5.1. Early diagenesis and porewater
After burial, diagenetic reactions affected the compositions of the porewater and sediments at a given depth, which happened between the water and minerals, and between solid minerals (Berner, 1980; Brown, 2001; Putnis and Ruiz-Agudo, 2013; Gorski and Fantle, 2017). Therefore, the elements from porewater, carbonate minerals and silicate minerals, are all underwent early diagenesis. The elemental concentrations in this study presumably represent the balance between minerals and porewater, and between solid minerals during early diagenesis, although any system is not at an absolute steady state.
Paleo-lake water was captured during the burial process of sediments and became the predominant composition of the porewater. Because of the exceptionally low deposition rate (0.2977–0.9697 mm/yr) in Guozha Co (Li et al., 2021a), it might have taken 15 to 33 years to deposit 1 cm-thick sediment. Therefore, the lake water trapped by the 1 cm–thick sediment represented the mixture of lake water during the past 15 to 33 years.
Previous studies reported that there would be a vertical flow of porewater in the upper 20 cm of the sediments, especially with the participation of microbial activity (e.g., Berner, 1980). However, the vertical flow of the pore water in this study could have been limited, which was supported by the following data. First, the contents of Nawater, Cawater, Mgwater and Kwater in the upper 20 cm of the sediments (i.e., first 5 data, Table supp., Fig. 2) varied considerably. A vertical flow of the pore fluid would cause that the ion content with depth profiles to be a straight line. It is clear from the profiles of the ionic content at Site GZLC15-1 (Fig. 2) that the vertical flow of pore water was limited. Second, the mean grain size was fine (most are < 40 μm), and the content of clay minerals was high with an average of approximately ~30% (Wang, 2021; Guo, 2021). These fine particles could have prevented the vertical flow of porewater. Third, low porosity is not conducive to the vertical flow of porewater. In general, rapid deposition leads to a high initial porosity (Berner, 1980). The low deposition rate of less than 0.1 cm/a (Li et al., 2021a) inferred that the porosity should be low in this study. Therefore, the early diagenesis in the water-sediment interface was likely to have occurred in situ.
5.2 Minerals and elements
Minerals controls on the elemental variations in lake sediments. The elements in the sediments basically exist in silicate minerals besides Ca and Sr (Fig. 5). Caacid of carbonate minerals is often replaced by Mgacid and Sracid. The contents of Caacid, Mgacid and Sracid are from the mixed carbonate minerals with calcite, aragonite and trace dolomite in Guozha Co. Both Mgacid and Sracid enter the lattice of calcite, while Sracid is preferred to the lattice of aragonite, which have been discussed by Li et al. (2021a). The correlation coefficients between Liacid and Sracid were high (R= 0.55, 0.8) during 8.7–1.5 kyr BP where there was no aragonite, while that was 0.006 from 1.5 kyr BP to present (Table 1), when the carbonate minerals dominated by calcite and aragonite (Li et al., 2021a). However, the coefficients between Liacid and Mgacid increased from 0.36 to 0.88 since 8.7 kyr BP (Table 1). These suggested that Liacid in the carbonate minerals should be primarily preserved in the calcite.
For the elements in residue leachate, most of them were from K–silicate minerals (such as microcline, clays), while the Naresidue was from Na-silicate minerals (such as albite), according to the high coefficients between Kresidue and the other elements (Table 3). The high correlation coefficients between Liresidue, Mgresidue and Kresidue (RLi-Mg=0.45–0.67, RLi-K=0.90–0.96; Table 3) suggested that Liresidue was closely related to clay minerals, especially illite and chlorite, because K and Mg are both the predominant cationic components of illite and chlorite. However, Liresidue and Naresidue appeared in an obvious negative correlation with correlation coefficients from -0.21 to -0.8 since 8.7 kyr BP (Table 3). The decreasing curves of the Mgresidue, Kresidue, Liresidue contents (Fig. 4) suggested the decreased input of K–silicate minerals, as a result of the decreasing recharge of glacier meltwater and weakening chemical weathering since 8.7 kyr BP (Li et al., 2021a). However, the Naresidue content increased. It was likely associated with the new formation of Na–silicate minerals, which was common during the early diagenesis (Wallmann et al., 2008; Andrews et al., 2020). The increasing salinity of lake water provided enough Na for the new formation of Na–silicate minerals.
The silicate mineral also controlled the element contents in the porewater. According to the plot of the molar ratios between Mg/Na and Ca/Na, Mg/Li and Ca/Li, Mg/K and Ca/K (Fig. 6), the data of porewater were closed to those of silicate minerals, and far away from those of carbonate minerals. These indicated that the chemical compositions of porewater were influenced significantly by silicate minerals during early diagenesis.
5.3 Elemental properties
Metal elements with similar properties can coexist in the same minerals and the ion exchanges between them occur easily during early diagenesis. The elemental properties here mainly include ionic valence, ionic radius, solubility and whether they are biological elements. For example, both Li+ and Na+ are easily dissolved, monovalent ions, with a little higher ionic radius of Na+ (rNa= 0.102 nm) than Li+ (rLi = 0.076 nm). The ions exchanges between Li+ and Na+ occurred easily not only in porewater, but also in carbonate minerals and silicate minerals according to the high correlation coefficients between Na and Li, although Naresidue and Liresidue in silicate minerals show a high negative correlation (Tables 1,2,3).
Li+ and Mg2+ have similar small-radius (rLi= 0.076 nm, rMg= 0.072 nm) with different ionic valences. But there were high correlation coefficients between them, especially in silicate minerals (Tables 1, 2,3). This is because that (a) Mg2+ is affected by microbiological activities, especially during the deposition of carbonate minerals (e.g., JimÈnez-López et al., 2001; Rivadeneyra et al., 2004; Li et al., 2008, 2009). (b) Mg replaces Al in the octahedra in clay minerals, such as illite and chlorite, resulting in an imbalance electron valence state. As small-radius ion, Li+ easily entered into the interlayer of clay minerals to balance the imbalance charge.
The ionic radii of Li and Mg are also similar to those of Nawater, but the coefficients between Mgwater and Liwater were much lower than those between Liwater and Nawater (Table 3). It was likely that the coexisting monovalent cations have a greater influence on the migration of Li+ than that of Mg2+ (Zhang et al., 2020).
Ca2+ and Sr2+ have the same valence and similar radius (rCa = 0.1 nm; rSr = 0.112 nm), which could explain the high coefficients between them in the core (Tables 1, 2, 3). Although they have similar radius with Na+ (rNa = 0.102 nm), their valent difference made the coefficients between Ca–Na and Sr–Na lower than those between Ca–Sr (Tables 1,2,3).
The Mg2+ (rMg= 0.072 nm) has smaller radius than Ca2+(rCa = 0.1 nm) and can easily enter the lattice of carbonate minerals such as calcite to replace Ca2+. In addition to the same valence, microbiological activities and types of carbonate minerals were also key factors (Li et al., 2009, 2021a). The high coefficients between Caacid and Macid only occurred during 8.7–4.0 kyr BP when there was no aragonite (Table 1, Li et al., 2021a), which could support this observation.
For Mg2+ and K+, they have different ionic valency and their ion radii vary considerably with rK = 0.138 nm and rMg= 0.072 nm, but they have high correlation coefficients, especially in porewater and silicate minerals (Tables 2,3). The possible reasons for this are likely to be that they are both biological elements.
5.4 Solution properties
Salinity, pH and temperature of solution/lake water played important roles on elemental variations and ionic exchanges (Berner,1980; Putnis, 2013; Swart, 2015; Marriott et al., 2004). The salinity and pH of lake water increased gradually since 8.7 kyr BP in this study (Li et al., 2021a). According to the increased contents of Nawater, Naacid, Liwater and Liacid (Figs. 2, 3), Na and Li are more sensitive to changes in the salinity than the others. The increased salinity could enhance ionic exchanges and their correlations (Tables 3,4,5). For example, the correlation coefficients between Nawater and Liwater increased from 0.67 to 0.94 since 8.7 kyr BP (Table 3), and those between Nawater and Liacid increased from 0.48 to 0.75 (Table 4), and those between Naacid and Liwater increased from 0.39 to 0.87 (Table 4). The correlation coefficients between Nawater and Liresidue increased from -0.3 to 0.39 since 8.7 kyr BP, which suggested that the Nawater substitution for Liresidue primary occurred since 4.0 kyr BP, especially in saline conditions from 1.5 kyr BP to present (Table 5). The decreased Liresidue content since 8.7 kyr BP also supported the release of Liwater from Liresidue.
Similarly, the correlation coefficient between Cawater and Srwater increased from 0.77 to 0.96 since 8.7 kyr BP (Table 3), and those between Cawater and Sracid increased from 0.25 to 0.49 (Table 4), and those between Srwater and Caacid increased from 0.24 to 0.56 (Table 4). The correlation coefficients between Cawater and Srresidue increased from -0.06 to 0.47 since 8.7 kyr BP, and those between Srwater and Caresidue increased from –0.22 to 0.48 (Table 5). These all suggested that the increased salinity enhanced the ionic exchanges in the interface of water–minerals.
The effects of pH on the elements were different. For example, more Mg2+water was bind to OH- along with the increasing pH values, while the Kwater remains K+. The low coefficient between Mgwater and Kwater was observed in brackish conditions with high pH from 1.5 kyr BP to the present (Table 3).
The influence of temperature on the elements was accompanied with the microbial activities, especially on the Mg2+ in solution and minerals. Microbial activity is stronger in a freshwater and/or warm environment than that in a brackish and/cold water environment. Therefore, due to low temperature and weak microbial activities, the correlation coefficients between Mgwater and the other elements in carbonate mineral were higher during 4.0–1.5 kyr BP than those in the other two stages (Table 4). For example, the coefficients between Liwater and Mgacid (r =0.80) and between Liacid and Mgwater (r =0.60) were higher in the low temperature stage during 4.0–1.5 kyr BP than those in the other two stages (Table 4).
However, the coefficients between Liwater and Mgresidue, and between Mgwater and Liresidue both exhibited an increasing trend since 8.7 kyr BP (Table 5). These suggested that (a) temperature have significant effect than salinity on the ionic exchanges between water–carbonate minerals, and (b) salinity have more significant effect than temperature on the ionic exchanges between water–silicate minerals.
5.5 Implication in environmental changes and mineral deposition
Because of the above factors, when the elemental ratios were used as environmental proxies, it should be considered that where they were from porewater, mixed carbonate minerals, single carbonate mineral or silicate minerals. For example, the curves of Mgacid/Caacid and Liacid/Caacid were similar, but varied with that of Sracid/Caacid, especially from 1.5 kyr BP to present (Figs.7,8). It was due to that the Mgacid and Liacid were primarily from calcite, while the Sracid was primarily from aragonite and minor from calcite. When aragonite formed from 1.5 kyr BP to present in this study, Mgwater and Liwater remained in lake water, and Srwater enter the lattice of aragonite and became the Sracid, as a result, both the molar ratios of Mgwater/Cawater, Liwater/Cawater and Sracid/Caacid increased, and that of Srwater/Cawater decreased during the period (Fig.7).
The Mg/Li difference between porewater, carbonate minerals and silicate minerals could be very useful for the industrial lithium extraction from brines. As we known, Li was incorporated into Li–carbonate minerals (Li2CO3), and then collected to be used in industrial production as a raw material. However, Mg will enter the lattice of Li2CO3, and lowered the quality of Li2CO3 as a raw material. High Mgwater/Liwater molar ratio hindered the Li–extraction from brine, and the separation of Mgwater and Liwater in aqueous solution has been a critical technical problem (Kesler et al., 2012; Liu et al., 2015; Nie et al., 2017; Zhang et al., 2020). In this study, the Mgwater/Liwater were much lower than those of Mgacid/Liacid and Mgresidue/Liresidue (Fig. 8). Along with the increased salinity, the values of Mgwater/Liwater showed a decreased trend, while those of Mgacid/Liacid and Mgresidue/Liresidue increased. This could be very useful for the lithium extraction from brines.