Introduction
Fifteen lanthanides, along with Sc and Y, make up a group of rare-earth elements (REE). These lithophile elements are predominantly trivalent, electropositive, and possess refractory properties (McLennan, Reference McLennan1994; Taylor and McLennan, Reference Taylor and McLennan1995). The orbital configuration of REE gives rise to their remarkably similar physical and chemical properties where subtle differences in chemical behavior arise from their systematic variations in ionic radii and complexations, as well as the distinct redox states of Ce and Eu (Dubinin, Reference Dubinin2004; Willis and Johannesson, Reference Willis and Johannesson2011; McLennan and Ross Taylor, Reference McLennan and Ross Taylor2012). From an application perspective, REE play a key role in the advancement of new technologies and the transition to a green, low-carbon economy (Binnemans et al., Reference Binnemans, Jones, Blanpain, Van Gerven, Yang, Walton and Buchert2013; Balaram, Reference Balaram2019). As these technologies gain broader acceptance, the demand for REE is expected to rise (Alonso et al., Reference Alonso, Sherman, Wallington, Everson, Field, Roth and Kirchain2012; Gismondi et al., Reference Gismondi, Kuzmin, Unsworth, Rangan, Khalid and Saha2022). Contrary to their name, REE are not truly rare. They are, on average, as abundant as Cu or Ni in the Earth’s crust (Sekine, Reference Sekine1963; Ronov and Yaroshevsky, Reference Ronov and Yaroshevsky1969). However, unlike the two metals, REE are not found in concentrated deposits, making their extraction more challenging (Gupta and Krishnamurthy, Reference Gupta and Krishnamurthy1992; Humphries, Reference Humphries2010).
Contrary to high-temperature geologic environments, the comportment of REE in low-grade regimes is still poorly understood (Cornu et al., Reference Cornu, Deschatrettes, Salvador-Blanes, Clozel, Hardy, Branchut and Le Forestier2005; Laveuf and Cornu, Reference Laveuf and Cornu2009; Zhang et al., Reference Zhang, Gao and Arthur Chen2014; Badurina and Šegvić, Reference Badurina and Šegvić2022). Their solubility, however, is found to increase generally during weathering and diagenesis (Burkov and Podporina, Reference Burkov and Podporina1967; Elliott, Reference Elliott2020). The complexing ability of REE by inorganic and organic ligands at near-surface aqueous conditions tends to increase from light to heavy lanthanides (Topp, Reference Topp1965), aligning with the trend of their solubility increasing as ionic radii decrease (Martin et al., Reference Martin, Høgdahl and Philippot1976). This relates to the linear relationship between the relative ionic potential of REE and their enthalpy of hydration (Dileep Kumar, Reference Dileep Kumar1984). Once released from their primary source, REE precipitate readily in secondary phases (e.g. decrespignyite-(Y), paratooite-(La), Mn minerals; Brugger et al., Reference Brugger, Ogierman, Pring, Waldron and Kolitsch2006; Berti et al., Reference Berti, Slowey, Yancey and Deng2022; Berti et al., Reference Berti, Slowey, Deng, Yancey and Velazquez2023) or get ion-sorbed by clay minerals, such as kaolinite (Cheshire et al., Reference Cheshire, Bish, Cahill, Kertesz and Stack2018; Ercan et al., Reference Ercan, Ece, Çiftçi and Aydın2022; Ou et al., Reference Ou, Chen, Chen, Li, Wang, Ren, Chen, Feng, Wang, Chen, Liang and Gao2022), halloysite (Li and Zhou, Reference Li and Zhou2020), and various 2:1 clay minerals (Liu et al., Reference Liu, Zhou, Chi, Feng, Ding and Liu2019; Borst et al., Reference Borst, Smith, Finch, Estrade, Villanova-de-Benavent, Nason, Marquis, Horsburgh, Goodenough, Xu, Kynický and Geraki2020; Li and Zhou, Reference Li and Zhou2020; Luo et al., Reference Luo, Liang, Yang, Gao, Chen and Mo2023; Wu et al., Reference Wu, Chen, Wang, Xu, Lin, Liang and Cheng2023). Although the specific nature of the chemical bonds between REE and the clay structure is still unknown, it has been observed that REE tend to be adsorbed weakly on broken edge sites of clay minerals and charged aluminol or siloxane surfaces through the processes of ion exchange and surface complexation. These mechanisms may result ultimately in REE fractionation and enrichment (Hao et al., Reference Hao, Flynn, Kashiwabara, Alam, Bandara, Swaren, Robbins, Alessi and Konhauser2019; Zhou et al., Reference Zhou, Li, Wang, Li and Liu2020; Schroeder et al., Reference Schroeder, Austin, Thompson and Richter2022; Wu et al., Reference Wu, Chen, Wang, Xu, Lin, Liang and Cheng2023; Zhao et al., Reference Zhao, Xu and Hao2023). Recent research has also reported that certain bacteria found in the diagenetic environment may secrete low molecular weight organic compounds known as metallophores. These compounds exhibit high selectivity and affinity for lanthanides (Zytnick et al., Reference Zytnick, Gutenthaler-Tietze, Aron, Reitz, Phi, Good, Petras, Daumann and Martinez-Gomez2023). The lanthanophore may therefore be used for recovery of REE (Pol et al., Reference Pol, Barends, Dietl, Khadem, Eygensteyn, Jetten and Op den Camp2014; Daumann et al., Reference Daumann, Pol, den Camp, Martinez-Gomez, Poole and Kelly2022); however, the relationship between the REE-bearing bacteria and clay minerals remains largely unknown (Neubauer et al., Reference Neubauer, Nowack, Furrer and Schulin2000).
The Carpathian-Pannonian Region (CPR) of central Europe experienced a significant extension of the continental lithosphere during the Miocene, which led to the most intense volcanic activity in Europe at the time (Szabó et al., Reference Szabó, Harangi and Csontos1992; Németh et al., Reference Németh, Martin and Harangi2001; Roşu et al., Reference Roşu, Seghedi, Downes, Alderton, Szakács, Pécskay, Panaiotu, Panaiotu and Nedelcu2004; Kovács et al., Reference Kovács, Csontos, Szabó, Bali, Falus, Benedek, Zajacz, Beccaluva, Bianchini and Wilson2007; Harangi et al., Reference Harangi, Lukács, Schmitt, Dunkl, Molnár, Kiss, Seghedi, Novothny and Molnár2015). As a result, large amounts of felsic to intermediate tephra covered areas of the Southwestern Pannonian Basin (SPB) (Mandic et al., Reference Mandic, de Leeuw, Bulić, Kuiper, Krijgsman and Jurišić-Polšak2012; Rybár et al., Reference Rybár, Šarinová, Sant, Kuiper, Kováčová, Vojtko, Reiser, Fordinál, Teodoridis, Nováková and Vlček2019; Gverić et al., Reference Gverić, Hanžel, Kampić, Pleša and Tibljaš2020; Marković et al., Reference Marković, Kuiper, Ćorić, Hajek-Tadesse, Kučenjak, Bakrač, Pezelj and Kovačić2021; Grizelj et al., Reference Grizelj, Milošević, Miknić, Hajek-Tadesse, Bakrač, Galović, Badurina, Kurečić, Wacha, Šegvić, Matošević, Čaić-Janković and Avanić2023; Šegvić et al., Reference Šegvić, Lukács, Mandic, Strauss, Badurina, Guillong and Harzhauser2023a) and adjacent Dinarides intramontane basins (DIB) (Krstić et al., Reference Krstić, Dumurdzanov, Jankonić-Golubović, Vujnović and Olujić2001; Mandic et al., Reference Mandic, Pavelić, Harzhauser, Zupanič, Reischenbacher, Sachsenhofer, Tadej and Vranjković2009; Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021) (Fig. 1). There, with prolonged weathering, the pyroclastic material altered into clay-rich assemblages (Grizelj et al., Reference Grizelj, Milošević, Miknić, Hajek-Tadesse, Bakrač, Galović, Badurina, Kurečić, Wacha, Šegvić, Matošević, Čaić-Janković and Avanić2023; Šegvić et al., Reference Šegvić, Lukács, Mandic, Strauss, Badurina, Guillong and Harzhauser2023a). To assess the potential for REE retention in argillaceous tuffs, understanding the post-emplacement trajectories of trace elements in different geological environments is crucial. This issue has been addressed in various studies through the analysis of fresh glass and its alteration products which were sampled from diverse locations (Christidis, Reference Christidis1998; McHenry, Reference McHenry2009; Kiipli et al., Reference Kiipli, Hints, Kallaste, Verš and Voolma2017). This required considerable amounts of analyzed material susceptible to potential contamination from allochthonous detritus. In this research, however, the novel method which examines the spatially associated fresh volcanic material and associated clay minerals to reconstruct REE mobility in a low-grade diagenetic environment was utilized (Badurina and Šegvić, Reference Badurina and Šegvić2022). Put differently, this study relies on comparing the REE chemistry collected by in situ laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) of glass shards with that of genetically related illite-smectite gathered in the form of fraction separates. This study examines nine altered Miocene tuffs from two SPB and four DIB basins which are marked by diverse basinal stratigraphy (Fig. 1; Table 1). The aim of this contribution is to bring further insights to the following research questions: (i) to what extent does the morphology, geochemistry, and mineralogy of clay alteration products impact REE mobility?, (ii) how does the depositional environment conditions of airborne distal tephra influence the REE fractionation and retention during weathering?, and (iii) could eogenetic clays formed on a tuffaceous substrate serve as a potential regolith-hosted REE deposit?
Geological background
The Alpine-Carpathian-Dinarides orogenic system and the Pannonian Basin System (PBS) are two of the most important geological features in central and southeastern Europe. The PBS is a wide back-arc extensional basin, floored by a thin continental lithosphere, which was formed as a result of the northward drift and the indentation of the Adriatic promontory (Roeder and Bachmann, Reference Roeder, Bachmann, Ziegler and Horváth1996; Schmid et al., Reference Schmid, Fügenschuh, Kissling and Schuster2004; Schmid et al., Reference Schmid, Bernoulli, Fügenschuh, Matenco, Schefer, Schuster, Tischler and Ustaszewski2008), the rotation of the Tiszia terrane (Balla, Reference Balla1986; Csontos et al., Reference Csontos, Nagymarosy, Horváth and Kovác1992; Bada and Horváth, Reference Bada and Horváth2001; Csontos and Vörös, Reference Csontos and Vörös2004) and their subsequent collision with the European margin (Schmid et al., Reference Schmid, Bernoulli, Fügenschuh, Matenco, Schefer, Schuster, Tischler and Ustaszewski2008; Schmid et al., Reference Schmid, Fügenschuh, Kounov, Maţenco, Nievergelt, Oberhänsli, Pleuger, Schefer, Schuster, Tomljenović, Ustaszewski and van Hinsbergen2020). Ultimately, the formation of PBS was caused by the eastward extension of an Alpine orogenic wedge and Tiszia toward the eastern Carpathians. This was a consequence of the subduction roll-back of the lithosphere of the Carpathian flysch basin (Horváth, Reference Horváth1995; Bada and Horváth, Reference Bada and Horváth2001; Horváth et al., Reference Horváth, Bada, Szafián, Tari, Ádám and Cloetingh2006; Šujan et al., Reference Šujan, Rybár, Kováč, Bielik, Majcin, Minár, Plašienka, Nováková and Kotulová2021). During the Early and Middle Miocene, the climate in the southern PBS area experienced warming, accompanied by contrasting humidity conditions along the Dinarides transection: humid toward the sea and arid inland (Andrić-Tomašević et al., Reference Andrić-Tomašević, Simić, Mandic, Životić, Suárez and García-Romero2021; Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022). Additionally, the synrift in the Southern Pannonian Basin and the related extension in the Dinarides reactivated reversed Paleogene faults, providing necessary depositional space (de Leeuw et al., Reference de Leeuw, Mandic, Krijgsman, Kuiper and Hrvatović2012; Pavelić and Kovačić, Reference Pavelić and Kovačić2018; van Unen et al., Reference van Unen, Matenco, Nader, Darnault, Mandic and Demir2019). Furthermore, the Middle Miocene time saw an initial transgression and subsequent sea level change of the Paratethys Sea in the southern Pannonian Basin (Mandic et al., Reference Mandic, Rundić, Ćorić, Pezelj, Theobalt, Sant and Krijgsman2019a; Mandic et al., Reference Mandic, Sant, Kallanxhi, Ćorić, Theobalt, Grunert, de Leeuw and Krijgsman2019b; Mandic et al., Reference Mandic, Hajek-Tadesse, Bakrač, Reichenbacher, Grizelj and Miknić2019c). These climatic, tectonic, and eustatic changes resulted in diverse depositional environments ranging from lacustrine to marine settings, creating a dynamic geological landscape (Pavelić et al., Reference Pavelić, Kovačić, Miknić, Avanić, Vrsaljko, Bakrac, Tisljar, Galovic and Bortek2003; Piller et al., Reference Piller, Harzhauser and Mandic2007; Bakrač et al., Reference Bakrač, Hajek-Tadesse, Miknić, Grizelj, Hećimović and Kovačić2010; Holbourn et al., Reference Holbourn, Kuhnt, Kochhann, Andersen and Meier2015; Pavelić and Kovačić, Reference Pavelić and Kovačić2018; Mandic et al., Reference Mandic, Rundić, Ćorić, Pezelj, Theobalt, Sant and Krijgsman2019a; Mandic et al., Reference Mandic, Sant, Kallanxhi, Ćorić, Theobalt, Grunert, de Leeuw and Krijgsman2019b; Grizelj et al., Reference Grizelj, Milošević, Bakrač, Galović, Kurečić, Hajek-Tadesse, Avanić, Miknić, Horvat, Janković and Matošević2020; Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022; Hajek-Tadesse et al., Reference Hajek-Tadesse, Wacha, Horvat, Galović, Bakrač, Grizelj, Mandic and Reichenbacher2023; Matošević et al., Reference Matošević, Marković, Bigunac, Suica, Krizmanić, Perkovic, Kovačić and Pavelić2023). The tectonic evolution of PBS was characterized by extensive volcanic activity, which produced large amounts of felsic or intermediate pyroclastic material (Harangi and Lenkey, Reference Harangi and Lenkey2007; Seghedi and Downes, Reference Seghedi and Downes2011; Lukács et al., Reference Lukács, Harangi, Guillong, Bachmann, Fodor, Buret, Dunkl, Sliwinski, von Quadt, Peytcheva and Zimmerer2018) now found in the form of intercalated tephra layers within the PBS clastic strata (Pavelić and Kovačić, Reference Pavelić and Kovačić2018; Gverić et al., Reference Gverić, Hanžel, Kampić, Pleša and Tibljaš2020; Grizelj et al., Reference Grizelj, Milošević, Miknić, Hajek-Tadesse, Bakrač, Galović, Badurina, Kurečić, Wacha, Šegvić, Matošević, Čaić-Janković and Avanić2023; Šegvić et al., Reference Šegvić, Lukács, Mandic, Strauss, Badurina, Guillong and Harzhauser2023a).
The Dinarides are part of the Alpine-Himalayan Orogeny, which experienced a significant uplift during the Middle Eocene and Early Oligocene (Pamić et al., Reference Pamić, Gušić and Jelaska1998; Tari, Reference Tari2002; Pamić and Hrvatović Reference Pamić and Hrvatović2003). Following this orogenic phase, extensional tectonics prevailed in the Late Oligocene as a result of the rifting in the Pannonian Basin (Matenco and Radivojević, Reference Matenco and Radivojević2012; Andrić et al., Reference Andrić, Sant, Matenco, Mandic, Tomljenović, Pavelić, Hrvatović, Demir and Ooms2017; Stojadinovic et al., Reference Stojadinovic, Matenco, Andriessen, Toljić, Rundić and Ducea2017). This led to the formation of numerous NW-SE-trending Dinaride Intramontane Basins (DIB; de Leeuw et al., Reference de Leeuw, Mandic, Krijgsman, Kuiper and Hrvatović2012; van Unen et al., Reference van Unen, Matenco, Nader, Darnault, Mandic and Demir2019). The Miocene climatic optimum, characterized by a phase of persistent humidity, probably contributed to the development of stable, long-lived lake conditions in the Intramountainous Basins of the Dinarides (Zachos et al., Reference Zachos, Pagani, Sloan, Thomas and Billups2001; Jiménez-Moreno et al., Reference Jiménez-Moreno, Mandic, Harzhauser, Pavelić and Vranjković2008). The DIB stratigraphic record, similar to that of the SPB, bears evidence of Miocene volcanism in the Pannonian Basin Region, manifested through numerous tuff layers documented throughout DIB (Mandic et al., Reference Mandic, Pavelić, Harzhauser, Zupanič, Reischenbacher, Sachsenhofer, Tadej and Vranjković2009; Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021).
The SPB and DIB distal tephra layers, depending on prevailing depositional conditions, are normally found altered into secondary hydrous minerals like clay minerals and zeolite (Gverić et al., Reference Gverić, Hanžel, Kampić, Pleša and Tibljaš2020; Andrić-Tomašević et al., Reference Andrić-Tomašević, Simić, Mandic, Životić, Suárez and García-Romero2021; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021; Simić et al., Reference Simić, Životić and Miladinović2021; Šegvić et al., Reference Šegvić, Lukács, Mandic, Strauss, Badurina, Guillong and Harzhauser2023a). The formation of clay minerals from pyroclastic substrate typically requires prolonged periods of warm and hydrolyzing conditions (McHenry, Reference McHenry2009; Cunningham et al., Reference Cunningham, Lowe, Wyatt, Moon and Churchman2016; Hong et al., Reference Hong, Fang, Wang, Churchman, Zhao, Gong and Yin2017) and is influenced by paleoclimatic and paleoenvironmental constraints (Huff et al., Reference Huff, Bergström, Kolata and Sun1998; Christidis and Huff, Reference Christidis and Huff2009; Huff, Reference Huff2016). The SPB and DIB tuffaceous clays chosen for this study (Table 1) reflect long and stable weathering periods and typically consist of illite, smectite, and mixed-layer illite-smectite (Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014; Gverić et al., Reference Gverić, Hanžel, Kampić, Pleša and Tibljaš2020; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021).
Materials and methods
Materials
Nine samples of altered tuff were selected for this study. They originate from eight SPB and DIB locations stretching from the North Croatian Basin (the area of Mt Papuk) of Eastern Croatia through the Tuzla, the Kamengrad, the Livno and the Gacko Basins of northern, western, and south Bosnia and Herzegovina, to the Sinj Basin in Dalmatia, southern Croatia (Fig. 1). The petrographic description, thickness, published age, depositional setting, and GPS coordinates are provided for each analyzed sample (Table 1).
The tuff from the North Croatian Basin was acquired from the section exposed at the Poljanska quarry of Mt Papuk (Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022). The Miocene depositional environment at the time of tuff deposition corresponded to a hydrologically closed salina lake (Šćavničar et al., Reference Šćavničar, Krkalo, Šćavničar, Halle and Tibljaš1983; Pavelić and Kovačić, Reference Pavelić and Kovačić1999; Mandic et al., Reference Mandic, de Leeuw, Bulić, Kuiper, Krijgsman and Jurišić-Polšak2012; Pavelić and Kovačić, Reference Pavelić and Kovačić2018). The felsic rhyolitic tuff at the base of the section is likely to be of Lower (to Middle) Miocene age and corresponds to similar occurrences found in Austria (Roetzel et al., Reference Roetzel, de Leeuw, Mandic, Marton, Nehyba, Kuiper, Scholger and Wimmer-Frey2014), Hungary (Pálfy et al., Reference Pálfy, Mundil, Renne, Bernor, Kordos and Gasparik2007), and Slovakia (Šarinová et al., Reference Šarinová, Hudáčková, Rybár, Jamrich, Jourdan, Frew, Mayers, Ruman, Subová and Sliva2021). The Poljanska quarry tuff consists of multiple beds, varying from 5 to 30 cm in thickness, intercalated with marls and dolomite (Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022).
Further to the south is the Tuzla Basin where the tuff was sampled at the Čaklovići locality about 7 km SSE of the city of Tuzla (Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021). The lithified dark grey tuff exposure there is ~3 m thick and forms the base of ~30 m thick Middle Miocene costal and deltaic lacustrine sediment overlain by the Badenian marine shallow-water massive marl (Ćorić et al., Reference Ćorić, Vrabac, Đulović and Babajić2018).
To the west, two tuff samples were taken from the Kamengrad Basin the deposits of which may be correlated with the main evolutionary phase (18–15 Ma) of the Dinarides Lake System (Kochansky-Devidé and Slišković, Reference Kochansky-Devidé and Slišković1978; de Leeuw et al., Reference de Leeuw, Mandic, Krijgsman, Kuiper and Hrvatović2012). The basinal infill is 2 km thick, with coal layers dominating the lower portion and limestone and marl in the upper portion (Sunarić et al., Reference Sunarić, Glišić, Dangić and Milivojević1976). Both tuff samples were collected along the road from the closed Gornji Kamengrad coal mine to Okreč and are part of the limestone and marl dominated succession. The first tuff (2–40) was taken from a 60 cm thick light gray layer distinguished by the presence of dark minerals. The second tuff (2–63) was obtained from a 2 m thick, light gray, and laminated/banded layer, which does not contain any dark minerals.
At the south of Bosnia and Herzegovina the tuff samples were acquired from the Gacko and Livno-Tomislavgrad Basins. The tuff from the Gacko Basin was obtained at the Gračanica locality, specifically from the uppermost part of a lacustrine succession exposed in a large active coal pit in the southwest region of the Gacko Basin. Sampling focused on the lowermost 20 cm of a 1 m thick tuff layer. The section description showing the stratigraphic position of the tephra layer can be found in Mandic et al. (Reference Mandic, de Leeuw, Vuković, Krijgsman, Harzhauser and Kuiper2011). 40Ar/39Ar measurements of feldspar separates revealed an age of 15.31±0.16 Ma.
Three tuff samples from three sections were recovered from the Livno-Tomislavgrad Basin. All sections were described and two were radiometrically dated by de Leeuw et al. (Reference de Leeuw, Mandic, de Bruijn, Marković, Reumer, Wessels, Šišić and Krijgsman2011). The older tuff, which dates to ~17 Ma, is situated in the Tušnica section. It consists of a laterally continuous bed, ~20 cm thick, exhibiting a gray to white coloration. The tuff is notably enriched with flakes of dark mica and is found in a distinct and sharp contact with the surrounding coal seam. The younger tuff, with an age of ~14.68 Ma, was collected from the Lake Mandek locality. A discrete layer of volcanic ash there is 6 m thick and displays a whitish color and weak lithification. The lower part of the tuff’s lithology remains uncertain due to dense vegetation covering the area. However, at the upper boundary, it transitions into lacustrine carbonates. The third tuff was collected in the upper part of the Ostrožac section at 762 m height. Its magnetostratigraphic age was estimated to be ~14.4 Ma (Table 1; de Leeuw et al., Reference de Leeuw, Mandic, de Bruijn, Marković, Reumer, Wessels, Šišić and Krijgsman2011). This tuff layer has a thickness of 50 cm and is found intercalated within lacustrine limestone.
DIB = Dinarides intramontane basins; SPB = Southern Pannonian Basin.
Finally, the sample of tuff from the Sinj Basin was taken from the Glavice composite section and was dated to 17.295±0.028 Ma (Brlek et al., Reference Brlek, Richard Tapster, Schindlbeck-Belo, Gaynor, Kutterolf, Hauff, Georgiev, Trinajstić, Šuica, Brčić, Wang, Lee, Beier, Abersteiner, Mišur, Peytcheva, Kukoč, Németh, Trajanova, Balen, Guillong, Szymanowski and Lukács2023). This vitroclastic to altered vitroclastic tuff is intercalated within the coal bearing lacustrine carbonates (Vranjković, Reference Vranjković2011) emerging in the form of a massive, homogenous, tabular, plastic to compact layer with sharp top and bottom contacts (Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014).
Methods
X-ray diffraction (XRD) measurements were carried out at the Texas Tech Department of Geosciences using a Phillips X’Pert Pro PW3040 diffractometer equipped with a variable divergence slit set to 16 mm exposure area, 15 mm incident beam mask, and 0.04 radian primary and secondary soller slits. For semi-quantitative XRD analysis, 3 g of sample material was weighed. The selected material was later combined with 1 g (20 wt.%) of corundum powder and then milled using a McCrone mill. Samples were gently backloaded into a sample holder of 27 mm internal diameter. Global sample measurements were conducted on a rotating sample holder (8 rpm) using continuous scanning mode in the Bragg-Brentano geometry with CuKα radiation (40 kV and 40 mA). The step size was set at 0.026°2θ in a 5 to 64°2θ range with a total measurement time of 62 min. Measurements of oriented clay fractions were carried out using the same parameters in a 5 to 33°2θ range, with a total measurement time of 23 min. Oriented clay mounts were produced by dispersing crushed material in an ultrasonic bath, followed by the sedimentation of the clay fraction (<2 μm) using centrifugation (Šegvić et al., Reference Šegvić, Zanoni and Moscariello2020). Subsequently, the specimens underwent treatment with ethylene glycol, and when necessary, subjected to heating at 400 and 500°C. The present phases were discerned utilizing the PDF 4+ database in conjunction with the Bruker DIFFRAC.EVA software suite. Clay mineral diffraction traces were analyzed following the guidelines provided by Moore and Reynolds (Reference Moore and Reynolds1997) and Środoń (Reference Środoń2013).
XRD traces were modelled using Sybilla© software, which integrated the formalism of Drits and Sakharov (Reference Drits and Sakharov1976). The fitting of diffraction patterns relies on a trial-and-error procedure, yielding optimal clay mineral structural and probability parameters to achieve the best fits with minimal differences between experimental and calculated patterns. This process also ensures the alignment of intensities of 00l reflections for each of the present clay phases. The number, nature, and stacking sequence of various compositional layers in mixed-layer minerals were considered as adjustable parameters (Uzarowicz et al., Reference Uzarowicz, Šegvić, Michalik and Bylina2012; Šegvić et al., Reference Šegvić, Zanoni and Moscariello2020). To produce experimental diffraction patterns, a single discrete clay mineral phase, namely illite, was introduced. Additionally, three disordered mixed-layer illite-smectite (I-S) phases (R0 I-S, R0 I-SS, R0 I-SSS) and one ordered mixed-layer illite-smectite phase (R1 I-SSS) were included. The second and third smectite in the I-S labeling signify different types of smectite components, distinguished from the preceding one by variations in water content and consequently, the d-spacing. In simpler terms, the expandable layers in I-S are bi-hydrated, in I-SS they are mono- and bi-hydrated, and in I-SSS, there can be up to three layers of water in the smectite component (Ferrage et al., Reference Ferrage, Lanson, Sakharov, Geoffroy, Jacquot and Drits2007). Finally, a disordered kaolinite-smectite (K-S) phase (R0 K-S) was introduced to accommodate the presence of kaolinite in one of the analyzed samples. The semi-quantification of phyllosilicates in the separated clay fractions was carried out using the Sybilla© quantification algorithm.
The BGMN program was employed for Rietveld refinement through the Profex graphical user interface (Doebelin and Kleeberg, Reference Doebelin and Kleeberg2015). The structural parameters of analyzed phases were sourced from the Crystallography Open Database (COD) (Vaitkus et al., Reference Vaitkus, Merkys and Gražulis2021) and the BGMN structure file repository (Doebelin and Kleeberg, Reference Doebelin and Kleeberg2015). For every identified phase, lattice parameters, peak-broadening parameters, and, if necessary, preferred orientation were refined, ensuring that they fell within physically acceptable ranges. Corundum served as an internal standard to assess the proportion of amorphous matter, including volcanic glass, in all analyzed sample (Dapiaggi et al., Reference Dapiaggi, Pagliari, Pavese, Sciascia, Merli and Francescon2015).
Tuff samples rich in glass shards were prepared in the form of thick (∼100 μm) sections and analyzed using LA-ICP-MS at Texas Tech University’s GeoAnalytical Laboratory using an Agilent 7500cs quadrupole mass spectrometer equipped with a New Wave Resolution 213 nm solid state laser with dual-volume cell. The laser was operated at a spot size of 100 μm and a measured fluence of between 6 and 7 J cm–2 with a frequency of 10 Hz. The NIST-610 glass was utilized as the calibration standard while Si served as the internal standard element (Jochum et al., Reference Jochum, Willbold, Raczek, Stoll and Herwig2005). The glass shards were measured from the thin sections whereas the clay fractions from the studied samples were separated using centrifugation (Badurina and Šegvić, Reference Badurina and Šegvić2022) and then investigated using an LA-ICP-MS line raster analysis which has been shown to be adequate for the analysis of an inherently heterogeneous clayey material (Vannoorenberghe et al., Reference Vannoorenberghe, Acker, Belza, Teetaert, Crombé and Vanhaecke2020). Careful polishing of thin sections ensured the removal of potentially altered shard surfaces containing micro- or nano-sized clay inclusions. The time-resolved signal from LA-ICP-MS was examined to detect potential contamination. The precision and accuracy of the method were assessed using secondary standards USGS BHVO-2 G, RGM-1 and NIST-614 (Hollocher and Ruiz, Reference Hollocher and Ruiz1995; Jochum et al., Reference Jochum, Willbold, Raczek, Stoll and Herwig2005; Schudel et al., Reference Schudel, Lai, Gordon and Weis2015) (see Table S1 in the Supplementary material). To inspect trace element mobility during the alteration of volcanic glass, the glass and clay REE abundances were firstly normalized to Al2O3 concentrations of the respective samples (Gifkins and Allen, Reference Gifkins and Allen2001). Such normalized values were than compared against each other using the equation of Nesbitt (Reference Nesbitt1979).
The measurements of the specific surface area (SSA) of clay minerals have been utilized commonly to assess quantitatively their reactive surface sites to predict the sorption and mineral dissolution process in clay-rich sedimentary rocks and soils (Sanders et al., Reference Sanders, Washton and Mueller2010; Macht et al., Reference Macht, Eusterhues, Pronk and Totsche2011). Consequently, higher SSA values of clay minerals were shown to be positively correlated with an increasing REE adsorption potential (Yang et al., Reference Yang, Liang, Ma, Huang, He and Zhu2019; Xu et al., Reference Xu, Zhang, Ding, Liu, Shi, Li, Dang, Cheng and Guo2020). Obtaining the SSA values involves measuring the adsorption of gaseous nitrogen by the material at its boiling point temperature. The adsorption isotherm is further fitted to the Brunauer–Emmett–Teller (BET) equation. This is the widely accepted technique for determining SSA values of phyllosilicates (Brigatti et al., Reference Brigatti, Galán, Theng, Bergaya and Lagaly2013). BET measurements were completed at Purdue Catalysis Center at Purdue University. Prior to the N2-physisorption experiments, 30–40 mg of powdered samples (US Standard No. 120, <125 μm) were firstly outgassed for 9 h at 120°C under vacuum (<6×10–3 mbar) to remove moisture and volatile organic compounds adsorbed to the material. The samples were then transferred to the measurement port attached to a gas/vacuum manifold. N2-physisorption isotherms were acquired by dosing nitrogen at N2 boiling temperature (77K, 196°C), using an adapted protocol issued by the American Society for Testing and Materials (ASTM, D4365-95, 2008). N2-physisorption isotherms were acquired using a Micromeritics 3-Flex instrument. The surface area values were calculated following the BET equation (Brunauer et al., Reference Brunauer, Emmett and Teller1938). The average pore size and pore volume were obtained by fitting the N2 desorption branch to the Barrett–Joyner–Halenda (BJH) equation (Barrett et al., Reference Barrett, Joyner and Halenda1951). The precision defined by the standard deviation of five repeated analyses was in the range of 2–7%.
Results
Mineralogy
The studied tuff samples predominantly exhibit freshness, with the amorphous glassy component making up 40.9–94 wt.% of their composition (Fig. 2; Table 2). The presence of volcanic glass and possibly amorphous silica (i.e. opal-A) is suggested by the prominent 15–35°2θ hump in most of the XRD traces. Notably, exceptions are observed in the Poljanska and Ostrožac samples, where the glass content is relatively small or absent, respectively. Ostrožac is characterized by a dominance of secondary calcite (84.3 wt.%), while Poljanska prominently features analcime (41 wt.%). Tuff samples rich in glass content are further characterized with the presence of illite-smectite (1.5–52.3 wt.%), along with minor quantities of quartz (0.5–7.8 wt.%) and plagioclase (3–16.9 wt.%). Illite-smectite, when identified, displays a prominent 001 reflection at approximately 13.5–12.5 Å. The periodicity of illite-smectite is small, and the sole remaining reflex is observed at around 4.55 Å, potentially corresponding to the asymmetric 02l (020) diffraction system of phyllosilicates as described by Brindley and Brown (Reference Brindley and Brown1980). The overall glass/clay ratio in the analyzed tuff categorizes most of it as either fresh vitroclastic or altered vitroclastic tuff (Table 1). Detailed optical and electron-beam petrographic studies of investigated tuff are available in the works of Badurina et al. (Reference Badurina, Šegvić, Mandic and Slovenec2021) and Badurina and Šegvić (Reference Badurina and Šegvić2022).
Modelling parameters of mixed-layer minerals consisted of: (1) the orientations of particles on the mounted slides (σ*), (2) the coherent scattering domain sizes expressed in number of layers (CSDS), and (3) the amounts of smectite, kaolinite, and illite in interstratified phases (see Table S2 in the Supplementary material). Considering the paragenesis and crystallinity of I-S, the mineralogy of the clay fraction can be categorized into four groups. Group 1 (Mandek and Glavice) is characterized by high crystallinity disordered and ordered illite-smectite (R0 I-SSS, R1 I-SSS; see Table S2 in the Supplementary material). The illite content in R0 I-SSS does not exceed 8%. This aligns with the relative positions of the I-S 002/003 reflexes, along with the 001/002 and 002/003 Δ°2θ values in ethylene glycol-treated traces (see Fig. S1 in the Supplementary material), indicating a smectite-rich (i.e. >90%) illite-smectite (Moore and Reynolds, Reference Moore and Reynolds1997). Group 2 (Ostrožac and Gračanica) is predominantly composed of multiple generations of well-crystallized and disordered I-S (R0 I-SS), featuring an average of about 20% of an illite component. Detrital illite is also present in this group, while kaolinite-smectite intermediates were observed solely in the Ostrožac sample (see Fig. S1 in the Supplementary material). The samples from Kamengrad2–4, Čaklovići, and Poljanska constitute Group 3. These samples consist of multiple generations of relatively poorly crystalline and disordered illite-smectite (R0 I-SS and R0 I-SSS) with an average illite content of 20%. Discrete illite is likely to be present in these samples. Finally, including samples from Tušnica and Kamengrad2–63, Group 4 exhibits the presence of a variety of I-S with very low crystallinity (see Fig. S1 in the Supplementary material). Modeling these species proved challenging, but it appears that their illite content is significantly greater (about 60%). It is important to note that the samples in this group are rich in amorphous matter (about 90%), while at the same time having the lowest documented abundances of I-S (around 2%; Table 2). Finally, the prominent I-S first basal reflexes documented in all ethylene glycol-solvated traces of analyzed samples, except Čaklovići, revolve around 5.2°2θ (see Fig. S1 in the Supplementary material), which corroborates the random, smectite-rich nature of the present I-S intermediates (Środoń, Reference Środoń1980).
N2 adsorption–desorption isotherm characteristics
The gas adsorption method serves as a routine approach to determine the specific surface area of fine-grained materials. Analyzing the adsorption isotherm of a non-polar gas such as nitrogen allows for the calculation of the surface area on which a monolayer of adsorbed gas molecules ideally accumulates (Brunauer et al., Reference Brunauer, Emmett and Teller1938). Furthermore, the volume of adsorbed gas within meso- and micropores can be assessed by examining hysteresis in the adsorption and desorption isotherms, as well as analyzing the adsorption isotherm at a low partial pressure (Aylmore and Quirk, Reference Aylmore and Quirk1967; Aringhieri, Reference Aringhieri2004; Michot and Villiéras, Reference Michot, Villiéras, Clay Science, Bergaya, Theng and Lagaly2006; Kaufhold et al., Reference Kaufhold, Dohrmann, Klinkenberg, Siegesmund and Ufer2010). By analyzing the configuration of the adsorption isotherms and the characteristics of hysteresis loops (Fig. 3), valuable insights can be derived concerning the adsorption mechanism within analyzed clay minerals, including qualitative indications of the types and geometries of the pores. Considering that the type of exchangeable cation in the interlayer of 2:1 clay minerals can significantly impact the specific surface area (SSAN2BET) and pore volumes of dried powdered samples (Cases et al., Reference Cases, Bérend, François, Uriot, Michot and Thomas1997; Rutherford et al., Reference Rutherford, Chiou and Eberl1997), it is worth noting that the clay samples studied had a relatively consistent earth and alkaline earth content (Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021; Badurina and Šegvić, Reference Badurina and Šegvić2022).
The SEM-EDS investigation revealed that the majority of examined tuffs exhibited a highly compact, laminar arrangement of illite-smectite crystallites. However, in the Čaklovići sample, the illite-smectite morphology seemed rather hairy with an arrangement that resembled a honeycomb or house of cards structure (Badurina and Šegvić, Reference Badurina and Šegvić2022). This is reflected by the differences in shapes of the N2 adsorption–desorption hysteresis curves between the said sample on one hand and most of the others on the other hand (Fig. 3). These differences become evident at relative pressures exceeding 0.5, which does not directly have an impact on the typical BET range. Additionally, the N2 adsorption data show notable variations in porosity and SSAN2BET values between the Čaklovići sample and the rest of the dataset. The rest of the dataset shows relative variations in SSAN2BET values which is, given their comparable shape arrangements, probably related to the particle size and the fact that fewer stacked 2:1 layers have more reactive surfaces accessible to the N2 molecules (Suárez et al., Reference Suárez, Lorenzo, García-Vicente, Morales, García-Rivas and García-Romero2022). The average pore size ranged from 19 to 38 Å, with the volume of mesoporosity (20–500 Å) spanning from 0.01 to 0.41 cm3 g–1, while the contribution of microporosity (<20 Å) was negligible (Table 3).
NCB = North Croatian Basin; LTB = Livno-Tomislavgrad Basins.
Rare-earth elements geochemistry
The concentrations of REE (Table 4) were normalized to chondrite (Boynton, Reference Boynton and Henderson1984) for both glass shards and the clay mineral matrix (Fig. 4). The REE curves exhibit parallel trends, spanning approximately 4–130 times the chondrite content, with total REE content ranging from 65.3 to 394.4 ppm. The geochemical characteristics of volcanic glass highlight its evolved igneous nature, marked by light rare-earth element (LREE) enrichment (LaN/YbN = 4.5–6.5) and an Eu anomaly (Eu* = 0.3–0.5). This pattern is also documented in the clay matrix, although certain clay samples show greater LREE/HREE ratios and slightly reduced Eu anomalies. Such trends align with the findings of Badurina et al. (Reference Badurina, Šegvić, Mandic and Slovenec2021), suggesting that the studied tuffs originate from an evolved, intermediate to felsic, magmatic source based on whole-rock tuff geochemistry.
Concentrations are expressed in ppm. *Data from Badurina and Šegvić (Reference Badurina and Šegvić2022); **glass shards not present. NCB = North Croatian Basin; LTB = Livno-Tomislavgrad Basin.
Apart from the Čaklovići and Glavice samples, all mobility curves indicate element loss during the argillitization of volcanic glass (Fig. 5). Moreover, the plots exhibit a range of LREE/HREE ratios, from comparable values (Ostrožac, Kamengrad 2.63) to significant LREE enrichment (Čaklovići, Glavice). Clay fractions extracted from Kamengrad, Glavice, Tušnica, and Ostrožac tuffs exhibit noticeable enrichment in Eu. These positive Eu anomalies are likely to be linked to the prevailing reducing paleoenvironment, causing the preferential mobilization of divalent Eu (Bau, Reference Bau1991; Yang et al., Reference Yang, Liang, Ma, Huang, He and Zhu2019).
Discussion
Depositional environment conditions and element mobility
Volcanic glass shows inherent instability under low-temperature diagenetic conditions. Upon interaction with water, it easily undergoes transformations into various hydrous phases, such as illite-smectite, which are thermodynamically more stable than glass (Cuadros et al., Reference Cuadros, Caballero, Huertas, Jiménez de Cisneros, Huertas and Linares1999; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021; Gong et al., Reference Gong, Li, Lu, Wang and Tang2021). The composition of secondary paragenesis is controlled by both the glass composition and environmental physiochemical conditions (Yamamoto et al., Reference Yamamoto, Sugisaki and Arai1986; Caballero et al., Reference Caballero, Reyes, Delgado, Huertas and Linares1992; dos Muchangos, Reference dos Muchangos2006; Namayandeh et al., Reference Namayandeh, Modabberi and López-Galindo2020). The mechanisms driving the glass alteration include ion diffusion and rearrangement on one hand, and dissolution and precipitation on the other (de la Fuente et al., Reference de la Fuente, Cuadros, Fiore and Linares2000). In lacustrine, low-temperature (15–25°C; Tütken et al., Reference Tütken, Vennemann, Janz and Heizmann2006) and circum-neutral to alkaline regimes (pH 7–8.5; Kříbek et al., Reference Kříbek, Knésl, Rojík, Sýkorová and Martínek2017), REE are mostly complexed; however, in such solutions LREE appear to be less complexed than HREE. The first form complexes with Cl–, SO42–, and CO32–, while the second are commonly associated with CO32– (Valsami and Cann, Reference Valsami and Cann1992; Li et al., Reference Li, Webb, Algeo, Kershaw, Lu, Oehlert, Gong, Pourmand and Tan2019; Gong et al., Reference Gong, Li, Lu, Wang and Tang2021; Šegvić et al., Reference Šegvić, Slovenec and Badurina2023b). This means that illite-smectite, a phase newly formed through the volcanic glass alteration of studied tuff (Table 2), will have to contend for REE in diagenetic environments with various dissolved anionic species (Cuadros et al., Reference Cuadros, Mavris and Nieto2023).
The chondrite-normalized REE curves of the clay separates exhibit similar shapes to those of the glass, albeit notable differences exist in terms of total concentrations (Fig. 4). This calls for a relatively large range of fluid/rock ratio environments, which are pivotal in generating fluctuations, both in the enrichment and depletion of REE within the clay fractions (Fig. 5). In instances of REE enrichment the clay mineral hosted adsorption sites clearly prevailed over those associated with dissolved anions. Conversely, REE depletion suggests a deficiency in solid phase binding sites. The larger amount of LREE relative to HREE observed in the studied glass and their clay derivatives is likely to be an artifact of the source magmatism. This interpretation considers the geochemistry of genetically linked Early to Mid-Miocene extension-related felsic and intermediate effusives of the Pannonian Basin (Harangi and Lenkey, Reference Harangi and Lenkey2007; Seghedi and Downes, Reference Seghedi and Downes2011; Lukács et al., Reference Lukács, Harangi, Guillong, Bachmann, Fodor, Buret, Dunkl, Sliwinski, von Quadt, Peytcheva and Zimmerer2018) and their pyroclastic manifestations in the North Croatian Basin and the Dinarides (Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014; Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021; Brlek et al., Reference Brlek, Richard Tapster, Schindlbeck-Belo, Gaynor, Kutterolf, Hauff, Georgiev, Trinajstić, Šuica, Brčić, Wang, Lee, Beier, Abersteiner, Mišur, Peytcheva, Kukoč, Németh, Trajanova, Balen, Guillong, Szymanowski and Lukács2023; Grizelj et al., Reference Grizelj, Milošević, Miknić, Hajek-Tadesse, Bakrač, Galović, Badurina, Kurečić, Wacha, Šegvić, Matošević, Čaić-Janković and Avanić2023; Šegvić et al., Reference Šegvić, Lukács, Mandic, Strauss, Badurina, Guillong and Harzhauser2023a; Trinajstić et al., Reference Trinajstić, Brlek, Gaynor, Schindlbeck-Belo, Šuica, Avanić, Kutterolf, Wang, Lee, Holcová, Kopecká, Baranyi, Hajek-Tadesse, Bakrač, Brčić, Kukoč, Milošević, Mišur and Lukács2023) which all adumbrate LREE/HREE enrichment and negative Eu anomaly. Taking into account the minor variations in the ΣLREE/ΣHREE ratios between fresh glass and their clay alteration products (Fig. 4), relatively equal net losses/gains are inferred, except for the Čaklovići and Glavice samples, which exhibit a notable LREE/HREE enrichment and an overall REE gain in the clay separates (Fig. 5). This cannot be explained by the source geochemistry and is more likely indicative of a preferential dissolution of HREE through carbonate complexation (dos Muchangos, Reference dos Muchangos2006; Badurina and Šegvić, Reference Badurina and Šegvić2022). Limited sorption of such complexes by clay mineral species further aids their depletion in the clay fraction (Byrne and Kim, Reference Byrne and Kim1990; Sholkovitz et al., Reference Sholkovitz, Landing and Lewis1994). Given the close proximity of the analyzed tuff beds and lacustrine marls, limestone and/or dolostone (Sunarić et al., Reference Sunarić, Glišić, Dangić and Milivojević1976; Mandic et al., Reference Mandic, de Leeuw, Vuković, Krijgsman, Harzhauser and Kuiper2011; Vranjković, Reference Vranjković2011; Ćorić et al., Reference Ćorić, Vrabac, Đulović and Babajić2018; Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022), it is reasonable to hypothesize that the paleolake waters in which the studied ash landed were carbonate saturated. This aligns with the conditions observed in many modern karstic lakes (Cantrell and Byrne, Reference Cantrell and Byrne1987; Christidis, Reference Christidis1998). Finally, a detailed inspection of REE mobility curves (Fig. 5) shows that the mobility, and therefore a net loss, of HREE increases proportionally with their greater atomic weight during the alteration of volcanic glass. This heightened mobility is, as already discussed, attributed to HREE complexation in the aqueous phase (Summa and Verosub, Reference Summa and Verosub1992). In contrast to the Čaklovići and Glavice tuffs, all the remaining ones exhibit a discernible degree of net REE loss in clay separates (Fig. 5), with the Poljanska and Ostrožac tuffs displaying the most pronounced depletion (50–90%). The significant REE loss may be attributed to the mineralogy of the mentioned two samples which are dominated by zeolite and calcite, respectively, with only a minor presence of clay minerals (Table 2). Compared with phyllosilicates, zeolite and calcite do not fractionate rare-earth elements during low-grade diagenetic processes and their abundances, if any, are generally at trace levels (Terakado and Nakajima, Reference Terakado and Nakajima1995; Cherniak, Reference Cherniak1998; Tanaka and Kawabe, Reference Tanaka and Kawabe2006).
The mineralogy of the studied tuff indicates a dominant presence of various illite-smectite intermediates (Table 2), which largely lean toward a smectite-rich composition (see Table S2 in the Supplementary material). This observation suggests a slightly basic character of the depositional environment, coupled with the removal of alkalis. These environmental conditions are conducive to the destabilization of glass and feldspars, favoring the preferential formation of smectite over zeolite (Christidis and Huff, Reference Christidis and Huff2009; Šegvić et al., Reference Šegvić, Mileusnić, Aljinović, Vranjković, Mandic, Pavelić, Dragičević and Ferreiro Mählmann2014; Krajišnik et al., Reference Krajišnik, Daković, Milić, Marković, Mercurio, Sarkar and Langella2019; Feng et al., Reference Feng, Kou, Tang, Shi, Tong and Zhang2023). In contrast, the Poljanska tuff stands out from this pattern due to its alteration in a saline lake environment (Pavelić et al., Reference Pavelić, Kovačić, Miknić, Avanić, Vrsaljko, Bakrac, Tisljar, Galovic and Bortek2003). In this specific setting, alkalis persist within the system (Hay, Reference Hay1964; Hay, Reference Hay, Murakami, Iijima and Ward1986). The elevated alkalinity in this environment probably contributed to an increased glass solubility (Mariner and Surdam, Reference Mariner and Surdam1970), leading to the large-scale formation of zeolite (Table 2).
The prevailing redox conditions during the eogenetic alteration of volcanic glass are readily reflected in the fractionation of redox-sensitive REE, such as Ce and Eu (Elderfield and Greaves, Reference Elderfield and Greaves1982; Bau, Reference Bau1991; Elderfield et al., Reference Elderfield, Whitfield, Burton, Bacon, Liss, Charnock, Lovelock, Liss and Whitfield1997; Chen et al., Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015). To illustrate a net change of Ce and Eu that occurred during the argillitization of the studied tuff, their δCe and δEu values were calculated (Liao et al., Reference Liao, Hu, Cao, Wang, Yao, Wu and Wan2016) (Table 5). When considering δCe, it is noteworthy that all the examined samples exhibit somewhat positive anomalies (ranging from 2.86 to 0.97), apart from the Čaklovići and Glavice samples, which display very weak negative anomalies (ranging from 0.93 to 0.87). This observation is typically associated with redox processes and the partial oxidation of soluble Ce3+ to Ce4+. The oxidized form then precipitates in situ as CeO2 (Liu et al., Reference Liu, Tournassat, Grangeon, Kalinichev, Takahashi and Marques Fernandes2022), accumulating in the solid phase, which is dominated by authigenic illite-smectite in the case of the studied altered tuff. The δCe values therefore serve as valuable proxies for assessing the redox environment during the alteration of glass and the subsequent eogenesis of clay minerals (Namayandeh et al., Reference Namayandeh, Modabberi and López-Galindo2020). Following the guidelines proposed by Chen et al. (Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015), it becomes evident that these redox conditions varied from oxic (δCe > 1.5, as observed in Kamengrad2.40 and Mandek), to suboxic (δCe ~1.1–1.4, as seen in Kamengrad2.63), and finally to anoxic (δCe < 1.1, observed in the remaining tuff; Table 5). Thicker layers of Kamengrad and Mandek tuffs (2 m and 6 m, respectively) are surmised to have offered a larger volume of porous material for the infiltration of freshwater, thereby fostering a more oxygenated environment during diagenesis. Conversely, thinner and more altered tuff horizons, abundant in less permeable fine-grained carbonate material and authigenic clays, exhibit reduced porosity, leading to limited influx of oxygenated eogenetic fluids (Martizzi et al., Reference Martizzi, Chiyonobu and Arato2020).
Concentrations are expressed in ppm. *Data from Badurina and Šegvić (Reference Badurina and Šegvić2022). ΣREE mob = sum of Al normalized ΣREE(clay)/ΣREE(shard) ratios; SSA = specific surface area of clay fractions; % Amorph. = share of amorphous matter (glass); % I-S = share of illite-smectite; % Sm in I-S = sum of Sme components in I-S in all present intermediates; Cryst. = crystallinity group (Table 3); δCe = CeN/(LaN×PrN)½; δEu = EuN/(GdN×SmN)½ (N denotes clay normalization to glass).
Regarding the δEu anomaly, the examined tuffs exhibit a broad spectrum of positive δEu values, ranging from 1.03 to 1.69, except for the Čalkovići sample, which registers a value of 0.87 (Table 5). The present authors acknowledge the fact that the Eu anomaly largely stems from inheritance (Fig. 4); this correlation is, however, not universally consistent, and the total intensities of the Eu anomaly were probably influenced by diagenetic conditions. Europium is another highly redox-sensitive lanthanide that, under the prevailing reducing conditions, can undergo transformation into a more soluble divalent form. This transformation ultimately results in the removal of Eu from the solid phase (Bau, Reference Bau1991; Chen et al., Reference Chen, Algeo, Zhao, Chen, Cao, Zhang and Li2015; Yang et al., Reference Yang, Liang, Ma, Huang, He and Zhu2019). The range of δEu values observed in the studied tuff corroborates the assessment of environmental conditions during tuff diagenesis, indicating a predominantly suboxic environment. However, in the case of the Čaklovići tuff, these conditions were notably more reductive. Additional consideration is needed in interpreting δEu values. Unlike Ce, the conversion of Eu3+ to Eu2+ requires high temperatures (>200°C) (Nagender Nath et al., Reference Nagender Nath, Bau, Ramalingeswara Rao and Rao1997). The comparatively greater mass of pyroclastic material in the case of Čaklovići tuff (~3 m thick; Ćorić et al., Reference Ćorić, Vrabac, Đulović and Babajić2018) probably sustained elevated temperatures for an extended duration, facilitating the reductive dissolution of Eu during tuff diagenesis (Christidis and Huff, Reference Christidis and Huff2009). Furthermore, the authors of this study posit that microbial activity might have played a role in the reduction of Eu (Castillo et al., Reference Castillo, Maleke, Unuofin, Cebekhulu, Gómez-Arias, Sinharoy and Lens2022) across all the examined tuffs, given the presence of redox-sensitive microorganisms in certain tuffs (Mandek; Badurina et al., Reference Badurina, Šegvić, Mandic and Zanoni2020). Finally, the δEu/δCe ratio (Table 5) may offer further insights into the weathering conditions of the studied glass. Consequently, greater values correspond to reduced weathering and a greater content of preserved fresh glass, where Eu is predominantly found in its trivalent form (Drake, Reference Drake1975). Conversely, smaller δEu/δCe ratios, not associated with an oxic environment (δCe > 2), generally indicate increased weathering intensities due to greater proportions of newly formed clays.
Clay mineral control on REE mobility
Clay minerals, being the predominant by-products of weathering and hydrothermal alterations, typically retain the majority of the REE budget in sedimentary environments (McLennan, Reference McLennan2001; Carloni et al., Reference Carloni, Šegvić, Sartori, Zanoni, Moscariello and Besse2021; Green et al., Reference Green, Šegvić, Luka, Omodeo-Salé and Le Bayon2024). This phenomenon is attributed to the smaller crystallite size and pronounced anisotropy of clay minerals, which eventually leads to the development of two distinct charged surfaces – the basal siloxane surfaces carrying a permanent negative charge and the edge or broken-bond surfaces hosting pH-dependent charge sites (Johnston and Tombácz, Reference Johnston and Tombácz2002; Tournassat et al., Reference Tournassat, Bourg, Steefel, Bergaya, Clay Science, Tournassat, Steefel, Bourg and Bergaya2015). Clay minerals, formed through volcanic glass diagenesis inherit the original magmatic REE signatures (Fig. 4). Additionally, they fractionate REE based on the prevailing physiochemical conditions of the environment, such as redox potential (Eh) and pH, in conjunction with factors like their speciation, morphology, crystallinity, or cation exchange capacity (Yang et al., Reference Yang, Liang, Ma, Huang, He and Zhu2019; Namayandeh et al., Reference Namayandeh, Modabberi and López-Galindo2020; Badurina and Šegvić, Reference Badurina and Šegvić2022; Li and Zhou, Reference Li and Zhou2023). This becomes apparent in the instance of the studied tuff, particularly with the Čaklovići and Glavice samples standing out as the sole examples of REE enrichment in the clay fraction (Fig. 5; Table 5). Despite originating in a slightly alkaline environment, akin to most investigated tuffs, these two samples underwent argillitization in a notably more reducing environment. This probably played a role in the adsorption of REE onto illite-smectite surfaces.
The geochemical nature of studied tuffaceous clays and their spatially related volcanic glass particles is relatively similar (Table 5). It follows that the variations in REE content observed in authigenic clay minerals across different tuffs primarily arise from diagenetic processes. The small particle size of illite-smectite directly corresponds to an increased external surface area of individual crystallites. This is attributed to expansive basal surfaces of I-S and a limited number of stacked T-O-T interlayers (Reid-Soukup and Ulery, Reference Reid-Soukup and Ulery2002). These areas are readily accessible to nitrogen molecules, making N2 adsorption a convenient method for determining the specific surface area (SSAN2BET) of the studied clays (Table 3). The SSAN2BET and calculated porosity values show significant differences between the Čaklovići tuff on one end and the rest of the dataset on the other. Upon comparing them with the cumulative mobile REE (ΣREEmob = Al-normalized (ΣREEclay/ΣREEshard); Nesbitt, Reference Nesbitt1979) it becomes evident that the Čaklovići tuff stands out prominently. In contrast, the remaining tuffs exhibit a less defined correlation between SSAN2BET and ΣREEmob (Fig. 6a). It follows that the pronounced capacity of the Čaklovići tuff to fractionate REE during diagenetic alteration of tuff is a function of its high SSAN2BET and porosity values (Table 5). The increase in SSAN2BET of smectite layers is probably associated with intraparticle porosity, stemming from the quasi-crystalline overlap region and accessible zones within the interlayer structure (Suárez et al., Reference Suárez, Lorenzo, García-Vicente, Morales, García-Rivas and García-Romero2022). Indeed, the particle morphology of the Čaklovići illite-smectite is rather fibrous or hairy compared with illite-smectite from the rest of studied tuff depicting a typical platelet or cornflake texture (Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021; Badurina and Šegvić, Reference Badurina and Šegvić2022). This wispy or hairy illite-smectite typically has a thickness of only a few hundred angstroms, but its length exhibits considerable variation, extending up to tens of micrometers (Huggett, Reference Huggett2005). This leads to the development of slit-shaped non-basal surfaces or porosity at the edges of particles, thereby influencing the overall SSAN2BET of illite-smectite (Kaufhold et al., Reference Kaufhold, Dohrmann, Klinkenberg, Siegesmund and Ufer2010). In analyzing the distinctive features of the Čaklovići tuff, a salient consideration pertains to its substantive MgO content, measuring 4.94 wt.%, a marked departure from the composition of other tuffs (Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021). Given the mineralogy of studied tuffs (Table 2), it may be hypothesized that Mg is largely hosted in illite-smectite, as felsic glass is virtually devoid of it (Christidis and Huff, Reference Christidis and Huff2009). Upon integration of Mg into the structural framework of illite-smectite, it will substitute Al in the octahedral sheet, which in turns leads to charge deficit delocalized over the siloxane planes (Sposito et al., Reference Sposito, Skipper, Sutton, Park, Soper and Greathouse1999). This creates a conducive adsorption environment for solvated cations like LREE, which propagate away from the glass in the guise of 8- or 9-fold hydrated outer-sphere complexes (Borst et al., Reference Borst, Smith, Finch, Estrade, Villanova-de-Benavent, Nason, Marquis, Horsburgh, Goodenough, Xu, Kynický and Geraki2020). Indeed, the large MgO content of Čaklovići clays might have contributed to their LREE retention potential (Fig. 5; Table 5).
To investigate further into the characteristics of I-S that potentially govern the adsorption of REE during tuff alteration, the smectite content of I-S (Table S2; Table 5) was plotted against ΣREEmob. This revealed a strong correlation between these two parameters (Fig. 6b). It appears that REE exhibit a pronounced affinity for the less charged smectite component within the I-S structure. This suggests a potential interaction between the smectite component and REE through surface complexation and/or ion exchange mechanisms. Additionally, the correlation between the particle size of mixed-layered clay minerals and their surface area exhibits an inverse proportionality (Nadeau et al., Reference Nadeau, Wilson, McHardy and Tait1984). The augmentation of smectite layer content within the smallest I-S crystallites therefore implies a concurrent increase in the REE-accessible adsorption surface area (Jaynes and Bigham, Reference Jaynes and Bigham1986; Sposito et al., Reference Sposito, Skipper, Sutton, Park, Soper and Greathouse1999). Beyond basal surfaces, the presence of variable pH-dependent sites along broken hydroxyl edges of I-S offers additional sites for the complexation of REE (Coppin et al., Reference Coppin, Berger, Bauer, Castet and Loubet2002; Wu et al., Reference Wu, Chen, Wang, Xu, Lin, Liang and Cheng2023). In solutions with a greater alkaline pH, these edges are more favorable to REE adsorption (Awual et al., Reference Awual, Kobayashi, Shiwaku, Miyazaki, Motokawa, Suzuki, Okamoto and Yaita2013). Their significance is particularly notable in the context of the Čaklovići tuff, where there is evidence of increased slit-shaped intraparticle porosity along the edges, as suggested by SSAN2BET (Table 5) and corroborated by SEM-EDS data (Badurina et al., Reference Badurina, Šegvić, Mandic and Slovenec2021). In connection with the noted phenomenon of REE fractionation being most pronounced in smectite-rich I-S, it is noteworthy that the clay fraction of those samples is devoid of discrete illite (Table S2; Table 5). Furthermore, there is no discernible correlation between the overall content of I-S or volcanic glass on one end and the total content of REE associated with authigenic illite-smectite on the other (Table 5). This underscores the leading role of the smectite component of I-S, and to some extent, the particle specific surface area and charge distribution, as primary determinants influencing REE fractionation during volcanic ash diagenesis.
From a regional perspective, it can be inferred that illite-smectite of the altered Miocene tuff of the Southwestern Pannonian Basin and the adjacent Dinarides intramontane basins show a notable capacity as reasonably effective REE scavenging lithologies. Typically, it retains no less than 50% of the REE budget originating from the volcanic source (Fig. 5). Over the course of the Early and Middle Miocene epochs, the areas along the southern perimeter of the Pannonian Basin experienced substantial climatic, tectonic, and eustatic transformations. These engendered a range of depositional environments, encompassing lacustrine to marine settings, with coal beds frequently defining the deposition environment for the tuffs (Pavelić et al., Reference Pavelić, Kovačić, Miknić, Avanić, Vrsaljko, Bakrac, Tisljar, Galovic and Bortek2003; Pavelić and Kovačić Reference Pavelić and Kovačić2018; Mandic et al., Reference Mandic, Rundić, Ćorić, Pezelj, Theobalt, Sant and Krijgsman2019a; Mandic et al., b; Pavelić et al., Reference Pavelić, Kovačić, Tibljaš, Galić, Marković and Pavičić2022). The prevailing suboxic to reducing conditions, and a relatively alkaline nature, proved favorable to the adsorption of REE on illite-smectite. This phenomenon was particularly pronounced in transitional lacustrine to marine settings, as exemplified by the Čaklovići tuff (Fig. 1), where highly reactive, smectite-rich I-S developed from the glassy substrate.
Conclusions
This study investigated the complex interplay between depositional environment conditions, volcanic glass alteration, and the subsequent authigenesis of clay minerals in Miocene SPB and DIB tuffs on one hand and then the ability of clay minerals to fractionate REE on the other hand. The normalized REE patterns of clay mineral separates demonstrate similarities to those of the glass, albeit with variations in total concentrations. These variations suggest a significant range of fluid/rock ratios, crucial in determining fluctuations in REE enrichment and depletion within the clay fractions. The source magmatism geochemistry contributes to the observed patterns of REE distribution. Notably, the distinct LREE/HREE enrichment and negative Eu anomaly align with the characteristics of Early to Mid-Miocene extension-related magmatism in the Pannonian Basin.
The redox conditions during the eogenetic alteration of volcanic glass are reflected in the fractionation of redox-sensitive REE such as Ce and Eu. The calculated δCe and δEu values provide valuable insights into the redox environment during tuff diagenesis, indicating variations from oxic to anoxic conditions across different samples. The subtle positive correlation between δEu and δCe hints at a potential influence of coal-bearing layers within the same diagenetic environment. This study demonstrated that illite-smectite, formed through volcanic glass diagenesis, inherits magmatic REE signatures and fractionates REE based on prevailing physiochemical conditions. The particle properties of illite-smectite, including specific surface area and porosity, also influence its REE fractionation capacity, with the Čaklovići tuff standing out prominently due to its unique characteristics. Moreover, a correlation between smectite content in illite-smectite and a total amount of fractionated REE draws attention to the significant role of smectite layers in REE fractionation during volcanic ash diagenesis. In conclusion, the altered Miocene tuffs in the SPB and DIB serve as a noteworthy REE scavenger, retaining a substantial portion of the REE budget from the volcanic glass. Additional research is needed, however, to assess the economic potential of such ion adsorption clays.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/cmn.2024.21.
Author contributions
Branimir Šegvić: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Resources, Data Curation, Supervision, Writing – Original Draft, Visualization, Funding Acquisition; Luka Badurina: Conceptualization, Methodology, Investigation, Data Curation, Resources, Writing – Review & Editing, Funding Acquisition; Adriano E. Braga: Formal Analysis, Investigation, Resources, Writing – Review & Editing; Oleg Mandic: Investigation, Writing – review & editing; Damir Slovenec: Writing – review & editing; Kevin Werts: Resources, Writing – review & editing; Emily Doyle: Formal Analysis; Frane Marković: Resources; Goran Slivišek: Resources; Vedad Demir: Resources.
Acknowledgements
The authors thank Svetlana Renovica (‘Rudnik i termoelektrana Ugljevik’), Boško Vuković (‘Rudnik Gacko’) and Tvrtko Ćubela (Livno) as well as Sejfudin Vrabac and Elvir Babajić from the University of Tuzla and Hazim Hrvatović from the Federal Geological Survey Sarajevo for their great help during the field work. Kevin Byerly is thanked for the constructive comments he provided during the preparation of this manuscript. We greatly appreciate the assistance of Ebunoluwa Olowolayemo for his fantastic support with the English, ensuring improved clarity, style, and grammar throughout this work. Finally, the critical comments and constructive reviews by Paul A. Schroeder and one anonymous reviewer, as well as editorial handling by Joseph W. Stucki have contributed significantly to the quality of this paper.
Financial support
The research received support from the Portnoy Geology Solidus Fund and the Texas Tech Geoscience Society, as well as the Geosciences Clay Laboratory of Texas Tech University.
Competing interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data availability statement
Additional data supporting the findings of this study are available from the corresponding author, Branimir Šegvić, upon request.