Solfataric alteration at the South Sulfur Bank, Kilauea, Hawaii, as a mechanism for the formation of sulfates, phyllosilicates, and silica on Mars

The surface of Mars is shaped by volcanism and impacts, many of which created hydrothermal environments and sulfate brine alteration. The minerals formed under hydrothermal conditions or through sulfate brine activity on Mars can help us trace back the ancient geochemical history of surface and near-surface environments. Investigating terrestrial hydrothermal environments provides geochemical context for a better understanding of martian processes that may have supported warmer settings with liquid water and habitable conditions (e.g., Walter and Des Marais 1993; Schulze-Makuch et al. 2007; Black and Hynek 2018) on an otherwise arid and cold planet. Evidence collected by the Mars Exploration Rover (MER) Spirit suggests the actions of hydrothermal processes contributed to the unusual sulfates at Paso Robles class outcrops (Johnson et al. 2007; Lane et al. 2008; Yen et al. 2008; Parente et al. 2009) and the high-silica environments at Home Plate (Squyres et al. 2007; Ruff et al. 2011; Ruff and Farmer 2016). Numerous outcrops of sulfates in the greater Valles Marineris region are products of sulfate brines, especially where gypsum and jarosite are observed (Bishop et al. 2009; Murchie et al. 2009; Roach et al. 2010; Weitz et al. 2011; Thollot et al. 2012; Bishop et al. 2021a). Smaller patches of Ca sulfates and jarosite/alunite are associated with phyllosilicates at Mawrth Vallis (Bishop et al. 2020, 2023), Noctis Labyrinthus (Weitz et al. 2011, 2013), and Cross crater (Ehlmann et al. 2016), indicating alteration with sulfate brines or hydrothermal processes.

Burns (1988, 1993) proposed clay-sulfate-iron oxide/oxyhydroxide assemblages on the surface of Mars that could be derived from aqueous alteration of volatile- and iron-rich basaltic rocks containing fayalitic olivine, iron-rich pyroxenes, and accessory sulfides. Burns (1988) emphasized the importance of these materials for understanding martian geochemistry by paraphrasing Jensen and Bateman (1979): These minerals “arrest attention and incite interest as to what they may mask …The finding of one may herald the discovery of buried wealth.” Acidic groundwater was proposed to react with igneous silicate minerals and iron sulfides to yield jarosite, alunite, and other minerals (Burns 1987; Burns and Fisher 1993). Iron sulfides and igneous silicate minerals are plentiful in martian meteorites (e.g., Bunch and Reid 1975; McSween 1994; McCubbin et al. 2009) and have been inferred as martian surface components since Vikings’ measurements (e.g., Toulmin et al. 1977; Clark and Baird 1979; McSween 2002), and dissolved volcanic gases on Mars would have produced acidic groundwaters (Gooding 1978). More recently, Moore and Szynkiewicz (2023) built upon this idea to describe a sulfur cycle whereby volcanism contributes sulfur to the groundwater and low-temperature chemical alteration produces sulfates in aqueous surface environments.

Elwood Madden et al. (2004, 2009, 2012) extended this early work by Burns through modeling the conditions supporting formation of jarosite. These studies support formation of jarosite together with iron oxides/hydroxides and clays on Mars as predicted by Burns (1988). Black and Hynek (2018) summarize a host of sulfates including jarosite and gypsum that form together with phyllosilicates and iron oxides/hydroxides in terrestrial analog environments. Recent brine experiments demonstrate that gypsum forms readily through the action of sulfate brines on smectites (Geyer et al. 2023). Coordinating field observations with lab investigations using instruments at Mars furthers our ability to connect the geochemical environments governing mineral formation with specific sites on Mars. Trace amounts of jarosite are observed in the Murray formation along the lower region of Mount Sharp in Gale crater (Rampe et al. 2020), while gypsum and Ca sulfates (anhydrite, bassanite, gypsum) are more commonly observed at higher elevations on Mount Sharp (Vaniman et al. 2018). CheMin XRD analyses of the lacustrine Marimba and Sebina samples indicate combinations of jarosite, gypsum, smectite, and amorphous phases (Achilles et al. 2020). Marimba contains 1.7 wt% jarosite, 3.5 wt% gypsum, 19 wt% smectite, and ~50 wt% amorphous phases, while Sebina includes 2.5 wt% jarosite, 6.6 wt% gypsum, 28 wt% smectite, and ~40 wt% amorphous phases (Achilles et al. 2020).

Solfataric alteration of basalts and volcanic ash may have played a role in sulfate mineralization on Mars through fumarolic action from vents (e.g., Morris et al. 2000; Schiffman et al. 2006; Bishop et al. 2007). Solfataric processes are defined here as alteration through fumarolic activity and/or “acid-fog” deposition occurring as the sulfuric fumes emanating from vents react with the ash and tephra (Schiffman et al. 2006). Solfataric alteration promotes silicification of basalt and also the formation of sulfates, amorphous silica, and layer silicates, resulting in partial consolidation of ash/tephra particles and/or crust formation on ash surfaces (Schiffman et al. 2006). Fumaroles in the Kilauea caldera, Hawaii, have created a solfataric bank on the south wall of the crater through alteration of the Keanakako’i ash that was previously deposited there. This study focuses on field observations of the “South Sulfur Bank” Kilauea solfatara site, lab measurements of samples collected there, and implications for hydrothermal alteration and the action of sulfate brines at solfatara sites on Mars.

The solfatara site studied here was located in a light-toned pyroclastic ash bed underneath the 1982 flow along the southern rim of Kilauea caldera at an elevation of 1104 m at 19°24′12.9″N and 155°16′28.7″W (Fig. 1). Lava flows in 1974 and 1982 near the Kilauea caldera are dominated by Pahoehoe lava and extend southwest to Mauna Iki and southeast to Mauna Ulu. These flows are rich in olivine and calcic feldspar and typically contain only small amounts of pyroxene (Wolfe and Morris 1996).

Our study site was termed South Sulfur Bank to differentiate it from the Sulfur Banks site at the north side of Kilauea inside Hawaii Volcanoes National Park. Both sites contain vents exuding sulfuric gases. The South Sulfur Bank outcrop formed east of the Halema’uma’u crater and west of the Keanakako’i crater and was part of the Keanakako’i Ash Member from the 1790 flow (Fig. 2). The solfatara outcrop was composed of thin layers of basaltic pumice and was present in some localities as 10-m thick pyroclastic deposits (Wolfe and Morris 1996; Schiffman et al. 2000). The sequence contained bedded vitric ash with cross-bedded deposits of lapilli and blocks that were coarser than the ash. At the base of the solfatara outcrop several small vents were present that periodically exhibited wisps of gases and/or steam. The Halema’uma’u crater became active again in 2008, forming a lava lake and steam. In 2018, seismic activity increased, fissures formed, and the wall southeast of Halema’uma’u crater containing this South Sulfur Bank site collapsed (https://www.usgs.gov/volcanoes/kilauea/2018-lower-east-rift-zone-eruption-and-summit-collapse; latest revision November 24, 2023).

Members of our team visited the site in 2003, 2004, and 2006 with assistance from the Hawaiian Volcanoes National Park (HAVO) staff and permits issued to Schiffman in 2003 and Bishop in 2004 and 2006. We accessed the site by hiking northeast across the 1982 flow from the rim road down toward the caldera. Our study site focused on the thick pyroclastic deposit of the South Sulfur Bank site (Figs. 2–3). Alteration of the basaltic flow above the South Sulfur Bank produced orange- to tan-colored fine-particulate material that seeped downhill in some places and pigmented portions of the solfatara unit (Figs. 2e and 3a).

The solfatara outcrop is composed of orange/tan-colored altered material below the brown basalt and wide expanses of white/beige material extending for ~20 m at the thickest location with a height of ~3 m. The brownish-orange altered basalt from the 1974 flow above our field site was investigated recently and contains silica, iron oxides, and sulfates in addition to basalt (Yant et al. 2017). The black 1982 flow covered the 1974 flow (altered to brownish color) above our field site and at the base of the field site, including portions of the white/beige material. A slight smell of sulfur at the site was noted at each visit to the site (2003, 2004, 2006) near several small, weakly active hydrothermal vents in the direction of the floor of the Halema’uma’u crater. Thin columns of steam and/or gases emanated periodically from a few small vents consistent with low activity levels in the caldera.

Samples were collected from the bedded pyroclastic deposit and the powdery friable deposit of the South Sulfur Bank site during three field seasons in 2003, 2004, and 2006 and are summarized in Table 1. The locations of these samples are shown in Figure 3. Layers of alternating white/light and orange/gray material were collected in 2003 (Fig. 3a) to investigate changes in chemistry and mineralogy of the layers. An ASD FieldSpecPro spectrometer (described below) was used at the field site in 2004 and 2006 to select samples from the layered pyroclastic deposits (tan/gray/orange materials), as well as the white/beige friable unit. These additional materials were collected in 2004 and 2006 to characterize the altered solfatara from multiple regions of the outcrop.

VNIR reflectance spectra were acquired both in the field and in the lab using an Analytical Spectral Devices (ASD) FieldSpecPro spectrometer with Spectralon as the white standard. Spectral sampling is 1.4 nm from 0.35 to 1.05 µm and 2 nm from 1.05 to 2.5 µm with a spectral resolution of 3 nm in the visible region and 10 nm in the near-infrared (NIR) region. Spectra were measured in the field using the 1° optics and the contact probe. The contact probe configuration collects spectra for a circular region ~1 cm in diameter and provided sufficiently clean spectra for mineral identification in the field. Additional spectra were measured in the lab of samples from multiple sites along the surface of whole rock samples to guide sampling of the rocks for further analyses.

Emission spectra were measured of rock samples at the Mars Space Flight Facility at Arizona State University (ASU) using a Nicolet Nexus 670 E.S.P. FTIR interferometric spectrometer, as in previous studies (Christensen et al. 2000). This spectrometer was modified for emission measurements and used a thermoelectrically stabilized DTGS detector and a CsI beam splitter for measurement of emitted radiation over the mid-infrared range of 4000 to 200 cm^–1 where thermal infrared (TIR) provides a strong signal for rocks and coarse-grained samples (Lane and Bishop 2019).

The soil and rock samples were divided by color and spectral properties and gently crushed and dry sieved to <125 µm for further study in the lab. These particulate materials were used for elemental analyses, XRD, SEM, Mössbauer spectroscopy, and additional reflectance spectroscopy measurements at the Reflectance Experiment Laboratory (RELAB) at Brown University.

Samples with sufficient material were sent to the Bureau Veritas labs (https://www.bvna.com/) for X-ray Fluorescence (XRF) determination of the major elements as oxides. Total C, total S, and sulfate abundance were also determined using a LECO elemental analyzer. These include JB629, JB631, JB632, JB775, JB776, and JB777. Loss on ignition (LOI) values were determined by heating to 1000 °C.

Wavelength-dispersive scanning electron microprobe (SEM) analyses were performed on epoxide-impregnated, polished sections or particles smear-mounted on double-sided sticky carbon tape coated with carbon for samples JB629, JB630, and JB631 at UC Davis as in previous studies (Bishop et al. 2002; Schiffman et al. 2000, 2002, 2006). Backscattered electron (BSE) images were acquired and evaluated for petrographic features and degree of consolidation of the textures. Quantitative, wavelength-dispersive analyses by electron microprobe analysis (EMPA) were conducted using a Cameca SX-100 microprobe operated at 15 keV, 5 or 10 nÅ beam current, and variable spot size to map out changes in elemental composition across the samples.

X-ray diffraction (XRD) analyses were performed at UC Davis on oriented aggregates of the particulate material from samples JB629, JB632, JB775, JB776, and JB777 mounted on glass slides with a Diano 8000 diffractometer (CuKα radiation) as in previous studies (Schiffman et al. 2002, 2006). Additional XRD analyses were performed on powdered aliquots of samples JB632 and JB633 at the Bavarian Geologic Survey using CuKα radiation on a Bruker D8 instrument equipped with a sample spinner and a diffracted-beam graphite monochromator as in previous studies (e.g., Murad and Rojik 2003; Bishop et al. 2004).

VNIR to mid-IR reflectance spectra were measured for the particulate samples at RELAB as described in previous experiments (Bishop et al. 2007). Bidirectional VNIR spectra were measured relative to Halon under ambient conditions from 0.3 to 2.5 µm. Biconical reflectance spectra were measured from ~1 to 25 µm relative to a rough gold surface using a Nicolet 740 Fourier transform infrared interferometer (FTIR) in a H₂O- and CO₂-purged environment to remove adsorbed water from the samples. Composite, absolute reflectance spectra were prepared by scaling the FTIR data to the bidirectional data near 1.2 µm. The spectral resolution is 5 nm for the bidirectional data and 4 cm^–1 for the FTIR data.

Mössbauer measurements were taken at 295 K on a Web Research Co. W100 spectrometer on samples JB632 and JB633 mixed with sucrose using a ~10 mCi ⁵⁷Co source in rhodium from –4 to +4 mm/s velocity. Run times ranged from 24–48 h. Spectra were collected in 2048 channels, and then all spectral baselines were corrected for the Compton scattering of 122 keV gamma rays by electrons inside the detector. The corrected data are equal to A/(1 – b), where A is the counts of the uncorrected absorption and b is the Compton fraction determined through recording the counts with and without a 14.4 keV stop filter (~2 mm Al foil) in the gamma-ray beam. While this correction does not change the results of the fits, it does allow for accurate determination of the percentage of absorption in each spectrum. Each spectrum was then folded and corrected for nonlinearity in WMOSS. Interpolation to a linear-velocity scale was accomplished using the room temperature spectrum of 25 μm α-Fe foil as a calibration standard.

The Mössbauer spectra were fit using the Dist3e program as in Dyar et al. (2008). The widths of quadrupole doublets or magnetic sextets were constrained to vary together, while doublets were constrained to have two equal peaks and sextets to have peaks in the ratio 3:2:1:1:2:3. Quadrupole splitting (QS) values of doublets and magnetic fields of sextets were varied over a range, and isomer shift (IS) values and peak widths were either fixed or left unconstrained. These IS and QS values can be used to identify or constrain the presence of Fe-bearing minerals in samples (Dyar and Sklute 2019).

Elemental trends and XRD analyses for samples of the white/beige friable materials and the orange/tan bedded units were similar for samples collected on different field trips. Bulk elemental analyses are provided in Table 2 for samples with sufficient material (JB629, JB630, JB631, JB632, JB633, JB775, JB776, and JB777), and some samples were measured twice to check for consistency among different sample aliquots. The white/light samples (Type 1) include primarily SiO₂ (~74–81 wt%), TiO₂ (3–11 wt%), and SO₃ (0–2 wt%) as sulfate and elemental S with additional minor/trace amounts of Al, Fe, Mg, Ca, and C (Table 2). These samples also include loss on ignition (LOI) values ranging from ~12–15 wt% (includes C and S values) that are consistent with the presence of opal in these samples. The layered orange materials (Type 2) have lower SiO₂ (~39–51 wt%) and TiO₂ (2–3 wt%) levels and higher amounts of the major geologic elements, consistent with the altered portions of Keanakako’i_ash from Schiffman et al. (2000). The Type 3 gray and orange sample has lower SiO₂ (~42–43 wt%) and higher MgO (10–13 wt%) levels compared to the layered orange Type 2 samples and is more similar to the glassy grains of the Keanakako’i_ash (Schiffman et al. 2000). Type 4 contains primarily elemental S with ~16 wt% SiO₂.

Microprobe analyses of several regions of samples JB629, JB630, and JB631 are reported in Online Materials¹ Table S1 and average values for each type of composition are presented in Table 3 for comparison with the bulk chemistry for those samples. SEM analyses of ash and tephra from the Kilauea region show replacement of glass and formation of rinds in the ash material (Schiffman et al. 2000). Much of the material in the glassy grains and rinds in the friable white/light material (Type 1 samples JB629, JB632, and JB776) was replaced by silica and the SiO₂ abundance is significantly elevated for these samples compared with that of the fresh glass and orange/tan samples JB630 and JB777 (Type 2), and JB631 (Type 3) (Tables 2–3). XRD shows these bright, friable samples to be primarily opal (Table 1; Fig. 4), which is consistent with the ~80 wt% SiO₂ observed. Microprobe analyses of an aliquot of sample JB629 displays these glassy grains and rinds that were replaced by silica (Fig. 5a). A portion of sample JB629 containing yellow grains reveals the presence of native sulfur (Fig. 5b). This is consistent with the ~66 wt% S and ~16 wt% SiO₂ for JB775 that is mostly yellow crumbles separated from the friable white/light material. XRD also identified anatase in samples JB629 and JB776, which is consistent with the elevated TiO₂ levels (Table 2). Limited C is observed in all samples (≤0.5 wt%), constraining the amount of organics or carbonates that can be present (Table 2).

XRD measured gypsum, jarosite, saponite, and nontronite in the layered light gray and orange materials (Table 1; Fig. 4). Microprobe analyses of Type 3 sample JB631 (Fig. 5c) found glassy grains and rinds as for Type 1 JB629 (white/light) and Type 2 JB630 (layered orange) as well as smectite rich in Mg and Fe, and bright phases extremely high in Fe, and also high in S, P, and Al. These bright materials include jarosite, as well as poorly crystalline iron oxide phases not detected by XRD and likely also amorphous phases containing P. The abundance of these bright Fe-sulfate grains is much higher in the microprobe analyses of Type 3 sample JB631 enabling determination of their chemistry, including ~55–60% Fe₂O₃, ~7–12% Al₂O₃, ~12–13% SO₃, and ~3% P₂O₅. Elemental variations across these bright spots point out nonlinear Fe/S correlations, indicating that S is present in other materials besides jarosite or that some of the jarosite contains high-Al levels, similar to alunite. Gypsum is also observed in selected Type 2 bedded orange materials through XRD (Fig. 4), microprobe analyses (Fig. 5), and VNIR reflectance spectra (Fig. 6) and is likely present in the lighter-colored grains in these primarily orange/tan samples. Sample JB777 has elevated SO₃ levels and also higher Fe₂O₃ and CaO, consistent with Ca sulfates in addition to jarosite in that sample.

The XRD results for Type 2 orange layered samples JB633 and JB777 show stronger peaks due to saponite or hectorite than those due to nontronite (Fig. 4), which is consistent with roughly three times more MgO than FeO^T. Interestingly, a combination of hectorite, saponite, and nontronite is needed to account for the smectite peaks in these samples. All three smectites include XRD peaks near 19.4–19.8, 34.5–35, and 53 °2Θ, while only saponite and nontronite have a peak near 5.7–6 °2Θ, only hectorite and nontronite have a peak near 29 °2Θ, and only saponite and hectorite have a peak near 61 °2Θ. Microprobe analyses of Mg-rich smectite in sample JB631 indicate 50.4% SiO₂, 17.3% Al₂O₃, 14.4% MgO, 4.5% Fe₂O₃ (as Fe³⁺ and/or Fe²⁺), 0.1% CaO, 0.9% MnO, 0.7% P₂O₅, and <0.1% of other oxides (Table 3; Online Materials¹ Table S1). Additionally, the glassy grains examined in microprobe analyses of JB630 and JB631 (Table 3; Online Materials¹ Table S1) contain ~13 wt% Al₂O₃, ~11 wt% FeO^T (where the total Fe is expressed as FeO, but both Fe²⁺ and Fe³⁺ are likely present), ~8 wt% MgO, and ~10 wt% CaO that are almost entirely lost through alteration to form the altered glassy grains and rinds of JB629. Ti is present in nearly all of the samples and is likely present in the basaltic glass for the orange/gray layered samples (Table 1).

VNIR reflectance spectra of rock samples measured under ambient conditions demonstrate compositional variations among the samples investigated, although the water bands are strong due to adsorbed H₂O molecules on the surfaces (Fig. 6). The lighter-colored samples contain broad bands centered near 1.41–1.47, 1.91–1.95, and 2.21–2.27 µm due to abundant opal and some montmorillonite, jarosite, and gypsum, but no alunite was observed (e.g., Bishop 2019). The spectral features near 1.4 µm include the Si/Al-OH stretching overtone at 1.41 µm, the bound H₂O stretching overtone at 1.41 µm, and the lower energy adsorbed H₂O stretching overtone near 1.44–1.45 µm (Bishop et al. 1994). A shoulder on the short-wavelength side of this band near 1.37 µm is attributed to the Si-OH stretching overtone for dehydrated opal (Anderson and Wickersheim 1964). The band at 1.91 µm is due to a combination of H₂O stretching and bending vibrations for tightly bound water molecules in opal and smectites, while the shoulder near 1.94–1.95 µm is due to the H₂O combination band for adsorbed or more loosely bound water molecules (Bishop et al. 1994). Additional H₂O overtones occur as weaker bands near 0.96 and 1.15 µm. The band near 2.21–2.27 µm is due to a combination of structural OH stretching and bending vibrations and occurs near 2.21 µm for Si/Al-OH and at longer wavelengths for Fe/Mg-OH (e.g., Bishop et al. 2008). The VNIR spectra are also consistent with the presence of Fe-bearing opal (Baker et al. 2014) that exhibits a broader band near 2.2 µm extending out toward 2.26 µm. Both jarosite (Bishop and Murad 2005) and gypsum (Bishop et al. 2014) also have features at 2.22 and 2.26 µm that contribute toward the broadening of this band. Jarosite spectra also include a sharp band at 1.47 µm that is likely contributing to the doublet-like shape at 1.41 and 1.47 µm for coarse material of samples JB629 and JB776 (Fig. 6b). The VNIR spectra of the light-toned solfatara rocks resemble the spectra of solfatara rocks measured in the Ka’u Desert as well as aerial spectra of solfatara regions collected by AVIRIS (Seelos et al. 2010).

The orange and darker colored samples are more variable in the VNIR region and include bands near 0.44 and 0.9–1 µm due to Fe, as well as the H₂O and OH bands described for the lighter colored samples (Figs. 6a, 6c, and 6d). The reflectance minimum near 0.91 µm and maximum near 0.73–0.74 µm is consistent with jarosite as well as akaganéite and schwertmannite (Bishop and Murad 1996), but only jarosite was observed by XRD. Ferrihydrite is also consistent with a band near 0.91 µm and is X-ray amorphous so it could also be present. The weak dip in reflectance near 0.63–0.64 µm and shoulder near 0.88 µm likely indicates that some nanophase hematite is present (e.g., Morris et al. 1985, 1989; Bishop and Murad 1996) that is too fine-grained or poorly crystalline to be observed with XRD. Fe in nontronite also produces bands near 0.64–0.65 and 0.95 µm (Bishop et al. 2008) that could be contributing to these features in the spectra of the orange samples. A triplet near 1.44, 1.49, and 1.53, and bands near 1.75 and 1.94 µm are characteristic of gypsum (Bishop et al. 2014), and these features are much more apparent in some of the orange-colored samples than the friable, light-toned ones. Weak bands due to jarosite are observed at 1.47 and 2.26 µm, but the jarosite band near 1.86 µm is not observed, indicating that jarosite abundance is <50 wt% (Perrin et al. 2018; Usabal and Bishop 2018). Spectra of orange soils from the sulfate hill at the Painted Desert include jarosite (52 wt%) and gypsum (26 wt%) that exhibit jarosite bands at 0.44, 0.91, 1.47, 1.86, and 2.26 µm (Perrin et al. 2018) and lab spectra of jarosite/gypsum mixtures (Perrin et al. 2018) and jarosite/nontronite mixtures (Usabal and Bishop 2018) with 50 wt% jarosite include a shoulder at 1.86 µm.

TIR emissivity spectra measured in the mid-IR region of the orange side of sample JB633 include features due to gypsum and jarosite (Fig. 7). The emissivity spectrum of jarosite includes features near 1110, 1020, 630, 525, 475, and 250 cm^–1 (or ~9, 10, 16, 19.6, 21, and 40 µm), while that of gypsum has features at ~1154, 673, 604, and 470 cm^–1 (or ~8.7, 15, 17, and 21 µm). These are similar to jarosite and gypsum bands observed in previous studies (Lane and Christensen 1998; Bishop and Murad 2005; Bishop et al. 2014). Sulfates exhibit strong SO₄ stretching and bending vibrations in this region (Lane and Bishop 2019) that often dominate the spectral features of phyllosilicates and opal. These strong bands in the emissivity spectrum are why the emissivity spectrum of the orange layered rock is primarily characterized by gypsum and jarosite, rather than poorly crystalline phases.

Reflectance spectra of fine-grained portions of the samples measured under controlled, dry conditions exhibit weaker water bands that are more characteristic of specific minerals in many cases (Fig. 8) compared to spectra of rocks measured under ambient conditions (Fig. 6). VNIR and mid-IR reflectance spectra of the white/light samples JB629-B and JB632-B are compared with spectra of opal, montmorillonite, and bytownite (Figs. 8a and 8b). Although opal is the most abundant component in these samples due to hydration of the silica replacing the glass and rinds, some features characteristic of montmorillonite and feldspar are also present. These spectra illustrate the NIR features due to opal and montmorillonite near 1.41, 1.91, and 3 µm (Bishop 2019), as well as mid-IR features due to opal near 1875, 1260, 1105, and 475 cm^–1 (Bishop and Rampe 2016), the montmorillonite water band near 1630 cm^–1, and the feldspar Si-O stretching band as a reflectance peak near 1000 cm^–1.

Spectra of fine-particulate portions of the darker orange and tan samples are compared to spectra of the minerals nontronite, saponite, opal, gypsum, and jarosite in Figures 8c and 8d. The spectra of these samples differ from the spectra of the light-toned samples in that phyllosilicates and sulfates contribute more to the spectral character than opal. A band near 0.91–0.95 µm is due to Fe-bearing minerals in these samples and varies from sample to sample with variations in nontronite, jarosite, and iron oxide minerals that are likely present as short-range ordered species that are not sufficiently crystalline to be detected by XRD. Bands near 1.41 and 1.91 µm are characteristic of water overtones and combinations in opal and smectites, while the band near 2.21 µm is due to an OH combination band in spectra of opal or montmorillonite, the band near 2.29 µm is due to nontronite, and the band near 2.31 µm is due to saponite or hectorite (Mg-rich smectites). Gypsum and jarosite also have features near 2.22 and 2.27 µm that overlap with these smectite bands. Gypsum also has bands near 4.5 and 4.7 µm observed for some of these samples. The mid-IR region for these samples is most consistent with spectra of Mg-smectite with some features due to gypsum. Expanded NIR views of samples JB633-B and JB643-A include several features due to nontronite, opal, and gypsum (Fig. 8e). The water combination band occurs near 1.95 µm in these samples, similar to that of gypsum, with a shoulder near 1.91 µm due to nontronite or opal. NIR spectra of samples JB631-B and JB631-D are consistent with Mg-smectite and opal, while NIR spectra of JB630-B includes weak features due to jarosite at 1.47, 2.22 and 2.27 µm, in addition to bands characteristic of opal near 1.37, 1.41, 1.91, and 2.21 µm (Fig. 8f).

Mössbauer spectra of samples of the white/light (JB632) and orange (JB633) materials are shown in Figure 9 together with spectra of selected minerals. The white/light and orange/tan materials are subtly different, but both have contributions from ferrihydrite, which has parameters of isomer shift (IS) = 0.35 mm/s and quadrupole splitting (QS) = 0.71 mm/s for single doublet fits (Murad and Schwertmann 1980). The spectrum of JB632 has a doublet with IS of 0.37 mm/s and QS of 0.55 mm/s, consistent with a poorly crystalline ferric oxide-bearing phase similar to ferrihydrite. The VNIR spectra of sample JB632 are consistent with the presence of Fe-bearing opal and that could also explain the ferrihydrite-like component of the Mössbauer spectrum. Another contribution to the Mössbauer spectrum of sample JB632 is from saponite, which is also observed in the VNIR spectra and microprobe analyses.

The spectrum of JB633 has a well formed doublet with isomer shift of 0.39 mm/s and quadrupole splitting of IS of 0.65 mm/s that is more similar to ferrihydrite than nontronite or jarosite, but may again include a contribution from saponite or alternatively a ferrous sulfate (Dyar et al. 2013).

Investigation of hydrothermally altered volcanic material at other field sites in Hawaii, Iceland, LaSolfatara, Yellowstone, Southwestern Australia, and the Atacama has identified similar mineral assemblages in some cases, although the presence of jarosite, gypsum, and smectite is uncommon. Silica was observed as the primary alteration product through hydrothermal alteration of basaltic lavas from the Kilauea 1974 flow (Yant et al. 2017) near our field site. Yant et al. (2017) also observed small amounts of jarosite, gypsum, phyllosilicates, and iron oxide phases in some samples from their study. Assemblages of jarosite, silica, and hydrated ferric oxides were identified in crusted material in the Kilauea region where plumes bearing sulfuric acid likely altered the volcanic ash in an acid-fog type mechanism (Schiffman et al. 2006). Hydrothermal alteration at Haleakala volcano on Maui produced jarosite and alunite near some of the cinder cones inside the caldera, along with poorly crystalline Al-Si phases, silica, smectites, and iron oxides/hydroxides (Bishop et al. 2007). Assemblages containing jarosite, alunite, and silica were observed near fumaroles at the Italian Solfatara, but this site was too acidic for the formation of smectite or gypsum (Flahaut et al. 2019). Another acid hydrothermal site at Yellowstone National Park produced jarosite, other low-pH sulfates, and kaolinite, but not gypsum or smectite (Hinman et al. 2021). The abundance of jarosite at the Italian Solfatara and Yellowstone sites could be due to the elevated temperatures and low acidity at these sites. Gypsum has been found in various environments on Earth but is more commonly observed in evaporite or high-saline environments (e.g., Benison and Bowen 2006; Bowen et al. 2013; Flahaut et al. 2017; Bishop et al. 2021b; Burton et al. 2023; Warren-Rhodes et al. 2023). Gypsum can also form in hydrothermal environments and is then often accompanied by carbonates (Burisch et al. 2018). Gypsum, jarosite, and smectite have been observed together at the Sulfate Hill in the Painted Desert region of the Petrified Forest (Perrin et al. 2018), although the genesis of this site is not yet well understood. Gypsum and jarosite have been detected together in saline lakes, where halite, silica, or kaolin minerals are frequently observed as well, but smectite is less common in these areas (e.g., Benison and Bowen 2006; Bowen et al. 2013).

Solfataric alteration at other fumarole sites has produced similar alteration trends to our observations at the Kilauea South Sulfur Banks site. Pyroclastic basalts altered by the Cerro Negro fumaroles in Nicaragua resulted in amorphous silica, gypsum, alunite/jarosite, and iron oxides/oxyhydroxides (McCollom et al. 2013). The Námafjall geothermal area in Iceland harbors a range of alteration environments and products, including amorphous silica, anatase, elemental S, smectite, kaolinite, goethite, sulfides, and multiple Fe²⁺ and Fe³⁺ sulfates, formed under acidic conditions near hot springs (Carson et al. 2023b). Pyrite and other sulfides were only found at depth in that study and not near the surface. A second study at the Námafjall geothermal area investigated changes in alteration with distance from the Hverir fumarole (Carson et al. 2023a) and found native S closest to the vent in soils enriched in silica and anatase, similar to the white/light-colored silica-bearing apron in our study that included yellow native S near the vent (Figs. 2, 3, and 10). Cristobalite was abundant in the light-toned soils close to the Hverir vent site, whereas opal-A was most abundant and only trace cristobalite was observed in our study, likely indicating lower temperatures at the Kilauea South Sulfur Bank site. Further from the Hverir fumarole vent jarosite and alunite were observed together with kaolinite and hematite, and still further out from the vent smectite was observed (Carson et al. 2023a). Temperatures ranged from ~14 to 97 °C, where samples were collected near the Hverir fumarole. Temperature was not recorded during sampling at the Kilauea South Sulfur Bank site, but elevated temperatures were not noticed while working at the Kilauea South Sulfur Bank site. During our visits to the South Sulfur Bank site in 2003–2006, small vents with periodic wisps of gases and/or steam emanating from one or more vents were observed, rather than a continuous outpouring of steam observed at other more active fumarole sites. Historic activity at the South Sulfur Bank site could have been different. The moderate temperatures of the vents observed during our fieldwork are consistent with formation of the assemblages of opal-A, gypsum, smectite, poorly crystalline ferric oxide phases, and minor jarosite, based on comparison with these other hydrothermal alteration studies. It is possible that the jarosite, trace cristobalite, and trace tridymite at the South Sulfur Bank site formed during past episodes of greater steam activity at higher temperatures. If this were the case, the jarosite, cristobalite, and tridymite would have been little altered by the mild steam activity observed at the time of sampling. Hydrothermal alteration at other sites at elevated temperatures produced higher quantities of jarosite and/or alunite together with cristobalite, kaolinite, and hematite instead of smectite and poorly crystalline ferric oxide phases.

Alteration at the South Sulfur Bank site includes oxidation and leaching of the basaltic lava covering the pyroclastic deposit, alteration of the orange/tan banded outcrop, and alteration of the white/light soil (Fig. 10). The four different alteration types observed are highlighted in a schematic diagram of the site in Figure 11. The friable white/light material is closest to the vents at the floor of the South Sulfur Bank, and the glass and rinds are both replaced by ~80 wt% silica. Hydrated silica, anatase, and native S are the primary components of this friable unit, similar to the highly altered material observed closest to the Hverir fumarole in Iceland (Carson et al. 2023a). The opal-A measured in our study likely results from lower temperature hydrothermal alteration compared to the Hverir fumarole, where temperatures close to 100 °C were observed and cristobalite formed rather than opal-A, indicating much higher temperatures at that site.

The orange/tan bedded units are further from the vents than the white/light/yellow friable material and are much more consolidated. This banded region contains a combination of alteration minerals in addition to some hydrated silica. Smectites include saponite, hectorite, and nontronite, and the type of smectite present is likely governed by the cations that were available—nontronite where Fe was readily available, hectorite where Mg and Li were available, and saponite where Mg and Al were available. Feldspar and gypsum are observed with Mg-rich smectite in the tan/lighter layers, whereas jarosite and poorly crystalline iron oxide phases are found with nontronite in the orange layers. However, gypsum and Mg-smectites are also observed in some of the orange material. Horizons of orange-colored beds are visible in several places, but patches of orange material are also evident (Fig. 10). This spatial relationship would suggest that there could have been variations in the chemistry of the pyroclastic material as it was deposited, with elevated Fe in some areas. Another possibility is that periodic episodes of greater leaching and alteration of the basalt above this unit occurred and that altered material rich in ferric iron percolated down from the altered basalt and was collected in layers and patches at the banded pyroclastic unit below. Possibly, boundaries at the beds in this unit could have helped retain the Fe³⁺ to form a layer of increased concentration of Fe³⁺ to react with sulfates during alteration. The orange layers are not just a thin surface coating, but extend at least 2 cm deep into the banded unit at the sampling location (Fig. 10c).

Characterization of analog sites with multiple mineralogy, chemistry, and spectroscopy techniques provides data for interpretation of orbital and rover data at Mars. This study of high-silica and smectite-sulfate outcrops near low-temperature vents at the South Sulfur Bank at Kilauea provides an opportunity to investigate the minerals formed under moderate hydrothermal conditions. These two types of hydrothermal outcrops are observed on Mars in ancient outcrops and could have formed on Mars in the past under environmental conditions similar to the moderate solfataric alteration that occurred at the Kilauea South Sulfur Bank region. The presence of high-silica outcrops and smectite-gypsum-jarosite mixtures is still observed today in some regions of Mars after billions of years. This indicates that limited liquid water has altered the surface since the emplacement of these samples.

Martian outcrops with extremely high silica, similar to the friable, white material observed near vents in the South Sulfur Banks region of Kilauea and near the Hverir fumaroles in Iceland are found at the Home Plate hydrothermal system in Gusev crater (e.g., Rice et al. 2010; Ruff et al. 2011), at the Nili Patera caldera associated with Syrtis Major (Skok et al. 2010; Fawdon et al. 2015), and within the Murray and Stimson formations at Gale crater (Rapin et al. 2018). Siliceous hot spring deposits are widespread on Earth (e.g., Campbell et al. 2015; Jones 2021) and have long been considered potential ecosystems for life on Mars (e.g., Walter and Des Marais 1993; Farmer 1996; Schulze-Makuch et al. 2007; Cady et al. 2018). Opaline silica formed near low-temperature vents similar to the Kilauea South Sulfur Banks region could also have provided a niche for potential microbes on Mars. An advantage in searching for life in ancient silica-rich outcrops on Mars is the high preservation potential in such environments (Walter and Des Marais 1993; Farmer and Des Marais 1999).

Jarosite has been detected at Mars by rovers and from orbit by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) and could play a critical role in understanding the aqueous geochemical history of Mars. Jarosite and its Al form alunite are common hydroxysulfate minerals with widespread occurrence on Earth (e.g., Van Breemen 1982; Desborough et al. 2010). These OH-bearing sulfates serve as indicators for salty and low-pH environments, and their dissolution products provide a record of temperature, pH, and longevity of liquid water on the surface of Mars (e.g., Elwood Madden et al. 2012), where they have been observed in multiple locations (e.g., Klingelhöfer et al. 2004; Farrand et al. 2009; Weitz et al. 2011; Ehlmann et al. 2016). These minerals can form in acidic waters associated with sulfide oxidation (e.g., Burns 1987, 1988; Alpers et al. 1989, 1992; Burns 1993; Burns and Fisher 1993; Bigham et al. 1996; Herbert 1997; Crowley et al. 2003; Murad and Rojik 2003), acid sulfate soils near volcanic vents (e.g., Zimbelman et al. 2005; Schiffman et al. 2006; Bishop et al. 2007; McCollom et al. 2013; Carson et al. 2023a, 2023b) and in acidic saline lake environments (e.g., Alpers et al. 1992; Risacher et al. 2002; Benison and LaClair 2003; Baldridge et al. 2009; Story et al. 2010; Flahaut et al. 2017). Jarosite and Ca sulfates can coexist in some environments (e.g., Elwood Madden et al. 2012; Dixon et al. 2015; Perrin et al. 2018) and various iron oxides/oxyhydroxides and clays can form through jarosite dissolution (e.g., Gasharova et al. 2005; Navrotsky et al. 2005; Elwood Madden et al. 2009, 2012). Recent lab studies also demonstrate smectite formation in acidic environments (Peretyazhko et al. 2018), which supports co-formation of smectite and jarosite in our study. The presence of these accessory minerals assists in defining the geochemical history governing the formation and reaction of jarosite and alunite on Mars. A combination of jarosite and gypsum in a smectite and hydrated silica matrix is observed at the South Sulfur Bank in the banded orange/tan units; this indicates that low-temperature hydrothermal alteration has taken place in the vicinity of volcanic vents but not directly proximal to the vents. Similar processes may have occurred on Mars near volcanic features. The mineralogy of these banded outcrops is most similar to parts of the Noctis Labyrinthus system (Weitz et al. 2011), Gale crater (Rampe et al. 2020), and Mawrth Vallis region (Bishop et al. 2023) on Mars, where both jarosite and gypsum group minerals are observed together with smectites. Additionally, both jarosite and gypsum were identified by instruments on Opportunity at Meridiani (Squyres et al. 2006). Jarosite and a nanophase iron oxide phase were identified by the Miniature Mössbauer instrument on Opportunity (Klingelhöfer et al. 2004; Morris et al. 2004, 2019) and could represent altered volcanic material there.

These low-temperature hydrothermal products at the Kilauea South Sulfur Bank contain a high abundance of opal in the friable white/light-colored unit and a mixture of smectites and sulfates in the orange-layered outcrops. This site provides an example of the minerals formed under moderate hydrothermal conditions similar to habitable environments on Mars where biosignatures could have been preserved.

This paper is dedicated to our dear colleagues Peter Schiffman and Enver Murad who did not live to see this work published. Peter discovered this intriguing site, and we remember him fondly for his excitement investigating the Kilauea region. Enver was always ready to analyze more samples, and his passion for understanding the mineralogy of Mars analogs provided significant advances for this project. Support for this work at the early stages of the project from the MFR program and, more recently, from MDAP grant 80NSSC19K1230 is much appreciated. Thanks are also due to the National Park Service and Hawaii Volcanoes National Park for access to the site (permit HAVO-2004-SCI-0018) and to E. Göschl for help with the fieldwork in 2006. We thank T. Hiroi for assistance with the reflectance measurements at RELAB, Brown University. Emission spectra were measured at the Arizona State University Mars Space Flight Facility, and Mössbauer spectra were measured at Mt. Holyoke College.