To investigate the particle size of the tephra, about 10 g of ash sample from each core was thoroughly cleaned and dried. The dry samples were treated with (1N) HCl to separate from carbonates and then thoroughly rinsed with Milli-Q water. Then the samples were sieved through multiple analytical sieves having pore sizes of 600 μm, 365 μm, 250 μm, 125 μm, and 63 μm, and thus, the weight percentages of each size fraction were estimated by using a microbalance. The volcanic glass shards available in each ash layer were examined with an X-ray diffractometer (Regaku Ultima IV) within the 2θ ranges between 10° to 57° to confirm their glassy nature. Several randomly selected glass particles of each ash layer were thoroughly examined for external morphologies with an optical microscope (Nikon’s SMZ-1500) and electron microscope (JEOL JSM-5800LV). The specific gravity of those glass separates from each core was also determined by the weight/volume (ASTM, 1995) method.

The composition of major elements in the glass shard was determined with an X-ray fluorometer (Axios PANanalytical). For this, the borate beads were made by using dry, powdered samples with Spectromelt A2 (Merck, Germany). Those beads were analyzed for major oxides by using the certified reference standard BE-N, JR-1, and JR-2 with an accuracy of (± 3 –4)%. For trace and rare earth element analyses, 0.05 g dry powdered samples were digested completely with 10 mL of an acid mixture (containing 7 : 3 HF-HNO₃); in 25 mL Savillex® screw-capped Teflon vials by following a closed digestion method. After swirling, the vials were kept on the hot plate at ~150℃ for 48 h. Following this, the contents were evaporated at 200℃ to dryness with a few drops of HClO₄. The residues were then dissolved in 10 mL 6.0(N) HNO_3, and the final volume was made to 250 mL with Milli-Q water. Solutions were analyzed for trace elements with HR-ICPMS (Nu Instruments Attom®, UK) by using (1 ng/mL) ¹⁰³Rh solution as an internal standard (Satyanarayanan et al., 2018). Certified reference materials JG-1a, JG-2, JR-2 from the Geological Survey of Japan, and G-2 from USGS, along with procedural blanks, were also analyzed by following the same protocol. Care has been taken to eliminate isobaric interferences, which were programmed and corrected automatically. Precision and accuracy were better than 3% RSD for most of the trace and rare earth elements.

The size, morphology, and specific gravity of glass shards from each core were studied separately, and the results are presented in Table 1. The data on particle size reveal that at all five sampling sites, the volcanic ashes were mostly dominated by silt-sized (2–63 µm) glass shards (48%-81% of the total), followed by larger sand-sized (63–125 µm) shards (16%–37%). While the abundances of relatively large (>125 µm) or very fineclay-sized (<2.0 µm) glass particles were substantially low (<10% of the total) (Table 1). It has also been noticed that the overall size distribution patterns of glass shards in ash layers considerably differ among the studied cores, but their specific gravity varies within a close range from 2.6 g/cm³ to 2.7 g/cm³ with an average of 2.64 g/cm³. Furthermore, the particle-size distribution of each sample shows multimodal patterns (M₁-M₅) having characteristic sizes of 2.84 µm, 5.5 µm, 9.25 µm, 25.32 µm, and 64.95 µm (Fig. 2). The particles of medium to fine sand size were transported ~2300 km away from the source. The shape of these shards ranges from blocky or platy to vesicular, most with a typical bubble-wall structure indicating atmospheric transport. This further suggests that the ash particles were deposited in the Mahanadi basin, bay of Bengal, directly from the atmosphere as resulting ash and were subsequently dispersed by bioturbation. Such multimodal size distribution can be attributed to tephra originating from a distal volcanic source and hence occur as poorly sorted ash aggregates (Carey and Sigurdsson, 2000; Lewis et al., 2012).

Like variable particle size, the range of color from colorless to deep brown of these glass shards also apparently indicates their compositional variability. Most of the glass shards displayed cuspate, flat, or curved platy morphology. SEM analyses showed that most of the glass shards have bubble-walled features and are characterized by smooth, slightly curved surfaces (Fig. 3B). An earlier study by Izeet (1981) indicated that usually less viscous G-type rhyolitic magma sources having temperatures more than 870℃ might develop such characteristic morphology in volcanic glass. Similar double wall morphology in Toba glass shards has also been reported in samples from other parts of the Indian Ocean (Gasparotto et al., 2000; Jumaila et al., 2017) and in the South China Sea (Rose and Chesner, 1987; Bühring and Sarnthein, 2000; Song et al., 2000).

The major element compositions of the glass shards from the ash layers of five cores from the Mahanadi basin are dominated by SiO₂ (75.2%–76.2%) and Al₂O₃ (10.5%–11.3%); relative to total alkalis (i.e., Na₂O+K₂O+CaO ~8.1%–8.9%) and Fe-oxides (0.7%–1.0%)(Table 2). The elevated silic a concentrations and higher K₂O contents (>4.5%) than CaO (<0.9%) in all these glass shards match well with typical rhyolitic magma with depleted iron content and differ from andesitic volcanic sources. The plot of alkali metals against silica contents (i.e., TAS diagram) in Fig. 4A also shows the glass compositions in all five cores were very consistent, with magma having typical rhyolitic composition. Moreover, the close range of major elemental composition in glass shards from different cores further suggests all these ash deposits across the Mahanadi basin might originate from the same rhyolitic sources. In the TAS diagram, the rhyolitic glass from the Mahanadi basin closely resembles the Toba volcanic deposits reported elsewhere (Pattan et al., 2010; Srivastava and Singh, 2020; Amonkar et al., 2020). Most of the other major oxide concentrations were also found compatible with the glasses associated with Toba tuffs from other parts of the Indian Ocean, South China Sea, and as well as with volcanic glasses from the Toba caldera (Table 2).

Till date, in many parts of the Indian Ocean and adjacent continental regions, the evidence of the past three major volcanic events at Toba, Sumatra (i.e., OTT of 800 ka; MTT of 500 ka, and YTT of 74 ka) have been recorded in sedimentary columns. In most cases, the ashes that originated through those events are distinctly identified based on volcanic glass compositions (Song et al., 2000; Smith et al., 2011; Pearce et al., 2020). In the present study, the estimated concentrations of major oxides like Na₂O, K₂O, CaO, Al₂O_3, and Fe₂O₃ in glasses from the Mahanadi basin lie close to the ranges reported in YTT from other marine environments (Table 2); but substantially differ from MTT and OTT deposits though not perceptible in the bulk composition. The comparison with proximal ash samples from Toba caldera also showed the Mahanadi basin samples also differ from the values in glasses from typical MTT and OTT deposits and resemble more YTT (Table 2). However, in contrast, the estimated values of SiO₂ (75.2%–76.2%) in present analyses were substantially less than average values in typical YTT deposits (SiO₂ > 76.5% ,Table 2), while matches well with MTT or OTT, though SiO₂ may not be the discriminating factor. Such contrasting SiO₂ contents could be the result of post-depositional alterations of glasses in marine sediment. The experimental results of Gatti et al. (2014) indicated that in aquatic environments, weathering of volcanic glasses could cause significant leaching of Si from glasses with time.

Based on the calcareous nanofossil bio-stratigraphy, Flores et al. (2014) has shown that during the late Pleistocene age, the estimated sedimentation rate was ~135 m/Ma (i.e.,13.5 cm/ka) near that site. Thus following the age-depth model based on radiocarbon dating, the ash layer at about 9.64 mbsf at the site MD-19 should correspond to the age between 70–75 ka, very similar to the YTT event. Therefore, based on major element composition and information from the age-depth model, it could ascertain the dispersion of ash from the youngest Toba events as the source for ash layers in the Mahanadi basin.

The concentrations of trace elements in volcanic glass samples from the 5 cores of the Mahanadi basin are presented in Table 3. The results show, unlike major oxides, most of those trace elements have substantial variations and apparently point toward the finer scale compositional variety among those glass samples. In particular, the lithophilic elements like Ba (187–812)×10⁻⁶ and Sr (31–90)×10⁻⁶ have a wide range of concentrations in these glass shards and were consistent with the values found in YTT deposits [Ba: (50–1 400)×10⁻⁶; Sr: (0–150)×10⁻⁶, Pearce et al., 2020]. The concentrations of other lithophiles, Rb (111–272 )×10⁻⁶ and Cs (3.4–10.4 )×10⁻⁶(Table 3) were also substantially high, and most of them are comparable to the range of values in YTT described elsewhere [Rb = (253 ± 45)×10⁻⁶; Cs = (8.3 ± 2.5)×10⁻⁶; Pearce et al., 2020]. Earlier studies also showed that besides the absolute concentrations of Ba and Sr, their relative enrichment with respect to incompatible elements like Y in volcanic glasses reflects the subtle differences between the composition and evolution of magma during those Toba events (Pearce et al., 2020). It is also evident that in Figs 5a and b, the plots of Ba and Sr concentrations against Y in analyzed glass shard samples from the Mahanadi basin are compared with the data from other Toba deposits, which indicates glasses of the present study are very similar to YTT and strikingly differs from OTT or MTT deposits. In other known YTT deposits, the multiple similar clusters of Ba and Sr concentrations in volcanic glasses has referred to as diverse YTT populations with different degrees of magmatic evolution (Figs 5a and b; Westgate et al., 2013; Pearce et al., 2014, 2020). The mostly heterogeneous nature of Toba magma at the time of YTT eruption was primarily responsible for the development of such diverse YTT glass populations (Gatti et al., 2014). Thus the lower Ba/Y ratios (10.3–14.9, Table 3) in glasses from the cores MD-19, MD-23, MD-24, and MD-27 apparently belong to the intermediate cluster-III of YTT; and those differ from the higher ratio in samples from the core MD-20 (Ba = 812×10⁻⁶; Ba/Y: 22.4, Table 3); representing more evolved cluster-IV (Fig. 5a). Earlier studies suggested that Toba glass shards with high Ba content correspond to the magma source equilibrate at the deeper part of the magma chamber, while lower Ba values originate from late-stage emission of the larger volume of magma from shallow depths (Pearce et al., 2020). Likewise, the elevated Sr/Y ratio (= 2.48) in glasses from the southernmost core MD-20 was also similar to those reported in cluster-IV glasses of YTT (Sr/Y = 2.16 ± 0.43, Pearce et al., 2020), but the values of Sr/Y = 1.4–1.7 in other cores (Table 3) matches well with cluster-III of YTT deposits (Sr/Y = 1.57 ± 0.27; Pearce et al., 2020). The glass samples from the MD-20 cores, which originated from equilibrated magma at deeper depth, also showed higher concentration of elements (e.g., Cu, Zn, V, Pb), usually have higher mobility at elevated temperatures than those found in glasses from cores MD-19, MD-23, MD-24 and MD-27 (Table 3).

In contrast to lithophiles, the concentrations of relatively immobile high field strength elements (HFSEs) like Sc (1.8–2.9)×10⁻⁶, Ga (6.8–14.4)×10⁻⁶, Zr (71–176)×10⁻⁶; Hf (1.6–5.8)×10⁻⁶, Nb (7.1–18.4)×10⁻⁶, Ta (0.9–4.1)×10⁻⁶, was less in those glass samples (Table 3), but have comparable ranges reported in YTT from other marine environments (Srivastava and Singh, 2020; Pearce et al., 2020). It has also been noticed that, even at lower concentrations, these trace elements showed considerable variation among glasses from those ash layers. Studies indicate subtle variability of some HFSEs in glasses with minor overlaps representing a variety of magma compositions originating during the youngest Toba event (Pearce et al., 2020). In the present study, most of the estimated HFSEs in Mahanadi samples (except MD-23), match well with ranges (e.g. Zr = (106–169)×10⁻⁶; Hf = (4.06–7.55) ×10⁻⁶; Nb = (15.9–21.5) ×10⁻⁶; Pearce et al., 2020) reported in more evolved magma sources (of clusters III and IV) of YTT.

The rare Earth elemental compositions of volcanic glasses from five cores of the Mahanadi basin are listed in Table 4. The total REE contents (∑REE) in these glass samples vary within the range of (61–164) ×10^–6 with the dominance of lighter rare earth elements (LREEs > 90% of the total, Table 4). Like other immobile trace elements, the concentrations of most of the rare earth in these glass shards lie within a range very similar to those reported in other YTT deposits. Moreover, the chondrite normalized REE patterns (Fig. 6) of all glass shards had very similar trends with significant LREE (La-Sm) enrichment and relatively flat HREE (Gd-Lu) profiles, might be indicative of felsic rhyolitic magma source, as mentioned before. Most of these REE-patterns (except MD161-23, which marginally differs from the rest); closely resemble the YTTs from Toba caldera and other oceanic sources. In an earlier study, it has also noticed that the relative enrichment of LREE over HREE is more prominent (Nd_n/Yb_n = 1.8–2.7, estimated with data from Pearce et al., 2020) in glass populations (PIII-PV) of YTT, originated from less voluminous deeper magma sources. Thus, in the present study, the higher REE cross-group fractionations (Nd_n/Yb_n = 2.0–2.6) in glass shards might also have a similar deep magmatic origin. All the REE patterns were characterized by very marginal negative Ce anomalies (Ce/Ce* = 0.91–0.95; Table 4) but very distinct negative Eu-anomalies (Eu/Eu* = 0.28–0.37; Table 4). The observed Eu anomaly in REE patterns of glass shards (Fig. 6) implies early crystallization of plagioclase which usually restrict Eu incorporation from the melt during magmatic evolution and thus cause depletion of Eu relative to adjacent elements (Nakamura, 1974; Westgate et al., 1998; Song et al., 2000). Belousova et al. (2006) also stated that, the changes in Eu is consistent with simple fractional crystallization during magma cooling, as the evolving liquid becomes enriched in incompatible elements and feldspar crystallization leads to enhancement of the negative Eu anomaly.