Variation in the structural order of kaolinite in regolith as an effective indicator of REE mineralization

Abstract

Regolith-hosted rare earth element (REE) deposits hosted by the granitic regolith in South China provide >90% of the world’s heavy REEs. Kaolinite is one of the major carriers of REE ions in regolith. The formation and transformation of kaolinite can be affected by chemical weathering and hydrodynamic conditions, but the contribution of each factor has not been evaluated. This study systematically investigated the variation in abundance of phyllosilicate minerals and structural order of kaolinite in the Renju regolith-hosted REE deposit. The total abundance of 1:1 phyllosilicate minerals increased upwards along the profile from Section I to Section III. However, semi-quantitative analyses indicated that Section III-1 (depth at 10–16 m) featured an evident decrease in both abundance and structural order of kaolinite upward along the profile. The morphological feature and abundance of kaolinite revealed intensive kaolinite-to-halloysite transformation and kaolinite dissolution in Section III-1. This suggests that the alternating wetting and drying zone in Section III-1 provided a favorable kinetic environment for the entry of water molecules into the kaolinite interlayer and the kaolinite-to-halloysite transformation, resulting in both lower structural order and abundance of kaolinite in Section III-1. Moreover, REE ions started to be enriched from the alternating wetting and drying zone and formed high-grade ores at the lower part of the water table, due to a significant increase in pore water and decrease in the seepage velocity. Therefore, the abundance and structural order of secondary kaolinite can serve as important indicators of hydrodynamic changes in regolith, as well as the mineralization of regolith-hosted REEs.

Introduction

Regolith-hosted rare earth element (REE) deposits in South China provide >90% of the world’s heavy REEs, which are critical to modern society and high-end technology (Bao and Zhao, 2008; Simandl, 2014; Weng et al., 2015; Jowitt et al., 2017; Riesgo García et al., 2017; Xu et al., 2017). These deposits are hosted by regolith developed from biotite and muscovite granites, syenites, monzogranites, granodiorites, granite porphyries, and rhyolitic tuffs (Li et al., 2017). During the formation of regolith, humic acid, organic acid and CO₂, which are produced by decay of organic matter and the activity of microorganisms and vegetation (McQueen and Scott, 2008), acidify surface and groundwater. REE-bearing minerals (e.g. synchysite and titanite) tend to be dissolved by the acidified water, and the released REE are leached downward vertically in the regolith and adsorbed onto the surface of secondary phyllosilicate minerals weathered from primary rock-forming minerals, forming regolith-hosted REE deposits (Bao and Zhao, 2008; Huang et al., 2021a).

During weathering, primary minerals (such as feldspar and mica) are first altered into the incipient 2:1 phyllosilicate minerals (e.g. illite and smectite), before further transforming into the 1:1 phyllosilicate minerals (e.g. halloysite and kaolinite) (Senkayi et al., 1984; Li and Zhou, 2020). Consequently, illite and smectite are predominant in the saprock and saprolite, while the abundance of kaolinite and halloysite increases in the more weathered shallow zones of the regolith (Mei et al., 2021; Tan et al., 2021). Compared with the lower pedolith zone, vermiculite and well-ordered kaolinite appear in the upper pedolith zone, whereas the abundance of halloysite decreases (Li and Zhou, 2023). Accordingly, the vertical zonation in well-developed profiles of regolith can be defined by the mineral abundance and structural order of kaolinite and other phyllosilicates (Huang et al., 2021a; Tan et al., 2021).

Kaolinite has been suggested as the primary carrier of REE ions in regolith (Yang et al., 2019; Borst et al., 2020). Notably, REEs tend to be enriched in the vicinity of the transition zone between upper and lower pedolith (Li and Zhou, 2023), possibly due to the different REE adsorption capacity of the clay minerals caused by significant changes in the structural order of kaolinite during chemical weathering (Fialips et al., 2000; Ndlovu et al., 2015; Li and Zhou, 2020, 2023). However, the changes in the crystallographic and physicochemical properties of kaolinite may also be affected by the hydrodynamic conditions. For example, dry conditions above the water table have been suggested as favorable for kaolinite crystallization (Costanzo and Giese, 1985; Inoue et al., 2012). The overlapping roles of chemical weathering and hydrodynamic effects on the formation and transformation of kaolinite, especially for the structural order of kaolinite, have not been explicated. Additionally, the downward migration and enrichment behaviors of REEs could also be controlled by the notable variations in hydrodynamic properties of regolith due to presence of the water table (Huang et al., 2021a). The underlying relationship between REE mineralization and the transformation behaviors of kaolinite in regolith also requires further elucidation.

This study established systematic variations in abundance of different types of phyllosilicate minerals and structural order of kaolinite in the Renju regolith-hosted REE deposits in South China. The mutual transformation between kaolinite and halloysite was also revealed via scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The obtained data sets were used to explore the factors governing the phyllosilicate mineral in the formation of regolith-hosted REE deposits. This study highlights the influence of hydrodynamic conditions on the transformation behaviors of kaolinite and the genesis of regolith-hosted REE deposits.

Renju regolith-hosted REE deposit

The Renju REE deposit is located in northeast Guangdong, South China. The total rare earth oxide (REO) resources of the Renju deposit have been estimated as ~20,000 tons, with an average grade of ore body ranging from 0.153 to 0.197 wt.% (Wang and Xu, 2016). More than 70% of the REEs in the deposit occur as exchangeable ions (Yang and Xiao, 2011). The deposits were developed in hills, with elevations varying from 250 to 350 m above sea-level and slope gradients generally <25° (Huang et al., 2021a). The Renju area is characterized by the subtropical monsoon climate, with an average temperature of 20.6°C, and annual precipitation of 1500–2000 mm (Huang et al., 2021b), which is favorable for the chemical weathering of granitoid rocks and the further formation of thick regolith. This monsoon climate formed before 41 Ma and became modern-like after 26 Ma, as recorded by stratigraphy and paleoclimate proxies, e.g. pollen content of xerophytic plants (Wu et al., 2022).

The regolith hosting the REE deposit developed on bedrock of Mesozoic quartz diorite, granite porphyry, and biotite granite (Huang et al., 2021b; Zhao et al., 2021). Ore bodies containing >500 ppm REEs occur in the completely weathered zone (depths of 10–34 m) of the regolith (Fig. 1; Table 1). Based on the systematic variations in Fe oxide and Ce anomaly, the boundary between the vadose and saturated zones in the Renju regolith was estimated at a depth of 10 m (Huang et al., 2021a).

Table 1. Summary of the geochemical and mineralogical characteristics of the Renju regolith-hosted REE deposit

The original data for chemical index of alteration (CIA), whole-rock rare earth element (REE), and ion-exchangeable REE (iREE) are from Huang et al. (2021a).

Figure 1. (a) Simplified geologic map of the Renju regolith-hosted REE deposit, revised according to the report of the Regional Geological Survey Team of Jiangxi Geological Bureau (1976). (b) Original data sets of REE concentrations, the chemical index of alteration (CIA) values, and depth of the vadose zone–saturated zone boundary were obtained from Huang et al. (2021a).

Sampling and methods

Sampling

Samples were selected from a 34 m drill core in the Renju REE deposit. The exposed topsoil (~1 m) was removed, and a total of 21 bulk samples were selected from depths ranging from ~2 to 34 m, with intervals of 1–2 m. The whole-rock samples were dried at room temperature and milled into 200 mesh powders. The clay fractions (equivalent spherical diameter <2 μm) (Schroeder, 2018) were concentrated by dispersing the whole-rock samples using a sodium hexametaphosphate solution in an ultrasonic bath and extracting the uppermost sedimentation according to the principles of Stokes’ law. The clay fractions were subsequently freeze-dried. The whole-rock samples were used for identification and semi-quantification of primary and secondary minerals. The clay fractions were used for phase identification, semi-quantitative analysis of the various phyllosilicate mineral phases, and evaluation of the structural order of kaolinite.

Methods

X-ray diffraction (XRD)

The analyses were performed using a Rigaku MiniFlex-600 X-ray diffractometer in the Key Laboratory of Mineralogy and Metallogeny, Guangzhou Institute of Geochemistry, Chinese Academy of Sciences. The diffraction patterns were recorded by a 1 D detector, using CuKα radiation filtered with Ni operated at a current of 15 mA and voltage of 40 kV. The mineral phases were analyzed using JADE 6.5, referring to standard PDF cards from the International Centre for Diffraction Data (ICDD^®) (Blanton and Gates-Rector, 2019).

Phase identification. Randomly oriented powders of the whole-rock samples were scanned from 3 to 80°2 at 10°2θ min^–1. The diffractions at ~3.3 and 4.3 Å were related to quartz, and the diffraction at ~3.3 Å was derived from orthoclase. The basally oriented powders of the clay fraction samples were prepared by pipetting the water-dispersed samples onto glass slides and air-drying. The basally oriented powders were then scanned at 3°2θ min^–1 from 3 to 30°2θ. The diffraction at 10 Å was related to illite and halloysite, and the diffraction at 7.2 Å was related to kaolinite and halloysite. The 120°C-heated samples were prepared by heating the basally oriented samples in an oven at 120°C for 6 h, so that the interlayer water in halloysite-10 Å was lost due to heating (Joussein et al., 2006). The diffraction at 10 Å was mostly derived from illite in the 120°C-heated samples. The formamide-intercalated samples were prepared by intercalating the 120°C-heated samples with formamide for 20 min (Churchman et al., 1984). The diffraction at 7.2 Å was mostly derived from kaolinite in the formamide-intercalated samples, as the halloysite-7 Å had expanded to 10 Å due to intercalation. The 120°C-heated and formamide-intercalated samples were scanned at 3°2θ min^–1 from 3 to 30°2θ.

Semi-quantification of mineral abundance. Semi-quantitative analyses were performed using a ratio method (Chung, 1974) based on the diffraction area of each mineral using:

(1) $$ {W}_{\unicode{x03C1}}=\frac{\frac{S_{\unicode{x03C1}}}{K_{\mathrm{cor}}^{\unicode{x03C1}}}}{\sum_{\mathrm{j}=1}^{\mathrm{n}}\frac{S_{\mathrm{j}}}{K_{\mathrm{cor}}^{\mathrm{j}}}}, $$

where W _ρ refers to the abundance of mineral ρ; S _ρ refers to the area of the most intense diffraction of mineral ρ; and K ^ρ_cor refers to the reference intensity ratio (RIR) value of mineral ρ.

Structural order of kaolinite. The clay fractions were used for characterizing the structural order of kaolinite. Randomly oriented powders of clay fractions were scanned from 3 to 70°2θ at 3°2θ min^–1 for evaluation the structural order of kaolinite, based on the XRD pattern in the ranges of 20–23°2θ and 35–40°2θ (Fialips et al., 2000; Bauluz et al., 2008; Ni et al., 2021). Effective XRD indices include the Hinckley index (HI), Liétard index (R2), Stoch index (IK), Range and Weiss index (QF), and full width at half maximum (Aparicio and Galan, 1999; Zadvernyuk et al., 2021). Among them, HI and R2 are commonly used to characterize the defect structures of kaolinites (Fialips et al., 2000; He et al., 2005; Ishida et al., 2018; Pineau et al., 2022). The kaolinite structural order indices used for this study were HI Eqn 2; Hinckley, 1962) and R2 Eqn 3; Liétard, 1977) according to the powder XRD patterns using:

(2) $$ \mathrm{HI}=\frac{A+B}{At}, $$

and

(3) $$ \mathrm{R}2=\frac{\frac{1}{2}(K1+K2){\textstyle - }k}{\frac{1}{3}(K1+K2+k)}, $$

where A is the height of the ( $ 1\overline{1}0 $ ) diffraction peak to the background line drawn from the trough between (020) and ( $ 11\overline{1} $ ); B is the height of the ( $ 1\overline{1}1 $ ) diffraction peak to the background line drawn from the trough between (020) and ( $ 1\overline{1}1 $ ); At is the height of the ( $ 1\overline{1}0 $ ) diffraction peak above the general background; K1 is the height of the (131) diffraction peak; K2 is the height of the ( $ 1\overline{3}1 $ ) diffraction peak; and k is the height of the trough between ( $ 1\overline{3}1 $ ) and (131).

Accordingly, kaolinite of different structural orders can be classified into low-defect kaolinite (HI=0.90−1.15; R2>1.20), medium-defect kaolinite (HI=0.50–0.90; R2=0.70–1.20), and high-defect kaolinite (HI<0.50; R2<0.70) (Aparicio et al., 2006).

REE adsorption capacity calculation

The REE adsorption capacity (AC) of regolith was calculated by summing the REE adsorption capacity of each phyllosilicate mineral in a regolith using:

(4) $$ \mathrm{AC}={\sum}_{\unicode{x03C1} =1}^{\mathrm{n}}{W}_{\rho }{C}_{\rho }, $$

where W _ρ refers to the abundance of mineral ρ; and C _ρ refers to the saturated REE adsorption capacity of mineral ρ. The REE AC was calculated mainly based on the AC of kaolinite and halloysite, which accounts for 85–100% of the total abundance of phyllosilicate minerals in the regolith. The kaolinite and halloysite REE AC values used for calculation were 3460 and 1610 ppm (at pH=5), which were experimentally determined by Yang et al. (2019).

The concentration of whole-rock REE (Table 1) was obtained by analyzing the whole-rock powders (200 mesh) using a Thermo iCAP Qc inductively coupled plasma-mass spectrometer (ICP-MS) (Huang et al., 2021a). Before the ICP-MS analysis, the samples were dried at 105°C for 3 h and then baked at 550°C for 3 h to eliminate organic material. Approximately 0.04 g of each solid sample was analyzed. Rh was added to each sample as an internal standard to calibrate the drift of the instrument during the measurements. The analytical precision for the whole-rock REE content was better than 3% RSD (Relative Standard Deviation).

The concentration of ion-exchangeable REE (iREE) (Table 1) represents the REE ions adsorbed by clay minerals in the regolith. The concentration of iREE was acquired by mixing 1.00 g of powdered whole-rock samples (200 mesh) with 25 mL of 1.0 mol L^–1 MgCl solution (pH=7±0.2) in centrifuge tubes. The centrifuge tubes were then shaken at room temperature (25±0.2°C) for 2 h, after which the supernatant was collected for ICP-MS analysis (Huang et al., 2021a).

Scanning electron microscopy

The bulk samples were embedded with epoxy and subsequently made into thin-section samples. The thin sections were coated with carbon for SEM analyses. The paragenesis characteristics of phyllosilicate minerals in the Renju REE deposit were investigated using a MIRA3 TESCAN scanning electron microscope in the secondary electron mode at a voltage of 20 kV. The surface elemental composition was characterized using an EDAX Element EDS detector for phyllosilicate identification.

Transmission electron microscopy

The separated clay fractions were ultrasonically dispersed in ethanol, and the suspension was transferred onto a porous carbon film supported by a copper grid. Nano-scale morphological features of the phyllosilicate minerals were examined using a transmission electron microscope (Thermo Scientific FEI Talos F200S) at 200 kV.

Results

Phyllosilicate minerals in the Renju regolith

Composition of phyllosilicate minerals

All XRD patterns for randomly oriented whole-rock samples showed sharp XRD peaks of quartz at ~3.3 and 4.3 Å (Fig. 2). The diffraction of orthoclase at ~3.3 Å presented in samples from the depth of 10–34 m in the regolith, but disappeared in the samples <10 m from the top (Fig. 2). The profiles featured characteristic diffractions at ~7.2 and 10 Å. The 10 Å diffraction corresponded to a mixture of the (001) plane of illite and the (001) plane of halloysite-10 Å, and the 7.2 Å diffraction corresponded to the (001) plane of kaolinite and the (001) plane of halloysite-7 Å. Specifically, for the heated samples, the decrease in diffraction intensities at 10 Å (I₁₀) was consistent with the increase in diffraction intensities at 7.2 Å (I_7.2), and the I_7.2/I₁₀ ratios of the heated samples were 3–15% higher than those of the basally oriented samples, suggesting the presence of halloysites at ~10 Å. Moreover, the formamide-intercalated samples exhibited a decrease in the diffraction intensities at ~7.2 Å and an equivalent increase at ~10 Å compared with the basally oriented samples (Fig. 2). Accordingly, the 28–72% decrease in the I_7.2/ I₁₀ ratios was attributable to the expansion of halloysite (001) plane from ~7.2 to 10 Å (Churchman and Theng, 1984; Churchman and Gilkes, 1989). The diffraction at ~7.2 Å of the formamide-intercalated samples and diffractions at ~10 Å of the heated samples in the profiles were contributed by kaolinite and illite, respectively.

Figure 2. X-ray diffraction (XRD) patterns of randomly oriented whole-rock samples and oriented clay fractions (<2 μm) from various sections. WR = whole-rock powder XRD samples; O = basally oriented samples; H = 120°C-heated samples; Fm = formamide-intercalated samples; Kln = kaolinite; Hly = halloysite; Ilt = illite; Qz = quartz; Or = orthoclase.

Abundance of phyllosilicate minerals

The abundance of total phyllosilicate minerals fluctuated between 40 and 85 wt.% (Fig. 3; Table 1). Kaolinite and halloysite were the dominant secondary phyllosilicate minerals, accounting for 40–80 wt.% of the regolith sample. According to the abundance and distribution of the phyllosilicate minerals, the investigated regolith can be divided into three sections, and Section III can be subdivided into two sections; these sections were numbered from the completely weathered zone upward to the topsoil.

Figure 3. Variation in the abundances of phyllosilicate minerals (clay), kaolinite (Kln), halloysite (Hly), illite (Ilt), ion-exchangeable REE (iREE), and REE adsorption capacity (AC) along the profile of the Renju regolith-hosted REE deposit. The iREE data were obtained from Huang et al. (2021a).

Section I was located at a depth ranging from 28 to 34 m, and was characterized by a relatively small abundance (43–54 wt.%) of phyllosilicate mineral. Phyllosilicate minerals commonly occurred along the micro-fractures and grain boundaries of the weathered primary minerals as aggregations of <10 μm (Fig. 4a). The abundance of kaolinite decreased from 46 to 33 wt.% upward along the profile, whereas the abundance of halloysite increased from 4 to 16 wt.%, and that of illite fluctuated between 2 and 4 wt.% (Fig. 3; Table 1).

Figure 4. Secondary electron micrograph of minerals in (a) Section I, (b) Section II, (c) Section III-1, and (d) Section III-2. The upper right inset in (c) presents the phyllosilicate mineral aggregations in Section III-1. Kln = kaolinite; Hly = halloysite; Ilt = illite; Or = orthoclase.

Section II was located at a depth ranging from 18 to 28 m and was characterized by a relatively large abundance (57–61 wt.%) of phyllosilicate minerals. In this section, the abundance of kaolinite fluctuated between 23 and 45 wt.%. Notably, the abundance of halloysite continually increased from 7 to 31 wt.% upward along the profile in Section II, whereas the abundance of illite decreased significantly from 14 to 2 wt.% (Fig. 3; Table 1). The size of phyllosilicate minerals examined by SEM exceeded 20 μm. The phyllosilicate minerals were identified based on the EDS results (Fig. S1). The 2:1 phyllosilicate mineral (illite) showed Si to Al ratios close to 2:1, and the 1:1 phyllosilicate mineral (kaolinite and halloysite) approaching 1:1. The 1:1 type kaolinite and halloysite microcrystals occurred at the edge of 2:1 type illite grains (Fig. 4b).

Section III was located at a depth ranging from 2 to 18 m, and the total phyllosilicate mineral abundance was 60–85 wt.%. The section was dominated by 1:1 phyllosilicate minerals, and the phyllosilicate mineral aggregations usually exceeded 100 μm. According to the distinctive variations in the abundances of kaolinite and halloysite, Section III can be subdivided into Sections III-1 and III-2.

Section III-1 was located at a depth ranging from 10 to 18 m. The abundance of halloysite in this section increased significantly from 12 to 55 wt.% upward along the profile, whereas the abundance of kaolinite decreased sharply from 58 to 14 wt.%. The abundance of illite ranged between 1 and 6 wt.% (Fig. 3; Table 1). The phyllosilicate minerals form irregular aggregates with massive 100–500 nm interstices, and the interstices between kaolinite plates were filled with curved halloysite tubes (Fig. 4c).

Section III-2 was located at a depth ranging from 2 to 10 m in the regolith. The abundance of kaolinite increased abruptly from 48 to 74 wt.% upward along the profile, while the abundance of halloysite decreased significantly from 19 to 6 wt.%. The abundance of illite decreased from 3 to 1 wt.% (Fig. 3; Table 1). The kaolinite plates featured regular pseudo-hexagonal edges and formed aggregates occurring as vermicular booklets (Fig. 4d).

Structural order of kaolinite

The HI and R2 values are shown in Fig. 5 and Table 1. For Section I, the HI and R2 values fluctuated along the section between 0.48 to 0.78 and 0.94 to 1.00, respectively. The corresponding TEM results revealed that kaolinite aggregates in this section were mainly composed of thinner plates in face–face contact and featured a relatively pseudo-hexagonal morphology (Fig. 6a,b).

Figure 5. (a) X-ray diffraction patterns of randomly oriented clay fractions (<2 μm) from various depths, and variation in (b) the Hinckley index (HI) and (c) the Liétard index (R2) along the profile.

Figure 6. Transmission electron microscopy (TEM) images of kaolinite in (a–b) Section I, (c–d) Section II, (e–f) Section III-1, and (g–h) Section III-2. The dashed lines indicate rolled kaolinite, and the short arrows indicate the roll and unroll directions. Kln = kaolinite; Hly = halloysite.

For Section II, the HI and R2 values slightly decreased upward along the profile, with HI varying from 0.77 to 0.45 and R2 varying from 0.98 to 0.87. The morphologies of kaolinite aggregates showed little change along the profile, and kaolinite plates with flat edges were stacked to form pseudo-hexagonal kaolinite aggregates (Fig. 6c,d).

For Section III-1, the HI and R2 values decreased abruptly from 0.65 to 0.42 and 0.95 to 0.84 upward along the profile, respectively. Some kaolinite aggregates exhibited complex surface morphologies, featuring rounded edges and uneven surfaces, and the aggregates contained fewer kaolinite plates. In addition, some kaolinite plates with relatively flat edges curled at the edges at different angles and formed halloysite tubes and spheres (Fig. 6e,f).

For Section III-2, the HI value increased from 0.27 at the bottom to 0.67 at the top. Similarly, R2 increased from 0.85 at the bottom to 1.02 at the top. Kaolinite plates with relatively straight boundaries and angular edges aggregated and formed regularly shaped aggregates, and the tubular halloysite began to unfold (Fig. 6g,h).

REE adsorption capacity

The simulative REE adsorption capacity results are shown in Table 1 and Fig. 3. REE adsorption capacity in Section I and Section II fluctuated between 1215 and 1692 ppm, then decreased from 2177 to 1542 ppm from the bottom to the top of Section III-1, and finally increased from 1971 to 2661 ppm from the bottom to the top of Section III-2.

Discussion

Effects of chemical weathering on the evolution of phyllosilicate minerals in regolith

The transformation process of phyllosilicate minerals in regolith layers is essentially related to chemical weathering intensity (Huang et al., 2022), which can be described using the chemical index of alteration (CIA = Al₂O₃/(Al₂O₃ + CaO + Na₂O + K₂O) × 100) (Nesbitt and Young, 1982). The CIA values of the Renju profile gradually increased at shallower depths (Fig. 1). Chemical weathering leads to the dissolution of primary minerals and the formation of secondary phyllosilicate minerals owing to the gradual loss of mobile elements, such as K, Na, Ca, and Si, and the accumulation of immobile elements such as Al (Banfield and Eggleton, 1990; Wilson, 2004). Consequently, the abundance of secondary phyllosilicate minerals increased with the progressive weathering of primary minerals in the regolith (Fig. 3).

Section I (CIA = 72–80) corresponded to the initial weathering stage. At this stage, secondary phyllosilicate minerals are formed in the following order, from primary minerals to 2:1 phyllosilicate minerals and then to 1:1 phyllosilicate minerals, or from primary minerals to 1:1 phyllosilicate minerals (Eggleton and Buseck, 1980; Senkayi et al., 1984; Banfield and Eggleton, 1990), depending on the order and rate of elements released from primary rock-forming minerals (Chou and Wollast, 1984; Chou and Wollast, 1985; Muir et al., 1990). Micas and plagioclase from bedrock were readily weathered into secondary phyllosilicate minerals in Section I, due to their relatively weak resistance to weathering (Banfield and Eggleton, 1990; Schroeder et al., 2022). Orthoclase was the major rock-forming mineral in Section I that contributed to the increase of secondary phyllosilicates from Section I to III (Fig. 2 & Fig. 4a). Previous studies have shown that Al–O bonds demonstrate lower dissolution activation energy than Si–O bonds in plagioclase tetrahedra; therefore, Al tends to be released from plagioclase, to form a Si-rich layer at the plagioclase surface (Chou and Wollast, 1984; Casey et al., 1989). The present study showed that the high Si/Al ratio of the Si-rich layer facilitated the formation of 2:1 phyllosilicate mineral (e.g. illite) when it reacted with the released Al cations in the ambient solution (Fig. 4a). The relatively low Si/Al ratio in the ambient solution possibly led to the formation of 1:1 phyllosilicate minerals (e.g. kaolinite and halloysite) (Chou and Wollast, 1984; Banfield and Eggleton, 1990; Devidal et al., 1996). As a result, both the 2:1 and 1:1 phyllosilicate minerals formed due to the chemical weathering of orthoclase, as demonstrated by the positive correlation between the abundances of illite (2:1 type) and halloysite/kaolinite (1:1 type) in Section I (Fig. 7a; R ² = 0.85).

Figure 7. Plots of (a) kaolinite and halloysite abundance versus illite abundance, and (b) halloysite abundance versus kaolinite abundance in regolith samples.

Section II (CIA = 73–84) corresponded to the intermediate weathering stage. At this stage, the abundance of 2:1 illite significantly decreased in more weathered depths, whereas the total abundance of 1:1 halloysite and kaolinite gradually increased (Fig. 3), resulting in a negative correlation between the illite abundance and the total halloysite and kaolinite abundance (Fig. 7a; R ² = 0.96). Accordingly, the 2:1 illite gradually transformed to 1:1 kaolinite and halloysite in the intermediate weathering stage (Fig. 4b), and the transformation could proceed via the mechanisms of dissolution-precipitation or solid-state transformation (Cuadros, 2012).

Effects of hydrological conditions on the mutual transformation between kaolinite and halloysite

Section III (CIA = 83–99) was mainly composed of kaolinite and halloysite with a small amount of illite, corresponding to the advanced weathering stage (Fig. 2). The weak correlation between the abundance of illite and the abundance of kaolinite and halloysite (Fig. 7a) was indicative of insignificant transformation of 2:1 phyllosilicate minerals to 1:1 phyllosilicate minerals at this weathering stage. Moreover, the abundance of kaolinite gradually decreased upward along the profile in Section III-1, but gradually increased in Section III-2 (Fig. 3). The negative correlation between the abundances of kaolinite and halloysite (Fig. 7b) suggested that kaolinite and halloysite underwent mutual transformation in Section III. As halloysite usually forms in a wet environment while kaolinite favors a largely dry environment (Churchman et al., 2016), the mutual transformation between kaolinite and halloysite can be attributed to variations in hydrological conditions.

Section III-1 was located in the vicinity of the vadose–saturated zone boundary (Fig. 1), where seasonal precipitation created an alternating wetting and drying zone (Pepper and Gentry, 2015). Kaolinite (Al₂Si₂O₅(OH)₄) is composed of one Si–O tetrahedral sheet and one Al–O octahedral sheet. Wetting can cause hydration of kaolinite by introducing interlayer water molecules (Churchman and Carr, 1975; Costanzo and Giese, 1985), thus disrupting the interlayer hydrogen bond and increasing the size of the Si–O tetrahedral sheet. The increased size of the Si–O tetrahedral sheet relative to the Al–O octahedral sheet possibly caused a mismatch between the apical oxygen plane and the inner OH plane (Radoslovich, 1963; Singh and Gilkes, 1992a; Singh, 1996; Bobos et al., 2001), leading to bending and rolling of the 1:1 layers (Fig. 6e–f). The rolling axis was suggested to be parallel to the b-axis with the Al–O octahedral sheet as the inner surface (Singh and Mackinnon, 1996). In addition, drying could have induced the loss of lumen water (Santagata and Johnston, 2022), thereby creating evacuated lumen space for the inward rolling of the hydrated 1:1 layers (Fig. 6e–f). Therefore, the alternate wetting and drying condition facilitated the kaolinite-to-halloysite (K-to-H) transformation, which is consistent with the abrupt increase in halloysite abundance in Section III-1 (Fig. 3).

Section III-2 was located in the vadose zone (Fig. 1), where the soil moisture significantly decreased owing to infiltration and evaporation of vadose zone water (Wang et al., 2017). The 1:1 layer of halloysite (Al₂Si₂O₅(OH)₄·nH₂O) features 0–2 interlayer H₂O molecules per unit cell. In the vadose zone, dehydration can cause the loss of interlayer H₂O molecules in halloysite under the sustained drying conditions (Kohyama et al., 1978; Costanzo and Giese, 1985; Inoue et al., 2012). With the loss of H₂O molecules, the interlayer hydrogen bond linking the OH groups on the octahedral basal surface and the tetrahedral oxygen could drive the rotation of the tetrahedron and resolve the volume mismatch between the tetrahedral and octahedral sheets (Radoslovich, 1963; Singh, 1996), thus facilitating the unrolling of tubular halloysite into lamellar kaolinite in Section III-2 (Fig. 6g,h). Therefore, the transformation of halloysite to kaolinite and the increase in kaolinite abundance in Section III-2 were probably induced by the dry environment of the vadose zone (Fig. 3).

Factors controlling variations in the structural order of kaolinite

Kaolinite structural disorder can be caused by layer displacement and octahedral vacancy displacement (Plançon et al., 1989). In the Renju regolith, the HI values (0.27–0.78) were indicative of a range of high to medium defect levels for the kaolinite, while R2 values (0.79–1.05) suggest medium defect levels (Fig. 8) (Aparicio et al., 2006). The ( $ 1\overline{3}1 $ ) and (131) diffraction used for calculating R2 values are related to the variation of octahedral vacancy (Plançon and Zacharie, 1990). The variations of R2 values in kaolinite from different sections (Fig. 5) indicated variation in proportion of the octahedral vacancy order of kaolinite along the Renju profile. Besides, the HI values can be significantly reduced due to the presence of halloysite, particularly when halloysite is abundant (Aparicio and Galan, 1999); thus the discrepancy between the HI and R2 can be ascribed to the interference from halloysite throughout the Renju regolith (Fig. 8). Accordingly, R2 is more suitable for evaluating the structural order of kaolinite in Renju weathering regolith, as it could be less influenced by halloysite.

Figure 8. Plot of the Hinckley index (HI) versus the Liétard index (R2) in regolith samples.

The kaolinite in Sections I and II was mainly composed of newly crystallized grains resulting from the dissolution of orthoclase and illite. The relatively stable physicochemical and hydrological conditions of Sections I and II provided a stable moisture and ionic strength environment (Huang et al., 2021a) for kaolinite crystallization, thereby facilitating the formation of euhedral kaolinite grains (Fig. 6a–d). Consequently, kaolinite exhibited a stable structural order, and its abundance gradually increased at shallower depths in Sections I and II (Fig. 5).

In Section III-1, kaolinite exhibited an abrupt decrease in structural order compared with Section II. The kaolinites could possibly become more defective with time due to radioactive decay of surrounding K and Th (Mathian et al., 2020), as kaolinite in Section III-1 was more aged than that in Section II. However, kaolinite from shallower Section III-2 was less defective than less weathered Section III-1, suggestive of the remarkable influence of hydrological conditions on the structural order of kaolinite in Section III. The decreased R2 values (Fig. 5c) indicated an increase in the disordered octahedral vacancies. During the K-to-H transformation, kaolinite reduced the distance between apical oxygens and the Si–O–Si angle in the tetrahedral sheet (Singh and Gilkes, 1992a; Singh, 1996) at the curly edges (Fig. 6f), and halloysite tubes were formed in the interstices of kaolinite (Fig. 4c). Kaolinite dissolution, as evidenced by the morphological features of round and irregular edges (Fig. 6e), probably facilitated the detachment and replacement of octahedral and tetrahedral cations (Wieland and Stumm, 1992; González Jesús et al., 2000; Tyagi et al., 2006), and tended to produce disordered octahedral vacancies (Sakharov et al., 2016), thereby reducing kaolinite structural order. Consequently, the K-to-H transformation induced kaolinite dissolution in Section III-1 and reduced both the structural order and abundance of kaolinite.

In Section III-2, the dehydration of halloysite (Fig. 6h) strengthened the hydrogen bond between the dehydrated layers of halloysite and drove the rotation of the Si–O tetrahedra of halloysite about their apices in opposite directions, causing Si and O to move closer to the ditrigonal ring center (Singh, 1996). The unrolling of halloysite along the crystalline axis possibly led to an increase in the regularity of vacancy stacking (Plançon et al., 1989; Fashina and Deng, 2021), as is displayed by the R2 values (Fig. 5c). Therefore, the decrease in the kaolinite defect density in Section III-2 was affected by the halloysite-to-kaolinite (H-to-K) transformation, and the stacking of well-crystallized triclinic kaolinite along the c-axis led to the formation of vermicular kaolinite in Section III-2 (Fig. 4d).

Underlying relationships between phyllosilicate mineral evolution and regolith-hosted REE mineralization

The vertical zonation of regolith can be defined by boundaries where the mineral composition and structural order of kaolinite became significantly varied. Previous studies suggested that changes in pH throughout the regolith largely influenced the adsorption capacity of phyllosilicate minerals and the distribution of REE mineralization (Bao and Zhao, 2008). However, the pH values change only between 4.7 and 6.3, and thus are unlikely to lead to significant changes in the adsorption capacity of phyllosilicate minerals of different sections in the studied regolith (Huang et al., 2021a). A previous study reported REE enrichment at the transition zone between upper and lower pedolith, spatially coincident with mineral transformation between halloysite and kaolinite (Li and Zhou, 2020, 2023). In Renju regolith, REE enrichment started at the interface between Section III-1 and Section III-2, where the type of mineral transformation changed from K-to-H to H-to-K. Li and Zhou (2020) suggested that mineral transformation led to REE desorption in regolith-hosted REE deposit due to a decrease in adsorption capacity, given the fact that poorly crystallized kaolinite in Section III-1 can provide more adsorption sites for REEs and facilitate the formation of REE ore body (Fialips et al., 2000; Ndlovu et al., 2015). However, the present study shows that the regolith above Section III-1 does not reach saturation regarding the adsorption of REEs (Table 1; Fig. 3), when evaluated with relatively under-estimated adsorption capacity of kaolinite and other phyllosilicate minerals (Singh and Gilkes, 1992b). Therefore, the desorption of REE from Section III-2 and their enrichment in Section III-1 were probably influenced by other factors.

Note that the interface between Section III-1 and Section III-2 was characterized by mineral transformation and significant kaolinite structural order variation. The present study demonstrated that both the K-to-H transformation and kaolinite structural order were essentially related to the alternating wetting and drying hydrological condition (Keller, 1978; Hurst and Kunkle, 1985; Metz and Ganor, 2001). The fluctuations in water table were largely controlled by precipitation (Wang and Pozdniakov, 2014; Hussain et al., 2022), thus creating the alternating wetting and drying hydrological condition between the highest and the lowest water table level within the regolith. As a result, REEs above the alternating wetting and drying interface were vertically transported by vadose zone water and fixed by phyllosilicate minerals around the water table owing to a significant decrease in the seepage velocity (Huang et al., 2021a), resulting in poor REE content in Section III-2 and initial REE enrichment at the interface (Fig. 1). The fixed REE ions can be partially desorbed and transported downward owing to the decrease in the water level. Additionally, REE-oxyhydroxide nanoparticles could also migrate via intense downward leaching and accumulate near the water table (Calabrese et al., 2018). Consequently, the REEs became significantly enriched in the lowest level of the water table (Fig. 1 & Fig. 9). Therefore, hydrodynamic conditions controlled both the REE enrichment and the kaolinite structural order in the Renju profile, forming the spatial coupling among the alternating wetting and drying zone, REE enrichment and mineral structural order (Fig. 9). Accordingly, the kaolinite structural order indices can be utilized to indicate the location of the regolith-hosted REE deposit in the weathering regolith (Fig. 9).

Figure 9. Schematic model illustrating the effects of the hydrodynamic conditions on the abundance of phyllosilicate minerals, kaolinite structural order, and REE enrichment. The shaded area represents the alternating wetting and drying zone. Kln = kaolinite abundance; Hly = halloysite abundance; Ilt = illite abundance; KC = kaolinite structural order; REE = REE concentration.

Conclusions

This study investigated the variations in the abundance of phyllosilicate minerals and the structural order of kaolinite in the regolith. The weathering of the primary mineral orthoclase and 2:1 phyllosilicate mineral was influenced by the intensity of chemical weathering. The hydrological conditions in the Renju regolith can explain well the abrupt changes in both the abundance and structural order of kaolinite in Section III. In the region around the vadose–saturated zone boundary, an alternating wetting and drying zone provided a favorable kinetic environment for intensive kaolinite-to-halloysite transformation and kaolinite dissolution. The decrease in abundance of kaolinite in Section III-1 coincided with the decrease in structural order of kaolinite during kaolinite-to-halloysite transformation and kaolinite dissolution. As soil moisture decreased in the vadose zone, halloysite lost its interlayer water and underwent dehydration to form well-crystallized kaolinite. Accordingly, this work would be the first documented case of the occurrence of mutual transformation between kaolinite and halloysite at Earth’s surface conditions. The REEs were transported vertically by vadose water to the highest water table level, then moved along with the fluctuating water tables and formed prominent REE enrichment at the lowest water table level in Section III-1. Hydrodynamic conditions affected both kaolinite transformation and REE enrichment; a low kaolinite abundance and structural order can serve as an indicator of alternating wetting and drying zones and the occurrence of regolith-hosted REE mineralization in a weathering regolith.