Ni isotope fractionation during coprecipitation of Fe(III)(oxyhydr)oxides in Si solutions

: The dramatic decline in aqueous Ni concentrations in the Archean oceans during the Great Oxygenation Event is evident in declining solid phase Ni concentrations in Banded Iron Formations (BIFs) at the time. Several experiments have been performed to identify the main removal mechanisms of Ni from seawater into BIFs, whereby adsorption of Ni onto ferrihydrites has shown to be an efficient process. Ni isotopic measurements have shown limited isotopic fraction during this process, however, most experiments have been conducted in simple solutions containing varying proportions of dissolved Fe and Ni as NO3 salts, as opposed to Cl salts which are dominant in seawater. Further, Archean oceans were, before the advent of siliceous eukaryotes, likely saturated with amorphous Si as seen in the interlayered chert layers within BIFs. Despite Si being shown to greatly affect the Ni elemental partitioning onto ferrihydrite solids, no studies have been made on the effects of Si on the Ni isotope fractionation. Here we report results of multiple coprecipitation experiments where ferrihydrite precipitated in mixed solutions with Ni and Si. Ni concentrations in the experiments ranged between 200 and 4000 nM for fixed concentrations of Si at either 0, 0.67 or 2.2 mM. The results show that Si at these concentrations has a limited effect on the Ni isotope fractionation during coprecipitation of ferrihydrite. At 0.67 mM, the saturation concentration of cristobalite, the isotopic fractionation factors between the precipitating solid and experimental fluid are identical to experiments not containing Si (0.34 ± 0.17‰). At 2.2 mM Si, and the saturation concentration of amorphous silica, however, the Ni isotopic composition of the ferrihydrite solids deviate to more negative values and show a larger variation than at low or no Si, and some samples show fractionation of up to 0.5‰. Despite this seemingly more unstable fractionation behaviour, the combined results indicate that even at as high concentrations of Si as 2.2 mM, the δ60Ni values of the forming ferrihydrites does not change much. The results of our study implicate that Si may not be a major factor in fractionating stable Ni isotopes, which would make it easier to interpret future BIF record and reconstruct ocean chemistry. of 0.1 Following 30 minutes of equilibration, the experiments were stopped and the contents were split into two ~50 ml fractions and immediately centrifuged to separate the ferrihydrite precipitates from the supernatant. One of the ferrihydrite fractions was kept as a solid for mineral determination using X-ray diffraction, and the other was dissolved in 20 ml 0.5M HNO for and Ni isotopic analyses.

Prior to the onset of large-scale global mantle convection after the Archean to Proterozoic Transition, low vertical mixing of the mantle restricted the capture and burial of elements (Andrault et al., 2017), resulting in higher elemental concentrations in the Archean oceans. It has been hypothesized that the depletion of nickel (Ni) in early Earth oceans, as a result of the onset of global mantle convection, had a direct negative influence on the strongly nickeldependent methanogenic biosphere, indirectly leading to the onset of the 'great oxygenation event' (GOE) (Konhauser et al., 2015(Konhauser et al., , 2009. It was suggested that successive removal of essential trace elements led to a dramatic decline in abundance of methanogens, paving the way for oxygen-producing cyanobacteria and the concomitant onset of atmospheric oxygenation. Although valuable clues can be found in Ni/Fe variability in banded iron formations (BIFs) which may have formed as a consequence of a gradual atmospheric oxygenation, anoxygenic photosynthesis or C-P-O-Fe cycling, both the source of the early Earth atmosphere and its composition is highly debated (Andrault et al., 2017;Bekker et al., 2010;Holland et al., 1986;Konhauser et al., 2017;Marakushev and Belonogova, 2019;Ozaki et al., 2019;Thibon et al., 2019). BIFs are chemical sedimentary deposits (usually of Archean and Paleoproterozoic age) characterized by layers of iron oxides alternating with chert layers.
The mechanisms of formation of these deposits are still debated and several explanations have been put forward, such as the formation through oxidation of dissolved Fe as a consequence of photosynthetic, cyanobacterial production of molecular oxygen (Cloud, 1973). Other, purely abiotic explanations have been suggested, such as the oxidation of ferrous iron to ferrihydrite or magnetite through UV oxidation or by reduction of carbon dioxide to methane (Thibon et al., 2019). Several abiotic processes show a preferential uptake of light Ni isotopes, such as the adsorption and coprecipitation of Ni with ferrihydrites (Eusterhues et al., 2011;Twidwell and Leonhard, 2008;Wasylenki et al., 2015). Ferrihydrite, a primary Fe(III)oxide, was likely the first Fe-oxide phase to form as the ferrous ocean oxidised and produced the extraordinary BIF deposits (Posth et al., 2013;Wasylenki et al., 2015). The large specific surface area of ferrihydrites promote adsorption of trace elements and possibly the fractionation of their stable isotopes. Several experiments have been conducted to evaluate the magnitude of abiotic fractionation of Ni isotopes in ferrihydrites, through adsorption and coprecipitation experiments (Gueguen et al., 2018;Wang and Wasylenki, 2017;Wasylenki et al., 2015). The studies demonstrated a strong coupling between iron oxide precipitation and light Ni, with fractionation factors between the experimental solution and ferrihydrite of ~+0.35‰. However, these fractionation factors were determined from experiments conducted in solutions containing only Fe and Ni dissolved in water or dilute NaNO 3 . The Archean J o u r n a l P r e -p r o o f 1. Introduction ocean is hypothesised to have had dissolved silicon, Si, concentrations as high as 2.2mM (Konhauser et al., 2009;Jones et al., 2015), evidenced by the microcrystalline quartz layers alternating with the iron bands within BIF deposits. The presence of Si has been shown to have a large effect on the partitioning of Ni into ferrihydrites (Konhauser et al., 2009), but no studies have yet been made on the effects on Ni isotopes during this process. Therefore, as a first step to investigate the effects of Si on Ni isotope fractionation, we conducted multiple coprecipitation experiments to investigate the isotopic fractionation of Ni sorption when ferrihydrite forms in mixed solutions with Si and Ni.

Mineral synthesis
All coprecipitation experiments were conducted in a clean lab at the Swedish Museum of Natural History in Stockholm, Sweden. The experiments were performed at room temperature (~24°C) in acid cleaned plastic beakers (PP) with constant agitation by a Teflon coated magnetic stirrer. Stock solutions of FeCl 2 , NiCl 2 and Si (as sodium metasilicate nonahydrate) were prepared by diluting concentrated solutions and powders with Milli-Q (18.2 MΩ) water.
The stock solutions were diluted further with Milli-Q for each experiment to a final experimental volume of 100 ml. The final solutions had an Fe concentration of 180 μM, Si concentrations of either 0, 0.67 or 2.2 mM and Ni concentrations ranging between 100 and 4000 nM, that covers the suggested Archean oceanic dissolved Ni concentrations ranging from approx. 200-400nM (Konhauser et al. 2009 , Table 1).
Ferrihydrite was synthesised through rapid hydrolysis of Fe 2+ by raising the solution pH by dropwise addition of 1M NaOH to a pH of 7. The pH was monitored throughout the experiment with a pH meter (Orion model 250A). Once the pH was stable, the solution was left to equilibrate for 30 minutes. The pH of the solution was not allowed to deviate more than ±0.1 from pH 7 and was controlled by adding small amounts (generally 5 to 10 μl) of 0.1 M HCl or NaOH. Following 30 minutes of equilibration, the experiments were stopped and the contents were split into two ~50 ml fractions and immediately centrifuged to separate the ferrihydrite precipitates from the supernatant. One of the ferrihydrite fractions was kept as a solid for mineral determination using X-ray diffraction, and the other was dissolved in 20 ml 0.5M HNO 3 for elemental and Ni isotopic analyses.
J o u r n a l P r e -p r o o f

X-Ray diffraction methods
Single crystal X-ray diffraction (XRD) was used to verify phase identities of ferrihydrites synthesised in the experiments. The XRD analyses were conducted on the same day as the precipitation experiments to avoid possible aging or alteration of the minerals. The analyses were performed on a Bruker D8 diffractometer using CuKα radiation, incidentand diffracted-beam Soller slits, and a Bruker Sol-XE detector. The samples were mounted on thin glass fibres attached to brass pins and mounted onto goniometer heads. Data were analysed using Bruker's EVA program.

Ni isotope analyses
Chemical purification of Ni and subsequent Ni isotope measurements were conducted at the Ifremer Centre de Bretagne in Brest, France, and follow the methods of Gueguen et al. (2013).
A double spike method was used where a 61 Ni-62 Ni spike (containing equal proportions of 61 Ni and 62 Ni) was added to all samples prior to chemical processing. The double spike allows correction of instrumental mass bias occurring during analysis as well as possible fractionation arising from less than 100% yields during column chromatography.
Samples and standards (NIST SRM 986) containing Ni concentrations of 0.5 to 1 μg Ni were evaporated on a hotplate. The spike was then added at a Ni sample /Ni spike ratio of 1 and reevaporated to dryness, ready to be taken up by the column elution matrix. Procedural blank measurements yielded Ni concentrations of 3-4 ng and are considered negligible compared to the high Ni concentrations of the processed samples.

Sample purification
The relatively simple matrix of ferrihydrite solids and experimental fluids allowed for a single step column procedure to be used, utilizing the Eichrome Ni specific resin (second column step in Gueguen et al., 2013). The Ni-spec resin contains a dimethylglyoxime (DMG) molecule, which at pH >8 forms a Ni-DMG complex which is retained in the resin. The dry sample-spike residues were dissolved in 1 ml 0.24M HCl, after which the pH was increased by the addition of 0.3 ml 1M ammonium citrate and 125μl concentrated ammonia. Between each addition step, 15 minutes were allowed for homogenization of the solution.
Approximately 0.5 ml wet Ni-spec resin was loaded into polypropylene columns. The resin was washed with 6.5 ml MilliQ and 4.5 ml of 0.2M ammonium citrate+0.45M ammonia. The J o u r n a l P r e -p r o o f columns were then capped and an additional 0.5 ml of the 0.2M ammonium citrate +0.45M ammonia solution was added to buffer the resin pH to ~9. The prepared sample and NIST standard solutions were then loaded onto the column and the resin was resuspended, increasing the active surface area and rate of complexation. The suspension was then left for 1 hour. The Ni yields through the column were close to 100% implying that the 1-hour equilibration time was sufficient. After 1 hour the columns were left to drain, and 2.2 ml of the 0.2M ammonium citrate+0.45M ammonia solution was added twice, in intervals of 15 minutes. This step washes out Fe and other matrix elements. Thereafter, 2.2 ml of MilliQ was added three times in intervals of 15 minutes, to wash out excess ammonium citrate and ammonia salts. Ni was then collected in 6.6 ml of 3M HNO3, which was added three times as 2.2 ml with 15 minutes between each time. The collected Ni cuts were then dried down on a hot plate at ~95°C overnight, and then refluxed in concentrated HNO 3 for 24 hours to fully oxidise and break down the Ni-DMG complex. Following evaporation, the sample residue was then dissolved in 1 ml of 0.3M HNO 3 ready for isotopic analysis.

Isotopic measurements
Ni isotopes were measured on the Thermo Scientific Neptune multi-collector inductively coupled mass spectrometer (MC-ICP-MS) at the Ifremer Centre de Bretagne in Brest, France.
The analyses were made in medium resolution with simultaneous measurements of 58 Ni, 60 Ni, 61 Ni and 62 Ni, as well as 57 Fe to monitor the isobaric interference of 58 Fe on 58 Ni. Each sample was measured with 50 cycles of 4 seconds integration time, separated by acid blank measurements of 10 cycles with 4 seconds integration time. Samples were introduced using an ApexQ MicroFlow PFA-50 self-aspirating nebulizer at an injection rate of ~60µl/min. The sample and skimmer cones are made of Ni, but 58 Ni intensities of the blanks indicate no Ni is contributed from the cones, as demonstrated previously (Fujii et al., 2011;Gueguen et al., 2013;Moynier et al., 2007).

Ni isotope data reduction
Nickel isotope compositions of samples are expressed using the delta (δ) notation where the isotopic ratio of the sample is normalised to that of the NIST SRM 986 standard, following: (1) J o u r n a l P r e -p r o o f where X denotes either the 60 Ni, 61 Ni or 62 Ni isotope ( 64 Ni was not considered in this study).
The data presented in this study is described using the 60 Ni/ 58 Ni ratio given as δ 60 Ni.
Prior to double spike calculation, the 58 Ni intensity was corrected for interference of 58 Fe following: where 58 Fe/ 57 Fe Natural is the ratio of the natural abundances of the 58 Fe and 57 Fe isotopes. The intensity of 58 Fe is then subtracted from the measured 58 Ni signal.
The nickel isotopic compositions of samples were then calculated using the double spike calculation template from Gueguen et al., 2013.
NIST SRM 986 was passed through columns four separate times yielding an average value and external reproducibility of the method of -0.01 ± 0.08‰.

X-Ray diffraction
Powder X-ray diffraction of the solid phase showed two peaks around 36 degrees and 63 degrees characteristic for and assigned to the (110) and (115)

Ni isotopic composition of coprecipitated solids and fluids
The Ni isotopic compositions of the ferrihydrite solids ranged from -0.37 to +0.02‰, while fluids were isotopically heavier with δ 60 Ni values ranging from -0.08 to +0.26‰. The isotopic composition of both fluids and solids increase with increasing fraction of Ni incorporated in the solids (Figure 2). The fraction of Ni sorbed to ferrihydrite in our experiments ranged from 0.32 to 0.86, however there was no systematic distinction between the varying Si concentrations (Table 1, Figure 2).

J o u r n a l P r e -p r o o f
The apparent fractionation factor between the residual fluid and forming ferrihydrite solids, Δ 60 Ni solution-solid (where Δ 60 Ni solution-solid = δ 60 Ni solution -δ 60 Ni solid ), ranges from -0.08 to +0.52‰, with an average of +0.28±0.33‰ (2 stdev, n=14). The variability is caused by two sets of samples which show the opposite sense of fractionation with solids being slightly heavier than fluids (Table 1, Figure 2). The mechanism of fractionation of these particular samples is discussed in section 4.2. Excluding these samples, the mean value is +0.34 ± 0.17‰, in good agreement with previous coprecipitation and adsorption experiments not containing Si (Wasylenki et al., 2015;Wang and Wasylenki, 2017;Gueguen et al., 2018).  (Figure 3).

Ferrihydrite precipitation in the presence of Cl
The main results of this study agree with the aggregated conclusions of previous Ferrihydrites from nitrate does not reach the same amount of Fe incorporated into the ferrihydrites even after 200 hours (Hiemstra et al., 2019). Since these reaction pathways are different in terms of uptake, the Ni fractionation factor could as well be influenced by this process, since it is commonly dependent of the Ni concentration (Hiemstra et al., 2019).
However, the similarities of our results and those of previously conducted experiments in nitrate environments suggest that either the resulting particle sizes were similar in the different environments, or that potential differences had little impact in the resulting Ni

Ni isotopic fractionation mechanism in the presence of Si
To test which fractionation mechanism best explains the distribution of our coprecipitation All models use -0.05‰ as the starting composition of the experimental fluid. Wang and Wasylenki (2017) showed that coprecipitated ferrihydrite consists of both surface adsorbed and incorporated Ni and that these pools have distinct isotopic compositions with δ 60 Ni values increasing in the order adsorbed < solution < incorporated. They determined that the fractionation factors between the solution and adsorbed Ni (Δ solution-adsorbed ) is ~+0.3 to +0.44‰ based on pure adsorption experiment from Wasylenki et al. (2015) and that the fractionation factor between the solution and incorporated Ni (Δ solution-incorporated ) is -0.18‰ (Wang and Wasylenki, 2017). The data is fitted using global models where fluid and solid data are regressed simultaneously. For the linear and Rayleigh models, the best-fit fractionation factor is modelled against measured δ 60 Ni and F Ferrihydrite values. The linear model gives best-fit Δ 60 Ni solution-solid of +0.32 ± 0.02‰ (1se) with a global R 2 of 0.90, while the Rayleigh model (following Wasylenki et al, 2015) gives +0.17 ± 0.02 ‰ (1se) and R 2 of 0.86.
For the three-pool Ni model the fractionation factors given by Wang and Wasylenki (2017) are used in the first instance but do not fit our dataset as well as the simpler models. Using higher pH promotes more adsorption of Ni onto the ferrihydrite particles (e.g. Gueguen et al., 2018). However, all samples in this study were synthesised at the same pH (7±0.1). Wang and Wasylenki (2017) show that pH, and in extension the Ni/Fe ratio of the solid have the largest effect on the ratio of adsorbed and incorporated Ni. However, the Ni/Fe ratio of these two sets of samples do not explain why they may have relatively more Ni incorporated. Overall, the simpler linear equilibrium isotope fractionation model best fit our dataset as a whole, yielding J o u r n a l P r e -p r o o f a best-fit value of Δ solution-solid of +0.32‰, and we cannot systematically distinguish between adsorbed or incorporated Ni. Although Si may influence the relative proportions of Ni being adsorbed and incorporated into the solid ferrihydrite structure, our data and modelling suggest that addition of Si did not appear to significantly change the Ni isotopic fractionation mechanism during ferrihydrite coprecipitation.

Implications for Archean seawater reconstruction using BIF record
Although previous studies have investigated the influence of Fe oxide precipitation on the fractionation of Ni in order to understand the formation environment of BIFs (Gueguen et al., 2018;Wang and Wasylenki, 2017;Wasylenki et al., 2015), the combined influence of Si and Fe precipitation on the Ni isotope fractionation has not previously been investigated. Konhauser et al. (2009) demonstrated that the presence of Si greatly reduced the partitioning behaviour of Ni into precipitating ferrihydrite particles, but no isotopic analyses were made.
The rationale for our added Si concentrations are, as in Konhauser et al. (2009), based on the assumption that Si was not readily removed from the seawater in the Precambrian oceans, due to the absence of Si-precipitating organisms. The source and sinks of Si to the oceans were therefore purely abiotic and thus, coprecipitation and adsorption may have had a much larger impact on the global Si cycle (Siever, 1992(Siever, , 1957. We used Si concentration saturations with respect to cristobalite (0.67 mM) and amorphous Si ( It is unclear at present why this is, but since coprecipitation involves Ni being both incorporated into the structure and surface adsorption it could be due to changing ratios of these processes. Wang and Wasylenki (2017) hypothesise that the fractionation factors are J o u r n a l P r e -p r o o f significantly different between adsorbed and incorporated Ni, and a more detailed study is currently undertaken to study this under the influence of Si. J o u r n a l P r e -p r o o f

Conclusions
The aggregated results from this study show that Ni isotope fractionation during coprecipitation of Fe(III)(oxyhydr)oxides in Si solutions show best fit lines for equilibrium fractionation for fluids as shown also by previous studies. The presence of Si seemed to influence the fractionation factor so that higher concentrations of Si resulted in a larger fractionation factor. However, the explanation for this result requires a more extensive study that is currently underway. The use of FeCl 2 rather than Fe(NO 3 ) 3 did not seem to have an effect on the fractionation factor or amount of Ni being incorporated into solids, even though using different reactants have been shown to have an impact on the particle surface area and

Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.