2.1 Caractérisation du sol

2.2 Dissolution chimique

2.3 Dissolution bactérienne

2.4 Observations en MET et microanalyses

3.1 Caractéristiques du sol

3.2 Réduction chimique

3.3 Réduction bactérienne

Quantin et al. [10] ont montré une forte corrélation entre la biodégradation des matières organiques du sol et la réduction des oxydes en conditions anaérobies, en soulignant que l'addition de glucose exerce un effet stimulant sur la dissolution des métaux [10]. En conditions biotiques, la solubilisation de Mn augmente pendant 24 j (97±2 mg l⁻¹), puis se stabilise, alors qu'elle reste très réduite en conditions abiotiques (Fig. 1a). La proportion de Mn dissous atteint 28 % du Mn total. Seulement 20 % du Mn est sous forme ionique libre (Mn²⁺), quantité estimée par formation d'un complexe coloré [10]. Une grande part du Mn est complexée en solution par les composés issus de la fermentation du glucose (acides acétique, propionique, butyrique) et de la biodégradation des matières organiques du sol [10]. La solubilisation de Fe augmente de façon continue jusqu'à 40 j environ, pour atteindre moins de 0,8 % du fer total. La solubilisation de Co, sous influence microbienne, montre une évolution en deux étapes : une solubilisation rapide pendant 24 j, qui atteint 7 % du Co total, puis une diminution lente (Fig. 1b). Le même comportement est observé pour Ni, avec un maximum de solubilisation à 14 j. L'insolubilisation de Co et Ni correspond sans doute à des phénomènes de précipitation et coprécipitation, incluant des phénomènes de sorption et la réorganisation des phases solides.

3.4 Caractérisation de la source de métaux

Des revêtements d'oxydes de Mn sont observés sur le profil de sol, mais ne sont pas détectables par DRX sur la terre fine. Néanmoins, les observations en MET confirment la présence d'un oxyde de Mn dans les échantillons de sol (Fig. 2), qui présente un aspect de papier froissé et une structure aciculaire. L'analyse élémentaire en EDXS montre que ce minéral est constitué principalement de Mn, Co et Al, avec 4,3 % de Ni (Tableau 2). Cette observation peut être rapprochée de celles de Manceau et al. [7], qui ont décrit des oxydes de Mn interstratifiés avec des feuillets de Mn(IV), Co(III) et Ni(II), Al(III) dans les horizons saprolitiques développés sur roches ultrabasiques de Nouvelle-Calédonie. Ce composé, caractérisé comme un oxyde mixte de lithiophorite–asbolane, a une composition très variable [7].

Table 2

EDXS elemental analysis of Mn-oxide particles.

Analyse élémentaire en EDXS des particules d'oxydes de Mn.

Element	Atomic %
Al	11.9 ± 3.3
Si	12.2 ± 9.9
Mn	47.1 ± 1.4
Fe	7.28 ± 2.75
Co	15.3 ± 8.4
Ni	4.30 ± 2.25

Les Geric Ferralsols de Nouvelle-Calédonie dérivés de péridotites présentent des teneurs exceptionnelles en métaux tels que Ni, Cr et Co. Les expériences de dissolution chimique montrent que Co est principalement associé aux oxydes de Mn et Ni aux oxydes de Fe. Au cours d'incubations en conditions anaérobies, l'activité bactérienne solubilise respectivement et au maximum 28,6, 7,8 et 0,4 % du Mn, Co et Ni total en 25 jours. Des observations en MET montrent la présence d'un oxyde de Mn porteur de Co, Ni et Al. Bien que la majeure partie de Ni soit associée aux oxydes de Fe, une phase minérale mineure d'oxyde mixte de Mn à feuillets de lithiophorite–asbolane apparaı̂t comme une source majeure de métaux biodisponibles. Dans les sols de plaine soumis à engorgement temporaire, ces oxydes de Mn, très sensibles aux variations des conditions d'oxydo-réduction, en raison du potentiel élevé du couple Mn⁴⁺/Mn²⁺, peuvent être réduits, à la fois par des processus chimiques et bactériens, plus rapidement et efficacement que les oxydes de Fe. Cette phase minéralogique mineure peut donc être la source principale de métaux facilement biodisponibles en conditions naturelles.

A subsurface soil sample (4–10 cm) was collected in a plain Ferralsol described elsewhere [1]. It was sieved at 2 mm, a part was air-dried for analysis and the other was kept moist (in the dark at 4 °C) until experiments were initiated.

Total C and N in the solid phase were quantified using a CHN 1108 Carlo Erba analyser. Total mineral elements of the sample were measured by ICP–AES (Jobin-Yvon 238) after dissolution in a lithium metaborate–tetraborate mixture at 1000 °C for Fe and Mn and a diacid digestion (2:1 concentrated HNO₃:HCl ratio) in a microwave oven for Ni, Cr and Co.

The hydroxylamine hydrochloride procedure of Chao [4] was used for Mn oxide extractions. One gram of soil was stirred with 25 ml of 0.1 M NH₂OH·HCl (pH = 2) for 30 min at 20 °C. Dissolution of iron oxides was carried out using a modification of the dithionite–citrate–bicarbonate (DCB) deferrification method of Holmgren [6]. 125 mg of soil sample were placed in centrifuge tubes with (or without) 350 mg of dithionite and 35 ml of citrate–bicarbonate solution (CB) and mixed gently in an end-over-end shaker at 20 °C for five days. Then, the samples were centrifuged and the supernatant analysed for Fe, Al, Mn, Ni and Co by ICP–AES.

Batch incubations were performed with 50 g (or 25 g for controls) soil in hermetically sealed 1000-ml (or 500-ml) plasma bottles, supplemented with 750 ml (or 375 ml) of glucose solution (1 g l⁻¹). Flasks with soil were incubated in the presence or absence of indigenous microorganisms to distinguish microbial from physicochemical processes. Abiotic conditions were obtained by addition of 1 g l⁻¹ of Na-thimerosal. Incubations were carried out anaerobically, with four replicates per treatment. The batch cultures were incubated in the dark at 28 °C for 11 weeks, without shaking except just before sampling.

Metal contents in solution were monitored at various times by collecting 15 ml (or 10 ml for controls) of suspension under sterile conditions. Aliquots of supernatants were filtered (0.2 μm membranes, cellulose acetate, Sartorius) and analysed by ICP–AES for total metal content. Fe and Mn speciation in solution was studied colorimetrically as described elsewhere [10]. pH and E_h of the medium were also measured.

Air-dried samples were suspended in ethanol under ultrasonication. A drop of suspension was then evaporated on a carbon-coated copper grid, and the preparation was observed with a Transmission Electron Microscope (Philips CM 20) at an accelerating voltage of 200 kV. The microscope was equipped with an EDAX energy dispersive X-ray spectrometer and provided elemental analysis of particles.

The soil was very rich in iron and relatively rich in Mn, Ni and Co (Table 1). The Fe-oxides were dominated by goethite, but hematite traces were observed [2]. The ferro-manganese coatings observed in the soil profile [1] suggested the existence of oxidation-reduction processes. However, as noticed elsewhere [2], the gleyic colour pattern of the soil is weak, indicating that reduction processes are probably slight. Soil pH was acidic. The organic matter content reached 5% and the C:N ratio was 22. Further details of the chemical and mineralogical characteristics of the soil are presented by Becquer et al. [2].

More than 85% of total Fe, Mn, Co and Ni were dissolved by DCB reagent (Table 1). Hydroxylamine hydrochloride solubilised 50% of total Mn and Co and less than 1% of total Fe and Ni. These results suggested that Mn and Co were mainly associated in Mn-oxides and Fe and Ni in Fe-oxides, because hydroxylamine hydrochloride dissolves specifically manganese oxides and only very low and limited amount of both iron oxides [4] and acido-soluble minerals. It can be considered that Ni included in substituted ferric oxides is dissolved congruently with Fe [13]. If we considered that all the solubilised Ni was originated from Fe-oxides having a Ni:Fe ratio of 0.02 (w/w), the amount of Ni extracted with hydroxylamine hydrochloride should amount to 37 mg kg⁻¹ according to the amount of Fe extracted by this reagent. Such a theoretically calculated amount is markedly lower than the amount of 86 mg kg⁻¹ really extracted with hydroxylamine hydrochloride (Table 1). Therefore, it can be proposed that nearly 40% of Ni extracted with hydroxylamine hydrochloride is associated with Fe oxide and 60% with Mn oxide.

A previous study has shown a correlation between organic matter biodegradation and oxide reduction processes under anaerobic conditions [10] and that the addition of an easily biodegradable compound like glucose increased the dissolution of the metals [10]. In abiotic treatment, Mn solubilisation was very low (Fig. 1a) and corresponded to the Mn exchanged with Na from thimerosal. In biotic conditions, Mn solubilisation increased continuously during the first 24 days, reaching 97±2 mg l⁻¹, then a stationary phase occurred (Fig. 1a). The proportion of total dissolved Mn reached then 28% of total Mn. Free ionic Mn²⁺, determined by the formation of soluble coloured complex, represented less than 20% of dissolved Mn, suggesting that a large part was complexed in solution by organic compounds (acetic, propionic, butyric acids...) originated from glucose fermentation and soil organic matter biodegradation [10]. Organic acids (acetic, propionic and butyric acids) were produced by bacteria in complement of other organic dissolved matter as fulvic-like compounds [10]. Iron reduction increased over time to reach less than 0.7% of total iron (i.e., 268±27 mg l⁻¹) in biotic treatment, whereas no detectable reduction appeared in abiotic controls (Fig. 1a).

Co and Ni solubilisation was very low in controls (1% and 0.9‰ of Co and Ni, respectively) and attributed to metal exchange with Na of the thimerosal (Fig. 1b). In biotic conditions, Co solubilisation showed a two-step process: first an increase until day 24, when solubilisation reached 7% of the total Co, and then a slow decrease. For Ni, the same trend was observed with a maximum solubilisation occurring at day 14. The non-solubilisation of Co and Ni may correspond to an adsorption–precipitation and coprecipitation phenomenon, including sorption and reorganisation of the solid compartments.

These experiments showed that bacteria growing in these anaerobic conditions [10] are significantly involved in the dissolution by reduction of Fe and Mn-oxides containing Co and Ni. Comparison with chemical dissolution showed that more Fe is solubilised by bacterial activity than by hydroxylamine. This suggests that crystallised Fe-oxide can be weathered by Fe-reducing bacteria [10], whereas it is not reduced by hydroxylamine. However, Fe-oxide solubilisation increased relatively slower and longer than Mn solubilisation (Fig. 1a), when the E_h of the solution was very low, whereas Mn-oxide solubilisation increased rapidly, from the beginning of the incubation, during E_h decrease (results not shown). The different trends of Mn- and Fe-oxides are more obvious when the soil organic matter is the only source of carbon and energy [10]. The lack of morphological evidence in the soil profile of severe reducing conditions lead us to consider that under natural conditions, Mn-oxides are more subjected to reduction process than Fe-ones. Only 60% of Mn solubilised by hydroxylamine was solubilised by bacterial reducing activity. Therefore all the Co and Ni associated to Mn-oxide was not microbiologically solubilised under these experimental conditions. Limitation can be due to the lack in energy source available as metabolisable organic matter to microorganisms.

Mn-oxide coatings that were observed in the soil profile were not detectable by powder X-ray diffraction on the bulk soil. Nevertheless, TEM observation revealed the presence of Mn-oxide in the sample (Fig. 2). This Mn-oxide is organised as screwed paper and acicular elementary particles. EDXS spectra showed that this mineral is composed mainly by Mn, Co and Al with 15.3% of Co and 4.3% of Ni (Table 2). Such observation can be related to the results of Manceau et al. [7], who observed Mn-oxides with large variation of their Ni, Co and Al contents in the saprolitic horizons on the ultramafic rocks from New Caledonia. They described these compounds as an irregular intergrowth of Mn(IV), Co(III) and Ni(II), Al(III) sheets, which they called lithiophorite–asbolane mixed layers Mn-oxide.