Stabilization of Ferrihydrite and Lepidocrocite by Silicate during Fe(II)-Catalyzed Mineral Transformation: Impact on Particle Morphology and Silicate Distribution.

Interactions between aqueous ferrous iron (Fe(II)) and secondary Fe oxyhydroxides catalyze mineral recrystallization and/or transformation processes in anoxic soils and sediments, where oxyanions, such as silicate, are abundant. However, the effect and the fate of silicate during Fe mineral recrystallization and transformation are not entirely understood and especially remain unclear for lepidocrocite. In this study, we reacted (Si-)ferrihydrite (Si/Fe = 0, 0.05, and 0.18) and (Si-)lepidocrocite (Si/Fe = 0 and 0.08) with isotopically labeled 57Fe(II) (Fe(II)/Fe(III) = 0.02 and 0.2) at pH 7 for up to 4 weeks. We followed Fe mineral transformations with X-ray diffraction and tracked Fe atom exchange by measuring aqueous and solid phase Fe isotope fractions. Our results show that the extent of ferrihydrite transformation in the presence of Fe(II) was strongly influenced by the solid phase Si/Fe ratio, while increasing the Fe(II)/Fe(III) ratio (from 0.02 to 0.2) had only a minor effect. The presence of silicate increased the thickness of newly formed lepidocrocite crystallites, and elemental distribution maps of Fe(II)-reacted Si-ferrihydrites revealed that much more Si was associated with the remaining ferrihydrite than with the newly formed lepidocrocite. Pure lepidocrocite underwent recrystallization in the low Fe(II) treatment and transformed to magnetite at the high Fe(II)/Fe(III) ratio. Adsorbed silicate inactivated the lepidocrocite surfaces, which strongly reduced Fe atom exchange and inhibited mineral transformation. Collectively, the results of this study demonstrate that Fe(II)-catalyzed Si-ferrihydrite transformation leads to the redistribution of silicate in the solid phase and the formation of thicker lepidocrocite platelets, while lepidocrocite transformation can be completely inhibited by adsorbed silicate. Therefore, silicate is an important factor to include when considering Fe mineral dynamics in soils under reducing conditions.

Introduction

Secondary oxides and oxyhydroxides of iron (Fe) are important sorbents for nutrients and contaminants in soils, influencing the chemical speciation and behavior of these elements.[1−3] In sub- or anoxic soils, dissimilatory Fe(III) reduction generates aqueous Fe(II). Interactions between Fe(II) and secondary Fe oxyhydroxides, such as ferrihydrite (Fe5HO8·4H2O) and lepidocrocite (γ-FeO(OH)), catalyze mineral recrystallization and transformation to more crystalline Fe minerals.[4−6] During the interaction of Fe(II) with Fe oxyhydroxides, adsorbed Fe(II) can be oxidized to Fe(III), leading to the reduction of structural Fe(III) through electron transfer and its release as Fe(II) into solution.[7,8] Thereby, Fe atoms are exchanged between the solid and the aqueous phase.[7,8] Recently, Sheng et al.[9] suggested that the accumulation of a labile Fe(III) species on the ferrihydrite surface may lead to the nucleation of ferrihydrite transformation products. Ferrihydrite interaction with Fe(II) leads to its transformation to lepidocrocite, goethite (α-FeOOH), and/or magnetite (Fe3O4), depending on the reaction conditions.[4,5] Model experiments have shown that the mineral products from ferrihydrite transformation are influenced by the Fe(II) concentration,[4,5] the Fe(II)/Fe(III) ratio,[4,5] pH,[10,11] temperature,[11] the solid to solution ratio,[12] and the ligand and buffer choice.[4] For example, while Fe(II)-catalyzed ferrihydrite transformation at pH <7 favors the formation of lepidocrocite and goethite,[4,5] high Fe(II) concentrations and pH >7 lead to the formation of magnetite.[4,10] During the Fe(II)-catalyzed transformation of ferrihydrite to goethite and magnetite, lepidocrocite often occurs as an intermediate phase[4,5,13] but has rarely been studied on its own (e.g., refs (4), (6), (14), and (15)). Some studies have shown lepidocrocite transformation to goethite[4,14] or magnetite;[6,15] however, results published to date are inconclusive, most likely due to the differences between experimental conditions (e.g., pH, Fe(II)/Fe(III) ratios, ligand, and buffer choice)[4] as well as variations in lepidocrocite particle size.[6] These results show that the stability of lepidocrocite during the interaction with Fe(II) and its role as an intermediate phase during the Fe(II)-catalyzed transformation of ferrihydrite is still poorly understood and warrants further investigation.

In natural environments, oxyanions, such as arsenic, phosphate, and silicate, and organic matter are abundant and can influence the formation and transformation of Fe oxyhydroxides in soils.[13,16−21] For example, Fe(II)-catalyzed transformation of ferrihydrite is hindered in the presence of arsenic[16] and coprecipitated organic matter.[13] Additionally, during the biomineralization of ferrihydrite, adsorbed phosphate reduced the mineral transformation rate and altered mineral transformation pathways compared to phosphate-free ferrihydrite.[17] However, the role of silicate during mineral transformations is less studied. Aqueous orthosilicate (H4SiO4) is ubiquitous in soil and sediment pore waters since silicate anions are released by chemical weathering of silicate minerals. Silicate anions interact with the positively charged surfaces of Fe oxyhydroxides at circumneutral pH and can coprecipitate with ferrihydrite, resulting in smaller crystallite sizes compared to pure ferrihydrite.[19,22,23] During Fe(III) hydrolysis, silicate has been reported to bind to free corner sites of FeO6 octahedra, hindering the formation of Fe oxyhydroxide polymeric complexes,[24] which may explain the smaller ferrihydrite crystallite sizes when formed in the presence of silicate. Nonetheless, ferrihydrite can precipitate at Si/Fe ratios of up to 0.68,[19] where silicate is integrated into micropores of ferrihydrite aggregates but not incorporated into the mineral structure.[25] In contrast, lepidocrocite formation is hindered at Si/Fe ratios >0.015,[21] and silicate-bearing ferrihydrite (Si-ferrihydrite) will precipitate instead.[21,26,27] Therefore, aqueous silicate in natural systems is likely adsorbed to lepidocrocite surfaces, whereas it is integrated into ferrihydrite aggregates. The reactivity and transformation of Si-ferrihydrites were studied by Jones et al.[18] who reacted Si-ferrihydrite coprecipitates with Fe(II) (Fe(II)/Fe(III) = 0.4 and Si/Fe(III) = 0.68) and observed complete Fe atom exchange (IAE) within 1 week but no transformation of Si-ferrihydrite coprecipitates. Similarly, they found that the presence of aqueous silicate can strongly reduce the IAE and mineral transformation of ferrihydrite and lepidocrocite during the reaction with Fe(II).[18] However, the effects of solid-associated silicate at low (<0.2) Si/Fe ratios and low (≤0.2) Fe(II)/Fe(III) ratios are not well studied. Especially for lepidocrocite formation, recrystallization, and/or transformation, the implications of solid-associated silicate remain unclear. Furthermore, the spatial redistribution of silicate in the solid phase during these processes, which may impact trace element sorption (e.g., As[22]), has not been investigated to date. Because ferrihydrite and lepidocrocite are closely linked to the biogeochemical cycling of nutrients and contaminants, we need to understand their reactivity in natural systems.

Therefore, we (i) examined the effects of solid-associated silicate on the stability and reactivity of ferrihydrite and lepidocrocite upon the reaction with Fe(II) and (ii) tracked the release of silicate to the aqueous phase and its spatial distribution in the solid phase. To this end, we reacted pure and silicate-associated ferrihydrite and lepidocrocite with isotopically labeled 57Fe(II) at pH 7 for 4 weeks, tracking trends in aqueous Fe and Si and the IAE between aqueous Fe(II) and structural Fe(III). In comparison to previous studies,[4,6,18] we chose low Fe(II)/Fe(III) ratios (≤0.2) and low Si/Fe ratios (0.05–0.18), which we consider relevant to natural environments. To better understand the impact of silicate on mineral transformation products, we also followed the kinetics of mineral transformations and characterized the products in terms of mineral morphology and changes in spatial silicate distribution.

Materials and Methods

Mineral Synthesis and Characterization

All reagents were prepared with doubly deionized (DDI) water (>18.2 MΩ·cm, Milli-Q, Millipore, Merck), and minerals were synthesized from isotopically natural-abundant (NA) Fe salts (5.8% 54Fe, 91.7% 56Fe, 2.2% 57Fe, and 0.3% 58Fe).[28] Ferrihydrite (Fh) was synthesized using the method of Schwertmann and Cornell.[29] In brief, a 0.2 M ferric nitrate (Fe(NO3)3·9H2O) solution (pH ∼1) was titrated (836 Titrando, Metrohm) with 1 M NaOH (Titrisol, Merck) to pH 7.5 ± 0.1 under vigorous stirring at room temperature. The Si-ferrihydrite coprecipitates were synthesized after Cismasu et al.[19] with slight modifications. The synthesis procedure followed that of ferrihydrite but used a NaOH solution containing 16.7 or 66.7 mM Si, added as sodium metasilicate granules (Na2SiO3, Sigma-Aldrich). Appropriate volumes of silicate-doped NaOH were added to the ferric nitrate solution to reach nominal Si/Fe ratios of 0.05 and 0.25, respectively. The pH of the resulting suspension was monitored and adjusted with silicate-free NaOH for 1 h after completion of the titration, and the Si-ferrihydrite suspensions were equilibrated in the dark for 24 h. The (Si-)ferrihydrite precipitates were centrifuge-washed with DDI water three times, shock-frozen dropwise in liquid N2, freeze-dried, and gently homogenized with a mortar and pestle. The shock-freezing was applied to ensure a fast and homogeneous freezing process of the precipitates, which reduces the formation of large and dense ferrihydrite aggregates.[30] Lepidocrocite (Lp) was synthesized using the method of Schwertmann and Cornell[29] with slight modifications. A 0.2 M Fe(II) solution (300 mL) prepared from FeCl2 was titrated to pH 6.7–6.9 with 1 M NaOH at room temperature and oxidized under vigorous stirring and gentle purging with air (approximately 400 mL min–1) for ∼90 min. The precipitates were washed, shock-frozen, freeze-dried, and homogenized, as described above.

To obtain silicate-adsorbed lepidocrocite, 1 g of the dried lepidocrocite was resuspended in 200 mL of a 13 mM Si solution to reach a final Si/Fe ratio of 0.23. The silicate was spiked in the form of an alkaline stock solution (200 mM Si), derived from Na2SiO3 dissolved in DDI water, and the pH of the suspension was adjusted to 7.5 with 1 M HCl (Suprapur). Due to the high silicate concentration, visible white flocs (up to 3 mm) formed after 48 h of equilibration under gentle stirring. The flocs were removed by passing the suspension through a 200 μm sieve. The silicate-adsorbed lepidocrocite was washed and freeze-dried as described above. Total Fe and Si contents in the mineral phases were analyzed after dissolution in concentrated HCl at room temperature with inductively coupled plasma optical emission spectrometry (ICP-OES). All mineral phases were characterized by X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), scanning transmission electron microscopy (STEM), and multipoint N2-BET (Brunauer–Emmett–Teller)[31] surface area analysis. Details of these methods are given in Section S1 of the Supporting Information.

The XRD analysis confirmed the expected mineralogical composition of the pure and silicate-associated ferrihydrites and lepidocrocites, without detecting mineral impurities (Section S1). The Si/Fe ratios in the Si-ferrihydrites were 0.05 and 0.18, named FhSi5 and FhSi18, respectively. For the Si-adsorbed lepidocrocite, the Si/Fe ratio was 0.08, from here on named LpSi8. Secondary electron images showed that Fh and FhSi5 consisted of nanometer-sized, spherical ferrihydrite particles in dense micrometer-sized aggregates (Section S9), and Lp and LpSi8 crystallites were lath- and finger-shaped (Section S9), which agrees with previous observations for ferrihydrite[4,32] and lepidocrocite.[4,14,32]

57Fe(II) Spike Experiments

All solutions were degassed with nitrogen (N2) for at least 1 h prior to transfer into an anoxic glovebox (N2 atmosphere, <5 ppm O2). All solutions, chemicals, equipment, and precipitates were equilibrated in the glovebox for at least 1 day before use. Suspensions of Fh, FhSi5, FhSi18, Lp, and LpSi8 containing 15 mM Fe(III) in 85 mL of anoxic 3-(N-morpholino)propanesulfonic acid (MOPS) buffer (50 mM, pH 7) were prepared in 100 mL septum glass bottles (in triplicates) and equilibrated inside the glovebox overnight. The corresponding solid to solution ratios were between 1.4 and 1.7 g L–1, depending on the Fe content of the respective materials (Section S2). An Fe(II) stock solution was prepared by dissolving isotopically labeled zero valent Fe (95.08% 57Fe, Isoflex USA) in 3.5 M HCl (Suprapur) overnight. The solution was passed through a 0.45 μm nylon filter and diluted to 0.2 M Fe. The 57Fe(II) concentration in the stock solution was determined by the 1,10-phenanthroline method after Walter[33] using UV–visible spectroscopy (λ = 510 nm). Aliquots of the 57Fe(II) solution were added to the mineral suspensions to achieve final Fe(II) concentrations of 0.3 mM (low Fe(II)) and 3 mM (high Fe(II)), corresponding to final Fe(II)/Fe(III) ratios of 0.02 and 0.2 (for low and high Fe(II) treatments, respectively). The pH was measured in all suspensions immediately after adding the 57Fe(II). In the first 5 min, suspensions of the high Fe(II) treatment were amended with 1 M NaOH (<200 μL) to readjust pH. Thereafter, the suspension pH remained constant at 7.0 ± 0.1. All bottles were crimp-sealed with a butyl rubber stopper and shaken (∼15 rpm) in the dark in an overhead shaker outside the glovebox. Control bottles containing the mineral phases in buffer solution without the Fe(II) spike were included to consider mineral transformation or dissolution in the absence of Fe(II).

At 2, 6, 24, 72, 168 (1 week), and 672 h (4 weeks) after the 57Fe(II) spike, septum bottles were returned to the glovebox for aqueous and solid phase sampling. Reactors were vigorously shaken until the solid phase was completely resuspended before the crimp-sealed bottles were opened, and an aliquot of the suspension was passed through a 0.45 μm nylon filter. Samples for aqueous phase analyses were additionally passed through a 0.22 μm nylon filter. The residual solid phase (≥0.45 μm) on the filter was rinsed with 3 mL of DDI water and dried in the glovebox in the dark. Solid phase triplicates were combined for analyses by manual homogenization. After sampling, the bottles were again crimp-sealed, removed from the glovebox, and returned to the shaker.

Aqueous Phase Analyses

Aqueous Fe concentrations in the 0.22 μm filtrates were measured by ICP-OES, and the dominance of Fe(II) over Fe(III) in solution was confirmed in 4 week samples (median: 99% Fe(II)) by the 1,10-phenanthroline method.[33] Therefore, all reported values for Fe(II) refer to total aqueous Fe. Aqueous Si concentrations and Fe isotope ratios in the 0.22 μm filtrates were measured with triple-quadrupole inductively coupled plasma mass spectrometry (ICP-MS), and 57Fe isotope fractions (f57Fe) were calculated relative to the sum of 54Fe, 56Fe, 57Fe, and 58Fe (details and quality controls in Section S3).

Solid Phase Analyses

The reacted solid phases were analyzed by XRD, and mineral phase fractions were determined by Rietveld quantitative phase analysis of the diffractograms, where ferrihydrite was included as a mass-calibrated PONKCS[34] phase (Section S6).[10,13] The Fe isotope composition of the reacted solid phases was determined after dissolution in HCl by ICP-MS. For the Fe isotope analysis, approximately 10 mg of the sample was dissolved in 1 mL of concentrated HCl in 2 mL Eppendorf plastic tubes and subsequently diluted to 30 mL with DDI water. The morphology of the transformation products and spatial distribution of Fe and Si in the solid phases were studied on selected 4 week reacted samples using STEM coupled with energy-dispersive X-ray analysis (EDX; details in Section S9).

Results and Discussion

Aqueous Fe(II) and Fe Atom Exchange

Temporal trends in aqueous Fe(II) concentrations showed that Fe(II) was rapidly removed from solution in all suspensions (Figure A,B and Section S4). Within the first 2 h, >0.2 mM of the initially spiked 0.3 mM Fe(II) and ∼1 mM of the initially spiked 3 mM Fe(II) were removed from solution in the pure Fh and Lp suspensions. At the same time, the aqueous f57Fe decreased similarly rapidly in Fh and Lp suspensions in the low and the high Fe(II) treatment (Figure C,D). This indicates that sorbed 57Fe(II) was oxidized by the transfer of an electron to the sorbent solid phase, resulting in the reduction and release of NAFe(III) as NAFe(II) to the solution. Adsorption of the 57Fe(II) alone would not lead to a detectable change in the Fe isotope composition of the aqueous phase. Within 2 h of reaction in the low Fe(II) treatment, the aqueous f57Fe in Fh (10.8%) and Lp (5.5%) suspensions dropped sharply and approached the isotopic composition of the total system (f57Fe = 3.9%), demonstrating that rapid atom exchange between aqueous 57Fe(II) and structural NAFe(III) in the solid phase occurred (Figure C). Similarly, in the high Fe(II) treatment, the aqueous f57Fe in the Fh (27.2%) and Lp (38.0%) suspensions approached the isotopic composition of the total system (f57Fe = 17.6%) within 2 h (Figure D).

Figure 1

Aqueous iron (Fe) concentrations (A, B) and 57Fe isotope fractions (f57Fe; C, D) in suspensions spiked with 0.3 mM (A, C) and 3 mM Fe(II) (B, D). Gray solid lines in panels (C, D) represent the f57Fe of the initial aqueous phase (Ainit) and unreacted solids (Sinit), and the dashed gray line shows the calculated f57Fe of the total system (Systot). Error bars show the standard error of experimental triplicates, and errors <5% (A, B) and <1% (C, D) are smaller than the symbols and are not shown. 672 h = 4 weeks. This data and a detailed view of the first 6 h can be found in Section S4.

After the first 2 h, Fe(II) concentrations in the low Fe(II) treatment continued to decrease slowly, and after 4 weeks, the Fe(II) concentrations were <40 μM in the Fh and Lp suspensions (Figure A). The aqueous f57Fe in low Fe(II) Fh and Lp suspensions matched the isotopic composition of the total system at 4 weeks (Figure C). In the high Fe(II) treatments, the Fe(II) concentration decreased sharply in Fh and Lp suspensions starting between 6 and 24 h and between 72 and 168 h, respectively (Figure B). After 4 weeks, Fe(II) concentrations were <20 μM in the Fh and Lp suspensions. The aqueous f57Fe after 4 weeks in the high Fe(II) Fh and Lp suspensions was lower than the isotopic composition of the total system (Figure D), suggesting that the solid phase became enriched in 57Fe, potentially in surface-associated crystalline products from Fh and Lp transformation.[13]

Coprecipitated silicate in FhSi5 and FhSi18 did not result in differences in Fe(II) removal from solution or IAE in the low Fe(II) treatment compared to pure Fh, whereas adsorbed silicate on LpSi8 resulted in slightly reduced Fe(II) adsorption and slower IAE (Figure A,C). In the high Fe(II) treatment, Fe(II) removal was similar for silicate-associated minerals (FhSi5, FhSi18, and LpSi8) compared to their corresponding pure mineral phases (Fh and Lp) in the first 6 h (Figure B). The impact of silicate became clearer after 24 h, where Fe(II) concentrations in the silicate-associated mineral suspensions (FhSi5, FhSi18, and LpSi8) remained relatively stable until 4 weeks, compared to a decrease in Fe(II) in pure Fh and Lp suspensions. After 4 weeks of reaction in the high Fe(II) treatment, the Fe(II) concentrations in the FhSi5 and FhSi18 suspensions were 2.1 and 1.8 mM, respectively. In comparison, the Fe(II) concentration in theLpSi8 suspension was 2.3 mM after 4 weeks (Figure B). The extent of IAE in the high Fe(II) treatment was similar for FhSi5 compared to pure Fh and only slightly reduced for FhSi18 (Figure D). Similarly rapid IAE has been reported for pure ferrihydrite[13] and Si-ferrihydrite coprecipitates up to a Si/Fe ratio of 0.68.[18] In contrast, IAE was strongly hindered by surface-adsorbed silicate in LpSi8, where the aqueous f57Fe was 76.3% at 4 weeks in the high Fe(II) treatment, indicating that only a minor fraction of aqueous 57Fe(II) exchanged with LpSi8 and that adsorbed silicate can reduce lepidocrocite reactivity (Figure D). Collectively, these results show that IAE is similarly rapid in pure Lp compared to pure Fh and that IAE is only slightly reduced by coprecipitated silicate in ferrihydrite (up to Si/Fe = 0.18) yet is strongly hindered by the presence of surface-adsorbed silicate in lepidocrocite.

Aqueous Silicate

Silicate was released from FhSi5, FhSi18, and LpSi8 to the aqueous phase within 2 h of reaction in both the low and the high Fe(II) treatment (relative Si release = 3% (FhSi5 and FhSi18) or 30% (LpSi8); Section S5). Within 4 weeks, the relative Si release from FhSi5 increased to 14 and 8% in the low and the high Fe(II) treatment, respectively. The aqueous Si fractions in FhSi18 suspensions, however, remained relatively stable after 2 h (3–5%). In comparison, the relative Si release from LpSi8 within 4 weeks of reaction in the low (58% of Si) and the high (46% of Si) Fe(II) treatment was much higher than from FhSi5 and FhSi18. This is likely related to the fact that silicate adsorbed to external mineral surfaces in LpSi8 but adsorbed to both external and internal surfaces (e.g., in micropores) of FhSi5 and FhSi18 aggregates during coprecipitation. Therefore, the silicate in FhSi5 and FhSi18 aggregates could only be released by mineral dissolution or micropore diffusion of silicate out of the aggregates, whereas silicate on the LpSi8 surface may desorb more rapidly. Furthermore, based on the specific surface area of each mineral (Section S1), the estimated relative coverage of the mineral surface by silicate was higher in the LpSi8 (∼19 μmol Si m–2) than in FhSi5 (∼4 μmol Si m–2) and FhSi18 (∼13 μmol Si m–2). As the BET specific surface area may be impacted by the aggregation of particles during flash-freezing and freeze drying,[30] the stated Si surface loadings might be slightly overestimated. However, the higher surface loading of LpSi8 likely facilitated silicate release despite some polymerization of silicate on the mineral surfaces, as shown by infrared spectroscopy (Section S1).

Mineral Transformation in the Absence of Silicate

The mineral fractions in reacted samples determined from Rietveld analysis of XRD data showed that, after the addition of Fe(II), Fh transformed into crystalline mineral phases within a few hours (Figure and Section S7). Within 6 h of reaction in the low Fe(II) treatment, traces of lepidocrocite (∼1%) were observed in the XRD patterns (e.g., small peaks at 14.14° 2θ (d = 6.269 Å) and 27.05° 2θ (d = 3.296 Å); Figure A). Thereafter, Fh transformation to lepidocrocite continued, reaching a maximum lepidocrocite contribution in the low Fe(II) Fh suspension of 10% at 1 week before decreasing to below the detection limit at 4 weeks. In contrast, goethite started to form within 24 h and dominated the transformation products until the end of the experiment, where 66% of Fh had transformed to goethite (Figure A). In the 3 mM Fe(II) treatment, the higher Fe(II) concentration facilitated faster transformation of Fh and enabled magnetite formation (Figure D). Lepidocrocite formation from Fh already started within less than 2 h in the high Fe(II) treatment and was the dominant transformation product until it reached its maximum at 24 h (38%). Thereafter, lepidocrocite contributions decreased at the expense of goethite and magnetite formation. During 4 weeks of reaction in the high Fe(II) treatment, Fh transformed to a mixture of lepidocrocite (19%), goethite (19%), and magnetite (28%), with 35% ferrihydrite remaining (Figure D). Based on the rapid IAE recorded in the Fh suspensions (Figure C,D), this remaining ferrihydrite may have recrystallized. The onset of magnetite accumulation between 6 and 24 h agrees well with the rapid removal of aqueous Fe(II) starting after 6 h (Figures B and 2D). Hansel et al.[4,35] suggested that magnetite accumulation can make the dissolution reprecipitation reaction from ferrihydrite to goethite less favorable due to the greater thermodynamic stability of magnetite. This agrees with our observation that the goethite fraction in the high Fe(II) Fh suspension only increased from 10 to 16% between 24 and 72 h, while the magnetite fraction increased from 8 to 26% (Figure D). After 72 h, the mineral composition in the high Fe(II) Fh suspension remained relatively constant until the end of the experiment, with 35–37% ferrihydrite, 19–25% lepidocrocite, 13–19% goethite, and 26–28% magnetite (Figure D). In general, the transformation of Fh into lepidocrocite, goethite, and magnetite during reactions with Fe(II) agrees well with the rapid IAE (Figure C,D).

Figure 2

Mineral phase fractions of solid samples before and during the reaction with 0.3 mM Fe(II) (A–C) and 3 mM Fe(II) (D–F) at six time points (h = hours, w = weeks) over 4 weeks (= 672 h), derived from X-ray diffraction patterns. Fractions <5% are not labeled but are reported with Rietveld fit data in Section S7. Main diffraction peaks are labeled with mineral abbreviations: Fh = ferrihydrite, L = lepidocrocite, G = goethite, M = magnetite. The marked XRD patterns were scaled by 5 (*) or 10 (**) in this figure. Corresponding figures for FhSi18 and LpSi8 are found in Section S7. Iron(II)-free control reactors showed no signs of mineral transformation (Section S7).

The Lp did not transform to more crystalline mineral phases in the low Fe(II) treatment over 4 weeks (Figure C). Combined with extensive Fe(II) uptake and rapid IAE recorded for Lp in the low Fe(II) treatment (Figure A,C), this suggests that Lp underwent recrystallization rather than transformation. Mineral recrystallization has also been reported for Fe(II) interactions with other crystalline Fe minerals (e.g., goethite[36,37] and hematite[38]). In contrast, during the reaction in the high Fe(II) treatment, 38% of Lp transformed to magnetite within 4 weeks without indication of goethite formation (Figure F). Whether lepidocrocite undergoes recrystallization versus transformation may be controlled by the supersaturation of labile Fe(III) on the lepidocrocite surface.[9,12] The transformation of lepidocrocite to magnetite occurs through a dissolution reprecipitation mechanism[15] and agrees with observations of magnetite accumulation during lepidocrocite interactions with Fe(II) at higher Fe(II)/Fe(III) ratios than used in this study (e.g., 2 at pH 6.5 in Pedersen et al.[6] versus 0.02 and 0.2 at pH 7 in this study; Section S2). However, the absence of goethite formation from lepidocrocite contrasts the strong dominance of goethite in transformation products originating from Fe(II)-reacted Fh in the low Fe(II) treatment. During Fh transformation, goethite may have nucleated from Fh and then, due to its greater thermodynamic stability, continued to grow at the expense of lepidocrocite. These results indicate that goethite nucleates more easily from Fh than from Lp under the given experimental conditions (Section S2), whereas magnetite can be formed from both Fh and Lp. This may be explained by the lower nucleation energy barrier for goethite formation from ferrihydrite than from lepidocrocite and the lower nucleation energy barrier for magnetite than goethite formation from lepidocrocite.[39] Few studies have reported goethite formation from lepidocrocite. For example, Yan et al.[14] found goethite and magnetite formation from lepidocrocite at low Fe(II)/Fe(III) ratios (0.02–0.06) but with strongly drifting pH values (7–3.7). Hansel et al.[4] found goethite but no magnetite formation from lepidocrocite-coated sand at high Fe(II)/Fe(III) ratios (0.8 and 8) at pH 7.2. However, this also demonstrated that the amount of goethite formed from lepidocrocite can vary widely (4–32%) depending on the Fe(II) concentration, ligand choice (chloride and sulfate), and buffer type (PIPES and bicarbonate).[4] Collectively, these results suggest that goethite formation from pure lepidocrocite is less favorable than from ferrihydrite and that ferrihydrite may provide the main goethite nucleation sites in mixed mineral systems.

Mineral Transformation in the Presence of Silicate

The mineral phase fractions during and after the reaction of Si-ferrihydrites (FhSi5 and FhSi18) with Fe(II) showed that coprecipitated silicate hindered the Fe(II)-catalyzed formation of crystalline minerals in both Fe(II) treatments and at both silicate loadings (Figure B,E and Section S7). The reaction of FhSi5 in the low Fe(II) treatment for 4 weeks resulted in the transformation to lepidocrocite (44%), with lepidocrocite first detected at 6 h (Figure B). In the high Fe(II) treatment, lepidocrocite (47%) was the major transformation product, with traces of goethite (4%) appearing at 4 weeks (Figure E). The reaction of FhSi18 in the low Fe(II) treatment resulted in no mineral transformation, while after the reaction in the high Fe(II) treatment for 4 weeks, traces of lepidocrocite (4%) and goethite (2%) were present (Section S7). In combination with the higher relative Si release from FhSi5 than from FhSi18 (Section S5), this suggests that the larger extent of mineral transformation (∼45% in FhSi5, <7% in FhSi18) led to higher Si release. The LpSi8 remained stable, with no mineral transformations recorded even in the high Fe(II) treatment (Section S7). In these suspensions, mineral transformation was hindered by surface-adsorbed silicate in LpSi8, which reduced Fe(II) adsorption and hindered IAE (Figure ).

The remaining ferrihydrite fractions at 4 weeks ranged between 56% (low Fe(II)) and 49% (high Fe(II)) for FhSi5 and between 100% (low Fe(II)) and 93% (high Fe(II)) for FhSi18, compared to 34% (low Fe(II)) and 35% (high Fe(II)) for pure Fh (Figure A,B,D,E and Section S7). The similar fractions of ferrihydrite remaining in comparable samples despite the increase in Fe(II)/Fe(III) ratios (from 0.2 to 2) indicate that the extent of ferrihydrite transformation was strongly influenced by the silicate loading of ferrihydrite, while the increase in the Fe(II) concentration had only a small effect.

In the FhSi18 suspension, ≥93% of the solid phase remained ferrihydrite even after 4 weeks of reaction in the low and the high Fe(II) treatment, yet Fe isotope data showed that FhSi18 underwent significant IAE (aqueous f57Fe deviated <6% from f57Fe of the total system after 4 weeks; Figure C,D). This indicates that mineral recrystallization of FhSi18 likely occurred despite the high silicate loading. Similar results were recorded for FhSi5, where within the first 2 h, the f57Fe dropped sharply (to 6.9% in low Fe(II) and 36.6% in the high Fe(II) treatment), yet only ≤7% of FhSi5 transformed to lepidocrocite (Figures C,D and 2B,E and Section S7). During the recrystallization process, silicate seemed to hinder nucleation and crystal growth of goethite and magnetite, most likely by hindering the polymerization of iron octahedra.[24] This resulted in ferrihydrite and, at a low silicate content (FhSi5), lepidocrocite as the only mineral transformation products. Recent research suggests the formation of a labile Fe(III) species on the ferrihydrite surface.[9] Considering this, silicate may have inhibited the polymerization of labile Fe(III) and therefore the formation of goethite and magnetite. Further, it has been shown that the presence of silicate can, depending on its concentration, hinder the precipitation of lepidocrocite and goethite, favoring ferrihydrite formation instead.[21,40] This is supported by Voegelin et al.[26] and Kaegi et al.[27] who demonstrated that abiotic Fe(II) oxidation at a Si/Fe ratio of 0.13 leads to the formation of poorly crystalline lepidocrocite (46%) mixed with a poorly crystalline silicate-rich hydrous ferric oxide phase (38%) and only small amounts of goethite (13%). They observed lepidocrocite formation only at initial aqueous Si/Fe ratios <0.24, suggesting that ferrihydrite and the silicate-rich hydrous ferric oxide phase precipitate until sufficient silicate has been removed from solution to enable lepidocrocite formation from excess Fe.[26] Thus, it is plausible that recrystallizing FhSi5 and FhSi18 continued to coprecipitate with aqueous silicate and continued to be stabilized against Fe(II)-catalyzed transformation, while lepidocrocite, if it forms, may only contain minor amounts of silicate.

Morphology of Transformation Products

The secondary electron images of 4-week reacted Fh (high Fe(II)) and FhSi5 (low and high Fe(II)) show that the morphology of lepidocrocite formed from Fe(II)-reacted ferrihydrite differed depending on the presence of both silicate and the spiked Fe(II) concentration (Figure A–F). With electron microscopy results, it needs to be considered that images can only represent a small fraction of the respective sample and therefore serve as indications. During the reaction of pure Fh in the high Fe(II) treatment, lepidocrocite crystallized as platelets and goethite formed clusters of small needles (Figure A), which agrees with previous observations for lepidocrocite[4,32] and goethite[4,14] morphology. In the presence of silicate, the reaction of FhSi5 in the low Fe(II) treatment led to large micrometer-sized lepidocrocite platelets (Figure C) and to large rectangular, thick platelets with rounded edges (Section S9). The latter contained spherical ferrihydrite particles inside and was covered by a smooth surface, which we suggest may be a crystallization front (Figure D and Section S9). The reaction of FhSi5 in the high Fe(II) treatment resulted in nanometer-sized lath-like lepidocrocites growing outward from ferrihydrite aggregates (Figure F) and independent, micrometer-sized, rough lath-like lepidocrocites with small, perpendicularly growing platelets (Figure E). These differences in morphology suggest that the growth of the lepidocrocite crystallites is strongly altered by the presence of silicate and Fe(II).[27]

Figure 3

Secondary electron images of Fh reacted with 3 mM Fe(II) (A, B), FhSi5 reacted with 0.3 mM Fe(II) (C, D), FhSi5 reacted with 3 mM Fe(II) (E, F), unreacted Lp (G), and Lp reacted with 3 mM Fe(II) (H); all reacted samples are 4 week samples, and scales are given in the respective images. The arrows point toward a <1 μm lepidocrocite platelet with a poorly defined crystal habit (A) and a micrometer-sized lepidocrocite platelet with a well-defined crystal habit (C). Additional images of FhSi5 reacted with 0.3 mM Fe(II) and of unreacted Fh and FhSi5 are found in Section S9.

The unreacted Lp consisted of finger-like structured small platelets with little morphological variation (Figure G). After the reaction in the high Fe(II) treatment, magnetite formed and residual lepidocrocite appeared shorter and thicker (Figure H). Southall et al.[41] and Joshi and Gorski[42] observed similar changes in goethite morphology during the reaction with Fe(II) and attributed this to preferential dissolution at goethite tips and preferential oxidation at long side edges. The secondary electron images of reacted and unreacted Lp suggest a similar mechanism for Lp during the interaction with Fe(II).

Morphological differences between the lepidocrocite crystallites formed via lepidocrocite synthesis versus through Fe(II)-catalyzed (Si-)ferrihydrite transformations were reflected in the crystallite thickness (Sections S8 and S9). Lepidocrocite thickness estimations were derived from fits of the 020 peak broadening in XRD patterns and refer to the crystal domain size perpendicular to the (010) plane[43] (details and the crystal structure in Section S6 and SEM images for comparison in Section S9). The thinnest lepidocrocite crystallites were observed for unreacted and Fe(II)-reacted Lp and LpSi8 (22–34 nm). In contrast, lepidocrocite crystallites that formed from FhSi5 in the low Fe(II) treatment were on average 185 nm thick, more than twice as thick as lepidocrocite crystallites formed from pure Fh (88 nm). This difference was less pronounced in the high Fe(II) treatment, where lepidocrocite crystallites at 4 weeks were only slightly thicker when formed from FhSi5 (87 nm) than when formed from pure Fh (68 nm). The determination of the crystallite thickness of FhSi18-derived lepidocrocite was prevented by the small amount of lepidocrocite in the respective samples (<5%). The differences in thickness of lepidocrocite crystallites formed from Si-associated ferrihydrite (FhSi5) compared to pure Fh are likely related to the hindering effect of silicate on lepidocrocite crystallization.[21,26,27] The edge surfaces ((100) and (101)) of lepidocrocite platelets are the most proton-reactive and, therefore, are preferential adsorption sites for Fe(II)[44] and potentially also for silicate.[45] If adsorbed silicate blocks these reactive sites, then the lepidocrocite may be less reactive toward Fe(II) adsorption and oxidation. The face surfaces (010) of lepidocrocite platelets are less proton-reactive and are, in perfect crystals, not reactive at all since all surface groups are doubly coordinated (≡Fe2OH).[44] However, mineral formation in the presence of impurities often leads to changes in the crystal structure,[46−49] and silicate has been reported to reduce linkages between FeO6 octahedra in Fe oxyhydroxide polymeric complexes.[24] Therefore, for the lepidocrocite formed from FhSi5, the face surfaces (010) of the lepidocrocite platelets likely had more surface imperfections than lepidocrocite formed from pure Fh. This would suggest that the face surfaces (010) of the FhSi5-derived lepidocrocite platelets were more reactive toward Fe(II) adsorption and oxidation, promoting the lepidocrocite growth perpendicular to the face surface (010). This may explain the increase in the lepidocrocite thickness when formed from ferrihydrite silicate coprecipitates (FhSi5) during interactions with Fe(II). In summary, these findings suggest that coprecipitated silicate can lead to the formation of thicker lepidocrocite crystallites during Fe(II)-catalyzed transformation of ferrihydrite.

Silicate Redistribution within the Solid Phase

The elemental distribution maps of unreacted FhSi5 and FhSi18 showed that the distribution of Si and Fe was fairly homogeneous throughout the Fe mineral phases (Figure A,B and Section S9). This is in line with the Si-ferrihydrite coprecipitation process described by Dyer et al.[25] with silicate adsorbing to the rapidly forming small ferrihydrite particles prior to aggregation. In silicate-adsorbed lepidocrocite (LpSi8), Si was present on the lepidocrocite surfaces, with no separate Si-rich particles visible (Figure C).

Figure 4

Elemental distribution maps of iron (Fe) and silicon (Si) in unreacted FhSi5 (A), FhSi18 (B) and LpSi8 (C), and in FhSi5 reacted with 0.3 mM Fe(II) (D, E) and reacted with 3 mM Fe(II) (F) for 4 weeks. Labels in panel (D) refer to the newly formed lepidocrocite (Lp) and remaining and/or recrystallized ferrihydrite (Fh). Arrows in panel (F) are pointing toward small platelets growing from a lath-like lepidocrocite. Corresponding HAADF images can be found in Section S9.

After 4 weeks of reaction with 0.3 mM Fe(II), much more silicate was associated with ferrihydrite than with the newly formed lepidocrocite, resulting in an uneven Si distribution within the samples (compare Figure A with Figure D,E). The Si content was also lower in smooth surfaces on ferrihydrite aggregates compared to the rough, poorly crystalline edges and spherical ferrihydrite particles inside the aggregates (Figure E). In lepidocrocite formed from FhSi5 in the high Fe(II) treatment, Si was enriched in small platelets growing at the edge of the lath-like lepidocrocite, compared to the rest of the crystallite (Figure F). The Si distribution within residual FhSi5 or recrystallized ferrihydrite in the low Fe(II) treatment remained similarly even as in unreacted FhSi5 (Figure D). Generally, the lower content of Si in newly formed lepidocrocite than in initial FhSi5 agrees with the hindered crystallization of lepidocrocite at high Si/Fe ratios. This suggests that the retention of Si in the recrystallizing ferrihydrite enabled the formation of lepidocrocite from Fe(II)-reacted FhSi5.[26]

Environmental Implications

Our results demonstrate that coprecipitated silicate impacts the stability of ferrihydrite under reducing conditions without decreasing its reactivity in terms of Fe(II) adsorption, electron transfer, and atom exchange. Since aqueous silicate is, among other oxyanions, ubiquitous in soil and sediment pore waters (0.1–0.5 mM Si[50]), it can influence the formation and transformation of Fe oxyhydroxides in soils. This interaction leads to the formation of silicate-associated iron (oxy)hydroxides, e.g., ferrihydrite with common Si/Fe ratios of ∼0.1 and up to 0.35.[51,52] The outcomes of this study suggest that, especially at higher silicate loading (Si/Fe = 0.18), silicate can stabilize ferrihydrite against mineral transformation in soils under Fe-reducing conditions, allowing the persistence of ferrihydrite even under conditions where ferrihydrite would not be expected. At lower silicate contents (Si/Fe = 0.05), Si-ferrihydrite transformation can result in lepidocrocite, which not only can be stabilized by the small amounts of associated silicate but also forms imperfect crystallite surfaces, potentially providing more reactive surface sites. However, during Si-ferrihydrite transformation, Si redistribution can lead to the accumulation of silicate in recrystallizing ferrihydrite, further increasing the ferrihydrite stability. Additionally, this study shows that pure lepidocrocite may be more stable than previously assumed, if Fe(II) concentrations remain low. Alternatively, lepidocrocite may become inactivated by adsorbed silicate, which hindered Fe(II) adsorption and thus limited IAE. However, our results further suggest that ferrihydrite could act as a nucleation site for goethite in mixed mineral systems, potentially inducing lepidocrocite dissolution due to the greater thermodynamic stability of goethite. Collectively, these findings imply a dual role of silicate during the formation and transformation of Fe oxyhydroxides in soils, not only favoring ferrihydrite over lepidocrocite formation during the oxidation of Fe(II)[21,26,27] but also stabilizing both ferrihydrite and lepidocrocite against mineral transformation under Fe-reducing conditions. Because Fe oxyhydroxides are important sorbent phases that influence the biogeochemical cycling of nutrients and contaminants, the altered mineral transformation dynamics in the presence of silicate anions need to be considered when assessing the iron mineralogy in Fe-reducing soil environments.