DAT) and Optically bleachable (OBAT) intensities of the Al centres along with Ti-Li and Ti-H intensities of each sample were measured on the same aliquot. The OBAT vs. DAT (Fig. 2c) are extremely variable between bedrock. First, main bedrock types (i.e. plutonic, metamorphic and sedimentary) clearly cluster into sub-groups, with the quartz vein (#01) clearly plotting apart (Fig. 2c). The latter also shows the lowest DAT intensities whereas the highest values are reported in the Bilstein deformed granites(#09A, 09B, 10). OBAT intensities display similar results: lowest value for the quartz vein vs. highest values for all types of granites. Second, the comparison between Ti-Li and Ti-H intensities shows a linear correlation (R2=0.721; Fig. 2d). It allows distinguishing two groups of bedrocks, similar to those based on BOSL_125 °C measurements: low Ti-Li and Ti-H intensities for the quartz vein, gneisses and the Bilstein granite vs. high Ti-Li and Ti-H intensities for all remaining granites and sandstones. Given this excellent correlation between Ti-Li and Ti-H intensities, only the former will be discussed in the following since our conclusions made on Ti-Li are considered as effective also for Ti-H. Third, intensity comparison of Al vs. Ti-Li shows no linear correlation. Whereas most of the Al values cluster around ∼ 600 a.u., Ti-Li intensities scatter between 0 and 60 a.u. (Fig. 2e).

The comparison between BOSL_125 °C sensitivities and Ti-Li intensities highlights two distinct groups: low sensitivities/intensities for the quartz vein, gneisses and Turckheim and Bilstein granites vs. high sensitivities/intensities for the remaining granites and sandstones (Fig. 2f) However no significant relationship can be calculated (R2 < 0.5). Sample #32 plots apart. By contrast, comparison of BOSL_125 °C sensitivities with Al intensities shows no linear correlation as the latter cluster around 600 a.u. and 900 a.u. (Fig. 2g). We noticed that depending on the type of measurements, sample #32 shifts within bedrock groups: its BOSL_125 °C sensitivity and ESR-Al intensity are similar to gneisses, while its ESR Ti-H and Ti-Li intensities place it in the granites/sandstones group. This sample was taken in granitic intrusions (γ1e, Table 1) into gneisses (z1e, Table 1), right at the boundary between granite and gneiss. It has acquired characteristics specific to both formations, as all the quartz grains were not recrystallized.

Al, Ti and Li concentrations are in the order of a hundred ppm for Al and around ten ppm for Ti and Li (Fig. 3).

Al contents do not discriminate between bedrock types, but median values seem slightly increasing from the quartz vein, gneisses and altered and deformed granites to those of granites. Li contents are scattered with different concentrations related to bedrock types: very low (< ∼ 5 ppm) in quartz veins/gneisses/altered granites (Turckheim), fairly homogeneous (∼ 15 ppm) in deformed (Bilstein) granites, and high (∼ 35 ppm) but with large scatter in granites (except #17) and sandstones. Ti content does not seem to discriminate between bedrock types. Gneisses and Altered granites showing a wide dispersion between samples from the same rock type whereas Ti content values for granites and sandstones are less scattered. The ratio between the median Al and Li contents shows a strong positive correlation (R2=0.891; Fig. 4a). In addition, two groups of samples can be distinguished here: one with low Li and Al contents and another with higher contents (Fig. 4a). These two groups of bedrocks were those previously reported with the trapped-charge methods (ESR/OSL) (Fig. 2a, d and g).

First, comparison between Al contents and BOSLf_125 °C sensitivities show a R2 of 0.514. Despite precautions to be taken on this R2 due to clustering of data points, it nevertheless indicates a rather good correlation (Fig. 4b). Again, it highlights the two usual groups: i.e. low sensitivity and low Al content for the quartz vein, gneisses, Turckheim and Bilstein altered granites vs. high sensitivity and high Al contents for the remaining granites and sandstones, except for granite sample #17 showing rather low Al content. Second – and interestingly – no linear correlation is observed between Al contents and ESR Al intensities (Fig. 4c). Third – by contrast – Ti contents correlate fairly well with ESR Ti-Li intensities (Fig. 4d, R2=0.398), with the same two groups as above except the sample #32 (gneiss) which includes the granite/sandstone group (Fig. 4d). Fourth, comparison between Li contents with ESR Ti-Li intensities do not allow to derive any correlation but highlight the same dichotomy, with samples #17 and 32 plotting as outliers (Fig. 4e) similar to Al contents vs. IRSL sensitivity (Fig. 4f).

Quartz aliquots with low OSL sensitivity signals are documented in plutonic or metamorphic bedrocks and the reported values generally fall within the following range between ∼ 10 to 500 counts Gy⁻¹ or slightly higher values for siliciclastic rocks (Sawakuchi et al., 2011a; Chithambo et al., 2007; Capaldi et al., 2022; Niyonzima et al., 2020; Mineli et al., 2021; Alexanderson, 2022, Constantin et al., 2025).

Our dataset also exhibits low quartz OSL sensitivities (i.e. ∼ 10 to ∼ 250 cts Gy⁻¹ mg⁻¹ or 15 to 615 cts Gy⁻¹, Table 3) which fall into the previous range. We also notice a strong correlation (R2 = 0.647, R2 = 0.804 without #32 outlier, Fig. 2b) between the TL_110 °C peak and the first second of the OSL signal (BOSLf_125 °C) sensitivities, a laboratory observation widely acknowledged in the literature (Murray and Wintle, 2000; Jain et al., 2003; Mineli et al., 2021). The sensitivity present characteristics which seem correlating with the lithologies of source bedrocks, also reported with ESR intensities derived from Ti-Li, Ti-H, OBAT and DAT centres (Fig. 2c, d and e). Importantly, our results indicate that ESR and OSL sensitivities may depend on intrinsic properties of the source bedrock, as already pointed out by other studies (Duttine et al, 2003, Lutoev, 2005; Kotova et al., 2007, 2008; Shimada et al., 2013; Sawakuchi et al., 2011a; Capaldi et al. 2022; Niyonzima et al., 2020). In this respect, quartz extracted from a quartz vein and metamorphic bedrocks record the lowest ESR and OSL sensitivities values, as well as plutonic samples collected within a shear zone (i.e. samples #9A and 9B in the Bilstein fault zone). Intermediate ESR and OSL sensitivities values are observed in quartz collescted from plutonic rocks and sandstones. So once again, our results thus clearly indicate that (i) inter-sample variability is characteristic of the geological origin, and that (ii) two main groups can be obviously distinguished in our present dataset based on the analysis of different measured signals.

The Ti-Li ESR and BOSLf_125 °C sensitivities appear to correlate (Fig 2F) but no correlation was observed between ESR-Al and BOSLf_125 °C sensitivities (Fig. 2g), which would indicate that different centres may be involved in ESR and OSL signals.

The potential relationship between quartz OSL/ESR sensitivities and geological sources in relation to different natures and concentrations of impurities (trace elements) and/or vacancies within the crystal lattice responsible for OSL emission centres and/or traps was preliminary discussed in Preusser et al. (2006). We thus aim to further explore this relationship by specifically looking at the type and the concentration of impurities using trace element contents in our samples. Trace elements are preferentially incorporated into the quartz lattice on substitutional (i.e. Al, Ti, Ge) or interstitial (i.e. Li, Na, K) positions. First, we report a significant linear correlation (R2=0.789) between quartz's Al and Li contents so that two groups of quartz-bearing bedrocks can be distinguished in the Strengbach catchment (Fig. 4a). Interestingly, this relation may be explained if Li is mostly structurally incorporated for compensating Al in the quartz structure. Our data also agree with previous results showing that Li may be exclusively incorporated into the structural channels parallel to the c-axis of the quartz lattice as charge balancing cation for Al in granite and pegmatite (Stavrov, 1978; Götze et al., 2004).

Because we assume that Al and Ti ESR centres from all treated samples have reached their own saturation level as crystallisation/diagenesis of all quartz-bearing bedrocks took place > 200 millions of years ago, we can compare measured ESR intensities with OSL sensitivities. One generally acknowledges that the large variability reported in bedrock OSL sensitivities is mainly linked to temperature and pressure conditions during metamorphism or magmatic crystallization (e.g. Sawakuchi et al., 2011b): quartz with low OSL sensitivity signals were detected in deformed plutonic or metamorphic bedrocks that have experienced high temperature and pressure conditions (Lukas et al., 2007; Jeong and Choi, 2012; Zular et al., 2015; Guralnik et al., 2015). Variability in ESR sensitivities for different bedrock types was likewise pointed out in different settings: Duttine et al. (2003) (Al, Ti-Li, Ti-H intensities); Kotova et al. (2008) (Al intensities); Shimada et al. (2013) (Al and Ti-Li intensities).

In this respect, two outcomes of this study well match previous observations. First, our observations in general agree with Kotova et al. (2008)'s suggestion: lower number ESR-Al and ESR-Ge defects (i.e. Ge and Ti defects being similarly ion compensated defects) are caused either by quartz vein crystallization from highly oversaturated silicium solutions or from solutions with low contents of Al or Ge, possibly reflecting the initial fluid composition. Second, similarly to Duttine et al. (2003), our quartz vein sample (#1) displays both (i) low ESR and OSL sensitivities and (ii) low Al and Li trace contents (Figs. 3, 4a, b, c).

Pietsch et al. (2008) showed that mature and recycled quartz presents higher OSL sensitivity values than parent bedrocks, which suggests that surficial transport processes produce sensitisation in quartz grains. However, this is still largely debated with contradictory observations on the sensitization of luminescence signals with transport distances (Sawakuchi et al., 2018; Capaldi et al., 2022; Magyar et al., 2024; Parida et al., 2025). In this respect, we can assume that the sandy sediments that later formed the Triassic sandstones in the Strengbach catchment are sensitive due do previous sedimentary cycles. Indeed, the sediments that make up the sandstones were deposited in a fluvial context (Bofill et al., 2024). The source of these sandstones is unknown, but the shapes of the grains and sorting of grains sizes indicate a long transport distance and therefore quartz grains underwent one or more cycles of exposure to light and natural irradiation when they were buried. Whilst the highest OSL sensitivity found in the sandstone sample #29 seems matching this assumption (Fig. 2a), the two other sandstone samples (#30 and 44) present similar OSL sensitivities than those of the granites. Reason may be either the quartz grains that make up the sandstones (#30 and #44) initially had very low sensitivity (comparable, for example, to that of gneisses), or these quartz grains underwent few regeneration cycles during transportation before diagenesis (unlikely), or the sandstones underwent (post-)diagenetic phenomena that caused a loss of sensitivity in the quartz. We have no evidence to support this last hypothesis. The sensitisation of ESR centres with optical bleaching-irradiation cycles is currently unknown. Our data show that none of the sandstone samples have higher ESR-Al sensitivities than granite and gneiss samples (Fig. 2g). The Ti-Li and Ti-H centres show ESR sensitivities for sandstones that are in the higher value of our dataset, but that remain within the range of undeformed granites. In fact, the latter are mostly indistinguishable from sandstones except for their bleaching rate (Fig. 2c).