Metal Exchangeability in the REE ‐ Enriched Biogenic Mn Oxide Birnessite from Ytterby, Sweden

: A black substance exuding from fractures was observed in 2012 in Ytterby mine, Sweden, and identified in 2017 as birnessite with the composition M x [Mn(III,IV)] 2 O 4 ∙ (H 2 O) n . M is usually cal ‐ cium and sodium, with x around 0.5. The Ytterby birnessite is unique, with M being calcium, mag ‐ nesium, and also rare earth elements (REEs) constituting up to 2% of the total metal content. The biogenic origin of the birnessite was established in 2018. Analysis of the microbial processes leading to the birnessite formation and the REE enrichment has continued since then. The process is fast and dynamic, as indicated by the depletion of manganese and of REE and other metals in the fracture water during the passage over the precipitation zone in the mine tunnel. Studies of the exchangea ‐ bility of metals in the structure are the main objective of the present program. Exposure to solutions of sodium, calcium, lanthanum, and iron led to exchanges and altered distribution of the metals in the birnessite, however, generating phases with almost identical structures after the exchanges, and no new mineral phases were detected. Exchangeability was more efficient for trivalent elements (REE) over divalent (calcium) and monovalent (sodium) elements of a similar size (ionic radii 90– 100 pm)


Introduction
The Ytterby mine in Sweden is known for the discovery of five rare earth elements (REEs), as well as scandium, yttrium, and tantalum. The mine is located on Resarö Island in the Baltic Sea, some 20 km NE of Stockholm. The site, which is described in detail by Sjöberg et al. [1] (geologic setting, hydrology), belongs to the Proterozoic Svecofennian domain, which covers most of the northern and central part of Sweden. The mine is located in a pegmatite body, which is bordered by amphibolite in the NW and by gneiss in the SE. The Ytterby pegmatite belongs to the NYF family, which is enriched in niobium, yttrium, and fluorine. The average hydraulic conductivity in the bedrock around the mine has been estimated to be around 3 × 10 6 m/sec. That would correspond to an inflow of water to the mine shaft of some 9 m 3 /day. Some 50 minerals have been identified in the mine, including 15 containing REE [2,3].
The mine was closed in 1933, after 170 years of mining for quartz and feldspar. The mine shaft was used as a fuel deposit from the 1950s until 1995. During this period, the shaft was covered, and a tunnel was opened through the bedrock into the shaft.
In 2012, a black substance was discovered in the opening of a water-bearing fracture in the tunnel wall, Figures 1 and 2, and a program for identification started. The composition of the substance (here denoted YBS) was determined from elemental analysis, as well as XRD and SEM with EDS. The YBS was identified as a manganese oxide of the birnessite type, ideally M0.5[Mn(III,IV)]2O4•(H2O)n, where M usually is sodium or calcium (mindat, 2019) [4]. Birnessite was discovered and defined as a new mineral phase in 1956 [5], and there are numerous reported studies of natural birnessite [6][7][8][9][10][11][12][13][14][15][16], as well as of synthetic birnessite, since then, in most cases on the adsorption properties and structure . There are also a few reported observations of birnessite that has precipitated on rock walls in tunnels and caves, similar to the Ytterby birnessite [39][40][41].  What makes the Ytterby birnessite unique is the enrichment of yttrium and REE, corresponding to up to 2% of the total metal content and replacing primarily sodium in the structure. The elemental composition and phase analysis are described in Sjöberg et al. (2017) [1]. Analysis of the organics, including IR and EPR spectroscopy, indicated that there was ca. 1% organic matter (lipids) associated with the YBS, and a biogenic origin was assumed. The microbial community inhabiting the YBS was characterized, known manganese oxidizers were identified, and the biogenic origin of the YBS was analyzed and verified [42][43][44][45]. The results revealed a startling bacterial diversity, and a model for microbe-mediated oxidation of manganese and the formation of birnessite was proposed [46]. The model was based on detailed studies and characterization of the microbial community that apparently is active in the process.
The objective of the present phase of the project is further studies of the processes leading to formation of the REE-enriched birnessite in the Ytterby mine tunnel. The exchangeability of the metals corresponding to M in the YBS birnessite phase (sodium, calcium, REE, etc.) is studied in particular.

Sampling of YBS and Fracture Water
The solid precipitate (YBS) was collected in the opening of a major water-bearing fracture in the wall of the tunnel leading into the mine shaft. The YBS sample was dried at ca. 30 °C until constant weight (dry weight), followed by washing with deionized water (18.2 MΩ) and then drying once more.
Water that was trickling out from the fracture was sampled above and below the YBS precipitation zone (cf. Figures 1 and 2) on the same day as the sampling of the precipitate and was filtered onsite (PP membrane filters, 0.20 mm pore diameter) and stored in polyethene bottles. Temperature, electric conductivity, and pH were measured in the field. Samples for element analysis (ICP-MS) were acidified (HNO3) and stored in a fridge until analysis.

Environmental Scanning Electron Microscopy (ESEM-EDS)
Scanning electron microscopy was carried out using a Quanta environmental scanning electron microscope (ESEM). An energy-dispersive X-ray spectrometer (EDS) was used for compositional information.

Sequential Washing
A sequential washing procedure was designed for separation and removal of organics and ion exchangeable and adsorbed ions, as well as fresh hydroxides and oxyhydroxides and any suspended mineral grains from the original YBS sample (denoted YBSNat). The three steps in sequence (L/S = 10, 24 h each step C1 and C2; 1 h step C3) are: The slurry after step C3 (water with the YBS grains, as well as suspended fine particles and colloidal matter) was left to settle for 1 min, and then the water phase with suspended material was discharged. The remaining solid was dried at 30 °C until constant weight.

Element Exchange
The washed YBSNat sample (denoted YBSRed) was divided into four fractions that were mixed with salt solutions and left for 48 h before separation (centrifugation), followed by washing and drying. The solutions were (1 mol/L) NaNO3, Ca(NO3)2, and La(NO3)3, as well as Fe(NO3)3 for a sample already exposed to Ca(NO3)2. All of the YBS samples were washed 2 times with deionized water after the exposure, prior to the characterization of the element composition (see below) and phases (XRD).

X-Ray Diffractometry
Mineral phases were determined by XRD (PANalytical X'Pert-Pro diffractometer, Malvern, Bristol, UK, equipped with X'celerator silicon strip detector and spinning sample holder; CuKα radiation from 5° to 70° 2θ with a step size of 0010° 2θ at 40 mA and 45 kV). The diffraction patterns were analyzed using PANalytical High Score Plus software connected to the Inorganic Crystal Structure Database (CSD), 2017 Version.

IR Spectroscopy
Infrared spectra were recorded on solid YBS samples (PerkinElmer Fourier-Transform Infrared (FT-IR) Spectrum Two, Perkin Elmer, Fairfield, US). The spectra were analyzed using the PerkinElmer Infrared Spectrum software that was provided with the instrument.

Analysis of Water
Elements were analyzed (fracture waters, washing solutions) by ICP-MS (Agilent 7500), following an established standard procedure (choice of internal standard, reference solutions, etc. [1]); anions by ion chromatography (Metrohm); and carbonate/bicarbonate by titration of total alkalinity (also dissolved organic carbon by DOC-analyzer; Schimadzu TOC-V CPH), following standard procedures [1].

Environmental Scanning Electron Microscopy (ESEM-EDS)
ESEM analyses of YBS have previously been presented and discussed [1]. Only one example is presented here-on the composition of the YBS precipitate (YBSNat) prior to washing and element exchange ( Figure 3). The sample was fine-grained and evidently contained suspended crystalline mineral fractions (silicates), but also amorphous metal phases and organics (lipids), according to the previous studies [1].

Fracture Water Chemistry and YBS Metal Enrichment
The composition of the YBS precipitate (YBSNat) and the fracture water above (AqA) and below (AqB) the precipitation zone are given in Table 1, as well as the concentration ratio of the fracture waters ([AqB]/[AqA]) and an enrichment factor ke, defined as [YBSNat]/[AqA]. Concentration levels of the major components were slightly lower in AqB than in AqA, except for calcium, carbonate, and sulphate with higher levels in AqB. This may reflect some interaction with the rock during the passage over the precipitation zone, where the calcium and carbonate levels may reach saturation, indicating the presence of solid calcium carbonate, considering the CaCO3(s) solubility product 3.3 × 10 −9 [47]. The enrichment factor ke demonstrates the efficient removal of manganese from the water phase, indicating high affinity for incorporation in the precipitating YBS, and also for iron (similar charge and size as for manganese), as well as the light REEs (LREEs). Lower enrichment factors are observed for copper, zinc, barium, vanadium, and yttrium, as well as the heavy REEs (HREEs). The behavior of yttrium is similar to the heaviest of HREEs, as expected for charge-and radius-controlled behavior (CHARAC) [48]. Zinc, copper, and vanadium, however, exhibit higher enrichment than expected, considering charge and size. The enrichment of REE is highest for cerium and decreasing with increasing atomic number. The strong enrichment of cerium may be due to oxidation of Ce(III) to Ce(IV). Oxidation of Ce(III) by Mn(IV) has previously been reported, and the oxidation processes have been discussed with examples from other Mn-oxide systems [27,29,[48][49][50][51][52][53][54].
The differences in concentrations of the trace elements between AqA and AqB (sampled within 1 h on the same day) indicate a progressive formation of the biogenic YBS phase. This is illustrated by the distribution of the elements in the water phase, Figure 4, showing the depletion as a function of ionic radii, using data from Table 1. This is evidently a dynamic process leading to changes in concentrations in a time span of hours, or less. The depletion is increasing with increasing radii for the REE. There is no evident depletion for aluminum and strontium among the trace elements. The composition of YBS species is given in Table 2: original YBS before washing (YBSNat); after sequential washing (YBSRed); and after exposure to sodium, calcium, lanthanum, and iron solutions, respectively (YBSNa, YBSCa, YBSLa, and YBSFe).  Gd  419  419  394  377  291  137  Pr  418  403  383  363  220  122  Sm  375  369  347  329  239  121  Dy  308  335  329  303  230  103  Er  150  171  162  155  118  51  Yb  99  116  110  106  90  38  Ho  60  65  61  58  45  20  Tb  56  65  60  60  42  19  Eu  31  31  30  28  20  10  Tm  18  21  20  19  16  7  Lu  15  17  16  16  13  6 The sequential leaching (steps C1-C3) is assumed to have removed the organic fraction, as well as exchangeable ions not needed for the charge balance, and most of the suspended and colloidal silicates. Concentrations of most of the elements in YBSRed are higher than in YBSNat since a significant fraction of the YBS mass has been removed. Concentrations of calcium and magnesium are lower, which may indicate losses during washing, possibly of carbonate species. The compositions of the modified YBS species, assuming a birnessite structure, Mx[Mn(III,IV)]2O4•(H2O)n, are given in Table 3. The stoichiometries of YBSNat and YBSRed are similar, despite the fact that a significant fraction of the total mass, including minerals not associated with the birnessite phase, haves been removed from YBSNat in the sequential washing process. The dominating metal (M) is calcium, followed by magnesium and REE + yttrium.   (Table 3). For Ce, expected to be tetravalent, the position in the diagram may be shifted corresponding to coordination number related to ionic radii other than the assumed r(6).
Exposure to sodium, calcium, lanthanum, and iron solutions led to corresponding changes of the proportions of cations (M), which are summarized in Table 4. Calcium is still the dominating exchangeable cation after exposure to 1 mol/L sodium solution, but calcium is partly exchanged by lanthanum. Iron is deviating from the others, which, in principle, could indicate the formation of a new phase. Fractions of the metals M in the assumed birnessite structure may be merely adsorbed on outer surfaces of the oxide phase (besides the dissolution of carbonates and hydroxides present in the natural YBS). However, exchange capacities below 10 meq/kg were reported for a selection of oxides (pyrolusite, hematite, magnetite, limonite, and gibbsite with particle sizes 0.044-0.063 mm; surface area ca. 2 m 2 /g for pyrolusite) in the pH range 7-9 [55], based on measurements by an isotopic dilution technique [56]. This may indicate that only a minor fraction of the major elements M would be adsorbed on outer-sphere positions. The effects of exposure of YBSRed to high concentrations of sodium and calcium on the levels of the trace elements (copper and zinc, as well as Y + REE, Table 4) are minor, which indicate that these elements are bound to the birnessite phase in inner-sphere, rather than outer-sphere, positions and are not readily exchangeable. An assumption-however, not verified experimentally-is that there is a charge balance; the positive charges of Mx and the Mn(III)/Mn(IV) ratio correspond to the negative charge from O4. Uptake of europium by birnessite in inner-sphere positions, independently of pH, has previous been reported [57]. The differences between outer-sphere and inner-sphere adsorption are discussed in some detail by Appelo and Postma (1999) [20], as well as by Kwon et al. (2013) [58].

Phase Analysis
X-Ray diffractograms are given in Figure 5. The peaks, denoted A-R in the figure, were determined using the PANalytical HighScore software. Silicates and calcite in YBSNat (D, F-H, I, K-O) were reduced or absent in YBSRed, which indicates and confirms that the sequential washing process was efficient in removing trace minerals, like feldspars, quartz, and carbonates, besides amorphous phases and adsorbates. No discrete iron phase was indicated in YBSNat or in YBSRed.
There are a minimum of 50 reported structure analyses of synthetic birnessite, but also some on natural birnessite, generally of biogenic origin (e.g., [5,6,10,15,40,51,[59][60][61][62][63]). Almost all studies identify the four peaks B, E, J, and R as being characteristic of hexagonal birnessite. Biological tissue, if present, would generate a peak in the area of E [10,61,62]. Saturation of a synthetic Na-birnessite with several metals, including calcium and lanthanum, is reported by Golden et al. (1987) [64]. XRD patterns indicate a difference, with a peak at 3.56 Å, in the La-birnessite, not present in the Ca-birnessite. A similar difference between YBSLa and YBSCa was, however, not observed (cf. peak E).

Conclusions
It was confirmed that the formation and precipitation of birnessite in the Ytterby mine tunnel is a fast, active process, as assessed already in [45]. Favorable conditions for the formation of microbially induced REE-enriched birnessite deposits are evidently provided in Ytterby mine by (1) the continuous supply of reduced Mn, as well as REEs, by the fracture water; (2) a well-buffered system keeping the pH stable, slightly above circumneutral; and possibly also (3) the sharp redox boundary between the anoxic environment in the water-bearing bedrock fractures and the oxygenated tunnel.
The composition of the birnessite reflects the presence of metals of suitable charge and size in the water phase, notably, sodium, calcium, and Y + REE. There is a strong preference for trivalent over divalent and monovalent ions in the structure for ions of similar size. The exchange of cations (sodium, calcium, lanthanum, iron) had no apparent effect on the birnessite structure. There is a suitable cation size, with an ionic radius around 90-100 pm (sodium, calcium, and Y + REE), but also, metals with a radius around 65-75 pm (copper, zinc, vanadium, and iron) appear to be part of the birnessite structure. Cations of proper size and charge in the water are most likely bound to the manganese oxide in inner-sphere positions, between the Mn-O planes, but a minor fraction may also be adsorbed on the outer surfaces, but not affecting the phase structure.
Iron that would be trivalent under the present redox conditions is not forming a separate detectable discrete mineral phase that could be identified in the XRD phase analysis. The radii of Fe(III) and Mn(III) are almost identical (around 64 pm), and an exchange with Mn(III) in the structure can not a priori be dismissed (almost identical size, charge, and coordination properties). There is also a cerium anomaly, most likely due to oxidation of Ce(III) to Ce(IV), leading to higher enrichment of cerium than expected for the trivalent state.