Accepted: April 01, 2023 | Published Online: April 03, 2023

Citation: Ugwu IM, Egirani DE (2023) Formation of Oxide and Silicate Nickel Laterites using Reactive Transport Simulations. Int J Earth Sci Geophys 9:067.

Copyright: © 2023 Ugwu IM, et al. This This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Abstract

Geochemical modelling of ore deposit gives preliminary information on the supergene enrichment of the metallic ore in the subsurface. With recent advances in analytical technology, thermodynamic data necessary for modelling rock-mineral weathering and water-mineral interaction have become available, thus enabling the prediction of geochemical processes. The PHREEQC code enables the simulation of weathering, secondary mineral formation, and metal transport using thermodynamic models for mineral dissolution and metal sorption. In Ni laterite deposits, goethite is the host mineral in the oxide type whereas both goethite and garnierite are the host minerals in the silicate type. In both types of Ni laterite, nickel and cobalt are associated with goethite via surface precipitates or surface complexes or by structural incorporation. This study presents a reactive transport simulation for the formation of both silicate and oxic Ni laterite weathering at ambient conditions using PHREEQC with a supplemental thermodynamic model. The PHREEQC simulation of Falcondo Ni laterite (silicate) revealed that the increasing trends of evolution of minerals are goethite, antigorite, sepiolite, falcondoite and enstatite. Similarly goethite, antigorite, enstatite and talc were reproduced for the Koniambo Ni laterites (oxic). For both laterite types, the pH at the limonite zone varies from 9.1 to 8.9; whereas the pH at the saprolite zone is 7.8 (for silicate laterite) and 8.3 (for oxic laterite). Starting from ultramafic rock consisting of forsterite with a mole fraction of 0.827, fayalite (0.1667) and Ni₂SiO₄ (0.006) co-existing as an ideal solid solution, the simulation of the oxic laterite show that 0.28 wt.% Co and 1.18 wt.% of Ni were sorbed on goethite. In contrast, 0.25 wt.% Co and 1.25 wt.% Ni were adsorbed on goethite for silicate Ni laterite simulation. The simulations produce a weathering profile, mineral assemblages and ore grade similar to that observed in natural Ni laterite deposits.

Keywords

Ni Laterite, PHREEQC, Nickel, Goethite, Garnierite

Introduction

The majority of Ni and Co used in manufacturing both super alloys and petrochemicals and in the aviation industry are obtained from Ni laterite deposits [1]. Consequently, the evolution, weathering profile, mineralogy and mode of occurrences of the alteration minerals present in both oxide and silicate types of Ni laterite have been studied by many researchers [2-8]. Goethite is the host mineral in the oxide type whereas both goethite and garnierite are the host minerals in the silicate type [5,6].

Reactive transport simulations, based on thermodynamics and mass transfer, can be used to investigate weathering processes, regolith formation, weathering profile thickness and metal leaching [9-21]. The PHREEQC (pH-REdox-Equilibrium Code) enables us to simulate weathering, secondary mineral formation, and metal transport using thermodynamic models for mineral dissolution and metal sorption [22]. However, previous attempts to simulate the formation of Ni laterite weathering using PHREEQC were unsuccessful, resulting in an unrealistic pH in the limonite zone containing goethite and variations between the hypothetical and actual layer thickness [17]. In particular, the pH above 7 required for total adsorption of Ni onto goethite was not attained in the previous published models [17,18]. Recent modelling of ultrabasic accounted for Ni adsorption on goethite as well as existence of Ni in the goethite structure and its occurrence as surface complex and precipitates [18]. However, the modelling was done using thermodynamic data obtained for ferrihydrite rather than goethite and unable to explain the Ni enrichment in present day laterites as well as unrealistic high amount of nickel. In addition, no model has been able to account for reproduction of cobalt adsorbed or incorporated in goethite present in Ni laterites Crucial to success of a reactive transport simulation is the thermodynamic (and potentially kinetic) model is it based on.

In the present work, a reactive transport modelling inclusive PHREEQC and the basic Stern model for electrostatics have been used to develop the secondary nickel ores profile with vertical progression of the alteration front. Also, the resulting thermodynamic and PHREEQC models have been used for the advective transport simulation of oxide and silicate laterites as applied for the sorption of Co and Ni by goethite [23,24]. The thermodynamic models used with PHREEQC with equilibrium constants have been used to simulate the precipitation of relevant secondary phases from [25]. The model inputs include those for garnierite (silicate) Ni laterite and limonite (oxide) Ni laterite. The outcome of the modelling provides an in-depth empathy of the weathering mechanism in the Nickel laterites. The validity of the model provides new insights into further processes that control the mobility of elements and formation of these Nickel laterites.

Method

PHREEQC Geochemical modelling of Ni Laterite Formation

The thermodynamic calculations involved the modelling of the Ni laterite formation from weathering of ultramafic rocks were carried out using the geochemical code PHREEQC. This program is well known for its capability for calculating complex sorption reactions between aqueous species and mineral assemblages both in 1D advection-diffusion-reaction and transport mode. Goethite associates with Ni and Co through surface complexation, surface precipitation and structural incorporation; however, existing thermodynamic models (e.g., llnl.dat) available with PHREEQC do have equilibrium constants for surface complexation or structural incorporation. Consequently, we supplemented the existing llnl.dat thermodynamic model with a surface complexation model derived from fitting to laboratory sorption experiments for Ni sorption up to 0.42 wt.% and 0.2 wt.% Co [23,24]. A two-site model was used to describe the surface complexes; the goethite surface area was assumed to be 50.4 m2g-1. The surface electrostatics was evaluated using a basic Stern layer model with a capacitance of the Stern layer set to 1.05 F/m2 with all charges at the 0-plane [26]. The surface complexation model is summarised in Table 1.

A number of important secondary phases associated with silicate laterites are not included in the distributed llnl.dat model (“database”). The chemical reactions, equilibrium constants and enthalpy of formation for dominant mineral phases in silicate Ni laterite were obtained from literature [25]. These are reproduced in Table 1. Thus, the llnl.dat thermodynamic model was modified to include thermodynamic parameters for népouite, falcondoite, kerolite and pimelite phases given by [25]. The resulting thermodynamic model was to simulate the formation of oxide Ni laterites (e.g., Koniambo, New Caleodonia, Figure 1) [5] and silicate Ni laterites (e.g., Falcondo, Dominican Republic, Figure 2) [3,4].

Model Input: Garnierite (Silicate) Ni Laterite

The parent rocks for the Falcondo Ni laterite are serpentinised peridotite, predominantly harzburgite with minor quantities of lherzolite, dunite and serpentinite [3,7]. The olivine in these rocks contains Fosterite 89 to 92 wt.% and 0.2 to 0.4 wt.% Ni [27,28]. These rocks weather to form Ni laterite with six distinct layers from top to base, namely: upper limonite, lower limonite, soft serpentine, hard serpentine, serpentinised peridotite and unaltered ultramafic protolith. The thickness of the layers are 0.8 m, 10 m, 13 m, 17 m, 19 m and 23 m from top to base. The secondary minerals identified are Ni-sepiolite-falcondoite, Ni serpentine, goethite, kerolite, pimelite, and lizardite-nepouite [3,7].

Our starting primary mineral assemblage consisted of olivine comprised of forsterite with a mole fraction of 0.827, fayalite (0.1667), Ni₂SiO₄ (0.005) and Co₂SiO₄ (0.001) co-existing as an ideal solid solution. This mole fraction was chosen to represent the composition of serpentinised peridotite described by Lewis, et al. [3]. The silicate laterite simulation was allowed to react with rainwater with pH 5.7 and equilibrate with antigorite, sepiolite, goethite, Ni₂SiO₄, Co₂SiO₄, enstatite, ferrosilite, nepouite, pimelite, falcondoite, kerolite and atmospheric CO₂. The simulation is made up of layers of 6 cells of variable thickness. The detailed input file for this simulation and the results obtained in molality and wt.% are shown in Appendix.

The oxic laterite profile consists of ten layers with different proportions of iron oxides and secondary silicates [5]. The layers are primarily rock containing forsterite with < 0.4 wt.% Ni, enstatite and serpentine. These units have variable thicknesses and an overall thickness of 64m starting from the top unit and finishing at the bedrock. The boundary between limonite and saprolite zones occurs at 34m depth. The minerals present in the profile are goethite, asbolane, serpentine, lithiophorite, forsterite, hematite, talc, quartz, enstatite, chromite and birnessite. Except at the bedrock, goethite occurs throughout the profile [5].

The simulation started with a primary mineral assemblage consisting of olivine comprised of forsterite with a mole fraction of 0.827, fayalite (0.1667), Ni₂SiO₄ (0.006) and Co₂SiO₄ (0.001) co-existing as an ideal solid solution. The simulation was allowed to react with rain water with pH 5.7 and equilibrate with antigorite, sepiolite, goethite, talc, enstatite, birnessite and atmospheric CO₂. The simulation is made up of layers of 10 cells of variable thickness and the no. of shifts used is 120. Therefore, the total number of years for the simulation is 1200 years (no. of cells x shifts). The input file is similar to that of the silicate Ni laterite expect that minerals that are formed on oxic Ni laterite are allowed to precipitate. The input parameters are illustrated in Figure 3 whereas the input values for the constants are shown in Table 1.

Ordinarily, mineral assemblage of phyllosilicates and Ni are components of solid solution series extending throughout the assemblage- Ni end-members [30]. Their effect is presented as an ideal solid solution instead of separate minerals. Some ground waters derived from the base of nickel laterite profiles are soaked with respect to talc-like mineral but not with serpentine [31]. Serpentine is seldomly form under ambient conditions and curbed from the phases of equilibrium allowed for precipitation. In real-life the rates of precipitation support the development of metastable sepiolite and kerolite in Al-poor environments [32]. The inclusion of nickel as a starter to the assemblage system enhances the stability of the Nepouite phase, thus preventing its development to Fayalite [33]. The surface complexation model of [34], takes in consideration the attachment of metals and protonation on strong and weak reactive sites of a mineral phase (Table 1b). The binding of these metals and subsequent protonation generates a charge on the ions sorbed and has been used to simulate the adsorption process. A charge depending on the ions sorbed, has been used in order to simulate adsorption process.

Results

Reactive transport simulation

The reactive transport simulations replicate the secondary mineralogy of natural silicate (Figure 1) and oxide laterites (Figure 2). The simulation of silicate Ni laterite revealed that the increasing trends of evolution of minerals are goethite, antigorite, sepiolite, falcondoite and enstatite. This is consistent with observation by [3,7,25]. Similarly, in order of increasing succession, the mineral reproduced for oxide laterites are goethite, antigorite, enstatite, talc and Ni₂SiO₄ which is consistent with that observed by [5].

Even though minerals such as pimelite, kerolite and nepouite were allowed to equilibrate, they did not precipitate out of solution. The formation of some secondary mineral phases may be thermodynamically favoured but limited by kinetics of precipitation e.g., birnessite, asbolane, serpentine. However, in the absence of kinetic models for these mineral, the only approach to is to prevent them from forming during the simulation. If the formation of antigorite is suppressed, then kerolite will form instead but prevent the formation of sepiolite and falcondoite with increase in the pH of the saprolite zone from 8.4 to 11.3. In the simulation reported here, however, no phases were suppressed and the resulting mineralogy is in agreement with field observations. For both laterite types, the pH (Figure 4) at the limonite zone varies from 9.1 to 8.9 whereas the pH at the silicate-saprolite zone is 7.8 and the oxide-saprolite zone is 8.3. Also, the pH-depth diagram for silicate-Falcondo Ni laterite in the limonite zone range from pH = 7-9 at a depth of 0-1.5 m and saprolite zone is from pH = 7-8 at a depth of 1.5- 4 m (Figure 4a and Figure 4b).

This is in agreement with the pH 6.2 to 9.2 in the ferricrete and coarse saprolite zones of natural Ni laterite [35]. Similarly, at Cerro Matoso Ni laterite deposit, pH 8.1 was observed at the spring located at the base of the weathering profile [36]. At pH above 8, garnierite minerals are slightly soluble [7] while all the Ni released sorbs to goethite at pH > 7.

Starting from ultramafic rock with forsterite with a mole fraction of 0.827, fayalite (0.1667) and Ni₂SiO₄ (0.006) co-existing as an ideal solid solution, the simulation of oxic Ni laterite containing both Ni and Co, yielded 0.0001% Co adsorbed at the weak but abundant site (> (FeOH)₂), as well as 0.28 wt.% Co at the strong but unabundant site (> Fe(OH)₂) of goethite. Similarly, 1.18 wt.% of Ni was adsorbed at the > (FeOH)₂ site while 0.003 wt.% Ni was adsorbed at the > Fe(OH)₂ site. In contrast, for the silicate Ni laterite containing both Ni and Co, 2.9 × 10^-5% Co was adsorbed on (> (FeOH)₂) site of goethite while 0.25 wt.% Co was adsorbed at the > Fe(OH)₂ site. Similarly, 1.25 wt.% Ni was adsorbed at the (> (FeOH)₂) site while 0.009 wt.% Ni was adsorbed at the > Fe(OH)₂ site of goethite. This is consistent with the findings of [5] for transition laterite (0 to 34 m) containing 0.8 to 1.5 wt% Ni and 0.1 to 0.3 wt.% Co although they found 2.44 wt.% Ni and 0.09 wt.% Co at depth of 46.5 m.

Weathering process and the formation of Falcondo Ni laterite, Dominican Republic

The simulation provides a spatial distribution of mineral phases with chemical species present after sorption of Ni and Co on goethite (Figure 5a, Figure 5b and Figure 5c).

Also, secondary mineral phases (i.e. Fe(OH)₂Ni), (FeOH)₂Ni, Fe(OH)₂Co, (FeOH)Co, antigorite, goethite, sepiolite and falcondite. The alteration mineral succession depicts that of natural silicate Ni laterite deposits.

The plot of elements versus depth for the silicate Ni laterite (Figure 6) indicate low amount of Ni, Mg, Fe and Si at the limonite zone as well as high Si, Ni and Mg at the saprolite zone. This trend closely matches that of natural Ni laterite. Magnesium, iron, nickel and silica are main components of the garnierite minerals.

Weathering process and formation of Koniambo Ni laterite, New Caledonia

The simulation provides a spatial distribution of mineral phases with chemical species present after Ni and Co sorption on goethite (Figure 7a, Figure 7b and Figure 7c). The secondary minerals generated include Fe(OH)₂Ni, (FeOH)₂Ni, Fe(OH)₂Co, (FeOH)Co, antigorite, goethite, sepiolite, talc, enstatite and Ni₂SiO₄. Herein, the alteration mineral succession depicts that of natural oxic Ni laterite deposits.

The concentration of key elements for different zones of oxide Ni laterite (Figure 8) shows that the amount of Ni and Fe at the limonite zone is directly proportional but their amount decreases at the saprorite zone. This indicates close association between Fe and Ni in goethite in the limonite zone. Silica is zero at the limonite zone but increases rapidly in the saprolite zone. The concentration of Ni varies from 0.003 to 1.18 wt.% from saprolite to limonite zone.

Discussion

These results indicate the presence of a wide-ranging solid solution between Mg and Ni end members. In general, Ni-Mg hydrous silicates from Falcondo and Koniambo Ni lateritehave similar chemical characteristics to garnierite minerals examined in other worldwide Ni-lateritic deposits, excluding the Ni-dominated serpentine-like phases, which have not investigated. It is suggested that Ni mobilization is initiated by fractures which enhances the circulation of water circulation. This process promotes weathering and preferential Ni concentration. There is evidence of garnierite precipitation in Falcondo Ni-laterite. Most of garnierite mineralization in the Falcondo mine compose of Ni-sepiolite and falcondoite solid solutions with different concentration [4] which consistent with the simulated Ni laterite (Figure 5). This precipitation is associated with deformation structure that affected the weathering profile [37,38]. The occurrence of these deposits as in fillings and coatings on fracture surfaces, thus precipitating in an energetic dynamic environment. The ultramafic rock suffered uplift before it as altered to several mineral assemblages. This characteristic is indicative of ceaseless dynamic uplift compared to weathering rates. A dynamic environment supported the development of hydrous silicate Ni-laterites, thus promoting cyclic, repeated processes of weathering and denudation. Subsequently, fluids and other reworking migrated at variable depth, remobilized and reconcentrated to form high concentration of Ni ore minerals.

The simulation model of Koniambo Ni laterite indicates the precipitation of silicates at local equilibrium, thus promoting simultaneous precipitation of thermodynamically favorable minerals. In the end, this pattern induces a narrow zone of Ni precipitation. Naturally, nickel, released due to Ni-Mg hydrous silicates dissolution, is partially lost by crosswise movement in solution. From Figure 4 the migration of transition zone from the limonite horizon to the saprolite, which is marked at 1.5 m and 38 m. The spatial distribution of the mineral phases is consistent with other reports [5,39]. Modelling results indicates that adsorbed nickel is governed by pH front, released into pore water at low pH as since silicates dissolution. It is difficult to explain high nickel content in mineral phases by adsorption alone. In nature, the pH of soil waters is lowered by the presence of dissolved organic acids, thus, it is suggested that nickel concentration is supported by stronger bonding. The stronger bonding is initiated by isomorphous substitution of Ni for Fe in Fe-mineral bearing phases. Surface complexation is useful in understanding metal sorption characteristics at a mineral surface. The simplified version of the complexation model considers a closed system of interaction for Ni and a fixed mineral phase. However, the interaction between Ni-Co and a mineral phase indicates competitive sorption of metals at reactive sites. Subsequent on mineral phase development, adsorption kicks in to regulate the migration of single metal and competitive metals, thus, impeding their migration). Previously, ferrihydrite is considered an initial phase precursor phase in the development of goethite. However, this phase is difficult to identify using standard techniques because of its poor crystallinity. However, the modelling was done using thermodynamic data obtained for ferrihydrite rather than goethite and unable to explain the Ni enrichment in present day laterites.

In summary, modelling the reactive transport simulation of oxide and silicate -type Ni laterite profilesis not limited to pH and depth of Ni laterite but the enormous influence of other cations such as Mg that is released during the alteration of ultramafic rocks. The dissolution process and subsequent competitive sorption are controlled by pH and depth of the profile.

Conclusion

The formation of Ni laterite profile from ultramafic parent rock, at 25 ℃, has been simulated by means of PHREEQC one-dimensional reaction-transport code. A reactive transport simulation for the formation of Koniambo Ni laterite, New Caledonia (oxic) and Falcondo Ni (silicate) laterite by weathering of the parent ultramafic rocks at ambient conditions produces a secondary mineral assemblage similar to that observed in the field. The main conclusions are listed below:

1. The descending progression of the pH front controls the mobility of the elements and dictates the Ni enrichment.

2. The precipitation of talc-like silicates at the cost of sepiolite is favoured by high pH due to the dissolution of olivine.

3. The alteration of mineral phases agrees with that of natural Ni laterite deposits.

4. Moreover, the sorption of Co and Ni sorption to goethite involves both surface complexation and irreversible non-ideal structural incorporation.

Hence, application of our surface complexation model to the simulated formation of Co, Ni oxide laterites is only a first-approximation and further refinements of thermodynamic models for the sorption of Co and Ni to goethite are needed.

Acknowledgement

Thanks to Prof Richard Herrington, Natural History Museum, London, Dr Heather Buss and Prof. David Sherman, University of Bristol, United Kingdom for their useful suggestions. This work was supported by the Tertiary Education Trust Fund (TET Fund) of Nigeria.

---