Sulawesi Island has a complex tectonic setting, which is a result of the collision between the Eurasian, Indian-Australian, and Pacific plates during the Late Cretaceous and Early Tertiary (Katili, 1978; Wilson and Moss, 1999; Hall and Wilson, 2000; Kadarusman et al., 2004). Consequently, the K-shaped Sulawesi island is surrounded by thrusts and has four distinct litho-tectonic belts from west to east: 1) West Sulawesi plutono-volcanic arc, 2) Central Sulawesi metamorphic belt, 3) East Sulawesi ophiolite, and 4) microcontinental blocks of Banggai-Sula and Buton-Tukang Besi (Hall and Wilson, 2000). The study area, Morowali deposit, is located in the East Sulawesi ophiolite belt (Figure 1).

According to the tectonic reconstructions of Southeast Asia, the Sulawesi ophiolite was accreted onto the Eurasian plate during subduction of the Indian-Australian plate in the Late Oligocene (Hall, 1997; Parkinson, 1998). The East Sulawesi ophiolite is considered as one of the largest ophiolites in the world, covering more than 15,000 km² with a total length of 700 km (Monnier et al., 1995; Kadarusman et al., 2004; Van der Ent et al., 2013). The East Sulawesi ophiolite comprises two lithological units: 1) Neogene and Quaternary sediments; 2) ultramafic and mafic rocks (Figure 1). The ophiolite belt is 70% peridotites, composed of lherzolite, harzburgite, and occasional dunite (Kadarusman et al., 2004). The Morowali area principally consists of harzburgite and lherzolite; however, these peridotites are highly altered to serpentinite, making it difficult to identify their original mineral composition and texture. The ophiolitic mélange zone belonging to the Central Sulawesi metamorphic belt is also exposed near the study area along with the active and earthquake-inducing Matano fault (Stevens et al., 1999).

The bedrock lithology, tectonic setting, age of weathering with respect to tectonic activity, climate, and topography control Ni-rich laterite formation (Freyssinet et al., 2005). The Morowali area has a tropical rainforest climate with annual average temperature of 27°C and 2,000 mm of annual precipitation, which makes the area thickly forested with a diverse vegetation in accordance with the elevation. The study area has a moderate relief and gentle slope. Furthermore, the well-developed drainage system of the bedrock promotes Ni-rich laterite profiles in a large area and aids in the formation of high-grade Mg-silicate minerals.

Before the powder XRD analysis, three different colors of garnierite (creamy yellow, grass green, and jade green) were separated from the saprolite samples by handpicking. Each sample was ground into the size of 200 mesh in an agate mortar. XRD analysis was carried out on the garnierite samples using the Rigaku SmartLab with 2θ angles from 5°to 80°, step size of 0.02°, and step duration of 2°s. The mineral phases were indexed using PDXL software and PDF-2 database.

Six thin sections from the bedrock and saprolite horizon were prepared for EPMA and SEM analyses. Quantitative mineral analyses were performed using field emission EPMA (JEOL JXA-8530F) and field emission SEM (JEOL JSM-7100F). The field emission EPMA is equipped with five WDS channels and one EDS channel. The applied accelerating voltage was 15 kV, the probe current was 20 nA, and the beam size was 3 μm. Calibration was performed using natural and synthetic standard materials: jadeite (Na), albite (Si), fayalite (Fe), K-feldspar (K), wollastonite (Ca), MgO (Mg), Al₂O₃ (Al), MnO (Mn), TiO₂ (Ti), Cr₂O₃ (Cr), and NiO (Ni). The field emission SEM had an accelerating voltage 15 kV, working distance of 10 mm and an emission current of 53 nA.

Twelve samples from the bedrock (B-2, B-3, and G-2), saprolite (G-1, G-3, S1-1, S1-2, S2-1, S2-2, S3-1, and S3-2), and limonite horizon (L-1) were powdered into 200 mesh in an agate mortar for whole rock analysis. Major (SiO₂, Al₂O₃, TiO₂, Fe₂O₃, MgO, CaO, Na₂O, K₂O, MnO, and P₂O₅) and trace elements (NiO and Cr₂O₃), and SO₃ were determined using Shimadzu XRF-1700. Loss on ignition (LOI), which is the weight loss of the sample after burning at 950°C was also measured. Samples were analyzed using lithium tetraborate (Li₂B₄O₇) glass beads. The determination limit was better than 10 ppm, and the analytical uncertainties are less than ±2%.

Table 1 summarizes the characteristics of garnierites and their structural formulae. According to XRD and EPMA analyses, garnierite of the Morowali deposit can be distinguished into serpentine- and talc-like. The serpentine-like garnierite occurs in creamy yellow and grass green varieties. XRD pattern of the creamy yellow garnierite (S1-2 and S3-1) shows chrysotile and lizardite phases (Figure 4A). The creamy yellow garnierite contains 42.17–44.26 wt% of SiO₂, 30.88–36.83 wt% of MgO, 4.14–6.43 wt% of FeO, and 0.76–8.41 wt% of NiO (Table 2). The concentrations of TiO₂, Al₂O₃, Cr₂O₃, MnO, Co, CaO, Na₂O, and K₂O were <0.1 wt%. The general chemical formula of serpentine-like garnierite is (Mg,Ni)₃Si₂O₅(OH)₄ (Brindley, 1980; Brand, 1998; Freyssinet et al., 2005), and its structural formulae was calculated using seven oxygens (Brindley and Hang, 1973). Each element concentration is represented in atoms per formula unit (apfu) in Table 2. Elements under 0.01 apfu were considered as negligible for the structural formulae calculation. The calculated structural formula for the creamy yellow garnierite is (Mg_2.22-2.63Ni_0.03-0.33Fe_0.17-0.26)Si_2.05-2.14O₅(OH)₄.

The grass green garnierite (S1-2, S1-3, and S2-2) exhibited mixed XRD patterns with serpentine-like and talc-like structures (Figure 4B). The grass green garnierite is composed of 41.23–48.25 wt% of SiO₂, 6.70–37.39 wt% of MgO, 0.14–7.03 wt% of FeO, and a higher amount of NiO (0.76–38.26 wt%) than creamy yellow and talc-like garnierites (Table 2). The structural formula of grass green garnierite was also calculated in the same way as that of creamy yellow garnierite. The calculated structural formula for grass green garnierite is (Mg_0.53-2.65Ni_0.03-1.63Fe_0.01-0.30)Si_2.02-2.41O₅(OH)₄.

The talc-like garnierite (G-1 and G-3) exhibited a jade green color, and consists of comparatively higher concentrations of SiO₂ (56.76–59.42 wt%), MgO (25.91–27.93 wt%), and NiO (4.02–8.02 wt%) while the other elements exist in trace amounts (<0.1 wt%) (Table 2). The general chemical formula for talc-like garnierite is (Mg,Ni)₃Si₄O₁₀(OH)₂·nH₂O (Brindley, 1980; Brand, 1998; Freyssinet et al., 2005). The structural formulae of talc-like garnierite were calculated on the basis of 11 oxygen atoms (Brindley and Hang, 1973). Mg and Ni were allocated to the octahedral site, while Si was allocated to the tetrahedral site. Due to the trace amounts of Ti, Al, Cr, Fe, Mn, Co., Ca, Na, and K (<0.01 apfu), only Si (3.94–3.99 apfu), Mg (2.64–2.80 apfu), and Ni (0.22–0.44 apfu) were considered for calculation of the structural formulae (Table 2). The estimated structural formula is (Mg_2.64-2.80Ni_0.22-0.44)Si_3.94-3.99O₁₀(OH)₂·nH₂O.

Pyroxene and serpentine are abundant minerals in the bedrock saprolite horizon with minor occurrence of chromite. The EPMA data for these minerals are presented in Table 3. Pyroxenes in the bedrock and saprolite horizon have similar amounts of SiO₂ (51.35–56.88 wt%), MgO (15.86–33.43 wt%), CaO (0.94–23.2 wt%), FeO (2.32–6.13 wt%), and NiO (0.02–0.15 wt%). The wide ranges of MgO and CaO are probably due to clinopyroxene exsolutions in orthopyroxene (Figures 3A,B). Serpentines show a narrow range of SiO₂ (40.12–43.35 wt%) in the bedrock and saprolite samples, while MgO and NiO show a wide variation (27.54–37.1 wt% and 0.97–7.84 wt%, respectively). It was also observed that the NiO content was inversely proportional to the MgO content. This variation indicates a high degree of cation exchange between Ni and Mg at this level. Additionally, chromites with variable Al₂O₃, FeO, and MgO contents were also observed. Such chromites were distinguished into high-Al and low-Al chromites. The high-Al chromites contains relatively lower FeO (16.09–17.37 wt%) and higher MgO (13.76–14.88 wt%) than the low-Al chromites (57.39 wt% FeO and 1.05 wt% MgO).

The major and trace elements concentrations in the bedrock and saprolite horizon are generally similar (Table 4). SiO₂, MgO, and Fe₂O₃ were the most abundant in the bedrock and saprolite horizon. SiO₂, MgO, and Fe₂O₃ contents for the bedrock and saprolite horizon vary between 38.79–43.91 wt%, 27.65–36.27 wt%, and 7.6–10.44 wt%, respectively. However, those for the limonite horizon exhibited lower SiO₂ (2.59 wt%) and MgO (0.74 wt%), and higher Fe₂O₃ (66.9 wt%) values. The contents of immobile elements, such as Al₂O₃, Cr₂O₃, and MnO, are below 6 wt% in all the horizons. Minor amounts of CaO (1.89–0.02 wt%) and Na₂O (0.21–0.09 wt%) with extremely low K₂O and P₂O₅ (<0.01%) were also observed. NiO (0.45–5.14 wt%) is abundant in saprolite horizon and depleted in the bedrock.

Serpentinization-related extensive water–rock interaction might occur before laterization because most of the laterite deposits that develop on the bedrock are already serpentinized due to chemical reactions with seawater (Golightly, 1981). At low temperatures, the primary minerals in peridotites, such as olivine and pyroxene, are altered faster than the hydrous phases, such as brucite, serpentine, and talc, because of their instability, and the alteration process continues until equilibrium is reached between the hydrous and primary minerals (Nesbitt and Bricker, 1978). Mesh-like textures are generally formed during serpentinization from seawater by the thermodynamically and kinetically controlled hydration processes (Figure 3C, E-H), and indicate olivine weathering (Viti and Mellini, 1998). Olivine hydration results in the formation of brucite and serpentine with minor magnetite (Tupaz et al., 2020). Magnetite formed during serpentinization develops at 200–300°C (Klein et al., 2014; Tupaz et al., 2020). The presence of magnetite associated with serpentine in the studied samples suggests that low-temperature hydration was the primary process that occurred before laterization.

Nickel laterite deposits are mostly developed on accretionary and cratonic terranes (Freyssinet et al., 2005; Butt and Cluzel, 2013). The study area (East Sulawesi ophiolite) is an accretionary terrane. Tectonic uplift plays an important role in the development of Ni-rich laterite profile since it forms faults and joints, which promote physical and chemical weathering along them (Cluzel and Vigier, 2008; Wells et al., 2009; Golightly, 2010; Butt and Cluzel, 2013; Furnes et al., 2014; Villanova-de-Benavent et al., 2014; Tauler et al., 2017). According to field observations, there is a thrust in the vicinity of the study area that uplifts the peridotite. The occurrence of garnierite, both as coatings and veins, along the joints and faults indicates that its precipitation was synchronous with the tectonic uplift. This process of garnierite precipitation is similar to the previous work on the supergene nickel deposit in New Caledonia (Cluzel and Vigier, 2008). Further, serpentine veins that crosscut the pre-existing serpentines and have a higher amount of Ni (Figures 3D,H) are a different type of serpentine that were probably formed after the bedrock was exposed to the surface (Tupaz et al., 2020). The occurrence of the crosscutting serpentine veins indicates that the bedrock was constantly weathered since its uplift. In addition, talc-like garnierite was not observed when the degree of weathering was low because the garnierite-forming solution did not become supersaturated (Tupaz et al., 2020b). This suggests that the Morowali Ni-laterite deposit is part of a good drainage system since it shows the presence of talc-like garnierite. The tropical weather of the study area supplies enough water, with an annual rainfall >1800 mm (Butt and Cluzel, 2013; Fu et al., 2014).

These mineralogical and textural characteristics of secondary alteration minerals, and uplifted tectonic setting indicate that the serpentinization had been occurred before uplift, and the uplift influenced the development of the Morowali Ni-laterite deposit.

During laterization, water moves downward along the joints and fractures causing fluctuations in the water table, and leaches soluble elements from the host rock causing chemical reactions within the joints and fractures. The enrichment of immobile elements such as Fe, Al, Mn, Cr, and Ti in the limonite horizon is the result of weathering since as the water steadily moves downward, it leaves the immobile elements in the upper horizon (Figures 7A–D,I). The main dissolution reactions for element leaching can be explained by the equilibrium between mineral solubility and ion exchange (Golightly, 1981). Olivine dissolves first, followed by pyroxene, serpentine, talc, Ni-bearing minerals (nepouite and kerolite), gibbsite, and goethite (Golightly, 1981; Golightly, 2010). These mineral solubilities are controlled by pH, and increase as pH decreases, except for quartz (Golightly, 2010). Therefore, the degree of weathering can be inferred from the minerals in the horizon. Weathering of the primary minerals cause leaching of soluble alkaline elements (Figures 7E,F), and poorly crystalline Fe oxides, mostly goethite, and smectites are precipitated, which contain the Ni released from olivine and serpentine (Humphris and Thompson, 1978; Freyssinet et al., 2005). This observation is reflected in the relatively high NiO of the limonite horizon compared to the bedrock (Figure 7H). Alkaline elements prefer to remain in solution rather than precipitate as secondary minerals (Aiuppa et al., 2000), while the relatively immobile Fe and Al replace the element present in the dioctahedral site of smectite (Manceau and Calas, 1985). Continued weathering hydrolyzes the smectites, which causes the enrichment of Fe oxides with Al in the upper horizon (Figure 8). Repetition of this process increases porosity, which allows the solution to percolate easily (Freyssinet et al., 2005). This solution might be Ni-rich because during the reprecipitation of Fe oxides in the upper horizon, the Ni contained in goethite is released and incorporated in the garnierite-forming solution.

The garnierite-forming solution percolates downward until it reaches the depth where garnierite can be formed. Formation of garnierite is influenced by the pH; for example, a sudden change in pH from acidic to alkaline oversaturates the garnierite-forming solution causing garnierite precipitation (Golightly, 1979). In this process, Ni replaces Mg in the octahedral site of serpentine or other weathering products of the primary minerals because of their similar ionic radii, leading to Ni enrichment in the saprolite horizon.

The index of lateritization (IOL) was used to verify the degree of laterization (Figure 9). The IOL values divide the samples according to the degree of weathering (weak, moderate, and strong). The range of IOL values for the studied bedrock, saprolite, and limonite were 19.87–20.05, 17.18–22.7, and 96.57, respectively. The similar low IOL values of the bedrock and saprolite imply that the Morowali Ni laterite deposit underwent weak weathering in the saprolite horizon. In the Kolonodale deposit, the weathering of the saprolite horizon is also weakly weathered but more evolved than the Morowali deposit. However, high IOL values for the limonite horizon in the two deposits suggest that it is strongly weathered, leading to the formation of a thick limonite horizon and high Ni concentration in the saprolite horizon (Tupaz et al., 2020). In the Morowali Ni-laterite deposit, the saprolite and limonite horizons are 2–7 m thick, and the boundary between the two horizons is undulatory. In the Kolonodale deposit, the high Ni concentrated talc-like garnierite is observed (Figure 6), but some limonite is weakly weathered. Therefore, Ni concentration in the saprolite horizon and the development of garnierite can differ depending on the location and degree of weathering.

Several generic evolution models have been proposed for laterite deposits (Golightly (1981); Freyssinet et al. (2005); Butt and Cluzel (2013). More specific models have been suggested based on these models. For example, Cathelineau et al. (2016) and Mongelli et al. (2019) insisted on the redistribution model for garnierite-forming elements in New Caledonia and Iran, respectively, while Fu et al. (2018) suggested a preferential flow model, in which preferential flow is the driving force for the Ni-rich solution. The redistribution model is that the pre-existing elements in the host rock are redistributed and formed nickel ore minerals according to changes in groundwater level or wet-dry season. The preferential flow model is that the precipitation of garnierite occurs mainly along preferential flow pathways, which are the pathways for water preferred to flow such as chemical and geological channels. The formation of the Morowali deposit corresponds to the prevalent models combined with the redistribution model.

Since its tectonic uplift in the Late Oligocene (Hall and Wilson, 2000), the serpentinized eastern Sulawesi ophiolite has undergone weathering because of the tropical climate of Indonesia due to large amount of precipitation, and the primary minerals in the bedrock have altered to smectite group minerals. In the process of laterization, immobile elements (Fe, Al, Mn, Cr, and Ti) remained in the upper horizons, and soluble elements (Mg and Si) were leached from the bedrock. The upper horizon has an oxidizing environment that results in the enrichment of redox-sensitive elements and the formation of Fe oxides (Robb, 2005). The Ni in olivine or serpentine is leached during weathering of the bedrock and reprecipitated in the Fe oxides, mostly goethite (Freyssinet et al., 2005).

As the dry and rainy seasons repeat, the water table fluctuates due to the meteoric water percolating into the bedrock that leads to redistribution of Ni preserved in goethite and formation of garnierite. The acidic meteoric water infiltrates the fractures and joints due to continued uplift. In this process, the Ni adsorbed in goethite, which is in the upper horizon, is leached and migrates downward with the solution (Freyssinet et al., 2005). The transported Ni reacts with the weathering products of ferromagnesian minerals by replacing Mg with Ni and finally producing a serpentine-like garnierite (Freyssinet et al., 2005; Butt, 2016). In the deeper horizon, the garnierite-forming solution percolates along the faults and joints and precipitates secondary talc-like garnierite with low FeO content (<0.5 wt% FeO), occurring as coatings between two blocks of serpentinized bedrock. According to Cathelineau et al. (2016), the talc-like garnierite-forming solution reaches the surface via vertical veins that were synchronous with faulting activity and joints, which contained early formed Ni-rich minerals, thus having high Ni content. However, the talc-like garnierite in this study had a low Ni content, suggesting that the talc-like garnierite-forming solution was a low-Ni-bearing fluid that originated in the saprolite horizon and precipitated garnierite from the outside of the horizontal vein. The garnierite and quartz mineralization occur in the dry season (Cathelineau et al., 2016; Mongelli et al., 2019).

However, the Ni-enriched vertical vein was not observed during the field investigation, which might be due to the frequent erosion of the laterite deposit during the rainy season. Cathelineau et al. (2016) found large Ni-bearing veins in their study area (New Caledonia) during open-pit mining, and they showed that changes in garnierite mineralization occurred within a few months. This means that large Ni-bearing veins can emerge from the Morowali deposit by continuing open-pit mining.