TephAta – an online collection of tephra data from the Atacama Desert

Tephrostratigraphy and tephrochronology are powerful tools using volcanic ash (tephra) layers for establishing stratigraphic correlations and/or to obtain chronological information for different kinds of sedimentary archives. To develop robust tephrostratigraphic frameworks it is essential to precisely characterize the physical, geochemical and chronological properties of tephra layers and their sampling sites. These frameworks allow newly discovered tephra layers to be correlated with established sequences, providing stratigraphic and chronological context and enhancing reconstructions of the spatial and temporal patterns of regional volcanic activity. Tephrostratigraphic frameworks have been developed or are currently emerging in many regions of the world. However, for some regions their potential remains underutilized. One such region is the Atacama Desert of northern Chile, despite its long history of frequent volcanism and associated tephra deposition. Within the framework of the Collaborative Research Centre “Earth – Evolution at the Dry Limit” (CRC1211), the glass phase geochemical compositions of Pleistocene tephra layers were recently systematically investigated to refine their stratigraphic and chronological placement. Alongside this effort, a dedicated tephra database, TephAta, was created to provide a long-term repository for tephrochronological research in the Atacama Desert. TephAta enables the comprehensive digital documentation of diverse tephra-related datasets within a single platform and offers search functionalities that support and streamline ongoing expansion of the regional framework, following FAIR (findability, accessibility, interoperability, and reusability) data principles. A first data release contains 106 tephra samples from 91 deposits, with continued spatial and temporal growth planned. All data are accessible via the TephAta portal (https://www.crc1211db.uni-koeln.de/tephata, last access: 29 May 2026) and datasets described in this paper are available at https://doi.org/10.60520/IEDA/114209 (Leicher, 2026). Beyond serving as a stratigraphic and chronological tool for paleoenvironmental or mapping studies, TephAta also facilitates investigations into the regions explosive volcanic history and supports volcanic hazard and risk assessment.

1 Introduction

Volcanic ash deposits are intercalated within many types of sedimentary sequences. This makes them highly valuable as isochronous marker horizons for stratigraphic and chronological studies in both geoscience and archaeology (Lowe, 2011). The geochemical composition of volcanic glass is one of the most commonly used and reliable parameters (Lowe et al., 2017) for correlating tephra layers and establishing tephrostratigraphic frameworks. During ash dispersal, aeolian fractionation often results in volcanic glass persisting as the dominant eruptive component with its relative abundance increases with distance from source. Owing to its quenched nature, the glass retains a geochemical composition that closely reflects the composition of the magma at the time of eruption, providing a distinctive fingerprint.

Tephrochronology is a well-developed and frequently used technique in many regions of the world including the Mediterranean (e.g., Sulpizio et al., 2003; Wulf et al., 2004; Giaccio et al., 2019; Leicher et al., 2021; Vakhrameeva et al., 2021), Northern Europe and the N-Atlantic (e.g., Griggs et al., 2014; Lowe et al., 2015; Abbott et al., 2018; Sáez et al., 2025), Central America (e.g., Kutterolf et al., 2016; Schindlbeck et al., 2018), New Zealand (e.g., Gehrels et al., 2006; Lowe et al., 2013; Hopkins et al., 2021) and eastern Asia (e.g., Sagawa et al., 2018; Albert et al., 2019; Portnyagin et al., 2020; Feng et al., 2022). Tephrochronological databases exist for specific regions, e.g., for (northern) Europe (TephraBase, Newton et al., 2007; Riede et al., 2011; RESET, Bronk Ramsey et al., 2015), the east African Rift (EarthD, Mana and Dimaggio, 2023), Alaska (GeoDIVA, Cameron et al., 2022), Kamchatka (TephraKam, Portnyagin et al., 2020), New Zealand (TephraNZ, Hopkins et al., 2021), southern Chile (BOOM!, Martínez Fontaine et al., 2023) or Antarctica (AnT, Kurbatov et al., 2014).

In some regions, however, tephrochronology remains underdeveloped, despite a long history of explosive volcanism that has produced frequent, widely dispersed tephra with diverse geochemical compositions. One such region is the western side of the Andes in Central South America, including the Atacama Desert. The Central Volcanic Zone of the Andes (14–28° S; De Silva, 1989a) has a long volcanic eruptive history (ca. 240 Ma; Oliveros et al., 2018) and is renowned for major explosive eruptive events during the Neogene (ca. 25–1 Ma, cf. Wörner et al., 2000; Burns et al., 2015; Van Zalinge et al., 2016). Geochemical variations characteristic of individual volcanoes and eruptions are influenced not only by the composition of the parent magma and degree of crystallization, but also by the geodynamic evolution of the Andes. Spatial and temporal differences in the crustal thickness and composition have produced to varying degrees of magma–crust interactions, leading to diverse and often additional pathways of magmatic differentiation for individual eruptions (Kay et al., 2010; Brandmeier and Wörner, 2016; Burns and De Silva, 2023). The Atacama Deserts' long-term aridity is thought to have initiated as early as ca. 25 Ma ago (Dunai et al., 2005; Evenstar et al., 2017) and sets with the low degree of erosion and weathering (e.g., Ritter et al., 2023) suitable climatic conditions for the preservation of volcanic deposits. Despite dominating easterly wind systems, tephra deposits can be found abundantly from the Andes to the Pacific coast (Marquardt et al., 2005; Vásquez and Sepúlveda, 2013; e.g., Breitkreuz et al., 2014). Tephra layers serve as crucial isochronous markers for dating sedimentary sequences within these arid environments, where other dateable materials are scarce and the temporal resolution of sediment records is often low. Thus, tephra layers of the Atacama region hitherto were mainly directly dated to constrain paleoenvironmental and tectonic processes (Sáez et al., 2012; Kirk-Lawlor et al., 2013; Jordan et al., 2014) or for use in extended geological mapping of the Chilean National Geology and Mining Survey (SERNAGEOMIN; e.g., Medina et al., 2012; Blanco and Tomlinson, 2013; Sepúlveda et al., 2014; Vásquez et al., 2018). Correlation of tephra layers by their glass geochemistry to establish local stratigraphic correlations is rare in the Atacama Desert (Placzek et al., 2009; Breitkreuz et al., 2014; Tapia et al., 2015). In contrast, intensive research, including comprehenisve geochemical (whole-rock compositions and isotopes) and chronological data, has been carried out on the magmatic evolution on the volcanoes of the Andes (e.g., De Silva, 1989b; Wörner et al., 2000; Kay et al., 2010; Mamani et al., 2010). Whole-rock analyses, however, have been found to be less suitable for tephrochronological alignments due to site-dependent variations in their main components (variable relative abundances of glass, crystals and lithics) caused by aeolian fractionation, temporal eruptive variations and the influence of alteration (Tomlinson et al., 2012a; Lowe et al., 2017). Thus, a tephrostratigraphic framework based on glass geochemical compositions and geochronological data is currently missing at the western side of the Andes.

Figure 1 (a) Structure of TephAta with its seven categories and different sources for data. (b) An example employing the sample filter function of TephAta showing the filtered samples on an interactive map and listing them below (CRC1211DB-TephAta, 2026). Map services and basemap provided by Esri via ArcGIS Online using data providers Esri, TomTom, Garmin, FAO, NOAA and USGS | Powered by Esri.

The Collaborative Research Centre “Earth – Evolution at the Dry Limit” (CRC1211) aims at disentangling how the shaping of land-surfaces by past episodes of wetter climate coevolved with the evolution of life in arid environments (Dunai et al., 2020). The main working area of the CRC is the Atacama Desert in Chile (Fig. 1a). For this multidisciplinary endeavor, establishing chronological control of the sedimentary climate archives and determining the timing and rates of surface processes are of fundamental importance. The inherent lack of dateable material in the hyperarid Atacama limits alternative chronological methods. Tephra layers represent a promising alternative, appearing across diverse sedimentary archives such as alluvial, fluvial, lacustrine, salar, and playa deposits (Sáez et al., 2012; Jordan et al., 2014; Vásquez et al., 2018; Medialdea et al., 2020; Ritter et al., 2022; Wennrich et al., 2025). Within the CRC1211, tephrochronology was used systematically to establish a time-stratigraphic framework for the Atacama Desert. These data are stored in a newly developed regionally focused tephra database, TephAta. TephAta can store a wide variety of tephra metadata, from field and lab context to physical, chemical and chronological characteristics. Storing the data in the dedicated TephAta database will facilitate the development of a regional tephrostratigraphic framework for the Atacama Desert. TephAta provides easy access to complex information that normally is distributed in several media (publications, private or public databases), hence fostering FAIR (findable, accessible, interoperable and reusable) data principals (Wilkinson et al., 2016). The design of TephAta is also meant to provide user friendly integration of suitable parts of the dataset (e.g. geochemical data) in global databases such as GEOROC (https://georoc.eu/, last access: 29 May 2026) or the tephra-data-optimized EARTHCHEM (https://www.earthchem.org/, last access: 29 May 2026) repository. This paper introduces the TephAta database and illustrates its usefulness by the presentation and discussion of an exemplary dataset obtained on widely distributed tephra deposits of Mid Pleistocene age from the central Atacama Desert.

2 The Central Volcanic Zone of the Andes

The subduction of the Farallon plate beneath the west coast of South America has caused continental arc volcanism within the Central Andes since Triassic times (Oliveros et al., 2018). The volcanic arc is characterized by low-flux, steady state andesitic magmatism and the construction of composite volcanoes (Burns et al., 2015). Starting about 25 Ma ago, an increase of the subduction rate and a shift in the subduction geometry of the Nazca Plate caused a eastward migration of the arc and associated crustal shortening and thickening (Allmendinger et al., 1997). This change is correlated with a shift to a high-flux, flare-up arc magmatism, which resulted in the production and eruption of large volumes of silicic “crustal” magmas (Wörner et al., 2018; Burns and De Silva, 2023). Ignimbrite volcanism is supposed to be not contemporaneous within the Central Volcanic Zone, indicating a north-south gradient towards younger major eruptive events (De Silva, 1989a; De Silva and Kay, 2018). The oldest ignimbrite deposits are incorporated within the Altos de Pica Formation (22–25 Ma; Jordan et al., 2014), the Oxaya Formation (19–23 Ma; Wörner et al., 2002; Van Zalinge et al., 2016) and the El Diablo Formations (11–16 Ma; Jordan et al., 2014). Between 10 and 1 Ma, magmatic fluxes strongly increased and numerous voluminous ignimbrites were erupted in pulses peaking at ca. 8.4, 5.5 and 4.0 Ma (De Silva, 1989a; Sáez et al., 1999; Kay et al., 2010; Salisbury et al., 2011; Burns and De Silva, 2023). One of the most prominent surface morphologies of these ignimbrite deposits is the Altiplano-Puna Volcanic Complex (APVC) between 21 and 24° S (De Silva, 1989b). Subsequently, the high-flux volcanic activity waned and the construction of composite volcanoes and small-volume lava domes indicates the return to steady-state conditions (Burns and De Silva, 2023).

3 Database structure, samples, and analytical data

3.1 Database Structure

TephAta is hosted within the CRC1211 database (https://www.crc1211db.uni-koeln.de, last access: 29 May 2026) and is focused entirely on storing and displaying information without data analysis by respective algorithms. In the backend, TephAta uses a relational database, MariaDB as data management system, communicating with a basic PHP instance hosted by the ITCC on virtual machines, while the frontend is handled by standard web technologies, with some display libraries (such as Tabulator or HighCharts) providing JavaScript interactivity. TephAta is designed to store field site and sample metadata of tephra samples in combination with respective analytical datasets. The data structure is adapted to the global tephra community guidelines for data acquisition and reporting outlined in Wallace et al. (2022). Data can originate from new investigations, but also from published work.

TephAta is organized into seven data categories: site, sample, physical properties, geochemistry, chronology, chronostratigraphy, and tephra correlations (Fig. 1a). A general introduction of the data stored in the different categories is given below and a full list and description of all input fields and upload functions is given in the Supplement.

The category “site” allows the documentation of sample location details including information about the site name, a general (log-) description, geographic position, physical habitus, and spatial context of an investigated location. The site category lists and links all samples taken from the same archive.

The section “sample” provides information when, how and by whom a sample from a specific site was taken. Further details about the sample type and material, its position within a site and the observed depositional processes can be defined. Information about sub- or split-samples and related (laboratory) labels can be listed together with the availability and accessibility of material and data. Specific analyses that were carried out can be indicated in an overview list.

The category “physical properties” collects field and microscopic observations. These include descriptions of the macrophysical properties of the stratigraphic layer (thickness, color, sedimentological structures, components), and its microphysical characteristics such as glass fragment morphology, mineral assemblages, or degree of alteration. Field sketches, images or notes and microscopic images can also be integrated.

The “geochemistry” category facilitates the classification of volcanic rock types using the total alkali vs silica (TAS) classification (Le Bas et al., 1986). Geochemical datasets of major (> 1 wt %), minor (1–0.1 wt %) and trace (< 0.1 wt %) element data, but also Sr, Nd, Pb isotope data can be archived. Unprocessed raw data, instrument log files and laboratory protocol data of geochemical analyses can be stored as supplementary files.

The “chronology” category collects age information related to a specific sample. Details on the applied dating technique(s) can be entered along with information about the dated material and specific age information (uncertainties, type of age calculation, xenocryst presence). Full analytical datasets including laboratory details as well as unprocessed and processed data-files (e.g., single crystal dating results) can be stored as supplementary files. For samples with multiple age determinations (e.g., by different dating techniques), users can designate a “best age”, to represent the most reliable age estimate in the sample overview.

Within the data category “chronostratigraphy” a specific sample can be categorized by its general (e.g. erathem, stage, etc.) and regional stratigraphic classifications. When available, this information is used to define minimum and maximum age constraints. For samples which could be correlated with other tephra samples or a specific volcanic eruption by tephrostratigraphic means, a link between samples can be documented within “tephra correlations” category. Groups of correlated tephra can be further defined by a common name and their most reliable age. In case a tephra layer can be traced back to a specific volcanic origin, both the source volcano and eruption can be linked to the respective correlation.

All data is entered via the TephAta web interface (https://www.crc1211db.uni-koeln.de/tephata, last access: 29 May 2026) using guided input forms, including unified predefined lists, as well as free-text fields for the input of individual data items. Geochemical data can be uploaded (and downloaded) using template data sheets provided on the website. The template files integrate sample and instrument metadata (e.g., sample IDs and types, related references, applied analytical protocols and instrument) with results from standards and samples, ensuring transparent documentation for interoperability and data reuse.

For all categories, additional data files (e.g., raw data and detailed instrument settings, field, and sample photo documentation) can be uploaded as zip-files. Datasets stored within TephAta can be continuously extended, as additional analyses become available. If no IGSN (International Generic Sample Number) was assigned to a site or sample before registration within TephAta, a new IGSN will be minted and connected with the internal metadata upon data submission in TephAta. The new IGSNs will be registered at the GFZ Data Services at the GFZ Helmholtz Centre for Geosciences and will then be associated with the TephAta URL, providing a link to the full metadata released within the database. Existing IGSN's can be linked to a sample during data submission. The IGSN enables a unique identification of a site and a sample and allows documentation outside TephAta including a connection to other databases. Data exchange to other global databases is also fostered by the structures of file templates, which are organized to match the needs of the global tephra-data repository EarthChem (https://www.earthchem.org/, last access: 29 May 2026; Kuehn et al., 2023).

Datasets stored within TephAta can be accessed through the categories: “sites” and “samples”. All data associated with a given sample or site are presented on a dedicated page that compiles and displays all shared entries. The sample list is linked to a map tool, which allows exploring entries within specific geographic areas (Fig. 1b). The sample list can be refined using combinable filters, including specific site sample and group of tephra correlations names (site name, sample name, lab ID, IGSN, name of group of tephra correlations). Users can also filter by the type of site (e.g., only alluvial fans), geochemical properties (TAS-classification, range of specific major, minor and/or trace element concentrations) and chronostratigraphic (series) and chronological (age range) values. Furthermore, users can select up to 15 samples to create compositional x-y plots. An overview of established tephra correlations is given by the “tephra correlations” list. These functions are designed to facilitate the identification of potential tephra correlations, providing the basis for detailed testing and subsequent establishment of tephrochronological frameworks.

3.2 Samples and chronological and geochemical data

TephAta focuses on tephra samples from the Atacama Desert and adjacent regions, namely the Coastal Cordillera, the Central Depression, Precordillera and parts of the Western Cordillera regardless of their age. To date, the dataset primarily includes samples from the late Middle Pleistocene to Late Pleistocene, largely collected during field campaigns within the CRC1211 (2013–2023). These samples are derived from sites, which were within the scientific focus areas of CRC1211 subprojects and adjacent sites to create a broad data foundation for characterizing major eruptive events and developing a tephrochronological framework. This set of samples was augmented by identifying tephra sites on geological maps and/or described in literature. Where tephra layers were identified near investigated CRC1211 archives or shared similar ages, sample splits were requested from the Chilean National Geology and Mining Survey (SERNAGEOMIN) or the respective researchers of literature sources. If material could not be provided, respective sample sites were revisited and resampled during field CRC1211 field campaigns. TephAta also allows the integration of legacy data for samples that are not accessible. Given their diverse origins across different projects and decades, the currently investigated samples lack a standardized initial collection protocol. Consequently, TephAta has been designed to accommodate this heterogeneity within a unified framework by offering a comprehensive set of optional data fields that can be utilized as information becomes available. Metadata and morphological field descriptions were extracted from references, field notes, or provided by personal communication.

Table 1 Overview table listing all sites and samples and respective IGSN included within TephAta along with information about their type of site, the availability of glass geochemical data (MME = major, minor elements, TE = trace elements), proposed age and dating details (if available, age and age uncertainty, dating method, dated mineral, dating laboratory, mineral standard and decay constant applied, type of age and reference of age). Further it is listed if the tephra was previously known and the sample was provided or (re-)sampled within the activities of the CRC1211, more details about the origin of samples are given within TephAta database. Sample AD-85 was provided by J. Quade (University of Arizona, USA) and sample TJ09-PdT-1 was provided by T. Jordan (Cornell University, USA). References are: 1 = Breitkreuz et al. (2014), 2 = Placzek et al. (2009), 3 = Sáez et al. (2012), 4 = Ritter et al. (2018), 5 = this study, 6 = Gardeweg and Sellés (2013), 7 = Astudillo et al. (2017), 8 = Blanco and Tomlinson (2013), 9 = Medina et al. (2012), 10 = Escribano et al. (2013), 11 = Marinovic et al. (1995), 12 = Carrizo et al. (2008), 13 = Vásquez et al. (2018), 14 = Vásquez and Sepúlveda (2013), 15 = Sepúlveda et al. (2014), 16 = Rodríguez et al. (2015), 17 = Horn (1991), 18 = Wörner et al. (2000), 19 = Medialdea et al. (2020), 20 = Fornari et al. (2001), 21 = Fritz et al. (2004), 22 = Sepúlveda et al. (2023), 23 = May et al. (2020).

Chronological data

Available ages for the samples currently included in TephAta have been extracted from published data and were complemented by new chronological results. Available ages from literature were obtained by ⁴⁰Ar $/$ ³⁹Ar, K-Ar and U-Pb dating, but also inferring indirect tephra ages by dating the host sediment. Available details of the respective dating techniques, age calculations and references are stored sample-specific within TephAta and are summarized in Table 1.

For two samples (PAG-T4 and IRU-1a) new ages were obtained by ⁴⁰Ar $/$ ³⁹Ar multi-grain fusion dating of biotite at the Vrije Universiteit Amsterdam and are presented here for the first time. Mineral separates were achieved by sieving, magnetic separation, and hand picking before the selected mineral separate was packed in a 6 mm ID AI packages and loaded in a 25 mm ID Al cup together with Fish Canyon Tuff sanidine (FCs) standard. Sample and standard were irradiated at the Oregon State University TRIGA reactor in the cadmium shielded CLICIT facility for 7 h (irradiation code VU114 and 15-OSU-06). At Vrije Universiteit Amsterdam after irradiation, samples and standards were unpacked and loaded in a 185 hole Cu tray and baked overnight at 250 °C under vacuum. This tray was then placed in a doubly pumped vacuum chamber with Zn-Se window and baked overnight at 120 °C under high vacuum. This chamber is connected to a ThermoFisher NGPrep gas purification line equipped with a hot GP50, a cold finger (Lauda at −70 °C) and hot St707 getter. Samples (1–3 grains/fusion) and standards (1 grain/fusion) are fused using a 25 W Synrad CO₂ laser. Released gas is analyzed on an ARGUS VI+ noble gas mass spectrometer, equipped with four Faraday cups at the H2, H1, AX and L1 positions and two compact discrete dynodes (CDDs) at positions L2 and L3. The system is equipped with a 1012 Ohm amplifier on H2 and 1013 Ohm amplifiers on H1, AX and L1 cups. Samples were run on H1–L3 collectors. Similar to Phillips and Matchan (2013), no bias corrections were applied, but samples and standards were analyzed in the same tray (and thus at more or less the same time) alternating with air pipettes with intensities in the same range as the samples and standards. Line blanks were measured every 2–3 unknowns and were subtracted from succeeding sample data. Data reduction is done in ArArCalc (Koppers, 2002). Ages are calculated with decay constants of Min et al. (2000) and 28.201 Ma for FCs (Kuiper et al., 2008). The atmospheric ⁴⁰Ar $/$ ³⁶Ar air value of 298.56 is used (Lee et al., 2006). The correction factors for neutron interference reactions are (2.64 ± 0.02) × 10⁻⁴ for (³⁶Ar $/$ ³⁷Ar) Ca, (6.73 ± 0.04) × 10⁻⁴ for (³⁹Ar $/$ ³⁷Ar) Ca, (1.21 ± 0.003) × 10⁻² for (³⁸Ar $/$ ³⁹Ar) K and (8.6 ± 0.7) × 10⁻⁴ for (⁴⁰Ar $/$ ³⁹Ar) K. All errors are quoted at the 2σ level.

Geochemical data

Depending on individual sample properties, different preparation methods were applied to enrich volcanic glass fragments for geochemical analyses. These methods include classical preparation techniques, like crushing of lithified/solidified samples, rinsing, (wet-) sieving, magnetic and density separation, and hand picking. Specific information for each sample is listed in the sample description and is also given along with the analytical results within the data-files provided for download (e.g., results of geochemical analyses). Respective aliquot fractions enriched in glass fragments were mounted on epoxy pucks and polished to remove potential surficial alteration and to avoid topographic effects causing compositional variations during subsequent glass geochemical analyses.

The pucks were carbon coated for major and minor element analyses of glass by electron probe microanalysis (EPMA) wavelength dispersive spectroscopy (WDS) and/or scanning electron microscope energy dispersive spectroscopy (SEM-EDS). Specific methodological details for each sample are provided alongside the geochemical datasets in TephAta.

A JEOL JXA-8900RL electron microprobe equipped with five-wavelength dispersive spectrometers was used for these analyses at the University of Cologne (UoC). The operation conditions were set to 12 kV accelerating voltage, 6 nA beam current and 5 µm beam diameter. Full details of calibration and measuring conditions are given in Leicher (2021). Data reduction included ZAF correction and substitution of halogens (Cl, F) for oxygen using the JEOL Ltd JXA8900 Basic Software V3.02. Only EPMA-WDS geochemical analyses of glass fragments with analytical totals > 90 wt % were considered and normalized to 100 % on a loss on ignition free basis, excluding volatiles (Cl, SO₃ and F).

A Zeiss Sigma 300-VP equipped with an OXFORD Instruments EDX detector controlled by the software Aztec 4.1 was used for SEM-EDS analyses, at the UoC. The instrument was set to 20 keV and an image resolution of 512 pixel × 512 pixel was chosen. The number of channels was set to 2048, processing time to 5, with a dwell time of 10 000 µs. The size of the scanning frame was adjusted to the respective grain size and typically exceeded 10 µm.

Trace-element analyses of glass fragments were performed at the University of Bonn (UoB) and UoC on the same sample pucks as the EPMA-WDS/SEM-EDS analyses, after removing the carbon coating. At the UoB trace element analyses were performed using a Resonetics Resolution M50E 193 nm excimer laser ablation system coupled to a Thermo Scientific Element XR (Leicher and Lagos, 2021). A spot size of 26 µm was chosen and analyses were performed at a repetition rate of 5 Hz and a count time of 35 s on the sample after 30 s on the gas blank (background). A He gas flow (0.75 L min⁻¹), mixed together with Ar sample gas (∼ 1.1 L min⁻¹) transported ablated material via an in-house signal-smoothing device to the ICP-MS. Maximum intensity as well as stability of the signal were obtained by tuning while taking account of concurrently low oxide ratios ($\mathrm{ThO}/\mathrm{Th}$ of ∼ 0.0012) to minimize potentially interfering oxide species prior to analyses in low-resolution mode. At UoC an NWRimageGEO 193 nm ArF excimer laser ablation system coupled to a Thermo Scientific iCAP quadrupole or iCAP triple quadrupole ICP-MS was used for trace-element analyses. Individual glass fragments were analyzed using 15 or 20 µm spot size, a repetition rate of 6 Hz and a count time of 40 s on the sample after 30 s of the gas blank (background) measurements. The sample aerosol was taken up by a He gas flow (0.9 L min⁻¹) and mixed with Ar (0.8 L min⁻¹), within a glass smoothing device before introduction into the plasma.

All data reduction was performed using the software Iolite 4.3 (Paton et al., 2011). The standard glass NIST SRM 612 was used for calibration and calculation of trace element concentrations based on ⁴³Ca, ⁸⁵Rb, ⁸⁸Sr, ⁸⁹Y, ⁹⁰Zr, ⁹³Nb, ¹³⁸Ba, ¹³⁹La, ¹⁴⁰Ce, ¹⁴¹Pr, ¹⁴⁶Nd, ¹⁴⁷Sm, ¹⁵³Eu, ¹⁵⁷Gd, ¹⁵⁹Tb, ¹⁶³Dy, ¹⁶⁵Ho, ¹⁶⁶Er, ¹⁶⁹Tm^, 172Yb, ¹⁷⁵Lu, ¹⁷⁸Hf, ¹⁸¹Ta, ²⁰⁸Pb, ²³²Th, ²³⁸U. As an internal standard ²⁹Si was incorporated using median SiO₂ concentrations obtained from EPMA-WDS data of respective samples. Since major and minor elements such as Ti, Na, K, or P were not analyzed by LA-ICP-MS, they could not be considered for detection of potential contamination by mineral phases during ablation. However, during data reduction in Iolite, the full analyzed trace element spectrum and Ca were screened to identify and exclude potential mineral inclusions (e.g. relatively elevated Ca, Sr, or Ba counts indicating feldspar inclusions) from data selection.

The here presented geochemical data can be either downloaded as sample-specific datasets from the TephAta database (https://www.crc1211db.uni-koeln.de/tephata, last access: 29 May 2026) or can be found as a combined dataset of all samples deposited at EarthChem (Leicher, 2026).

Data quality of geochemical analyses

During EPMA-WDS, SEM-EDS and LA-ICP-MS analysis MPI-DING standard glasses (ATHO-G, StHs6/80-G; Jochum et al., 2006) and SRM NIST SRM 610 (Jochum et al., 2011) were used as secondary standards to evaluate the accuracy and precision of individual measurement sessions. Sample-specific results of secondary standard analysis are stored within TephAta along with the sample results and are provided for download within the sample-specific data-file.

Over the full analytical period, EPMA-WDS analyses conducted at UoC yielded mean intermediate precision (expressed as relative two standard deviation) and accuracy (expressed as bias relative to the preferred reference value) of up to 2.5 % and 0.6 %, respectively, for elemental concentrations > 75 wt %, up to 4.4 % and 0.6 % for 75 wt %–12 wt %, up to 13.33 % and 2.0 % for 12 wt %–1 wt % and up to 58.4 % and 6.4 % for < 1 wt %. Mean values for intermediate precision and accuracy of SEM-EDS analyses, are up to 0.6 % and 0.3 % for elemental concentrations > 75 wt %, up to 3.8 % and 0.3 % for 75 wt %–12 wt %, up to 18.0 % and 3.7 % for 12 wt %–1 wt % and up to 60.0 % and 7.9 % for < 1 wt %.

Secondary standard analyses of MPI-DING glasses ATHO-G and StHs6/80-G during LA-ICP-MS analyses at the UoB revealed median intermediate accuracies of < 5 % for Sm, Ho, Rb, Ba, Pr, Ca, Sr, Ce, Yb, Lu, Hf, and Er, 5 %–12 % for La, Dy, Nd, Tm, U, Tb, Th, Eu, Y, Zr, Gd, Nb and Ta and 19 % for Pb. The median intermediate precision of the secondary standard analyses at UoB was typically < 10 % for Sr, Zr, Ce, Nb, Rb, Ba, La, Ca, Pr, Y, and Th and 10 %–28 % for Nd, U, Hf, Er, Ho, Pb, Dy, Eu, Gd, Ta, Sm, Tb, Lu, Yb, and Tm. At the UoC, secondary standard analyses of MPI-DING glasses (ATHO-G, StHs6/80-G) glasses during LA-ICP-MS analyses revealed median accuracies of 5 %–10 % for Rb, Sr, Ba, Ca, Ce, La, Sm, Ho, and U, 10 %–15 % for Zr, Nd, Th, Hf, Y, Pr, Eu, Tb, Er, Yb, Nb, Dy Pb and 15 %–17 % for Gd, Lu, Tm and Ta. The median intermediate precision of the secondary standard analyses at UoC was typically < 10 % for Sr, Ba, Zr, Ce, La, Rb, Nb, Y, Ca, Pr, Nd, Th, Hf, Yb, U, Pb, Er, and Ta and of 10 %–16 % for Ho, Gd, Sm, Dy, Tb, Eu, Lu, and Tm.

4 Results and Discussion

4.1 Site and sample data

Currently, the TephAta dataset lists 106 tephra samples from 91 tephra deposits sampled at 74 sites between approximately 20–27° S to 70–67° W. The specific origin (locality, reference) of samples is listed in Table 1. The set of samples includes 65 samples taken before the CRC-research activities of which 23 samples originate from mapping campaigns of the SERNAGEOMIN and 25 samples from other regional geoscientific studies. In addition, 17 previously known sample sites were revisited and re-sampled as part of the CRC's field work during which 41 newly identified tephra samples were gathered and included here.

Tephra samples are sourced from all types of sediment deposits of the Atacama Desert including alluvial fans, river terraces as well as clay pan and halite deposits. In addition, a small set of samples originating from the proximal deposits associated with the Irruputuncu volcanic complex were included, because they represent a potential volcanic source. Sampling sites are mostly outcrops formed by the incision of channels, gravimetric movement, tectonic activity, and roadcuts, and also include pits and drill cores. Tephra deposits show a broad variety in their physical properties, which represents the variable influence of eruption parameters and (post-) depositional related processes. The typical thickness of tephra layers is in the scale of centimeters, but also include several m-thick to sub-cm thin tephra deposits, whose lateral extent ranges from decimeter to hectometer. In addition, cryptotephra horizons have been identified in sediment cores. Visible volcanic ash layers are most commonly of bright color. Sedimentological texture properties, such as the grain-size, are in the range of ash (< 2 mm diameter) for most tephra deposits, due to their distance to potential source volcanoes. Internal sedimentological structure of these tephra deposits, like grading, are only rarely observed due to their generally well-sorted grain-size distribution and are, if observed, commonly related to post-depositional weak water current transport processes. The mineralogical composition of tephra layers is dominated by glass, but also primary (biotite, feldspar, zircon) and secondary minerals (gypsum, halite) were observed. The level of details of the available morphological description of individual layers varies, as for some samples morphological and mineral compositional details are incomplete.

4.2 Chronology of samples

Biotites of sample PAG-T4 were ⁴⁰Ar $/$ ³⁹Ar dated in two analytical sessions (19T04: n=10; 1–3 grains/hole; 19T11: n=20; 10 grains/hole) yielding individual weighted mean ages of 318.8 ± 18.9 ka (n=7 ages) and 298.4 ± 7.4 ka (n=18 ages), with a combined weighted mean age of 301.3 ± 13.8 ka. The weighted mean age is obtained by including as many individual analyses with mean squared weighted deviation (MSWD) < T-test statistic at 95 % confidence level. The ⁴⁰Ar $/$ ³⁶Ar inverse isochron intercept is 297.9 ± 1.0 and overlaps with air values of Lee et al. (2006). The analytical errors in the first series are larger due to small beam intensities. The radiogenic ⁴⁰Ar yield is on average ∼ 4 %. Biotites of sample IRU-1a were analyzed during the same analytical sessions (19T04: n=10; 1–3 grains/hole; 19T11: n=20; 4–5 grains/hole) as samples of PAG-T4 and their weighted mean age computation followed the same criteria. The first series yielded a weighted mean age of 188.7 ± 12.1 ka (n=6 ages) and the second series 171.2 ± 12.6 ka (n=7 ages), which gives a combined weighted mean of 180.5 ± 8.7 ka. The ⁴⁰Ar $/$ ³⁶Ar inverse isochron intercept is 298.1 ± 0.6 and overlaps with air values of Lee et al. (2006). The radiogenic ⁴⁰Ar yield is very low with ∼ 2 % on average. In the second series, parts of the experiments were discarded because the ⁴⁰Ar signal was too high for the detector and runs were aborted.

In combination with the data available in literature, 42 direct tephra ages obtained by absolute dating methods and 8 indirect tephra ages obtained by luminescence or U-Th dating of the host sediment, are available for 47 of 106 samples (Table 1, 3 samples have multiple ages). Of the 42 direct ages, the majority derives from biotite ⁴⁰Ar $/$ ³⁹Ar dating (n=29), whereas the other ages were obtained by biotite K-Ar (n=10), feldspar K-Ar (n=1), feldspar ⁴⁰Ar $/$ ³⁹Ar (n=1) and zircon U-Pb (n=1) dating methods.

Figure 2 (a) Compilation of available chronological information for the samples of the Pleistocene being currently available within TephAta. For references of the respective ages see Table 1. (b) Result of combined chronological and geochemical information illustrating the ages available for compositional groups CG1, 2, 3, 9, and 12. The legend illustrates the different dating techniques used for the respective samples in (a) and (b).

The samples so far included in TephAta cover an age range from 22.90 to 0.06 Ma. Most samples (n=45) have an age < 1 Ma, but few samples (n=5) have Early Pleistocene to Miocene ages (Ritter et al., 2018; Medialdea et al., 2020). The ages of tephra layers > 1 Ma suggest several distinct eruptive events, whereof only the ages of two samples partly overlap due to their uncertainties of up to 21 % (CH22-NL-T1 and -T2, cf. Table 1). The ⁴⁰Ar $/$ ³⁹Ar sanidine of tephra layer SALAR T6 age (1.297 ± 0.018 Ma) is contradicted by a much younger IRSL host sediment age (< 0.1 Ma), which is discussed in detail below. Among the < 1 Ma old tephra layers, two main periods of explosive volcanic activity can be tentatively identified within the dataset (Fig. 2a). The older period includes tephra with ages of 0.9–0.5 Ma (n=7) and clustering between 0.75 and 0.60 Ma. The younger period includes samples with continuously overlapping ages between 0.43 and 0.06 Ma (n=36). One tephra deposit (PAG-T4) could not be assigned to one of these two periods, as different dating techniques provide very different ages with the new ⁴⁰Ar $/$ ³⁹Ar biotite age of 0.301 ± 0.007 Ma and an U-Pb zircon age of 0.98 ± 0.04 Ma (Ritter et al., 2018). In general the active periods were also identified based on a partly overlapping dataset of chronological data focusing on tephra layers of the Coastal Cordillera of 20–21° S (Sepúlveda et al., 2013; Quezada et al., 2018).

Using solely the chronological data within the TephAta dataset to disentangle and identify individual volcanic events < 1 Ma within the two periods of volcanic activity is challenging. The heterogeneous dating techniques applied within several decades (1978–2023) used a variety of instruments, protocols, internal standards, and respective reference values, which limit the accurate differentiation of similar ages. A homogenization, e.g., through recalculation of ages based on the same reference values for mineral standards and decay constants, to increase the comparability of the complete dataset is, however, hampered by scarcity of metadata and is further limited by the fact that different dating methods have been applied. Besides the uncertainties related to the different methodological approaches, also the sample-specific aspects (quantity and quality of minerals, type of target minerals, plateau vs isochrone ages) infer further ambiguity in comparing their ages. 20 samples have relatively large age uncertainties of more than > 15 %, which is especially evident within the group of K-Ar ages. Furthermore, due to uncertainties exceeding the respective dated ages, some (n=2) ages have to be classified as unreliable.

The ages of 7 samples assigned to the older period (0.5–0.9 Ma) overlap among each other within their uncertainties and do not allow a further differentiation. Among the younger sample group (< 0.5 Ma), relatively small uncertainties of a few ages (SMS-15d: 0.375 ± 0.017 Ma, GSQ-08d: 0.300 ± 0.020 Ma, GSQ-027d: 0.255 ± 0.017 Ma, TdSU: 0.191 ± 0.005 Ma and IRU-1a: 0.181 ± 0.009 Ma) indicate a potential chronological differentiation into multiple eruptive events. All other ages of tephra layers, however, widely overlap with the aforementioned more precise ages and among each other, which hampers the assignment of tephra to a specific eruptive event. The age differences between some samples are within a range of 1–2 ka (e.g., GSQ-158d: 0.361–0.319 Ma vs GSQ-08d: 0.320–0.280 Ma; Fig. 2a), which is too small for a clear differentiation between eruptive events. Moreover, also the accuracy of the (more precise) ages is difficult to assess as their ages are based on a single, multi-grain incremental heating experiment, from which a plateau age was calculated. This limits the detection of sources for potential biases such as xenocryst contamination, incorporation of altered minerals or minerals with a complex crystallization history (Hora et al., 2010; Kern et al., 2016). Such biases are suggested for some samples, whose ⁴⁰Ar $/$ ³⁶Ar values are above the uncertainties of the atmospheric air value and thus indicate excess Ar, so that only inverse isochrone ages could be computed (e.g., IA-11, SMS-15d). Considering the few available total fusion ages within the dataset (PAG-T4, IRU-1a), their individual ages show a natural scatter in ages and thus provide additional information to extrapolate more robust eruption ages, while potential biased ages (e.g. several age populations caused by xenocrysts) can be filtered. Among the here listed samples, only very few samples have been dated by different dating approaches to further test for potential biases. The available results indicate inconsistencies such as by the zircon U-Pb age for tephra PAG-T4 of 0.98 ± 0.04 Ma (Ritter et al., 2018), which is in conflict with the ⁴⁰Ar $/$ ³⁹Ar biotite age of 0.301 ± 0.06 Ma obtained from the same sample. The youngest tephra ages included in the dataset derive from a series of indirect infrared stimulated luminescence (IRSL) ages of the respective outcrop. The stratigraphic oldest tephra layer of this deposit (SALAR T6) has been also dated by sanidine ⁴⁰Ar $/$ ³⁹Ar (Medialdea et al., 2020) yielding a weighted mean age of 1.297 ± 0. 018 Ma is based on a robust population of 25 individual total fusion ages. However, this age is much older than the proposed IRSL maximum age < 0.098 ± 0.015 Ma and was explained by the authors as a potential reworking of the tephra layer (Medialdea et al., 2020). New tephrostratigraphic conclusions based on new geochemical signatures (discussed below), however, question this hypothesis. Further, the geochemical fingerprint of an overlying tephra (SALAR-T2) in the succession likely corresponds to a tephra dated at 0.75 ± 0.06 Ma, thus also indicating an older age of the sequence. A similar situation is observed for a tephra found in drill cores from Salar de Uyuni, where the directly obtained biotite ⁴⁰Ar $/$ ³⁹Ar age of the tephra is in conflict with a younger age derived from the U-series and ¹⁴C-based age model of the succession (Fornari et al., 2001; Fritz et al., 2004). Overall, the uncertainties and inconsistencies between the different dating approaches make a chronological differentiation between all available ages only of a tentative nature and suggest the need for additional criteria (e.g., other dating methods, geochemical fingerprinting) for a robust separation into different eruptive events.

4.3 Geochemistry

For 79 of the 91 sites being listed within TephAta, glass geochemical compositions of tephra layers are available, which were obtained from 82 of the 106 samples. Sixty-five of these samples were geochemically characterized for the first time and 15 samples had been characterized within the scope of previous projects within the CRC (May et al., 2020; Medialdea et al., 2020; Ritter et al., 2022). In addition, EPMA-WDS legacy data and new measurements are available for two samples (2/3/5/2, AD-01; Placzek et al., 2009; Breitkreuz et al., 2014), whereas TephAta only includes the new data. Comparing the legacy data with the respective new EPMA-WDS analyses reveals that the major and minor glass compositions of AD-01 are undistinguishably overlapping. However, the legacy data of sample 2/3/5/2 has significantly higher SiO₂ values (79.25 wt %–81.22 wt %) compared with its reanalysis data (74.40 wt %–77.93 wt % SiO₂), which results also in differences observed for the other element concentrations. A significant alkali loss and a SiO₂-enrichment were already noted in Breitkreuz et al. (2014), but the reason for such could not be explained. With regard to the high SiO₂ concentrations of the $\mathrm{2}/\mathrm{3}/\mathrm{5}/\mathrm{2}$ legacy data, similarly high values were not observed for other samples within TephAta. Since no secondary reference data are given in the original article, a technical issue cannot be excluded

Figure 3 Geochemical classification and differentiation based on major and minor element glass compositions of the investigated tephra layers. (a) TAS-classification of all samples according to Le Bas et al. (1986). (b, c) TAS-diagram extraction for differentiation between type I, II and III rhyolites according to their silica content. (d, g–i) x-y oxide plots indicating a differentiation of type I rhyolites according to differences in the CaO, TiO₂, FeOtotal compositions (wt %) and ${\mathrm{Na}}_{\mathrm{2}}\mathrm{O}/{\mathrm{K}}_{\mathrm{2}}\mathrm{O}$ alkali-ratio. (d, e) x-y oxide plots indicating internal differentiation of type II and III rhyolites according to differences in the CaO (wt %) composition of respective samples.

Major and minor element compositions of the samples allow a general classification of their composition based on the Total Alkali versus Silica diagram (TAS, Le Bas et al., 1986) suggesting that all samples are dominated by glass shards with rhyolitic compositions (Fig. 3). Five of these samples (SGL3, IRRU13, PAG17 ID4, ID14, ID24) have a more heterogeneous composition including also less evolved trachyandesitic, trachytic and/or dacitic shards. The samples of rhyolitic compositions exhibit silica concentrations which suggests a subdivision into three silica-types: type I (n=66) includes samples dominated by shards with silica concentrations between ∼ 76–79 SiO₂ wt %, type II (n=7) between ∼ 73–76 SiO₂ wt % and type III (n=9) represents a mix of both types having less homogenous rhyolitic compositions. This classification is also seen within the other major and minor element data of the samples investigated, as samples of the same silica type have very similar, partly overlapping major and minor oxide compositions (Fig. 3). The majority of the rhyolitic type I samples have a similar major and minor element geochemical composition and cannot be further separated into different geochemical clusters. Some samples of type I, however, show specific variations in TiO₂, FeO_(TOT), CaO, K₂O and/or Na₂O concentrations, which suggest a division into several clusters (Fig. 3b, d, and g–i). Tephra samples of type II and type III cannot be further separated into compositional clusters based on their major element glass geochemistry, except for individual differences observed within the CaO content (Fig. 3d and e; e.g., SALAR T4, SG17/002, TJ09-PdT CH18-T3, CH18-T5).

Due to their generally more incompatible behaviour and lower concentrations of trace elements compared to major and minor elements within the melt, trace elements effectively amplify compositional variations driven by differences in the parental magma source and magmatic history (e.g. fractional crystallization, degree and depth of melting, or residual mineral assemblages). Thus, trace element geochemistry of glass shards have been shown to be a reliable parameter for distinguishing between eruptions with similar major element composition (Tomlinson et al., 2012; Pearce, 2014; Hopkins et al., 2021).

Figure 4 Geochemical classification and differentiation based on trace element glass compositions of investigated tephra layers. (a–c) Spidergrams demonstrating different degrees of enrichment/depletion of REE of the different composition groups. For each tephra sample the mean is given. Samples of the same composition group share the same color and are sorted according to their assignment to rhyolites I–III. Trace-element concentrations are normalized to values of the primitive mantle (pyrolite; Mcdonough and Sun, 1995). (d) x-y plot for trace element ratios $\mathrm{La}/\mathrm{Th}$ vs. $\mathrm{Ba}/\mathrm{Nb}$ to demonstrate the clustering of the different individual samples. (e–h) High-low plot of selected trace element ratios illustrating the differences between the different CG1-16. For each CG, all individual sample data are plotted as a stack.

Table 2 Overview of samples and their classification based on the Total Alkali vs. Silica composition (TAS; Le Bas et al., 1986), their assignment into rhyolites types I–III, and trace element compositional groups (TE-CG). If no trace element compositions were available, samples are listed at the end of each respective rhyolitic type. Samples which have not been geochemically investigated are listed at the bottom of the table.

For better differentiation among the identified geochemical clusters and to identify potential additional compositional differences, trace element data of 59 samples were considered. The individual samples show a mostly homogenous composition or indicate geochemical trends supporting that samples represent primary deposits and not a mixture of different eruptions. Trace-element concentrations normalized to values of the primitive mantle (pyrolite; Mcdonough and Sun, 1995) reveal for all samples (Fig. 4a–c) a relative enrichment of large-ion lithophile elements (LILE: Rb, Ba, Pb) against high field strength elements (HFSE: Ta, Nb, Hf, Zr), and among the latter an enrichment of light REE (LREE) compared to heavy REE (HREE), which is a pattern typical for the geodynamic continental arc setting. Combining the major, minor and trace element data enables the definition of 16 clusters, which are shortly discussed below and are listed in Table 2.

Among the 52 individual trace-element data sets of type I rhyolites, a differentiation into 11 different clusters is suggested due to distinct differences in their trace-element compositions (Fig. 4). Among them, 35 samples have a homogenous overlapping composition and are grouped in compositional group (CG) 1. CG2 comprises four samples of which three are compositionally identical (CG2.1), whereas one sample shows a generally similar pattern, although with differences in composition (CG2.2). All samples can be clearly distinguished from the other CGs by their lower HFSE and REE concentrations (Fig. 4a–c). Compositional groups 3, 4, 5, 6, 7, and 8 include one sample each, which all differ among each other (Fig. 4a–c and f–i). CG3 tephra has low Th concentrations (5–10 ppm) and a certain enrichment in HFSE (Y, Nb) and HREE compared with the other type I rhyolites, which is also observed for CG4. A characteristic depletion in Sr and Ba accompanied by elevated HFSE (except for Zr, Hf) and HREE separate CG5 from the other clusters. CG6 has slightly enriched Ba concentrations as well as HFSE and LREE, if compared to CG1. CG7 and CG8 both have a wider range of composition and have slightly lower LILE, but higher HFSE and (H+L)REE concentrations. CG9 includes four different tephra samples with a common composition, having generally higher Th concentrations (25–40 ppm, compared to 15–26 ppm of CG1), and further differ e.g. from CG1 by higher HFSE (Y, Nb, U) and LREE (La, Ce, Pr, Nd). CG10 has, relative to other rhyolite I tephra, lower Sr and Ba, but elevated HFSE (Y, Nb, Ta, U) and HREE concentrations, similar to tephra composition of CG11.

The individual trace element compositions of type II rhyolites (n=3) overlap, such as their major-element compositions, and thus are grouped as CG12. Their trace element compositions clearly differ from those of the other rhyolites (type I + III) as seen e.g., in lower LILE (expect for Ba, Sr) and higher HREE. Trace-element compositions of type III rhyolites (n=4) are heterogeneous among each other and allow a differentiation into four individual clusters (CG13, 14, 15, 16). CG13 has low HFSE and REE concentrations relative to other type III rhyolites. CG14 and CG15 are slightly similar and partially overlap with CG12. However, they clearly differ in lower HFSE (Zr, Hf), higher U concentrations, lower HFSE (Zr, U), and higher Ba concentrations.

Tephrochronological implications

The geochemical fingerprinting approach of tephra layers from the Atacama Desert enables the identification and characterization of distinct volcanic events, which promotes the investigation of inter-site tephra correlations and their spatial dispersal. A first assessment of the combined major, minor, and trace-element compositions of the tephra layers, allows a definition of 16 different geochemical clusters, which likely represent at least the same number of volcanic events. Some tephra layers were found in different sedimentary archives (e.g., tephra layers of CG1, CG2, CG9). The widespread dispersal of these tephra layers points to major eruptive events, which have the potential to be marker horizons within the sedimentary archives of the Atacama Desert. Even though for several tephra layers no correlations with other tephra deposits could be identified, their heterogenous geochemical clustering highlights their potential to contribute as geochemically distinct markers within the developing stratigraphic framework.

Although many of the investigated tephra layers have already been dated, the available chronological data often exhibit large uncertainties making it difficult to distinguish between one or multiple volcanic events. These uncertainties can be reduced by incorporating geochemical clustering results, which help to verify or challenge the proposed age distinctions and identify equivalent tephra layers derived from the same volcanic source. For 7 of the 16 geochemically defined clusters, chronological information is available and allows the assessment of 39 individual ages. Accordingly, at least 9 volcanic events have been identified for which no chronological information is available at present. As samples with ages > 1 Ma have not been in the main focus of this exemplary dataset, an in-depth tephrostratigraphic evaluation of these older events is not possible to date. Overall, the available chronological and geochemical data are a valuable basis for further exploring the explosive volcanic history of the region between 22 and 1 Ma. First implications are that the geochemical fingerprinting of samples < 1 Ma supports the identified bifurcation of active periods within the geochronological data and further allows characterization of individual events within the two activity periods (0.06–0.43 and 0.5–0.9 Ma). For the samples > 0.5 Ma, their assignment to different geochemical clusters (CG9 + 12) indicates at least two different eruptive events. Although their age uncertainties overlap, the combined age limits of correlated samples suggest two eruptive events at ca. 0.75 Ma (CG9) and ca. 0.6 Ma (CG12). For the remaining samples, the clustering of their geochemical data implies three potential volcanic events.

Samples belonging to CG1 (n=35) appear to be the most widespread cluster identified so far, but their available ages do not allow a distinct age constraint at present. Ages of CG1 include 14 ages with an age between 0.4–0.1 Ma, but also three ages exceeding 0.5 Ma (Fig. 2b). Their geochemically indistinguishable compositions imply a single volcanic eruption, but some samples with more precise ages within this group tentatively suggest multiple eruptions (PAG-T4, SMS-15d, GSQ-08d, GSQ-027d). Thus, it cannot be fully excluded that CG1 represents a group of repeating eruptions with identical geochemical composition that occurred in short succession. Regarding the other two identified eruptive events, no precise age can be given at present. For the older one (CG2), two differing ages of 0.191 ± 0.005 Ma (L5/TdSU) and 0.280 ± 0.014 Ma (ATA19-023/CR-006A) exist, whereas for the younger (CG3) only an imprecise age of 0.140 ± 0.057 Ma (GSV-132d) is available.

In some archives, tephrochronology also provides an independent validation of direct dating of host sediment. A tephra layer (SALAR T2) of the SALAR GRANDE hillslope section (Medialdea et al., 2020) now could be correlated with a tephra layer dated in other archives, introducing an additional age of 0.75 ± 0.06 Ma for the succession. This age, however, is in conflict with the much younger IRSL ages from the section (< 0.1 Ma), but rather supports the direct age obtained for the lowermost tephra of the succession (SALAR T6: 1.297 ± 0. 018 Ma; Medialdea et al., 2020). The distinct geochemical compositions of all tephra layers in the succession associated with different geochemical clusters supports a primary deposition of those layers and rather contradicts a reworking of the dated sanidine crystals form older material (Medialdea et al., 2020). In a sediment core recovered from Salar de Uyuni, the biotite ⁴⁰Ar $/$ ³⁹Ar age of a tephra layer (0.191 ± 0.005 Ma; Fornari et al., 2001), is questioned by a much younger, indirect U-Th age of the host sediment (0.059 Ma; Fritz et al., 2004). This tephra layer could now be geochemically correlated with two other pyroclastic deposits being located westward of Salar de Uyuni (POR-3, ATA19-023) and of which the age of ATA19-023 (resampled CR-006A: 0.280 ± 0.014 Ma; Gardeweg and Sellés, 2013) is supporting also an older age. However, the large difference in their ages highlights the need for refining their chronology.

5 Data availability

Table 3 lists all samples included within this manuscript and provides their IGSN number and a link to download the individual geochemical data, which is stored within TephAta: https://www.crc1211db.uni-koeln.de/tephata, last access: 29 May 2026 (CRC1211DB-TephAta, 2026). In addition, the full dataset (meta and chronological data overview and geochemical data including individual major, minor and trace element glass compositions) is also made available for download at the EarthChem repository: https://doi.org/10.60520/IEDA/114209 (Leicher et al., 2026).

Table 3 Overview of samples included within the TephAta database. The sample name is given along with their IGSN and the information about which kind of data is available.

6 Conclusions

TephAta represents a new platform to support the development of a comprehensive tephrochronological framework for the Atacama Desert in northern Chile. Its structure is strongly aligned with scientific tephra guidelines (Wallace et al., 2022) providing a centralized online repository for tephra related data. As the first platform of its kind to integrate geochemical, geochronological, and detailed outcrop metadata, TephAta preserves both new and legacy datasets to maximize data reusability. Due to the automatic implementation of IGSNs, samples become citable and have a unique identification, facilitating interoperability with global repositories. At present the dataset contains information from 106 tephra samples. For samples listed with an incomplete or missing geochemical characterization, the pending major, minor, and/or trace element analyses data will be complemented within the successive analytical work of the CRC1211. Future expansions will progressively incorporate over 400 additional samples from SERNAGEOMIN and current CRC activities. This set of samples will cover a time interval from the Early Miocene to the Holocene and has a spatial extent from the Chilean coast to the Andes along 18–30° S.

Based on the initial datasets compiled in TephAta, several data-specific insights emerge. The integrity of sample and site metadata reveals that field notes and laboratory documentation currently vary widely in detail and often lack key information, likely because many samples were not originally collected with tephrostratigraphic objectives in mind. The standardized structure provided by TephAta, in combination with dedicated field-support tools (such as sample-report templates, and app-based logging platforms like StraboSpot) will streamline and strengthen future documentation and data archiving. This improved framework will also support consistent public reporting of analytical information that aligns with bestpractice metadata standards established by the global tephra research community.

In this study, most tephra layers have rhyolitic glass compositions. While major- and minor-element glass data compositions distinguish only a subset of tephra layers, trace-element glass compositions provide the most robust means of differentiation. Based on these geochemical differences, at least 16 distinct eruptive events have been identified, allowing for potential regional correlations. Despite these advances, the uncertainties in existing geochronological data require careful reassessment. Existing ages should be supported by additional, higher precision dating, before used as tephrochronological tie points. Implementing methods such as zircon double-dating (Danišík et al., 2017), will help to account for pre-eruptive mineral histories and improve discrimination among tephra layers of similar age and geochemical composition. Furthermore, obtaining new chronological data for the numerous identified eruptive events (CG1, 4–8, 10, 11, 13–16), is essential to strengthening the tephrochronological framework and overcoming the inherent dating challenges of arid environments.

Overall, this initial dataset demonstrates the value and potential of a tephrochronological database for the region. It also highlights that establishing a robust regional tephrochronological framework requires a large set of samples with adequate spatial and temporal coverage. This includes the tephrochronological characterization of proximal major eruptive products to link widespread distal tephra layers with their volcanic sources. As the database grows, it will enhance our understanding of the Andean volcanic history, particularly in a region where distal archives play a key role in unraveling eruptive sequences, as demonstrated elsewhere (Schindlbeck et al., 2018; Leicher et al., 2023; Vineberg et al., 2024). In addition, the individual geochemical data for the tephra samples offer valuable opportunities to investigate magmatic processes, such as the varying degrees of element depletion and enrichment of characteristic elements observed in this dataset.