Not too old to rock: ESR and OSL dating reveal Quaternary activity of the Periadriatic Fault in the Alps

Abstract

The Periadriatic Fault system (PAF) ranks among the largest post-collisional structures of the European Alps. Recent geodetic data suggest that a fraction of the Adria–Europe convergence is still being accommodated in the Eastern Alps. However, the historical seismicity records along the easternmost segment of the PAF are ambiguous, and instrumental records indicate that seismotectonic deformation is mostly concentrated in the adjacent Southern Alps and Dinarides. Both electron spin resonance (ESR) and optically stimulated luminescence (OSL) dating methods can be used to date coseismic slip (with a combined range covering a few decades to a couple of million years) in slowly deforming fault zones, such as the PAF. Since the saturation doses of the quartz ESR signals are larger than quartz and feldspar OSL, ESR enables establishing a maximum age of the last resetting event of the system, while OSL allows constraining their minimum age when the signal is in saturation. We collected fault gouge samples from three localities along the easternmost segment of the PAF. For ESR, we measured the signals from the Al center in quartz comparing the results from the single aliquot additive dose (SAAD) and single aliquot regenerative (SAR) dose protocols. For OSL, we recorded the infrared stimulated luminescence signal at 50°C (IR₅₀) and post-infrared infrared-stimulated luminescence signal at 225 °C (pIRIR₂₂₅). Our dating results indicate that the studied segment of the PAF system accommodated seismotectonic deformation during the Quaternary, with a maximum age for the last resetting event of the system ranging from 1075 ± 48 to 552 ± 26 ka (ESR SAR) and minimum ages in the range from 196 ± 12 to 281 ± 16 ka (saturation of pIRIR₂₂₅). The obtained ages suggest that the studied segment of the PAF could be considered at least as a potentially active fault.

Graphical abstract

Introduction

This article deals with the Quaternary activity of one of the longest faults in the European Alps that has not had strong instrumental earthquakes (i.e., recorded via instruments). A recurring problem in characterizing past and present regional deformation patterns in many regions is constraining the timing of fault activity. Notwithstanding purely academic aspects, the identification and characterization of active faults (e.g., in terms of slip rates and recurrence rates of damaging earthquakes) are of paramount importance from a societal perspective to assess the regional seismic hazard. Instrumental earthquake records do not necessarily help to identify active faults, and geodetic monitoring may fail to recognize active fault segments in case of locked faults and low deformation rates. The problem may be exacerbated by long or unknown recurrence rates of large earthquakes in regions with low overall deformation rates such as the Alps and the Dinarides (Cheloni et al. 2014; Hintersberger et al. 2018; Grützner et al. 2021; Oswald et al. 2022), or intra-continental regions (e.g., 1811–1812 New Madrid earthquakes: Tuttle et al. 2002; 2001 Gujarat earthquake: Rajendran et al. 2008; 2016 Petermann Ranges Earthquake: King et al. 2021).

To date, there is no single definition for an “active fault”. According to the review of Slemmons & McKinney (1977), the most accepted definition is “a structure that shows evidence of Holocene displacement”, which conflicts in terms of temporal resolution with faults that have recurrence intervals of tens of thousands of years (Machette 2000; Grützner et al. 2016, 2017). Four main criteria presented in Slemmons & McKinney (1977) and originally proposed by Willis & Wood (1924) could be used to assess the activity of a structure. These are either (1) offset during the present tectonic regime (i.e., configuration of the tectonic plate movements and interactions), (2) the potential for future offset, (3) evidence of recent activity, and (4) associated earthquake activity. Other definitions have been provided by the U. S. N. R. Commission (1997), which classifies a tectonic structure as of interest for seismic hazard when there is geological evidence for recurring deformation within approximately the last 0.5 Ma, at least once in the past 50 ka, or when it has been involved in deformation associated with the current tectonic regime. The I. A. E. Agency (2010) proposes guidelines to distinguish active structures depending on the tectonic setting, considering an Upper Pleistocene–Holocene time frame in interplate regions and a Pliocene–Quaternary time frame in intraplate regions.

Since instrumental data acquired via seismometers, satellite images, and Global Navigation Satellite Systems (GNSS) is only available for the last few decades, dating of recent brittle deformation can help determine whether a fault is (potentially) active. The most common techniques for finding the age of brittle deformation episodes in fault zones are either dating deformation associated with earthquakes in displaced or deformed soft sediments with luminescence or radiocarbon methods (e.g., McCalpin 2009; Kondo et al. 2022), dating aseismically formed precipitations such as calcite veins and fibers using ²³⁰U/Th (e.g., Nuriel et al. 2012), or K/Ar dating of authigenic illite in fault gouge (e.g., Kralik et al. 1987; Pleuger et al. 2012; Torgersen et al. 2014; Ring et al. 2017; Vrolijk et al. 2018). However, these techniques either suffer from temporal limitations or rely on relative means to date the deformation episodes.

In geoscience, both electron spin resonance (ESR) in quartz and optically stimulated luminescence (OSL) in quartz and feldspar have been applied for dating sediments (e.g., OSL: Wintle and Huntley 1982; Preusser et al. 2008; Mahan et al. 2022; ESR: Schellmann et al. 2008; Tsukamoto et al. 2018; Richter et al. 2022), and recently applied also as ultra-low temperature thermochronometers (e.g., Guralnik et al. 2015; Fang and Grün, 2020; King et al. 2020). These methods utilize unpaired electrons trapped within crystal lattice defects in minerals and measure a signal proportional to the accumulation over time by exposure to natural irradiation (Rink 1997; Aitken 1998). The trapped charges can be released either during periods of increased temperature due to shear heating, elevated stress (normal or shear), and frictional grain comminution (Fukuchi 1992; Yang et al. 2019). Therefore, they potentially allow direct dating of seismotectonic deformation in fault zones at near-surface conditions (e.g., Ikeya et al. 1982; Fukuchi 1992; 2001).

In this work, we present the first attempt to apply direct dating methods to constrain active deformation along the Periadriatic Fault system (PAF) in the European Alps, by combining quartz ESR and feldspar OSL dating retrieved from fault gouges sampled in the core zone of the structure. The PAF represents an orogen-scale post-collisional fault system that was active mostly in the Oligocene to Miocene, but whose Quaternary activity is not well understood. Large instrumental earthquakes have not occurred on this fault system, and the historical earthquake records are ambiguous (Hammerl et al. 1994; Guidoboni et al. 2019; Caracciolo et al. 2021; Kázmér et al. 2023). Given the potential of the ESR and OSL dating methods to directly constrain the timing for seismogenic faulting episodes in the Quaternary, especially for slow-moving faults that are potentially still active, the PAF provides an opportunity to study the last seismogenic activity. For ESR, we compare the measurements and ages obtained with the single aliquot regenerative (SAR) dose protocol and the single aliquot additive dose (SAAD) protocol. For OSL, we employ the post-infrared infrared-stimulated luminescence (pIRIR) protocol, with infrared (IR) stimulations at 50 °C and 225 °C. Our findings indicate that at least a partial reset of the system was achieved during the Quaternary due to seismogenic faulting along the sampled portions of the structure.

The Periadriatic Fault system

The PAF system is an orogen-scale array of post-collisional, predominantly dextral, and mostly transpressive strike-slip faults in the European Alps. It formed in response to the NW motion and counterclockwise rotation of the Adriatic plate with respect to Europe during the Alpine orogeny (Schmid et al. 1989; Fig. 1). The fault system can be traced for about 700 km along-strike the entire width of the orogen, starting from N of Turin, Italy, to the Pannonian Basin in Hungary where it becomes buried below Plio–Pleistocene basin sediments. The eastern PAF system (east of the Giudicarie Fault) also represents the division between the high-grade metamorphic Austroalpine units to the north and the low-grade South Alpine units to the south (Fig. 2). A suite of calc–alkaline intrusive bodies, usually referred to as Periadriatic plutons, occurs along much of the length of the PAF. These plutons are considered to have resulted from a magmatic episode triggered by slab breakoff of the southward subducting European plate during the Eocene–Oligocene (von Blanckenburg and Davies 1995; Rosenberg 2004).

Fig. 1

a Tectonic map of the Alps. Modified after Schmid et al., (2004). b Simplified N–S-trending cross section across the Alps at 13° E showing the structural position of the different tectonic units (black lines delimit the boundaries between the units) and main structures with their respective kinematics (red lines). Fault zones: AFT Alpine Foreland Thrust, PAF Periadriatic Fault, SFS Sangone Fault System, SEMP Salzach–Ennstal–Mariazell–Puchberg Fault, ZWD Zwischenbergen–Wöllatratten and Drautal Faults. Modified after Schmid et al. (2013) and Hetényi et al. (2018)

Fig. 2

a Simplified geological map of the Pustertal–Gailtal segment of the PAF. Modified after Heberer et al. (2017). Fault zones: Fe Fella Fault, Ho Hochstuhl Fault, La Lavanttal Fault, Mö Mölltal Fault, NK North Karawanken Thrust, PG Pustertal–Gailtal Fault, Sa Sava Fault, Šo Šoštanj Fault. b Detail of the Pustertal–Gailtal segment of the PAF in the Karawanken mountains

According to different estimates, the PAF could have accommodated between 150 and 300 km of displacement between about 35 and 15 Ma (Laubscher 1971; Schmid et al. 1989; Schmid & Kissling 2000; Handy et al. 2005). The structure contributed to lateral extrusion in the Eastern Alps towards the Pannonian Basin during the late Oligocene–Miocene (Ratschbacher et al. 1991a, b). The northward motion of the southern Alpine indenter (the rigid northern tip of the Adriatic microplate) between 22 and 17 Ma led to an offset on the PAF by the NNE–SSW-trending sinistral Giudicarie Fault, modifying the originally straight E–W trend of the structure defined by the Insubric and Pustertal–Gailtal segments (e.g., Viola et al. 2001; Pomella et al. 2012; Sieberer et al. 2023; Fig. 1a). The minimum age for the offset has been established by K/Ar dating targeting clay minerals in fault gouges by Pleuger et al. (2012), which indicates displacement along the segment at around 17 Ma. This offset was simultaneous with south-directed thrusting in the eastern Southern Alps (Schönborn 1999; Castellarin and Cantelli 2000; Verwater et al. 2021). The post-Oligocene displacement along the PAF segments east of the Giudicarie Fault possibly did not exceed about 20 km (Ustaszewski et al. 2008), and was concurrent with at least 40–50 km N–S convergence accommodated along south-directed thrusts in the easternmost Southern Alps in Friuli (Nussbaum 2000; Merlini et al. 2002; Moulin & Benedetti 2018).

The Neogene deformation along the Pustertal–Gailtal segment of the PAF system has a prominently dextral transpressive character, especially in the region of the Karawanken Mountains (Polinski and Eisbacher 1992; Nemes et al. 1997; Fodor et al. 1998; Vrabec and Fodor 2006; Fig. 2a). In this segment, shortening is accommodated by slip partitioning between dextral strike-slip faults and N- and S-directed thrusts. On the northern edge of the Karawanken Mountains, Mesozoic successions overthrust the Middle Miocene to Pliocene sedimentary infill of the Klagenfurt basin (Polinski & Eisbacher 1992; Nemes et al. 1997; Fig. 2b). Apatite (U–Th–Sm)/He ages obtained along the fault fall between around 12 and 5 Ma and indicate Late Miocene to Pliocene cooling below 110–50 °C, likely triggered by fault-related basement exhumation (Heberer et al. 2017). K/Ar dating of clay minerals in fault gouges hints at displacement along the structure in the westernmost part of the segment (Mauls, Italy) at around 16 Ma (Zwingmann & Mancktelow 2004).

Available GNSS data (D’Agostino et al. 2008; Métois et al. 2015; Serpelloni et al. 2016; 2022; Sánchez et al. 2018a, b) suggest that a part of the present NW–SE convergence between stable Adria and Europe is still being accommodated along contractional faults in the Southern Alps and across the eastern PAF (vectors in Fig. 3). Still, the seismic moment release along the eastern PAF system is considerably smaller than in the Southern Alps of Friuli, located south of the structure (Serpelloni et al. 2016; Reiter et al. 2018). It is also possible that this segment of the PAF system hosted a few strong historical earthquakes, such as the Mw∼7.0 event in 1348 and the Mw∼6.6 event in 1690 (AHEAD Group: Albini et al. 2013; Guidoboni et al. 2019), both with epicenters in Carinthia. However, recent studies place the epicenter to the south of the PAF towards Friuli (Hammerl 1994; Caracciolo et al. 2021). In a recent archaeoseismology study, Kázmér et al. (2023) inferred a damaging earthquake at the eastern tip of the PAF near Celje (Slovenia) in the fourth century AD based on deformed Roman buildings.

Fig. 3

Historically recorded earthquakes (Grünthal and Wahlström, 2012), instrumentally recorded earthquakes (Stucchi et al. 2012; Jozi Najafabadi et al. 2021; Bagagli et al. 2022), and GPS velocities (D’Agostino et al. 2008; Metois et al. 2015; Serpelloni et al. 2016; Sánchez et al. 2018a, b) in the Eastern and Southern Alps

ESR and OSL for dating active deformation

ESR and OSL dating methods rely on the accumulation of unpaired electrons within traps (i.e., paramagnetic lattice defects) in mineral grains that can be released by different stimuli. These electrons are produced over time by the natural decay of radiogenic nuclides (such as ²³⁸U and ²³²Th, their daughter nuclides, and ⁴⁰K) and the incidence of cosmic rays. Ikeya et al. (1982) and Miki and Ikeya (1982) studied the potential of ESR to constrain the timing of recent earthquakes in geologic timescales for the first time, using quartz grains from rocks in fault zone cores. Fault gouges are a suitable lithological target for the method because their mineral grains experience frictional heating and grain size comminution directly at the fault core during coseismic slip (Sibson 1977; 1986; Fukuchi 1989). Furthermore, recent experimental studies suggest that the shear heating produced during an earthquake may reset at least partially OSL (Kim et al. 2019; Yang et al. 2019; Oohashi et al. 2020), thermoluminescence (Hiraga et al. 2002; Yang et al. 2019), and ESR (Al center, Yang et al. 2019) signals in fault gouges.

Establishing whether the system was totally or partially reset during a past earthquake event is crucial for determining the significance of the obtained ages. The multiple center and grain-size plateau approaches were developed to aid this assessment. The multiple center approach (Fukuchi 1988) relies on the assumption that discrepancies in the ages obtained for different centers could be the result of uneven reset of the signals during more recent fault movements. Therefore, the agreement between ages obtained from different centers could be considered a criterion for a complete reset of the system during the last seismogenic faulting event along a fault. The grain-size plateau method was proposed by Schwarcz et al. (1987), and further developed and applied in subsequent studies (e.g., Lee & Schwarcz 1994b; Lee 1994). Its basic principle states that after experiencing frictional deformation, grains below a critical radius would become fully reset. Thus, the distribution of ages from different grain size fractions would resemble a plateau ranging from younger ages in smaller particles to older ages in larger ones. The experimental basis for this distribution was initially presented by Buhay et al. (1988) and later expanded by other authors (Lee and Schwarcz 1993, 1996; Lee 1995). An additional criterion for complete zeroing of the signal during an earthquake was presented by Lee and Schwarcz (1994b), adopting the basic principle behind the multiple center method. This consisted of measuring the distribution from different signals and observing if the youngest ages from at least two plateaus converge, implying that said ages could be interpreted as the time of the last activity of the fault.

ESR dating of quartz has been applied for dating seismotectonic deformation since the 1980s. The studied faults were usually active structures, and the obtained ages indicate at least a partial system reset by fault motion. The earliest applications of ESR to date seismogenic activity using minerals extracted from fault gouges were made in Japan. Ikeya et al. (1982) targeted the Atotsugawa Fault (coseismic slip in 1858; Takeuchi et al. 2003), where they obtained a Pleistocene age of seismogenic faulting (~ 65 ka) measuring the E’ center. Matsumoto et al. (2008) used core samples (389.4 m depth), from the Nojima Fault (1995 Kobe/Great Hanshin Earthquake 6.9 M_w) and observed a decrease in the signal intensity from the Al and E’ centers, which was attributed to shear heating partially resetting the signals during earthquakes. Other applications have targeted known active faults outside of Japan such as the San Andreas Fault Zone (San Jacinto segment: Buhay et al. 1988, San Gabriel branch: Lee 1994; Lee and Schwarcz 1996) and the fault zones in the Underground Rock Laboratory in Canada (Lee 1994), which have shown a consistent partial reset of the system.

Only a few studies used OSL for dating seismic slip using minerals from fault gouges. The method has been applied as a geothermometer by Rink et al. (1999) using the sensitization effect of quartz to study the signal resetting related to shear heating during earthquakes. Studies targeting active structures show that the signal has been at least partially reset in fault gouges at the surface and at depth conditions (e.g., Tsakalos et al. 2020). Spencer et al. (2012) studied drill core gouge samples (2604 m depth) from the San Andreas Fault zone, using infrared stimulated luminescence (IRSL) and thermoluminescence signals on poly-mineral and feldspar aliquots. Their dating results (139 ± 12 years) suggest an effective reset of the system by the 7.9 Mw Fort Tejon Earthquake.

Study area and sampling

Suitable sampling localities were scouted using geological maps (Anderle 1977a, b; Bigi et al. 1990a, b; Geološki zavod Slovenije 2003; Rockenschaub & Nowotny 2011; Schuster et al. 2015) and digital elevation models of spatial resolutions ranging from 30 to 1 m (SRTM: NASA JPL 2013; LIDAR ARSO, 2021; KAGIS 2020) from the study area delimited in Fig. 1. In total, three localities along the eastern PAF were studied, with their respective location details reported in Table 1 and the geology summarized in Fig. 4.

Table 1 Location, altitude, and overburden of the sampled localities

Fig. 4

a Sampling localities along the PAF. Fault traces after Schmid et al. (2004). b Geological map modified after Schuster et al. (2015), the legend is available in the supplements

The first sampling locality is Mauls in Italy, to the east of the transition between the Giudicarie and the Pustertal–Gailtal segments of the PAF. In the area, the PAF marks the division between high-grade metamorphic Austroalpine units to the north and a tonalite body belonging to the Periadriatic plutons in the south (4384 in Fig. 4b). The sampled outcrop contained gray to greenish cataclastic tonalites within a fault plane oriented 015/55 (nomenclature here and in the following: dip azimuth/dip angle) with slickensides oriented 302/28 (trend/plunge) indicating dextral slip (Fig. 5a). The main fault plane contained ca. 1 m wide, pale gray fault gouge produced by the comminution of the tonalites (Fig. 5b). The second sampled locality is in Finkenstein, Austria. There, the PAF separates the Periadriatic tonalitic lamellae from limestones belonging to the South Alpine units, and it is overlain by potentially displaced Miocene sedimentary units (4427 in Fig. 4b). The sampled outcrop consisted of a body of limestone from the South Alpine units surrounded by blocks of tonalite. At the location, the fault core hosted a dark gray to green fault gouge between tonalites and limestones (Fig. 5c). The third locality is in Plešivec, Slovenia. Here, the PAF placed a tonalitic lamellar body in contact with Triassic limestones from the South Alpine units (4429–4430 in Fig. 4b). In the outcrop, two fault planes hosted a fault gouge within cataclastic tonalites in the vicinity of the contact with the limestones to the south (Fig. 5d, e). The fault planes were oriented 155/73 (Fig. 5e, f), consistent with the regional strike of the PAF at map scale. The gouge exhibited a similarly oriented mm-to-cm-spaced foliation (140/65), forming a compositional layering defined by changes in coloration and possibly mineral components.

Fig. 5

Sampled localities along the PAF. a Mauls locality outcrop (sample 4348). b Detail of the sampled fault gouge at the Mauls locality outcrop. c Finkenstein locality outcrop (sample 4427). d Northern section of the Plešivec locality outcrop (sample 4429). e Southern section of the Plešivec locality outcrop (sample 4430). f Detail of the sampled gouge at the Plešivec locality outcrop showing the foliation in the gouge and the fault plane. g Detail of the sampled fault gouge at the Plešivec locality outcrop showing the foliation of the gouge

To sample the material in the field, we first removed the outermost, weathered portion of the gouge under direct sunlight exposure from each outcrop. Unweathered material was then collected using stainless steel cylinder tubes of 4.2 cm diameter and 15 cm length to avoid sunlight exposition. We collected additional material to determine the natural dose rate via gamma-ray spectrometry. The samples were collected at least 30 cm away from the adjacent wall rock to simplify the assumptions for the dose rate calculation. In cases where the gouges were sampled at a distance below 30 cm from the host rock, a dedicated wall–rock sample was collected to calculate the total dose rate.

Methodology

Sample preparation

All samples were prepared at the facilities of the Leibniz Institute for Applied Geophysics (LIAG)—Hannover, in a dark laboratory under red light conditions to avoid optical bleaching of the signals. The samples were first separated into two grain-size groups by adding water and decanting the finer fraction (< 40 μm) while it was in suspension into a separate container. The coarser portions of the samples (> 40 μm) were dry-sieved and divided into different grain-size ranges. Later, the grain-size fractions between 100 and 250 μm (63–300 μm in cases where the amount of material was considered insufficient) were selected for chemical treatment. Each sample was treated with sodium oxalate (> 24 h, for deflocculation and dissolution of remaining clay aggregates), 30% hydrochloric acid (1–2 h, for carbonate removal), and 30% hydrogen peroxide (> 24 h, for organic material removal). The resulting material was then separated into groups of mineral phases using heavy liquid solutions of Na-polytungstate with densities between 2.62 g/cm³ and 2.7 g/cm³ for quartz and below 2.58 g/cm³ for K-feldspar. Afterward, the fraction corresponding to quartz was etched using 40% hydrofluoric acid for 1 h, and any resulting fluoride precipitates were removed with 20% hydrochloric acid. Finally, the resulting quartz and the fraction corresponding to potassium feldspar were sieved separately into three grain-size ranges corresponding to 100–150 μm, 150–200 μm, and 200–250 μm.

Gamma-ray spectrometry sample preparation and dose rate measurements

The samples used to obtain the natural dose rates were collected simultaneously with the ESR/OSL samples from the same sampled spots within the fault gouge outcrops. These samples were prepared and measured at the facilities of LIAG. The initial material was dried at 110 °C for 24 h to obtain the natural water content by comparing the weight of the samples before and after drying. We assumed a 5% error in the natural water content. Due to the high amount of clay minerals in the samples, the material was aggregated after drying. Therefore, the samples were deconsolidated using a jaw crusher with an aperture of around 1–3 mm. Finally, 700 g of the resulting material was deposited in a sealed plastic container for 4 weeks to ensure equilibrium between ²²²Rn and its daughter nuclides.

An estimation of the radionuclide concentrations (²³⁸U, ²³²Th, and ⁴⁰K) was obtained by measuring the samples with an Ortec N-type HPGe gamma spectrometer. The dose rates for quartz and feldspar were calculated using the compositions and the conversion factors from Liritzis et al. (2012), the alpha attenuation factor calculation after Brennan (1991), the beta absorption factors from Guerin et al. (2012), the etching factor for beta absorption from Brennan (2003), the absorbed dose fraction of Rb from Readhead (2002), and the cosmic ray calculation from Prescott & Hutton (1994). We assumed no Rn loss and an etching depth for quartz of 10 μm. For feldspar, we calculated the internal dose rates using a K content of 12.5 ± 1.0% after Huntley and Baril (1997). For sample 4427, the distance between the gouge sample and the wall rock was 25 cm. Therefore, the gamma dose rate of the sample was calculated using the scaling factor following Aitken (1985).

ESR measuring parameters and protocols

The ESR measurements were conducted using an X-band JEOL FA100 series spectrometer. The aliquots consisted of 60 mg quartz in the grain size range of 100–150 μm contained in quartz glass tubes of 2 mm inner diameter. During the measurements, the microwave power was set to 10 mW using a 100 kHz modulation, with magnetic sweeps of 1 min and a sweep width of 30 mT. We measured the spectra under temperatures of – 160 °C and − 156.5 °C. The intensities from the Al center (Fig. 6) were calculated from the first positive peak (g ~ 2.02) to the last negative peak of the spectra (g ~ 1.99) and were then normalized using the 6th line of the Mn marker (g ~ 1.88), permanently inserted in the cavity. The Ti center could also be observed under the same conditions, but it was not recorded due to a very weak intensity in the analyzed samples. To account for the crystallographic anisotropy of quartz, each aliquot was measured three times, manually rotating the sample tube by 60° before recording the spectra. A mean intensity was calculated from these three measurements of the Al center and the uncertainty was estimated as ± 1σ standard error.

Fig. 6

ESR spectra indicating the position of the Al and Ti centers, the Mn marker, and peaks used for the calculation of the intensities. The spectra examples include measurements from a natural aliquot, a thermally reset aliquot, the undulated pattern produced by the Fe contamination, and the results of the baseline correction

A custom-built external X-ray radiation machine (Oppermann and Tsukamoto 2015) was used to irradiate, preheat, and thermally reset the signals in the aliquots. The X-ray tube provides a dose rate of 0.302 ± 0.014 Gy/s with a mean energy of 26 keV (Tsukamoto et al. 2021). Following the SAR protocol (Table 2a, left) established by Tsukamoto et al. (2015), all aliquots were preheated at 160 °C for 4 min (preheating conditions test explained in ‘‘Preheat test’’Sect. ) after each laboratory irradiation to eliminate the thermally unstable portion of the signal.

Table 2 a) Description of the applied ESR SAR and SAAD protocols

We heated the aliquots to a temperature of 420 °C for 4 min before administering regenerative doses, following the procedure after Tsukamoto et al. (2015), which indicates that at these conditions the signal from the Al center can be effectively removed (green spectrum in Fig. 6). Measurements applying the SAAD protocol (Table 2a, right) were conducted to verify the validity of the applied SAR protocol. Detailed explanations of the applied protocols and the quality controls used for ESR are provided in the following sections. For both applied protocols, a single aliquot from each sample was used to obtain an equivalent dose (D_e) value.

SAR protocol

For this protocol, we first preheated and measured a natural intensity in a single aliquot of the sample. Then, the aliquot was thermally reset and measured to record the zero-dose intensity. A dose response curve (DRC) was then constructed by measuring the intensities after increasing irradiation steps (at around 300 Gy, 900 Gy, 2700 Gy, 5100 Gy, 7500 Gy, and 10,600 Gy) until the saturation intensity was reached (i.e., the intensity at which the addition of further laboratory dose increments does not yield a stronger signal) and subtracting the zero-dose intensity (including the overlapping Peroxy center) from each measurement. The data points were fitted to a single saturating exponential function (Eq. 1), where ${I}_{sat}$ (a. u.) is the saturation intensity and ${D}_{0}$ describes the characteristic saturation dose in Gy. The D_e (Gy) was obtained by projecting the natural intensity onto the DRC (Fig. 7a). The uncertainty of the D_e was obtained from the median absolute deviation of the distribution resulting from a Monte-Carlo simulation of 2000 repetitions to fit the curve to randomized data points within the standard error of the measurements:

Fig. 7

Estimation of the D_e for: a SAR protocol. b SAAD protocol. Modified after Tsukamoto et al. (2020)

$$I\left(D\right)={I}_{sat}(1-{e}^{\frac{-D}{{D}_{0}}})$$

(1)

SAAD protocol

This protocol relies on the extrapolation of a DRC built starting from the natural intensity and with measurements recorded after subsequent steps of increasing irradiation (starting at around 600 Gy, then 1800 Gy, 4200 Gy, 6600 Gy, 9600 Gy, up to 12,700 Gy) on the same aliquot (Fig. 7b). The zero-dose intensity (obtained from the SAR protocol) was subtracted from each measurement, and the dose–response curve was constructed by fitting the obtained data points to a single saturated exponential function (Eq. 2), where ${I}_{sat}$ (a. u.) is the saturation intensity, D_e is the equivalent dose in Gy, and ${D}_{0}$ is the characteristic saturation dose in Gy. The uncertainty of the D_e was obtained from the median absolute deviation of the distribution resulting from a Monte-Carlo simulation of 2000 repetitions. For each iteration of the simulation, a DRC was fit to randomized data points within the standard error of the measurements and a single D_e was obtained from the x-intercept (extrapolation) of the curve:

$$I\left(D\right)={I}_{sat}(1-{e}^{\frac{-D+{D}_{e }}{{D}_{0 }}})$$

(2)

Preheat test

A preheat test was conducted to verify the temperature necessary to remove the thermally unstable part of the signal introduced after laboratory irradiation. In this case, a natural and a laboratory-irradiated (thermally reset, then irradiated to ~ 3000 Gy) aliquot from the same sample were selected. The Al center intensity of each aliquot was initially measured without preheating (room temperature preheat). Then, the intensity was measured after heating steps of 4 min at temperatures of 100 °C, 130 °C, 160 °C, 190 °C, 220 °C, and 250 °C. For both aliquots, the intensity from the Al center was compared to the respective preheating temperature of the step to obtain the curves displayed in Fig. 8a. The plateau shows that the Al center intensity for the natural aliquot starts decaying at preheating temperatures larger than 190 °C. On the other hand, there are two-step decays for the laboratory-irradiated aliquot. The first one starts after preheating at 100 °C where the thermally unstable portion of the signal starts to be evicted, and the second one occurs above 160 °C where partial resetting of the signal starts. The comparison of the laboratory-irradiated and the natural aliquot in Fig. 8a shows that the intensities at the 160 °C preheating temperature are comparable between both aliquots. Moreover, the anomalous peak introduced by laboratory irradiation is effectively removed after preheating at 160 °C (Fig. 8b), which confirms it as the optimal preheat temperature (Additional file 1: Figure S1, Additional file 2: Figure S1).

Fig. 8

Preheating conditions test for a natural and a regenerated aliquot from one of the Plešivec locality samples (4430). a Preheating plateau test for a naturally irradiated aliquot and a thermally reset and subsequently irradiated aliquot. b Comparison of the spectra before and after preheating at 160 °C for the regenerated aliquot

Isothermal decay test

Thermally assisted escape of the electrons from the traps can result in signal loss, and therefore, it may lead to an age underestimation. An isothermal decay test helps to assess how significant is the effect of thermal de-trapping. We conducted a test based on the study by Toyoda and Ikeya (1991) by selecting four aliquots from the same sample, each corresponding to a different isothermal holding temperature (160 °C, 190 °C, 220 °C, and 250 °C). The aliquots were thermally reset at 420 °C for 4 min, irradiated to a dose of around 5000 Gy, then preheated at 160 °C for 4 min (according to the results of the preheating conditions test) to normalize all signal intensities, and measured to record the corresponding initial intensity of the Al center. Successive steps of heating and measuring were conducted, maintaining the respective isothermal heating temperature for each aliquot but doubling the heating time from the previous iteration, starting at 4 min until reaching a heating time of about 2 h.

After the measurements were completed, each intensity was normalized to the initial irradiated intensity from the respective aliquot recorded after the preheating step. We calculated the kinetic parameters using the ESR MATLAB toolboxes after King et al. (2020). The thermal lifetime (${\tau }_{th}$) was obtained using the Gaussian model after Lambert (2018) described by the convolution integral in Eq. 3, where ${P}_{({E}_{a})}$ (Eq. 4) and ${\tau }_{\left({E}_{t}\right)}$ (Eq. 5) are the probability of electrons escaping from the trap due to thermal effects (a. u.) and the corresponding thermal lifetime in seconds at a specific trap depth $\left({E}_{t}\right)$, respectively. $\sigma ({E}_{t})$ is the Gaussian distribution of activation energies in eV around the mean trap depth $\mu \left({E}_{t}\right)$ in eV, s is the frequency factor in s⁻¹, ${k}_{B}$ is the Boltzmann constant in eV*K⁻¹, and T is the ambient temperature in K:

$${\tau }_{th}={\int }_{0}^{{E}_{t} max}{P}_{({E}_{t})}*{\tau }_{\left({E}_{t}\right)} d{E}_{t}$$

(3)

$${P}_{({E}_{t})}=\frac{1}{\sigma ({E}_{t})\sqrt{2\pi }}{e}^{\left[\frac{1}{-2}{\left[\frac{{E}_{t}-{\mu (E}_{t})}{\sigma ({E}_{t})}\right]}^{2}\right]}$$

(4)

$${\tau }_{\left({E}_{t}\right)}= {s}^{-1}{e}^{\frac{{E}_{t}}{{K}_{B}*T}}$$

(5)

Dose recovery test

We conducted a dose recovery test to confirm whether the thermal reset step in the SAR protocol induces a sensitivity change. Beerten et al. (2008) originally introduced a dose recovery test using quartz for ESR dating. However, they heated the sample at 500 °C for 1 h before a dose was given, and its effectiveness in monitoring any sensitivity change on natural aliquots in the SAR protocol was unclear. Tsukamoto et al. (2017) proposed a dose recovery test using modern sediment samples, and a similar approach was applied in subsequent studies (Richter and Tsukamoto 2022; Richter et al. 2022). In this study, we followed a modified procedure proposed by Toyoda et al. (2009), which is a regenerative-additive method and has been demonstrated as a means for conducting the dose recovery test by Fang and Grün (2020) and Richter et al. (2022). For this, the intensities measured using the SAAD protocol were compared to the calculated DRC from the SAR protocol. Apparent D_e values were calculated by projecting the SAAD intensities onto the SAR curve. The obtained D_e values were then compared to the added dose from each SAAD protocol measurement following the regenerative-additive dose plot. The slope of this plot is equivalent to the dose recovery ratio, and if this is close to unity (slope is 1), the SAR protocol is regarded to be robust. However, unlike the dose recovery test in luminescence dating, in which the dose recovery test is conducted using laboratory-bleached aliquots, it is difficult to reset the quartz ESR signals without heating the aliquots before a given dose is administered. Therefore, the dose recovery test may be affected by the signal saturation (i.e., the slope may not be linear) Additional file 3: Table S1, Additional file 4: Table S2).

OSL equipment, protocol, and data processing

For OSL, we used K-feldspar aliquots with a diameter of 2 mm and material in the grain size fraction of 100–150 μm, mounted on stainless steel disks with a diameter of 9.80 mm. The measurements were recorded with a Risø TL/OSL DA-20 reader available in the facilities of LIAG—Hannover. The reader is equipped with a beta ⁹⁰Sr/⁹⁰Y source that provides a dose rate of 0.119 ± 0.006 Gy/s. For stimulation, the reader includes LEDs emitting infrared (870 ± 40 nm, 3 clusters of 7 LEDs with a total maximum power of 145 mW/cm²). The light detection system of the reader includes a photomultiplier tube (bialkali EMI 9235QB) with a maximum detection efficiency between 200 and 400 nm and blue detection filters (Schött BG-39 and Corning 7–59) with a transmission window in the blue–violet region between 320 and 450 nm. We used IR optical stimulation at 90% power, and in cases where the light level was too bright for the detector, an extra neutral density filter (ND 1.0). For acquisition and initial processing, we used the Risø software suite. The R luminescence package after Kreutzer et al. (2012) was used to calculate the fading rate and corrected ages.

SAR protocol

We measured the IR₅₀ signal after Thomsen et al. (2008) and the pIRIR₂₂₅ signal after Buylaert et al. (2009), applying a SAR protocol summarized in Table 2b. The first step of the procedure consisted of preheating the aliquots at a temperature of 250 °C for 60 s. Next, the measurements were conducted during a first IR stimulation for 100 s at 50 °C, followed by a second IR stimulation for 200 s at  °C. Five aliquots from each sample were measured to obtain a mean natural intensity with the uncertainty calculated using the standard error of the measurements. Three of these aliquots were selected for measurements to build the DRCs. The aliquots were separately measured after giving doses of approximately 0 Gy, 110 Gy, 220 Gy, 560 Gy, 1130 Gy, and 2260 Gy, with a test dose of 90 Gy. The IR₅₀ intensities were calculated using the first 5 s of stimulation for the signal and the last 10 s for the background. The pIRIR₂₂₅ intensities were calculated using the first 10 s of stimulation for the signal and the last 20 s for the background.

The acceptance of the measurements was evaluated using two criteria after Murray and Wintle (2000). The first is the recycling ratio, which was calculated by dividing the sensitivity-corrected intensity of the recycling dose measurement by the corresponding intensity obtained at the same dose value during the dose–response curve measurement run. An acceptable recycling ratio should be close to 1, with a margin of error of 10%. The second is the recuperation rate, which was obtained by dividing the sensitivity-corrected intensity of the zero-dose measurement by the sensitivity-corrected natural intensity. The recuperation rate should not exceed 5%.

The dose–response curves were constructed by fitting the data from at least three measured aliquots from each sample to a single saturating exponential function (Eq. 6), where ${I}_{sat}$ is the saturation intensity (a. u.) and ${D}_{0}$ is the characteristic saturation dose in Gy. A fading test was performed using three bleached aliquots from each sample that were irradiated to a dose of around 240 Gy. Then, the signals were measured at different time intervals after the irradiation steps. The fading test data were processed using the R Luminescence package after Kreutzer et al. (2012) to calculate the g value and recombination center density (“analyse_FadingMeasurement” function) after the model from Huntley (2006) and to apply the fading correction (“calc_Huntley2006” function) after Kars et al. (2008), modeling the natural simulated DRCs and correcting the D_e. Saturation levels were estimated by dividing the natural intensity by the saturation intensity of the fading-corrected DRC:

$$I\left(D\right)={I}_{sat}(1-{e}^{\frac{-D+c}{{D}_{0 }}})$$

(6)

The reliability of the applied protocols was confirmed by implementing a dose recovery test conducted using six aliquots from each sample. The aliquots were bleached for 4 h using a Hönle UVACUBE 400 solar simulator. To test the dose recovery ratio, half of the bleached aliquots were then given a known dose which was subsequently measured using their respective DRC. A residual dose was obtained by comparing the remaining dose after bleaching to DRCs measured from the other half of the aliquots. Finally, the dose recovery ratio was calculated by subtracting the residual dose from the irradiated D_e and then dividing the value by the irradiated dose.

Age calculation

We calculated the ages following the standard procedure in trapped charge dating. Each D_e was divided by the respective dose rates for feldspar in OSL and for quartz in ESR. The uncertainty of the ages was calculated by propagating the error from the dose rate and the D_e. We calculated a minimum age using the double of the characteristic saturation dose (2D₀) instead of the D_e for the samples that could be considered in saturation (i.e., saturation level > 86% or D_e > 2D₀).

Results

Dose rates

The calculated dose rates for feldspar fall between 1.92 ± 0.12 and 4.72 ± 0.21 Gy/ka, and for quartz between 1.34 ± 0.06 and 3.99 ± 0.18 Gy/ka. Since the dose rates calculated for feldspar depend on the internal K content and the external alpha value, they are generally larger than the ones calculated for quartz. However, the low alpha content indicates that in this case, the dose rates are more influenced by the internal K content. The complete list of the results, including the respective contents of radiogenic nuclides, water, and contribution by cosmic rays, is reported in Table 3. The dose rates from the Mauls and Finkenstein localities are overall lower than the ones in the Plešivec locality samples. This is most likely due to differences in mineralogy between the gouges associated with the lithologies involved in comminution.

Table 3 Radiogenic nuclide compositions, water content, and calculated dose rates for the fault gouge samples

ESR dating

A strong signal from the Al center was detected in the measured ESR spectra of the samples, usually accompanied by a very weak signal from the Ti center (orange spectrum in Fig. 6). Due to this effect, the measurements were recorded only from the Al center. The aliquots from the Mauls sample (4384) presented a subtle Fe³⁺ contamination, observed as a long-wavelength undulation superimposed on the spectra and a slight negative baseline slope (blue spectrum in Fig. 6). However, the effect of the overprint was not strong enough to completely mask the signal of the Al center, meaning that the intensity could still be calculated. An additional baseline correction was performed on the spectra measured from this sample using the cwESR processing software (purple spectrum in Fig. 6).

The dose–response curves of the measured aliquots from each locality using the SAR and SAAD protocols are presented in Fig. 9, with the respective fit parameters available in the supplementary material. Most of the SAAD protocol DRC had a considerable amount of extrapolation to obtain the D_e, and there is a large uncertainty in the fit of the curve in the extrapolated portions. Overall, the D_e obtained using the SAAD protocol are larger than the ones obtained with the SAR protocol. For the latter, all D_e belong to measurements with intensities below the saturation points of the DRCs. The D_e obtained using the SAR protocol range between 1232 ± 23 Gy and 3026 ± 47 Gy, while for the SAAD protocol, they range between 1871 ± 55 Gy and 3426 ± 239 Gy. The difference in D_e values between the measurements from both protocols is carried over to the calculated ages. The SAR protocol ages range from 552 ± 26 to 1075 ± 47 ka, while the ages obtained with the SAAD protocol span from 530 ± 33 to 1522 ± 73 ka. A complete overview of the calculated D_e values, their respective ages using both protocols, and the saturation level of the natural intensity is presented in Table 4.

Fig. 9

ESR dose–response curves for the Al center obtained with the SAR and SAAD ESR protocols for the samples: a 4384—Mauls locality. b 4427—Finkenstein locality. c 4429—Plešivec locality. d 4430—Plešivec locality. The colored area surrounding the respective best fit of the dose response curve contains the curves fitted in 2000 simulations

Table 4 Results from ESR dating measuring the SAAD and SAR protocols on quartz aliquots

The applied ESR dose recovery test shows that the relationship between the apparent D_e and the added dose is sample-dependent. The Mauls (4384, Fig. 10a) and Finkenstein (4427, Fig. 10b) samples present a good correlation with dose recovery ratios of 0.88 ± 0.04 and 1.07 ± 0.23, respectively. However, for the Finkenstein sample, the uncertainty of the linear fit is large. Both Plešivec locality samples present a lower correlation, with dose recovery ratios of 0.64 ± 0.09 (sample 4429; Fig. 10c) and 0.78 ± 0.08 (sample 4430; Fig. 10d). Changes in sensitivity between the protocols can also be evaluated by shifting the SAAD curves by the respective SAR D_e and comparing them to the SAR DRCs (Fig. 11). For the studied samples, the shifted SAAD curves present an overall good match to the SAR measurements and curves, and most of the differences lie in the extrapolated portions of each SAAD DRC.

Fig. 10

Dose recovery test for ESR dating. Projection of the SAAD measurements onto the SAR dose response curves (left) and comparison of the apparent SAAD D_e and SAAD added doses (right) for the samples: a 4384—Mauls locality. b 4427—Finkenstein locality. c 4429—Plešivec locality. d 4430—Plešivec locality

Fig. 11

SAAD protocol dose response curves shifted by their respective SAR protocol D_e for the samples: a 4384—Mauls locality. b 4427—Finkenstein locality. c 4429—Plešivec locality. d 4430—Plešivec locality

The results from the isothermal decay test performed using aliquots from sample 4430 (Plešivec locality) are presented in Fig. 12. From the fit to the Gaussian model, we obtained an activation energy of 1.20 ± 0.08 eV, a distribution around the trap depth $\sigma ({E}_{t})$ of 0.16 ± 0.01 eV, and a frequency factor of 10^9.09±0.88 s⁻¹. Using these values and assuming an environmental temperature of 10 °C, the calculated lifetime of the Al center is 4.20*10⁷ Ma.

Fig. 12

ESR isothermal decay test performed on an aliquot from the Plešivec locality (sample 4430) at four different holding temperatures

OSL dating

The IR₅₀ and pIRIR₂₂₅ signals were measured for the samples obtained from Mauls (sample 4384 in Fig. 13a) and Plešivec (samples 4429 and 4430 in Fig. 13b, c, respectively). The sample from the Finkenstein locality (4427) could not be measured due to the absence of potassium feldspar in the fault gouge. After correcting the laboratory DRCs for anomalous fading, the samples show either a saturated state (i.e., D_e > 2D₀ or saturation level > 86%) or are very close to saturation. The respective DRC before and after accounting for anomalous fading are displayed in Fig. 13. According to the saturation levels presented in Table 5, all the pIRIR₂₂₅ signal measurements indicate saturation, whereas for the IR₅₀ signal the level is below but close to the threshold value. The effect of anomalous fading is more drastic for the IR₅₀ signal because of the larger g values (3.43 ± 0.31–3.48 ± 0.37%/decade) compared to those of the pIRIR₂₂₅ signal (1.97 ± 0.36–2.24 ± 0.34%/decade). The obtained D_e values from both signals and the respective 2D₀ values from the DRCs are reported in Table 5. The complete list of parameters of the DRCs before and after correcting for fading with the respective ρ’ and g values used for the correction is available in the supplementary material.

Fig. 13

OSL dose–response curves from the measured pIRIR₂₂₅ and IR₅₀ signals from the samples: a 4384—Mauls locality, b 4429—Plešivec locality, c 4430—Plešivec locality

Table 5 Results from OSL dating measuring the pIRIR₂₂₅ and IR₅₀ signals on potassium feldspar aliquots

Due to the saturation of the pIRIR₂₂₅ signal, we calculated minimum ages using the 2D₀ instead of the D_e. For the IR₅₀ signal, the combination of a near saturated state and a larger fading correction factor of the signal (g value) implies a less reliable D_e estimation. Therefore, minimum ages were also calculated using 2D₀ instead of the D_e. The obtained ages are reported in the respective columns of Table 5. For the Mauls sample (4384), the difference between the 2D₀ values in both measured signals results in a large minimum age variation, where from the pIRIR₂₂₅ signal the calculated age is 196 ± 12 ka and approximately 335 ± 21 ka from the IR₅₀ signal. The ages from both signals for the two Plešivec samples fall within the margin of error of each other. For sample 4429, the ages are 281 ± 16 ka and 279 ± 13 ka for the pIRIR₂₂₅ signal and the IR₅₀ signal, respectively, while for sample 4430, the ages are 280 ± 13 ka for the pIRIR₂₂₅ signal and 295 ± 13 ka for the IR₅₀ signal.

The dose recovery ratio was calculated with and without the subtraction of the residual dose and presented in Table 5. The results of the dose recovery test present little variation before and after subtracting the residual, and the values indicate an overall good quality of the measurements, with recovery ratios larger than 85% for both signals.

Discussion

In general, the equivalent doses and ESR ages (Table 4) obtained with the SAAD and SAR protocols are not in agreement. The SAAD DRCs in Fig. 9 involve a large extrapolation that introduces more uncertainty at the time of the D_e calculation, especially when dealing with older samples that present natural intensities closer to the saturation point of the curve. This results in SAAD ages that, when compared to their SAR counterparts, are older and have a larger uncertainty. The main advantage of the SAR protocol over the SAAD protocol is that for the SAR protocol, it is not necessary to extrapolate a large portion of the DRC to obtain the D_e. Despite the difficulty in accounting for sensitivity changes during the thermal reset step when applying the SAR protocol, our results from the dose recovery in Fig. 10 test show dose recovery rates close to unity and indicate that the curves are comparable to those of the SAAD protocol (Fig. 11). Therefore, it is unlikely that a large sensitivity change was introduced to the measurements after the thermal reset step in the SAR protocol. Taking into account all the ESR dating results and the limitations of the SAAD protocol, especially when dealing with old samples, we consider the ages obtained with the SAR protocol to be more reliable than the SAAD protocol ages.

The ESR kinetic parameters and thermal lifetime are largely sample-dependent. The obtained kinetic parameters for the Al center fall within the range of parameters compiled by Fang & Grün (2020) and Richter et al. (2020) from ESR dating studies. The thermal lifetime for the Al center reported in this study (4.20*10⁷ Ma) falls within the values calculated by Richter et al. (2020) from other works in the literature at the same ambient temperature conditions. The lifetime we obtained is also larger than the one calculated using the kinetic parameters from King et al. (2020) assuming a temperature of 10 °C (3.25*10⁴ Ma). In any case, our thermal lifetime is still above 10 times the dating limit for the method (Aitken 1985), meaning that we can assume that no significant loss of signal due to thermal de-trapping has occurred during the last couple of million years.

Considering the results from the frictional heating experiments formulated by Yang et al. (2019), suggesting that the Al center in quartz is most likely partially reset by earthquakes of M_w 7–8, we assume that the shear heating was potentially not sufficient to fully reset the system during the last earthquakes released by the PAF. Consequently, we interpret the obtained ESR ages as the maximum ages of the last resetting event of the system, and therefore, the maximum age of the last large earthquake (Tsukamoto et al. 2020) along the specific fault segment. Still, an age overestimation due to an incomplete system reset could be accounted for by measuring multiple centers in ESR or targeting different grain size fractions (e.g., Fukuchi 1988; Schwarcz et al. 1987; Lee & Schwarcz 1994a, b; Lee 1994) while applying the SAR protocol. However, the signal from the Ti center in our samples is weak, and the centers such as E’ and OHC do not strongly depend on shear heating to be effectively reset compared to the Al center (Yang et al. 2019). Therefore, only the grain size plateau approach using the Al center could still be applied. Furthermore, establishing a temperature estimate for shear heating along the fault plane is difficult. A potential way to obtain such estimates would be for instance, to measure the sensitivity changes in the OSL signal with increasing heating temperature (Rink et al. 1999), or to apply a Raman spectroscopy geothermometer for organic matter (RSCM, e.g., Kouketsu et al. 2013; Furuichi et al. 2015; Lünsdorf et al. 2017; Henry et al. 2019) on cataclasites and gouges from the fault zone.

Our results from the OSL dating show that the pIRIR₂₂₅ signals in all samples are in saturation (Table 5, Fig. 13). In the case of the IR₅₀ signals, the analyzed samples are close but below saturation. However, for older samples (i.e., close to saturation) the fading correction factor for the IR₅₀ signal is larger than for the pIRIR₂₂₅ signal, meaning that the corrected D_e are less reliable than those obtained for the pIRIR₂₂₅ signal. In general, other studies dealing with OSL faulting of fault gouge have identified a reset of the system associated with fault activity (e.g., Spencer et al. 2012; Tsakalos et al. 2020). However, these studies have been conducted at depth, under higher ambient temperatures and confining pressure, where the material is more likely to be effectively reset during coseismic slip. The saturated state of the measured signals could be the result of a large recurrence interval of the PAF combined with only low-magnitude events that do not fully reset the system at near-surface conditions. Another possibility would be the lack of events that effectively reset the system during the datable time range of the method. Therefore, we consider the ages calculated from the pIRIR₂₂₅ signal (Table 5) as the minimum ages of the last resetting event of the system. A potentially helpful approach to identify likely resetting of the traps would be to analyze thermoluminescence signals collectively with IRSL, as proposed by Spencer et al. (2012).

By taking advantage of the difference in saturation limits of the ESR and OSL dating methods, a combined approach using both allows us to narrow down the temporal range of the last resetting event of the system (Fig. 14) and establish whether a fault has experienced seismotectonic deformation during the Quaternary. This is useful in cases where geodetic monitoring or instruments are incapable of yielding significant data due to the slow deformation rates. In the case of the studied portion of the PAF, the obtained ages suggest that coseismic slip occurred along the fault system during the Quaternary with a maximum age for the last resetting event of the system ranging from 1075 ± 47 to 552 ± 26 ka (ESR SAR) and minimum ages in the range from 196 ± 12 to 281 ± 16 ka (pIRIR₂₂₅ using 2D₀ estimates). The individual ages are relatively congruent (Fig. 15), and their differences could be attributed to different amounts of partial resetting of the system, differences in the timing of coseismic slip along the specific segment of the PAF, or reactivation of different fault planes as suggested by the difference in the ESR age from the Plešivec locality samples (4429 and 4430). In general, previous studies hint at fault activity along the PAF of at least Late Miocene to Pliocene age (e.g., Zwingmann and Mancktelow 2004; Heberer et al. 2017). Our findings suggest a much younger activity.

Fig. 14

Comparison between the datable range of OSL and ESR dating from the samples: a 4384—Mauls locality, b 4429—Plešivec locality, c 4430—Plešivec locality

Fig. 15

Maximum (ESR SAR) and minimum (OSL pIRIR₂₂₅) ages of the last resetting event from the studied localities along the eastern PAF

Considering most of the definitions of what makes a fault considered active, the narrowed-down period of earthquake activity we obtained for this segment of the PAF implies that by some definitions the fault cannot be determined active. Our ages cannot account for Holocene displacement (Slemmons and McKinney 1977), and although our minimum limit suggests activity younger than 0.5 Ma it is not possible to confirm how recurrent it was (U.S. National Regulatory Commission 1997). However, there are four reasons why it could be considered at least potentially active. (1) The Eastern PAF is characterized by low seismicity and low deformation rates, similar to what usually occurs in intraplate settings. According to the definition of the International Atomic Energy Agency (2010), the seismically active period we dated for the structure fits within what could be considered active for such settings. (2) Earthquake ages obtained with trapped charge dating methods are generally older than the last earthquake triggered on active faults (e.g., Fukuchi et al. 1986; Buhay et al. 1988; Tsakalos et al. 2020). They do not necessarily reflect the last coseismic slip along the structures but can still be used to assess whether a structure has been active during the Quaternary. (3) The GPS velocities displayed in Fig. 3 show there is still a transpressional component to the PAF, and the orientation of the maximum horizontal compressive stress is oriented at a high angle to PAF, but not entirely perpendicular to the structure (Aiman et al. 2023). As a consequence, low resolved shear stresses along the PAF might be the cause behind its low deformation rates, but it does not necessarily mean the structure is inactive since it could take a long time to accumulate enough strain. In addition, a very weak fault (i.e., one with low static friction, for instance, the San Andreas Fault: Carpenter et al. 2015) may slip even when principal stresses look unfavorably oriented and resolved shear stresses are low. Besides, strong historical earthquakes have occurred in the vicinity of the structure (Roman earthquake at Celje: Kazmer et al. (2023). Overall, the PAF has the potential for future offset, which is one of the criteria to identify active faults defined by several authors (Willis and Wood 1924; Slemmons and McKinney 1977; U.S. Nuclear Regulatory Commission 1997). (4) Other structures under similar tectonic settings (i.e., low deformation rates and scarce to inexistent historical and instrumental earthquake records) where earthquakes recently occurred (e.g., Provence region, France: Ritz et al. 2020; Thomas et al. 2020; Eastern US: Figueiredo et al. 2022), imply that long recurrence intervals could hinder their interpretation as active or potentially active faults. Spooner et al. (2019) have suggested that present-day seismicity with M > 6 events in the Alps correlates with regions of smaller crustal thickness and lower density. However, in the interior of the Alps strong historical earthquakes are documented (Stucchi et al. 2012), and additional pre-historical events have been identified in lake paleoseismology (Daxer et al. 2020, 2022a, b; Oswald et al. 2022) and cave studies (Baroň et al. 2022). These observations are in line with our findings—large earthquakes do occur in the interior of the mountain chain, but they are too rare to be captured by instrumental and historical records alone. In the regional tectonic context, an active PAF will likely accommodate a small fraction of the Adria–Europe convergence, probably with a component of right-lateral slip, such as during the Neogene lateral extrusion of the Eastern Alps towards the Pannonian Basin (Ratschbacher et al. 1991a, b).

Conclusions

Trapped charge dating methods allow dating seismotectonic deformation at near-surface conditions in places where deformation rates are low or where neither instrumental nor historical earthquake records help to identify active faults. A lack of large instrumentally recorded earthquakes on the PAF system, combined with an ambiguous historical earthquake record, made the structure a suitable target to apply the methods to unravel the timing of seismotectonic deformation produced by the fault during the Quaternary. In this work, we systematically compared two available methodological approaches in ESR dating to establish a meaningful direct age of active deformation. Furthermore, we introduced the first systematic approach of combined ESR and OSL dating on quartz and feldspar from fault gouges to date active deformation. Finally, we presented the first successful attempt at absolute dating for active deformation along the PAF in the Alps.

Due to the limitations of the ESR SAAD protocol when dealing with old samples and the systematic advantages of the SAR protocol, we consider the ages obtained with the SAR protocol ages more reliable than the SAAD protocol ages. The main advantage of the SAR protocol over the SAAD protocol is a shorter uncertainty of the DRC parameters and the D_e, both associated with the lack of extrapolation for a large part of the curve. Large extrapolation steps largely affect the age calculation for aliquots with natural intensities closer to a saturated state. However, changes in sensitivity after bleaching the signal during the SAR protocol in ESR dating can be difficult to identify or quantify. Therefore, we employed a dose recovery test to verify the validity of the SAR DRC and D_e, which yielded recovery rates close to unity for most samples.

Our dating results suggest that the PAF accommodated seismotectonic deformation associated with Adria–Europe convergence during the Quaternary, with a maximum age for the last resetting event of the system (and the last large magnitude earthquake) in the range of 1075 ± 47 to 552 ± 26 ka (ESR SAR) and minimum ages in the range from 196 ± 12 to 281 ± 16 ka (pIRIR₂₂₅ using 2D₀ estimates). Therefore, in places where activity is not so high, ESR dating on quartz could be considered the preferred method for unraveling periods of seismogenic faulting, while OSL can be additionally applied to constrain the minimum ages for the last resetting event of the system, even when the signals are in saturation. Furthermore, a combined approach of the methods allows to narrow down the timing of Quaternary seismotectonic deformation along structures where geodetic monitoring and instruments do not yield results, such as the PAF. Considering the different definitions of what constitutes an active fault, the ages we obtained, and regional observations of the geometrical compatibility with the current stress regime, it would be possible to classify the easternmost segment of the PAF as a potentially active fault.