Introduction

Chromium exists in nature in two stable and predominant oxidation states, Cr(III) and Cr(VI) that have different chemical and toxicological profiles. Under slightly acidic to alkaline conditions of natural waters, Cr(III) species precipitate out of solution, forming hydroxide precipitates (Cr(OH)3(s) or FexCr1-x(OH)3) of very low solubility (Fendorf, 1995; Hausladen et al., 2018; Rai et al., 1989). In the environment, Cr(VI) exists as oxyanions (HCrO4−, CrO42−) that are thermodynamically stable at oxic and alkaline conditions (Rai et al., 1989). Chromates may adsorb onto positively charged Fe hydrous oxides and clay minerals, a process that is favored by acidity (Fendorf, 1995). The anionic nature of its species renders Cr(VI) very mobile and potentially bioavailable in the environment (Oze et al., 2004). In terms of toxicity, Cr(VI) hexavalent Cr is isostructural with sulfate and phosphate at physiological pH, and as such, it is readily taken up by cells and transported throughout the body and even brain. Inhalation, ingestion, and dermal exposure to Cr(VI) may cause severe adverse health effects, including cancer (Costa & Klein, 2006; Saha et al., 2011). Contrarily, Cr(III) is largely excluded from cells and thus is much less toxic than Cr(VI) (Sun et al., 2015). Until recently, Cr(III) was considered as an essential element, but recent studies suggest that it is rather a pharmacological agent (Zoroddu et al., 2019).

Ultramafic rocks, mainly peridotites and pyroxenites, and their weathering products (serpentinites) are the most important natural repositories of Cr, with Cr contents exceeding 3000 mg kg–1 (Kabata-Pendias, 2010). Prolonged and intense weathering of ultramafic rocks under warm and humid climates resulted in the formation of Fe–Ni laterite deposits, where Cr contents may reach values up to 90,000 mg kg–1 (Eliopoulos et al., 2012). In ultramafic settings, Cr exists as Cr(III), predominantly in Cr spinel mineral groups (with the general formula (Mg,Fe)O⋅(Cr,Al,Fe)2O4), Cr-bearing silicates (e.g., olivine, pyroxene, chlorite) and Fe (oxyhydr)oxides (goethite, hematite) (Kierczak et al., 2020; Vithanage et al., 2014). However, Cr(III) can be oxidized to Cr(VI) by mixed valence Mn(III/VI) oxides (Bartlett & James, 1979; Fandeur et al., 2009a; Garnier et al., 2013; Oze et al., 2007). Generated Cr(VI) could be reduced back to the trivalent state abiotically by organic matter, Fe(II), and S2– and by microbially mediated processes (Bishop et al., 2014; Hausladen & Fendorf, 2017; James & Bartlett, 1983; Oze et al., 2007). As a combined result of redox reactions, elevated Cr(VI) concentrations in water and soil have been reported in numerous areas of ultramafic geological background (Chrysochoou et al., 2016; Kierczak et al., 2020; Vithanage et al., 2019). Thus, geogenic Cr(VI) contamination is a pragmatic environmental issue, deserving further research from a geochemical, ecotoxicological, and managerial point of view.

Laboratory experimental studies, involving aqueous Cr(III) (Eary & Rai, 1987; Fendorf & Zasoski, 1992; Fendorf et al., 1992) or Cr(III) precipitates (Apte et al., 2006; Dai et al., 2009; Hausladen & Fendorf, 2017) and synthetic Mn oxides, have offered a great deal of knowledge on the oxidation and fate of Cr(VI). Owing to the low solubility of Cr(III), most of these experiments were designed at acidic pH values (pH 3–5). However, such low pH values may not be common in natural settings, and additional research is needed to understand the extent and rates of Cr(VI) formation in natural systems and its subsequent release into the environment (Chrysochoou et al., 2016; Garnier et al., 2013; Hausladen & Fendorf, 2017; Kazakis et al., 2015; Lilli et al., 2019). In complex mineral assemblages of ultramafic rocks and derived products, the formation of Cr(VI) is dependent on the dissolution of Cr(III) host minerals, the accessibility of Cr(III) to Mn oxides, and retention–release processes of oxidized Cr(VI) from the soil surfaces (Fandeur et al., 2009a; Garnier et al., 2013; McClain et al., 2019; Mills et al., 2011; Oze et al., 2007). Retention–release processes are eventually determinative on the extent of Cr(VI) release to aquatic and soil systems and associated environmental impacts. Yet, the potential mobility of the generated Cr(VI) remains largely unresolved (Garnier et al., 2013; Lilli et al., 2019).

Single and sequential extractions have been widely used to assess the fraction of total metal content associated with specific mineral phases (Becquer et al., 2003; Garnier et al., 2006; Hseu et al., 2018; Kierczak et al., 2016; Rajapaksha et al., 2013; Vithanage et al., 2014). The extractants employed are considered to attack different “pools” of metals in the solid media with different solubilities, thus environmental mobility under specific pH and redox conditions (Davidson, 2013; Gleyzes et al., 2002). Therefore, they could provide useful information on Cr species mobility and their potential impact on the environment. In the case of Cr(VI), the water-soluble and exchangeable fractions are commonly determined to evaluate Cr(VI) mobility from the mineral phases to surface or groundwater (Becquer et al., 2003; McClain et al., 2019; Mills et al., 2011; Raous et al., 2013). Less studied are the non-exchangeable forms of Cr(VI); despite their limited potential to be released to groundwater, the environmental hazards pertinent to these insoluble forms of Cr(VI) are directly associated with airborne, respirable dust and probably with colloid movement in groundwater (James et al., 1995).

The present study focuses on the formation of geogenic Cr(VI) and the factors affecting its release into the environment. It has been motivated by the widespread occurrence of ultramafic rocks from the northern to the southern boundaries of Greece, the exploitation of Fe–Ni laterite deposits generating large amounts of metal-rich wastes, and the progressively increasing number of publications linking the elevated Cr(VI) concentrations in groundwater to oxidative reactions of Cr(III) of the Cr-rich bedrock (Dermatas et al., 2015; Economou-Eliopoulos et al., 2017; Kaprara et al., 2015; Megremi et al., 2019; Moraetis et al., 2012; Papazotos et al., 2019; Pyrgaki et al., 2016, 2020, 2021; Vasileiou et al., 2019). The study examines the levels and potential mobility of Cr species in a natural, complex environment, consisting of serpentinized peridotites and a Fe–Ni sedimentary deposit in an active mining site. The specific objectives are threefold: first, to determine the mineralogy, total chemistry, and Cr contents of the samples; second, to assess the mode of Cr species fixation onto the solid surfaces, hence their potential mobility, by means of sequential (total Cr) and single-step (Cr(VI)) chemical extractions; and finally, to determine the amounts and rates of Cr(VI) released from the solid surfaces into aqueous media at pH values that are commonly encountered in natural settings. The major physicochemical and geochemical factors that exert control on the leachability of Cr(VI) are then explored.

Materials and methods

Geological setting of the study area and sampling

The study was carried out at the Tsouka lateritic deposits of the Lokris province, central Greece (Fig. 1). The nickeliferous laterite deposits are developed on the Late Jurassic–Early Cretaceous ophiolites of central Greece (Papanikolaou, 2009). The geological characteristics of the deposits have been initially mapped by the Institute of Geology and Mineral Explorations of Greece (Maratos, 1965) and later described in detail by several authors (Alevizos, 1997; Eliopoulos & Economou-Eliopoulos, 2000; Eliopoulos et al., 2012; Gamaletsos et al., 2018; Samouhos et al., 2019). Briefly, Tsouka ore deposits are placed on weathered peridotites (serpentinites), and they are transgressively covered by Cretaceous limestones (Alevizos, 1997; Eliopoulos & Economou-Eliopoulos, 2000; Skarpelis, 2006). The Fe–Ni laterite deposit is mainly composed of goethite, hematite, spinels, and Ni-bearing phases—chlorite, serpentine, and talc (Alevizos, 1997; Gamaletsos et al., 2018; Samouhos et al., 2019).

Fig. 1
figure 1

(a) Location of the study area in Greece, (b) aerial image of the Tsouka mine in Lokris province of Central Greece (Google images), (c) close photograph of the studied area of the mine, and (d, e) photographs of the studied samples

Full size image

Sampling was designed to evaluate whether the different rocks and weathered soil that are susceptible to erosion, mostly by runoff events and wind, constitute sources of geogenic Cr(VI) contamination. Samples were collected from the weathered bedrock (serpentinite samples SH1 and SH2), which were exposed due to the excavations in the site, as well as the overlying soil. Samples were also collected from the exposed nickeliferous laterite deposit. Field observations during sampling in the laterites revealed a textural heterogeneity from coarser to finer grained spheroidal particles even at very short distances (often less than a meter). Samples were collected from the coarser (Lat1)- and the finer-grained material (Lat2 and Lat3). Approximately 3 kg of each sample was collected, homogenized, crushed, and subsequently ground in a mechanical mill, equipped with agate mortar and balls.

Mineralogical and geochemical analysis

The mineralogical and textural characteristics of the samples were studied in polished-thin sections using optical and scanning electron microscope (SEM). The bulk mineralogy of the samples was also determined by X-ray diffraction (XRD) on a D8 Advance diffractometer (Bruker AXS) equipped with a LynxEye strip silicon detector, and data were evaluated with the DIFFRACplus EVA v12.0® software of Bruker AXS GmbH (see Supplementary Material; SM). Determination of the chemical composition of the minerals present was performed using a JEOL JSM-6300 SEM equipped with energy-dispersive spectrometer (EDS) and INCA software, at the Laboratory of Electron Microscopy and Microanalysis, University of Patras, Greece. Operating conditions are described in the SM.

Total major and trace elements contents of the powdered samples were determined by LiBO2/Li2B4O7 fusion followed by ICP-ES analysis at the laboratories of Bureau Veritas Commodities, Canada. The detection limit of all elements reported as oxides was 0.01%, except for Cr2O3 which was 0.02%, and for all minor elements 1–5 μg g–1, except for Ni (20 μg g–1).

Sequential extractions were performed to identify the elements–solid phase associations and determine their relative mobility under changing environmental conditions. Apart from Cr, two other elements were determined by sequential extractions: manganese, as a measure of native Mn oxides that could oxidize Cr(III) to Cr(VI), and Fe as a measure of amorphous and crystalline Fe oxides that could sequester Cr(III) during weathering and serve as substrates for Cr(VI), as well as a constructive element of minerals, in order to gain insights into mineral phases transformations along the weathering process. A five-step sequential extraction procedure was employed, under the experimental conditions that are summarized in Table 1 and described further in the SM. Chromium, Fe, and Mn in the supernatants were measured by Flame (Varian SpectrAA 200), or Graphite Furnace (Varian GTA 100-Zeeman 640Z) Atomic Absorption Spectrometry.

Table 1 Description of sequential extraction procedure
Full size table

The pH of the samples was measured by a Jenway combination electrode connected to a digital pH meter (Thermo Orion 3 Star meter) in water suspensions, using a 1:2.5 solid/solution ratio (U.S. EPA, 2004). Moisture content was determined by oven drying at 105 °C for 24 h with repeated weighing at 16 h and 24 h to ensure constant weight.

Chromium(VI) determinations

Water-soluble Cr species were determined in 1:10 solid samples to deionized water (MilliQ, resistivity 18.2 MΩ.cm at 25 °C) extracts (James, 1994). The accuracy of the method was tested by analyzing potassium dichromate spikes in deionized water and was found to be 92% (n = 4). Exchangeable Cr species were extracted from the rock samples by 0.1 M dipotassium phosphate–potassium dihydrogen phosphate (KH2PO4-K2HPO4) buffer solution of pH 7.2 (James & Bartlett, 1983). Total Cr(VI) content of the solid samples was determined using the US EPA method 3060A (U.S. EPA, 1996). The accuracy of the method was tested by analyzing lead chromate spikes on quartz (SiO2; Sigma-Aldrich), which was previously found to be free of Cr(VI), and the recoveries were 101 ± 7% (n = 4). Analyses of samples by all methods were done in triplicate. Details of the methods employed are given in the SM.

Hexavalent Cr concentrations of all extractants were determined by the 1,5 diphenylcarbazide (DPC) method (Bartlett & James, 1979) with a Varian Cary 1E UV–vis spectrophotometer. The absorbance was corrected for the color of native sample by the simultaneous analysis of a sub-sample without the addition of 1,5 DPC reagent. The method’s detection limit (MDL) was 0.90 μg L–1 and precision at concentrations levels of 5 μg L–1 and 50 μg L–1 was 6.9 and 0.9%, respectively. Aliquots of water extracts were acidified with few drops of HNO3 (suprapur® Merck) and stored at 4 °C before analysis for total Cr by GFAAS.

Quick oxidation test

Chromium(III) oxidation capacity of the samples by native oxidants was tested according to following method (Bartlett, 1991): Twenty-five milliliters of 1 mM CrCl3· 6H2O (Acros Organics, 98%) was added in 50-mL centrifuge tubes containing 2.5 g of each sample. After shaking for 30 min, centrifugation, and filtration through a 0.22-μm syringe filter (Fioroni PES-type syringe filters), the concentration of Cr(VI) in the supernatants was determined by the DPC method. To test whether any of the oxidized Cr(VI) remained bound onto the solid surfaces, the solid residues were extracted with 10 mL of phosphates solution, as described previously, and the extracts were analyzed for their Cr(VI) concentration. Because the samples contained a significant water-soluble fraction that could had been released from the solids during the 30 min experimentation time, a slight modification of the original method was introduced: A second series of samples were analyzed for their Cr(VI) contents released in the solution or bound to surfaces as before with deionized water without the addition of CrCl3. The net Cr(VI) produced by the oxidation of Cr(III) was determined by subtraction.

Batch experiments

One sample of each category, SH1, soil, and Lat 3, was incubated at a slightly acidic (pH = 5.7) and a slightly basic pH (pH = 8.5) in four replicates for 502 h (21 days) at room temperature. These two series of experiments (named hereafter pH 5.7 and pH 8.5, respectively) aimed at exploring the release of Cr(VI) in aqueous environments, at pH values that represent typical boundaries of natural surface and ground waters. The design of the batch experiments was largely based on the previous works of Oze et al. (2007) and Rajapaksha et al. (2013) and is described in detail in the SM. Briefly, an aliquot of sample was loaded in polypropylene bottles and mixed with a dilute (0.01 M) buffered acetic acid–sodium acetate solution (pH = 5.7), or 0.01 M ammonium acetate–ammonium hydroxide solution at pH 8.5. The bottles were shaken twice for 1 h in 6-h intervals every day in an end-over-end mechanical shaker and then allowed to stand overnight. Aliquots of exact volumes were extracted at six periods of time, filtered through a 0.22-μm syringe filter, and then measured for Cr(VI) concentrations by the DPC method at the same day. Changes in volumes over the course of the experiments were considered in the calculations of concentrations. At the end of the incubation period, the solid samples were analyzed for their phosphate-exchangeable Cr(VI) and total Cr(VI) contents in three replicates, according to the procedures described previously.

Results and discussion

Mineralogical observations

Mineralogical results revealed that serpentinite samples consisted of serpentine as a major constituent, and Cr spinels, magnetite, calcite, and chlorite as minor components (Table 2 and characteristic XRD patterns in Fig. S1). Observations with optical microscope showed that serpentine was the main alteration product forming mesh, ribbon, and bastite textures. The presence of Ni-bearing serpentine was also confirmed by SEM–EDS analyses (Table S1). Chromium spinels were euhedral to subhedral with lobate boundaries, often fractured, and usually surrounded by thin rims of magnetite (Fig. S2a). Locally, ferrian chromite had also formed along fractures and rims of Cr spinels.

Table 2 Mineralogy of the rock samples identified by XRD and confirmed by SEM–EDS analyses
Full size table

The soil sample, overlying the serpentinites, comprised serpentine, quartz, hematite, chlorite, talc, Cr spinels, goethite, and traces of plagioclase, calcite, and clay minerals (smectite and illite) (Table 2, Fig. S1). The sample was characterized by the presence of angular fragments of serpentinite, composite spheroidal particles, and fine clay minerals aggregates. Chromium spinels were in the form of small subhedral to anhedral grains and showed very little compositional variation (Fig. S2c, Table S1). Hematite and goethite grains were observed mostly as angular fragments within composite granules and showed enrichment in Cr (Table S1).

The samples collected from the Fe–Ni laterite deposit had distinct textural and mineralogical characteristics. The coarser, lower part of the deposit (sample Lat1) was mainly composed of hematite, and goethite; Cr spinels, and clay minerals were present as secondary phases. The upper part of the deposit (samples Lat2 and Lat3) consisted of hematite, goethite, calcite, chlorite, smectite, serpentine, and talc, usually in the form of fine to coarse-grained spheroidal particles, or fragments of spheroidal particles. In all lateritic samples a fine-grained fraction, which consisted of Fe (hydr)oxides (hematite and goethite), chlorite, smectite, serpentine, and talc, filled the spaces between the spheroidal particles (Fig. S2d, Table 2). Cr spinels were mostly subhedral dispersed in the fine matrix, or within peloids (Fig. S2d). In all samples, calcite is related to the presence of Cretaceous limestone.

The major Cr-bearing mineral in all studied samples was Cr spinel, with mean Cr2O3 contents 48 wt% in the serpentinites and ~ 50 wt% in the laterites and soil (Table S1). In the serpentinite samples, significant amounts of Cr were also found in ferrichromite (mean Cr2O3: 25 wt%) and to a lesser degree in magnetite (mean Cr2O3: 0.4 wt%). In the secondary Fe oxides, goethite and hematite, of the weathered units (soil and laterites), mean Cr2O3 contents ranged from 2.1 to 2.9 wt%, whereas in chlorite mean Cr2O3 content was 1.6 wt% (Table S1).

Given the crucial role of Mn oxides in Cr(III) oxidation, the distributions of Mn, Cr, and Fe within the solid fine matrix were investigated by means of elemental mapping followed by EDS analyses on hotspots (Fig. 2a, b). The maps revealed that Mn coexisted with Fe and Cr in several areas. Manganese was detected mostly in Cr spinels in the laterites (mean MnO content 0.7 wt%) and in secondary Fe (hydr)oxides (goethite and hematite) with mean MnO contents 0.3 wt% in the soil and laterites samples, which were also enriched in Cr and Al as a result of the weathering process (Table S1, Fig. 2).

Fig. 2
figure 2

(a) Typical back-scattered electron image of a lateritic sample showing hematite (hm) (light gray grain in the center) surrounded by rims comprising Mg–Al silicates and goethite (goe) forming a composite concentric nodule. The differences in the gray colored areas in the grains as shown in the photomicrograph are due to differences in the iron content (lighter gray areas are richer in iron content than darker gray areas and clay minerals matrix). chr: Cr spinel, chl: chlorite, tlc: talc. (b) Elemental mapping of the previous photomicrograph showing the elemental distribution of Cr (green color), Fe (blue color), and Mn (red color). Note the presence of Mn in Cr spinel grains; as asbolane-type Mn oxides (top left marked with square and Fig. 2c), as intergrowths occupying spaces formed in the fine-grained matrix of clay mineral silicates and as coatings at the edges of the Fe (hydr)oxides in the laterite matrix. (c) Magnified back-scattered electron image of the area marked in Fig. 2b showing aggregates of platelets of asbolane-type Mn oxides (Mn-ox), hematite (hm), and chlorite (chl)

Full size image

Manganese also existed in discrete concretion-like aggregates, as fine intergrowths with clay mineral silicates, and as boundaries of the grains or coatings at the edges of the Fe (hydr)oxides in the laterite matrix (Fig. 2c) corresponding to Mn oxides. EDS analyses (Table S2) revealed that the Mn oxides contained Ni (range: 1.7–4.2 wt% NiO) and sometimes Ni and Co (max: 4.2 wt% NiO and 4.0 wt% CoO) and were classified as asbolane-type Mn oxides [(Ni,Co)1-x(Mn(IV)O2)2-y(OH)2-2x+2y ·nH2O)]. Chukhrov et al. (1982) described asbolanes as ordered mixed-layer minerals in which the layers of Mn(VI) octahedra alternate with octahedral layers of Co, Ni, and Co–Ni that have an island-like (imperfect) appearance. Such fine-dispersed, poorly crystallized minerals could not be easily identified by conventional powder-XRD analysis, and EDS analysis (Chukhrov et al., 1982), or more advanced techniques (particularly for the nano-sized ones) are much more effective (Caraballo et al., 2015). The presence of asbolane was also reported in the Lokris province (Kalatha & Economou-Eliopoulos, 2015; Kalatha et al., 2017), as well as other lateritic settings (Economou-Eliopoulos et al., 2017; Fandeur et al., 2009b; Manceau et al., 1987; Ulrich et al., 2019) as Co and Ni released during the serpentinization and weathering of the parent rock are usually incorporated into new minerals formed (e.g., asbolane, Ni–Mg silicates). Interestingly, Cr was also detected in several spots of asbolane (Fig. 2b, Fig. S3), where its contents ranged from 1.2 to 1.7 wt% Cr2O3 (Table S2).

Although Cr speciation could not be accurately determined with the employed technique, this finding clearly suggests that oxidation of Cr by asbolane-type Mn oxides is actually taking place, adding evidence on the crucial role of asbolane as oxidants of Cr in natural settings (Fandeur et al., 2009b). In the serpentinite samples, asbolane was not detected. Manganese was present in Cr spinels grains (mean MnO: 0.3 wt%). A scarce network of Mn-bearing magnetite (mean MnO: 1.5 wt%) was also present in the serpentinites (Table S1, Fig. S2).

Geochemical composition

The geochemical composition of the samples was consistent with the mineralogical observations and is summarized in Table S3. Serpentinites (SH1 and SH2) exhibited similar chemical composition: they had high SiO2 and MgO contents (39–41% and 30–32%, respectively) and low Fe2O3 and Al2O3 contents (6.9–7.4% and ~ 0.5%, respectively), as a result of the high degree of serpentinization. Chromium and Ni contents were approximately 2000 mg kg−1, in full accordance with the presence of Cr spinels, ferrian chromite, Cr-bearing magnetite, and Ni-bearing serpentine. The formation of ferrian chromite and magnetite, which occurred as the alteration products of the spinel-group minerals, was characterized by depletion in Cr, Mg, and Al, while Fe and Mn were significantly enriched (Table S1). Manganese contents ranged from 697 to 774 mg kg–1.

The soil sample was enriched in Fe2O3 (36%) and Al2O3 (8.4%), as well as Cr (12,103 mg kg–1), Ni (4376 mg kg–1), and Mn (2477 mg kg–1) compared to the serpentinites, due to weathering and subsequent formation of secondary Fe (hydr)oxides (hematite and goethite) and clay minerals (Table S3). Cr spinels and serpentine, which occurred in the serpentinites, were also preserved in the soil sample (see also Table 2).

The substitution of Si and Mg by Fe, Al, Cr, Ni, and other minor elements during laterization was clearly reflected on the chemical composition of samples Lat1, Lat2, and Lat3, representing the laterite ore, where Fe2O3 (as FeT) became the major constituent (48–74%) as secondary hematite and goethite. Laterites were substantially enriched in Cr (18,528–21,929 mg kg–1), Ni (4538–6611 mg kg–1), and Mn (1393–1935 mg kg–1) as a result of the laterization process.

Speciation of Cr, Fe, and Mn assessed by sequential extraction

Total Cr

The residual fraction, represented mainly by Cr spinel and silicates, accounted for 90% of total Cr (CrT) in the serpentinites, 72% of CrT in the soil sample and 60–69% in the laterite samples (Fig. 3; Table S4). The observed decrease of the residual fraction from serpentinites to soil and laterites suggests that even the resistant Cr-bearing minerals can undergo dissolution, in line with other studies reporting Cr mobilization during weathering (Fandeur et al., 2009a; Garnier et al., 2013).

Fig. 3
figure 3

Relative distribution of Cr, Fe, and Mn as determined by sequential extractions in the sum of F1–F6 fractions and in the sum of non-residual fractions, F1–F5. Major targeted phases are: weakly bound, exchangeable, and carbonates phases (F1); Mn oxides (F2); amorphous Fe (hydr)oxides (F3); crystalline Fe (hydr)oxides (F4); residual magnetite (F5); residual (F6). See also Table 1 for details

Full size image

Crystalline Fe (hydr)oxides (extracted by the CDB solution; F4) constituted the second most abundant Cr-bearing mineral phase in the soil (27% of CrT) and laterite samples (29–38% of CrT), in line with sequestration of Cr by hematite and goethite during weathering (Table S1, Sect. Geochemical composition). Residual magnetite (F5) hosted 2–4% of CrT in the serpentinites and 1–2% of CrT in the soil and laterites. Amorphous Fe (hydr)oxides (F3) accounted on average for 2% of CrT in the serpentinites, but for less than 0.2% in the soil and laterites. A similar decrease of Cr contents from the serpentinites to soil and laterites was observed in the first two steps of sequential extractions (F1 and F2 see Fig. 3), suggesting the leaching out of the more labile fractions of Cr during the enhanced stages of weathering. The above results are relevant to the formation of Cr(VI), given that Cr oxidation is likely controlled by the relative ability of Cr(III) to migrate to high valence Mn oxides (Garnier et al., 2013; Mills et al., 2011; Oze et al., 2007).

Iron

Iron in the serpentinite samples was predominantly found in the residual solid phases (54–64% of FeT) (Fig. 3). Non-residual Fe was almost equally distributed in amorphous Fe (hydr)oxides (F3; 11–17% of FeT), crystalline Fe (hydr)oxides (F4; 10–11% of FeT) and magnetite (F5; ~ 15% of FeT). All other solid phases were of minor importance, accounting for less than 2% of FeT. In the soil and laterite samples, crystalline Fe (hydr)oxides were responsible for the retention of more than 82% of FeT. The observed distribution pattern clearly shows the decline of the residual Fe from ~ 60% in the serpentinized bedrock to less than 15% of FeT in the overlying weathered units. The dominance of crystalline Fe oxides in the soil and laterites is consistent with the oxidation of released Fe(II), combined with removal of the soluble elements (Mg and Si) during the laterization process. In the case of soil, the CBD- extractable Fe oxides could also account for pedogenically formed magnetite (Hunt et al., 1995) (see also SM).

Manganese

Manganese oxides, targeted by the HA-A extraction (F2), were present in all samples in amounts ranging from 204 (SH2) to 1227 mg kg–1 (soil) (Table S4). In the serpentinite and soil samples, Mn oxides represented the majority of Mn, accounting for 29–42% and 50% of MnT, respectively, whereas in the laterite samples, the extraction yields were lower, ranging between 13 and 26% of MnT (Fig. 3). The CDB extraction (F4), targeting crystalline Fe (hydr)oxides, released 450 mg kg−1 from the soil (18% of MnT) and 278–603 mg kg−1 (20–31% of MnT) from the laterite samples (Table S4; Fig. 3). Manganese was observed to substitute into the structures of Fe (hydr)oxides in the soil and laterites samples (Table S1). Manganese-substituted goethite can also oxidize Cr(III), at rates increasing as Mn substitution increases (Wu et al., 2007).

Hexavalent Cr

Table 3 summarizes Cr(VI) contents determined by deionized water extraction, targeting water-soluble forms, nitrates, and phosphates extraction, targeting both exchangeable and soluble forms, and the alkaline digestion, which determines total Cr(VI) and includes the soluble, exchangeable, and insoluble chromates. Hexavalent Cr was present in all samples of the studied sequence, yet in varying contents among the different extractions, as well as the different geological categories of samples (serpentinites, soil, and laterites).

Table 3 Cr(VI) contents (mean of three replicates and standard deviation in μg kg−1) determined by various methods
Full size table

Total Cr(VI)

Total Cr(VI) contents accounted for less than 0.1% of CrT in all samples, evidencing that only a small fraction of the dominant Cr(III) is oxidized to Cr(VI) (Mills et al., 2011; Rajapaksha et al., 2013). Hexavalent Cr contents were lower in the serpentinites (492 and 1080 μg kg–1) and substantially increased in the soil (5932 μg kg–1) and laterite samples (3218–11,519 μg kg−1). This result, combined with the increase of non-residual (and more labile) fractions of total Cr from the serpentinized bedrock to the weathered units (laterite, soil) (see sequential extractions), suggests that more Cr(VI) was generated during the advanced stages of oxidative weathering, which were accompanied by increased Cr mobility. Increased contents of Cr(VI) in the laterite unit in relation to the bedrock were also reported by Fandeur et al. (2009b) in a lateritic regolith in New Caledonia.

Similar to Mills et al. (2011), a correlation between total Cr(VI) and the amounts of Mn extracted by HA-A was not encountered. However, a statistically significant positive correlation (Spearman’s rho: 1) was found between Cr(VI) contents and Mn bound to Fe crystalline (hydr)oxides. This finding suggests that Mn-bearing Fe oxides deserve further research as a potential oxidizing agent of Cr(III).

Water, nitrate, and phosphate extracted Cr(VI)

Water-soluble Cr(VI) contents were low in the serpentinite samples (26–54 μg kg–1) and increased by an order of magnitude in the laterites (477 and 798 μg kg–1 in Lat2 and Lat3, respectively) and the soil sample (312 μg kg–1) (Table 3). The quite low values of the water-soluble Cr(VI) contents found in sample Lat1 (61 μg kg–1) compared to the other laterite samples are consistent with the textural and mineralogical characteristics of the sample.

Chromium(VI) extracted by the phosphates solution ranged between 84 and 1033 μg kg–1. These values are much lower than the ones reported for laterites and acidic soils in Niquelandia, Brazil (Garnier et al., 2009; Raous et al., 2013), or laterites and soils in New Caledonia (Becquer et al., 2003; Fandeur et al., 2009a), but higher than those of alkaline serpentinized soils of the Sacramento Valley (Mills et al., 2011). The pH values of the samples in our study ranged from 7.7 to 8.4 (Table S3). Under alkaline pH, part of the generated Cr(VI) could be easily leached out, since Cr(VI) association with solid phase surfaces is limited to positively charged exchange sites, the number of which decreases with increasing soil pH. By contrast, acidic pH favors the retention of Cr(VI) onto the solid surfaces (McLean & Bledsoe, 1992).

Nitrates and phosphates extractions of the soil and laterites samples released comparable amounts of Cr(VI) with the deionized water extraction, suggesting that soluble chromates predominate over exchangeable forms in these samples. Nitrates-extracted Cr(VI) in the soil (150 μg kg–1) was lower than the water-soluble fraction (328 μg kg–1), probably due to adsorption of nitrate ions on the solid surfaces. Contrary to soil and laterite samples, serpentinites contained much higher amounts of phosphate extracted chromates compared to the water-soluble and nitrates-extracted fraction. Unbuffered neutral salts, such as NaNO3, desorb electrostatically adsorbed ions (Jiang et al., 2008). Phosphates adsorb specifically onto mineral surfaces; thus, extraction by these anions may additionally desorb specifically adsorbed anions (Bartlett & Kimble, 1976). Thus, the relative differences of Cr(VI) contents determined by the different extractions indicate that chromates occurred in various forms (i.e., precipitated/surface complexed) in the samples studied and were sorbed on the surfaces with a variety of binding strengths (i.e., high/low energy bonds).

The competitive ion displacement results of this study could infer the negative consequences on Cr(VI) mobility arising from agricultural pollution. The presence of phosphate ions in water, originating from phosphorus-based fertilizers, could enhance the desorption of previously immobilized Cr(VI) by retention to the solid surfaces (Kierczak et al., 2020). The highest Cr(VI) concentration determined in soil solutions of phosphorus fertilized croplands is 700 μg L–1 (Becquer et al., 2003). As far as nitrate ions are concerned, it has been previously proposed that acidification, driven by nitrification process, could promote the release of Cr(III) and its subsequent oxidation (Mills & Goldhaber, 2012; Mills et al., 2011; Papazotos et al., 2019; Vasileiou et al., 2019). Our results also show that, although not as effective as phosphate ions, nitrates could promote the release of Cr(VI) by desorption processes.

Water-soluble, exchangeable, and non-exchangeable forms of Cr(VI)

To better estimate the potential mobility of Cr(VI) from the samples, the distribution of Cr(VI) into water-soluble, nitrate- and phosphate-exchangeable, and non-exchangeable forms was examined in Fig. 4. The non-exchangeable forms were determined by subtracting the phosphates-extracted fraction from total Cr(VI). This fraction includes precipitated chromates, as well as chemisorbed, non-exchangeable by phosphates Cr(VI). The phosphate-exchangeable forms were determined by subtracting the nitrates-extracted from the phosphates-extracted fraction of Table 3, whereas the nitrate-exchangeable forms by subtracting the water-soluble from the nitrate-extracted fraction of Table 3.

Fig. 4
figure 4

Distribution of Cr(VI) into non-exchangeable, phosphate-exchangeable, nitrate-exchangeable, and water-soluble forms

Full size image

In the serpentinite samples, Cr(VI) occurred predominantly in non-exchangeable (SH1) and phosphate-exchangeable forms (less weathered SH2), accounting for 52% and 86% of total Cr(VI), respectively. Water-soluble and nitrate-exchangeable forms were of minor importance, accounting for 3–5% of total Cr(VI). In the soil and laterite samples, Cr(VI) occurred predominantly in non-exchangeable forms, at percentages ranging from 82 to 99% of total Cr(VI). A small proportion of total Cr(VI), ranging from 1 to 17%, was water-soluble. The exchangeable forms, either by nitrate or phosphate ions, were totally absent from these samples. The observed substantial increase of the non-exchangeable fraction from the bedrock to the soil and laterite samples suggests a transition of Cr(VI) into immobile forms with increasing degree of weathering.

Cr oxidation capacity of samples

According to the quick oxidation test, all samples oxidized the added Cr(III) to Cr(VI) to some extent. The soil sample had the highest Cr oxidation capacity, and the serpentinite sample SH2 had the lowest (Table 4). These samples contained the greatest (soil) and lowest (SH2) amounts of Mn oxides, as estimated by the HA-A extraction (F2; Table S4). However, the Cr oxidation capacity of all other samples did not correlate with their Mn oxides contents. Instead, a statistically significant correlation was found (Spearman’s rho: 0.829; p = 0.042) only when Mn-bearing crystalline oxides (F4) were also considered. This is in agreement with our previous finding on total Cr(VI) contents and further evidences the role of Mn bound to hematite/goethite on the oxidation of Cr(III).

Table 4 Generated Cr(VI) (in nmoles per mole of added Cr(III)) and its distribution into dissolved and phosphate-exchangeable fractions
Full size table

In the serpentinites and Lat1 sample, the majority of the generated Cr(VI) was found to be sorbed on the mineral surfaces, whereas in the soil and the other laterites (Lat2 and Lat3) the majority of the generated Cr(VI) was in the dissolved phase (Table 4). By comparing the distribution pattern of Fig. 4 and that of the oxidized Cr(VI) (Table 4), it is evident that serpentinites showed a much higher tendency to sorb chromates than the soil and laterites (Lat2 and Lat3). It is noteworthy that the serpentinites contained significant amounts of hydroxylamine hydrochloride (HA-B)-extractable Fe, which were higher than those found in the soil and laterite samples. Laterite 1 contained more HA-B extractable Fe that the other laterite samples (Table S4). These amorphous or poorly crystalline Fe (hydr)oxides are well known for the defects in their crystal structure, where the charge balance is maintained by the incorporation of elements such as chromium, nickel, and aluminum in the structure (Kuhnel et al., 1975). Moreover, due to their higher water content they could provide surface hydroxyl functional groups (Kühnel et al., 1975) in addition to well-crystallized Fe oxides, for the complexation of Cr(VI) species (Fandeur et al., 2009b; Garnier et al., 2013; Johnston & Chrysochoou, 2014). It has been recently suggested that amorphous Fe (hydr)oxides may adsorb more Cr(VI) than goethite because of their higher specific density (Shi et al., 2020). Iron (hydr)oxides may be found in the form of mineral nanoparticles and/or as nanoparticle aggregates, either poorly ordered or partially amorphous, and its environmental significance, is critically arising due to their important role in the mobility and (bio)geochemical cycles of hazardous elements in nature (Caraballo et al., 2015; Hochella et al., 2008). The substantially differentiated fractionation patterns of the generated Cr(VI) among the studied samples point out the importance of the composition of the phases to retain Cr(VI) on the solid surfaces, hence, limiting aqueous concentrations and related human health and ecosystem risks.

Cr(VI) release into solution

To assess the extent and rates of Cr(VI) release from Cr-rich phases into an aqueous media of the minimum and maximum pH values typically encountered in nature, a serpentinite (SH1), soil, and a lateritic sample (Lat3) were incubated over a period of 500 h at a slightly acidic (pH-5.7) and a slightly alkaline pH (8.5).

The concentrations of aqueous Cr(VI) as a function of time are illustrated in Fig. 5. At the end of the pH-5.7 experiments, the serpentinite and soil sample released approximately 5 times more Cr(VI) in the solution than at the slightly alkaline pH, whereas the laterite sample released similar amounts of Cr(VI) at both pH values. This is despite the fact that increased Cr(VI) aqueous concentrations are thermodynamically favorable under the alkaline and oxidizing conditions (Rai et al., 1989). Total Cr(VI) (sum of sorbed and dissolved forms) after the incubations of the serpentinite at both pH values, as well as the laterite and soil samples at the acidic pH (Table 5,) was 1.3–11 times higher than total Cr(VI) of the initial samples (Table 3). The greater Cr(VI) values determined after the incubation experiments clearly demonstrate that excess Cr(VI) was generated during the incubation period. The release of newly formed Cr(VI) into solution explains the fact that aqueous Cr(VI) concentrations in the acidic pH were higher than in the alkaline pH.

Fig. 5
figure 5

Aqueous Cr(VI) concentrations as a function of time during the incubation of serpentinite (SH1), soil, and the lateritic (Lat3) samples at the two pH values. Mean values and standard deviation (n = 4) are marked by triangles and bars, respectively

Full size image
Table 5 Cr(VI) contents (mean of four replicates and standard deviation in μg kg−1) after incubation of serpentinite (SH1), soil, and laterite (Lat3) at the slightly acidic (pH 5.7) and alkaline (pH 8.5) solutions
Full size table

At the pH-5.7 series of experiments, the slightly acidic pH values promoted the acidic dissolution of Cr(III) from Cr-bearing minerals and its subsequent oxidation by native oxidants. The presence of efficient Cr(III) oxidants in the samples has been previously evidenced by the oxidation test. The F1 fraction of Cr (Table S4), which is susceptible to acidic dissolution, could account for the excess Cr(VI) generated during incubations. Therefore, the two prerequisites for Cr(VI) formation, i.e., the potential of Cr mobilization and the presence of native oxidants, are well met for the incubated samples.

The oxidation reaction was characterized by two stages: First, a fast kinetic (0–159 h) in which 67% (SH1), 86% (Lat3), and 83% (Soil) of the maximum amount of Cr(VI) was released in solution (Fig. 5). These results show the ease by which water contamination by Cr(VI) is likely to occur. Second, a slower kinetic (159–494 h) that leads to a plateau of rather constant Cr(VI) concentrations in solution. The depression of the initial rapid rate of Cr(III) oxidation has been also reported in previous studies (Dai et al., 2009; Eary & Rai, 1987; Rajapaksha et al., 2013; Weaver et al., 2002) and has been ascribed to secondary reactions and formation of surface precipitates and complexes (Rajapaksha et al., 2013). The solubility of Cr(III) is limited at pH 4.5–5.5 (McLean & Bledsoe, 1992); thus, at the pH 5.7 of the incubation experiment any Cr(III) in solution should be in equilibrium with the Cr(OH)3(s) phase. Indeed, monitoring of total Cr aqueous concentrations over the course of the experiments showed that Cr(III) accounted for 0–15% only (data not shown).

The majority of the generated Cr(VI) resided onto the solid surfaces, accounting for 52, 79, and 76% of total Cr(VI) in the serpentinite, laterite, and soil incubations, respectively (Table 5). The distribution pattern of Cr(VI) into exchangeable and non-exchangeable forms after oxidation was identical to the initial (pre-incubated) samples (Figs. 4 and 6, respectively), as well as that determined by the quick oxidation test (Table 4): the generated Cr(VI) was much more prone to adsorption in the serpentinite sample, whereas in the soil and laterite samples the majority of solid-bound Cr(VI) resided in non-exchangeable by phosphates forms, i.e., insoluble phases, including sparingly soluble Cr(VI) precipitates. Thus, retention processes varied among the samples and were strongly depended on the chemical composition of Cr-bearing rocks. Retention processes are particularly important as they limit aqueous Cr(VI) concentrations, subsequently, greatly reduce the environmental and human health hazards related to Cr(VI) contamination.

Fig. 6
figure 6

Comparative distribution patterns of Cr(VI) into the dissolved, phosphate-exchangeable, and non-exchangeable forms before and after the incubation experiments

Full size image

Hexavalent Cr in the SH1 sample was produced even in the slightly alkaline pH (Tables 3 and 5), though at a five times lower rate than that for the acidic pH (0.21 nM h–1 at pH-8.5 compared to 1.07 nM h–1 at pH-5.7 experiments). At alkaline pH the dissolution of Cr(III) minerals decreases, and surface precipitation of Cr(III) hydroxides is enhanced (Rajapaksha et al., 2013). In the case of soil and Lat3 incubations at the slightly alkaline pH, the range of total Cr(VI) at the end of experiments overlapped with that of the initial samples. Thus, there is no clear evidence that oxidation had taken place in these series of experiments, and the evolution of dissolved Cr(VI) concentrations over time (Fig. 5) is treated as resulting from release processes. The release rates for Cr(VI), estimated for the first 159-h experimental period, were 0.38 nM h–1 for the soil and 0.42 nM h–1 for the Lat3 (compared to 1.20 and 0.32 nM h–1, respectively, at the acidic pH). As shown in Fig. 6, the increase of soluble Cr(VI) in the Lat3 from 17% at the initial conditions to 36% at the end of the pH-8.5 incubation period was paralleled by the decrease of non-exchangeable chromates from 82 to 63% of total Cr(VI). The decrease of non-exchangeable chromates together with the continuous release of Cr(VI) over the experimental period suggest that Cr(VI) released in the solution derived from the dissolution of chromate precipitates. Although desorption processes have been relatively well studied in the literature, our results point out the need to study in more detail the formation–dissolution processes of Cr(VI) precipitates in order to better understand the mechanisms of ground and surface water contamination by geogenic Cr(VI). Finally, it should be noted that lower rates are expected in nature, as continuous mixing, increased solution-to-solid ratio, and exposure of larger surface areas due to the grinding of samples are encountered in the batch experiments.

Implications on groundwater quality

Since the increasing number of studies reporting elevated Cr(VI) concentrations in surface and groundwater of UM settings worldwide has been the major motive of this research, here we discuss some critical issues pertaining to water contamination by geogenic Cr(VI). The results of this study revealed that oxidative production of Cr(VI) occurred readily in serpentinites, laterites, and deriving soil. The rates calculated from the laboratory experiments support previous findings (Lilli et al., 2019; Oze et al., 2007; Rajapaksha et al., 2013) that Cr(VI) can be formed rapidly enough providing measurable concentrations in the soil solution and groundwater. However, the extent and rate of oxidation, as well as the lability of the generated Cr(VI), were greatly influenced by the degree of weathering, geochemical composition and mineralogy of samples. Thus, exposed material interacting with water (in the saturated zone), pH (as shown through the incubation experiments), Eh, and the presence of reducing agents that could reduce back Cr(VI) to Cr(III) are among the straightforward controlling factors on groundwater Cr(VI) concentrations.

Leaching of Cr(VI) through the unsaturated zone is another pathway of groundwater contamination (McClain et al., 2019). In this case, the thickness of the unsaturated zone, its recharge source and rate and water residence time, also control the levels of aqueous Cr(VI) concentrations (Manning et al., 2015). For example, rapid recharge limits interaction between groundwater and Cr-bearing minerals (Izbicki et al., 2015; Kazakis et al., 2015), while a thick unsaturated zone (and longer residence time of water) could promote the genesis and mobilization of Cr(VI) (Manning et al., 2015).

The presence of manganese oxides, as grain coatings, deposits in cracks or fractures (Frei et al., 2014), or finely disseminated grains (as in this study) is a prerequisite for the oxidation of native Cr(III). Although we proved that all samples contained Mn oxides at adequate amounts to oxidize Cr(III), other studies have shown that the rate of oxidation could be accelerated with the increase of Mn oxide concentration (Rajapaksha et al., 2013). Under this aspect, water table oscillations, resulting from seasonal variation or water abstraction for irrigation and public supply, favor alternate redox reaction and enhance Mn cycling. The re-generation of Mn oxides, either biotically or abiotically, facilitates the transport of Mn oxide surfaces on or near solid phase Cr(III) accelerating eventually the formation of Cr(VI) (Hausladen & Fendorf, 2017; McClain et al., 2019).

Finally, the vicinity of an aquifer to ultramafic outcrops generally controls aqueous Cr(VI) concentrations (Morrison et al., 2011). Considering that ultramafic outcrops exist in various areas around the world, geogenic Cr(VI) may be widely distributed within groundwaters (Oze et al., 2007). Nevertheless, alluvial transport and wind-blown fine particles (particularly from exposed sites of mining operations) may also redistribute Cr(III)-bearing minerals (or Cr(VI)-bearing particles) at long distances away from their source formations (Hausladen et al., 2018).

Conclusions

The present study provided information about the genesis and the mode of fixation of Cr(VI) in weathered serpentinite rocks, soils, and laterites and the behavior of Cr(VI) when such formations get in contact with aqueous media at pH values commonly encountered in nature. The main conclusions reached are summarized as follows:

During the serpentinization and laterization processes, Cr-bearing minerals dissolve and Cr repartitions into secondary minerals. The more weathered products (soil and laterites) contained more Cr(VI) than serpentinites due to increased Cr mobility. All of the studied samples, although in varying extent, had a self-capacity to oxidize Cr(III) into Cr(VI). Asbolane-type Mn oxides, as well as Mn-bearing crystalline Fe (hydr)oxides seem to play a significant role in Cr(VI) formation. Therefore, the extent of surface and groundwater contamination with geogenic Cr(VI) would primarily depend on the availability of Cr(III) species, rather than the presence of non-native and/or additional oxidants. Furthermore, the mode of Cr(VI) fixation on solid surfaces is another critical factor that could limit water contamination. Although adsorption has received much attention in the literature, relatively little is known about precipitation processes and further research is needed in order to understand their role on the immobilization of Cr(VI).