Fe–Ni-bearing serpentines from the saprolite horizon of Caribbean Ni-laterite deposits: new insights from thermodynamic calculations

Fe–Ni-bearing serpentine from the saprolite horizon is the main Ni ores in hydrous silicate-type Ni laterites and formed by chemical weathering of partially serpentinized ultramafic rocks under tropical conditions. During lateritization, Mg, Si, and Ni are leached from the surface and transported downwards. Fe2+ is oxidized to Fe3+ and fixed as insoluble Fe-oxyhydroxides (mostly goethite) that incorporate Ni. This Ni is later leached from goethite and incorporated in secondary serpentine and garnierite. As a result, a serpentine-dominated saprolite horizon forms over the ultramafic protolith, overlapped by a Fe-oxyhydroxide-dominated limonite horizon. The serpentine from the protolith (serpentine I) is of hydrothermal origin and yields similar Ni (0.10–0.62 wt.% NiO) and lower Fe (mostly 1.37–5.81 wt.% FeO) concentrations than the primary olivine. In contrast, Fe–Ni-bearing serpentine from the saprolite (serpentine II) shows significantly higher and variable Fe and Ni contents, typically ranging from 2.23 to 15.59 wt.% Fe2O3 and from 1.30 to 7.67 wt.% NiO, suggesting that serpentine get enriched in Fe and Ni under supergene conditions. This study presents detailed mineralogical, textural, and chemical data on this serpentine II, as well as new insights by thermodynamic calculations assuming ideal solution between Fe-, Ni- and Mg-pure serpentines. The aim is to assess if at atmospheric pressure and temperature Fe–Ni-bearing serpentine can be formed by precipitation. Results indicate that the formation of serpentine II under atmospheric pressure and temperature is thermodynamically supported, and pH, Eh, and the equilibrium constant of the reaction are the parameters that affect the results more significantly.


AQ3
Fe-Ni-serpentine represents one of the main sources of Ni in many Ni-laterite deposits worldwide (Freyssinet et al. 2005 ). Ni-laterite deposits account for about 40 % of the world's annual production of Ni, they host over 60 % of the world land-based Ni resources (Gleeson et al. 2003 ;Kuck 2013 ) and the amount of Ni being mined from laterite ores is increasing steadily (Mudd 2010 ).
Hydrous Mg silicate Ni-laterite deposits are characterized by a thick serpentine-dominated saprolite horizon covered by a Fe-oxyhydroxidedominated limonite horizon.
The combination of tropical climate with intense rainfall, low water table and continuous tectonic uplift produces physical and chemical weathering that leads to the dissolution of the primary ferromagnesian minerals of the ultramafic protolith and the formation of Fe-oxyhydroxides (Roqué-Rosell et al. 2010) and secondary Ni-bearing Mg phyllosilicates (Villanovade-Benavent et al. 2014 ;Cathelineau et al. 2015 ).

AQ4
Alteration begins in crystal joints and cracks in peridotite that provides access for water (Pelletier 1996 ). Among peridotite forming minerals, olivine is considered the most unstable mineral and the first to weather following the hydrolysis reaction 1 (modified from Freyssinet et al. 2005 ): Mg is ultimately leached out of the profile, whereas Fe is oxidized to Fe and subsequently concentrated as Fe-oxyhydroxides (mainly goethite) in the limonite horizon (Golightly 2010 ).
As olivine represents also the main Ni-bearing mineral in the protolith, Ni gets released during weathering but is retained in the profile by absorption onto goethite. As weathering continues, goethite may be eventually dissolved and reprecipitated, leaching some Ni following reaction 2 (Freyssinet et al. 2005 ): Ni released by goethite can be incorporated in primary (hydrothermal) serpentine (hereafter serpentine I) from the protolith, formed prior to weathering and lateritization of the peridotite body, thus leading to the formation of a secondary Fe-Ni-enriched serpentine (serpentine II) following reaction 3 (Freyssinet et al. 2005 ) or the precipitation of garnierites in open fractures (Pelletier 1996 ).
Whereas recent studies have focused on Ni-rich garnierite mineralization (Tauler et al. 2009 ;Wells et al. 2009 ;Villanova-de-Benavent et al. 2014 ;Cathelineau et al. 2015 ), only three studies investigated compositional variations of these serpentines II of the saprolite horizon in detail (Golightly and Arancibia 1979 ;Pelletier 1996 ;Chen et al. 2004 ). Golightly and Arancibia ( 1979 ) studied several samples of unweathered and weathered serpentines from Indonesia, where 6 % of the world's Ni reserves are located (Kuck 2015). They observed, in the Ni-laterite profile of Soroako, that weathered serpentine (serpentine II in this contribution) had a low Mg content and increased Fe (4-8 wt.% Fe O ) and Ni (1-8 wt.% NiO) contents relative to the unweathered serpentine (here serpentine I). Pelletier ( 1996 ) performed a study on hydrous silicate Ni-laterite deposits of New Caledonia, which represent 15 % of the world's Ni reserves (Kuck 2015). Pelletier ( 1996 ) observed that secondary nickeliferous serpentine (serpentine II), associated with garnierite and resulting from weathering of serpentine I, was enriched in Ni (1-7 wt.% NiO) and also in Fe (7-17 wt.% Fe O ).
The Caribbean region hosts 10 % of the world's Ni resources, mostly in the northern part, which includes the Moa Bay and Punta Gorda deposits in eastern Cuba and the Falcondo deposit in central Dominican Republic (Dalvi et al. 2004 ;Lewis et al. 2006 ;Nelson et al. 2011 ;Aiglsperger et al. 2016 ). Preliminary mineralogical, textural and chemical studies on serpentine from the saprolite horizon of Caribbean Ni-laterites revealed two different serpentine generations Proenza et al. 2007 ;Gallardo et al. 2010 ). Serpentine I, which surrounds olivine grains in a mesh texture arrangement, has low Ni and Fe contents, comparable with those of the primary olivine (mostly less than 0.5 wt.% NiO). In contrast, the serpentine II that replaced serpentine I is enriched in Ni (up to 3 wt.% NiO) with respect to serpentine I. In addition, serpentine II yielded higher Fe contents than serpentine I  ).
The replacement of Mg by Ni or Fe in serpentine would stabilize the silica excess in serpentine caused by the incongruent dissolution of serpentine that enhances the leaching of Mg (Golightly 1981 ). Freyssinet et al. ( 2005 ) suggested that Ni exchanges for Mg in octahedral sites of serpentine II without solution, recrystallisation, or neoformation.
Given the economic importance of Ni, these studies mostly focus on the incorporation of this element in serpentine in the lateritic environment. In contrast, the presence of Fe in serpentine has recently been studied as a key factor during serpentinization (Klein et al. 2009 ) but has not been studied in lateritic environments. In addition to these authors, Streit et al. ( 2012 ) reported that serpentine may contain significant amounts of Fe . The incorporation of ferric iron in serpentine takes place in the form of a ferri-Tschermak substitution, where Fe substitutes for Mg (in octahedral sites) and for Si (in tetrahedral sites) to form the cronstedtite component of serpentine (Fe Fe )(SiFe )O (OH) ). It also may substitute in the octahedral site, that would be charge balanced by some vacancies, forming a ferrian serpentine with the formula (□Fe )Si O (OH) , the Fe-analog of kaolinite (Wicks and Plant 1979 ;Evans 2008;Evans et al. 2009 ).

AQ7
Whereas Golightly and Arancibia ( 1979 ) and Golightly ( 1981 )  Thermodynamic modeling may give further insight into the Fe enrichment in a Ni-laterite environment. Therefore, the aim of the present study is to shed light on the process of Fe and Ni enrichment in Fe-Ni-serpentine of Ni-laterite deposits by combining the knowledge on compositional variations of serpentine II and detailed mineralogical analyses, with thermodynamic and geochemical calculations concerning serpentine II stability. This is used to discuss if serpentine II might form under atmospheric pressure and temperature conditions through dissolution/precipitation processes.

Geological setting
This study considers two hydrous Mg silicate Ni-laterite deposits located in the Caribbean region: Yamanigüey (Eastern Cuba) and Falcondo (Central Dominican Republic). The Ni-laterite deposits in the Caribbean region developed from ophiolite-related peridotites, which were emplaced during the Cretaceous, with lateritization and subsequent ore deposit formation commencing in the Miocene (Lewis et al. 2006 ). The Yamanigüey Ni-laterite deposit, located in Moa Bay mining area, eastern Cuba, developed on serpentinized harzburgite and dunite from the Mayarí-Baracoa ophiolite belt (Proenza et al. 1999;Marchesi et al. 2006). The ultramafic protolith consists of olivine (Fo ) with 0.16-0.39 wt.% NiO, orthopyroxene (En with NiO below 0.1 wt.%), and clinopyroxene (Mg from 78 to 92, Wo , Al O from 2.2-4.2 wt.%; Proenza et al. 2007 ). The thickness of the profile in Yamanigüey is about 12 m. The main mineral phases in the saprolite horizon are serpentines I and II, with minor magnetite and goethite  ). According to Normando ( 2006 ), the Ni grade in the saprolite horizon can reach values of 1.76 wt.% Ni.

AQ8
The Falcondo Ni-laterite mining area comprises different ore deposits, and samples were taken from the Loma Ortega and Loma Caribe deposits. These deposits developed on the serpentinized Loma Caribe peridotite belt, which was interpreted by Lewis and Jiménez ( 1991 )  by orthopyroxene (En ) with NiO below 0.1 wt.%, and minor clinopyroxene (Haldemann et al. 1979 ;Lewis et al. 2006 ;Lithgow 1993 ;Marchesi et al. 2012 ). The thickness of the Falcondo Ni-laterite profile varies from 1 to 60 m (Haldemann et al. 1979 ). The major Ni-bearing mineral phases are located in the lower saprolite horizon, and comprise serpentine II and garnierites (Tauler et al. 2009 ;Villanova-de-Benavent et al. 2014

Materials and methods
For this study, one sample from the saprolite horizon of the Ni-laterite deposit of Yamanigüey, two of Loma Caribe and one of Loma Ortega were selected. These four samples were characterized mineralogical and chemically by means of powder X-ray diffraction (XRD), optical and scanning electron microscopy (SEM-EDS), and electron microanalysis (EPMA) in the Centers Científics i Tecnològics of the Universitat de Barcelona (CCiT-UB). The data were used to perform geochemical calculations in order to determine the stability field of serpentine II and to study whether this Ni and Fe enrichment can take place in supergene conditions. The samples were powdered with an agate mortar and pestle. The instrument used is a PANalytical X'Pert PRO MPD Alpha1 powder diffractometer in Bragg-Brentano θ/2 θ geometry, using Cu kα1 radiation (λ = 1.5406 Å) and The structural formulae of serpentine were calculated on the basis of seven oxygens. In the case of serpentine II, mass and charge balance was only achieved if all iron was disposed as Fe in the octahedral layer. This is in agreement with Roqué-Rosell et al. (2016), who determined that most of iron in serpentine II is in the ferric form, by Fe K-edge XANES on samples from the saprolite horizon of the Falcondo Ni-laterite. This observation is coherent with Golightly and Arancibia ( 1979 ) who considered Fe to be in its oxidized form in serpentine II from Indonesia, and with Streit et al. ( 2012 ), who concluded that only the substitution of Fe in octahedral sites in serpentine from Oman could explain the compositions they observed.
Thermodynamic calculations were carried out with the MEDUSA software package (Puigdomènech 2010 ), based on chemical equilibrium, and PHREEQC v.3 (Parkhurst and Appelo 2012 ). Thermodynamic data from Thermochimie v.9 database (Giffaut et al. 2014 ) were used in the calculations. Its domain of application is mainly within a pH range of 6 to 13, Eh of −0.5 to +0.5 V and temperatures below 80 °C. Si, Ni, Mg, and Fe aqueous speciation has been calculated considering the aqueous species and corresponding logK values.

Results and discussion
Mineralogy and chemistry of Fe-Ni-serpentine in the saprolite horizon  Observations under the optical microscope show that serpentine forms yellowish rims around brown, rounded, silicified Fe-oxyhydroxide cores, and fills micron-thick veinlets (Figs. 3 a-d). The examination under SEM confirms that serpentine is Fe-Ni rich and therefore can be labeled as serpentine II. These textures are in agreement with the formation model described by Trescases ( 1973Trescases ( , 1979 and Streit et al. ( 2012 ). Trescases ( 1973 ), in New Caledonia, observed that serpentine in unweathered, serpentinized peridotite (serpentine I) was colorless, whereas serpentine in the saprolite (serpentine II) was strongly colored by Fe hydroxides, despite showing the mesh texture intact. Furthermore, Trescases ( 1979 ), on saprolite fragments from various Ni-laterite localities, found that olivine, surrounded by serpentine in a mesh arrangement, was replaced by Fe-oxyhydroxides, and the serpentine mesh remained intact after olivine weathering. Streit et al. ( 2012 ), when studying carbonatized, serpentinized peridotites from Oman, described that Fe-rich secondary serpentine, formed from a primary Fe-poor serpentine, develops mesh textures around quartz and Fe-oxyhydroxide cores, which were formerly olivine fragments. The difference between Streit et al. ( 2012 ) and this study is that the serpentinite from Oman did not undergo weathering under lateritic conditions and therefore did not experiment Ni enrichment from an upper horizon. The secondary serpentine is Fe rich but with low Ni and comes from the alteration of a Fe-poor serpentine precursor.

Fig. 3
Optical and scanning electron micrographs of typical yellowish serpentine II mesh texture surrounding brownish Fe-oxyhydroxide and quartz rounded aggregates of the saprolite horizon: a-b plane polarized light (a) and crossed polarized (b) image of serpentine II from Loma Ortega; c-d plane polarized light (c) and detail backscattered electron image (d) of serpentine II (Srp II) and silicified goethite (Gth) from Loma Caribe; e plane polarized light (left) and crossed polarized (right) images of primary pyroxene (Px) and olivine (Ol) relicts crosscut and surrounded by yellowish serpentine II in the saprolite sample from Yamanigüey; f backscattered electron image depicting the high variation of Fe and Ni contents in serpentine II from Loma Ortega No traces of serpentine I were found in any of the four studied samples. However, partially altered pyroxene and olivine relicts are in the Yamanigüey sample ( Fig. 3 e), surrounded and crosscut by yellowish serpentine II (formerly serpentine I), along fractures and cleavage planes.
As seen in Figs. 3 f and 4 and Si O (OH) . Thirty-nine EPMA serpentine II analyses were carried out in the two Loma Caribe samples. Serpentine II has 3.84-13.51 wt.% Fe O (0.14-0.50 apfu Fe ) and 1.75-7.00 wt.% NiO (0.07-0.28 apfu Ni), with an average structural formula of Mg Fe Ni Si O (OH) . These Ni and Fe contents are higher than those of serpentine I from samples of less weathered saprolite rocks collected at greater depths in the Ni-laterite profiles of Yamanigüey, Loma Ortega and Loma Caribe (Fig. 4 ). The composition of serpentine I from Yamanigüey ranges from 3.55 to 5.81 wt.% FeO (0.14-0.24 apfu Fe ) and from 0.10 to 0.34 wt.% NiO (below 0.01 apfu Ni), and the average structural formula from 10 analyses is Mg Fe Ni Si O (OH) . Serpentine I from Loma Ortega has 1.37-1.87 wt.% FeO (0.06-0.08 apfu Fe ) and 0.15-0.53 wt.% NiO (0.01-0.02 apfu Ni), and the average structural formula from 12 analyses is Mg Fe Ni Si O (OH) . In addition, one analysis of serpentine I from Loma Caribe was obtained, with 2.91 wt.% FeO (0.11 apfu Fe ) and 0.62 wt.% NiO (0.02 apfu Ni) and the corresponding structural formula Mg Fe Ni Si O (OH) .

Fig. 4
Comparison between the major octahedral elements (Ni-Fe (a); Ni-Mg (b); Fe-Mg (c); Ni + Fe-Mg (d)) in serpentine II of the studied samples from Yamanigüey, Loma Ortega, and Loma Caribe, compared with serpentine I from other samples of the same occurrences, and the data of serpentines I and II from Indonesia (Golightly and Arancibia 1979 ;Chen et al. 2004 ), and from New Caledonia (Pelletier 1996 ;Chen et al. 2004 ). In (a), the curves indicate the general trend of the analyses obtained in serpentines II from Yamanigüey (red), Loma Ortega (yellow), and Loma Caribe (blue), and the corresponding arrows mark the Ni-Fe-enrichment paths between serpentines I and II in each deposit  The high Fe contents in serpentine could be explained by contamination of the analyses by micro-or nano-magnetite inclusions. Magnetite usually forms during serpentinization of ultramafic rocks: olivine in the presence of water is transformed into serpentine and ferroan brucite, which in turn can react with aqueous silica to become serpentine and magnetite (Sleep et al. 2004 ).
However, magnetite was not detected by powder XRD or when examining the samples under SEM and TEM. In addition, serpentines are not stoichiometric and displayed vacancies in the octahedral layer (possibly as a consequence of Mg -Fe substitution). This non-stoichiometry was also observed by Goligthly and Arancibia ( 1979 ) and Orberger et al. ( 1990 ). Therefore, Fe is considered to be contained in the octahedral layer of serpentine. The relationship between the octahedral cations (Mg, Ni, and Fe) is plotted in Fig.  4 , which includes data from serpentine I and II from Yamanigüey, Loma Ortega and Loma Caribe and data from Golightly and Arancibia ( 1979 ), Pelletier ( 1996 ), and Chen et al. ( 2004 ). According to Fig. 4 a, Ni and Fe in serpentine II from the Caribbean are positively correlated. This correlation indicates that serpentine II is enriched in Ni and Fe with respect to serpentine I, and therefore the incorporation of Fe and Ni can be linked to weathering. Furthermore, the slope is steeper in Loma Caribe than in Loma Ortega and Yamanigüey. This indicates that Ni increases more rapidly than Fe in serpentine II from Loma Caribe than in the other two occurrences. This was also observed by Golightly and Arancibia ( 1979 ) (Pelletier 1996 ).

Figure 4 b-d displays a negative correlation between Ni and Mg, between Fe
and Mg, and between Ni + Fe and Mg, respectively; this correlation is better defined in the Ni + Fe-Mg diagram (Fig. 4 d). The steepest slopes are for serpentine II from Loma Caribe (Fig. 4 b) and Loma Ortega (Fig. 4 b, c). The Ni-Mg data from Loma Ortega and Loma Caribe are more similar to those of Golightly and Arancibia ( 1979 ) and Pelletier ( 1996 ) than those from Yamanigüey (Fig. 4 b). In contrast, the Fe-Mg data from Yamanigüey are close to those of Pelletier ( 1996 ) whereas Fe-Mg data from Loma Ortega and Loma Caribe are similar to those of Golightly and Arancibia ( 1979 ). The negative correlation confirms that Ni and Fe exchange for Mg in the octahedral layer of the serpentine structure.

Thermodynamic modeling
Thermodynamic data selection for serpentines The solubility constants for serpentine II depend on their composition in terms of Mg, Ni, and Fe(III) contents. These serpentines can be described as a solid solution between three end members: a pure Mg, a pure Ni and a pure Fe(III) serpentine. Assuming that the solid solution is behaving ideally (Boschetti and Toscani 2008 ), the equilibrium constant of serpentine of a given composition can be calculated following Eq. 5 , where K is the equilibrium constant of end member i (Mg-, Ni-, or Fe(III)-end members) and χ , the molar fraction of this end member in the serpentine.  ( 2004 ) and Robie and Hemingway ( 1995 ). The logK (reaction 6 ) values calculated from these thermodynamic and auxiliary data of ThermoChimie v.9 database are  Available ∆G values for népouite and Fe(III)-serpentine in the literature are estimated following different methods. The estimations proposed in Nriagu ( 1975 ) and Chermak and Rimstidt ( 1989 )  Solubility constants for all serpentine II analyses were then calculated using Eq. 5 . Calculated values range from 31.4 ± 0.9 to 25 ± 1 and correspond to those samples with higher and lower Mg respectively. Most of logK values are in the range between 28.0 and 32.0, with an average of 29 ± 1 (Fig. 5 a). No significant differences in logK values have been observed between serpentines of Loma Ortega, Loma Caribe and Yamanigüey, which is coherent with the low variation of serpentine compositions among the three localities (Fig. 5 b). In addition, logK values calculated for serpentines II from Indonesia reported in Golightly and Arancibia ( 1979 ) are also within the range of values calculated for Caribbean serpentines, and similar to the values calculated for the samples from New Caledonia of Pelletier ( 1996 ). The minimum logK value obtained in a serpentine from Pelletier ( 1996 ) corresponds to an analysis of a serpentine II with extremely low Mg content in comparison with the other samples provided by Pelletier ( 1996 ). Taking into account that calculated logK values for the different serpentines are all within a narrow range (31.4 ± .0.9 to 25 ± 1), only three samples have been selected to carry out the thermodynamic calculations. Si O (OH) ) from Yamanigüey, have been selected as representative of the highest (31.4 ± 0.9) and the lowest logK (25 ± 1), respectively.

Saprolite porewaters
Information on groundwater composition associated with the saprolite horizon is scarce. Landauro Sotelo ( 2008 ) groundwater samples from the Falcondo Ni-laterite deposit (Dominican Republic). The samples are from groundwater flowing through the saprolite horizon (Table 2 ). pH values varies from 8.28 to 9.26. The most abundant cations are Mg, Si, and Na, with concentrations between 13 and 42 mg/L (~1 × 10 M), 25 and 55 mg/L (~4 × 10 M), and between 4 and 7 mg/L (~2 × 10 M), respectively. Nickel concentrations are between 2 (4 × 10 M) and 120 µg/L (4 × 10 M) while iron concentrations are between 100 and 220 µg/L (2 to 4 × 10 M). Bicarbonate concentration is about 100 mg/L (~2 × 10 M) and Cl concentrations are between 4 to 9 mg/L (~2 × 10 M). Speciation calculations using PHREEQC v.3 and the Thermochimie v.9 database (Giffaut et al. 2014 ) indicate that the ionic strength values of the groundwater are between 2.0 × 10 and 5.0 × 10 mol/L (Fig. 6 a). C, Cl, S, N, and P are mostly forming HCO , Cl , SO , and HPO , respectively, although they also form soluble complexes with Mg, Na, Ca, and in minor amount with Ni. Ca, K, Na, and Mg are mostly presented in their free form as Ca , K , Na , and Mg although they secondarily form aqueous complexes with CO . Ni is mainly as free ion Ni but also forming Ni(OH) (aq) and Ni(CO )(aq) aqueous complexes. Since no redox data are available, the speciation of Fe cannot be computed. Calculated saturation indices show that the waters are near equilibrium with calcite and quartz (Fig. 6 b).

Fig. 6
Ionic strength (mol/L) (a) and saturation indices of quartz and calcite (b) of each porewater sample listed in Table 2 . Dotted line in (b) represents equilibrium Thermodynamic calculations Figure 7 shows the solubility of serpentines LC-107,   as carbonate has been considered equal to those of water sample A9a (Table  2 ), the groundwater with the highest pH. Additionally, calculations have also been performed using porewater composition of sample A6a (lowest pH) but no significant differences arised. The solubility curves are compared with Si, Mg, Fe and Ni concentrations measured in saprolite groundwaters from the Falcondo Ni-laterite.  solubility of serpentine II varies over several orders of magnitude for small changes in pH, when pH is in the range 7.5-8.5. This is evidence of the strong effect of slight variations in pH on serpentine II solubilities. No significant differences are observed between solubilities calculated at Eh = 0.4 V and at Eh = −0.3 V except for serpentine YAM-179. This represents the serpentine with lower logK , and those with higher Fe(III) content. Solubility at Eh = 0.4 V is lower than at Eh = −0.3 V, indicating that stability of serpentine increases when redox values become more oxidizing. This is consistent with the fact that these serpentines are richer in Fe(III).
In Fig. 7 , quartz solubility is shown. Si concentrations measured in groundwater fall close to the solubility of the different serpentines (except that of YAM-179 calculated at 0.4 V) and quartz, indicating that groundwater could be near equilibrium with these minerals, as suggested by the saturation indices calculated for groundwater samples (Fig. 6 b). This is coherent with the observation of quartz (mixed with goethite) replacing olivine grains in the saprolite samples, surrounded by serpentine (Fig. 3 d).
In addition, Fig. 7 displays goethite solubility, calculated at Eh = −0.4, −0.3, and 0.4 V, given the high dependence of goethite solubility on this parameter. As can be seen, Fe concentrations measured in groundwater are similar to the solubility of serpentine at the pH range measured in the field, but also to the solubility of goethite at reducing Eh (−0.4 and −0.3 V). Porewater samples are far from solubility of goethite at 0.4 V. Therefore, it can be stated that the saprolite horizon of the Falcondo Ni-laterite, where goethite and serpentine have been observed close to equilibrium, must be under slightly reducing conditions rather than under slightly oxidizing conditions. Mg and Ni aqueous concentration are also close to the solubility of serpentine calculated at −0.3 V. Figure 8 shows the predominance diagram pH-Eh(V) (25 °C, 1 atm) of the Fe system for water compositions of the Falcondo Ni-laterite deposit. The stability field for LC-107 sample, representing the most common serpentine composition analyzed (Mg Fe Ni )Si O (OH) ), indicates that this phase is stable at pH higher than 8.0 and Eh lower than −0.2 V. This stability field slightly decreases for serpentines with compositions similar to LO-17, that is with more Mg and less Fe(III) ((Mg Fe Ni )Si O (OH) ) (Fig. 8 ) serpentine towards higher, even oxidizing, Eh values (Fig. 8 ). The Eh value of each porewater sample has been calculated assuming that porewater is simultaneously in equilibrium with goethite and serpentine II. In these calculations, performed for LC-107, LO-17, and YAM-179 samples, pH has also been recalculated to allow water to be in equilibrium with the minerals. Results (Fig. 8 ) show that pH should be between 8.0 and 8.5 and Eh between −0.24 and −0.31 V, in agreement with the previous calculations.
According to the results presented here, which show agreement between serpentine mineralogical characteristics, saprolite horizon porewater analyses and thermodynamic calculations at 25 °C, the formation of a Fe(III)-Ni-bearing serpentine in the saprolite horizon of a Ni-laterite deposit may take place under atmospheric pressure and temperature, suggesting that these processes can be occurring near the surface.

Conclusions
The detailed mineralogical study of the saprolite horizon of Ni-laterites from the Caribbean region reveals that it is mainly composed of lizardite surrounding silicified cores of Fe-oxyhydroxides (goethite Despite compositional variations of serpentine II, logK only ranges from 31.4 ± 0.9 to 25 ± 1. No significant differences are observed for logK from Caribbean serpentines or even with serpentines II from the literature. The thermodynamic calculations performed with three different serpentine II compositions covering the range of logK calculated indicate that serpentine II 2 3 0 0 0 0 0 solubility is highly dependent on pH, especially in the range 7.5-8.5, while the effect of Eh is negligible except for those cases with higher Fe(III). In those cases, the stability field of serpentine II increases towards oxidizing conditions.
Solubility calculations provide Fe, Si, Ni, and Mg concentrations similar to those measured in groundwater samples from the saprolite horizon of the Falcondo Ni-laterite from the literature. Eh and pH values of the porewater samples, calculated assuming that porewater is simultaneously in equilibrium with goethite and serpentine II, show that pH should be between 8.0 and 8.5 and Eh between −0.24 and −0.31 V.
Geochemical calculations indicate an agreement between serpentine mineralogical characterization, saprolite horizon porewater analyses, and thermodynamic calculations at 25 °C, suggesting both that the formation of a Fe(III)-Ni-bearing serpentine in the saprolite horizon of a Ni-laterite deposit may take place under atmospheric pressure and temperature and that these processes are currently occurring near the surface.