Satellite images were acquired before and after the emplacement of the sulfur flows in 2019 (Figure 2), and videos were recorded during fieldwork while the flows were still moving (Supplementary Material S1). Satellite images were acquired in stereo mode by Pleiades, a French satellite tasking Lastarria, in 2016 and in 2022 (Figures 2A–D). The satellite captures grey scale panchromatic images (PM) with a resolution of 0.5 m, which is sufficient to depict prominent changes in fumarolic field 1 (Figure 2). We processed the data using the satellite processing function in Agisoft Metashape 2.0, using the rational polynomial coefficient (RPC) information for georeferencing. To ensure the accurate stacking of the two datasets, we defined ground control points at 24 identified locations. Following the dense cloud generation of 24 million points, digital elevation models and orthomosaics were generated. For this study, we concentrated only on fumarolic field 1 and searched for changes by visually comparing the 2016 and 2022 datasets (Figure 2). The location, dimensions, and topography of the sulfur flow area are important constraints for the further analysis and contextualization of other observations.

Video files of one active and advancing sulfur flow (flow #1 of the 2019-flows, Figures 3, 4) were recorded using a mobile phone (Xiaomi Redmi Note 7; Supplementary Material S1). The videos were stabilized and time-stamped image files were extracted for geometric reconstruction and velocity estimation. The pixel-to-meter transformation was performed using the widths of the flows observed in the orthomosaic data. For the geometric reconstruction, we used 767 extracted images in the structure-from-motion workflow using Agisoft Metashape 2.0, defined GCPs from the field to constrain the dimension (width of the front of 108 cm), and calculated 110,000 tie points. From this, an orthomosaic with a 0.5 cm pixel size, and a digital elevation model with a resolution of 2 cm were created. These results were used to constrain the morphometry of flow #1. For velocity estimation, we imported the image database into DAVIS (Lavision Inc.), a particle image velocimetry (PIV) approach that aims to search for point and region transformations in fluid dynamics regimes.

Temperatures were measured using a K-type thermocouple thermometer from Hanna Instruments (HI935002) connected to a Hanna Instruments flexible probe (HI766Z) built-in stainless steel to measure temperatures of up to 1,100°C. Temperatures were measured at the sulfur pools and at the proximal and distal sections of the sulfur flow (Figure 3).

The sulfur flow was sampled twice; once during the active flow in January 2019 (Figure 3) and again 3 months later (April 2019) when the flow had cooled and allowed closer inspection (Figure 4). First, the samples were introduced into glass flasks (Figures 3D,E) and cooled at ambient temperature (Supplementary Material S1). Once solidified, the samples were stored in sealed plastic bags and sent to the laboratory for X-ray diffraction, scanning electron microscope, and trace element analysis. In April 2019, a physical description of the solidified sulfur flows was provided.

X-ray diffraction was performed using a Bruker D8 Advance diffractometer at Unidad de Equipamiento Científico (MAINI, Universidad Católica del Norte, Chile). Before analysis, a representative fragment of the sample (∼20 g) was powdered with an agate mortar, sieved (<0.075 mm), and mounted in plastic holders. Analyses were done with an accelerating voltage of 40 kV and a current intensity of 30 mA, with Cu Ka radiation (I = 1.5406 Å) using a graphite monochromator and scintillation detector. Samples were diffracted at a 2θ angle of 3–70°, with steps of 0.020° and a 5 s integration time. Diffractograms were processed using the Bruker DIFRACT-SUITE software, which identifies and semi-quantitatively calculates the mineral phases and degree of crystallinity (amount of crystalline phases versus non-crystalline or amorphous phases).

Texture and quantitative microanalyses of the collected samples performed using a Hitachi TM-1000 environmental scanning electron microscope equipped with an energy dispersive spectrometer (ESEM-EDS) at the Laboratorio de Petrografía y Microtermometría of the Instituto de Geofísica UNAM. Selected samples and individual crystals were analyzed in detail using a field–emission scanning electron microscope (FE-SEM, Hitachi SU5000) to retrieve high-resolution images at the MAINI facilities. This device also included an Energy Dispersive X-Ray Spectrometer (EDS) for acquiring detailed chemical maps and accurate point analyses. Before the analysis, a representative fragment of the sample (<3 cm in size) was mounted on Al holders and coated with carbon. Samples were analyzed using an accelerating voltage of 20 kV and a 20–40 s integration time through the Hitachi SU5000 while an accelerating voltage of 15 kV and 37–43 s integration time were employed using the Hitachi TM-1000 equipment.

Trace element concentrations (As, Cd, Co., Cs, Cu, Li, Mo, Ni, Pb, Rb, Sb, Sn, Th, U, and Zn) were determined using high-resolution inductively coupled plasma mass spectrometry (HR-ICP-MS; Element 2XR Thermo Fisher Scientific). First, the samples were powdered and sieved (#270) at the Universidad Católica del Norte and then sent to the geochemistry laboratories of the Helmholtz Centre, Potsdam (GFZ, German Research Centre for Geosciences, Germany) for acid digestion and trace element analyses. The acid digestion procedure included the drying of powders (105°C); weighing into 15 mL Teflon vials (Savillex^®); and decomposed using HF, HNO₃, and HClO₄ (ULTREX^® II), following the protocol of Romer and Hahne (2010). Analytical results were supervised using internal (SCO-1; Romer and Hahne, 2010) and external (Imai et al., 1995) standards.

Stable sulfur isotope ratios were measured using a femtosecond laser ablation (fs-LA) system at GFZ Potsdam in combination with a multi collector ICP-MS (Thermo Fisher Scientific Neptune, equipped with a Neptune Plus Jet Interface) (Oelze et al., 2021). A laser beam (∼25 µm in diameter) was continuously scanned over the sample surface (ablation area of 100 × 100 µm) for 100 s, with subsequent background measurement for 150 s. Samples were quantified using the standard sample bracketing (SSB) approach with IAEA-S1 as the primary reference material. Every measurement session contained a range of reference materials (e.g., MASS-1 and Balmat pyrite) that were repeatedly analyzed between sample measurements for comparison with published S isotope values. Raw isotope data processing and background corrections were performed after applying several data rejection and acceptance criteria (e.g., Oelze et al., 2021). The most imperative data rejection/acceptance criteria were as follows: i) only the ^34/32S and ^33/32S ratios were used for the calculation, which deviates by < 3 s (standard deviation) from the sample mean; ii) only results that follow the mass-dependent terrestrial fractionation line in a three-isotope-plot of ^34/32S vs. ^33/32S within analytical uncertainties; and iii) had a mass bias drift between the two bracketing calibrators of <0.30‰ were accepted and reported in this study. We reported the sulfur isotope values (δ³⁴S) in delta notation as: δ 34 S = δ 34 / 32 S V C D T = ( 34 S / 32 S ) s a m p l e / ( 34 S / 32 S ) V C D T –   1 (1)

Satellite images acquired in 2016 and 2022 showed two sulfur flows that emerged at approximate elevations of 5,100 and 5,114 m above sea level (m asl), referred to as the 2016–2022- and 2019–flows, respectively (Figure 1C, 2D). The exact dates of the 2016–2022-flows are unknown, but can be constrained between 2016 and 2022, most likely between our field visits in April 2019 and February 2020. The 2016–2022-flows comprise several partially overlapping flow units, up to 55-m-long and 5.3 m wide, and emerge from a 16-m-wide fumarole cluster (at 68.52189° W, 25.15458° S). They terminated at 5,074 m, thus descending with an average dip of 25°.

At 5,114 m asl, the 2019-flows were better identified in our field photos and videos taken in January and April 2019 (Figures 5A, B). Four flows can be identified (Figure 4), and flow #1 is well recorded, allowing us to construct a three-dimensional model and determine the geometric characteristics. Sulfur flow #1 (cf. Records in Supplementary Material S1) reached a 9.5 m distance from the source, a maximum width of 108 cm, and an average thickness of 3 cm (Figure 5). The three-dimensional model shows that the thickness was not uniform with a deposition zone at the lower front (up to 4 cm thick). Colored orthomosaics show a dark brownish central area, surrounded by pale-gray sulfur deposits (Figure 5C). The dimensions of the flow structure were well determined, although the source at the fumarole vent is not visible here because of poor image quality. The morphology at the middle distance shows defined flow channels, which are most visible in slope maps and topographic profiles (Figures 5D,E), and shows minor deposition of material up to a thickness of approximately 1 cm and central erosion into the deposited materials (Figures 5D,E). The levees are up to 0.8 cm high, traceable over 180 cm, and oriented in the flow direction, representing traces of prominent shear zones during flow. The flow front fans out and thickens in the depositional zone with an irregularly shaped rope-like frontal thrust (Figures 3C, 5D).

Sulfur flow #2 presented two branches (east and west) reaching a distance of 12 and 9.8 m distance from the source, respectively, up to a width of 0.8 m and a maximum thickness of 5 cm. Sulfur flow #3 was shorter and thicker than the previous flows, being 7.8 m long, 7 cm thick, and up to 1 m wide near the front. Finally, sulfur flow #4 was the shortest and thinnest flow, 7 m in length, 1 cm in thickness, and up to 18 cm in width. Furthermore, contiguous areas (Supplementary Figure S2.2) show several sulfur flows sourced from the same site as the four flows shown in Figure 4. The sulfur flow (Supplementary Figure S2.2) exhibited poorly developed rope-like morphologies with an average thickness of only 2 cm. Field measurements of the #1, #2, #3, and #4 flows (Figure 4), and flows in the contiguous area (Supplementary Figure S2.2) are summarized in Table 1. Based on the maximum length, average thickness, and average width (Table 1), we concluded that the total volume of the 2019-flows was 1.45 ± 0.29 m³. The most important source of uncertainty regarding this volume is the average thickness because we have limited point measurements that vary between 1 and 7 cm, ignoring thickness fluctuations along the flow. Therefore, a relative uncertainty of 20% was considered for volume computations.

Cross-sections of the solidified 2019-flows revealed a generally high vesicularity (visual estimation of 30%–50%; Supplementary Figure S2.3), although fewer vesicular areas occasionally appeared towards the base of the flow, close to the contact with the yellowish substrate (Supplementary Figure S2.3C). Furthermore, the incorporation of accessory fragments of lithics and sulfur crystals of minor sizes (less than 0.5 cm) is also depicted in Supplementary Figure S2.3 and field observations.

The January 2019 video recordings showed the sulfur flow #1 in reddish-brown colors, with a continuous bubbling due to the constant gas input at the pool bottom, presenting a temperature of 158°C (Figure 3A). In the upper part, the sulfur pool overflow produced two molten sulfur channels moving downslope (11–15° on average, 24° maximum; Figure 3B) at an average speed of 0.069 m/s. Molten sulfur gradually slowed until it stopped because of the diminished terrain slope and partial solidification of the sulfur flow surface, as evidenced by the grayish crusts and rope-like textures (Figures 3, 5C; Supplementary Material S1). Subsequently, flow #1 showed a delta-type morphology and a measured temperature in the front of 124°C (Figure 3A, Figure 3C). Higher velocities were found at the center of the flows, where the margins showed large rotational shear components and the formation of small levees. As the surface is blocky and uneven, steep slopes represent higher velocities, locally reaching even 0.4 m/s. In contrast, the low–slope sections were related to lower velocities and the formation of sulfur puddles of up to 108 cm wide (Figure 3C). We note that, similar to sulfur pools, the sulfur flows showed intense bubbling as they moved downslope, especially in areas with fewer slopes where flows stagnated (Supplementary Material S1).

The total trace element concentrations in sulfur flow #1 (Figure 4) ranged from a few ppm to 42,268 ppm (Table 2). Arsenic reached the highest concentrations, especially in sulfur pools, with values up to five orders of magnitude higher than those of the other analyzed chemical elements. Lead, Bi, Cu, Zn, Rb, Sr, Zr, Sb, and Sn presented average concentrations between 1 and 56 ppm. In contrast, Li, Cd, Th, U, Co., Sc, Ni, Ga, Nb, and Ga had average concentrations of less than 1 ppm.

X-ray powder diffraction analysis of the sulfur pool (F1A, F1B) and sulfur flow (F2) samples showed the presence of orthorhombic native sulfur (Supplementary Material S3) with crystallinity degrees of 45% and 64% in the sulfur pool and sulfur flow samples, respectively. A more detailed inspection using high-resolution BSE (back-scattered electrons) imaging allowed us to identify the microtextures of the samples and the size/shape of the accessory fragments. SEM-EDS analysis detected Al, As, Fe, K, O, S, Si, Ti, I, and Pb, with S being the most abundant and displaying reticular and arborescent textures (Figure 6A–F). Additionally, distinctive crystalline phases <100 µm in size containing As, I, and Pb were observed. The As-bearing phases were identified as possible orpiments with iodine impurities, realgar, and arsenolite, whereas the Pb-bearing minerals were linked to galena (Figures 7C,D, Supplementary Material S4). Fragments containing Si and O, plus minor Fe, K, Al, and Ti of 20–200 µm in size, were ascribed to rock fragments (RF in Figure 6) incorporated during the movement of the sulfur flow over the substrate. Conversely, the presence of As, I, and Pb (Figure 7) can be ascribed to magmatic degassing.

The sulfur isotopic composition in the F2 sample was -6.10‰ vs. VCDT. This value is the first result for δ³⁴S reported for Lastarria volcano; it is fairly light compared to average MORB glasses (–0.91‰ ± 0.50‰; Fischer et al., 1998). The obtained value of –6.10‰ is heavier than the range observed for elemental sulfur at Poás (–12.3 to –9.4‰; Oppenheimer, 1992; Rowe, 1994) and within the range (–9 to +7‰; Ueda et al., 1979) obtained for 44 different volcanic sites in Japan. Then, our δ³⁴S value is closer to the lighter values found in floating spherules at Kawah Ijen (–4.2 to –1.4‰; Delmelle et al., 2000; Kusakabe et al., 2000) and subsurface native sulfur at Campi Flegrei (–5.5‰ to 0‰; Piochi et al., 2015). Conversely, isotope values at Lastarria contrast with those of 0 to +10‰ found at Nea Kameni Islet in Santorini (Greece; Hubberten et al., 1975).

Field observations and measurements allowed for better understanding of the physicochemical conditions controlling the emplacement mechanisms of molten sulfur. Measured temperatures of 124°C in sulfur flows and 140 and 158°C in sulfur pools (Figure 3) represent one of the few records of temperatures from these types of manifestations, which are higher than those reported by Ikehata et al. (2019) of 124.7°C in a yellow-amber molten sulfur pool at Hakone volcano. The temperature of molten sulfur has a direct relationship with viscosity. For example, pure sulfur melts at 119°C, and its viscosity decreases until 159°C, increasing rapidly by four orders of magnitude at approximately 160°C because of the polymerization of sulfur molecules (MacKnight and Tobolsky, 1965; Greeley et al., 1990; Oppenheimer, 1992). Although viscosity measurements were not available in this study, there were evident differences in this physical parameter between the flows and pools of sulfur, which could be associated with the in situ measured temperatures (Figure 3). For example, the sulfur pools showed variable viscosities, with those located in the upper part of fumarolic field 1 being more viscous than those feeding sulfur flow #1 (Figure 3A). This difference can be explained according to variable measured temperatures (140 vs. 158°C) and then the stronger influence of high-temperature gases feeding the bubbling and less viscous sulfur pool at 158°C. In contrast, sulfur flow #1 showed an intermediate viscosity compared to the two sulfur pools (Figure 3, Supplementary Material S1) and lower recorded temperatures (124°C), increasing its viscosity gradually with distance from the source and decreasing slope. In general, sulfur flows observed in Lastarria had similar bubbling processes and colors to the sulfur pool in the 15-2A fumarole at the Hakone volcano (e.g., Ikehata et al., 2019).

One of the most important features of molten sulfur observed at the Lastarria volcano in January 2019 was its dark brownish to reddish color. When molten sulfur occurs, the color is directly related to the temperature and viscosity (Takano et al., 1994). For example, according to the simple temperature-viscosity-color correlation, the dark-brownish color of our samples could suggest sulfur temperatures of ∼200°C and very high viscosities. However, our measured temperatures are within a narrow range of 124–158°C, and the sulfur flows seemed to be low viscous. Therefore, impurities within the molten sulfur can explain the dark brownish color and lower measured sulfur temperatures (Kargel et al., 1999). For instance, shades of red, brown, and orange in fumarolic deposits at the Lastarria volcano correlate well with As enrichment and As-bearing minerals (Inostroza et al., 2020). Therefore, high As contents (Table 2) and As-bearing minerals found in molten sulfur (Figure 7 and Supplementary Material S4) suggest that the brownish to reddish shades in the 2019-flows were more likely due to As impurities instead of an increased molten sulfur temperature. A similar conclusion was obtained by Kargel et al. (1999), who revealed substantial viscosity changes in sulfur when it contained impurities of As and other elements such as Cl and I. Consequently, impurities of As must be considered when studying the natural occurrence of reddish-to-orange molten sulfur manifestations (e.g., Greeley et al., 1990; Kargel et al., 1999).

The 2019-flows had significant concentrations of trace elements, with extremely high concentrations of As. The enrichment factor (EF) is one of the most important parameters to describe the enrichment of trace elements in a geochemical medium relative to a reference material (EF; Zoller et al., 1974). For the Lastarria volcano, the EF values for a given element from the sulfur pools and sulfur flow were compared with volcanic pristine host rocks according to the following expression: E F = X / Y s a m p l e / X / Y h o s t − r o c k s , (2) where “X” is the chemical concentration of a given element in the sample (i.e., molten sulfur) and host rocks while “Y” is the concentration of a reference chemical element in both the sample and host rocks. In this case, the reference element “Y” should be a refractory and non-volatile element with low concentrations in the sample and host rocks to enhance the EF values. Previous studies have selected Ti (Symonds et al., 1987), Al (Zoller et al., 1983; Varrica et al., 2000), Mg (Aguilera et al., 2016), Be (Moune et al., 2010), Th (Calabrese et al., 2011), or Sc (Olmez et al., 1986) as a chemical element of reference. In this case, Ti, Al, and Mg were not in our database (Table 2), whereas Th appeared to be slightly enriched compared to Sc, particularly in the sulfur flow sample. Therefore, Sc was selected as the reference element. The rock reference database was constructed from the average concentrations of lava and pyroclastic rocks from the Lastarria volcano (Table 2; Naranjo, 1992, Naranjo, 2010; Trumbull et al., 1999; Robidoux et al., 2020).

The Log EF values for the 22 trace elements varied between ∼0 and 6.4. The elements were divided into two groups (Figure 8). Group 1 includes those chemical elements with Log EF < 1 (Nb, Cs, U, Zr, Th, Rb, Ni, Co., Li, Ga, V, Sr, and Sc), characterized by volatilities close to zero at temperatures <400°C and, in general terms, affinity with silicate melts instead of S-rich fumarolic fluids. Similar (X/Y)_sample and (X/Y)_host-rocks ratios (Eq. 2) suggest that these elements were incorporated into the flow as rock particles, likely because of erosive processes in the fumarolic conduit or erosion of the substrate surface as the flow moved downslope. Given the vigorous fumarolic activity, these processes are highly plausible for this volcano. Accidental fragments were observed in the SEM-EDS images (Figure 6) and cross-sections (Supplementary Figure S2.3), which agrees with the opalized basalt fragments in sulfur flows at the Azufre volcano (Galapagos; Colony and Nordlie, 1973) and rock fragments in old sulfur flows at Lastarria (Naranjo, 1988).

Group 2 corresponds to trace elements with Log EF > 1 (As, Bi, Sb, Sn, Cd, Pb, Cu, Mo, and Zn), mainly dominated by chalcophiles; they have a strong affinity with sulfur. The high EF of Group 2 elements indicates the influence of a deep and/or shallow magma chamber beneath the volcano, i.e., feeding chalcophile elements, and that these elements show volatile behavior in fumarolic emissions at the Lastarria volcano. Among these elements, As stands out, with EF values six orders of magnitude higher than those found in the host rocks. Such enrichments agree with the As-bearing minerals found in the SEM-EDS images (Figure 7; Supplementary Material S4), in addition to the As-enrichments and As-bearing minerals found in reddish fumarolic deposits at the Lastarria volcano (Aguilera et al., 2016; Inostroza et al., 2020). Similarly, Sb and Bi also showed a very high EF, even though no mineral phases containing these elements were found in the Lastarria volcano. In addition to the trace elements mentioned above, such as Sn, Cd, and Pb, chalcophile elements that are usually enriched in subduction-related fumarolic gases (Edmonds et al., 2018) also showed a high EF. We highlight iodine, which was detected forming mineral phases through SEM-EDS analyses (Figure 7), suggesting that it is present in high concentrations in fumarolic deposits and gases. This agrees with the chemical analyses reported by Aguilera et al. (2016), where it appears as the most enriched trace element in condensed gases. Aguilera et al. (2016) found similar EF patterns in fumarolic deposit samples and volcanic gas condensates, demonstrating that Lastarria samples were enriched in Sb, As, Cd, Se, Ni, Pb, and Cu. Such a good correlation between the chemical composition of molten sulfur and fumarolic deposits suggests that the molten sulfur observed in January 2019 corresponds to the melting of previously existing fumarolic deposits and/or that all of these species were transported by the same gas phase (e.g., Symonds et al., 1992; Taran et al., 1995).

Arsenic, Se, Sb, and Cu have been found at high concentrations in molten sulfur, either in subaerial sulfur flows or at the bottom of acid crater lakes (e.g., Stoiber and Rose, 1974; Oppenheimer and Stevenson, 1989; Kargel et al., 1999; Xu et al., 2000; Daga et al., 2017). Moreover, Hg, Mo, Au, Co., and Fe are also present in significant concentrations in floating S-rich spherules in the Poás and Copahue crater lakes (Xu et al., 2000; Daga et al., 2017). Molten sulfur found at the Lastarria volcano contains significant amounts of chalcophile trace elements, in agreement with similar manifestations worldwide, such as Poás or Copahue (Oppenheimer and Stevenson, 1989; Daga et al., 2017). The high EF of chalcophiles in subaerial and subaqueous molten sulfur can be ascribed to their selective scavenging during sulfur formation, which is favored by their chemical affinity and continuous supply from the gas phase (Symonds et al., 1992). Furthermore, chalcophile scavenging by sulfur was identified in sulfur precipitates during the sampling of condensed gases (e.g., Fischer et al., 1998). Unfortunately, there is a lack of quantitative trace element analyses in molten sulfur samples, which hampers detailed comparisons between different volcanic systems.

As listed in Table 2, the As concentrations in the sulfur pools were six times higher than the same concentration in the sulfur flow, despite the fact that they were connected at a very short distance (<12 m between the sulfur pool and sulfur flow #1; Figure 3). This difference appears intriguing given the similar molten sulfur temperatures and colors of the pool and flow (Figure 3). The As-bearing crystalline and amorphous phases condense and precipitate or sublimate at temperatures lower than 300 °C (Mambo and Yoshida, 1993; Mandon et al., 2020), thus they can be efficiently trapped in S-rich melts. Consequently, the higher As concentration in the sulfur pools can be related to the condensation of the gas phase, whereas the lower concentration in the sulfur flow could be ascribed to As partitioning into the gas phase favored by continuous sulfur flow bubbling. This process should decrease the As concentration towards the flow front.

Considering the four sulfur flows shown in Figure 4 and the contiguous area with molten sulfur (Supplementary Figure S2.2), the total volume of molten sulfur approaches 1.45 ± 0.29 m³. This volume can be regarded as a modest value compared with the sulfur flows previously reported at the Lastarria, Azufre, Shiretoko-Iozan, and Poás volcanoes (Table 1). The volume estimated for the 2019-flows could increase considering the 2016–2022-flows (Figures 1, 2), which occurred during the same period; therefore, the accumulated sulfur volume may be significantly higher, although we do not consider that it could reach the volume of old flows at Lastarria. Pure sulfur was characterized by a density of 2.07 g/cm³. However, the number of vesicles and accessory lithics decrease and increase this value, respectively. The approximate density of sulfur flows can be obtained using the percentages of sulfur, rock particles, and vesicles, as shown in Figure 6 and Supplementary Figure S2.3. Considering 50, 15, and 35% of sulfur, rock particles, and vesicles, respectively, a density of 1.43 g/cm³ was calculated, which implies a mass of molten sulfur of 2.07 ± 0.41 tons for the 2019-flows.

Although several volcanoes are known for the presence of flows or pools of sulfur, the emplacement and speed of molten sulfur have only been observed in a few of them (e.g., Shiretoko-Iozan; Watanabe and Shimotomai, 1937; Watanabe, 1940). Video records at Lastarria volcano (Supplementary Material S1) indicate that the sulfur flow #1 (Figures 3, 4) was emplaced at an average speed of 0.069 m/s over a slope of 11–15°. However, the speed decreased to ∼0.001 m/s when the flow reached flatter areas (<10°). These speeds are lower than those of 0.12 m/s estimated by Naranjo (1988) and also lower than those of 0.1–1.05 m/s reported in the case of pure molten sulfur (99.6% of purity) produced through the Harsh method (industrial sulfur flows produced by the injection of superheated water into wells drilled in sulfur-rich sediments; further details in Greeley et al., 1990). For the 2019-flows, field measurements and sampling took approximately 120 min to complete while sulfur flow #1 moved slowly downslope by approximately 2 m, indicating an average speed of 0.00028 m/s. Independent of these subtle speed variations, forward speeds at the sulfur flow front decreased notably because of the solidification of sulfur at temperatures close to 119°C (Figure 3C), producing layer stacking of semi-molten sulfur, forming lobes, and rope-like textures (Supplementary Video S1; Figure 5).

The computed speeds for the 2019-flows are clearly lower than those previously calculated/reported for other volcanoes (e.g., Shiretoko-Iozan, Lastarria, and Azufre volcanoes), suggesting that the time at which the longest sulfur flows occurred in these volcanoes may be notably underestimated. Accordingly, information on these times is limited. Indeed, recorded times have been estimated to be in the range of 30 min to 4 h depending on the volume of the sulfur melt, or by up to 5 days of discontinuous emission of molten sulfur, which occurred at Shiretoko-Iozan (e.g., Watanabe and Shimotomai, 1937; Watanabe, 1940; Naranjo, 1988).

By assuming that the high-speed zone (i.e., slope of approximately 20°) is located within the first 3 m from the source and maximum speeds were only reached in this area, whereas the low-speed zone was located at distances greater than 3 m, we can obtain an approximation of the time it took for the four flows presented in Figure 4 to reach the site where they were found in April 2019. For the first 3 m, the high-temperature and poorly viscous sulfur flowed down in only 44 s and then entered the low-speed zone until stopping. Within this flatter zone, sulfur flow #1 remained in motion for at least 108 min, sulfur flow #2 for 150 and 113 min (east and west branches, respectively), sulfur flow #3 for 80 min, and sulfur flow #4 for 67 min (Table 1). These computations suggest that the molten sulfur flows were active for at least 408 min (6.8 h), assuming that each flow occurred immediately, one after the other, without pauses. Using computed volume and “displacement times” of the molten sulfur, emission rates result in a range of 0.13–0.22 m³/h for flows #1, #2, and #3 while flow #4 showed a speed of only 0.009 m³/h (Table 1). Such emission rates differ notably from the 64.8 m³/h estimated for the old flows at Lastarria (Naranjo, 1988) and ∼20 m³/h established for the Shiretoko-Iozan flow (Table 1; Yamamoto et al., 2017).

In general terms, the active sulfur flows observed at the Lastarria volcano appear to have different rheological properties from those reported for other volcanoes because they have moved slower and for a more extended time range. Then, the 2019-flows appeared in a very specific area with a limited total volume, suggesting the occurrence of local processes, such as fracture opening, instead of general heating of the fumarolic field. However, records of unavailable temperature variations could have helped constrain these hypotheses. Therefore, possible changes in the fumarole temperature cannot be completely ruled out.

Elemental sulfur can be formed via several pathways. For instance, at relatively low temperatures (<350°C), native sulfur condenses in the presence of SO₂ and H₂S according to the following reaction 3) (Mizutani and Sugiura, 1966; Giggenbach, 1987): S O 2   g + 2 H 2 S   g = 3 S 0   l / s + 2 H 2 O   g . (3)

Under similar temperature conditions, solid/liquid native sulfur can also be produced by SO₂ disproportionation in the presence of H₂O according to reactions 4) or 5) (Oppenheimer, 1992): 3 S O 2   g + 2 H 2 O   g / l = 2 H 2 S O 4   a q + 1 / x   S x   l / s   a n d (4) 4 S O 2   g + 4 H 2 O   g / l = 3 H 2 S O 4   a q + H 2 S   g . (5)

Reaction 3) has been primarily used to explain the formation of S-rich subaerial fumarolic deposits, whereas reactions (4–5) describe the formation of molten sulfur in subaquatic environments or volcanoes with well-developed hydrothermal systems (Oppenheimer, 1992). Furthermore, reactions (3–5) produce sulfur enriched in ³²S and then with δ³⁴S values <0 (Oana and Ishikawa, 1966; Ohmoto and Lasaga, 1982; Delmelle and Bernard, 2015), with the SO₂ disproportionation process (Reactions 4–5) producing native sulfur with strongly negative δ³⁴S values (as light as –12.3‰). However, for Lastarria volcano, the least negative δ³⁴S value (δ³⁴S = –6.10‰) differed from more negative values at Poás volcano. Additionally, reactions (4–5) require significant amounts of water inside the volcanic edifice (Delmelle and Bernard, 2015), contrary to what is thought for the Lastarria volcano, where the amount of water seems to be very limited (Aguilera et al., 2012). Therefore, reaction 3) can be postulated as an appropriate candidate for explaining the formation of molten sulfur, whereas SO₂ disproportionation (reactions 4 and 5) is an implausible process during the 2019-flows.

Inspection of historical data on fumarolic gases discharged at the Lastarria volcano (Aguilera et al., 2012; Layana et al., 2023) permits a better understanding of the physicochemical processes behind the 2019-flows and the feasibility of reaction 3) to form solid/liquid sulfur. Gas species involved in reaction 3) had the following behavior in 2006 and 2022 (excluding 2019 samples; all concentrations expressed in mmol/mol): SO₂ = 5.9 ± 0.9; H₂S = 2.5 ± 0.4; H₂O = 876 ± 4.6; SO₂/H₂S = ∼2.3 (Layana et al., 2023). However, fumarolic gases collected during the 2019-flows contained the following concentrations: SO₂ = 38 ± 2; H₂S = 4.6 ± 0.6; H₂O = 695 ± 5.2; SO₂/H₂S = ∼8.4 (Table 3). Notably, the samples collected in January 2019 (Table 3) showed the highest SO₂ and lowest H₂O concentrations ever recorded at the Lastarria volcano while the H₂S concentration almost doubled. The very high SO₂ compared to H₂S concentrations can be explained by the consumption of 2 mol of H₂S for every mole of SO₂ in reaction 3), depleting volcanic gases in H₂S and increasing the SO₂/H₂S ratios from an average of ∼2.3, according to historical data, to ∼8.4 in the 2019 samples (Table 3).

Following reaction 3) from Giggenbach (1987), solid/liquid sulfur formation is favored when fumarolic gases contain high concentrations of sulfur compounds and lower steam concentrations. Using the historical fumarolic gas dataset from Layana et al. (2023), detailed information was obtained by plotting the equilibrium constant for the deposition of elemental sulfur (Reaction 3) and the outlet gas temperatures (Figure 9). From this plot, the correspondence between the chemistry of fumarolic gases and molten sulfur observed in 2019 can be depicted, given that the gas samples fell into the liquid sulfur field. Remarkably, all of the >120°C gas samples collected after 2014 are similar to the 2019-samples in terms of the chemical composition (Figure 9), suggesting that the deposition of sulfur (liquid) could be favored by the chemistry of the fumarolic gases since this date. Simultaneously, Reaction 3) should produce significant amounts of H₂O, which contradicts the low H₂O concentrations in the 2019 samples. This contrasting evidence could be ascribed to subsurface steam condensation or that the 2019-samples were collected in the presence of a depleted hydrothermal system, a process induced by an increase in degassing from a shallow magma chamber (Lopez et al., 2018; Layana et al., 2023).

The question that arises is why liquid sulfur was not frequently observed during the sampling procedures after 2014. Only the January 2019 flows were active, besides another 55-m-long sulfur flow that was active sometime between 2016 and 2022 (Figure 2D). Occasional visits to this volcano (only a few days every year) owing to access difficulties reduce the possibility of observing this phenomenon. Despite favorable physicochemical conditions for native sulfur formation, we believe that molten sulfur manifestations are sporadic events triggered by mechanical effects, such as the opening of new fractures or changes in host rock permeability (e.g., Rouwet et al., 2017). This sporadicity also explains the occurrence of molten sulfur only in specific sections of the fumarolic field, weakening the theory that sulfur flows are produced by the general melting of previously formed sulfur crusts owing to the heating of the fumarolic field. However, partial melting of sulfur crusts was recorded on the walls of the 158 °C sulfur pool (Supplementary Material S1; Figure 3A), implying that favorable physicochemical conditions for fumarolic gases and then partial melting of fumarolic deposits are behind the formation of pools and flows of molten sulfur.

Volcanoes such as Poás, Turrialba, and Copahue (González et al., 2015; Daga et al., 2017; Salvage et al., 2018) have shown molten sulfur before or during eruptive periods, especially during phreatic activity, prompting researchers to link these phenomena and even suggesting the occurrence of molten sulfur as a precursor of volcanic activity. For the Lastarria volcano, there are no historical records of eruptive activity. However, ongoing ground deformation in the Lazufre area (uplift of up to ∼3 cm per year) and significant changes in the chemistry of fumarolic gases place this volcano in continuous unrest, which has been attributed to a pressurized magma chamber located at depths of 7–15 km; changes in the chemistry of fumarolic gases could be explained by a sequence of events that produced the acidification and depletion of the hydrothermal system, allowing the passage of less scrubbed magmatic fluids (Layana et al., 2023). Coincidentally, the 2019-flows occurred during a period that seemed to have started in 2014, when the volcano reached its peak activity, emitting high concentrations of SO₂, H₂S, HCl, and HF (Layana et al., 2023). The ongoing unrest satisfies the physicochemical conditions necessary for the formation of molten sulfur, and the occurrence of new sulfur flows cannot be ruled out. Consequently, molten sulfur witnessed at Lastarria volcano seems to be more related to ongoing unrest than to precursory eruptive activity, so it cannot be directly linked with eruptive activity.