Cryptomelane formation from nanocrystalline vernadite precursor: a high energy X-ray scattering and transmission electron microscopy perspective on reaction mechanisms
- 2.7k Downloads
Vernadite is a nanocrystalline and turbostratic phyllomanganate which is ubiquitous in the environment. Its layers are built of (MnO6)8− octahedra connected through their edges and frequently contain vacancies and (or) isomorphic substitutions. Both create a layer charge deficit that can exceed 1 valence unit per layer octahedron and thus induces a strong chemical reactivity. In addition, vernadite has a high affinity for many trace elements (e.g., Co, Ni, and Zn) and possesses a redox potential that allows for the oxidation of redox-sensitive elements (e.g., As, Cr, Tl). As a result, vernadite acts as a sink for many trace metal elements. In the environment, vernadite is often found associated with tectomanganates (e.g., todorokite and cryptomelane) of which it is thought to be the precursor. The transformation mechanism is not yet fully understood however and the fate of metals initially contained in vernadite structure during this transformation is still debated. In the present work, the transformation of synthetic vernadite (δ-MnO2) to synthetic cryptomelane under conditions analogous to those prevailing in soils (dry state, room temperature and ambient pressure, in the dark) and over a time scale of ~10 years was monitored using high-energy X-ray scattering (with both Bragg-rod and pair distribution function formalisms) and transmission electron microscopy.
Migration of Mn3+ from layer to interlayer to release strains and their subsequent sorption above newly formed vacancy in a triple-corner sharing configuration initiate the reaction. Reaction proceeds with preferential growth to form needle-like crystals that subsequently aggregate. Finally, the resulting lath-shaped crystals stack, with n × 120° (n = 1 or 2) rotations between crystals. Resulting cryptomelane crystal sizes are ~50–150 nm in the ab plane and ~10–50 nm along c*, that is a tenfold increase compared to fresh samples.
The presently observed transformation mechanism is analogous to that observed in other studies that used higher temperatures and (or) pressure, and resulting tectomanganate crystals have a number of morphological characteristics similar to natural ones. This pleads for the relevance of the proposed mechanism to environmental conditions.
KeywordsVernadite δ-MnO2 Cryptomelane Phyllomanganate Tectomanganate Pair Distribution Function Bragg rod High-energy X-ray scattering X-ray diffraction Transmission electron microscopy
Vernadite (and δ-MnO2, its synthetic analogue) is a nanocrystalline turbostratic birnessite, a phyllomanganate whose layers are built of (MnO6)8− octahedra connected through their edges and separated by hydrated interlayer cations. Vernadite is ubiquitous in the environment, and probably results mainly from the aqueous oxidation of Mn2+ by bacteria [1, 2, 3], fungi [4, 5, 6], and higher living forms , as abiotic oxidation catalyzed by mineral surfaces is about two orders of magnitude slower [8, 9, 10, 11]. Vernadite layers frequently contain vacancies and (or) isomorphic substitutions (substitution of layer Mn4+ by a foreign cation, e.g., Co3+, Mn3+, or Ni2+ [12, 13, 14]), both types of defects inducing a layer charge deficit. For example, layer charge was 0.86–1.22 and 1.58 valence unit (v.u.) per layer octahedron for samples produced by fungal strains  and by grass roots , respectively. Layer charge induces a high chemical reactivity, which is reinforced by the nanometric size of vernadite (typically 5–50 nm in the layer plane—[1, 15, 16]) and the induced proportion of reactive edge sites [3, 17, 18]. Vernadite also presents a high affinity for many trace elements such as transition metals (e.g., Co, Ni, Zn), actinides and rare earth elements [14, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36]. For example, vernadite is the main sink for Ni in mixed mineral/biotic systems (vernadite and Pseudomonas Putida biofilm [37, 38]). Vernadite can also oxidize organic pollutants and redox-sensitive elements such as arsenic [39, 40, 41, 42], chromium [39, 40, 41, 42, 43], and thallium [34, 44], possibly because of the common coexistence of heterovalent Mn cations in its structure. Specifically, Mn3+ cations can be present both within octahedral layers and (or) adsorbed above layer vacancies, forming triple-corner-sharing complexes (TCMn3+; Fig. 1 in ).
Additional interest in understanding vernadite structure arises from its frequent association with tectomanganates (i.e., tunnel structures) in the environment. Tectomanganates include a variety of minerals with [n × m] tunnel sizes, where n stands for the number of octahedra connected to form the “walls” of the tunnels, whereas m stands for the number of octahedra forming the “ceiling” and the “floor”. For example, vernadite is frequently found with todorokite and cryptomelane, [3 × 3] and [2 × 2] tunnel structures, respectively [26, 46, 47, 48]. Tectomanganates can form from phyllomanganate precursors [46, 49, 50] provided that they possess a particular crystal-chemistry, for example possess Mn3+ [51, 52]. Phyllomanganate-to-tectomanganate reaction mechanisms have been widely investigated owing to the tectomanganate potential as octahedral molecular sieves. Reaction products must have homogenous tunnel size for this purpose to achieve optimal efficiency and the relation between the layered precursor and the resulting tunnel structure is studied with special care (e.g.,  and references therein and [53, 54]). Relations between layer and tunnel structures is also of interest in natural settings, mainly because defective todorokite (i.e., having mainly [3 × 3] tunnel size, together with [3 × m] size, m varying from 2 to 5) is found associated with vernadite in oceanic ferromanganese nodules. These nodules consist of alternating layers of iron and manganese oxides [31, 34, 55], the latter containing typically over 1 wt% of Ni [56, 57, 58], and are increasingly considered for their potential as a source of strategic trace metals including rare-earth elements. A comprehensive understanding of phyllomanganate to tectomanganate transformation, at the atomic scale, thus appears key to an improved prediction and modeling of the impact of structure defects (layer vacancies and isomorphic substitutions) on the fate of trace elements.
Observations of vernadite-to-todorokite transformation in natural samples are scarce, but suggest a topotactic reaction . It was first reported by Golden and coworkers  for synthetic analogues and has since been repeatedly described for a variety of hydrothermal protocols. Recently, these protocols were reviewed by Atkins and coworkers , who also characterized the structural mechanisms of the δ-MnO2 to todorokite transformation at 100 °C and atmospheric pressure. These authors described the transformation as a four-stage process, starting with the formation of todorokite tunnel walls via layer kinking, followed by the growth of the particles along the  direction (i.e., along the tunnel direction) during a dissolution–recrystallization step and subsequent oriented attachment of the resulting particles along the  direction to form todorokite laths which then stack. A last step is described as the growth of todorokite crystals by Ostwald ripening. These authors mentioned also that the density of structural Mn3+ has to be “significant” to allow for a complete transformation and linked this condition to the ability of Mn3+ to induce layer kinking, owing to its Jahn–Teller distorted environment. Another important parameter appears to be the presence of an interlayer cation enforcing a ~10 Å layer-to-layer distance (e.g., Mg2+).
Similar to todorokite, cryptomelane formation in environmental conditions is little documented. A recent study  focused on δ-MnO2 samples equilibrated at pH values ranging from 3 to 10. Initially, all samples had a similar number of layer Mn3+ per octahedron (0.10 ± 0.02–0.14 ± 0.02), whereas the number of interlayer TCMn3+ increased with decreasing equilibration pH. Although such evolution might seem counter-intuitive because, by analogy with clay minerals, decreasing pH should increase proton competition for sorption [61, 62, 63], it is made possible by the fact that pH decrease is accompanied by partial layer dissolution and thus by an increase of Mn concentration in the equilibrium solution, making it possible for dissolved Mn to adsorb above vacancies . Structural formulae from samples initially equilibrated at pH 3 and 10 were Na 0.06 + (H2O)0.30Mn 0.185 3+ [Mn 0.12 3+ Mn 0.71 4+ Vac0.17O2] and Na 0.27 + (H2O)0.30Mn 0.095 3+ [Mn 0.10 3+ Mn 0.76 4+ Vac0.14O2], respectively. In these formulae, species within brackets form the layer, those to the left the interlayer, Vac stands for layer vacancies, and all interlayer Mn sites (TC and triple-edge sharing  configurations) are summed. Samples were then dried, and aged in the dark. With time, it was observed that layer Mn3+ leaves the layer to form TCMn3+ above the newly generated vacancy. Partial transformation to cryptomelane was observed only for pH 3 sample that had the highest initial number of Mn3+ (layer plus interlayer). Samples initially equilibrated at pH 4–10 did not show evidence for transformation to cryptomelane, but had contrasting crystal chemistry (Na+ 0.12(H2O)0.30Mn 0.315 3+ [Mn 0.74 4+ Vac0.26O2] and Na+ 0.27(H2O)0.30Mn 0.205 3+ [Mn 0.79 4+ Vac0.21O2], respectively) after 8 years of ageing. The structure of pH 3 sample was not precisely determined, but crystals that transformed certainly had ~1/3 of TCMn3+ per layer octahedron . Still, many questions remain open as to the transformation mechanisms. In particular, it was unclear if the transformation was homogeneous at the crystal scale (if some crystals transformed to cryptomelane while others remained lamellar) or if it only affected portions of the crystals (topotactic transformation). Finally, tectomanganates observed in natural environments are larger than typical vernadite [65, 66]. The mechanisms leading to crystal growth during or following vernadite to cryptomelane transformation (e.g., dissolution/recrystallization, aggregation, oriented attachment, Ostwald ripening) could not be elucidated.
The present study focuses further on the structure of aged δ-MnO2 samples studied by Grangeon and coworkers . Samples that were initially equilibrated at pH values of 3, 4, 8, and 10, and then left ageing in the dark in the dry state for 10 years are hereafter referred to as MndBiXX_10y, where XX stands for the equilibration pH. These samples were selected because of their contrasting structures after 8 years of ageing (i.e., 2 years before the present study). MndBi3_10y and MndBi10_10y respectively represent the highest degree of transformation and an aged δ-MnO2 that retained its original lamellar structure. In the present study, synchrotron X-ray diffraction data is analyzed with both Bragg-rod and pair distribution function (PDF) methods to determine the sample structure as a function of pH, after two more years of ageing. Bragg-rod formalism is used to determine whether samples were partially transformed to cryptomelane, and to determine the size of coherent scattering domains (CSD) and the number of TCMn3+. PDF analysis is used to determine if layer vacancies are ordered and to cross-check the number of TCMn3+. In addition, transmission electron microscopy (TEM) will be used to determine actual crystal sizes and to provide textural and structural constraints on the vernadite-to-cryptomelane transformation mechanisms. Scanning TEM will be used to reveal structural heterogeneities at the atomic scale.
Analysis of X-ray diffraction patterns and Bragg-rod modeling
Compared to MndBi8_10y and MndBi10_10y, additional reflections are observed at 29.0 nm−1 (2.17 Å), 34.4 nm−1 (1.83 Å), and 40.6 nm−1 (1.55 Å) in MndBi3_10y and MndBi4_10y. These reflections are more intense in MndBi3_10y than in MndBi4_10y (Fig. 1) and can be attributed to a cryptomelane-like structure. MndBi3_10y and MndBi4_10y are thus well-suited to study initial stages of the phyllomanganate-to-tectomanganate transformation, with part of MndBi3_10y and, to a lesser extent, of MndBi4_10y crystals having a cryptomelane-like structure.
Main structural parameters extracted from analysis of high energy X-ray scattering data, in the PDF and Bragg-rod approaches
TCMn (per layer octahedron)
Analysis of PDF data
As PDF analysis will hereafter focus on short-range order, MndBi3_10y and MndBi4_10y, which contain a minor amount of cryptomelane, will be treated as pure δ-MnO2 owing to the structural similarities between the two species. Indeed, Mn–Mn pairs from δ-MnO2 layers are similar to those in cryptomelane walls or floor/ceiling, and pairs formed by layer Mn and TCMn in δ-MnO2 are similar to those formed by Mn atoms from adjacent walls and floor/ceiling in cryptomelane. An implication of this similarity is that PDF data cannot be used, in the present study, to detect a minor amount of cryptomelane in the samples.
Morphological and structural evolution with pH as seen by transmission electron microscopy
Bragg-rod and PDF data analysis allowed probing structure of coherent scattering domains but could not determine their distribution within crystals, and in particular the possible coexistence of δ-MnO2 and cryptomelane domains within crystals. As they are representative of samples undergoing the maximum and minimum degree of structural conversion, MndBi3_10y and MndBi10_10y were investigated by TEM to gain further insights into transformation mechanisms.
Mechanism of the phyllomanganate to tectomanganate transformation
To our knowledge, this study is the first to document the transformation of vernadite to cryptomelane under conditions that can be considered relevant for soils, i.e. at room temperature, ~105 Pa, under unsaturated conditions, and in the dark. As previously discussed , this structural transformation requires a locally high number of Mn3+ (~0.33 per layer octahedron) in the octahedral layer. This can be obtained experimentally by equilibrating phyllomanganates at low pH values (≤4) that can be observed in forest soils, soils developed on parent granite or gneiss, or organic-rich soils . Such assumption is confirmed by independent laboratory experiments which show that δ-MnO2 to cryptomelane transformation is favored at low pH .
From many viewpoints, the studied transformation is similar to that of δ-MnO2 to todorokite described by Atkins and coworkers . The contrasting reaction products obtained in the two studies are most likely due to the different nature of interlayer cation in the initial phyllomanganate (Na+ in the present study and Mg2+ for Atkins and coworkers). The first reaction step differs however as Atkins and coworkers propose that the presence of layer Mn3+ leads to layer kinking, owing to their Jahn–Teller distorted coordination sphere. In our opinion, reaction first step rather corresponds to the migration of Mn3+ from the layer to the interlayer, which allows also releasing strains related to the Jahn–Teller distortion of Mn3+ octahedra. This interpretation is consistent with the infrared data of Atkins and coworkers that showed an increased density of TCMn during the initial steps of the transformation. The proposed mechanism accounts also for their observation, corroborated by Feng and coworkers , that transformation occurs at constant mean Mn oxidation degree. In contrast to the hypothesis of Atkins and coworkers this initial step is likely not thermally triggered but reaction kinetics is enhanced with increasing temperature. The next steps of the reaction described by Atkins and coworkers are also observed in the present study. Needle-like crystal type 2 results indeed from the crystal growth along tectomanganate tunnels proposed by Atkins and coworkers, whereas lath-like crystal type 3 results from crystal growth perpendicular to tectomanganate tunnels (by coalescence of type 2 crystals along their long dimension). Finally, crystal type 4, built of lath-like units rotated by n × 120° (n being equal to 1 or 2), results from stacking of crystal type 3. The crystallographic axes along which crystal growth/aggregation takes place could not be determined in the present study, as SAED patterns could not be collected on type 2 and 3 crystals, and as instrumental limitations hampered the observations of in-plane lattice fringes. A last transformation step was described by Atkins and coworkers as Ostwald ripening and would involve cryptomelane crystal growth. This final step could not be observed in the present samples as evidence for a cryptomelane-like signal could only be found in crystal type 4 (Fig. 8). To date, only Bodeï and coworkers  have investigated in detail the mechanisms of phyllomanganate (vernadite) to tectomanganate (todorokite) transformation in natural samples. Collation of the present study with that of these authors is hampered however by numerous unknowns inherent to natural systems, such as the density of layer and interlayer Mn3+ in the initial phyllomanganate and the structural variety of natural tectomanganates [77, 78]. However, several clues suggest that the observed transformation is similar to that occurring in natural systems. First, proposed transformation mechanisms are valid both in surface soil conditions (present study) and in saturated conditions [50, 72], making them relevant to a variety of natural systems. Second, naturally occurring tectomanganates frequently exhibit textures similar to those observed in experimental studies (Figs. 7, 8; [51, 53, 60, 79, 80]) with rotations of aggregated crystals by n × 120° [46, 66, 79]. Third, transformations occur via topotactic transformation in both experimental and natural systems and are heterogeneous at the crystal scale. In particular, transformation systematically affects only part of the crystals and the transformed parts show heterogeneous tunnel sizes. This latter point is certainly related to an imperfect distribution of Mn3+ atoms (TCMn3+ + layer Mn3+) in the initial phyllomanganate structure.
Growth mechanisms proposed from previous experimental studies performed at higher temperatures were confirmed for conditions relevant to surface soil. The four-stage transformation begins with the migration of Mn3+ from layer to the interlayer to form TCMn3+ that further connect to hydration spheres belonging to TCMn3+ from adjacent layers. Crystals then grow first along the “tunnel” direction to form needle-like crystals. These crystals coalesce along their long dimension to form lath-like crystals which, in a final step, stack along c* with rotation by n × 120° between adjacent laths. Cryptomelane structure was observed only in these latter crystals (type 4), but is certainly present also in crystals type 2 and 3.
Samples used for the present study are those studied by Grangeon and coworkers  and Manceau and coworkers . They were synthesized using the redox method  and were left for equilibration at pH values ranging from 3 to 10 immediately after synthesis. They were then left to age for 10 years in the dark and in the dry state. For consistency with Grangeon and coworkers , samples are labeled MndBiXX_10y, where XX is the equilibrium pH and “10y” stands for “10 years of ageing”.
X-ray diffraction pattern (XRD) analysis and modeling
XRD patterns were modeled using the software developed by Plançon , based on the formalism developed by Drits and Tchoubar . This specific routine allows for the simulation of (lamellar) structures affected by various nature and density of layer defects (e.g., layer vacancies, isomorphic substitutions) and stacking defects (e.g., well-defined or random stacking faults, interstratification). It has been previously applied to the study of nanocrystalline manganese or iron oxides [69, 84, 85], nanocrystalline calcium silicate hydrates [86, 87, 88] and phyllosilicates [89, 90]. As discussed by Manceau and coworkers  the simulation of the [11, 20], [31, 02] and [22, 40] bands (using a C-centered hexagonal unit-cell, with γ = 90°) is sufficient to accurately determine the structure of synthetic phyllomanganates, and only these bands were modeled here. All structure parameters but crystallite size, a and b lattice parameters, abundance of TCMn3+, layer vacancies and of interlayer water molecules were kept identical to those determined by Grangeon and coworkers . An example of the sensitivity of calculated XRD patterns to the number of TCMn3+ is available in the Additional File 2: Data S2. The cryptomelane pattern was calculated using the structure model from Vicat and coworkers , assuming a CSD size of 6 nm.
High energy X-ray scattering coupled with pair distribution function (PDF) analysis and modeling
X-ray diffraction patterns were collected at ID15B high-energy beamline of the European Synchrotron Radiation Facility (ESRF, Grenoble, France), using energy of 87 keV and a PerkinElmer flat panel detector. Data were acquired on randomly-oriented powders packed in polyimide capillaries having a diameter of 1 mm and on empty capillary used for background subtraction. 40 frames of 5 s, corrected for detector’s dark current, were collected for each sample. After instrumental calibration using a NIST certified CeO2 powder sample, frames were integrated to one dimensional patterns  and averaged. Data were then transformed to PDF patterns using PdfGetX3 , and fit using PDFGui . The model from Manceau and coworkers  was used to refine the patterns. The only modification was that TCMn3+ was allowed to sorb on both sides of a layer vacancy. The refined parameters were the abundance of TCMn3+, layer vacancies, lattice parameters and Debye–Waller factors. q broadening and q dampening factors were retrieved from simulation of a CeO2 pattern and found to be equal to, respectively, 0.044 and 0.048.
Transmission electron microscopy (TEM)
TEM was performed using a Philips CM20 operated at 200 kV. Samples were deposited on a copper grid prior to observation. When samples were first dispersed in ethanol or in water, and deposited on the grid from the suspension, they were rapidly altered under the beam, with amorphous products occurring within a few seconds. To circumvent this problem, samples were first embedded in epoxy resin and left in the dark for 48 h until full polymerization. Cutting was performed using an ultramicrotome Reichert-Jung Ultra-cut E. Five to ten thin sections having thicknesses of about 100 nm were collected on the surface of the water contained in the boated knife and picked up on a lacey carbon film loaded on copper grids.
Scanning transmission electron microscopy (STEM)
STEM experiments were performed using a Nion Ultra-STEM 200 operated at 100 kV. Sample preparation was identical to that applied for TEM measurements, except that the thickness of the slices containing the sample embedded in epoxy resin was reduced to ~50 nm. Image presented in the present study was acquired in high-angular annular dark-field (HAADF) mode. In order to ease visualization of the structural features, the raw image was Fourier-transformed and filtered using a band pass with a window set for spatial frequencies between 1.6 and 40 nm−1, and a color threshold was then applied to reduce the contribution from pixels having a grey color lower than half of the mean image grey value.
AFM, AP, SG and NM participated to high-energy X-ray scattering experiments (HEXS). SG processed HEXS data. FW and SG did TEM experiments. SG and FW processed and analyzed TEM data. AG conducted STEM experiments to which FW and SG participated. AG, FW and SG interpreted STEM data. SG wrote the main parts of the manuscript. AFM, FW, NM, AG and BL participated to manuscript writing. All authors read and approved the final manuscript.
S.G. acknowledges funding from the ANR (NACRE—ANR-14-CE01-0006). AFM was supported by a grant from Labex OSUG@2020 (Investissements d’avenir—ANR10 LABX56). Synchrotron experiments were performed on the ID15B beamline at the European Synchrotron Radiation Facility (ESRF), Grenoble, France. STEM experiments were performed at the LPS (Laboratoire de Physique des Solides, Orsay, France) in the frame of a research project granted to SG and FW by the national network for transmission electron microscopy and atom probe studies in France (METSA). This article benefited from comments and suggestions made by two anonymous reviewers and by Xionghan Feng (associate editor).
Compliance with ethical guidelines
Competing interests The authors declare that they have no competing interests.
- 1.Jürgensen A, Widmeyer JR, Gordon RA, Bendell-Young LI, Moore MM, Crozier ED (2004) The structure of the manganese oxide on the sheath of the bacterium Leptothrix discophora: an XAFS study. Am Miner 89:1110–1118Google Scholar
- 12.Lanson B, Drits VA, Silvester E, Manceau A (2000) Structure of H-exchange hexagonal birnessite and its mechanism of formation from Na-rich monoclinic buserite at low pH. Am Miner 85:826–838Google Scholar
- 13.Manceau A, Drits VA, Silvester E, Bartoli C, Lanson B (1997) Structural mechanism of Co2+ oxidation by the phyllomanganate buserite. Am Miner 82:1150–1175Google Scholar
- 24.Isaure M-P, Manceau A, Geoffroy N, Laboudigue A, Tamura N, Marcus MA (2005) Zinc mobility and speciation in soil covered by contaminated dredged sediment using micrometer-scale and bulk-averaging X-ray fluorescence, absorption and diffraction techniques. Geochim Cosmochim Acta 69:1173–1198CrossRefGoogle Scholar
- 30.Manceau A, Tommaseo C, Rihs S, Geoffroy N, Chateigner D, Schlegel M, Tisserand D, Marcus MA, Tamura N, Chen Z-S (2005) Natural speciation of Mn, Ni, and Zn at the micrometer scale in a clayey paddy soil using X-ray fluorescence, absorption, and diffraction. Geochim Cosmochim Acta 69:4007–4034CrossRefGoogle Scholar
- 64.Silvester E, Manceau A, Drits VA (1997) Structure of synthetic monoclinic Na-rich birnessite and hexagonal birnessite; II, Results from chemical studies and EXAFS spectroscopy. Am Miner 82:962–978Google Scholar
- 65.Buatier MD, Guillaume D, Wheat CG, Herve L, Adatte T (2004) Mineralogical characterization and genesis of hydrothermal Mn oxides from the flank of the Juan the Fuca Ridge. Am Miner 89:1807–1815Google Scholar
- 75.Post JE, Veblen DR (1990) Crystal structure determinations of synthetic sodium, magnesium, and potassium birnessite using TEM and the Rietveld method. Am Miner 75:477–489Google Scholar
- 77.Burns RG, Burns VM (1977) Mineralogy of manganese nodules. In: Glasby GP (ed) Marine manganese deposits. Elsevier, New YorkGoogle Scholar
- 78.Burns RG, Burns VM (1979) Manganese oxides. In: Burns RG (ed) Marine minerals. Mineral Society of America, Washington, D.CGoogle Scholar
- 92.Hammersley A (1997) FIT2D: an introduction and overview. European Synchrotron Radiation Facility Internal Report ESRF97HA02TGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.