Impact of Zn Substitution on Fe(II)-induced Ferrihydrite

Iron oxide minerals are ubiquitous in soils, sediments, and aquatic systems and influence the fate and availability of trace metals. Ferrihydrite is a common iron oxide of nanoparticulate size and poor crystallinity, serving as a thermodynamically unstable precursor to more crystalline phases. While aging induces such phase transformations, these are accelerated by the presence of dissolved Fe(II). However, the impact of trace metals on Fe(II)-catalyzed ferrihydrite phase transformations at ambient temperatures and the associated effects on trace metal speciation has seen limited study. In the present work, phase transformations of ferrihydrite that contains the trace metal zinc in its structure were investigated during aging at ambient temperature in the presence of two different Fe(II) concentrations at pH 7. X-ray diffraction reveals that low Fe(II) concentration (0.2 mM) generates hematite plus minor lepidocrocite, whereas high Fe(II) concentration (1.0 mM) promotes the production of a magnetite-lepidocrocite mixture. In both cases, a substantial fraction of ferrihydrite remains after 12 days. In contrast, Zn-free ferrihydrite forms primarily lepidocrocite and goethite in the presence of 0.2 mM Fe(II), with minor hematite and a trace of ferrihydrite remaining. For 1.0 mM Fe(II), magnetite, goethite, and lepidocrocite form when Zn is absent, leaving no residual ferrihydrite. Transformations of Zn-ferrihydrite produce a transient release of zinc to solution, but this is nearly quantitatively removed into the mineral products after 1 hour. Extended X-ray absorption fine structure spectroscopy suggests that zinc partitions into the newly formed phases, with a shift from tetrahedral to a mixture of tetrahedral and octahedral coordination in the 0.2 mM Fe(II) system and taking on a spinel-like local structure in the 1.0 mM Fe(II) reaction products. This work indicates that substituting elements in ferrihydrite may play a key role in promoting the formation of hematite in low temperature systems, such as soils or sediments. In addition, the retention of zinc in the products of ferrihydrite phase transformation shows that trace metal micronutrients and contaminants may not be mobilized under circumneutral conditions despite the formation of more crystalline iron oxides. Furthermore, mass balance requires that the abundance and isotopic composition of iron oxide-associated zinc, and possibly other trace metals, in the rock record may be retained during diagenetic phase transformations of ferrihydrite if catalyzed by dissolved Fe(II).

of ferrihydrite phase transformation shows that trace metal micronutrients and contaminants may 46 not be mobilized under circumneutral conditions despite the formation of more crystalline iron 47 oxides. Furthermore, mass balance requires that the abundance and isotopic composition of iron 48 oxide-associated zinc, and possibly other trace metals, in the rock record may be retained during 49 diagenetic phase transformations of ferrihydrite if catalyzed by dissolved Fe(II). 50 51

INTRODUCTION 52
Ferrihydrite is a ubiquitous, naturally occurring iron oxide that is commonly found in soils 53 and aquatic environments (Combes et al., 1990;Jambor and Dutrizac, 1998;Cornell and 54 Schwertmann, 2003). It has high surface area and is nanocrystalline, producing greater reactivity 55 compared to other iron oxides (Hiemstra, 2013;Hiemstra, 2015). The fate and transport of metal 56 ions is often controlled by sorption to ferrihydrite and other iron oxide minerals ( incorporation (Schultz et al., 1987;Ainsworth et al., 1994;Ford et al., 1997). In addition, metals 60 present during ferrihydrite formation may coprecipitate into the mineral structure (Martínez and 61 McBride, 1998;Ford et al., 1999;Dai et al., 2017). 62 Ferrihydrite is thermodynamically unstable with respect to other more crystalline iron 63 oxides such as lepidocrocite, goethite, and hematite (Navrotsky et al., 2008), and transforms over 64 time to these more stable phases (Schwertmann and Taylor, 1972;Cornell and Schwertmann, 65 2003). Such transformations play essential roles in determining iron oxide mineralogy in soils 66 (Kämpf and Schwertmann, 1983;Nørnberg et al., 2009;Jiang et al., 2018) as well as modern and 67 ancient sedimentary systems (Chan et al., 2007;Bekker et al., 2010). Ferrihydrite phase 68 transformations involve competitive processes influenced by various factors, such as temperature, 69 ligand type, and pH (Fischer and Schwertmann, 1975;Schwertmann et al., 1999;Cornell and 70 Schwertmann, 2003). Hematite formation from a ferrihydrite precursor occurs rapidly at 71 temperatures of 50 to 100°C (Fischer and Schwertmann, 1975 the relative amounts of goethite and lepidocrocite produced, and the rate of ferrihydrite 91 ferrihydrite, a Zn-hematite standard, and the solid-phase products formed from reaction of Zn-207 ferrihydrite with 0.2 mM Fe(II) for 2 h and 12 d with 1.0 mM Fe(II) after 12 days. Preparation of 208 the Zn-hematite standard was described in a prior study . Data collection 209 was performed on beamline 12-BM-B at the Advanced Photon Source (APS) at Argonne National 210 Laboratory. The beamline employed a Si (111) fixed offset monochromator which was detuned 211 30% to reduce the harmonic content of the beam. Toroidal focusing and flat mirrors were used to 212 increase usable X-ray flux and further reduce harmonics, with focusing effects resulting in a beam 213 of ~700 µm diameter. Zn K-edge data were collected in fluorescence yield with a 13-element 214 energy-dispersive Ge detector. Aluminum foil was used to selectively reduce the Fe fluorescence 215 intensity in order to prevent detector saturation. An additional standard of Zn adsorbed to hematite 216 was prepared by reacting 0.2 mM zinc chloride with 4 g L -1 synthetic hematite in a 0.01 M sodium 217 chloride solution for 5 days at pH 7.5. Details of the hematite synthesis and sample preparation 218 follow procedures described previously (Frierdich et al., 2011). The Zn K-edge XAFS spectrum 219 of this additional sample was measured at APS beamline 20-BM-B using similar optics and 220 detector details as those described above. 221 The X-ray energy for all measurements was calibrated by setting the maximum in the first 222 derivative of the X-ray absorption near-edge structure spectrum of a Zn metal foil to 9659 eV for 223 the Zn K-edge. XAFS spectral scans were averaged using the Athena (Ravel and Newville, 2005) 224 interface to IFEFFIT (Newville, 2001). The normalized and background subtracted k 3 -weighted 225 EXAFS spectra of Zn were fitted to structural models in SixPACK (Webb, 2005) using phase and 226 amplitude functions generated from the structure of franklinite (Verwey and Heilmann, 1947) 227 using FEFF 7.02 (Ankudinov and Ravel, 1998). Spectra were fit in k-space over a range of 3.0 to 228 11.3 Å and in R-space from 1.0 to 4.0 Å for all samples and standards. The coordination number 229 (N), interatomic distances (R), σ 2 (a Debye−Waller-type factor based on a Guassian distribution 230 of interatomic distances), and ΔE0 were refined using nonlinear least-squares fitting. The 231 amplitude reduction factor (S0 2 ) was fixed at 0.9 for spectral fitting. Linear combination fitting in 232 select cases was conducted in Athena. 233 234

Characterization of Zn-ferrihydrite and Zn-free Ferrihydrite 236
XRD reveals that Zn-ferrihydrite contained no detectable crystalline impurities, yielding a 237 pattern consistent with that of 2-line ferrihydrite (Fig. 1a). Zn-free ferrihydrite generated an XRD 238 pattern indistinguishable from the Zn-substituted phase. The Zn content of the substituted 239 ferrihydrite determined by acid digestion was 2.2 mol.%, close to the target substitution level of 240 2.0 mol.%. The BET specific surface areas for Zn-free ferrihydrite and Zn-ferrihydrite are 297 241 m²/g and 233 m²/g, respectively. Structural analysis of the Zn K-edge EXAFS spectrum (Fig. 1b) 242 of Zn-ferrihydrite shows that Zn substitutes in tetrahedral configuration, as indicated by the Zn-O 243 interatomic distance and coordination number (Table 1). The Zn-Fe distance of ~3.45 Å (Table 1)  244 indicates corner-sharing between a zinc tetrahedron and an iron octahedron, a geometry also 245 consistent with zinc substituting into the proposed tetrahedral iron site in ferrihydrite (Michel et

Transformations in 0.2 mM Fe(II) 250
Upon contact with 0.2 mM dissolved Fe(II), Zn-ferrihydrite showed no evidence of 251 converting to more crystalline phases over the first few hours of reaction (Fig. 2). By 1 d of reaction, 252 initial hematite peaks appeared in the XRD pattern, with both hematite and lepidocrocite clearly 253 present after 3 d of reaction. These continued to grow in through the 12 d duration of the 254 experiment (Fig. 2). The broad ferrihydrite peaks persisted throughout much of the reaction, with 255 weak background features present at ~35° and ~63° 2θ. Rietveld refinement of the XRD data (Fig.  256 S2) quantified the crystalline phases formed and provided a semi-quantitative estimate of 257 ferrihydrite abundance (Fig. 3) using an empirical approach described in the Electronic Annex. 258 This analysis shows that hematite was the dominant crystalline product formed from Zn-259 ferrihydrite over the course of reaction, with lepidocrocite present as 10-20 wt.% of the crystalline 260 phases. Ferrihydrite transformed slowly and only partially converted to crystalline phases after 12 261 d of reaction, with more than 50 wt.% of the initial ferrihydrite remaining (Fig. 3). 262 263

Transformations in 1.0 mM Fe(II) 264
In a 1.0 mM Fe(II) solution, Zn-ferrihydrite transforms more rapidly (Fig. 2). After 2 h of 265 reaction, the first time point sampled, substantial lepidocrocite peaks are present. Magnetite peaks 266 appear in the XRD pattern after 1 d and become substantial components after 3 d of reaction. The 267 relative peak intensities of lepidocrocite and magnetite are stable between 5 and 12 d of reaction. 268 Similar to the 0.2 mM Fe(II) experiment, the data continued to have a background feature near 269 ~35° 2θ, suggesting that some ferrihydrite remained unreacted. Rietveld refinement of the XRD 270 data (Fig. S3) confirms that lepidocrocite dominated the crystalline products over the first day of 271 reaction but this evolved to a subequal mixture with magnetite that appeared to stabilize in relative 272 proportions by 5 d of reaction (Fig. 4). While Zn-ferrihydrite transformation was more rapid 273 compared to the 0.2 mM Fe(II) experiment, a substantial residual component (~35 wt.%) remained 274 after 12 d (Fig. 4). 275

Control Experiments 276
Control experiments were conducted to evaluate the role of Zn and Fe(II) in generating the 277 observed mineral products (Fig. 5). Aging Zn-ferrihydrite for 12 d in the absence of dissolved 278 Fe(II) yielded no detectable phase transformation, with the XRD pattern preserving the features of 279 2-line ferrihydrite (Fig. 5). Zn-free ferrihydrite reacted with 0.2 mM dissolved Fe(II) for 12 d 280 produced substantial lepidocrocite and goethite peaks and minor hematite peaks in the XRD 281 pattern (Fig. 5). Notably, goethite was absent in the Zn-ferrihydrite experiment under the same 282 conditions and aging time (Fig. 2). Full-pattern fitting via Rietveld refinement (Fig. S4) required 283 a mixture of lepidocrocite, goethite, and hematite to reproduce the data. This showed that the 284 crystalline products contained ~35 wt.% goethite, which did not form when the starting solid was 285 Zn-ferrihydrite. The lepidocrocite content of the crystalline phases was also substantially increased 286 for Zn-free ferrihydrite. In addition, hematite was only ~25 wt.% of the crystalline fraction of the 287 solids compared to >80 wt.% of the crystalline phases in the Zn-ferrihydrite experiment. Only ~10 288 wt.% ferrihydrite remained after reaction, compared to >50 wt.% for Zn-ferrihydrite. 289 Reaction of Zn-free ferrihydrite with 1.0 mM dissolved Fe(II) for 12 d also produced 290 distinct mineralogy compared to Zn-ferrihydrite. While a mixture of magnetite and lepidocrocite 291 still formed, goethite was also present, similar to the 0.2 mM Fe(II) control experiment. Magnetite 292 abundance was ~50 wt.% of the crystalline products, approximately the same percentage as in the 293 Zn-ferrihydrite experiment, but less lepidocrocite formed at the expense of goethite (Fig. 3). 294 Ferrihydrite was below detection limit, producing greater conversion of Zn-free ferrihydrite 295 compared to Zn-ferrihydrite. 296 297 298

Transformations in 0.2 mM Fe(II) 300
Zinc release and uptake and the removal of dissolved iron were monitored during the 301 transformation of Zn-ferrihydrite in the presence of 0.2 mM Fe(II). At the beginning of the reaction, 302 Zn is rapidly released to solution, with a dissolved concentration of ~20 µM Zn at the first time Zn-ferrihydrite. Within one day, the dissolved Zn concentration displayed a drastic drop. After the 318 third day of reaction, Zn concentration in the solution was ~7 µM, which is 3% of the initial Zn 319 concentration. More than 97% of Zn was retained in the solid phase. 320 The dissolved Fe concentration experienced a sharp drop during the first hour of reaction 321 followed by a slower decline to less than 10 µM by day 5, with >99% of the initial dissolved Fe(II) 322 partitioning into the solid phase. During the period of large declines in dissolved Fe(II) 323 concentration, the pH drifted down from pH 7.0 to as low as 6.6, despite the presence of a buffer 324 in the experiment. At each sampling point the pH was thus manually adjusted to back to 7.0±0.1 325 by dropwise addition of 0.1 M NaOH (Fig. S5). This adjustment was performed after removal of 326 an aliquot of the suspension for fluid and solid-phase analyses. 327 328

Control Experiments 329
Control experiments were carried out in parallel to both 0. The EXAFS spectra show that Zn coordination changes substantially during Fe(II)-355 catalyzed phase transformations of ferrihydrite (Fig. 7). During transformations induced by 0.2 356 mM Fe(II), the EXAFS spectrum of a sample collected after 2 h show little variation from the 357 initial unreacted Zn-ferrihydrite (Fig. 7). Structural analysis confirms that Zn coordination does 358 not detectably change over this time period (Table 1). In contrast, after 12 d of reaction with 0.2 359 mM Fe(II) the EXAFS spectrum is clearly different from the initial Zn-ferrihydrite. Spectral fitting 360 (Table 1) show that a mixture of octahedral and tetrahedral Zn is present and a second Zn-Fe 361 distance occurs at 2.97 Å. In addition, the longer Zn-Fe distance shortens slightly to 3.40 Å. These 362 distances are similar to those observed for Zn substituting into hematite (Table 1) in the EXAFS spectrum of Zn (Fig. 8). Notably, the spectrum has similar fine-structure as the 381 spectrum of franklinite, but with weaker oscillations. Structural model fitting (Table 1)  was not detected via XRD (Fig. 2). Linear-combination fitting reproduces the data well with a 387 mixture of franklinite and Zn-ferrihydrite (Fig. 8)  with all products consisting of ferric iron minerals. The higher concentration explored, 1.0 mM, 406 generated substantial magnetite via reaction between Fe(II) and ferrihydrite. Note that while the 407 fractional abundance of crystalline products of Zn-ferrihydrite transformation was dominated by 408 lepidocrocite at early times, the absolute abundance (Table S1) increases to 37±3 wt.% after 1 d 409 before stabilizing at 32±3 wt.% by 12 d. There thus appear to be little to no lepidocrocite 410 conversion to magnetite; lepidocrocite simply forms first with ample Zn-ferrihydrite remaining. 411 The absolute abundance of lepidocrocite at 1 d is affected by the amount of ferrihydrite identified 412 using our semi-quantitative method via Rietveld refinement, and the potential small decline in 413 absolute lepidocrocite abundance over time may reflect systematic errors in the analysis rather 414 than real changes in mineralogy. 415 416 4.2 Impact of Zinc on Ferrihydrite Transformation Pathways 417

Promotion of Low-Temperature Hematite Formation 418
While hematite is the most thermodynamically stable phase among the common iron 419 oxides in aqueous environments, its formation from ferrihydrite at ambient temperatures is slow 420 (Cornell and Schwertmann, 2003)  systems may be limited because it is unlikely that ferrihydrite in the environment will contain Zn 506 at the level (~2 mol.%) explored in the present study, except perhaps near weathering zinc sulfide 507 ore deposits. However, a wide array of substituting elements are commonly associated with iron 508 oxides in nature (Schwertmann and Cornell, 2000;Cornell and Schwertmann, 2003). The present 509 results suggest a general mechanism where substituting elements in ferrihydrite enhance the 510 conversion to hematite, helping to nucleate this phase at temperatures lower than is typically 511 observed in laboratory studies. While further study is warranted, impurities may play key roles in 512 promoting low-temperature hematite formation in the environment. 513 514

Potential Impact on Metal Stable Isotope Records 515
The and Catalano, 2012) may later enhance metal availability. It is unclear whether similar behavior 544 will be displayed by other bioessential metals, such as Co and Ni, as their solid-water partitioning 545 may be distinct from Zn because of differences in ionic radii and chemical properties. 546 547

548
The trace metal Zn alters the Fe(II)-induced transformation pathways of ferrihydrite and 549 partially incorporates into the resulting crystalline iron oxides. This study is the first to report that 550 Zn promotes the formation of hematite at 22°C, which occurs as the dominant reaction product for 551 systems containing 0.2 mM dissolved Fe(II). This suggests that impurity ions may be critical to 552 the formation of hematite in low-temperature environments. In addition, Zn inhibits both goethite 553 formation and the overall transformation of ferrihydrite to more crystalline phases. A small 554 fraction of the ferrihydrite-bound Zn is released to solution during reaction but is rapidly taken 555 back up into the solid phase, partially incorporating into hematite and magnetite. Near-quantitative 556 retention of Zn demonstrates that ferrihydrite phase transformations at pH 7 from interactions with 557 dissolved Fe(II) will preserve the Zn isotopic composition. The Zn to Fe ratio will also be 558 Zn K-edge XANES spectra of Zn-ferrihydrite, its reaction products, and associated standards, and 588 a       . Only the lattice parameters and a peak broadening term were allowed to vary, and 28 the values from this initial analysis were then fixed for all subsequent analyses. Next, a series of 29 mixtures of ferrihydrite and zinc oxide, including both pure endmembers, were analyzed via 30 Rietveld refinement as a two-component mixture. The recovered ferrihydrite abundances (Fig. S1a) 31 varied linearly with true abundance but overestimated the ferrihydrite content as its concentration 32 decreased. A calibration curve between fitted and actual ferrihydrite abundance was generated 33 using linear regression, with the uncertainties on the slope and intercept propagated through 34 calculation of actual ferrihydrite abundance for unknown samples. 35 The accuracy of this calibration curve was next tested using data collected for synthetic 36 mixtures of ferrihydrite and hematite. The values derived from fitting the XRD patterns of these 37 mixtures via Rietveld refinement and then applying the calibration curve described above 38 generally well-reproduced the true ferrihydrite abundance within ±5 wt.%, and within the one-39 sigma fitting uncertainty for all but one sample (Fig. S1b,c). Fitting uncertainties are likely 40 underestimated at low ferrihydrite abundances because the calibration curve was generated using 41 unweighted linear regression. While a weighted linear regression to account for the different 42 uncertainties derived from Rietveld refinement of the initial ferrihydrite-zinc oxide data would 43 provide a more accurate estimate of the confidence interval, it is not trivial to then propagate this 44 through the calculation to derive actual ferrihydrite abundances. The ferrihydrite determination 45 should thus be viewed as semi-quantitative, primarily because of underestimated uncertainties as 46 the ferrihydrite-hematite mixtures suggest minimal systematic bias in the analysis. The absolute 47 abundances of crystalline phases when ferrihydrite is present should thus also be considered semi-48 quantitative. The relative abundances of crystalline phases, however, remain quantitative.