Monazite transformation into Ce- and La-containing oxalates by Aspergillus niger.

Monazite is a naturally-occurring lanthanide (Ln) phosphate mineral [Lnx (PO4 )y ] and is the main industrial source of the rare earth elements (REE), cerium and lanthanum. Endeavours to ensure the security of supply of elements critical to modern technologies view bioprocessing as a promising alternative or adjunct to new methods of element recovery. However, relatively little is known about microbial interactions with REE. Fungi are important geoactive agents in the terrestrial environment and well known for properties of mineral transformations, particularly phosphate solubilization. Accordingly, this research examined the capability of a ubiquitous geoactive soil fungus, Aspergillus niger, to affect the mobility of REE in monazite and identify possible mechanisms for biorecovery. It was found that A. niger could grow in the presence of monazite and mediated the formation of secondary Ce and La-containing biominerals with distinct morphologies including thin sheets, orthorhombic tablets, acicular needles, and rosette aggregates which were identified as cerium oxalate decahydrate (Ce2 (C2 O4 )3 ·10H2 O) and lanthanum oxalate decahydrate (La2 (C2 O4 )3 ·10H2 O). In order to identify a means for biorecovery of REE via oxalate precipitation the bioleaching and bioprecipitation potential of biomass-free spent culture supernatants was investigated. Although such indirect bioleaching of REE was low from the monazite with maximal lanthanide release reaching >40 mg L-1 , leached REE were efficiently precipitated as Ce and La oxalates of high purity, and did not contain Nd, Pr and Ba, present in the original monazite. Geochemical modelling of the speciation of oxalates and phosphates in the reaction system confirmed that pure Ln oxalates can be formed under a wide range of chemical conditions. These findings provide fundamental knowledge about the interactions with and biotransformation of REE present in a natural mineral resource, and indicate the potential of oxalate bioprecipitation as a means for efficient biorecovery of REE from solution. This article is protected by copyright. All rights reserved.


Introduction
Monazite is a group of monoclinic phosphates mainly comprising rare earth elements (REE), e.g. cerium, lanthanum and neodymium, and has long been regarded as a strategic resource coveted by the World's great powers such as China, Europe, USA and Japan because of its critical involvement in high-technology sectors including consumer electronics, clean energy, hybrid electric vehicles and weapons systems (Humphries, 2012;Massari and Ruberti, 2013;Goodenough et al., 2016). More than 70% of the global supply of monazite is restricted to only a few rich mine deposits and the content of Ce in commercial monazite concentrate ranges from 42.7% (Guangdong, China) to 51.0% (Mount Weld, Australia) while that of La ranges from 17.5% (Green Cove Springs, USA) to 26.0% (Mount Weld, Australia) of the total REE present (Massari and Ruberti, 2013;Moss et al., 2013;Kumari et al., 2015).
Monazite concentrate is usually separated from monazite-bearing ores by crushing and flotation or through gravity and electrostatic separation (Moustafa and Abdelfattah, 2010;Khanchi et al., 2014;Chelgani et al., 2015). Since monazite occurs as lanthanidephosphate minerals of general formula Ln x (PO 4 ) y , it was thought to have limited interactions in terrestrial environments due to its high chemical and thermal stability and extremely low solubility: solubility products for CePO 4 and LaPO 4 are K sp = 1 × 10 −23 and K sp = 3.7 × 10 −23 respectively (Zhenghua et al., 2001). Traditional methods for recovering REE from concentrate include direct leaching using strong acids, e.g. sulphuric, nitric, and hydrochloric acid or strong alkalis, e.g. sodium hydroxide (Panda et al., 2014;Kumari et al., 2015): The resulting products can be recovered using precipitation agents before they are finally thermally decomposed to oxides. Although traditional leaching approaches are efficient, the use of strong acids and alkalis can be environmentally hazardous. In recent years, sustainable bioprocessing alternatives have been proposed, most focusing on element bioleaching and biorecovery from REE-bearing minerals by applying microbial systems (Zhuang et al., 2015;Liang and Gadd, 2017). Such biohydrometallurgical approaches are based on the biological mobilization of REE from solid-state materials including ore concentrate, electronic waste, mining residues and other metal-containing substrates with most attention being devoted to bacterial systems because of their well-known efficiency and established exploitation in bioleaching of metals such as copper and cobalt (Rawlings et al., 2003;Barmettler et al., 2016;Ng et al., 2016). However, in some previous research, citric acid-overproducing Aspergillus ficuum was used to treat Egyptian monazite (Th-U) which resulted in high bioleaching efficiencies of 75.4% and 63.8% under pulp densities of 0.6% and 1.2% (w/v) respectively (Hassanien et al., 2013). Other work showed that biomass-free spent medium of A. terreus ML3-1 and a Paecilomyces sp. WE3-F were 1.7-3.8 times more efficient in leaching REE from monazite than the amounts recovered using abiotic controls (Brisson et al., 2016). It has been established that one of the most important mechanisms underlying fungal bioleaching is the excretion of low-molecular-weight organic acids (Deng et al., 2013;Reed et al., 2016).
Aspergillus niger is a ubiquitous environmental fungus known for many industrial applications including citric acid production (Jernejc et al., 1982). It also possesses significant biogeochemical properties that enable interactions with metals and minerals such as mineral dissolution and biologically-induced mineralization (BIM), both of which can be mediated by the extracellular excretion of organic acids (Gadd, 2010;Li et al., 2014;Yang et al., 2019a;Yang et al., 2019b). Aspergillus niger is an efficient producer of oxalic acid (H 2 C 2 O 4 ), which can play a critical role in the BIM process for the bioprecipitation of a number of metals from soluble and insoluble sources . For example, A. niger can transform Co 3 (PO 4 ) 2 , Zn 3 (PO 4 ) 2 and ZnO into corresponding insoluble metal oxalates via an intermediate solubilization phase (Sayer and Gadd, 1997). Other studies have shown that A. niger was able to solubilize and precipitate calcium from natural gypsum (CaSO 4 Á2H 2 O) as calcium oxalate (Gharieb et al., 1998). This fungal species is also capable of transforming more complex minerals such as malachite (Cu 2 (CO 3 )(OH) 2 ) into copper oxalate (Fomina et al., 2017). Another example showed that lead oxalate was precipitated by A. niger in Pb 2+ -containing media when supplemented with organic phosphate as the P source (Liang et al., 2016). Aspergillus niger has been shown to be an effective agent for the recovery of lanthanides including Ce and La from solution (Zhuang et al., 2015). Kang et al. (2019) described a novel method of recovering lanthanum as lanthanum oxalate from aqueous solution using biomass-free culture supernatants after the growth of A. niger. In general, it is assumed that biogenic oxalate can interact with phosphate through the following mechanisms (Chi and Xu, 1999;Furrow et al., 2012): ] = $ 10 −26 , is lower than that of LnPO 4 ($10 −23 ), speciation of oxalate-containing secondary minerals during interaction of monazite-containing rare earth phosphates in an acidic fungal-induced environment could be a profitable approach for biorecovery and worthy of examination. Bioprocessing is currently viewed as a promising alternative or adjunct to new methods of element recovery to ensure the security of supply of valuable strategic elements but little is known about microbial interactions with REE. Since fungi are important geoactive agents in the terrestrial environment and well known for properties of mineral transformations, particularly phosphate solubilization, the objective of this research was to examine the capability of a ubiquitous geoactive soil fungus, Aspergillus niger, to affect the mobility of REE in monazite with a view to identifying possible mechanisms for biorecovery. The results contribute a new understanding of the interactions between geoactive fungi and REEcontaining minerals with potential practical applications.

Fungal growth on solid medium
It was found that the presence of monazite had an influence on the growth of A. niger over 2 weeks of incubation. The surface pH of the monazite-containing agar plates ranged from pH 3.55 to 3.74, which was significantly higher (P < 0.05) than the monazite-free control, which was at pH 2.75 (Fig. 1A). Fungal biomass yield, however, showed a significant declining trend as the monazite concentration increased. In particular, plates with 0.5% and 1.0% added monazite yielded 30% and 39% less biomass respectively, compared with the untreated controls (Fig. 1B). In plates supplemented with 0.5% monazite, a significant (P < 0.05) rise of surface pH was observed after 2 weeks of incubation compared with the first week (pH 2.94). The surface pH was maintained around pH 4.3 for the following 3 weeks (Fig. 1C). The biomass yield on these plates showed a general increasing trend as the incubation time increased (Fig. 1D).

REE concentrations and pH changes in liquid media
Changes in total REE concentration and pH were recorded over the course of a 5-week reaction of 2% monazite with spent Modified Czapek-Dox (MCD) culture medium (Fig. 2). The amount of REE leached out was low (7.3 mg L −1 ) over the first week but dramatically increased to 39.9 mg L −1 by the second week and reached the maximal value of 43.1 mg L −1 by the third week. However, the fourth and fifth weeks saw a significant reduction of 53.5% and 58.7% respectively, compared with the maximum value. The overall pH followed a slow upward trend over the whole length of the liquid interaction, reaching a maximum (pH 2.92) by the fourth Changes of (C) surface pH and (D) biomass yield of A. niger following 5 weeks growth on MCD plates supplemented with 0.5% (w/v) monazite under the same conditions. Data are averages of at least three replicates and error bars show the standard error of the mean. Different lowercase letters between treatments indicate differences are significant at P < 0.05 based on Tukey's test. Total rare earth concentration and pH of biomass-free spent liquid MCD culture medium supplemented with 2% (w/v) monazite and maintained on a roller mixer at 25 C in the dark. MCD medium was inoculated with A. niger and incubated at 25 C in the dark for 14 days prior to harvesting of the biomass-free culture supernatant.
, total REE; , pH. Data are averages of at least three replicates: error bars representing the standard error of the mean are smaller than the symbol dimensions. week and this only slightly decreased in the last week of incubation.

SEM and EDXA of produced biominerals
After growth of A. niger on 0.5% MCD agar plates, the formation of secondary biominerals was observed using scanning electron microscopy (SEM), which revealed distinctive morphological features varying in shapes and sizes at different incubation times. Large single crystals measuring $200 μm in length and of a thin lamellar structure were found to be the only morphotype of the secondary minerals after incubation of A. niger for 1 week ( Fig. 3A and B). After 2 weeks of incubation, the predominant biominerals were square-shaped tabular plates that grew out of the grains of the substrate ( Fig. 3C and D). These structures ranged from $20 to 60 μm in width but were much thicker than those crystals produced in the first week samples. Well-formed acicular structures consisting of radiating needles were discovered after 3 weeks of incubation. These needle-like crystals had disphenoid tips and formed large clusters measuring $100 μm across ( Fig. 3E and F). Acicular crystals were still the major structure at the fourth week of incubation. However, the needles tended to become broader and shorter and the clusters appeared to be in a less organized fashion ( Fig. 3G and H). After 5 weeks of incubation, there was a complete change in the crystal habit of the biominerals, which were all of a rosette structure composed of thin layers and forming large aggregates ( Fig. 3I and J). However, morphological changes for the samples that were mixed with the biomass-free spent MCD liquid culture medium were not so varied and secondary minerals with distinct features were observed only in the first and second weeks. Diamond-shaped tabular structures were most abundant in 1-week samples and these formed large aggregates ($200 μm) by stacking together ( Fig. 4A and B). The second week saw a change in these aggregates into a radiating arrangement ( Fig. 4C and D). After 5 weeks of mixing, the secondary minerals further changed into smaller single crystals with monoclinic features (see arrows in Fig. 4E and F).
The energy-dispersive X-ray analysis (EDXA) spectra showed a difference in REE composition for these secondary minerals compared with the untreated monazite ( Fig. 5B-D). Ce and La were found to be the only REE in the secondary minerals at every growth stage. The presence of Nd and F, which were present in low amounts in the untreated monazite, was not detected in the newly formed biominerals. The 1-, 2-and 5-week samples all had a high O:C ratio and a similar relative abundance of Ce-La, which was similar to the untreated samples ( Fig. 5B and D). However, the 3and 4-week samples showed a low O:C ratio and large peaks for Ce and La (Fig. 5C). EDXA of the secondary minerals from the spent MCD liquid culture reaction showed a similar elemental composition to those resulting from the MCD plates except for larger peaks for Ce and La (Fig. 6).

XRD analysis of the secondary mycogenic biominerals
Biominerals of the first and final week's treatment were subjected to XRD analysis. It was revealed that the sample incubated for 1 week with A. niger on MCD agar shared similar mineral phases to the untreated control except for a generally lower peak intensity (Fig. 8b). In addition to the mineral phases in the substrate, weak peaks corresponding to Ce and La oxalates were observed in samples from the fifth week of incubation (Fig. 9b). In the interaction with spent liquid culture medium, the 1-week incubation sample also showed similar results to the control (Fig. 8a). Patterns linked to Ce and La oxalates of prominently higher intensity were observed in the 5-week sample (Fig. 9a).

Geochemical modelling
Solubility diagrams predicting the fate of lanthanide under three scenarios were constructed. The solubility constant chosen to input for lanthanide was K sp = 5.07 × 10 −26 , which was determined by measuring the solubility of chemically precipitated cerium oxalate under laboratory conditions. The amount of oxalic acid in the simulation system was set at 55 mM, an actual oxalate concentration found in the liquid MCD medium after 2 weeks incubation with A. niger in preliminary experiments (Kang and Gadd, unpublished). The system of log a[(C 2 O 4 ) 2− ] versus pH (Fig. 10A) showed the transformation of lanthanide as three insoluble minerals in the presence of different activities of oxalate over the full range of pH. The results showed that at pH 0, the minimum concentration of oxalate required to form Ln 2 (C 2 O 4 ) 3 was 10 -4.98 M. It was also revealed that Ln oxalate could be a possible mineral at a low pH and that higher amounts of oxalate would be required to precipitate Ln as the pH increased. Likewise, the system of log a[(PO 4 ) 3− ] versus pH (Fig. 10B) showed      oxalate, an increase of phosphate must be matched by an increasing amount of oxalate (Fig. 10C).

Discussion
The Gakara REE deposit where the monazite sample originated is located in the East Africa Rift, which, in recent years, has been well known for supplying the rare earth industry (Lehmann et al., 1994). Geochronological data have shown that the REE veins in the area mainly consist of primary bastnaesite (REE-CO 3 F) and monazite as secondary occurring minerals dating back to 602 and 589 Ma ago respectively (Ntiharirizwa et al., 2018). The presence of REE-carbonates was also confirmed by XRD in our work, which showed small amounts of CeCO 3 F and LaCO 3 F. Our XRF results are in agreement with the elemental composition revealed in a survey of African monazite ores that found light REE in the following order of abundance: Ce > La > Nd > Pr (Harmer and Nex, 2016), and also showed that our monazite sample, as in other relevant studies on monazite, contains a slightly higher amount of Ce and La than other REE (Zhu and O'Nions, 1999;Galvin and Safarzadeh, 2018). Fungal interactions with monazite have hitherto not been reported. The presented results from incubation of A. niger on solid medium suggest that the growth of the organism was retarded by the presence of monazite as the biomass showed a continuous decline with increasing amounts of monazite. The increase of surface pH on monazite-containing plates can probably be explained by the use of NO 3 − as the nitrogen source, buffering effects of the monazite, and lower acid production because of reduced growth. The general upward trend of biomass yield throughout the 5-week continuous incubation at a 0.5% monazite concentration indicated that A. niger could still grow at a lower rate. After 2 weeks of incubation, the surface pH showed a significant increase, which was the same as the pH trend in the liquid medium experiments. These results are consistent with recent work where similar pH shifts were observed during the bioleaching of laterite using A. niger grown in liquid media (Yang et al., 2019a). It is known that external medium pH can sometimes increase during incubation of A. niger on mineral-containing solid media and especially when nitrate is supplied as the inorganic nitrogen source (Sayer and Gadd, 1997;Ceci et al., 2015a;Ceci et al., 2015b). This has been commonly observed for fungal growth on nitrate because nitrate uptake in fungi occurs by NO 3 − /H + symport leading to alkalization of the external environment, while many fungi can also excrete NH 4 + , derived from NO 3 − metabolism, which also elevates the external pH (Zhou et al., 2000;Galvan and Fernández, 2001;Takasaki et al., 2004;Kramer-Haimovich et al., 2006;Stief et al., 2014;Watkinson et al., 2015).
Examination of the novel mycogenic biominerals revealed that, at each stage of fungal growth, the morphologies of the REE-containing secondary biominerals varied considerably. In solid media, at least four distinctive features for the crystals were observed including thin lamellar sheets, tabular plates, radiating needle clusters and rosette aggregates. Two additional morphologies involving tabular aggregates and individual monoclinic bodies were observed when monazite was incubated with biomass-free spent culture medium. These welldeveloped monoclinic or orthorhombic crystalline morphologies were observed in previous studies, which used calcite marble, Ca-phosphate rock and gypsum as the substrate for the precipitation of calcium by oxalic acidproducing fungi (Gharieb et al., 1998;Schneider et al., 2010;Sturm et al., 2015). However, there may be a variety of formation types for a given metal oxalate. For example, mycogenic calcium oxalate crystals can be categorized into four groups, i.e. tetragonal bipyramids, prisms, tablets and needles (Arnott, 1995). A difference in the mineral substrate, though containing the same metal, can also influence the size and shape of the secondary minerals produced. Fomina et al. (2005a) revealed that mycogenic lead oxalate precipitated by Beauveria caledonica exhibited tetragonal spikes in the presence of lead phosphate while irregular spikes occurred with lead tetraoxide, and octahedral bodies formed when lead carbonate was present. Mycogenic oxalates can also assume the form of orthorhombic tablets (glushinskite -magnesium oxalate), acicular clusters (strontium oxalate hydrate), rosettes and spheres (whewellite -calcium oxalate monohydrate and moolooite -copper oxalate hydrate) (Gadd, 2007), which were similar observations to our results. The only previous studies of lanthanides, to which our results could be comparable, are few and generally involve the transformation of a single element rather than multiple elements in a natural mineral substrate (Li and Gadd, 2017;Kang et al., 2019). Chemically synthesized Ce-or La-oxalate invariably take the form of plate-like single crystals and their sizes can vary from 10 to 300 μm in length depending on the precipitation method and concentrations of reagents (Claparede et al., 2011;Maslennikov et al., 2017;Yu et al., 2017). A rare case of flower-like aggregates of tabular crystals was reported for cerium oxalate decahydrate [Ce 2 (C 2 O 4 ) 3 Á10H 2 O] which were abiotically precipitated using pure oxalic acid (Liu et al., 2013). Few previous studies have paid attention to the mycogenic transformation of lanthanide elements. Large tabularshaped crystals with a layered texture, which bore a resemblance to those in the present study, were formed as a result of the bioprecipitation of La oxalate in a solid medium containing lanthanum chloride (Kang et al., 2019). These crystals also showed prominent differences in morphology from the lanthanum oxalate that was precipitated using liquid spent culture media (Kang et al., 2019).
The secondary minerals obtained in this study contained only Ce and La as REE and the only difference was in the relative proportion to C and O. The absence of other metals, including Nd, Pr and Ba, indicated these secondary minerals may be of high purity. XRD showed that, in both solid and liquid media, A. niger was able to transform monazite into Ce-and La-oxalates. The only difference was in the relative yield, which was clearly higher for the spent liquid culture medium as indicated by the higher peaks in the XRD patterns. The formation of oxalates in the liquid medium was accompanied by a sharp decrease in the total REE concentration after 3 weeks (Fig. 2). The excretion of organic acids, e.g. citric and oxalic acids can play a significant role in the dissolution of insoluble minerals and can be preceded by the formation of soluble metal-ligand complexes before oxalate-mediated precipitation occurs (Sayer and Gadd, 1997). Despite the involvement of organic acids, the bioleaching ability remained at a low level, which was in agreement with another study which found that other phosphate-solubilizing microbes showed quite a low efficiency in leaching Ce and La from monazite ores ranging from only 0.005% to 0.13% recovery (Shin et al., 2015) and also that direct rather than indirect contact may be essential for more efficient fungal bioleaching of elements from ores (Yang et al., 2019a).
Geochemical simulation is a useful method to investigate the transformation mechanisms for minerals and has been widely applied to different scenarios regarding fungal-mineral interactions (Ceci et al., 2015b;Li et al., 2019). Since the monazite contained a larger amount of Ce than La, our modelling systems were based on the solubility product of Ce-oxalate, which was measured under laboratory conditions. We did not carry out geochemical simulation for speciation of La-oxalate, which will produce very similar results to Ce because of its similar physicochemical properties and K sp value. All three diagrams that were produced showed the possibility of oxalate transformation from phosphates. According to the modelling results, the possible mechanism through which Ln oxalates would be formed under these circumstances could be: Such a mechanism may be of importance to the biogeochemical cycling of REE because P-solubilizing fungi and bacteria are ubiquitous in soil environments (Jacobs et al., 2002;Fomina et al., 2004;Fomina et al., 2005b;Fomina et al., 2006). The solubilization of REE-bearing phosphate compounds also releases inorganic phosphate (P i ) and thus can improve soil fertility (Zhang et al., 2018). In a metallurgical process, REE-oxalate can be readily converted into REE-oxides through hightemperature treatment and therefore may also serve as a precursor for other useful REE materials (Kang et al., 2019). In conclusion, we have demonstrated the ability of A. niger to transform natural monazite into Ce-and Lacontaining crystalline oxalates [Ln x (C 2 O 4 2− ) y ÁzH 2 O] of distinct morphologies in solid media and also after interaction with biomass-free spent liquid culture medium, the latter method allowing Ln biorecovery in high purity. Our findings provide new knowledge on the interactions between REE and fungi, which also have potential application for the biorecovery of these strategically important elements from REE-containing solutions, process streams and leachates.

Microorganism, media and mineral
The microorganism used in our study was a wild-type strain of A. niger (ATCC 1015), which was routinely maintained on malt extract agar (MEA) (Lab M Limited, Bury, UK) in the dark at 25 C. Modified Czapek-Dox liquid medium (MCD) consisted of (l −1 Milli-Q water): sucrose 30 g, NaNO 3 2 g, Na 2 HPO 4 1 g, MgSO 4 Á7H 2 O 0.5 g, KCl 0.5 g and FeSO 4 Á7H 2 O 0.01 g. The final pH was adjusted to pH 5.5 with sterile 1 M HCl prior to autoclaving for 15 min at 115 C. For the preparation of solid media, 15 g of Agar No. 1 (Lab M Limited, Bury, UK) was added to 1 L of the liquid MCD medium. The medium was then autoclaved at 115 C for 15 min. The monazite concentrate, which was originally procured from the Gakara deposit, Burundi, East Africa, was kindly provided by Rainbow Rare Earths Limited (London, UK). The monazite was pulverized using a mortar and pestle, sieved through a 90-μm mesh and autoclaved for 15 min at 115 C before use.

Solid media experiments
Monazite-containing agar plates were made by incorporating 0.25%, 0.50%, and 1.0% (w/v) sterilized monazite powder into molten MCD agar medium when cooled to around 60 C. Cellophane membranes, which were used to separate mycelia from the agar, were prepared by autoclaving in Milli-Q water at 115 C for 15 min followed by washing three times in sterilized Milli-Q water. A sterile cellophane membrane (90 mm diameter) was placed on top of the agar surface and a plug (0.5 cm diameter), cut from a freshly grown A. niger colony using a sterile cork borer, was inoculated onto the centre of the plate. The plates were incubated in the dark at 25 C for a total period of 5 weeks. Biomass was harvested and surface pH measured at weekly intervals at five points across the diameter of the agar plate using an Orion 3 Star benchtop pH meter (Thermo Fisher Scientific, Loughborough, UK) equipped with a flat-tip electrode (VWR International, Lutterworth, England, UK). Minerals were collected by gently homogenizing the agar in Milli-Q water at 80 C in a crystallizing dish after settling and washed at least three times with Milli-Q water. All solid media experiments were carried out at least in triplicate.

Liquid media experiments
Spore suspensions of A. niger were prepared by washing off spores from fully grown colonies on MEA using 0.1% (v/v) sterile TWEEN ® 80 (Sigma-Aldrich, St Louis, MO, USA) and filtering through a sterile muslin cloth. Aspergillus niger spores were initially inoculated in 200 ml liquid MCD medium at 1 × 10 6 spores ml −1 in a 250 ml Erlenmeyer flask and shaken at 125 rpm for 2 weeks at 25 C in the dark. After this time, biomass-free spent medium was collected by vacuum filtration through 0.45 μm pore diameter cellulose acetate membrane filters (Whatman, Maidstone, UK). The reaction system contained 1.0% or 2.0% (w/v) monazite sand, which was achieved by adding sterilized monazite to 50 ml biomass-free medium in a 50 ml centrifuge tube. The tubes were mixed on a roller mixer for 1, 2, 3, 4 and 5 weeks at 25 C in the dark. Minerals were collected by centrifugation at 2553g for 30 min and washed with Milli-Q water at least three times. The supernatant pH was measured using the above-mentioned equipment. Measurements for total rare earth concentration in the liquid system after reaction were carried out using the Arsenazo III colorimetric method (Hogendoorn et al., 2018). This was achieved by mixing 1 ml of an appropriately diluted sample with 1 ml 0.02% (w/v) Arsenazo III solution (Sigma-Aldrich) and 8 ml pH 2.8 potassium hydrogen phthalate buffer. After the chromogenic reaction, the OD 658 nm of the mixture was measured using an Ultrospec 2100 pro spectrophotometer (Biochrom, Holliston, MA, USA).

EDXA, SEM, XRD and XRF
The elemental composition of the mineral samples was determined using EDXA. SEM was used to examine morphological features. XRD was used for determining mineral phases, and XRF was used for the measurement of elemental composition. After collection, the mineral samples were dried in a desiccator for at least 1 week and mounted on adhesive carbon tape on 25 × 5 mm 2 electron microscopy aluminium stubs (Agar Scientific, Essex, UK) before being examined using EDXA (Oxford Inca, Abingdon, Oxon, UK) operating at an accelerating voltage of 15 kV for 100 s. For SEM, samples on the stub were coated with a layer of 10 nm gold and platinum using a Cressington 208HR sputter coater (Ted Pella, Redding, CA, USA) prior to examination using a field emission scanning electron microscope (Jeol JSM7400F) operating at an accelerating voltage of 5 kV. Mineral phases were determined using a Hiltonbrooks X-ray diffractometer (XRD) (HiltonBrooks, Crewe, UK) furnished with a monochromatic CuKα source and curved graphite, single-crystal chronometer (30 mA, 40 kV). To prepare for XRD, samples were ground to a fine powder using a ceramic mortar and pestle and compacted tightly on the reverse side of an aluminium specimen holder (15 × 20 × 2 mm 3 ) held against a glass slide. The back cover was then snapped into place and the glass slide was removed from the holder. Duplicate samples were analysed over the range 3-60 2θ at a scan rate of 1 min −1 in 0.1 increments.
To determine the elemental composition of the minerals, XRF spectroscopy was carried out using a Philips PW2424 sequential spectrometer with an RhKα source and calibrated with certified standard materials. The samples were placed in a 32-mm diameter pellet mould and were compacted under loads of 75 kN for 5 min and 150 kN for a further 10 min and then transferred to a specimen cup that had a 27-mm-diameter viewing aperture. The results are expressed as oxides.

Geochemical modelling
Geochemist's Workbench software, version 12.0.4 (Aqueous Solutions LLC, Champaign, USA) was used for simulating the geochemical fate of monazite in the fungal-mediated reaction systems. Three subprograms included in the software, i.e. Act2, SpecE8 and TEdit were also used for this purpose (Ceci et al., 2015a;Kang et al., 2019). To model the system in solid medium experiments, which contain a known amount of monazite (LnPO 4 ), the pH and concentration of oxalate were set as variables and a diagram of pH versus oxalate was created. To simulate the biomass-free supernatant reactions with monazite, in which pH and PO 4 3− were assumed to be the variables, a diagram of pH versus phosphate was constructed. In the case of a fixed pH in a solid medium, both PO 4 3− and oxalate were set as variables and a diagram of PO 4 3− versus oxalate was created. In these three scenarios, the molarities of all ions and anions were entered into the subprogram SpecE8 (the Visual MINTEQ database) as their actual concentrations in the MCD medium and the solubility diagrams were drawn using Act2. The activity of all the chemical species at 25 C and 1.013 bar was calculated using SpecE8 based on thermodynamic equilibria with the relationship between activity and the concentration of aqueous species being expressed as: where a is the activity, c is the actual concentration and γ is the activity coefficient (c ! 0, γ = 1.0 at infinite dilution). In the simulation systems, the activity of monazite (LnPO 4 ) was set as 1.0 because of its very low solubility. Since the database does not include the solubility product of hydrated Ce oxalate, a calculation was carried out according to the actual molarity of Ce ions in a saturated solution. For the determination of Ce 3+ activity at equilibrium, Ce oxalate was completely precipitated by adding an excessive amount of oxalic acid into a 5 mM CeCl 3 solution. The resulting precipitate was washed multiple times with Milli-Q water and maintained on a roller mixer for 3 days until equilibrium was achieved. The actual concentration of Ce 3+ in the supernatant of the saturated solution was measured using the Arsenazo III method at 25 C (Hogendoorn et al., 2018). The calculation was based on the equilibrium equation 6 and the following formula: where a(Ce 3+ ) and a(C 2 O 4 2− ) represent the activities for Ce 3+ and C 2 O 4 2− according to Eq. 7. The K sp of Ce 2 (C 2 O 4 ) 3 ÁxH 2 O at 25 C was manually added into the database using the subprogram TEdit.

Statistical analysis
Data regarding biomass yield, REE concentration and pH were subjected to statistical analysis using IBM SPSS Statistics 22.0. Tukey's HSD post hoc test in one-way ANOVA was applied to compare the significance of difference at P < 0.05 between treatments. All data were from at least three replicates for each treatment.