The global distribution of clay minerals on the Martian surface demonstrates the liquid water activity on ancient Mars (Poulet et al., 2005; Bibring et al., 2006; Bishop, 2018). Among clay members of the phyllosilicate group identified from the orbital datasets of Mars, the content of chlorite is higher than that of all the other members with the exception of smectite (Ehlmann et al., 2011a; Ehlmann et al., 2011b). Similar to smectite, Martian chlorite is also rich in Fe and Mg (Ehlmann et al., 2011a). Chlorite is a hydrous phyllosilicate-type mineral with an interlayer hydroxide octahedral sheet between two structural layered units, forming a special 2:1:1 layered structure (Kohut and Warren, 2002). According to different enriched metallic elements that occupy the octahedral sheet, chlorite can be divided into four sub-types—clinochlore (Mg²⁺-rich), chamosite (Fe²⁺-rich), nimite (Ni²⁺-rich), and pennantite (Mn²⁺-rich) (Kohut and Warren, 2002). However, due to element substitutions, clinochlore can also be rich in Fe²⁺ (Smyth et al., 1997), as Fe²⁺ and Mg²⁺ are the dominant divalent cations in the octahedral sheets (Aja et al., 2015). Chlorite can be formed through the diagenesis of smectite, hydrothermal alteration, and metamorphism, and it could be further weathered to vermiculite in warm and wet terrestrial surface conditions (Ehlmann et al., 2011a).

On Mars, chlorite has been identified mainly by visible/near-infrared (VNIR) spectra obtained by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) onboard the Mars Reconnaissance Orbiter (MRO) and Observatoire pour la Minéralogie, l'Eau, les Glaces et l'Activité (OMEGA) onboard the Mars Express orbiter (Poulet et al., 2005; Ehlmann et al., 2011b). In some cases, chlorite coexists with prehnite near craters, indicating the presence of hydrothermal activity driven by impacts (Ehlmann et al., 2009; Ehlmann et al., 2011a). Alteration mineral assemblages are also discovered in central peaks of craters where chlorite is associated with Fe, Mg-smectite, analcime, and hydrated silica (Milliken et al., 2008; Ehlmann et al., 2009).

Another phyllosilicate, illite, is structurally a 2:1 type of phyllosilicate composed of two tetrahedral sheets sandwiched by one octahedral sheet (T-O-T) (Ehlmann et al., 2009). Although illite is indistinguishable from muscovite in CRISM and OMEGA data (Ehlmann et al., 2009), it may coexist with chlorite as the main weathering-derived clay member on Mars. In the region of Nili Fossae, both chlorite and illite are identified based on the CRISM data (Ehlmann et al., 2009). Considering the mineral composition and the aqueous history of this region, (Ehlmann et al., 2009), suggest that the fluvial activity and sapping channel formation could affect the distribution of these minerals. Therefore, the post-depositional effects caused by later water activity are significant for understanding the aqueous history and the mineral transformation during the long evolution of the Martian paleoclimate.

The Qaidam Basin, located on the northern edge of the Tibetan Plateau (Figure 1A), has formed a vast area of playas and eolian landscapes due to the blocking of moisture by surrounding mountain ranges of >5,000-m altitude (Figure 1B) and the resultant prolonged hyperarid climate (Han et al., 2014). After a similar climate transition from wet to dry, the surface landforms and changing surface environments of the Qaidam Basin are analogous to those proposed for ancient Mars (Kong et al., 2018). The basin has maintained high aridity, with a current evaporation rate of 2,590 mm/yr and a precipitation rate of <20 mm/yr (Kong et al., 2018). The average annual surface temperature of the western Qaidam basin is 3.5°C, with the maximum monthly average air temperature of 16°C in July and the minimum monthly average air temperature of −10.7°C in January (Kong et al., 2018). Thus, numerous previous studies exemplify Qaidam as one of the highest environmental Mars analogs (Xiao et al., 2016; Anglés and Li, 2017; Sun et al., 2019; Sun et al., 2021). Here, we report alteration of chlorite illite-bearing deposits in two areas: weathered yardang (basin center) and dune-covered (basin margin) deposits in the western Qaidam Basin. By evaluating the post-depositional effects on these deposits in a dry and cold Mars analogous environment, this study seeks to provide constraints on the evolution of widespread iron-bearing phyllosilicate on Mars.

Four samples of this study were collected in lacustrine-fluvial deposits from western Qaidam Basin: the yardang samples (named 0718 and 0719) at N38°2′6.4″, E91°51′8.39″, and N38°12′40.716″, E92°4′15.276″, respectively; the dune-covered samples (named 1222-3A and 3B) at N37°50′58.6″, E91°4′6.5″ (Figure 1C). These samples were collected in the vast area of the Holocene playa (Figure 1C), representing the last massive liquid water event (Han et al., 2014) that has been shaping the depocenter of the western Qaidam Basin ever since. After the disappearance of the northwestern center of the mega-lake, the surface is featured by strong wind-sculpted yardang fields (Xiao et al., 2016). Thin layers of carbonate ooids were found actively formed 38,000 years ago near the yardang sampling area (Sun et al., 2019), indicating that these deposits should be exposed later than this period. Otherwise, evaporites would have cemented the deposits. In the western Qaidam Basin, yardang fields display various morphologies and sizes due to winds with different strengths and directions. The yardang fields now cover an area of 3.88 × 10⁴ km² in the Qaidam Basin (Xiao et al., 2016). The two yardang samples were collected on the lower side of the weathered yardangs (Figures 1C, 2A). Samples displayed grey surfaces with rusty small concretions (Figures 2B,C). Sand dunes in the desert, however, have only developed in particular zones in the basin. The dune-covered samples were collected on a fluvial fan of Kunlun Mountain, which is partially covered by a series of barchan dunes (Figures 1C, 2D). These deposits were layered and loose (Figure 2E). Iron-bearing veins exist in these deposits (Figure 2F). Two samples were collected with clean tools from the same site but in different layers and sealed with parafilm tapes to prevent exposure to air moisture during transport to the laboratory.

In addition to the four sediment samples, we collected two purified clay minerals (see Supplementary Figure S1 for more component information), clinochlore (from Liaoning, China) and illite (from Zhejiang, China), as positive controls in spectral measurements.

The mineral compositions of these samples were determined using a Rigaku MiniFlex-400 X-ray diffractometer (XRD) at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (CAS). The diffractometer is equipped with a Hybrid pixel array detector, a Cu source, and a Ni filter. Prior to bulk XRD measurements, samples were ground into fine powder (70 μm) using agate mortar. The samples were loaded into rectangular Al holders and tableted for measurements. Data were collected between 3° and 70° with a step size of 0.01 and a dwell time of 1°s in each step. To identify the composition of the rusty parts from these deposits, the rusty concretions were separated with clean tweezers based on different colors and ground for further XRD measurements.

Spectroscopic analysis was conducted on a Malvern Panalytical TerraSpec 4 Hi-Res spectrometer at the Institute of Geology and Geophysics (IGG), CAS. Before the spectroscopic analysis, samples were oven-dried at 80°C overnight and separated into bulk clay and rusty parts with tweezers, followed by mild grinding into fine powder. Microstructure and morphology of small particles were recorded using an FEI Apreo S scanning electron microscope (SEM) equipped with a Bruker XFlash 60 energy dispersive spectrometer (EDS) detector and a JEOL JEM-2100 transmission electron microscope (TEM) at IGGCAS. To obtain the natural morphology of the minerals, we use small tweezers to pick out the rusty parts and fix them on the conductive adhesive for direct SEM observation. The samples for TEM observation were ground with an agate mortar for 5 min, dispersed in alcohol, and dropped on the copper grid for analysis.

The rusty concretions were separated from the samples and were ground into fine powder for ⁵⁷Fe Mössbauer spectroscopy measurements at room temperature in the transmission mode operating in constant acceleration mode using a Silver Double Limited WSS-10 spectrometer. The Mössbauer source was a ⁵⁷Co in the Rh matrix. The velocity drive transducer was operated in a triangular waveform mode over energy ranges of ±15 mm/s. The spectroscopic center was calibrated using an α-Fe foil. The spectrum was analyzed with the Mosswinn 4.0 program. Subspectra attributed to different phases were determined based on their hyperfine parameters [i.e., isomer (IS), quadrupole splitting (QS), and linewidths].

As shown by the XRD patterns, the mineral components of these deposits were quartz (d₍₁₀₁₎ = 3.34 Å), albite (d₍₀₀₂₎ = 3.18 Å), minor carbonate (d₍₁₀₄₎ = 3.03 Å), and clay minerals identified as Fe-clinochlore and illite in accordance with their main d values (d₍₀₀₁₎ = 13.90 Å and d₍₀₀₂₎ = 9.90 Å) and the overall pattern features (Figure 3). The yardang samples contained minor gypsum (d₍₀₂₀₎ = 7.6 Å), reflecting the evaporation of liquid water in the basin center. The dune-covered deposits displayed characteristic XRD peaks of amphibole (d₍₁₁₀₎ = 8.4 Å) (Figure 3), implying that the debris from nearby mountains was only partially weathered.

The XRD patterns of the rusty parts (named 0718-Fe and 0719-Fe) display the same features as the bulk samples (Figure 4), indicating no new well-crystallized phase was formed. The intensity ratio of d₍₀₀₁₎/d₍₀₀₃₎ in XRD patterns was used to determine the Fe content in clinochlore (Ryan and Reynolds, 1997). The ratios are 8.84, 4.6 for 0719-Bulk and 0719-Fe; 4.4, 4.2 for 0718-Bulk and 0718-Fe. The decrease of the d₍₀₀₁₎/d₍₀₀₃₎ ratio indicated that Fe migrated from clinochlore to the rusty parts.

SEM micrographs of the bulk samples showed poor morphology of these clay minerals with grain sizes <2 μm (Figures 5A–C), indicating a debris origin (Wilson, 1999). Halite was observed in the yardang-type samples (Figure 5B). In the rusty parts, nanoparticles are distributed on the surface of the clay grains (Figure 5D). Backscattered electron imaging associated with the elemental mapping of the same area illustrated that these materials were Fe-enriched (Figures 5E,F) but Mg, Al, Si-depleted compared to the surrounding clay grains (Figures 5G–I). EDS results confirmed that clinochlore in these samples was enriched with Fe, and the content of Fe in the clay mineral was decreased near the Fe enriched area (see Supplementary Table S1).

Under TEM (Figure 6A), nanoparticles were intimately mixed with clinochlore. The magnified image of these minerals again displayed nanophase aggregates with a diameter of <5 nm (Figure 6B). The high-resolution TEM image revealed that the d-spacing of the nanoparticle was 2.5 Å (Figure 6B), which corresponded to the 110 panels of ferrihydrite or nano-hematite (Jiang et al., 2018).

Mössbauer spectroscopy of the rusty parts provided further identification of the Fe-bearing materials. The spectrum was fitted with two subspectra (Figure 7). The doublet with IS, QS values of 1.12 mm/s, and 2.73 mm/s (green) corresponded to the Fe²⁺ in the octahedron of chlorite (Pollak and Stevens, 1986). Another doublet with IS, QS values of 0.37 mm/s, 0.61 mm/s (pink) indicated that the octahedral sites of Fe³⁺ were relatively disordered, which should be attributed to ferrihydrite as additionally supported by the TEM observations (Murad, 1996).

The NIR spectra of these Qaidam samples displayed combinative features of both clinochlore and illite (Figure 8). The -OH overtone at 1.414 μm was much weaker than the pure clinochlore and illite. The H-O-H combination band at 1.931 μm should be attributed to the coupling of the signals from both clinochlore and illite, which moved to a higher wavelength indicative of the increasing content of illite (Figure 8). The 2.211 μm absorption was due to the Al-OH in illite (Ehlmann et al., 2009), while the ∼2.347 μm absorption was a combination of M-OH (M could be Fe, Mg, or the mix of these cations) in clinochlore and illite (Bishop et al., 2008). The CRISM data reported by (Ehlmann et al., 2009) demonstrated a similar NIR pattern with 1.4, 1.9, 2.2, and ∼2.3 μm absorptions as compared to the Qaidam samples in this study, indicating the deposition of the chlorite illite assemblage in the Nili Fossae region on Mars.

Through the mineralogical study of chlorite illite deposits in the western Qaidam Basin, we observed the post-depositional effects on the change of chlorite composition and structure under hyperarid climatic conditions. Due to the ephemeral water activity, iron is released from Fe-clinochlore and forms amorphous ferric oxyhydroxides in the sediments. Similar NIR spectra are characterized in the Nili Fossae region on Mars, indicating the presence of chlorite illite deposits. Diagenesis in the Qaidam Basin and the resulting chemical weathering of chlorite may inform the evolution of chlorite and precipitation of secondary iron phases on the Martian surface.