Cobalt (Co) has exceptional high-temperature strength, corrosion resistance, catalytic, and electrochemical properties, which make it a crucial element for high-performance applications. Due to the development of these applications, and in particular lithium-ion batteries, the cobalt demand is growing rapidly. Co was thus classified by the European Commission as a critical raw material (European Commission, 2020). In this context, the CROCODILE project aimed at developing innovative metallurgical systems based on advanced pyro-, hydro-, bio-, iono-, and electrometallurgy technologies for the recovery of Co and the production of Co metal and upstream products from a wide variety of secondary and primary European resources (Crocodile Project, 2018).

During primary mining, Co is with few exceptions a by-product of copper and nickel (Ni) primary production. The majority of global Ni resources are hosted by laterite deposits and the remaining portion in sulfide deposits (Mudd and Jowitt, 2014). Owing to the exhaustion of higher grade sulfidic deposits as well as to the increase in demand for Ni, the production of Ni from laterites continues to increase. Laterite deposits are characterized by the presence of two distinct zones:

A limonite zone near the surface where Ni is mainly hosted in ferric iron oxides, typically goethite (FeOOH) and to a lesser extent in Mn oxides. Ni content usually varies between 1 and 2%.

Saprolite is usually treated via pyrometallurgical reduction. In contrast, limonite cannot be economically treated via pyrometallurgy due to the combination of a high Fe and a low Ni content. Limonite ore is traditionally processed through pressure acid leaching (PAL). Unfortunately, PAL plants have often proven to be very costly, plagued with technical challenges (Oxley et al., 2016). Still today, a large proportion of limonite ore is stockpiled for future processing. Reductive bioleaching is one of the metallurgical options that are currently studied to extract Co and Ni from limonite.

During reductive bioleaching, the main following reactions occur:

Goethite equilibrium in the acid environment is shifted toward its dissolution (Eq. 1), due to acid generation and bioreduction of Fe(III) to Fe(II) (Eq. 2). This bioreduction also produces the ferrous iron involved in Mn oxide dissolution (Eq. 3). The concept of reductive bioleaching was first introduced by Hallberg et al. (2011). Since then, several studies dealing with the proof of concept of reductive bioleaching have been released (Johnson and du Plessis, 2015; Marrero et al., 2015; Santos et al., 2020, 2022; Johnson and Pakostova, 2021; Stankovic et al., 2022). These studies are mainly dedicated to the study of microorganisms, with a microbiological approach. Very few studies are dealing with process engineering approaches (kinetics data, mass transfer phenomena, etc.). In anaerobic conditions, existing studies only report bioreduction kinetics data obtained with Acidithiobacillus (At.) ferrooxidans. In particular, Kucera et al. (2012) revealed a bioreduction rate of 26.1 mg·L⁻¹·h⁻¹ while in the study of Pronk et al. (1992) bioreduction rate reached 25.6 mg·L⁻¹·h⁻¹. A recent study from Breuker and Schippers (2024) reported a value of 19.9 mg·L⁻¹·h⁻¹ in anaerobic conditions in the presence of 10⁷ cells·mL⁻¹ for At. ferrooxidans after tetrathionate growth in aerobic conditions. However, it is admitted that the use of microbial consortia generally increases the kinetics compared to pure cultures, due to the synergistic effects between microorganisms (Mahmoud et al., 2017). Thus, determining bioreduction kinetics in the presence of microbial consortia is of great interest. Obtaining kinetics data (and in particular Fe(III) bioreduction kinetics as one of the key reactions for the process) is necessary for reactor sizing and other engineering aspects, which will impact the economic and environmental assessments of such processes.

The objective of this study was thus to determine Fe(III) bioreduction kinetics in anaerobic conditions (Eq. 2) in the presence of a microbial consortium containing Acidithiobacillus and Sulfobacillus species. Due to the limited growth of microbial communities under anaerobic conditions, microbial growth was first performed in aerobic conditions using sulfur as an electron donor, followed by Fe(III) bioreduction kinetics experiments. The microbial consortium was initially composed of different microorganisms and was grown at 35°C. The ferric iron sulfate (Fe₂(SO₄)₃) bioreduction rate was first measured in batch mode. It was then determined in the presence of different biomass concentrations in the liquid phase (10⁸ cells·mL⁻¹ and 10⁹ cells·mL⁻¹). Finally, the Fe(III) bioreduction rate was evaluated in continuous mode in the presence of two different growth media.

Table 5 gathers the results of the different experiments. It showed that the presence of microorganisms highly increased the reduction rate compared to the abiotic experiment (0.13 mg·L⁻¹·h⁻¹). A mean bioreduction rate of 1.7 mg·L⁻¹·h⁻¹ was obtained in batch mode with 10⁸ cells·mL⁻¹ in the liquid phase.

Bioreduction experiments highlighted an increase of the bioreduction rate with an increase of the initial biomass concentration in the liquid phase (up to 9.8 mg·L⁻¹·h⁻¹ of mean rate and 12.2 mg·L⁻¹·h⁻¹ of maximal rate when biomass concentration was approximately 10⁹ cells·mL⁻¹ in the liquid phase, compared to 1.7 mg·L⁻¹·h⁻¹ in the presence of 10⁸ cells·mL⁻¹). Similar results were observed in continuous mode. The increase of inorganic nutrient concentrations in 0Km medium enabled to detect a higher biomass concentration in the liquid phase than the ABS-TE medium, resulting in a higher Fe(III) bioreduction rate. A value of 20 mg·L⁻¹·h⁻¹ was observed in continuous mode and 0Km medium. The higher consumptions in 0Km medium compared to the ABS-TE medium showed that larger nutrient concentrations favored their consumption (both from microbial growth and activity and from precipitation phenomena). The availability of the nutrients may not be the only parameter that is at stake. In bioleaching systems, dissolved oxygen is known to be a key factor to obtain high performances. It was demonstrated that below a certain value of dissolved oxygen (approximately 2 mg/L), even though dissolved oxygen was available, the bioleaching kinetics were negatively affected (Morin and d'Hugues, 2007). A similar phenomenon may appear for the nutrients. This phenomenon may be due to mass transfer limitations or complexation.

The bioreduction kinetics obtained at 35°C in our study are in the same order of magnitude than the ones of Kucera et al. (2012) (26.1 mg·L⁻¹·h⁻¹), Pronk et al. (1992) (25.6 mg·L⁻¹·h⁻¹), and Breuker and Schippers (2024) (19.9 mg·L⁻¹·h⁻¹) but was measured in the presence of Sb. thermotolerans as the main species found after the bioreduction. This species previously demonstrated its ability to reduce Fe(III) (Malik and Hedrich, 2022), but there were, to our knowledge, no data on the kinetics obtained. It should be noticed that no microbial growth was observed in the liquid phase in all experiments at 35°C under anaerobic conditions, despite the fact that Sulfobacillus species were reported to grow under anaerobic conditions.

A linear correlation was observed between biomass concentration in the liquid phase and bioreduction rates, whether in batch or continuous mode. Even if a consequent proportion of microorganisms may be attached to sulfur particles, the determination of biomass concentration in the liquid phase by cell count is still a reliable indicator of the biomass available for Fe(III) bioreduction, both in batch and continuous mode. This result could also be improved by considering the biomass attached to sulfur particles, in particular in the case of higher sulfur content than in our study (1% or 2.4% (w/w)) which should increase the proportion of attached bacteria. This may be achieved by collecting solid particles and detaching bacteria before cell counts. The microbial DNA could also be extracted from planktonic bacteria and from the ones attached to the solids after their separation from the liquid phase, and gene abundance of total bacteria, determined by quantitative PCR, used as a marker of cell abundance. When optimized, these approaches have given comparable results to evaluate biomass of bacteria in batch bioleaching reactors (Hedrich et al., 2016).

At. caldus was detected in large quantities after sulfur oxidation. Since it is not known to reduce Fe(III), its presence might have provided organic carbons in the bioreduction reactor. Both Sulfobacillus and Ferroplasma genera detected in this study scavenge carbon from the excretion of chemolithotrophs or dead cell remains. This might also be the case for Alicyclobacillus which was found after sulfur oxidation but was a minor strain during bioreduction despite its ability to develop and reduce Fe(III). Like some other primary producers, At. caldus produces glyconic acid as an exudate, an organic compound that can be metabolized by Sulfobacillus species, including Sb. thermosulfidooxidans, as a specific metabolic trait that would give Sulfobacilli a competitive advantage in bioleaching by relation to other mixotrophs (Ñancucheo and Johnson, 2010; Justice et al., 2014).

Complementary molecular methods based on the widely used 16S rRNA gene marker have been applied to monitor community diversity dynamics in this study. They gave comparable results and evidenced Sulfobacillus strains as the iron reducers in bioreduction conditions. tRFLP fingerprinting has the advantage of being fast and easy to implement for a daily monitoring and is still acknowledged as powerful for profiling microbial community dynamics in relation to varying conditions in processes (De Vrieze et al., 2018). This is particularly true in bioleaching, because on the one hand such a selective environment leads to reduced microbial diversity and on the other hand the diversity of bioleaching strains is well documented. However, since only terminal restriction fragments are considered, a disadvantage is that two distinct sequences carried by different strains and sharing a terminal restriction site will not be distinguished. This happened here for both Sulfobacillus species present in the reactors and which have the same tRF. Metabarcoding of the 16S rRNA gene provided a more detailed description of the community composition and notably highlighted the presence of two species from the genus Sulfobacillus. This approach is more difficult to implement on a daily base but gives the needed knowledge on community diversity and the taxa to follow. By crossing the results, metabarcoding can be used to further improve tRFLP, e.g., by applying multiple restriction enzymes to better distinguish two species of a same genus, so that the main identified taxa can be detected and their evolution monitored. The two methods are complementary, and community monitoring in bioleaching will be improved if they are more generally combined.

Reductive bioleaching is one of the metallurgical options that are currently studied to extract Co from limonite. To design the bioprocess, Fe(III) bioreduction rates were determined in this study in the presence of different microbial consortia at different temperatures. In STR run in batch mode with a consortium cultivated at 35°C, a maximal bioreduction rate of 12.2 mg·L⁻¹·h⁻¹ and a mean value of 9.8 mg·L⁻¹·h⁻¹ were obtained. It was demonstrated that Fe(III) bioreduction rate was proportional to the biomass concentration in the liquid phase. Almost no evolution of the biomass in the liquid phase was noticed during bioreduction experiments. In continuous mode, the bioreduction rate was higher than the one measured in batch mode. It was estimated approximately 9 mg·L⁻¹·h⁻¹ with ABS-TE growth medium and increased up to 19.5 mg·L⁻¹·h⁻¹ with 0Km medium (richer medium). Even if nutrients were in excess in both growth media, the increase of nutrient concentrations sustained a larger biomass concentration in the liquid phase after sulfur oxidation. The microbial community structure switched from a bacterial community essentially composed of At. caldus after microbial growth on sulfur, to a community in which Sulfobacillus was largely dominant at the end of the bioreduction.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/genbank/, PRJEB71394.

AH: Conceptualization, Investigation, Methodology, Supervision, Visualization, Writing – original draft. CJ: Methodology, Validation, Visualization, Writing – original draft. HT: Investigation, Writing – review & editing. DP-H: Investigation, Methodology, Validation, Writing – review & editing. CB: Investigation, Writing – review & editing. A-GG: Conceptualization, Project administration, Supervision, Validation, Writing – review & editing.