Absorption mechanisms of Cu2+ on a biogenic bixbyite-like Mn2O3 produced by Bacillus CUA isolated from soil

Background Although most reported biogenic Mn oxides are hexagonal birnessites, other types of biogenic Mn oxides also commonly occur in the environment. However, sorption characteristics and underlying mechanisms of the adsorption of heavy-metal ions on these biogenic Mn oxides are still rarely addressed. In this study, the sorption mechanisms of Cu(II) on a low valence biogenic Mn oxide, poorly crystallized bixbyite-like Mn2O3 (α-Mn2O3), were investigated. Results The maximum adsorption capacity of Cu(II) onto this biogenic Mn oxide at pH 6.00 was 796 mmol/kg (0.45 mol Cu mol−1 Mn). The complex structure of adsorbed Cu(II) was constrained using Cu extended X-ray absorption fine structure (EXAFS) analysis, combined with structural parameters of the biogenic Mn oxide with alternately arranged regular and distorted MnO6 octahedra obtained through multiple-FEFF fitting of Mn EXAFS data. The sorbed Cu(II) was found to coordinate with the biogenic Mn oxide particle edges as inner-sphere complexes. At a relatively low Cu2+ loading (233 mmol/kg, pH 6.00), Cu(II) adsorbed onto the biogenic Mn oxide with two types of coordinated complexes, i.e., (1) coordinated with one regular/distorted MnO6 octahedron as a monodentate-mononuclear complex and (2) with two adjacent MnO6 octahedra as a bidentate-binuclear complex. While, at a relatively high Cu2+ loading (787 mmol/kg, pH 6.00), only one type of coordinated complex was constrained, the adsorbed Cu(II) coordinated with one regular/distorted MnO6 octahedron as a monodentate-mononuclear complex. Conclusions This research extends further insight into the bacterial Mn(II) oxidation in the environment and serves as a good reference for understanding the interactions between metal ions and biogenic low valence Mn oxides, which are still poorly explored either theoretically or practically. Electronic supplementary material The online version of this article (doi:10.1186/s12932-015-0020-6) contains supplementary material, which is available to authorized users.


Background
Copper is one of the most ancient metals of human civilization, and a trace metal in most natural environments, posing no serious threat to biota and vegetation at its background levels. However, it often poses a threat to the environment in areas with elevated levels of copper as a result of contamination from, for example, mining of copper ore minerals, untreated industrial waste waters and its widespread use in agriculture, chemical and electronics industry [1][2][3][4]. Apart from organic matter [5,6] and clay minerals [7], the chemical speciation and mobility of Cu(II) are linked in part to Mn oxides in soils and sediments [8][9][10][11]. Mn oxides are environmentally ubiquitous with multiple Mn oxidation states and an important source of reactive mineral surfaces in the environment. They usually exist as fine granular, spherical or film-like particles distributed in soils and sediments. As natural highly active sorbents and oxidants, Mn oxides widely participate in a variety of adsorption and/or redox reactions [12][13][14]. These reactions control the concentration, speciation, behavior and bioavailability of many heavy-metal ions and organic pollutants in the environment.
The formation of Mn oxides in the environment is generally considered to be closely related to the microbial activity. Up to now, microbially mediated Mn(II) oxidation processes have been studied mainly using three phylogenetically different bacteria: a marine Bacillus sp. Strain G-1, Leptothrix discophora strains SS-1 and SP-6, and Pseudomonas putida strains MnB1 and GB-1 [15][16][17][18][19][20] and a variety of fungi [13,[21][22][23][24]. Accordingly, our knowledge of the adsorption and oxidation of metal cations by biogenic Mn oxides and the transformation of biogenic Mn oxides into other Mn oxide minerals is mainly based on the Mn oxides produced by these microorganisms [12,[25][26][27][28]. However, the primary products of Mn(II) oxidation by the aforementioned microorganisms (bacteria and fungi) are exclusively nanoparticulate, poorly-crystalline hexagonal birnessites with the presence of Mn mainly as Mn(IV). Considering the diverse mechanisms of biooxidation of Mn(II) [20,29], other types of biogenic Mn oxides could also be produced microbially in the environment. Hosseinkhani and Emtiati [30] first reported a gramnegative Acinetobacter sp. strain obtained from ocean water, the Mn(II)-oxidizing product of which was a bixbyite-like Mn 2 O 3 . Then, Zhang et al. [31] reported a Bacillus CUA isolated from soil, the Mn(II) oxidation product of which is also poorly crystallized bixbyitelike Mn 2 O 3 (α-Mn 2 O 3 ), a low valence of biogenic Mn oxides. However, compared to high valence biogenic Mn oxides (e.g., hexagonal birnessites), low valence biogenic Mn-oxides and especially their properties have been rarely reported, despite a plethora of studies on non-biogenic low valence Mn oxides. Webb et al. [32] suggested that Mn(III) is formed as an intermediate during the oxidation of Mn(II) by bacteria and can serve as both oxidant and reductant in one-electrontransfer reactions with other redox species in the biogeochemical processes. Therefore, the knowledge of formation, properties and surface reactivity of low valence biogenic Mn oxides could contribute to a new understanding of the biogeochemistry of Mn-oxides with various Mn valence states and the relevant elements (e.g., Cu(II)).
The sorption mechanisms of Cu(II) adsorbed to Mn oxides also has been extensively investigated [8,11,[33][34][35][36][37][38][39]. Manceau et al. [35] claimed that the adsorbed Cu(II) on birnessite was six-fold coordination with four oxygen atoms at 1.96 Å and two at 2.23 Å, while, Sherman and Peacock [36] maintained that the adsorbed Cu(II) on birnessite was four-fold coordination. However, they are in agreement that the adsorbed Cu(II) coordinated with three surface oxygen atoms at vacancy sites of birnessite, forming a triple-corner-sharing complex with a Cu-Mn interatomic distance of 3.39-3.43 Å [35][36][37]. In addition, in aqueous solution, Cu(II) was reported to prefer a fivefold elongated square pyramidal coordination that is likely to be competitive with six-fold distorted octahedral coordination [40][41][42]. Sherman and Peacock [36] proposed that some Cu(II) was adsorbed in Cu-incorporation (Cu-INC) mode at a high pH value (pH ≈ 8.0), based on the appearance of a peak near 2.9 Å in the Fouriertransformed EXAFS spectrum for δ-MnO 2 sample with a loading of 0.068 mol Cu mol −1 Mn. However, in a density functional theory (DFT) study, Kwon et al. [39] argued that the incorporation of Cu into a vacancy site strongly inhibits the Jahn-Teller distortion of Cu, destabilizing the Cu-INC species relative to Cu-TCS species at any pH value. More recently, Peña et al. [11] showed that Cu(II) was dominantly adsorbed at particle edges of δ-MnO 2 as dimers or polynuclear surface species with a surface loading of 0.01 to 0.26 mol Cu mol −1 Mn.
However, previous studies on Cu(II) sorption onto Mn oxides are mainly focused on chemically synthesized layered Mn oxides, including some high-valence biogenic hexagonal birnessite. There are no reports about Cu(II) sorption onto low-valence biogenic Mn oxides and the issue of how the adsorbed Cu(II) coordinates with the Mn oxides is rarely addressed. Here, a low-valence biogenic Mn oxide was prepared, and the coordination mechanisms for the sorption of Cu(II), as a model of heavy metals, onto the biogenic Mn oxide were investigated using X-ray absorption spectroscopy (XAS).

Sorption of Cu(II) onto the biogenic Mn oxide
Adsorption isotherm of Cu(II) on biogenic Mn oxide at pH6.0 is presented in Fig. 1. At the given pH, the final adsorbed amount (mmol/kg) increased with increasing initial Cu 2+ concentration. The data were fitted to a Langmuir model with a correlation coefficient of R 2 = 0.9904, indicating an adequate fit of the equation to the data. The maximum adsorption capacity was 796 mmol/kg (0.45 mol Cu mol −1 Mn). Therefore, the biogenic Mn oxide (α-Mn 2 O 3 ) could also be a promising adsorbent for eliminating Cu(II) from soil and water environments. Two sorption samples, pH6.0-Cu0.4 and pH6.0-Cu4.0, as indicated by their initial pH and Cu 2+ concentrations (mM), were used to collect EXAFS spectra. The adsorption densities of the two samples were 233 mmol/kg (0.13 mol Cu mol −1 Mn) for pH6.0-Cu0.4 and 787 mmol/kg (0.44 mol Cu mol −1 Mn) for pH6.0-Cu4.0, respectively. Figure 2 shows X-ray absorption near-edge structure (XANES) spectra and second derivatives of the XANES spectra of the sorption samples and some reference compounds. The two sorption samples have similar Xray absorption edges characteristic of divalent Cu(II) aqueous solution (Fig. 2a), indicating that the oxidation state of Cu on the surface of biogenic Mn oxide (Bixbyite) remains unchanged. In general, the sorbed Cu prefers to bind with Mn oxides, for example, birnessite, with three mechanisms: inner-sphere surface at cation vacancy sites and/or particle edges, incorporation into the MnO 2 sheet and Cu polymerization or precipitation [11]. The two sorption samples differ from Cu(OH) 2 not only in the shape and position of the XANES spectra, but also in second derivative (Fig. 2b) and EXAFS (Fig. 3), thereby eliminating the possibility of polymerization or precipitation. In tetrahedral coordination, the pre-edge feature in Cu K-edge XANES should show greater intensity and be shifted towards slightly lower energies [11,43,44], which meant that the sorbed Cu(II) was not in tetrahedral coordination. The second derivative of copper acetate monohydrate, a model compound with isolated dimeric Cu units, was used to determine whether dimeric Cu units were present in sorbed Cu(II) [45]. Here, the two sorption samples have second derivative differ with copper acetate monohydrate, which means no dimeric Cu unit exist in the sorbed Cu(II) on the surface of biogenic Mn oxides. Figure 3 shows EXAFS and Fourier transforms (FT) of the sorption samples and those of reference compounds. Generally, the d 9 electronic configuration of Cu(II) leads to the Jahn-Teller distortion of the first coordination shell and this structure distortion leads to two distinct Cu-O shell subshells, where the equatorial sub-shell dominates the EXAFS signal, but no reliable coordination numbers and distances of axial sub-shell could be fitted due to the anisotropic disorder [11,45,46]. Thus, we fit the first Cu-O shell EXAFS spectra of Cu 2+ (aq) with coordination number fixed at 4 and got the amplitude reduction factor (S 0 2 = 0.96). Then, the other EXAFS spectra were fitted with S 0 2 set at 0.96, and the fitting results are listed in Table 1. It shows that the main EXAFS contributions of the reference compounds are from oxygen atoms at approximately~1.95 Å, with coordination number close to 4 of this first Cu-O shell, which agrees well with several previous reports [11,36,45]. Additionally, all of the Cu-O shells with a low Debye-Waller factor (DWF, 0.005-0.007 Å 2 ), indicates a low local structural disorder and these Cu-O bonds are the typical equatorial bonds of the distorted Cu polyhedron. Manceua et al. [35] obtained an improved fit by 4 equatorial oxygen atoms at 1.96 Å and two axial oxygen atoms at 2.23 Å, with a much larger DWF, as expected for Cu in a Jahn-Teller distorted environment. We cannot confirm whether Cu has one or two axial ligands in our sorption samples, due to no obvious improvement with the addition of one or two axial oxygen atoms. However, in principle, both the results are reasonable, because Cu in aqueous solution prefers a five-fold coordination that is likely to be competitive with a six-fold distorted octahedral coordination [36,[40][41][42].  [45]. The fitting results are given in Fig. 4 and Table 1. Table 1 shows that the Cu-Mn distance of the sample pH6.0-Cu0.4 is 3.37 Å with a coordination number of 1.7 (0.3) and that of the sample pH6.0-Cu4.0 is 3.44 Å with a coordination number of 0.7 (0.3). Sherman and Peacock [36] showed that the Cu-Mn interatomic distance of sobred Cu incorporation into the MnO 2 sheet was in the range of 2.81-2.94 Å. In a density functional theory study, Kown et al. [39] showed that the theoretical values of Cu-Mn distance in incorporation mode are 2.84 Å, 2.95 Å and 2.98 Å. Though the biogenic Mn oxide (Bixbyite) in this study differs in structure with birnessite, the Cu-Mn distance (3.37 Å and 3.44 Å) in the two sorption samples should be unlikely in incorporation mode. Thus, we deny the possibility of Cu incorporation into the biogenic Mn oxide in our study. In birnessite, the adsorption Cu(II) at cation vacancy sites involves the coordination of the copper to the three surface oxygen atoms surrounding the vacancy site, thus forming a triple-corner-sharing (TCS) complex [11,35,47]. The Cu-Mn interatomic distances of Cu-TCS complexes in birnessite were inferred in the range of 3.39-3.43 Å with a Cu-Mn shell coordination number of 3 [35][36][37]. Based on the Cu-Mn distance, the sorbed Cu(II) on the two samples could be in Cu-TCS complexes, but the coordination numbers of the two sorption samples are far less than 3. Additionally, there is no report of cation vacancy site in the structure of bixbyite [31]. Therefore, the sorbed Cu(II) on the two samples are unlikely in Cu-TCS complexes. Based on the above analysis, the most possible way of sorbed Cu(II) on the biogenic Mn oxide is coordinated with the particle edges as inner-sphere surface complexes. To find out the specific ways in which the sorbed Cu(II) coordinated with the MnO 6 octahedra, we used the EXAFS analysis results of our samples and the polyhedral approach for further investigation.

Constraints on the structure of Cu 2+ complexes on the low valence biogenic Mn oxide
We cannot confirm whether Cu has one or two axial ligands in our sorption samples, but, in principle, both the results are reasonable [11,39], because the five-fold-   [31]. Based on these coordination distances and the geometry coordination model (Fig. 5), we analyzed the possible linkages between a Cu(II)O 6 octahedron and a MnO 6 octahedron as follows: (1)A Cu(II)O 6 octahedron and one MnO 6 octahedron can link as a monodentate mononuclear (MM) complex (Fig. 5a). This structure could exhibit Cu-Mn distances in the range of approximately 2.800-4.300 Å. The range of Cu-Mn distances is large because Cu can be coordinated by either equatorial or axial oxygen. Furthermore, it is theoretically possible that a Cu(II)O 6 octahedron and a MnO 6 octahedron could link at a flexible angle. The Cu-Mn distances observed in the two samples (Table 1) are all within the indicated range, suggesting that MM is the first likely linkage. complex (Fig. 5c). Based on a spatial geometry, the theoretical Cu-Mn distance for this BB linkage can be calculated using the following trigonometric function: where d is the atomic distance of the indicated interatomic bond, and θ is the Mn-O-Cu angle. The Cu-Mn distance in this mode is approximately in the range of 2.73 to 3.98 Å (Additional file 1: S1). The Cu-Mn distances of the two samples (Table 1) are all within the indicated range, implying that BB is the second likely linkage. However, the coordination number of the sample pH6.0-Cu4.0 is 0.7 (0.3), which is far less than the theoretical coordination number (CN = 2) of BB mode, indicating that the possibility of a bidentate binuclear (BB) complex in the sample pH6.0-Cu4.0 can be rejected. Therefore, at a low Cu 2+ loading, the sorbed Cu 2+ could link with the MnO 6 octahedra with two types of complexes: (1) one CuO 6 octahedron and one MnO 6 octahedron (regular or distorted) linked as MM (Fig. 6a); (2) one CuO 6 octahedron and two adjacent MnO 6 octahedra linked as BB (Fig. 6b). At a high Cu 2+ loading, Cu 2+ (CuO 6 ) could link with MnO 6 octahedra only in one possible type of complex, one CuO 6 octahedron and one MnO 6 octahedron (regular or distorted) linked as MM (Fig. 6a). There is a little difference in the coordination mode between the low and high Cu 2+ loading samples, probably because there are still some adsorption sites in the low Cu 2+ loading sample (233 mmol/kg), resulting in the shift of some sorbed Cu 2+ in the monodentate mononuclear (MM) complex to the bidentate binuclear (BB) complex, a more stable geometric structure. However, in the high Cu 2+ loading sample (787 mmol/kg), there is almost no additional adsorption site in it, because the adsorption density (787 mmol/kg) is almost in the maximum adsorption capacity (796 mmol/kg) of the biogenic Mn oxide.
It should be noted that, in all of the above analyses, mainly the bulk structure of the biogenic Mn oxide (bixbyite) was considered, implying that the bulk and the surface structure of the biogenic Mn oxide were assumed to be the same. When the surface structure becomes available, the options for the Cu 2+ complex structures on the surface of the biogenic bixbyite-like phase might be further constrained. The aforementioned results could provide a good reference for understanding the interactions between metal cations and biogenic or abiogenic lowvalence Mn oxides, which are still poorly explored both theoretically and experimentally.

Biogenic Mn oxide
The biogenic Mn oxide was reported in our previous work [31]. Briefly, biogenic Mn oxides were produced by culturing an isolated Mn-oxidizing bacterium using Leptothrix discophora medium with initial Mn(II) concentration at 1.0 mM. The biogenic Mn oxide was harvested after 12 days via centrifugation at 10,000 RCF (25°C, 5 min), because at which time the Mn-oxide concentration was at its maximum. The biogenic Mn oxide used in this study is a poorly-crystalline bixbyite (JCPDS 01-078-0390), with a bulk-averaged Mn oxidation state (AOS) of 2.73. It is composed of uniform aggregates of granular particles with sizes of 25-50 nm, and a specific surface area of 49 m 2 /g [31].
Bixbyite is a cubic, granular manganese oxide mineral of the cubic crystal system Ia-3 (SG.NO.206) with regular and distorted MnO 6 octahedra alternately arranged in the structure [48,31]. In the structure of crystalline bixbyite, the cations occupy the 8a site at 0, 0, 0, etc. and the 24d site at x, 0, 0.25, etc., with x = 0.2855 [48]. The anions occupy the general 48e site, x, y, z (0.1293, 0.1471, −0.0835), whereas the 16c site, x, x, x, is unoccupied, being the site of the "anion vacancies", with x = 0.125. The cell formula is Mn 32 O 48 □ 16 (□ denotes an anion vacancy) [48]. In the ideal structure of bixbyite, the Mn1 atoms of the 8a site are incorporated into the regular Mn1O 6 octahedra and the Mn2 atoms of the 24d site are part of the highly distorted Mn2O 6 octahedra/polyhedron due to the distortion of the 3d 4 Mn(III) ions. The six Mn-O distances of the regular Mn1O 6 octahedra in the prepared biogenic Mn oxide of this study are 2.095 Å, while the Mn-O distances in the distorted Mn2O 6 octahedra have three different pairs at 1.903 Å, 2.178 Å and 2.350 Å [31].

Sorption of Cu 2+ on biogenic Mn Oxide
Sorption experiments of Cu 2+ on the biogenic Mn oxides were carried out as previously reported by Zhao et al. [49] and Zhang et al. [31]. Briefly, 5.0 g/L mineral suspensions were prepared and equilibrated for a few days, during which the pH was adjusted to 6.00 (±0.05) with 0.1 M HNO 3 or 0.1 M NaOH. Sorption experiments were initiated by adding 20 mL of the final suspension to a 100-mL polypropylene bottle containing 0.8 mL to 16 mL of 15-mM Cu(NO 3 ) 2 solution. The total volume of the solution in each bottle was increased to 60 mL by adding 0.15 M NaNO 3 . Therefore, the sorbent concentration was 1.67 g/L, the ionic strength was approximately 0.1 M and the Cu 2+ initial concentrations were 0.2 to 4.0 mM, respectively. The polypropylene bottles with suspensions were immediately capped and shaken, and the pH was adjusted twice during the equilibration period to 6.00 (±0.05). After 24 h of equilibration, the suspensions were centrifuged. Approximately 10 mL of the clear supernatant liquid was used for the determination of the Cu 2+ concentration using a Varian AAS 240FS atomic absorption spectrometer (USA). The precipitates were collected and conserved at 4°C for the XAS analysis.

XAS data collection
The XAS spectra of all of the samples were collected on the 1W1B beamline at Beijing Synchrotron Radiation Facility (BSRF) [31,[50][51][52]. Cu sorption samples were obtained from centrifugation and prepared by the procedures of Zhang et al. [31]. The EXAFS spectra (8979 eV) were collected with a Lytle ion-chamber detector in fluorescence mode. The reference compounds, solid Cu(OH) 2 powder, copper acetate monohydrate and copper aqueous solutions (0.1 M Cu(NO 3 ) 2 ) were measured in transmission mode. The energy was calibrated with Cu foil. Reduction and analysis of all XAS data were performed using IFEFFIT/ SIXPACK [53]. The Cu K-edge averaged spectra were background-subtracted using the following parameters: E 0 = 8979 eV, Rbkg = 1.0 Å and k-weight = 3. Phase and amplitude functions for single-scattering paths were calculated using FEFF7 [54]. An amplitude reduction factor (S 0 2 ) of 0.96 was determined by fitting the Cu-O shell of Cu(NO 3 ) 2 (aq) spectra in the R-range 1.0-2.0 Å, and the other samples were fitted with S 0 2 fixed at 0.96. The calculations of the phase shift and amplitude functions of Cu-Mn were based on the crystal structure of CuMnO 2 (Crednerite). Final fits of the two sorption samples were made in Artemis in R-range 1.0-3.8 Å.

Conclusions
In this study, a low valence biogenic Mn oxide, a poorly crystallized bixbyite-like Mn 2 O 3 (α-Mn 2 O 3 ), which has barely been studied in the past several decades, was used as an adsorbent and the adsorption mechanisms of Cu(II) on it were studied. The maximum adsorption capacity of Cu(II) adsorbed to the biogenic Mn oxide at pH 6.00 was 796 mmol/kg, indicating that the biogenic Mn oxide could be an promising adsorbent for eliminating Cu(II) from soil and water environments. At a relatively low Cu(II) loading (233 mmol/kg) sample, Cu(II) was adsorbed onto the biogenic Mn oxide as innersphere complexes and linked with MnO 6 octahedra by two types of complexes: MM and BB. At a relatively high Cu(II) loading (787 mmol/kg), the adsorbed Cu(II) linked with MnO 6 octahedra only by MM type of complex. These results provide further insight into the bacterial Mn(II) oxidation in the environment and a good reference for understanding the biogeochemistry of low valence of biogenic Mn oxides and their interactions with metal ions in the natural environments, which are still poorly explored, either theoretically or practically. Furthermore, this research could also enrich or contribute a new understanding of the biogeochemical processes of Mn oxides and their environmental significance.