Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanoparticles

Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanoparticles

Accepted Manuscript Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanop...

5MB Sizes 0 Downloads 9 Views

Accepted Manuscript Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanoparticles Birger Rasmussen, Janet R. Muhling, Alexandra Suvorova, Bryan Krapež PII: DOI: Reference:

S0301-9268(16)30264-9 http://dx.doi.org/10.1016/j.precamres.2016.07.003 PRECAM 4546

To appear in:

Precambrian Research

Received Date: Revised Date:

23 February 2016 15 June 2016

Please cite this article as: B. Rasmussen, J.R. Muhling, A. Suvorova, B. Krapež, Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanoparticles, Precambrian Research (2016), doi: http://dx.doi.org/10.1016/j.precamres.2016.07.003

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Dust to dust: Evidence for the formation of “primary” hematite dust in banded iron formations via oxidation of iron silicate nanoparticles

Birger Rasmussen1*, Janet R. Muhling1,2, Alexandra Suvorova2 and Bryan Krapež1

1

Department of Applied Geology, Curtin University, Kent Street, Bentley WA 6102, Australia

2

Centre for Microscopy, Characterisation and Analysis, The University of Western Australia, Stirling

Highway, Crawley, WA 6009, Australia *Corresponding author - E-mail: [email protected].

ABSTRACT Conventional models for the deposition of banded iron formations (BIFs) envisage the oxidation of upwelled ferrous iron and the precipitation of ferric oxide/hydroxide particles in surface waters that settled to form laterally extensive layers of iron-rich sediment. A fundamental tenet of this model is that fine-grained hematite (so-called dusty hematite) in least-altered BIFs represents the dehydration product of original oxide/hydroxide precipitates. However, this premise has never been proven. We have investigated the origin of the earliest-formed iron oxides in chert in well-preserved BIFs of the 2.63-2.45 billion-year-old Hamersley Group, Australia. We find that laminated chert in BIFs show progressive stages of in situ alteration from grey-green chert, containing iron-silicate nanoparticles, to red chert with abundant hematite dust. Analysis

1

of textures by transmission electron microscopy of samples from the transition zone between grey-green and red chert reveals that dusty hematite formed after partial dissolution of iron-silicate nanoparticles by the precipitation of iron oxides in resulting cavities. These observations suggest that hematite dust is not a relict of an original seawater precipitate but the end-product of post-depositional oxidation. Our observations are consistent with paleomagnetic results from the Hamersley Group, which record two major phases of magnetic remanence carried by hematite that post-date deposition by more than 200 million years. Our results may provide an alternative explanation for the origin of jasper in BIFs deposited before the start of the Great Oxidation Event about 2.4 billion years ago. If correct, it follows that hematite dust is not a reliable proxy for paleoenvironmental conditions or biological processes in early Precambrian seawater. Furthermore, our results suggest that the primary iron precipitate in BIFs was iron-silicate mud that was silicified at or just below the sediment-water interface, a hypothesis that requires neither dissolved oxygen nor photosynthetic life, but was an inorganic, chemical process, reflecting anoxic oceans enriched in iron and silica.

Keywords: Banded iron formations; hematite; greenalite; Hamersley Group; Paleoproterozoic; ocean chemistry

1. Introduction Banded iron formations (BIFs) are interpreted to be chemical sediments comprising silica and iron-rich minerals precipitated from seawater. As such, they have the potential to record the evolution of the hydrosphere, atmosphere and biosphere

2

through Precambrian time. However, in order to correctly interpret the signals recorded by isotopes and trace elements in BIFs it is essential to determine what the original precipitates were and how they were modified after deposition. The challenge of identifying primary minerals is increased by the vulnerability of iron-rich minerals to changes in redox conditions, which can overprint paleoenvironmental and biological signatures during diagenesis, metamorphism and weathering (Chan et al., 2006). The original precipitates of BIFs are difficult to identify. Unweathered rocks comprise combinations of oxides (magnetite and hematite), carbonates (siderite, ankerite, dolomite and calcite) and silicates (greenalite, stilpnomelane, minnesotaite and riebeckite) with quartz (chert), but most of the grains are interpreted to be diagenetic or metamorphic. Weathered and mineralized BIFs are even further removed from the original mineralogy, and are dominated by secondary iron oxides. It is generally agreed that the deposition of BIFs required anoxic seawater to transport iron in large quantities, as dissolved ferrous iron, over large distances. The iron is generally thought to have been derived from mid-ocean ridge volcanism, although a contribution from continental weathering is also possible (Li et al., 2015). Conventional models propose the oxidation of ferrous iron that upwelled into the surface ocean, by photoferrotrophs or cyanobacteria, to produce particles of ferric oxides or hydroxides that settled to form the primary iron deposit in BIFs (Cloud, 1965, 1973; Konhauser et al., 2002, 2007; Kappler et al., 2005; Bekker et al., 2010; Sun et al., 2015). The oxidation of ferrous iron by ultraviolet radiation (Cairns-Smith, 1978) has been discounted as a major source of ferric oxides and hydroxides as it is considered incapable of producing the vast quantities of iron preserved in Archean and early Paleoproterozoic BIFs (Konhauser et

3

al., 2007; Pecoits et al., 2015). An alternative to the conventional models favors inorganic precipitation of iron silicates as the primary iron deposit in at least some BIFs (Rasmussen et al., 2015b), a model that is supported by experiments on analogs of Precambrian seawater rich in iron and silica (Harder, 1978; Konhauser et al., 2007; Tosca and Guggenheim, 2014; Tosca et al., 2016). It is possible that both processes contributed to the formation of BIFs, with one process dominant over the other at various times (e.g., before and after the GOE) and in different depositional settings (e.g., deep water vs shallow water). The origin of hematite is enigmatic but central to understanding how BIFs formed. Whereas some coarse-grained hematite is clearly secondary (Kaufman et al., 1990; Rasmussen et al., 2014a; Sun et al., 2015), hematite dust particles (typically <1 µ m) in chert are widely viewed as the products of primary iron deposition, primarily because of their small size (Spencer and Percival, 1952; James, 1954; Ayres, 1972; Dimroth and Chauvel, 1973; Ahn and Buseck, 1990; Beukes and Gutzmer, 2008; Sun et al., 2015). The occurrence of sub-micron-sized hematite particles “floating” in chert is generally accepted to record the co-precipitation of precursor iron oxide/hydroxide particles and silica from seawater (Beukes and Gutzmer, 2008; Sun et al., 2015) but textural evidence at the nanometer scale is lacking. Therefore, the challenge is to find samples containing hematite that have been least modified by post-depositional processes and use these to define mineral paragenesis based on preserved textural relationships. In this work, we examine the case for primary hematite in BIFs. Firstly, we review the historical evidence and arguments presented for the origin of hematite in iron formations. It is important to evaluate this evidence as it underpins conventional models

4

for the formation of BIFs. We then present our results, which are based on standard light and scanning electron microscopy complemented with in situ nanometer-scale charaterisation of hematite dust in BIFs from the Hamersley Group, Australia (Fig. 1). Our results have implications for the depositional models and the use of BIFs as proxies for paleoenvironmental conditions and biological processes during the early Precambrian.

2. Hematite in iron formations – A historical perspective “The origin of the iron-bearing rock has long been a subject of doubt, discussion and speculation, for the original characters are not plainly seen, even in a comparatively careful examination” (Spurr, 1894). These words are still relevant after more than 120 years and despite numerous studies on these enigmatic and remarkable rocks. Early workers interpreted the primary sediment of iron formations to have been an iron silicate mineral similar to glauconite (Spurr, 1894; Winchell et al., 1899) or greenalite (Leith, 1903). Iron oxides were largely viewed as oxidation products of ferrous silicates. The first comprehensive study of Lake Superior iron formations concluded that the original sediments comprised mainly greenalite, siderite and chert (Van Hise and Leith, 1911, p. 500). In contrast, Wolff (1917, p. 34, 35) in his study of ore deposits of the Mesabi Range, Minnesota, states: “The writer believes that the great mass of iron was laid down as original oxide (hematite and magnetite or an intermediate oxide), cemented together in an amorphous silica matrix. While it is true that under the microscope numerous oxidized greenalite granules can be seen, it does not follow that all the oxide grains were greenalite originally.” Until the 1950s, there was no consensus as to the mineralogy of the primary precipitates. Hematite was not widely accepted as an important primary

5

mineral in the Lake Superior region because, amongst other things, the amount of hematite in most iron formations is small (LaBerge, 1964), and it is a common product of later weathering and mineralization. In 1951, James reported three main iron phases (sulfide, carbonate and silicate) in iron formations from the Iron River district, Michigan. A fourth assemblage (oxides), dominated by magnetite and hematite, was postulated based on the reports of primary oxides in other iron districts. A year later, Krumbein and Garrels (1952) published a geochemical classification scheme for marine chemical sediments, including iron-rich sediments, based on Eh and pH values. They calculated stability fields for pyrite, hematite and siderite but not for iron silicates. Instead, based on natural assemblages, chamosite (representing all iron silicates) is shown co-existing with hematite and with siderite. Shortly afterwards, James (1954) published his influential paper on the four sedimentary facies of iron-formations: sulfide (pyrite), oxide (hematite and magnetite), carbonate (siderite) and silicate (greenalite). Although he acknowledges that “much of the rock [iron formation] characterized by iron oxides is due to later oxidation … the recognition of these processes [e.g., oxidation] and products [e.g., iron oxides] in no way negates the proposition that iron-rich rocks characterized by … oxides also were primary sedimentary types.” However, no petrographic evidence was presented to support this proposition. In the absence of new textural evidence for primary hematite, James (1954) cites a series of studies of Phanerozoic ironstones (Smyth, 1911; Hallimond, 1925; Taylor, 1949) and iron formations (Grout, 1919; Gruner, 1922; Gill, 1926, 1927; Geijer, 1938;

6

Dorr, 1945; du Preez, 1945), in which workers imply that some of the hematite is probably primary. However, the cited papers do not present petrographic proof for primary hematite (or silicates and carbonates), and indeed, many concede that the origin of the iron-rich sediments is uncertain. A key piece of evidence for primary hematite cited by James (1954) is a statement by Gruner (1946, p. 31): “Siderite is another important constituent which is interlayered with any of the silicates or oxides. Hematite occurring with siderite in this manner is unquestionably primary.” This assertion is not supported by further explanation or additional information. It is possible that the siderite formed after the hematite, but that does not prove that the hematite is primary, merely that it formed before the siderite. James (1954) goes on to say: “… the known presence of hematite layers in unoxdizied carbonate rock …. as described by Gruner (1946), demonstrates that hematite was definitely deposited in some areas. There is, therefore, no a priori reason for not assuming that at least part of the premetamorphic hematite is of primary origin; certainly such a facies is a predictable one ….” As an example of hematite oxide facies iron formation, James (1954, p. 259) states: “jaspilite of the Marquette range, a spectacular example, consists of interlayered, even bedded reddish jasper and fine-grained steel grey hematite …. The appearance of the jaspilite in outcrop is shown in Figure 13, although a black and white photograph does not do justice to this colorful rock”. His Figure 13 shows “jaspilite in outcrop at Jasper Knob, Ishpeming (Marquette range).” That outcrop, which is cited as a spectacular example of primary “hematite oxide facies iron formation”, contains finely laminated jasper and hematite-rich bands from the upper Negaunee Iron Formation (Fig.

7

2A-C). The iron formation has been complexly folded and brecciated, with microplaty and specular hematite in cross-cutting veins, bedding-parallel layers, fault breccias and cleavage planes. Petrographic examination of the iron formation shows that it mostly comprises chert, quartz, martite (i.e., hematite that has pseudomorphed magnetite), microplaty hematite, specular hematite and hematite dust. In thin section, the BIF is folded and cut by hematite veins (Fig. 2D), and contains rotated breccia fragments with crack-and-seal veins comprising hematite and quartz. The jasper bands contain finegrained hematite plates, which are commonly aligned with the dominant cleavage direction (Fig. 2E). The metallic steel grey bands contain microplaty hematite, which defines a crenulation cleavage, as well as martite crystals with elongate quartz pressure shadows parallel with the cleavage direction (Fig. 2F, G). Most of the hematite at Jasper Knob is clearly secondary. Rather than representing a spectacular example of primary hematite oxide facies iron formation, it appears to be an exceptional example of iron formation that has undergone hematite mineralization. This is borne out by the proximity of the outcrop to a number of hematite-martite deposits around Ishpeming, including a historic mine <100 m from the main outcrop. Nevertheless, James (1954) concluded “the preceding discussion makes it clear that primary hematite iron-formation is a recognized major rock type in both Precambrian rocks and younger rocks in many parts of the world”. James (1954) proposed his model before the recognition of mid-ocean ridge volcanism as the major source of iron for BIF and before models for oxygenation of the atmosphere and hydrosphere were developed. His model therefore does not explain how iron oxides and hydroxides derived by continental weathering, his preferred source of iron, were

8

dissolved and then re-precipitated in oxic water to form oxide facies BIF. Nevertheless, the paper by James (1954) led to a paradigm shift in the understanding of how iron formations were deposited, and the facies model was widely applied in studies of BIF (e.g., Beukes, 1973; Klein and Fink, 1976; Klein and Bricker, 1977), although no lateral variation of facies was recognized in the BIFs of the Hamersley Group (Trendall and Blockley, 1970; Morris, 1993). In order to explain the precipitation of iron oxides from an anoxic water column laden with reduced iron, Cloud (1973) proposed a new mechanism for the deposition of BIFs involving the upwelling of seawater enriched in dissolved ferrous iron, and oxidation in the photic zone by microbes. He stated: “I suggest, therefore, an interaction between primitive oxygen-releasing photosynthesizers and ambient ferrous iron. … iron in the ferrous state was thus converted to ferric iron and precipitated as ferric oxides or hydroxides. The alternation of iron-rich and iron-poor chert that comprises the usual oxide facies of BIF implies an episodic balance…”. With this paper, Cloud (1973) transformed the postulated oxide facies of James (1954) into the “usual oxide facies of BIF”. In 1982, Button et al. concluded in a State of the Art report on sedimentary iron deposits from the Dahlem Conference of that year: “In the past there has been little agreement on the genesis of cherty iron formation of the Superior and Hamersley-types. We were therefore surprised to find essential unanimity within the group regarding some of the most important aspects of the genesis of these sediments”. The favored model is essentially that of Cloud (1973), and it has remained so ever since. The work of Krumbein and Garrels (1952) and the facies model of James (1954) legitimized hematite as a primary mineral of BIF although textural evidence for a primary

9

origin has, to this day, never been presented. The most-cited evidence for a ferric oxide/hydroxide precursor is the presence of hematite dust (Spencer and Percival, 1952; Ayers, 1972; Dimroth and Chauvel, 1973; Beukes and Gutzmer, 2008; Sun et al., 2015), which fulfills the petrographic criteria articulated by LaBerge (1964) diagnostic of primary iron phases, including: i) uniform, fine grain size (typically <0.5 µm), ii) occurrence as individual particles, and iii) close association with, and delineation of, primary features such as bedding. The pivotal role of hematite dust in models of the deposition of iron formations can be gleaned from this statement in the review by Beukes and Gutzmer (2008): “… the observation that microcrystalline (“dusty”) hematite, with grain sizes <1 µm, is the paragenetically oldest iron mineral, leads to the conclusion that ferric oxyhydroxide represented the primary iron precipitate in all of the mineralogical facies of banded micritic iron formation macrocycles …. The most likely explanation is that this dusty hematite formed from dehydration of ferric oxyhydroxide particles that settled from suspension at the same time that gelatinous silica accumulated on the floor of the depository. Such primary ferric oxyhydroxides may have settled from suspension or formed along the sediment-water interface.” This interpretation of the origin of hematite dust underpins conventional models for the deposition of iron formations. Although hematite dust may fulfill the petrographic criteria for a primary mineral (e.g., LaBerge, 1964), it “could also be a product of supergene and hydrothermal alteration” (Beukes and Gutzmer, 2008, p. 9). This opinion is shared by Klein (2005, p. 1478), who cautioned: “… many BIFs with a red color are the result of hematite that is a secondary oxidation product.” However, no criteria exist for distinguishing hematite dust that is

10

interpreted to be primary from that which formed via weathering or hydrothermal alteration. A potential test of the origin of hematite dust is detailed examination of its paragenetic relationship with other minerals, using nanometer-scale microscopy.

3. Aims The aim of this study is to investigate the origin of hematite dust in samples of BIF from the Hamersley Group, Australia. Previous studies indicate that the primary minerals of BIF are very fine grained (<1 µm) and difficult to characterize by light microscopy. The techniques of modern nanotechnology were not available to early researchers of the origin of BIF and traditional petrographic studies to identify primary sediments were inconclusive. Recent developments in Focussed Ion Beam (FIB) technology, that allow targeted areas of thin sections to be excised for examination by Transmission Electron Microscopy (TEM), and the comparative accessibility and ease of operation of modern TEMs, have enabled the textural relationships of very fine grained minerals in BIFs to be routinely examined. In the study reported here, we use standard light techniques combined with scanning electron microscopy (SEM) and TEM to investigate nano-scale minerals in BIF. The study focuses on laminated chert in which hematite dust co-exists with iron-silicate nanoparticles, and seeks to establish whether the oxide and silicate minerals represent co-precipitates or parent-product phases of mineral replacement reactions.

4. Geological Background and Samples

11

The 2.63-2.45 Ga Hamersley Group contains some of the best-preserved and most-studied BIFs in the early Precambrian rock record (LaBerge, 1966; Trendall and Blockley, 1970; Ayres, 1972; Ewers and Morris, 1981; Morris, 1993; Krapež et al., 2003; Pecoits et al., 2009; Rasmussen et al., 2013, 2014a, 2015a, 2015b; Haugard et al., 2016). They are by far the largest, most extensive and thickest, and are only rivalled by the coeval Transvaal BIFs in South Africa (Trendall, 2002). Most of the Hamersley Group has undergone only very low- to low-grade metamorphism and mild deformation (Smith et al., 1982). Because of the great areal extent of the Hamersley BIFs and their high level of preservation, they are regarded as archetypes (Trendall, 2002). The BIFs are interpreted to be chemical sediments that were deposited when dissolved, deep-water ferrous iron upwelled into surface waters and was oxidized by photosynthetic organisms or UV photolysis (Trendall, 2002; Bekker et al., 2010). The main precipitates are interpreted to have been silica and ferric oxides/hydroxides that were deposited across the seafloor (Trendall and Blockley, 1970; Morris, 1993; Trendall, 2002). Krapez et al. (2003) argue that the original precipitates were subsequently reworked on the sea floor by bottom currents. Recent work by Rasmussen et al. (2013; 2015a; 2015b) suggests that the precursor sediments included iron-rich clay nanoparticles and microgranules that were cemented on the seafloor by amorphous silica during hiatuses in deposition. Sedimentological analysis suggests a basin-floor environment beyond a continental slope, in water depths possibly up to several kilometres (Krapež et al., 2003; Pickard et al., 2004). The micro-scale paragenetic history of the BIFs has been described previously (Trendall and Blockley, 1970; Ayres, 1972; Morris, 1980; Ewers and Morris, 1981;

12

Kaufman et al., 1990; Morris, 1993; Pecoits et al., 2009; Rasmussen et al., 2013, 2014a, 2015a, 2015b, Haugard et al., 2016), and the primary iron precipitates were considered to include silicates and siderite, as well as ferric oxides/hydroxides. After deposition, the hydrous, silica- and iron-rich mineral phases were recrystallized in response to increasing temperature and pressure, producing a diverse group of minerals, including quartz, iron oxides (hematite, magnetite), iron silicates (stilpnomelane, minnesotaite, riebeckite, greenalite), and carbonates (siderite, ferroan dolomite, ankerite, calcite). In order to address questions about the primary precipitates, this study targeted samples from drill-holes that were distal to iron-ore mines and at depths below the modern weathering profile. Samples of BIF and iron-rich chert were collected from three diamond drill-holes (ABDP9, Mitchell-2, Silvergrass), intersecting the Hamersley Group (Fig. 1). A fourth drill-hole, DDH44, which was collared near the Paraburdoo mine, was also sampled because of its historical significance. The drill-hole comprises a complete section of the Dales Gorge Member, which has been extensively sampled in numerous studies seeking to reconstruct the chemistry of the ancient ocean and the history of life. Despite its location, Ewers and Morris (1981) note: “throughout its length the core from DDH44 satisfied the criteria for freshness. No martite, goethite, or other evidence of oxidation or corrosion of silicates, carbonates, and apatites was noted.” Laminated chert was targeted because it formed shortly after deposition, is chemically inert and is relatively impermeable to fluids. It is interpreted to preserve depositional and early diagenetic structures such as bedding and shrinkage cracks in chert. These qualities make it an optimal target for studying the earliest mineralogical and textural components of BIFs (Rasmussen et al., 2015b).

13

5. Analytical Methods In this study, sedimentological logging of drill-core was followed by sampling for petrography and microanalysis. Polished thin sections were prepared and examined using transmitted light and reflected light microscopy, and backscattered-electron imaging (BSE) on TESCAN VEGA 3 (LaB6 source), Zeiss 55 Supra (field-emission source) and FEI Verios (field-emission source) SEMs located in the Centre for Microscopy, Characterisation and Analysis (CMCA) at the University of Western Australia (UWA). Each SEM is fitted with an Oxford Instruments X-Max energy dispersive X-ray spectrometer (EDS), which was used for quantitative and qualitative chemical analysis of mineral grains. Grains <1 µm are too small for quantitative analysis by EDS and for these particles spectra with only O, Si and Fe peaks in green chert were interpreted to be greenalite [Fe2-3Si2O5(OH)4], whereas grains with minor peaks for Mg, Al and K, in addition to O, Si and Fe, were interpreted to be stilpnomelane [K(Fe,Mg)8(Si,Al)12(O,OH)27]. Minnesotaite [Fe,Mg)3Si4O10(OH)2] is mostly coarser grained and can be analysed quantitatively. Foils for TEM studies were cut from areas of chert enclosing silicate nanoparticles located with BSE imaging in polished thin sections. FIB techniques using an FEI Helios NanoLab DualBeam instrument located at Adelaide Microscopy, the University of Adelaide, were used to prepare 100 nm thick TEM foils. Areas selected for TEM analysis were first coated with a strip of Pt ~1 µm thick to protect the surface, then trenches ~5 µm deep were milled on either side of the strip using a Ga ion beam with 30 kV voltage and 21 nA current. The foil was then cut away from the sample and welded to

14

an Omniprobe Cu TEM grid with a Kleindiek nanotechnik micromanipulator. The foils were thinned with the Ga ion beam at 30 kV and 0.28-0.92 nA, before cleaning at 5 kV and 47 pA, and polishing at 2 kV and 28 pA. TEM data were obtained at 200 kV using an FEI Titan G2 80–200 TEM/STEM with ChemiSTEM technology located at CMCA, UWA. Bright-field TEM and high-angle annular dark-field (HAADF) STEM images, and qualitative EDS spectra and maps were collected to identify the iron-rich nanoparticles. High-resolution TEM (HRTEM) and electron diffraction were used to confirm the identity of particles differentiated by HAADF and EDS data.

6. Petrographic evidence for origin of hematite Microcrystalline hematite (or hematite dust) is present in BIFs throughout the Hamersley Group, although it is less common in diamond drill-hole intersections than it is in outcrops or near hematite ore bodies. Light and scanning electron microscopy suggests that there are at least two types of hematite dust, which are recognized based on mineralogical affinities and textural characteristics: Type 1 is associated with the alteration of coarse-grained, iron-rich carbonates and silicates, and Type 2 occurs as finely dispersed particles in laminated chert. Type 1 hematite dust is localized around grain boundaries, within iron-rich minerals or along mineral cleavage planes of carbonate crystals (Fig. 3). It also occurs around the grain boundaries of iron silicate minerals, forming minute hematite particles (see Figs. 9 and 10 in Rasmussen et al., 2014a). The petrographic textures indicate that this generation of hematite dust formed after ferroan dolomite-ankerite rhombs, siderite

15

nodules and stilpnomelane, and is probably the product of late-stage fluid-mediated alteration. The second type of hematite dust, which occurs in laminated chert, is widely regarded as relicts of the original iron oxide/hydroxide precipitate. It has a uniform fine grain size (typically <1 µm), comprises individual particles, and is closely related to, or defines, primary features such as bedding. This combination of features is regarded as indicative of a primary origin (e.g., LaBerge, 1964). However, vertical and lateral transitions from green chert to red chert in iron formations (Fig. 4; Fig. 5C and 5D in Fischer and Knoll, 2009) suggest that hematite dust in laminated chert may have another origin. There are at least two possibilities: either the hematite dust is primary and was partly reduced to ferrous minerals some time after formation, or it is the product of secondary oxidation of pre-existing iron-bearing minerals. Light microscopy of finely laminated green chert shows the presence of densely packed, submicron-sized particles (Fig. 5). The dust particles are generally randomly distributed throughout the chert cement, comprising between 5-10% of it. SEM, HRTEM, EDS and electron diffraction analysis reveal that the particles are either stilpnomelane or greenalite (cf. Rasmussen et al., 2015b). In the best-preserved samples of chert, the ironsilicate nanoparticles are present as randomly oriented, evenly distributed plates, ranging in size from <50 nm to 500 nm (Fig. 5). The grain boundaries between the quartz cement and iron-silicate particles are straight or curved with no evidence of cavities (Fig. 5). In samples of green chert from the Dales Gorge Member (DDH44, drill-depth 513.25 m) hematite dust occurs in a set of vertical hairline fractures (~1 µm wide and 100

16

µ m long) that are perpendicular to bedding (Fig. 6). The green chert contains abundant iron-silicate nanoparticles, but the vertical fractures are delineated by trails of hematite dust. These textures indicate that in this sample the iron-silicate nanoparticles were replaced by hematite, probably as fluids infiltrated the hairline fractures. Transition zones between green chert and red chert occur on a scale of <1 mm (Figs. 3, 7, 8). These transition zones may be quite dramatic (Fig. 3) or subtle in appearance, depending on the density of dust particles and their mineralogy. Figure 7 shows a laminated chert from the Dales Gorge Member (DDH44, drill-depth 363.15 m), including individual laminae that are both green and red (Fig. 7C; Fig. 8). The boundary between the red and green chert is diffuse and crosscuts sedimentary lamination (Fig. 7AC). The green chert comprises a dense cluster of iron-silicate dust (Fig. 7G-I) whereas the red chert contains both hematite dust and iron-silicate nanoparticles (Fig. 7D-F). Typically, red chert contains minute pits and cavities on the surface of the polished thin section, which are associated with hematite dust particles. Laminated chert from the Brockman Iron Formation (Silvergrass, drill-depth 313.65 m) also contains both red and green laminae, as well as more diffuse bands of blue-grey riebeckite (Fig. 8A). Individual laminae may pass from green to red along the length of the lamina as in the sample from DDH44 (Fig. 8B-E). A TEM foil was cut from an area showing the presence of both hematite and silicate nanoparticles in BSE images (Fig. 9A). HAADF images also show two types of particles within chert: lamellae and hexagonal plates of high brightness (hematite), and lower brightness particles of more irregular shape (Fig. 9B-D). STEM EDS shows that these are iron silicate nanoparticles that seemingly “float” in chert cement. The silicates occur as fine-grained (<50 nm to

17

1000 nm), randomly oriented plates (Fig. 9). Most of the larger nanoparticles have one or more cavities in their cores (Fig. 9B-D), which comprise 50-90% of the particles. In contrast, hematite dust particles occur as equant to platy particles, between 100 nm and 700 nm in size (Fig. 9B-D), commonly occupying cavities in the iron-silicate nanoparticles. Locally, the larger hematite crystals extend beyond the walls of cavities into adjacent chert. HRTEM and electron diffraction show that the silicates in this sample are stilpnomelane and STEM EDS show that they contain minor Mg, Al and K in addition to Si, Fe, and O (Fig. 10). Figures 9 and 10 show that many of the cavities in the larger stilpnomelane nanoparticles have been partly filled with hematite euhedra. While the Fe from the dissolved stilpnomelane may have been incorporated into hematite, the element maps show no evidence for the reprecipitation of Mg, Al, and K, suggesting that these elements were transferred out of the zone of silicate dissolution. A sample of laminated chert from the Bee Gorge Member, Wittenoom Dolomite (ABDP9, drill-depth 219.27-.33 m) also contains a transition between green and red chert. In the transition zone, iron-silicate nanoparticles occur in chert between coarsegrained fans of minnesotaite, which is generally considered a late diagenetic or metamorphic phase in BIF (Fig. 11A-D). The minnesotaite has partly replaced laminated chert containing iron-silicate nanoparticles and carbonate. The relict pockets of ironsilicate dust are partly replaced by fine-grained hematite dust (Fig. 11C, D). The occurrence of hematite dust coincides with a darkening in the color of the adjacent ironsilicate nanoparticles (Fig. 11C). In the same interval, hematite occurs around the grain margins and interior regions of coarse-grained ferroan dolomite-ankerite rhombs (Fig.

18

3G-I; 11E-I). Some of the smaller carbonate rhombs have been almost completely dissolved, with irregular relict carbonate grains surrounded by, and impregnated with, hematite and magnetite (Fig. 11I). In some thin sections from this interval, hematite is present within minnesotaite fans (Fig. 11J, K), indicating that this generation of hematite postdated late-stage minnesotaite and coarse-grained carbonate. In red chert from the same sample from the Bee Gorge Member of the Wittenoom Formation (Fig. 12), the abundance of hematite dust is greater than in the transition zone. HAADF images reveal randomly oriented plates of greenalite between 200-500 nm in length dispersed throughout the chert cement (Figs. 12, 13), as well as hematite dust particles that are typically coarser grained (0.5-10 µm) than the greenalite. There are also abundant cavities in the chert that contain minnesotaite and hematite (Figs. 12, 13). The textures suggest that greenalite nanoparticles have been dissolved and that minnesotaite and hematite have precipitated in the resultant cavities.

7. Post-depositional origin of hematite The petrographic information collected regarding the origin of hematite dust in Hamersley Group BIFs is based on examination of more than 400 polished thin sections. The following observations indicate that the hematite dust formed after the iron formations were deposited: i) The occurrence of hematite dust along corroded grain boundaries, in mineral cleavage planes and within altered grains of diagenetic or metamorphic iron-bearing carbonates and silicates (Figs. 3, 11). These textures provide clear evidence for in situ growth of

19

late-stage hematite dust related to dissolution-precipitation reactions of iron-bearing minerals. ii) The presence of transition zones between red and green chert that are discordant to bedding (Figs. 4, 7). The change from green to red chert along individual sedimentary laminae (Fig. 8) cannot be reconciled with depositional processes involving coprecipitation of ferrous silicates and ferric oxides/hydroxides from the same water column, and therefore reflects post-depositional alteration. iii) The replacement of iron-silicate nanoparticles by hematite dust in hairline fractures cross-cutting laminated chert (Fig. 6) provides unequivocal evidence for later hematite growth. iv) The presence of cavities in iron-silicate nanoparticles and chert, some of which contain hematite euhedra, from the red-green chert transition zones (Figs. 9, 10, 12, 13). These textures provide compelling evidence for the dissolution of precursor ironsilicate nanoparticles and the precipitation of hematite dust in the resultant voids. v) The presence of relict iron-silicate nanoparticles in red cherts (Figs. 11, 12). All red chert examined in this study contains residual iron-silicate nanoparticles, commonly darkened and pitted, which are interstitial to larger hematite dust particles. Our study of iron formations of the Hamersley Group, which includes some of the best-preserved samples of iron formation in the world, shows that hematite dust formed via dissolution-precipitation reactions probably involving late-stage oxidizing fluids. The development of reaction-induced cavities, as observed in the silicate nanoparticles, is a common feature of dissolution-precipitation mineral reactions (Putnis, 2009). This observation implies that chert, which is commonly regarded to be impermeable, was

20

infiltrated by fluid in order to explain the transport of elements to and from the reaction site. The fluid pathway is uncertain but it is conceivable that crystal dislocations in the chert around the edges of the nanoparticles formed a three-dimensional network for fluid migration and element transfer. Such a mechanism may have favored the dissolution of the larger silicate particles while preserving many of the smaller particles not intersected by dislocations. In addition, the release of water during dehydration may also have played a role in establishing fluid pathways in the enclosing chert. The transition zones between green and red chert, containing both silicate and hematite nanoparticles, are interpreted to represent “frozen” reaction fronts. The presence of both parent (silicate) and product (hematite) phases indicates that dissolutionprecipitation reactions halted before completion. The partial dissolution of iron silicates suggests that the fluid did not fully equilibrate with the rock, possibly reflecting insufficient fluid, reactant or permeability, resulting in the development of disequilibrium mineral assemblages. The replacement of iron-silicate nanoparticles by hematite dust in laminated chert may be cryptic in hand specimen or by light microscopy. Laminated chert devoid of hematite dust is typically green, however, with increasing hematite content, dusty chert becomes increasingly brown and eventually bright red. However, even in the most brilliant red jasper beds, residual iron-silicate nanoparticles are present between the larger hematite dust particles (Rasmussen et al., 2014b). A further indication that hematite dust is post-depositional lies in the absence of primary magnetic remanence in Hamersley BIFs (Li et al., 2000), despite the high blocking temperature of fine-grained hematite (>600° C). Acquisition of secondary

21

magnetization is a common process in sedimentary rocks, being linked to hematite growth during orogeny in Paleozoic red beds and ironstones (e.g., McCabe and Elmore, 1989). The presence of pervasive secondary magnetization postdating structures of the ~2.2 Ga Ophthalmian Orogeny (Li et al., 2000) indicates that hematite growth in Hamersley Group BIFs was a large-scale process that affected the southern Pilbara Craton at least 300 million years after sediment deposition. The presence of multiple generations of secondary magnetization suggests that hematite growth occurred repeatedly after 2.2 Ga, reflecting multiple episodes of fluid flow through the BIFs (e.g., Rasmussen et al., 2007).

8. Implications for the depositional model of iron formations As part of a re-examination of the precursor sediments of iron formations using instruments capable of nanometer-scale microscopy and microanalysis, we find that bedded chert, which contains hematite and silicate dust, preserves textural evidence for late-stage dissolution of silicates and precipitation of hematite. Two potential reaction pathways have been identified: i) the dissolution of stilpnomelane and the precipitation of hematite in the resulting cavities, and ii) dissolution of greenalite and precipitation of minnesotaite, followed by the precipitation of hematite. The different alteration pathways for greenalite and stilpnomelane nanoparticles may relate to differences in their stability. Phase equilibria studies and petrographic results suggest that greenalite reacts with silica to form minnesotaite and releases water (Fe6Si4O10(OH)8 + 4SiO2 = 2Fe3Si4O10(OH)2 + 2H2O) at temperatures above ~150 °C (Klein, 1974; Miyano, 1978), whereas stilpnomelane is stable to temperatures of 430-470

22

°C (Miyano and Klein, 1989). Both reaction pathways ultimately lead to the replacement of iron-silicate nanoparticles by hematite dust. What is remarkable about these results is that the late-stage oxidation of iron-silicate nanoparticles produced hematite dust that fulfills all the petrographic criteria of a primary mineral in iron formations (cf. LaBerge, 1964), even though the hematite dust formed long after deposition. Hematite dust is widely considered to provide a glimpse of the primary precipitate of iron formations, namely ferric oxide/hydroxide particles. Our results suggest that hematite dust can form via dissolution-precipitation reactions of iron-silicate nanoparticles. These findings raise important questions about the identity of the primary precipitates of iron formations. Clearly, the presence of hematite dust cannot be used with confidence to infer that the primary precipitates were hydrated ferric oxides/hydroxides without nanometer-scale textural evidence. Instead, our results show that iron-silicate nanoparticles were the precursors to hematite dust in BIFs of the Hamersley Group. The replacement process observed here is probably not unique and it therefore seems reasonable to suggest that hematite dust in other BIFs may have formed via the same process. While more work is required, if our results are confirmed in other iron formations, it raises important questions about hematite dust: is bona fide primary hematite dust preserved in iron formations, and if not, did it ever form? Our work on hematite dust shows that it does not occur without relict iron-silicate nanoparticles that display evidence of alteration and leaching. This observation leads us to propose that hematite dust in laminated chert and jasper is probably derived from the oxidation of iron-silicate nanoparticles.

23

A secondary origin for hematite dust provides a simple explanation for its coexistence with ferrous-silicates in the same laminae and transitions between red and green chert. Iron minerals can be easily oxidized or reduced by fluids after deposition, which can result in changes to the original redox state and mineralogy. Although these processes can be difficult to identify and are commonly overlooked in ancient rocks, we propose that the juxtaposition of hematite dust with ferrous silicate nanoparticles, combined with dissolution-precipitation textures, can best be understood as the partial oxidation of ferrous silicates. The fluid-mediated alteration of iron formations identified here has led to changes in mineralogy and an increase in the average valence state of Fe. It is, therefore, important that the effects of post-depositional overprinting by fluid-mineral interactions in iron formations are considered and distinguished from primary paleoenvironmental and biological signatures.

9. Summary and Conclusions The assumption that hematite dust establishes that the primary precipitates of BIFs were ferric oxides/hydroxides underpins interpretations of the composition and redox history of the Archean ocean-atmosphere system and the evolution of photosynthetic life. The physicochemical and biological processes invoked to explain the precipitation of ferric oxides/hydroxides are significantly different to those required for the precipitation of ferrous/ferric silicates, that is, anoxic seawater enriched in Fe(II)(aq) and Si(aq) (Harder, 1978; Konhauser et al., 2007; Tosca and Guggenheim, 2014; Tosca et al., 2016). Most models for the precipitation of ferric oxides/hydroxides in early Precambrian oceans involve a biological oxidation mechanism operating in the photic

24

zone, either via O2 produced by cyanobacteria or by the actions of photoferrotrophs (Cloud, 1965, 1973; Konhauser et al., 2002, 2007; Kappler et al., 2005). However, we find no evidence for primary hematite in the least-altered BIFs from the Hamersley Group. Instead, our results show that iron-silicate nanoparticles, which exhibit all the characteristics of primary precipitates, were replaced by hematite dust via dissolution and precipitation reactions. These findings indicate that hematite dust cannot be assumed to be primary, and cannot be used with confidence as evidence for ferric oxide/hydroxide precipitates. If the primary precipitates were iron silicates, then neither dissolved oxygen nor photosynthesis was required to deposit BIFs before 2.4 Ga. While our results do not exclude the possibility that ferric oxides/hydroxides formed or contributed to the precursor sediments, the most parsimonious interpretation is that the original precipitates to BIFs were iron-silicate muds (Rasmussen et al., 2015b). An iron-silicate precursor to BIFs suggests that silicates, and not oxides, played a key role in marine geochemical cycles prior to 2.4 Ga. The replacement of ferrous-bearing minerals by hematite may be much more widespread in early Precambrian rocks than previously recognized. Our results provide an alternative mechanism for the formation of red chert in the sedimentary record prior to the GOE (2.4-2.2 Ga). A post-depositional origin for hematite overcomes a major geochemical puzzle inherent to BIFs: the apparent precipitation and preservation of vast deposits of iron oxides/hydroxides in a predominantly anoxic ocean. Our results support a critical re-evaluation of the original mineralogy and deposition of iron formations, which may have major implications for interpretations of the evolution of the oceanatmosphere-biosphere system during the Archean and early Paleoproterozoic.

25

Acknowledgements Financial support for this research was provided by ARC grants DP110103660 and DP140100512. We acknowledge the staff and facilities at the Centre for Microscopy, Characterisation and Analysis at UWA and Adelaide Microscopy, the University of Adelaide. We thank Greg Lester, Andrew Putnis, Martin Saunders and Steve Sheppard for comments and discussion. We thank A. Basak for Focused Ion Beam operation and T. Waggoner for his time and assistance during fieldwork in the Marquette Range, Michigan. We are grateful to N. Tosca for helpful comments.

References Ahn, J.H., Buseck, P.R., 1990. Hematite nanospheres of possible colloidal origin from a Precambrian banded iron formation. Science 250, 111–113. Ayres, D.E., 1972. Genesis of iron-bearing minerals in the Brockman Iron Formation mesobands in the Dales Gorge Member, Hamersley Group, Western Australia. Econ. Geol. 67, 1214-1233. Bekker, A., Slack, J.F., Planavsky, N., Krapež, B., Hofmann, A., Konhauser, K.O., Rouxel, O.J., 2010. Iron formation: The sedimentary product of a complex interplay among mantle, tectonic, oceanic, and biospheric processes. Econ. Geol. 105, 467– 508. Beukes, N.J., 1973. Precambrian iron-formations of southern Africa. Econ. Geol. 68, 960-1004. Beukes, N.J., Gutzmer, J., 2008. Origin and paleoenvironmental significance of major

26

iron formations at the Archean-Paleoproterozoic boundary. Rev. Econ. Geol. 15, 5– 47. Button, A., Brock, T.D., Cook, P.J., Eugster, H.P., Goodwin, A.M., James, H.L., Margulis, L., Nealson, K.H., Nriagu, J.O., Trendall, A.F., Walter, M.R., 1982. Sedimentary iron deposits, evaporate and phosphorites – State of the art report In: Holland, H.D., Schidlowski, M. (Eds.), Mineral Deposits and the Evolution of the Biosphere. Dahlem Konferenzen, 1982, Berlin, Heidelberg, New York, Springer Verlag, 259-273. Cairns-Smith, A.G., 1978. Precambrian solution photochemistry, inverse segregation, and banded iron formations. Nature 276, 807–808. Chan, M.A., Johnson, C.M., Beard, B.L., Bowman, J.R., Parry, W.T., 2006. Iron isotopes constrain the pathways and formation of terrestrial oxide concretions: A tool for tracing iron cycling on Mars? Geosphere 2, 324-332. Cloud, P., 1965. Atmospheric and hydrospheric evolution on the primitive Earth. Science 160, 729-735. Cloud, P., 1973. Paleocological significance of the banded iron-formation. Econ. Geol. 68, 1135-1143. Dimroth, E., Chauvel, J.-J., 1973. Petrography of the Sokoman Iron Formation in part of the central Labrador Trough, Quebec, Canada. Geol. Soc. Am. Bull. 84, 111-134. Dorr, J.V.N., II., 1945. Manganese and iron deposits of Morro do Urucum, Mato Grosso, Brazil. U.S. Geol. Surv. Bull. 946-A, 1-47. du Preez, J.W., 1945. The structural geology of the area east of Thabazimbi and the genesis of the associated iron ores. Stellenbosch University Annals, vol. 22, sec. A,

27

no. 1-14, p. 263-360. Ewers, W.E., Morris, R.C., 1981. Studies of the Dales Gorge Member of the Brockman Iron Formation. Econ. Geol. 76, 1929–1953. Fischer, W.W., Knoll, A.H., 2009. An iron shuttle for deepwater silica in Late Archean and early Paleoproterozoic iron formation. Geol. Soc. Am. Bull. 121, 222–235. Geijer, P., 1938. Stripa Odalfilts Geologi. Sveriges geologiske undersøkning, vol. 28, set. Ca. Gill, J.E., 1926. Gunflint iron-bearing formation, Ontario. Geol. Surv. Can. Summary Report for 1924, pt. C., 28-88. Gill, J.E., 1927. Origin of the Gunflint iron-bearing formation. Econ. Geol. 22, 687-728. Grout, F.F., 1919. Nature and origin of the Biwabik iron-bearing formation of the Mesabi range, Minnesota. Econ. Geol. 14, 452-464. Gruner, J.W., 1922. The origin of the sedimentary iron-formations; the Biwabik formation of the Mesabi range. Econ. Geol. 17, 407-460. Gruner, J.W., 1946. The mineralogy and geology of the taconites and iron ores of the Mesabi range, Minnesota. Iron Range Resources and Rehabilitation, St. Paul, Minnesota, 127p. Hallimond, A.F., 1925. Iron ores: Bedded ores of England and Wales. Great Britain Special Report of Mining and Resources, Petrography and Chemistry, vol. 29, 139p. Harder, H., 1978. Synthesis of iron layer silicate minerals under natural conditions. Clays Clay Minerals 26, 65-72. Haugard, R., Pecoits, E., Lalonde, S., Rouxel, O., Konhauser, K.O., 2016. The Joffre banded iron formation, Hamersley Group, Western Australia: Assessing the

28

palaeoenvironment through detailed petrology and chemostratigraphy. Precambr. Res. 273, 12-37. James, H.L., 1951. Iron formation and associated rocks in the Iron River district, Michigan. Geol. Soc. Am. Bull. 62, 251-266. James, H.L., 1954. Sedimentary facies of iron-formation. Econ. Geol. 49, 235−293. Kappler, A., Pasquero, C., Konhauser, K.O., Newman, D.K., 2005. Deposition of banded iron formations by anoxygenic phototrophic Fe(II)-oxidizing bacteria. Geology 33, 865–868. Kaufman, A.J., Hayes, J.M., Klein, C., 1990. Primary and diagenetic controls of isotopic compositions of iron-formation carbonates. Geochim. Cosmochim. Acta 54, 34613473. Klein, C., 1974. Greenalite, stilpnomelane, minnesotaite, crocidolite and carbonates in a very low-grade metamorphic Precambrian iron-formation. Can. Mineral. 12, 475498. Klein, C., 2005. Some Precambrian banded iron-formations (BIFs) from around the world: Their age, geologic setting, mineralogy, metamorphism, geochemistry, and origin. Am. Mineral. 90, 1473–1499. Klein, C., Bricker, O.P., 1977. Some aspects of the sedimentary and diagenetic environment of Proterozoic banded iron-formation. Econ. Geol. 72, 1457-1470. Klein, C., Fink, R.P., 1976. Petrology of the Sokoman Iron Formation in the Howells River area, at the western edge of the Labrador Trough. Econ. Geol. 71, 453-487.

29

Konhauser, K.O., Hamade, T., Raiswell, R., Morris, R.C., Ferris, F.G., Southam, G., Canfield, D.E., 2002. Could bacteria have formed the Precambrian banded iron formations? Geology 30, 1079-1082. Konhauser, K.O., Amskold, L., Lalonde, S.V., Posth, N.R., Kappler, A., Anbar, A., 2007. Decoupling photochemical Fe(II) oxidation from shallow-water BIF deposition. Earth Planet. Sci. Lett. 258, 87–100. Krapež, B., Barley, M.E., Pickard, A.L., 2003. Hydrothermal and resedimented origins of the precursor sediments to banded iron formations: Sedimentological evidence from the early Palaeoproterozoic Brockman Supersequence of Western Australia. Sedimentology 50, 979–1011. Krumbein, W.C., Garrels, R.M., 1952. Origin and classification of chemical sediments in terms of pH and oxidation-reduction potentials. J. Geol. 60, 1-33. LaBerge, G.L., 1964. Development of magnetite in iron-formations of the Lake Superior region. Econ. Geol. 59, 1313-1342. LaBerge, G.L., 1966. Altered pyroclastic rocks in iron-formation in the Hamersley Range, Western Australia. Econ. Geol. 61, 147-161. Leith, C.K., 1903. The Mesabi iron-bearing district of Minnesota. U.S. Geol. Surv. Monograph 43, 316p. Li, W., Beard, B.L., Johnson, C.M., 2015. Biologically recycled continental iron is a major component in banded iron formations. Proc. Nat. Acad. Sci. USA 112, 81938198. Li, Z.X., Guo, W., Powell, C.McA., 2000. Timing and genesis of Hamersley BIF-hosted iron deposits: A new palaeomagnetic interpretation. MERIWA Project M242 Final

30

Report, 283p. McCabe, C., Elmore, R.D., 1989. The occurrence and origin of late Paleozoic remagnetization in the sedimentary rocks on North America. Rev. Geophys. 27, 471494. Miyano, T., 1978. Phase relations in the system Fe-Mg-Si-O-H and environments during low-grade metamorphism of some Precambrian iron formations. J. Geol. Soc. Japan 84, 679-690. Miyano, T., Klein, C., 1989. Phase equilibria in the system K2O-FeO-MgO-Al2O3-SiO2H2O-CO2 and stability limit of stilpnomelane in metamorphosed Precambrian ironformations. Contrib. Min. Petrol. 102, 478-491. Morris, R.C., 1980. A textural and mineralogical study of the relationship of iron ore to banded iron-formation in the Hamersley Iron Province of Western Australia. Econ. Geol. 75, 184-209. Morris, R.C., 1993. Genetic modelling for banded iron-formation of the Hamersley Group, Pilbara Craton, Western Australia. Precambr. Res. 60, 243–286. Pecoits, E., Gingras, M.K., Barley, M.E., Kappler, A., Posth, N.R., Konhauser, K.O., 2009. Petrography and geochemistry of the Dales Gorge banded iron formation: Paragenetic sequence, source and implications for palaeo-ocean chemistry. Precambr. Res. 172, 163-187. Pecoits, E., Smith, M.L., Catling, D.C., Phillippot, P., Kappler, A., Konhauser, K.O., 2015. Atmospheric hydrogen peroxide and Eoarchean iron formations. Geobiology 13, 1-14. Pickard, A.L., Barley, M.E., Krapež, B., 2004. Deep-marine depositional setting of

31

banded iron formation: Sedimentological evidence from interbedded clastic sedimentary rocks in the early Paleoproterozoic Dales Gorge Member of Western Australia. Sediment. Geol. 170, 37−62. Putnis, A., 2009. Mineral replacement reactions. In: Oelkers, E.H., Schott, J. (Eds.), Thermodynamics and Kinetics of Water-Rock Interaction: Reviews in Mineralogy and Geochemistry 70, 87-124. Rasmussen, B., Fletcher, I.R., Muhling, J.R., Thorne, W.S., Broadbent, G.C., 2007. Prolonged history of episodic fluid flow in giant hematite ore bodies: Evidence from in situ U-Pb geochronology of hydrothermal xenotime. Earth Planet. Sci. Lett. 258, 249-259. Rasmussen, B., Krapež, B., Meier, D.B., 2014a. Replacement origin for hematite in 2.5 Ga banded iron formation: Evidence for post-depositional oxidation of iron-bearing minerals. Geol. Soc. Am. Bull. 126, 438-446. Rasmussen, B., Krapez, B., Muhling, J.R., 2014b. Hematite replacement of iron-bearing precursor sediments in the 3.46-b.y.-old Marble Bar Chert, Pilbara craton, Australia. Geol. Soc. Am. Bull. 126, 1245-1258. Rasmussen, B., Krapez, B., Muhling, J.R., 2015a. Seafloor silicification and hardground development during deposition of 2.5 Ga banded iron formations. Geology 43, 235238. Rasmussen, B., Krapež, B., Muhling, J.R., Suvorova, A., 2015b. Precipitation of iron silicate nanoparticles in early Precambrian oceans marks Earth’s first iron age. Geology 43, 303-306. Rasmussen, B., Meier, D.B., Krapez, B., Muhling, J.R., 2013. Iron silicate microgranules

32

as precursor sediments to 2.5-billion-year-old banded iron formations. Geology 41, 435-438. Smith, R.E., Perdrix, J.L., Parks, T.C., 1982. Burial metamorphism in the Hamersley Basin, Western Australia. J. Petrol. 23, 75–102. Smyth, C.H., 1911. The Clinton type of iron-ore deposits. In: Bain, H.F. (Eds.), Types of Ore Deposits: Dewey Publishing Company, 33-52. Spencer, E., Percival, F.G., 1952. The structure and origin of the banded hematite jaspers of Singhbhum, India. Econ. Geol. 47, 365-383. Spurr, J.E., 1894. The iron-bearing rocks of the Mesabi Range in Minnesota. Geol. Nat. Hist. Surv. Minnesota, Bull. 10, 268p. Sun, S., Konhauser, K.O., Kappler, A., Li, Y-L., 2015. Primary hematite in Neoarchean to Paleoproterozoic oceans. Geol. Soc. Am. Bull. 127, 850-861. Taylor, J.H., 1949. Petrology of the Northampton Sand Ironstone Formation: The Geol. Surv. Great Britain Mem 111p. Tosca, N.J., Guggenheim, S., 2014. Experimental constraints on Precambrian seawater chemistry: Solubility in the Fe2+-SiO2(aq) system: Goldschmidt 2014 Abstract, 2502 (http://goldschmidt.info/2014/abstracts/abstractView?abstractId=3783). Tosca, N.J., Guggenheim, S., Pufahl, P., 2016. An authigenic origin for Precambrian greenalite: implications for iron formation and the chemistry of ancient seawater. Geol. Soc. Am. Bull. 128, 511-530. Trendall, A.F., 2002. The significance of iron-formation in the Precambrian stratigraphic record. In: Altermann, W., Corcoran, P.L. (Eds.), Precambrian Sedimentary Environments: A Modern Approach to Ancient Depositional Systems. Special

33

Publication no. 33, International Association of Sedimentologists, 33-66. Trendall, A.F., Blockley, J.G., 1970. The iron-formations of the Precambrian Hamersley Group, Western Australia. Geol. Surv. Western Australia Bull. 119. Van Hise, C., Leith, C.K., 1911. Geology of the Lake Superior region. U.S. Geol. Surv. Monograph 52, 641p. Winchell, N.H., Grant, U.S., Todd, J.E., Upham, W., Winchell, H.V., 1899. The Geology of Minnesota. Volume 4 of the Final Report, 1896-1898. The Geological and Natural History Survey of Minnesota, 629p. Wolff, J.F., 1917. Recent geologic developments on the Mesabi iron range, Minnesota. Proc. Lake Superior Min. Inst. 21, 33-40.

34

FIGURE CAPTIONS Figure 1. Locality map of the four drill-holes sampled in this study, intersecting banded iron formations the Hamersley Group, southern Pilbara Craton, Western Australia: ABDP9, DDH44, Mitchell 2 and Silvergrass.

Figure 2. (A). Photograph of outcrop of banded iron formation from Jasper Knob, Ishpeming, Michigan, consisting of alternating bands of bright red jasper and metallic hematite. (B). Outcrop of iron formation from Jasper Knob showing complex folding and brecciation along fold hinges. (C). Polished thin section of banded iron formation from Jasper Knob showing red jasper layers and hematite-rich layers. (D). Reflected light (RL) image of jasper layer crosscut by vein containing hematite, martite (i.e., hematite pseudomorphs of magnetite) and quartz. (E). RL image showing alignment of minute plates of hematite in jasper layer. (F). RL image of hematite-rich layer showing strong fabric defined by aligned hematite plates and quartz pressure shadows around euhedral martite crystals. (G). RL image of crenulation cleavage defined by platy hematite. Note well-developed quartz pressure shadows around martite euhedra.

Figure 3. (A). Transmitted light (TL) image of randomly oriented ferroan dolomiteankerite crystals (dark brown) in laminated chert-carbonate rock. (B). Close-up of carbonate rhomb. (C). Reflected light (RL) image of carbonate rhomb in Fig. 3B containing numerous minute specks of hematite (hem: see arrows). (D). TL image of carbonate nodule in hematite-magnetite rich band. (E). TL image of carbonate rhomb showing dark trails along mineral cleavage plane. (F). Incident light (IL) image of minute

35

hematite particles (hem) in carbonate mineral cleavage. (G). Coarse-grained ferroan dolomite-ankerite crystal lined by thin black rim. (H). Close-up of carbonate rhomb lined by black rim comprising hematite dust particles (hem; see arrows). (I). IL image of hematite dust particles (red) lining the irregular outer margin of a carbonate rhomb. (J). TL image of an iron-rich mudrock comprising stilpnomelane, siderite nodules, magnetite (black) and hematite (black). (K). TL image of siderite (sid) nodule in stilpnomelane (stp) band mantled by hematite (hem). (L). RL image of pitted siderite nodule (in 3K) containing numerous minute particles of hematite (see arrows; hem). Figure 3A-C, DDH44, 516.3 m; 3D-F, DDH44, 488.47 m; 3G-I, ABDP9, 219.27-.33 m; 3J-L, Silvergrass, 394.47 m.

Figure 4. Drill-core of chert mesoband from the Dales Gorge Member of the Brockman Iron Formation showing relic rafts of grey-green laminated chert surrounded by red chert. Note the lateral and vertical transition between red and grey-green chert, which is discordant to bedding. Drill-hole DGM4.

Figure 5. (A). Transmitted light (TL) image showing finely laminated green chert. (B, C). TL and incident light (IL) images of finely laminated green chert (from Figure 5A). (D). High-magnification TL image of green chert containing numerous minute inclusions. (E). Transmission electron microscope (TEM) image of green chert showing randomly oriented iron-silicate nanoparticles (medium to dark grey) “floating” in chert cement (light grey). Drill-hole ABDP-9, 288.23–288.36 m.

36

Figure 6. (A). Image of finely laminated chert mesoband. (B). Transmitted light (TL) image of laminated chert with abundant silicate dust particles. (C). Densely packed silicate dust particles in laminae transected by vertical hairline fractures. (D). TL image of vertical hairline fractures in dust-impregnated chert. (E). Incident light (IL) image of Figure 6D showing minute hematite dust particles (red) in vertical trails. (F). Reflected light image of Figure 6E with hematite dust particles (white) concentrated along vertical fractures. Drill-hole DDH44, 363.15 m.

Figure 7. (A). Transmitted light (TL) of portion of polished thin section showing chert mesoband with lateral and vertical transition between red and green chert. (B). TL image of Fig. 7A. (C) Incident light image of Figure 7B showing the transition from red chert to green chert along a single lamina. (D-F). Incident light (IL), TL and reflected light (RL) images of red chert from Figure 7C showing dense clusters of hematite dust in a matrix of green chert. (G-I). IL, TL and RL images of green chert with densely packed inclusions of silicate nanoparticles devoid of hematite dust. Drill-hole DDH44, 513.25 m.

Figure 8. (A). Transmitted light (TL) images showing laminated red and green chert and diffuse blue-grey bands of riebeckite. (B, C). TL and IL images showing abundant hematite (red) in chert. (D, E). TL and IL images of iron-rich laminae in chert, showing the transition between silicate dust particles (left) and red hematite dust particles (right). Drill-hole Silvergrass, 313.65 m.

37

Figure 9. (A). Back-scattered electron image of where a foil was cut from polished thin section by focused ion beam. (B). Transmission electron microscope (TEM) image of randomly oriented iron-silicate nanoparticles “floating” in chert cement. Note the presence of cavities in the silicate particles, some of which contain euhedral hematite crystals. (C, D). HAADF STEM images showing stilpnomelane particles “floating” in chert cement with cavities (black; arrows) and hematite euhedra. Drill-hole Silvergrass, 313.65 m.

Figure 10. (A, B). HAADF STEM images of partly dissolved stilpnomelane particles (stp) with cavities (black) and euhedral hematite (hem) “floating” in chert (ch) cement. (C-F). Si, Fe, Mg and O element maps, respectively, clearly showing the presence of Mg in stilpnomelane. Drill-hole Silvergrass, 313.65 m.

Figure 11. (A). TL image of chert layer with abundant iron silicate nanoparticles, hematite dust, minnesotaite and ankerite. (B). Incident light (IL) image of 11A. (C). TL image showing randomly oriented fans of minnesotaite in a matrix of chert with densely packed iron silicate and hematite dust particles. (D). Incident light (IL) image of Figure 11C showing distribution of hematite (red) in the chert cement. (E). TL image of ankerite (ank) rhomb in chert band with black rim. (F). IL image of ankerite (ank) highlighting the distribution of iron oxides (red-brown) in the chert and around ankerite. (G). TL image of part of ankerite (ank) rhomb with black rim comprising hematite dust (hem). (H). RL image of rim in 11G showing minute hematite particles (hem; see arrows) lining carbonate. (I). Chert laminae containing densely packed iron silicate nanoparticles, hematite dust and altered ankerite rhombs (ank), surrounded by hematite dust (opaque). 38

(J). TL image of minnesotaite fans with hematite (opaque). (K). Reflected light image of Figure 11C showing hematite (white) within centre of a minnesotaite (min) crystal. Drillhole ABDP-9, 219.27-.33 m.

Figure 12. (A). Transmitted light (TL) image of laminated chert mesoband comprising green and red laminae. (B, C). TL and IL images showing abundant hematite in chert (opaque in 12 B and red in 12C). (D). TL image of randomly oriented hematite plates (orange) in chert cement with abundant silicate dust particles. (E). Reflected light image of Figure 12D showing hematite plates (white) in chert cement (grey) containing abundant silicate dust particles (dark pits). (F). HAADF STEM image of randomly oriented iron-silicate nanoparticles (light grey) and hematite (white) in chert (dark grey). Drill-hole ABDP-9, 219.27-.33 m.

Figure 13. (A). HAADF STEM image of greenalite particles (gre) with cavities (black), minnesotaite (min) and hematite (hem) “floating” in chert (ch) cement. (B-F). Si, Fe, Mg, O and K element maps, respectively. Drill-hole ABDP-9, 219.27-.33 m.

39

Figure 1

Indian Ocean o 21 S

o

o

117 E

119 E

Weeli Wolli Formation

Karratha

Mitchell 2

Mt McRae Sh Mt Sylvia Fm Wittenoom Formation Marra Mamba Iron Fm

Joffre Member

Perth

ABDP9

Silvergrass

o

23 S

Whaleback

Wittenoom Tom Price

Hamersley Province

Paraburdoo

100 km

Dales Gorge

Brockman Iron Formation

Pannawonica

500 metres

Western Australia

Newman

DDH44

Upper Wyloo Group and cover sequences Lower Wyloo Group

Turee Creek Group Hamersley Group Fortescue Group

Granite Greenstone belt Diamond drill-hole

Figure 1, Rasmussen et al.

Figure 2

B

A

5 cm

10 cm

DD

C

E

E

E F G 5 mm

50 µm

10 µm 200 µm

F

G

100 µm

100 µm

Figure 2, Rasmussen et al.

Figure 3

B

A

C C

hem

D D

F

E E

F

hem

0.5 mm

G

10 µm

50 µm

0.2 mm

H

H

50 µm

100 µm

I

hem

K

J

hem

50 µm

100 µm

stp

L

stp

25 µm

hem

sid 1 mm

hem

50 µm

50 µm

Figure 3, Rasmussen et al.

Figure 4

5 mm Figure 4, Rasmussen et al.

Figure 5

C

B

A

D

2 mm

D

50 µm

50 µm

E

D

10 µm

400 nm

Figure 5, Rasmussen et al.

Figure 6

A

B

B

C

1 mm

5 mm

D

C D

100 µm

E

20 µm

F

20 µm

20 µm

Figure 6, Rasmussen et al.

Figure 7

A

B

B

1 mm

5 mm

C

1 mm

D-F

E

D

50 µm

G-I

F

50 µm

G

10 µm

I

H

50 µm

50 µm

10 µm

Figure 7, Rasmussen et al.

Figure 8

A

C

B

D

50 µm

2 mm

D

E

50 µm

E

10µm

10 µm

Figure 8, Rasmussen et al.

Figure 9

B

A

hem ch

ch

D

C

10 µm

C

stp

ch

D

ch hem

cavity hem

stp

stp

cavity

ch

200 nm

Figure 9, Rasmussen et al.

Figure 10

B

A

stp cavity hem

ch

stp hem

B

500 nm

C

ch

1 µm

D

Fe E

300 nm

Mg F

300 nm

Si

300 nm

O

300 nm

Figure 10, Rasmussen et al.

Figure 11

A

C

Fig. 10E

D

min

min

1 mm

B 1 mm

E

ch

F

F ank ch

I

ch

G

G

ank 0.1 mm

0.2 mm

ank

ank

ch

min

0.1 mm

min

0.1 mm

H

ch

ch

hem

hem

50 µm

ank

C J

ch

ank

K min

hem

50 µm

ch

ch min

ank 0.1 mm

ch 0.1 mm

min

50 µm

Figure 11, Rasmussen et al.

Figure 12

B

A

C

K

50 µm

2 mm

50 µm

F F

D D

10 µm 10 µm

E 400 nm 10 µm

500 nm

Figure 12, Rasmussen et al.

Figure 13

gre

A

B

Si

ch

cavity

hem ch min

ch

1 µm

gre

hm

min

hm 1 µm

D

Fe

C

O E

1 µm

Mg F

1 µm

1 µm

FK

1 µm

Figure 13, Rasmussen et al.

Highlights

• Hematite dust in major BIF formed by replacement of iron-silicate nanoparticles. • Hematite dust is not a reliable proxy for relict iron oxide/hydroxide precipitates. • Oxidizing fluids modified the redox state and mineralogy of BIFs after deposition. • Replacement origin for hematite negates its use to infer ancient redox conditions. • Secondary origin for hematite explains lack of primary magnetic remanence in BIFs.

40