sedimentology and geochemistry of archean silica...

Sedimentology and geochemistry of Archean silica granules

Geological Society of America Bulletin, v. 1XX, no. XX/XX 1


Elizabeth J.T. Stefurak1,†, Donald R. Lowe1, Danielle Zentner1, and Woodward W. Fischer2

1Department of Geological and Environmental Sciences, Stanford University, Stanford, California 94305, USA2Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California 91126, USA

ABSTRACT

The production of biogenic silica has dominated the marine silica cycle since early Paleozoic time, drawing down the concentra-tion of dissolved silica in modern seawater to a few parts per million (ppm). Prior to the biological innovation of the fi rst silica biomineralizing organisms in late Protero-zoic time, inputs of silica into Precambrian seawater were balanced by strictly chemi-cal silica and silicate precipitation processes, although the mechanics of this abiotic ma-rine silica cycle remain poorly understood. Cherty sedimentary rocks are abundant in Archean sequences, and many previous au-thors have suggested that primary precipita-tion of amorphous silica could have occurred in Archean seawater. The recent discovery that many pure chert layers in early Archean rocks formed as sedimentary beds of sand-sized, subspherical silica granules has pro-vided direct evidence for primary silica deposition. Here, we provide further sedi-mentological and geochemical analyses of early Archean silica granules in order to gain a better understanding of the mechanisms of granule formation. Silica granules are com-mon components of sedimentary cherts from a variety of depositional settings and water depths. The abundance and widespread dis-tribution of silica granules in Archean rocks suggest that they represented a signifi cant primary silica depositional mode and that most formed by precipitation in the upper part of the water column. The regular oc-currence of silica granules as centimeter-scale layers within banded chert alternating with layers of black or ferruginous chert containing few granules indicates episodic granule sedimentation. Contrasting silicon iso topic compositions of granules from dif-ferent depo sitional environments indicate that isotopic signatures were modifi ed during early diagenesis. Looking to modern siliceous sinters for insight into silica precipitation, we suggest that silica granules may have formed

via multiple stages of aggregation of silica nanospheres and microspheres. Consistent with this hypothesis, Archean ocean chemis-try would have favored particle aggregation over gelling. Granule formation would have been most favorable under conditions pro-moting rapid silica polymerization, includ-ing high salinity and/or high concentrations of dissolved silica. Our observations suggest that granule sedimentation was often epi-sodic, suggesting that granule formation may have also been episodic, perhaps linked to variations in these key parameters.

INTRODUCTION

The abundance and sedimentary style of chert in early Archean rocks highlight a fundamental distinction between the modern and Archean silica cycles. The early Archean silica cycle lacked the key components that dominate the modern silica cycle: continental weathering as the dominant source and silica biomineraliza-tion as the dominant sink for dissolved silica in marine waters (Maliva et al., 1989, 2005; Siever, 1957, 1992; Treguer et al., 1995). Although the volume of continental crust during Archean time was much less than on modern Earth (Cawood et al., 2012; Dhuime et al., 2012), Archean oceans had abundant silica sources in the form of weathering and alteration of mafi c and ultramafi c rocks (Siever, 1992); mass balance requires that abundant sinks must also have been present. Previous authors have suggested that banded iron formations, diagenetic silicifi cation, and authigenic clay precipitation (Maliva, 2001; Maliva et al., 1989, 2005; Siever, 1992; Stefurak et al., 2015) were important sinks in the Archean silica cycle. It has also long been suggested that primary chemical precipitation of amorphous silica phases also played a major role as a silica sink during Precambrian time (Lowe, 1999a; Maliva et al., 2005; Siever, 1992), which is sup-ported by the abundance of chert, especially in pre–3.0 Ga Archean sequences. However, many cherts also formed by diagenetic replacement and meta somatism of nonsiliceous primary material (Lowe, 1999a), making primary silica precipitates diffi cult to identify unambiguously.

Many pre–3.0 Ga cherty sequences include <10-cm-thick layers of white-weathering or translucent chert composed of nearly pure SiO2 (commonly >99 wt%), now microcrystalline quartz (Lowe, 1999a). These white chert beds are interbedded with compositionally contrast-ing beds containing primary organic sedimen-tary grains (Lowe, 1999a; Walsh and Lowe, 1999), mixtures of siderite and fi ne aluminosili-cates, or hematite. These alternating lithologies are known as banded black-and-white chert, banded ferruginous (iron-bearing) chert, and banded iron formation, respectively (Fig. 1).

Stefurak et al. (2014) reported that many of the white chert bands in these units are distinct sedi-mentary beds composed entirely of sand-sized subspherical primary silica granules (Fig. 1) and that silica granules were a common component in other sedimentary layers as well, suggesting that silica granules were a widespread grain type and represented a signifi cant mode of silica deposi-tion in early Archean time. This observation aug-mented the emerging view that the deposition of mud-, silt-, and sand-sized chemical sand grains composed of silica (Stefurak et al., 2014), iron silicates (Rasmussen et al., 2013), and siderite (Köhler et al., 2013) was involved in the deposi-tion of early Precambrian cherts and iron forma-tions. This study presents additional petrographic and geochemical analyses of silica granules with the goal of providing further insight into controls on their formation and deposition.

GEOLOGIC BACKGROUND

This study focuses on chert samples collected in early Archean strata from both the Pilbara block of Western Australia and the Barberton greenstone belt of South Africa. Samples from the Pilbara block are from the Antarctic Creek Member of the 3470 ± 1 Ma Apex Basalt (Byerly et al., 2002), which forms the middle part of the Warrawoona Group (Fig. 2; Van Kranendonk et al., 2002). The Antarctic Creek Member is a 4–14-m-thick unit composed largely of silicifi ed felsic volcaniclastic sediments that also include an event bed of impact spherules and current-deposited layers of silica granules (Stefurak et al., 2014). The spherule bed has been corre-

GSA Bulletin; Month/Month 2015; v. 1xx; no. X/X; p. 1–18; doi: 10.1130/B31181.1; 10 fi gures; 1 table; Data Repository item 2015105.; published online XX Month 2014.

†E-mail: [email protected]

For permission to copy, contact [email protected]© 2015 Geological Society of America

Stefurak et al.

2 Geological Society of America Bulletin, v. 1XX, no. XX/XX

lated with a similar 3470 ± 2 Ma unit, spherule bed S1, in the Barberton greenstone belt (Byerly et al., 2002; Glikson et al., 2004; Hickman and Van Kranendonk, 2008; Lowe and Byerly, 1986).

Samples from the Barberton greenstone belt were collected from the Buck Reef Chert Mem-ber of the Kromberg Formation, the upper Men-don Formation, and the basal Mapepe Forma-tion (Fig. 2; Fig. DR11). Both the Kromberg and

Mendon Formations are part of the Onverwacht Group, an 8–10-km-thick sequence of mafi c and ultramafi c rocks with thin, interbedded sedimentary chert units representing sediments deposited during periods of volcanic quiescence (Lowe and Byerly, 1999). The overlying Fig Tree Group is an ~180-m-thick sequence of immature terrigenous clastic rocks and felsic volcaniclastic units (Condie et al., 1970; Lowe and Byerly, 1999). The Mapepe Formation is the basal unit of the Fig Tree Group south of the Inyoka fault (Lowe and Byerly, 1999).

The Buck Reef Chert is a 200–400-m-thick unit of banded black-and-white and banded

ferruginous chert at the base of the Kromberg Formation (Lowe and Byerly, 1999). The age of the Buck Reef Chert is bracketed by two sets of U/Pb zircon ages of 3416 ± 5 Ma from a con-glomeratic unit in the lowest part of the Buck Reef Chert (Krüner et al., 1991) and 3334 ± 3 Ma from a felsic tuff from the uppermost Kromberg Formation (Byerly et al., 1996). The basal unit of the Buck Reef Chert was deposited in a shallow, restricted setting, indicated by the common occurrence of wave ripples and evapo-rite pseudomorphs after nahcolite (Lowe and Fisher Worrell, 1999). The middle Buck Reef Chert, composed of banded black-and-white

1 mm

500 μm1 cm

1 cm 500 μm2 cm

1 mm

2 cm

A B C

F

G H I

D E

1 cm

Figure 1. Layers of silica granules at different scales. (A–C) Black-and-white banded chert: (A) outcrop photo, Buck Reef Chert (locality Buck Reef Chert), (B) polished slab of white chert band showing faintly granular texture, upper Mendon Formation (locality BH-03), and (C) thin section of silica granules from a white chert band, upper Mendon Formation (locality SAF 521). (D–F) Lenticular granular layers within ferruginous shale, lower Mapepe Formation (locality SAF 183): (D) polished slab with lenses of two distinct granule types (arrows), both shown in thin section—a lower lens of mini-mally compacted, slightly ferruginous granules (E) and an upper lens of more compacted non-ferruginous granules (F). (G–I) Banded iron formation, lower Mapepe Formation: (G) outcrop photo showing chert bands (arrows) within jasper , (H) core photo highlighting ~1-cm-thick granular layer within banded iron formation (dashed outline) (sample SAF 649–14), and (I) thin-section image of pure silica granules within slightly ferruginous matrix from the layer shown in H. See supplementary Table DR1 for additional stratigraphic information for samples shown and supplementary Figure DR1 for map of sample localities (see text footnote 1).

1GSA Data Repository item 2015105, Figures DR1-DR8 and Tables DR1 and DR2, is available at http:// www .geosociety .org /pubs /ft2015 .htm or by re-quest to editing@ geosociety .org.



chert, was deposited in a relatively open shelf environment (Tice and Lowe, 2004). The upper Buck Reef Chert represents a deeper, quieter set-ting marked by fi nely laminated, banded ferru-ginous chert (Lowe and Byerly, 1999; Tice and Lowe, 2004). The Buck Reef Chert is exposed in the southernmost structural belt of the Bar-berton greenstone belt and outcrops for over 15 km along strike, indicating that deposition of banded black-and-white chert and banded ferru-ginous chert was occurring continuously across relatively large areas on the Archean seafl oor.

Previous studies have identifi ed in aggregate at least fi ve episodes of volcanism within the Mendon Formation, punctuated by the depo-sition of now-silicifi ed sedimentary layers. However, most sections do not contain the full sequence, potentially due to diachronous depo-sition and/or local topographic highs imped-ing lateral continuity of volcanic units (Byerly, 1999; Decker, 2013; Lowe and Byerly, 1999; Thompson Stiegler et al., 2010). The overall age of the Mendon Formation is constrained by U/Pb zircon dating: a maximum age of 3334 ±

3 Ma from the underlying Kromberg Formation (Byerly et al., 1996); intermediate ages mea-sured within the Mendon Formation of 3298 ± 3 Ma (Byerly et al., 1996), 3287 ± 3 Ma (Decker, 2013), and 3280 ± 9 Ma (Decker, 2013); and several ages from the base of the overlying Mapepe Formation ca. 3259 ± 3 Ma (Byerly et al., 1996; Decker, 2013; Krüner et al., 1991).

Throughout the Barberton greenstone belt, the Mendon Formation is capped by a 5–70-m-thick unit of black chert, banded black-and-white chert, and banded ferruginous chert (Lowe and Byerly, 1999). The upper Mendon Formation chert unit is overlain with apparent conformity by the sediments of the Mapepe For-mation of the Fig Tree Group, which includes ferruginous shale, locally coarser clastic units, banded ferruginous chert, and hematitic banded iron formation (Lowe and Byerly, 1999). The Mendon-Mapepe contact refl ects a change from a relatively quiet, deep-water regime marked by chemical, biogenic, and volcaniclastic sedimen-tation to a regime dominated by terrigenous and volcaniclastic sediments deposited in a range of depositional environments refl ecting the initial stages of uplift and deformation in and around the greenstone belt (Lowe and Byerly, 1999; Lowe and Nocita, 1999).

METHODS

Outcrops, polished hand samples, and pol-ished petrographic thin sections were used to examine silica granule properties and occur-rence at a variety of scales. Some lower Mapepe Formation samples were collected from the BARB4 core of the 2011 International Conti-nental Scientifi c Drilling Program (ICDP) in the Barberton greenstone belt.

Granule sizes were measured using thin sec-tions of granules that were only slightly to mod-erately compacted. For aggregates of granules, individual granules were measured rather than the dimension of aggregates. Apparent grain sizes were corrected using a Matlab script based on the work of Schäfer and Teyssen (1987) to estimate the true grain sizes from observations of spherical segment diameters in thin section cuts through grains. Grains were approximated as oblate spheres. Aspect ratios were measured and averaged in order to compare diameters of volume-equivalent spheres.

Gray-scale cathodoluminescence (CL) images of areas of interest were mapped using a JEOL JSM-5600 scanning electron microscope (SEM) fi tted with a Hamamatsu photo multi-plier tube (PMT), at the Stanford–U.S. Geo-logical Survey Microanalysis Center at Stanford University (Stanford, California). Samples were coated with 15 nm of Au. Analysis settings were

South Africa

Australia

Barberton

Greenstone Belt

Pilbara

Block

Mendon Formation ~3.33 Ga

Buck Reef Chert ~3.42 Ga

Wa

rrawo

on

a G

rou

pK

elly G

rou

pS

SG

So

an

esv

ille G

rou

p

Antarctic Chert Member

of the Apex Basalt ~3.47 Ga

MG

FTG

MK

rom

be

rg F

m.

Ho

og

ge

no

eg

Fm

.K

om

ati

Fm

.

On

verw

ach

t G

rou

p

0 km

5

10 Mapepe Formation ~3.26 Ga

Swaziland

Sandstone

Conglomerate

Chert

Various fine-grained

terrigenous rocks

Felsic volcaniclastic rocks

Basalt

Komatiite

Mafic volcaniclastic rocks

Felsic volcanic rocks

Granite

25° S

30° S

20° E 30° E

200 km

120° E 135° E

15° S

30° S600 km

Figure 2. Location maps and general stratigraphic sections of the pre–3.0 Ga sequences included in this study: the Barberton greenstone belt of South Africa (left; modifi ed after Lowe and Byerly, 1999) and the Pilbara block of Western Australia (right; modifi ed after Van Kranendonk, 2011). Abbreviations: M—Mendon Formation; FTG—Fig Tree Group; MG—Moodies Group; SSG—Sulfur Springs Group.

Stefurak et al.


15 kV accelerating voltage, 38 µm spot size, and 39 mm working distance.

Elements of interest (Ca, Mg, Fe, Al, and P or Ti) were mapped in carbon-coated (~14-nm-thick) polished thin sections using a JEOL JXA-8200 advanced electron probe micro-analyzer at the Division of Geological and Planetary Sciences Analytical Facility at the California Institute of Technology (Pasadena, California) and the JEOL JXA-8230 SuperProbe electron probe micro-analyzer at the School of Earth Sciences Mineral Analysis Facility at Stanford University (Stanford, California). Qualitative intensity maps without background correc-tions were collected, operating the electron probe in wavelength-dispersive X-ray spec-trometer (WDS) mode at 15 kV accelerating voltage, 100 nA (Caltech) or 10 nA (Stanford) beam current, and 100 ms (Caltech) or 10 ms (Stanford) dwell time. Color CL images were mapped simultaneously with WDS elemental maps using the JEOL JXA-8230 at Stanford. The sensitivity of the CL of samples to beam damage made collecting high-quality CL maps challenging.

For Si isotope analysis, select samples were made into 1 in. (2.54 cm) circular polished thin sections with standard grains embedded in epoxy as close to the analytical domains of interest as possible, and then carefully polished to obtain a fl at analytical surface for light and electron microscopy, electron microprobe, and secondary ion mass spectrometry (SIMS) analyses. The sample set of silica granule layers includes three samples of banded black-and-white chert from the Mendon Formation, one sample of ferruginous chert with lenses of silica granules from the Mapepe Formation, and one sample of banded iron formation from the Mapepe Formation. We used Caltech Rose Quartz (δ30Si = –0.02‰ ± 0.10‰; Georg et al., 2007) and NBS-28 pure quartz sand (δ30Si = 0‰) as standards. High-resolution silicon iso-tope analyses were performed with the Cameca IMS-7f-GEO ion microprobe at the Caltech Microanalysis Center (California Institute of Technology, Pasadena, California) over three sessions. Samples were sputter-coated with 30 nm of Au for SIMS analysis. Samples were sputtered with an O– primary beam of 3–4 nA with 9 kV acceleration. Analysis spot diam-eters were 30–40 µm, varying slightly between sessions. Sample images in transmitted and refl ected light, in addition to elemental maps acquired with the electron microprobe, were used to avoid placing analytical spots in areas with impurities (particularly iron oxides). A mass resolving power of ~2400 (R = M/ΔM) was achieved, suffi cient for excluding contribu-tions from the 29Si1H hydride peak in 30Si mea-

surements. The 28Si and 30Si ions were detected using fast peak switching with one Faraday cup (resistor value 10 × 10–11 ohms). The primary beam aperture size was ~400 μm in diameter. Critical illumination was used, with the diame-ter of the focused beam of ~15 μm. The focused beam was rastered across an area for the mea-surements, producing effective analyzed areas of ~30 μm in width. The energy bandwidth was 45 eV. Count rates were typically between 5 and 7 × 107 counts/s. Total analysis time for each spot was 8 min, including 60 s presput-tering, fi eld and beam centering, and analyses (20 cycles). Count times were 0.96 s for 28Si and 8.00 s for 30Si. Each set of 12–16 sample measurements was bracketed with 3–4 standard measurements. Each grain of NBS-28 quartz sand accommodated 4–10 separate spot analy-ses, while the larger grains of Caltech Rose Quartz typically accommodated >20 separate spot analyses. Isotope ratios are reported as per mil deviations from the NBS-28 quartz standard (defi ned as 0‰) using delta notation: δ30Si = 1000 × ([Rsample/RNBS-28] – 1). Typical 1σ external (standard-corrected) errors were ±0.30‰ determined by the variation of multi-ple spot measurements on known standards.

RESULTS

Depositional Environments

Antarctic Creek MemberSilica granules in the Antarctic Creek Mem-

ber occur within a sequence of volcaniclas-tic sandstone, laminated gray chert, granular chert, and translucent chert. The volcaniclastic sandstone is current-deposited, moderately- to well-sorted, and well-rounded sandstone, often including accretionary lapilli and less often admixed silica granules. Units of vol-cani clastic sandstone are tens of centimeters to about a meter thick, with individual beds of 5–10 cm in thickness. Cross-laminations and large-scale cross-stratifi cation can be observed locally. Some beds fi ne upward and grade up into centimeter-scale units of fi ne-grained laminated black, gray, or translucent chert, which sometimes contain thin lenses of fi ne- to medium-grained siliciclastic sediment. Similar fi ne-grained, fi nely laminated trans-lucent, gray, and/or black chert also occurs without closely associated underlying vol-cani clastic sandstones and these are typically <10 cm thick. Silica granules occur in cur-rent-deposited, well-sorted, typically massive layers of coarse- to very coarse-grained gran-ules. Sequences of granule beds are commonly ~1 m thick, but they can be as thin as 10 cm and up to several meters thick. Some granule

zones have lenticular geometries, thinning laterally into fi nely laminated black or gray chert on a scale of ~10 m. The bases of granule zones are often characterized by ferruginous granules, while the tops are characterized by iron-free granules. Dune-scale cross-stratifi ca-tion (~20–30 cm scale) is often observed, but, in contrast with the associated volcaniclastic sandstones, grading is absent from granule layers . Some granule layers have erosional caps with evidence of scour and detrital lag deposits containing translucent chert rip-up clasts. Although silica granules occur within some volcaniclastic sandstones, volcaniclastic grains are not observed in granule layers.

Overall, the Antarctic Creek Member where studied is characterized by current-deposited sediments composed of volcaniclastic sand or silica granules interbedded with thin beds of fi nely laminated chert (Fig. DR2 [see footnote 1]). The large-scale cross-stratifi cation and lenticularity of granule beds suggest that they formed in a shallow-water setting characterized by periods of moderate- to high-energy wave or current activity. The volcaniclastic sandstones represent an adjacent current-deposited facies; the fi ning-upward character of some volcani-clastic units suggests that they could be chan-nel deposits. The fi nely laminated, fi ne-grained chert represents lower-energy depositional envi-ronments adjacent to these more current-active settings. The overall sequence is consistent with deposition in a shallow, current-active setting indicating that granules were forming and being deposited in very shallow water.

Buck Reef ChertThe sedimentology, lithofacies, and deposi-

tional environments of the Buck Reef Chert have been extensively studied by previous investiga-tors (Lowe and Byerly, 1999; Lowe and Fisher Worrell, 1999; Tice and Lowe, 2006). The samples in this study are from the lower banded black-and-white chert facies of Tice and Lowe (2006; cf. fi g. 4 of Tice and Lowe, 2006). Black chert bands are composed of coarse sand-sized carbonaceous grains. Brecciation and/or plastic deformation of white chert bands are common, and coarse quartz cavity-fi lling cements often occur beneath white chert bands or plates. The lower banded black-and-white chert facies of the Buck Reef Chert overlies the basal evapo-ritic unit, which includes pseudomorphs after nahcolite and wave ripples (Lowe and Fisher Worrell, 1999). The lower banded black-and-white chert facies is in turn overlain by an upper banded black-and-white chert facies (as defi ned in Tice and Lowe, 2006; this unit is included in the lower Buck Reef Chert facies of Lowe and Byerly, 1999) and is characterized by



fi ne-grained, fi nely laminated, and slightly fer-ruginous banded chert with less common brec-ciation of the white bands. The succeeding unit consists of the banded ferruginous chert facies, which is composed of intergrown microquartz and siderite in unweathered subsurface samples (Tice and Lowe, 2006). Locally, the Buck Reef Chert is capped by another unit of banded black-and-white chert up to 60 m thick (Lowe and Byerly, 1999). The Buck Reef Chert samples included in this study are from the lower black-and-white chert unit. Our observations support the interpretation of Tice and Lowe (2006) that this facies represents a shelfal environment above storm wave base at water depths that were probably between 15 and 200 m.

Upper Mendon FormationAcross the Barberton greenstone belt, the

uppermost Mendon Formation is composed of ~20–75 m of black, banded black-and-white, and banded ferruginous chert (Lowe and Byerly, 1999). The upper Mendon Formation black chert and the black layers in banded black-and-white chert are dominated by fi nely plane-lami-nated, silica-cemented deposits of fi ne-grained carbonaceous matter and micaceous grains, perhaps representing clays or altered volcani-clastic particles. Many black cherts show a fi ne-grained granular texture on weathered surfaces and probably represent mixtures of fi ne ash and carbonaceous matter. Many sections contain <2-cm-thick graded beds of fi ne ash, but these generally lack current structures and prob-ably represent air-fall material. Graded beds of accretionary lapilli occur in some sections. Cur-rent structures are represented mainly by small, widely spaced cross-laminations. Scour features include small, 2–5-cm-deep and 1–15-cm-wide scour pits, but large-scale cross-stratifi cation, scour, and evidence for string current activity and erosion are absent.

Brecciation of white bands in banded black-and-white chert is less common than in the lower black-and-white banded chert facies of the Buck Reef Chert, likely refl ecting a lower-energy paleoenvironment. Banded ferruginous chert of the upper Mendon Formation is defi ned by alter-nating white chert bands, and fi nely laminated ferruginous chert bands. Ferruginous chert is often slightly rust-colored on the surface, prob-ably representing a sideritic component at depth.

The upper Mendon Formation represents a relatively quiet depositional setting below storm wave base and therefore a slightly deeper water environment (>~200 m water depth) than the lower black-and-white banded chert facies of the Buck Reef Chert.

Lower Mapepe FormationWithin the study area in the central part of

the Barberton greenstone belt, the Mapepe Formation has an apparently conformable contact with cherts of the upper Mendon For-mation (Lowe and Byerly, 1999). This contact is widely marked by a bed of spherules formed through a large meteorite impact (Lowe and Byerly, 1986). The Mapepe Formation itself shows a wide variety of facies deposited in a range of environments from shallow to deep water that vary from one structural belt to another within the central and southern domains of the Barberton greenstone belt. These complex facies and depositional set-tings refl ect the initiation of uplift and defor-mation within the belt, providing sources of clastic sediment and adjacent basins for its deposition and the accumulation of associated deep-water, generally iron-rich chemical sedi-ments. Samples from the Mapepe Formation included in this study are from ferruginous lithofacies representing relatively quiet, deep-water sedimentation—ferruginous cherts and banded iron formation.

Petrography and Sedimentology of Silica Granules

Despite some variability, several key proper-ties characterize granules from all depositional settings. Granules are everywhere sand-sized subspherical grains with shapes ranging from nearly spherical to oblate spheroidal. The mean diameters for volume-equivalent spheres from representative samples from a variety of deposi-tional settings range from 216 to 623 μm (fi ne- to coarse-grained sand; Table 1). In addition, all granules appear to have been relatively soft and easily compacted. Granules are now composed of relatively pure microquartz, and the major-ity of granules lack internal structures such as concentric laminations, irregular distributions of trace impurities, relict textures refl ecting a different primary mineralogy, or cracks formed during dehydration and shrinkage of an initially hydrated gel phase. Some iron-associated gran-ules do display a crude structuring, with interi-ors containing trace iron oxides and rims with more abundant iron phases.

Some white bands within banded black-and-white chert of the Buck Reef Chert and the upper Mendon Formation are composed exclusively of granules. These bands are gen-erally 1–10 cm thick. Both matrix-supported (Fig. 1C) and grain-supported (Fig. 3B; Fig. DR1 [see footnote 1]) textures are observed within granule bands. Although structures like cross-laminations and grading are not com-mon, they do occur in rare examples (Fig. DR1 [see footnote 1]). While many white bands are tabular and laterally continuous, others lens out laterally, and local brecciation is common (Fig. 4A). The bands of silica granules have sharp contacts with the adjacent black chert bands, which are composed primarily of silica containing detrital organic grains (Fig. 3B). Compaction of granules is common within the

TABLE 1. RAW AND CORRECTED GRANULE DIAMETERS

Sample name* LithologyDepositional environment n

Aspect ratio

μ,†

raw#

(µm)

σ,§

raw (µm)

Φ,** raw

μ, corrected

(µm)

σ, corrected

(µm)

Equivalent sphere

diameter (µm) Φ** Grain size

AUS 36-1 Granular chert Subtidal 110 1.7 498 206 0.69 622 202 517 0.95 Coarse sand

AUS 36-5 Granular chert Subtidal 50 1.9 639 268 0.37 772 288 623 0.68 Coarse sand

dnaseniF12.261273165279.18114027.1003laflehStrehckcalB31-20-RHB

dnasesraoC78.094527383656.05233056.1201laflehStrehckcalB61-20-RHB

SAF 521-16 Banded black-and-white chert Shelfal 52 2.0 236 84 1.79 289 81 230 2.12 Fine sand

SAF 183-2 Ferruginous chert Basinal 150 2.2 294 159 1.45 365 181 281 1.83 Medium sand

SAF 649-14 Banded iron formation Basinal 300 2.2 302 166 1.37 386 187 299 1.74 Medium sand*See supplementary Table DR1 for more detailed stratigraphic and location information for these samples (text footnote 1).†μ—mean granule diameter.§σ—standard deviation of granule diameters.#“Raw” values refer to the grain diameters measured from thin sections prior to being corrected for the effects of the thin section providing random cuts through grains,

as described in the methods section.**Φ = –log2 (grain diameter in mm); Φ is the Krumbein phi scale, a standard logarithmic expression of grain size.

Stefurak et al.


white bands (maximum aspect ratios of ~9:1; Fig. 4), often obscuring primary textures. The surrounding matrix is microquartz, visually distinguishable due to trace organic matter or sider ite inclusions. In contrast with the com-mon compaction of silica granules in white bands, carbonaceous grains in adjacent black chert bands typically show little to no com-paction. This is seemingly in confl ict with the sequence of cementa tion implied by chert plate breccias, where white chert bands are often deformed as relatively rigid and impermeable plates while the surrounding carbonaceous sediment was still unlithifi ed. These observa-tions suggest that, unlike silica granules, many organic grains were at least partially cemented very early, prior to deposition, making them more resistant to compaction. At the same time, in situ silica cementation occurred much earlier in pure layers of silica granules compared to layers of carbonaceous sediment.

500 μm

2 cm

AA BBB

CC DD

1 mm

500 μm

500 μm 500 μm

EE FFF

GG

1 mm

Figure 3. Outcrop (A) and thin-section (B–G) images of representative occurrences of silica granules associated with black-and-white banded chert and black chert. (A) Outcrop photo of white chert plate breccia, Buck Reef Chert (locality Buck Reef Chert). (B) Sharp contact between silica granule layer (white chert band) and adjacent detrital organic layer (black chert band), upper Mendon Formation (locality SAF 521). (C) Silica granules mixed with fi ner-grained detrital organic grains, lower Mapepe Formation (locality BHR-02). (D) Large silica granules associated with a variety of organic and volcanic grain types, upper Mendon Formation (locality SAF 521). (E) Silica granule intraclast within an intraclast breccia, upper Mendon Forma-tion (locality SAF 186). (F) Silica granules, intraclasts, detrital organic grains, and volcanic grains within an intraclast brec-cia bed from the same sample as E, upper Mendon Formation (locality SAF 186). (G) Organic grains and silica granules from a black chert, upper Mendon Formation (locality CQ-01). Silica granules (arrows) occur within the bed and at the base of the overlying graded bed. These granules are similar in size to and in some cases slightly coarser than the associated detrital organic grains. See supplementary Table DR1 for additional stratigraphic information for samples shown and supplementary Figure DR1 for map of sample localities (see text footnote 1).



Within nonbanded chert of the upper Men-don Formation, granules also commonly occur mixed in detrital beds with other sedimentary grain types, including: carbonaceous grains and intraclasts of carbonaceous chert, volcanic grains and intraclasts of volcaniclastic sediments, and chert band intraclasts (Figs. 3C–3G). In occur-rences with organic grains, the mean aspect ratios of silica granules are similar to other types of occurrences (2:1; Fig. 4), although in some cases, particularly when the organic material occurs as dispersed, very fi ne-grained inclusions rather than as clearly detrital grains, the silica granules can be quite compacted (~10:1). Within intraclast breccia units, maximum compaction observed (~2.5:1) is much less severe, likely refl ecting differences in timing of silica cemen-tation between these lithofacies.

In the Antarctic Creek Member, coarse sand-sized, slightly compacted silica granules (aspect ratios in the range of 1.2:1–5:1, mean aspect ratio 1.9:1) occur in 10-cm- to 1-m-thick, grain-supported and often cross-stratifi ed beds. Some granules from the Antarctic Creek Member are composed purely of microquartz surrounded by drusy quartz cements (Figs. 5A–5B). Other granules contain iron minerals closely associ-ated with the granules as rims of fi ne iron-oxide grains (Figs. 5C–5D) and/or dispersed evenly throughout granule interiors (Figs. 5E–5F).

Lenticular centimeter-scale layers of granules occur within ferruginous chert of the Mapepe Formation, which consist of fi nely laminated sideritic chert (Figs. 1D–1F). Some of these granules are relatively pure silica (Figs. 1F and 5G–5H), while others have trace inclusions of micron-scale iron oxides (Fig. 1E). Most gran-ules are slightly to moderately compacted with mean aspect ratios of ~2:1 (Fig. 4). The sur-rounding matrix or cement is primarily micro-quartz, although often containing fi nely dis-persed hematite. The lenticular granule layers could represent cross sections of starved ripples, although the examples examined in this study do not display any cross-lamination.

Silica granules also occur as centimeter-thick lenticular layers within banded iron formation in the lower Mapepe Formation (Figs. 1G–1I and 5I). Excluding the granule lenses, the Mapepe banded iron formation is defi ned by fi nely laminated hematitic, sideritic, and cherty layers with some interbedded layers of claystone or mudstone. The banded iron formation–associ-ated granules are composed of relatively pure silica, while the surrounding matrix/cement contains trace hematite and siderite. Granules are slightly to moderately compacted with mean aspect ratios of ~2:1 (Fig. 4). As with granules in the ferruginous chert, the lenticular layers in the banded iron formation may represent starved ripple forms, but we have not observed cross-laminations that would confi rm this inter-pretation, and some granule layers appear to be matrix-supported (e.g., Fig. 1I).

Geochemistry

Electron Microprobe MapsElectron probe elemental maps confirm

petrographic observations showing that gran-ules are essentially pure SiO2 (Fig. 6; Figs. DR4 and DR5 [see footnote 1]). Some gran-ules associated with other iron-bearing facies contain trace inclusions of hematite (Fig. 6). Iron-bearing carbonates, primarily siderite with minor ankerite and ankeritic dolomite, occur within the matrix to the granules, but not within the granules themselves. Granule samples were

mapped for either P or Ti, two poorly soluble elements that might reveal possible precursor mineralogy, but there was essentially no P or Ti in any of the samples analyzed. Occurrence of Al was limited to matrix material with trace amounts of clay minerals, but these phases do not occur in granule layers.

CathodoluminescenceSilica granules and the adjacent matrix or

cement are not distinguishable in gray-scale CL (Fig. DR6 [see footnote 1]), but some granules are distinguishable using color CL (Fig. 6C). In the example shown in Figure 6, granules con-taining fi ne trace inclusions of hematite are dis-tinguishable from the surrounding cement by their brighter luminescence. Petrographically, the granules appear to be compositionally vari-able; similarly, CL brightness varies between granules. CL maps do not reveal any internal textures or structuring (e.g., concentric layer-ing) apart from mirroring petrographic textures refl ecting slight variations in the abundance of hematite inclusions. The peak emission wave-length is ~640 nm (red). Apart from intensity, there were no distinct or systematic differences between CL spectra within the granules versus within the surrounding cement.

Si Isotope GeochemistrySi isotopes are fractionated by many low-

temperature precipitation processes and can be useful in differentiating different generations of silica and in tracking the cycling of silica (Chakrabarti et al., 2012; Heck et al., 2011; Marin-Carbonne et al., 2012; Stefurak et al., 2015; van den Boorn et al., 2007, 2010). The Si isotopic compositions of silica granules mea-sured in this study had a mean δ30Si of –0.28‰ and standard deviation of 1.42‰ (n = 338), although analysis of variance (ANOVA) and multiple comparison tests indicated that sam-ple means of granules associated with banded black-and-white chert, banded iron formation, and ferruginous chert were signifi cantly dif-ferent (Fig. 7; Tables DR2 and DR3 [see foot-note 1]). Banded black-and-white chert–associ-ated silica granules had mean δ30Si of 0.77‰ and standard deviation of 0.45‰ (n = 189), while banded iron formation–associated granules had mean δ30Si of –0.70‰ and standard deviation of 0.67‰ (n = 68), and slightly ferruginous granules occurring in a lens within ferruginous shale had a mean δ30Si of –2.36‰ and standard deviation of 0.83‰ (n = 81). While neither the banded iron formation–associated nor the ferru-ginous chert–associated silica granules had sig-nifi cantly different mean δ30Si than the adjacent matrix or cement (Fig. 7), ANOVA and multiple comparison tests indicated a small, but statisti-

2

4

6

8

10

0

shal

low

BBW

Cb

lack

ch

ert

FC BIF

Gra

nule

asp

ect r

atio

intr

acla

st b

recc

ia

Figure 4. Measurements of aspect ratios of silica granules associated with different litholo gies. Mean (circle), standard deviation (line), minimum (diamond), and maximum (square) are given for each data set. Ellipses next to the y-axis indicate dimensions for as-pect ratios of 2, 4, 6, and 8. Abbreviations: BBWC—banded black-and-white chert, FC—ferruginous chert, and BIF—banded iron formation.

Stefurak et al.


cally signifi cant difference in the sample means of banded black-and-white chert–associated granules (mean δ30Si = 0.77‰) and the associ-ated microquartz matrix (mean δ30Si = 1.00‰; Table DR2 [see footnote 1]). The three banded black-and-white chert samples and the fer-ruginous chert sample were all collected from the same continuous stratigraphic section (Fig. DR7; Table DR1 [see footnote 1]; the banded black-and-white chert samples were collected within the upper 3 m of the Mendon Formation, and the ferruginous chert sample was collected from 54 m into the lower Mapepe Formation. The granules analyzed in the three banded black-and-white chert samples have similar δ30Si distributions and nonstatistically signifi -cant means (Fig. 7B), but they contrast strongly with the ferruginous chert sample, indicating in this case there is some relationship between observed granule Si isotopic composition and lithofacies and/or stratigraphic position. Silica granules from banded black-and-white chert have similar Si isotopic compositions to bulk analyses of samples from van den Boorn et al. (2007, 2010) interpreted to have formed via silicifi cation by seawater silica, while the iron-

associated silica granules in banded iron for-mation and ferruginous chert are broadly con-sistent with many previous analyses of various Archean banded iron formations fi nding them to have distinctly low δ30Si compositions (Fig. 7C; Andre et al., 2006; Chakrabarti et al., 2012; Del-vigne et al., 2012; Ding et al., 1996; Heck et al., 2011; Steinhoefel et al., 2009, 2010).

DISCUSSION

Petrography and Environmental Distribution

The unstructured nature of iron-free gran-ules and lack of relict textures resulting from dia genetic transformations differentiate silica

500 μm

200 μm

200 μm200 μm200 μm200 μm

200 μmPPL XPL

C D

E F

GG H

1 mm 1 mm

1 mm

PPL XPL

500 μm 500 μm

A B

II

PPL XPL

DD

DD

QQ

QQ

Figure 5. Thin-section images of represen-tative occurrences of silica granules associ-ated with ferruginous chert, banded iron formation, and chert closely associated with ferruginous lithofacies. (A–B) Pure silica granules from the Antarctic Creek Member (locality AUS 36) in plane-polarized light (PPL; A) and cross-polarized light (XPL; B). Note drusy quartz cement (red D) in some pore spaces and the replacement of some granule interiors with coarse quartz (yellow Q). (C–D) Iron-oxide rims on silica granules similar to those shown in (A–B), in plane light (C) and cross-polarized light (D), from the Antarctic Creek Member (locality AUS 36). (E–F) Varied, somewhat ferruginous granules from the Antarctic Creek Member (locality AUS 36); image in F shows a close-up of the iron-rimmed grain from the lower-left corner of E. (G–H) Lens of compacted silica granules within ferruginous shale of the Mapepe Formation (locality SAF 183), in plane light (G) and cross-polarized light (H). (I) Thin lens of silica granules within Mapepe Formation banded iron formation (locality SAF 649). See supplementary Table DR1 for additional stratigraphic informa-tion for samples shown and supplementary Figure DR1 for map of sample localities (see text footnote 1).



granules from other grain types like accretionary lapilli (Lowe, 1999b), impact spherules (Lowe and Byerly, 1986), volcanic grains (DiMarco and Lowe, 1989), ooids, or other primary car-bonate grains (Maliva et al., 2005). Although silica diagenesis, which involves several stages of dissolution and reprecipitation (Williams et al., 1985), could theoretically have obscured some primary textures, it is unlikely that textures defi ned by insoluble impurities like organic mat-ter, iron oxides, or clays, and elements like P, Ti, and Al would have been completely obscured in every granule occurrence. At the same time, we cannot completely rule out the possibility that the granules initially contained some primary heterogeneities that have been homogenized during diagenesis, such as structuring of the original amorphous silica phase or inclusions of more soluble minerals like carbonates. In the latter case, the strongest evidence that the granules did not originate as carbonate grains is the lack of carbonate inclusions, ghosts, or even rare examples of only partially silicifi ed gran-ules such as those commonly observed in later Precambrian silicifi ed carbonates (e.g., Maliva et al., 2005). The ubiquitous homogeneity of iron-free silica granules throughout a range of depositional and dia genetic settings suggests that they originated as compositionally homo-geneous particles.

Some iron-associated granules do show a rudimentary structuring defi ned by rims of iron oxides (Figs. 1E and 5C–5F). Some of these granules also include trace iron phases within the granule interiors (Figs. 1E and 5E). Although the distribution of iron within gran-ules is slightly heterogeneous, it does not defi ne concentric laminations as in oolitic iron granules that occur in some granular iron for-mations (Beukes and Klein, 1990; Dimroth and Chauvel, 1973; Goode et al., 1983; Gross,

1972; Simonson, 1987, 2003). This indicates the coprecipitation of silica and iron phases, rather than alternating deposition of silica and iron layers. The iron-rich rims, on the other hand, appear to have accumulated after silica deposi-tion ceased. These distribution patterns of iron in and around silica granules are both consistent with the hypothesis that granules formed rap-idly, in some cases incorporating iron, likely as iron oxyhydroxides or carbonates. Afterwards,

additional iron phases could collect on the gran-ule surfaces, perhaps as they were reworked by currents.

Silica granules occur in many depositional settings and water depths, from shallow subtidal (Antarctic Creek Member) to shelfal in the Buck Reef Chert to deep basinal (Mapepe banded iron formation). They also occur in different types of deposits, from lenticular or tabular layers com-posed purely of granules (Fig. 1), to intraclast breccia units with a variety of grain types (Figs. 3D–3F), to black chert composed mostly of organic grains with a minor component of silica granules (Figs. 3C and 3G). The similarities of granule properties, including size, shape, sus-ceptibility to compaction, lack of internal struc-ture, and composition, across a wide range of depositional environments also suggest that the granules share a common origin. The uniformly sand-sized, subspherical, well-rounded granules are inconsistent with formation as rip-up clasts, which would be expected to display a range of grain sizes, shapes, and internal textures. The occurrence of granules in shallow-water depositional environments indicates that gran-ules probably formed within the shallow water column. Within the deepest-water litho facies considered here, the banded iron formation of the lower Mapepe Formation, granule layers are relatively thin and irregularly spaced (Fig. 1H). This pattern suggests either that there were some processes preventing the delivery or pres-ervation of granules in deep-water environments (e.g., obliteration during diagenesis) or that con-ditions for granule formation were more favor-able in the shallow water column above shelfal environments, such that current transport was required for granules to be delivered to more basinal environments.

One notable aspect of silica granules is their relationship to the phenomenon of banding: many relatively pure, whitish to translucent chert bands or lenses within banded black-and-white chert and banded ferruginous chert show granular textures and were therefore deposited as beds of silica granules. In theory, hydraulic sorting could explain this juxtaposition of pure granule layers with layers of contrasting com-position if both layers were current-deposited and there were density contrasts between gran-ules and other grain types. These conditions are inconsistent with a number of observations, which suggest that the two band types represent separate depositional events. The compositional contrasts between white chert bands composed of silica granules and the adjacent black chert bands (Fig. 3B) or laminated ferruginous sedi-ment (Figs. 5G–5I) are sharp rather than grada-tional, and silica granules commonly occur as a minor component within black chert bands,

14

0

12

0

424

37

Si (

cou

nts

)

Fe (

cou

nts

)

CL,

55

3-7

17

nm

500 μm

500 μm

500 μm

A

B

C

PPL

XPL

Figure 6. Petrographic textures in slightly ferruginous granules (A–B) correspond with relative intensity maps of Si and Fe (using wavelength-dispersive X-ray spectroscopy [WDS]) and color cathodoluminescence (CL) (C). Thin-section images in plane-po-larized light (PPL; A) and cross-polarized light (XPL; B) show differences in composi-tion between the granules and a contrast in crystallinity among granules and between granules and cement. The CL data (C, red) show that the granules have slightly higher CL intensity than the adjacent cement. See supplementary Table DR1 for additional stratigraphic information for sample shown (see text footnote 1).

Stefurak et al.


neither of which is consistent with hydraulic sorting of compositionally distinct layers as parts of the same depositional event. Evidence of deposition by currents does not occur consis-tently throughout banded chert facies: Although the carbonaceous grains in some black chert bands (e.g., within the lower facies of the Buck Reef Chert; see Tice and Lowe, 2006) appear to have been deposited and/or reworked by cur-rents, granular white chert bands can also be jux-taposed with more fi ne-grained carbonaceous or ferruginous chert facies representing quiet sus-pension settling. Hydraulic sorting alone cannot explain the occurrence of relatively pure granule layers juxtaposed with different primary sedi-ment types and in different depositional envi-ronments. Instead, deposition of silica granules must have occurred only episodically.

Composition of Granules and Associated Grains, Matrix, and Cements

The composition of other grains and matrix materials associated with silica granules depends on the depositional setting. When cements are present, they are always silica cements, support-ing the inference that the concentration of dis-solved silica in pore fl uids was high, either due to the occurrence of primary amorphous silica (e.g., the granules themselves) or as a consequence of a high concentration of silica in seawater. Black cherts are commonly dominated by silicifi ed organic grains and a microquartz matrix often containing trace carbonaceous material, while deeper-water iron-rich sediments juxtapose len-ticular granule layers with much fi ner-grained laminated ferruginous sediment composed of mixtures of hematite, siderite, phyllosili-cates, and silica. Similarly, the matrix to silica granules within black chert is microquartz that commonly contains trace organic matter, while the matrix within ferruginous settings contains fi ne hematite, siderite, and silica. The Antarctic Creek granules are notable in the co-occurrence of iron-bearing minerals as trace constituents within granules. Most ferruginous chert–asso-ciated and banded iron formation–associated granules in Barberton samples contain little to no iron (Figs. 1F, 1I, and 5G–5I), except for one isolated example, a lens of loosely packed, grain-supported, well-sorted, subangular iron-bearing granules (Fig. 1E) occurring within laminated ferruginous chert (Fig. 1D) within 2 cm of a more typical silica granule layer (Fig. 1F). These granules appear to be current-depos-ited and may represent a turbidite composed of grains transported from shallower water, which could explain the contrast in composition, sort-ing, rounding, packing, and compaction with the adjacent granule lens.

FC cht cement (N=1; n=24)

FC Fe granules (N=1; n=81)

BIF Fe matrix (N=1; n=92)

BIF granules (N=1; n=68)

BBWC carbonaceous

matrix (N=1; n=17)

BBWC cht matrix

(N=2; n=110)

BBWC granules

(N=3; n=189)

0.4

−4 −2 0 2 4

0

0.8

1.2

1.6

δ SiNBS-28 (‰)

Den

sity

SAF 521−14 (n=18);

μ=0.6‰, σ=0.4‰

SAF 521−15 (n=110);

μ=0.7‰, σ=0.4‰

SAF 521−16 (n=60);

μ=0.9‰, σ=0.5‰

BIF, previous studies

(n=359)

van den Boorn et al.

S−cherts (n=22)


BIF granules (N=1; n=68)

BBWC granules

(N=3; n=189)

0.4

0

0.8

1.2

1.6

Den

sity

0.4

0

0.8

1.2

1.6

Den

sity

A

B

C Data from previous studies:

Data from this study:

BBWC granules by sample:


μ=-2.4‰, σ=0.8‰

FC granules from samestratigraphic section:

Figure 7. Kernel-smoothing probability density estimates of δ30Si data from silica granules and associated matrix and/or cement associated with different lithologies. (A) All data from this study, divided by lithofacies and texture. The samples included are from banded black-and-white chert (BBWC; black), banded iron formations (BIF), and ferruginous chert (FC). Spot analyses from within granules (solid lines) are differentiated from the associated matrix or cement (dotted or dashed lines). The number of samples (N) and the total number of spot analyses (n) are distinguished for each category. The three samples of the banded black-and-white chert are from within the same section of the upper Mendon Formation. The ferru-ginous chert sample is from the lower Mapepe Formation higher in the same stratigraphic section as the upper Mendon samples. The banded iron formation sample is from the BARB4 core drilled in the Manzimnyama syncline. (B) Spot analyses from within silica granules of each of the banded black-and-white chert samples and the ferruginous chert sample collected from ~55 m higher in the same stratigraphic section. The means of the banded black-and-white chert samples (listed in key to the right) are not statistically distinct, but the ferruginous chert sample is clearly different. (C) Comparison of granule data from this study (solid lines) with Si isotope data from previous studies (dashed lines) of cherts interpreted to record a sea-water δ30Si signature (van den Boorn et al., 2007, 2010) and many measurements of Archean banded iron formations (Andre et al., 2006; Chakrabarti et al., 2012; Delvigne et al., 2012; Ding et al., 1996; Heck et al., 2011; Steinhoefel et al., 2009, 2010).



The overall CL intensity of microquartz from the cherts in this study is low, but color CL allows the recognition of some contrasts in CL intensity that mirror petrographic textures (Fig. 6). The peak CL wavelength observed using color CL (Fig. 6) is ~640 nm, an intrin-sic wavelength observed in most quartz due to nonbridging oxygen hole centers (Götze et al., 2001). Red emissions strongly dominate, and overall CL intensity is relatively low, both of which are common features in low-grade meta-morphic rocks (Boggs and Krinsley, 2006; Boggs et al., 2002). The samples do not display spectral peaks indicative of structural substitu-tions of other elements like Fe3+ (705 nm) or Al (several possible peaks in the green or blue range; Götze et al., 2001).

Granule Compaction

Most iron-free granules appear to have been soft when deposited, based on the common occurrence of slight to severe postdepositional compaction. The degree of compaction varies between samples (Fig. 4). Silica granules within white bands of banded black-and-white chert include the most severely compacted examples, even though independent textural evidence indi-cates that the white bands lithifi ed early (Knauth and Lowe, 1978, 2003; Lowe, 1999a). The lesser degree of compaction of granules in other lithologies might indicate either that granules had already begun to lithify before being depos-ited or that some local aspect of depositional conditions prevented the granules from being as severely compacted. Alternately, organic-rich sediments may have been more fl uid-like during compaction, with the result that granules mixed with black chert precursor sediments would have experienced lower compaction stresses than in grainstone-like granule layers. In the case of intraclast breccias, the abundant lithifi ed or partially lithifi ed intraclasts (Figs. 3E–3F), often including angular chert plate fragments, could have helped protect the relatively minor component of silica granules from the effects of compaction. In the case of deeper-water ferruginous chert–associated or banded iron formation–associated granule lenses, the less-extreme compaction likely resulted from some combination of aging and strengthening of the amorphous silica during transport and/or set-tling, earlier cementation due to differences in diagenetic conditions, and a slower background depositional rate than that of banded black-and-white chert, resulting in smaller forces exerted by overlying sediment prior to cementation.

Silica granules containing iron from the Ant-arctic Creek Member (Figs. 5E–5F) and an iso-lated lens in ferruginous chert of the Mapepe

Formation (Fig. 1E) include some oblong grain shapes, but these grains are loosely packed, and the axes of grain elongations are often not aligned, indicating that many of these grains were irregular in shape when they were depos-ited, rather than being deformed during burial. In contrast with the uniformly rounded nature of iron-free silica granules, some iron-bearing silica granules are angular to subangular. These observations both suggest that coprecipitation of silica and iron resulted in earlier lithifi cation of iron-bearing silica granules, making them more likely to be broken or abraded by grain collisions but less susceptible to postdeposi-tional compaction.

Relationship to Granular Iron Formation

The silica granules described here bear some resemblance to granules in granular iron forma-tion in that they are well-rounded, uniquely Pre-cambrian chemical sand grains (Bekker et al., 2010, 2014; Klein, 2005; Simonson, 2003; Trendall, 2002). Some granular iron formations contain oolitic grains with concentric layering, although non-oolitic iron granules are much more common (Bekker et al., 2010; Beukes and Klein, 1990; Dimroth and Chauvel, 1973; Goode et al., 1983; Gross, 1972; Simonson, 1987, 2003). Although granular iron formations are rich in SiO2 (~50 wt%), they also contain a signifi cant amount of iron (>15 wt%, but com-monly ~30 wt%) in the form of iron oxides or iron silicates (Bekker et al., 2010, 2014; Klein, 2005; Simonson, 2003; Trendall, 2002). While the silica granules described here co-occur with iron oxides or carbonates in some settings, they also commonly occur without associated iron minerals. Even when iron is associated with the silica granules, it is only a trace constitu-ent within the chert. The silica granules here are clearly distinct from the iron-oxide and iron-silicate granules that occur in granular iron formations.

At the same time, many previous authors have reported the occurrence of silica-rich granules in addition to iron-rich granules within granu-lar iron formations (Bekker et al., 2010; Beukes and Klein, 1990; Dimroth and Chauvel, 1973; Goode et al., 1983; Gross, 1972; Simonson, 1987, 2003). These silica-rich granules appear to commonly have trace inclusions of iron phases, much like the iron-bearing granules we observe in the Antarctic Creek Member (Figs. 1E–1F) and the isolated lens in the Mapepe Formation (Fig. 1E). Furthermore, most granu-lar iron formations are also current-reworked deposits of sand-sized chemical grains repre-senting relatively shallow shelfal environments (Beukes and Klein, 1990; Gross, 1972; Simon-

son, 1985, 2003). The iron-free and iron-bearing silica granules detailed in this study may there-fore represent points on a spectrum of silica granule compositions, with shallow-water gran-ules apparently containing more iron than those deposited in deeper-water environments. The main distinction, therefore, between the shal-low-water ferruginous chert–associated silica granules in this study and granular iron forma-tions is the absence of associated iron-oxide or iron-silicate granules. Since most granular iron formations are latest Archean or Proterozoic in age (Bekker et al., 2014; Simonson, 2003), this distinction could be related to the rise of oxygen.

Si Isotope Geochemistry

The contrasts between the δ30Si composi-tions of silica granules and associated matrix and/or cement from banded black-and-white chert (three samples, 316 spot analyses), banded iron formation (one sample, 160 spot analy-ses), and ferruginous chert (one sample, 105 spot analyses) are striking (Fig. 7), although it is important to point out that this is a relatively small data set and therefore may not refl ect the full range of δ30Si variability within each of these lithofacies. At the same time, the data pre-sented here are broadly consistent with analyses from previous Si isotopic studies. The granules from banded black-and-white chert are similar in Si isotopic composition to previous analyses of chert not associated with iron, interpreted to refl ect a seawater δ30Si signature (Fig. 7C; van den Boorn et al., 2007, 2010; Stefurak et al., 2015). Similarly, the banded iron formation and ferruginous chert samples are isotopically depleted relative to the mean value of bulk sili-cate earth (δ30Si = –0.4‰), consistent with pre-vious analyses of Archean banded iron forma-tions (Fig. 7C; Andre et al., 2006; Chakrabarti et al., 2012; Delvigne et al., 2012; Ding et al., 1996; Heck et al., 2011; Steinhoefel et al., 2009, 2010). There are two main explanations for these distinctions: Either (1) the mecha-nisms forming these different granule types had drastically different Si isotopic fractionation factors (ε = δ30Siprecip – δ30Sifl uid) between fl uid and precipitate, or (2) a substantial amount of silica from isotopically contrasting fl uids was contributed to each granule association during initial granule formation and/or diagenesis (e.g., the granules associated with ferruginous chert incorporated silica from fl uids of much lower δ30Si than the banded black-and-white chert granules).

Previous authors have suggested that Si iso-topic fractionation associated with coprecipita-tion of silica and iron oxyhydroxides (hematite precursor) was responsible for the low δ30Si

Stefurak et al.


values observed in banded iron formations (Chakrabarti et al., 2012; Fischer and Knoll, 2009). In theory, this could explain why silica granules with ferruginous inclusions (Fig. 7A, “FC Fe granules”) have lower δ30Si than the iron-free silica granules (Fig. 7A, “BIF gran-ules”), i.e., that Si isotopic fractionation was enhanced by coprecipitation with iron in the former type but not the latter. However, the iron-poor silica granules in the banded iron forma-tion sample still have signifi cantly lower δ30Si than the banded black-and-white chert granules, so differences in fractionation during primary granule formation cannot explain the full range of variability observed in our data.

The different granule types must therefore have incorporated silica from isotopically con-trasting fl uids: higher δ30Si for the shallower-water banded black-and-white chert–associated granules, intermediate δ30Si for the banded iron formation–associated granules, and surprisingly low δ30Si for the ferruginous chert–associated granules. Since banded iron formation repre-sents the deepest-water depositional environ-ment of these three lithofacies, there is no linear relationship between paleo–water depth and Si isotope composition of silica granules. The lack of isotopic contrast between granules and the adjacent matrix or cement in the iron-associated lithologies (Fig. 7A) furthermore indicates that the cements precipitated from fl uids of similar Si isotopic composition to the adjacent gran-ules, perhaps due to exchange of silica during the stages of dissolution and reprecipitation characterizing phase transformations from opal-A to opal-CT and fi nally to microquartz. These inferences fi t with the previous suggestion that there was a δ30Si chemocline in Precambrian oceans (Chakrabarti et al., 2012), but they are seemingly in confl ict with the conclusion that all granules originated in relatively shallow water (Stefurak et al., 2014), which would suggest that they should have shared similar initial isotopic compositions. There are two main scenarios that could have produced the observed patterns: (1) primary Si isotopic variability inherited from depth-related and/or lateral variations in seawater δ30Si, or (2) overprinting of a smaller range of primary granule δ30Si compositions by distinct, locally derived δ30Si signatures during diagenesis.

In reality, both primary and diagenetic pro-cesses probably contributed to the observed Si isotopic variability of silica granules, although a larger sample set is required to better evalu-ate their relative contributions. In one example (Fig. DR8 [see footnote 1]), δ30Si does appear to be sensitive to primary textural boundaries, with lower δ30Si in cements and some granules (mean δ30Si = –3.9‰), while other adjacent granules

have higher δ30Si (mean δ30Si = –2.0‰). This distinction indicates that, at least in this case, fl u-ids with a particularly light δ30Si value interacted with these granules during diagenesis. Given the typical fl uid-precipitated Si fractionation factor of –1‰ to –2‰ (Basile-Doelsch et al., 2005; Ding et al., 1996; Geilert et al., 2014), cements with δ30Si of –3.9‰ likely precipitated from a fl uid with a δ30Si value in the range of –2‰ to –3‰, which is signifi cantly lighter than esti-mates for Archean seawater, with δ30Si in the range of ~+2‰ to +3‰ (Abraham et al., 2011; van den Boorn et al., 2007, 2010), or analyses of modern hydrothermal fl uids, which yielded a δ30Si of –0.3‰ (De La Rocha et al., 2000). Si isotopic compositions this depleted require light δ30Si generated via one or more previous stages of precipitation/dissolution or adsorption/desorption. Both banded iron formation– and ferruginous chert–associated granules are adja-cent to layers that are rich in iron. Adsorption of silica onto iron oxides is known to fractionate Si isotopes (Delstanche et al., 2009). Release of this adsorbed silica with light δ30Si could have contributed to the light isotopic signatures observed in our banded iron formation and fer-ruginous chert granule samples. The laminated ferruginous chert adjacent to the lenses of gran-ules also contains phyllosilicates. Authigenic clays can have fairly light δ30Si compositions (mean δ30Si = ~–1.4‰) due to preferential incorporation of 28Si (De La Rocha et al., 2000; Ding et al., 1996; Douthitt , 1982; Georg et al., 2009; Opfergelt et al., 2010, 2012; Ziegler et al., 2005), so any release of silica from these phases during diagenesis could also have contributed to pore fl uids with low δ30Si compositions. The behavior of Si isotope compositions of chert during diagenesis has important implications for determining which silica phases record sea-water chemistry rather than diagenetic fl uids. The variability of δ30Si compositions of silica granules observed here suggests that future analyses of large silica granule data sets can pro-vide important insights into seawater δ30Si and the fractionation of Si isotopes under different diagenetic conditions.

Processes of Granule Formation

Silicic acid molecules (H4SiO4) combine and grow into particles by polymerization. This polymerization usually occurs via dehydration reactions that convert two silanol groups into a siloxane bond (i.e., R-Si-O-H + H-O-Si-R ↔ R-Si-O-Si-R + H2O). The aqueous chemistry of silica at low temperatures is dominated by kinetics: Although a solution is technically satu-rated with respect to quartz at 6–10 ppm SiO2, monosilicic acid is relatively stable in solution

up to 100–200 ppm (the solubility of opal-A, depending on temperature; Iler, 1979). In solu-tions supersaturated with respect to opal-A, sil-ica monomers polymerize fi rst to dimers, then cyclic polymers, and fi nally to compact spheri-cal particles, which grow via Ostwald ripening to stable colloidal particles 5–10 nm in diameter (above pH 7; Iler, 1979).

Modern siliceous sinters provide a natural system for studying silica precipitation that is potentially analogous to Archean oceans. Some modern and ancient sinter deposits contain sili-ceous oncoids 1–10 cm in diameter (Guidry and Chafetz, 2003; Jones and Renaut, 1997; Jones et al., 1999, 2001; Lynne et al., 2008; Renaut et al., 1996) or pisoids (Lowe and Braunstein, 2003; Walter, 1976; Walter et al., 1996). These large siliceous grains are not appropriate ana-logues for silica granules: Apart from their size discrepancy, these grains are irregular in shape and often contain relatively insoluble impuri-ties (e.g., organic matter, iron oxides, clays) that distinguish the wrinkly or irregular concentric internal laminations visually and geochemi-cally. Although some of these features could be overprinted during diagenesis, it is unlikely that traces of a complex original structure, if similarly defi ned by insoluble phases, would be eliminated in all grains during lithifi cation.

Opal-A “microspheres” commonly occur in siliceous sinter deposits (Braunstein and Lowe, 2001; Jones and Renaut, 2007; Lowe and Braun-stein, 2003; Lynne and Campbell, 2004; Lynne et al., 2005, 2008; Rodgers et al., 2004; Smith et al., 2001, 2003; Tobler et al., 2008). These microspheres are generally <5 μm in diam-eter, with mean diameters of <2 μm (Jones and Renaut, 2007; Rodgers et al., 2004). Some microspheres show nanospheric substructure (Rodgers et al., 2004; Smith et al., 2003), indi-cating that they formed via aggregation of nano-meter-scale colloidal particles. Once silica col-loids reach a certain size (~5 nm), aggregation reduces the interfacial free energy more effec-tively than individual particle growth, resulting in the formation of microspheres (Iler, 1979; Smith et al., 2003). Smith et al. (2003, p. 577) suggested that the apparent maximum size limit of microspheres in sinter deposits (~10 μm) is a consequence of a similar principle: “Once silica particles grow to a certain size, there is very little change in energy content with surface area, and there is an optimum upper limit on the size to which particles may grow.”

This consistent upper bound on the diameters of microspheres from modern siliceous sinters is two orders of magnitude less than the aver-age size of the Archean silica granules described here. Although there is some systematic size variation observed in sinter microspheres,



thought to be related to variations in environ-mental conditions (e.g., pH, salinity, tempera-ture, dissolved silica concentration; Rodgers et al., 2004), these variations are still on the order of a few microns. To the extent that the optimum microsphere size appears to be a func-tion of surface free energy and a small range of

environmental factors, silica granules cannot be easily explained as extraordinarily large opal-A microspheres.

In the modern Akinomiya siliceous sinter deposit (Japan), examined and sampled as part of this study, sand-sized subspherical granules of amorphous silica occur along the edges of

some pools (Fig. 8). Some granules clearly formed around nuclei of β-quartz or other local lithic material (Fig. 8B), but many do not have nuclei identifi able in thin section (Figs. 8A and 8C), although this could be a result of random cuts through grain interiors rather than a true absence of nuclei. The cortices of these granules

100 μm

100 μm200 μm

500 μm

200 μm

1 mm

A B

C D

E F

Figure 8. Thin-section images of subspherical, sand-sized amorphous silica granules from the modern Akinomiya siliceous sinter deposit in Japan. The sinter is located at 38°57′32″N, 140°33′26″E, and is 0–80 yr old (Nakanishi et al., 2003). (A) Loosely packed sinter granules. The blue coloration in the pore spaces between the samples is epoxy. (B) Sinter granule with a euhedral volcanic β-quartz grain as its nucleus and thin layers of isopachous silica cement along the grain edges. (C) Somewhat irregularly shaped granule. Silica microspheres of 30–50 μm in diameter can be distinguished along the edges of the granule, but the interior is structureless. (D) Distinguishable silica microspheres within a sinter granule. (E) Geopetal fi ll composed of silt-sized silica microspheres, possibly distinguishable due to extremely thin coatings of impurities. (F) Close-up of layered microsphere fi ll from E.

Stefurak et al.


appear to have formed by aggregation of silica spheres 10–50 μm in diameter; these spheres are often clearly visible along granule edges (Fig. 8C) and, in limited cases, in incompletely cemented granule interiors (Fig. 8D). Silica microspheres also settled out of suspension as small geopetal fi lls (Fig. 8E–8F). Many of the microspheres observed here are larger than the apparent maximum microsphere size (~10 μm) observed in other sinter deposits. This could be due to differences in environmental condi-tions or to an additional intermediate stage of aggregation that produced spherical particles tens of microns in diameter. The Akinomiya sinter deposit is only ~80 yr old (Nakanishi et al., 2003), and most of the granules have little discernible internal structure apart from the nucleus/cortex boundary—they lack con-centric laminations analogous to ooids. Some granules have fi nely layered rims of isopachous silica cements that appear to have precipitated after the granules were deposited (Fig. 8B). The Akinomiya sinter granules could potentially be analogous to Archean silica granules, although whether the modern sinter granules can form without nuclei is still an open question.

Polymerization and aggregation are com-mon processes in silica precipitation at multiple spatial scales, be it the initial polymerization of nanometer-scale colloids, the aggregation of nanometer-scale colloidal particles to form microspheres, or, as observed in the Akinomiya sinter deposit, the aggregation of rather large microspheres to form sinter granules. We there-fore hypothesize that Archean silica granules may have formed via multistage aggregation analogous to processes observed in siliceous sinters, but including an additional stage or stages of aggregation of microspheres to form macroscopic silica granules (Fig. 9).

These additional stages of aggregation are not common in modern siliceous hot springs—most do not contain silica granules like those in the Akinomiya sinter deposit. Several aspects of the depositional conditions make modern sinters imperfect analogues to Archean marine environ-ments. For example, the maximum water depths in siliceous hot springs are shallow, and currents are variable but generally unidirectional, both of which could affect the length of time during which particles of a particular size stay in sus-pension. Many siliceous hot springs also have brackish alkali-chloride waters with very high concentrations of silica (400–500 ppm; Jones and Renaut, 2007; Rodgers et al., 2004); neither of those conditions is a particularly good fi t for Archean oceans, which are thought to have been perhaps supersaline (Knauth, 2005) and only sat-urated to moderately supersaturated with respect to amorphous silica (Siever, 1992). Salinity and

silica concentration are both important param-eters for the silica system and could well explain the differences between modern siliceous sinters and Archean marine environments.

Previous authors have suggested that the con-centration of silica in the Archean ocean was at or above the saturation of amorphous silica (Maliva et al., 2005; Siever, 1992), meaning that silica would have existed in colloidal, rather than monomeric form (Iler, 1979). Proposed models for Archean primary silica precipitation must therefore take the polymerization behavior of colloidal silica into consideration. Polymer-ization is most rapid in the range of pH 4–8 (Fig. 10), and higher concentrations of silica translate to more rapid polymerization (Iler, 1979). Polymerization is inhibited from pH 8 to 10 (Fig. 10) because of the increasing particle-particle repulsion caused by negative surface charges due to dissociation of H+ from surface silanol groups (R-Si-O-H; Iler, 1979). When electrolyte concentrations are low, circum-neutral pH favors gelling, in which colloidal particles link together to form a branched net-work of chains such that the liquid plus colloid system does not change in silica concentration but becomes increasingly viscous (Iler, 1979). However, when cations are present (Na+, Ca2+, etc.) in the pH range 6–10, the negative surface charges on colloidal particles are mitigated, and coagulation is favored, in which colloidal parti-cles clump together into relatively dense aggre-gates (Baxter and Bryant, 1952; Iler, 1979). Although the pH of Archean seawater is poorly constrained, we can reasonably assume that it was not strongly acidic (pH < 4) or alkaline (pH > 10). In addition, the salinity of Archean seawater was almost certainly higher than in the modern alkali-chloride siliceous hot springs dis-cussed here (Knauth, 2005). We can therefore expect that aggregation played a more signifi -cant role in Archean marine silica precipitation than in modern siliceous sinters, which perhaps explains why sand-sized silica granules are not common in these modern systems.

Episodicity of Granule Formation and Deposition

One of the most striking characteristics of early Archean banded cherts is the regular alter-nation of layers of contrasting composition, one component of which is the relatively pure chert layer of silica granules, indicating that deposi-tion and probably formation of silica granules occurred episodically. This repetitive nature of granule deposition and the local consistency in thicknesses of silica granule layers and the inter-vening black or ferruginous chert layers suggest that this cyclic deposition of granules and con-

Si

OH

O

O O

H

HH

SiO

HOO

OH

H H

Si

O

OO

H

H

Silicic acid

monomer

Dimers and

oligomers

Colloidal

nanospheres

Aggregation

of nanospheres

Microspheres

Aggregation

of microspheres

Granules

?

5 nm

5 nm

2 μm

2 μm

200 μm

Figure 9. Schematic outline of the hypoth-esis that silica granules formed via multiple stages of aggregation. Silicic acid monomers (H4SiO4) polymerize to form dimers and oligomers, which eventually grow to form spherical nanospheres with characteristic diameters on the order of 5 nm. Within mod-ern siliceous sinters, colloidal nanospheres aggregate to form silica microspheres with characteristic diameters on the order of 2 μm. We propose that aggregation of micro-spheres could produce spherical granules with characteristic diameters similar to those observed in this study, ~200 μm.



trasting sediments occurred with regular fre-quency. Granule formation may therefore have been driven by periodic fl uctuations in key envi-ronmental parameters occurring over geologi-cally short time scales. Temperature, salinity, pH, and concentration of dissolved silica, the principal parameters infl uencing silica polymer-ization, could all have plausibly varied regularly on seasonal, annual, or multi-annual time scales. A rapid drop in temperature can cause a fl uid to become supersaturated with respect to silica, promoting silica precipitation (Williams and Crerar, 1985). However, periodic fl uctuations in seawater temperature would not have been the most effective trigger for rapid granule forma-tion because temperature changes would have been gradual and small in magnitude, resulting in only small changes in the degree of saturation with respect to amorphous silica.

As discussed already, there are complicated relationships among pH, salinity, and stability of the colloidal silica-water system (Fig. 10). In general, polymerization of silica is most rapid at moderate pH, and aggregation is favored over gelling in the presence of cations, which would have been abundant in Archean sea-water (Knauth, 2005). Unless pH was fl uctuat-ing between circumneutral and either strongly acidic (pH < 4) or alkaline (pH > 10) values, pH

variations are unlikely to have triggered granule formation. At slightly alkaline pH (8–10), how-ever, increases in salinity can strongly infl uence polymerization rates (Fig. 10). Finally, the rate of silica precipitation is proportional to the con-centration of dissolved silica, so higher silica concentrations produce more rapid precipitation (Rimstidt and Barnes, 1980) and polymerization rates (Iler, 1979). Salinity and silica concentra-tion can therefore both affect polymerization rates and could have fl uctuated over seasonal, annual, and/or multi-annual time scales via vari-ations in evaporative concentration and/or fl uc-tuations in inputs (cf. Bekker et al., 2014). We hypothesize that these two parameters are more likely to have triggered silica granule formation than periodic changes in temperature or pH.

If the production and deposition of silica granules were seasonal or annual, an impor-tant corollary is that the depositional rates of banded chert must have been quite rapid—sev-eral centimeters per year as refl ected by both the background sedimentation (black chert bands or ferruginous chert bands) and the sil-ica granule event beds (white chert bands). We could expect to observe a characteristic time scale required to return silica concentration to its critical level after being diminished by the previous granule-forming episode. This time

scale would depend on the mixing time of shal-low granule-forming waters with deeper silica-enriched waters and variations of the other critical environmental variables (e.g., seasonal/annual variations in salinity). Granule layers in the deepest-water lithofacies examined here, the lower Mapepe Formation banded iron forma-tion, are thinner, less regular, and often farther apart stratigraphically than those observed in the shelfal Buck Reef Chert. As we have argued herein, this seems to suggest that granules either formed exclusively over shelf environments or that granules formed in the water column over basinal environments did not make it to the seafl oor as effi ciently. Either way, it seems that not all granule formation episodes recorded in shallower-water environments had equivalents in deeper-water facies. The depositional rates in the deeper-water facies, banded ferruginous chert and banded iron formation, could there-fore still be relatively slower than shallower-water facies (cf. Tice and Lowe, 2004).

Effects of Postdepositional Alteration

The observations presented here suggest that silica granules have been affected physically and chemically by their extended histories of dia-genesis and metamorphism. Some of the conse-quences of this postdepositional alteration relate to the appearance of the silica granules, most notably including widespread compaction. A more subtle but signifi cant change is the oblitera-tion of any primary textural features with smaller length scales than the average size of microquartz domains, ~30 μm. This makes it impossible to directly test our granule formation hypothesis using traditional transmitted or refl ected light petrography of Archean silica granules: Micro-scopic substructures like those observed in the modern granules have been destroyed.

Similarly, the data presented here suggest that the Si isotopic compositions of some silica granules have likely been altered during dia-genesis. One likely explanation for the distinct Si isotopic compositions observed in silica granules from different lithofacies is that they resulted from the incorporation of contrasting Si isotopic signatures from local pore fl uids, perhaps infl uenced by the mineralogical and Si isotopic composition of adjacent sediments. In theory, Si isotopic compositions recorded by primary silica precipitates should provide some key insights into any spatial variations in the Si isotopic composition of seawater. At this point, however, we cannot control for the magnitudes and mechanisms of any changes in Si isotopic composition; in other words, the initial Si iso-topic compositions of the granules analyzed here are unknown. Notably, if this interpreta-

pHn

o s

alt

0.1 M NaCl

0.2 M NaCl

silica solstability

4 6 8 10 12

gelling stable sols

increasing negative surface charge

SiO2

dissolves

coagulation brackish/saline

fresh water

polymerizationmode

Figure 10. Schematic diagram illustrating the effects of pH and sa-linity on polymerization of colloidal silica. The y axis represents sta-bility of the silica sol (the colloidal silica-water system), where higher values correspond to silica being more stable in colloidal form, while lower values correspond to systems favoring polymeri zation, resulting in coagulation of larger aggregate particles or gelling of branched networks. In the absence of electrolytes, particle repulsion due to negative particle surface inhibits polymerization above pH 7, and the colloidal system is relatively stable. Added cations , however, mitigate surface charges over a greater range in pH and promote coagulation over gelling. For comparison, the concentration of Na+ in the modern ocean is 0.47 M (DOE, 1994). Figure is modifi ed after Iler (1979) and Williams and Crerar (1985).

Stefurak et al.


tion is correct, this behavior would contrast with observations that Si isotope compositions of Archean cherts were largely rock-buffered and relatively resistant to resetting during diagen-esis and metamorphism (Marin-Carbonne et al., 2012; Stefurak et al., 2015), perhaps signaling differences in the timing and/or mechanisms of silica diagenesis experienced by silica granules compared to silica cements and other early dia-genetic chert-precursor silica phases. The most productive future research directions may there-fore be experimental: evaluating the silica gran-ule formation hypothesis presented here and documenting the physical and chemical changes associated with diagenesis and metamorphism.

CONCLUSIONS

Silica granules were a common grain type in early Archean cherty sediments and appear to have represented a signifi cant mode of silica deposition in the Archean silica cycle. Their sand size, occurrence in non-ferruginous set-tings, nearly pure SiO2 composition, and uni-formly subspherical, well-rounded shapes dif-ferentiate silica granules from other associated detrital grain types or from granular iron forma-tion. The sharp contacts between pure granule layers and the adjacent detrital organic layers in banded black-and-white chert suggest that the two types of layers were formed from distinct depositional events and that production of silica granules was episodic. We suggest that the gran-ules were formed via multistage aggregation of silica, in a manner somewhat analogous to silica precipitation observed in modern siliceous sinter deposits. In systems with near-neutral to moderately alkaline pH and high concentrations of dissolved silica (saturated to supersaturated with respect to amorphous silica), increased salinity raises the rate of silica polymerization and favors aggregation of silica particles over gelling. Variations in silica concentration and salinity could therefore have triggered episodic production of silica granules.

ACKNOWLEDGMENTS

We thank the following for their assistance: Y. Guan, secondary ion mass spectrometry; E. Gottlieb and M. Coble, sample preparation, scanning electron micros-copy, and cathodoluminescence; C. Ma, electron probe (Caltech); and R. Jones, electron probe (Stanford). Stefurak was supported by a National Science Foun-dation Graduate Fellowship. Fischer was supported by National Aeronautics and Space Administration (NASA) Exobiology program (NNX09AM91G) and the David and Lucile Packard Foundation. This study was carried out using funds provided by the School of Earth Sciences, Stanford University, to Lowe. We are grateful to Sappi Forest Products, the Mpumalanga Parks Board (J. Eksteen and L. Loocks), and Taurus Estates (C. Wille) for access to private properties. We thank Nic Beukes, Andrey Bekker, and Assoc. Editor Henning Dypvik for their helpful reviews.

REFERENCES CITED

Abraham, K., Hofmann, A., Foley, S.F., Cardinal, D., Harris, C., Barth, M.G., and Andre, L., 2011, Coupled silicon-oxygen isotope fractionation traces Archaean silici-fi cation: Earth and Planetary Science Letters, v. 301, p. 222–230, doi: 10 .1016 /j .epsl .2010 .11 .002 .

Andre, L., Cardinal, D., Alleman, L.Y., and Moorbath, S., 2006, Silicon isotopes in ~3.8 Ga West Greenland rocks as clues to the Eoarchean supracrustal Si cycle: Earth and Planetary Science Letters, v. 245, p. 162–173, doi: 10 .1016 /j .epsl .2006 .02 .046 .

Basile-Doelsch, I., Meunier, J.D., and Parron, C., 2005, Another continental pool in the terrestrial silicon cycle: Nature, v. 433, p. 399–402, doi: 10 .1038 /nature03217 .

Baxter, S., and Bryant, K.C., 1952, 577. Silica sols. Part II. Conditions of stability: Journal of the Chemical Soci-ety, p. 3021–3024, doi: 10 .1039 /jr9520003021 .

Bekker, A., Slack, J.F., Planavsky, N., Krapez, B., Hofmann, A., Konhauser, K.O., and Rouxel, O.J., 2010, Iron formation: The sedimentary product of a complex in-terplay among mantle, tectonic, ocean, and biospheric processes: Economic Geology and the Bulletin of the Society of Economic Geologists, v. 105, p. 467–508, doi: 10 .2113 /gsecongeo .105 .3 .467 .

Bekker, A., Planavsky, N.J., Krapež, B., Rasmussen, B., Hofmann, A., Slack, J.F., Rouxel, O.J., and Konhauser, K.O., 2014, Iron formations: Their origins and impli-cations for ancient seawater chemistry, in Turekian, H.D.H.K., ed., Treatise on Geochemistry (2nd ed.): Oxford, Elsevier, p. 561–628.

Beukes, N.J., and Klein, C., 1990, Geochemistry and sedi-mentology of a facies transition—From microbanded to granular iron-formation—in the early Proterozoic Transvaal Supergroup, South Africa: Precambrian Research, v. 47, p. 99–139, doi: 10 .1016 /0301 -9268 (90)90033 -M .

Boggs, S., Jr., and Krinsley, D., 2006, Application of Cath-odoluminescence Imaging to the Study of Sedimentary Rocks: Cambridge, UK, Cambridge University Press, 176 p.

Boggs, S., Jr., Kwon, Y.-I., Goles, G.G., Rusk, B., Krinsley, D., and Seyedolali, A., 2002, Is quartz cathodolumines-cence color a reliable provenance tool? A quantitative examination: Journal of Sedimentary Research, v. 72, no. 3, p. 408–415, doi: 10 .1306 /102501720408 .

Braunstein, D., and Lowe, D.R., 2001, Relationship between spring and geyser activity and the deposition and mor-phology of high temperature (>73°C) siliceous sinter, Yellowstone National Park, Wyoming, U.S.A.: Journal of Sedimentary Research, v. 71, no. 5, p. 747–763, doi: 10 .1306 /2DC40965 -0E47 -11D7 -8643000102C1865D .

Byerly, G.R., 1999, Komatiites of the Mendon Formation: Late-stage ultramafi c volcanism in the Barberton greenstone belt, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 189–211.

Byerly, G.R., Kroner, A., Lowe, D.R., Todt, W., and Walsh, M.M., 1996, Prolonged magmatism and time con-straints for sediment deposition in the early Archean Barberton greenstone belt: Evidence from the Up-per Onverwacht and Fig Tree Groups: Precambrian Research, v. 78, p. 125–138, doi: 10 .1016 /0301 -9268 (95)00073 -9 .

Byerly, G.R., Lowe, D.R., Wooden, J.L., and Xie, X., 2002, An Archean impact layer from the Pilbara and Kaapvaal cratons: Science, v. 297, no. 5585, p. 1325–1327, doi: 10 .1126 /science .1073934 .

Cawood, P.A., Hawkesworth, C.J., and Dhuime, B., 2012, The continental record and the generation of conti-nental crust: Geological Society of America Bulletin, v. 125, no. 1–2, p. 14–32.

Chakrabarti, R., Knoll, A.H., Jacobsen, S.B., and Fischer, W.W., 2012, Si isotope variability in Proterozoic cherts: Geochimica et Cosmochimica Acta, v. 91, p. 187–201, doi: 10 .1016 /j .gca .2012 .05 .025 .

Condie, K.C., Macke, J.E., and Reimer, T.O., 1970, Petrol-ogy and geochemistry of early Precambrian graywackes from the Fig Tree Group, South Africa: Geological So-ciety of America Bulletin, v. 81, p. 2759–2776, doi: 10 .1130 /0016 -7606 (1970)81 [2759: PAGOEP]2 .0 .CO;2 .

Decker, N.B., 2013, High Resolution Tephra and U/Pb Chronol ogy of the 3.33–3.26 Ga Mendon Formation, Barberton Greenstone Belt, South Africa [M.S. thesis]: Baton Rouge, Louisiana, Louisiana State University, 70 p.

De La Rocha, C.L., Brzezinski, M.A., and DeNiro, M.J., 2000, A fi rst look at the distribution of the stable iso-topes of silicon in natural waters: Geochimica et Cos-mochimica Acta, v. 64, no. 14, p. 2467–2477, doi: 10 .1016 /S0016 -7037 (00)00373 -2 .

Delstanche, S., Opfergelt, S., Cardinal, D., Elsass, F., Andre, L., and Delvaux, B., 2009, Silicon isotopic fractionation during adsorption of aqueous monosilicic acid onto iron oxide: Geochimica et Cosmochimica Acta, v. 73, no. 4, p. 923–934, doi: 10 .1016 /j .gca .2008 .11 .014 .

Delvigne, C., Cardinal, D., Hofmann, A., and Andre, L., 2012, Stratigraphic changes of Ge/Si, REE + Y and sil-icon isotopes as insights into the deposition of a Meso-archaean banded iron formation: Earth and Planetary Science Letters, v. 355–356, p. 109–118, doi: 10 .1016 /j .epsl .2012 .07 .035 .

Department of Energy (DOE), 1994, Physical and thermo-dynamic data, in Dickson, A.G., and Goyet, C., eds., Handbook of Methods for the Analysis of the Vari-ous Parameters of the Carbon Dioxide System in Sea Water, Version 2: Oak Ridge, Tennessee, Oak Ridge National Laboratory, Carbon Dioxide Information Analysis Center, 22 p.

Dhuime, B., Hawkesworth, C.J., Cawood, P.A., and Storey, C.D., 2012, A change in the geodynamics of conti-nental growth 3 billion years ago: Science, v. 335, no. 6074, p. 1334–1336, doi: 10 .1126 /science .1216066 .

DiMarco, M.J., and Lowe, D.R., 1989, Petrography and provenance of silicifi ed Early Archaean volcaniclastic sandstones, eastern Pilbara block, Western Australia: Sedimentology, v. 36, p. 821–836, doi: 10 .1111 /j .1365 -3091 .1989 .tb01748 .x .

Dimroth, E., and Chauvel, J.-J., 1973, Petrography of the Sokoman Iron Formation in part of the Central Lab-rador Trough, Quebec, Canada: Geological Society of America Bulletin, v. 84, no. 1, p. 111–134, doi: 10 .1130 /0016 -7606 (1973)84 <111: POTSIF>2 .0 .CO;2 .

Ding, T., Jiang, S., Wan, D., Li, Y., Li, J., Song, H., Liu, Z., and Yao, Z., 1996, Silicon Isotope Geochemistry: Bei-jing, Geological Publishing House, 125 p.

Douthitt, C.B., 1982, The geochemistry of the stable isotopes of silicon: Geochimica et Cosmochimica Acta, v. 46, p. 1449–1458, doi: 10 .1016 /0016 -7037 (82)90278 -2 .

Fischer, W.W., and Knoll, A.H., 2009, An iron shuttle for deepwater silica in Late Archean and early Paleo protero-zoic iron formation: Geological Society of America Bulletin, v. 121, p. 222–235.

Geilert, S., Vroon, P.Z., Roerdink, D.L., Van Cappellen, P., and Van Bergen, M.J., 2014, Silicon isotope fraction-ation during abiotic silica precipitation at low tem-peratures: Inferences from fl ow-through experiments: Geochimica et Cosmochimica Acta, v. 142, p. 95–114, doi: 10 .1016 /j .gca .2014 .07 .003 .

Georg, R.B., Halliday, A.N., Schauble, E.A., and Reynolds, B.C., 2007, Silicon in the Earth’s core: Nature, v. 447, p. 1102–1106, doi: 10 .1038 /nature05927 .

Georg, R.B., Zhu, C., Reynolds, B.C., and Halliday, A.N., 2009, Stable silicon isotopes of groundwater, feldspars, and clay coatings in the Navajo Sandstone aquifer, Black Mesa, Arizona, USA: Geochimica et Cosmo-chimica Acta, v. 73, p. 2229–2241, doi: 10 .1016 /j .gca .2009 .02 .005 .

Glikson, A.Y., Allen, C., and Vickers, J., 2004, Multiple 3.47-Ga-old asteroid impact fallout units, Pilbara cra-ton, Western Australia: Earth and Planetary Science Letters, v. 7040, p. 1–14.

Goode, A.D.T., Hall, W.D.M., and Bunting, J.A., 1983, The Nabberu Basin of Western Australia, in Trendall, A.F., and Morris, R.C., eds., Iron-Formation: Facts and Prob-lems: Amsterdam, Netherlands, Elsevier, p. 295–323.

Götze, J., Plötze, M., and Habermann, D., 2001, Origin, spectral characteristics and practical applications of the cathodoluminescence (CL) of quartz—A review: Mineralogy and Petrology, v. 71, p. 225–250, doi: 10 .1007 /s007100170040 .

Gross, G.A., 1972, Primary features in cherty iron-forma-tions: Sedimentary Geology, v. 7, no. 4, p. 241–261, doi: 10 .1016 /0037 -0738 (72)90024 -3 .



Guidry, S.A., and Chafetz, H.S., 2003, Anatomy of siliceous hot springs: Examples from Yellowstone National Park, Wyoming, USA: Sedimentary Geology, v. 157, no. 1–2, p. 71–106, doi: 10 .1016 /S0037 -0738 (02)00195 -1 .

Heck, P.R., Huberty, J.M., Kita, N.T., Ushikubo, T., Kozdon, R., and Valley, J.W., 2011, SIMS analyses of silicon and oxygen isotope ratios for quartz from Archean and Paleoproterozoic banded iron formations: Geochimica et Cosmochimica Acta, v. 75, no. 20, p. 5879–5891, doi: 10 .1016 /j .gca .2011 .07 .023 .

Hickman, A., and Van Kranendonk, M. J., 2008, Marble Bar Geological Sheet 2855: Geological Survey of Western Australia, scale 1:100,000.

Iler, R.K., 1979, The Chemistry of Silica: New York, John Wiley & Sons, 866 p.

Jones, B., and Renaut, R.W., 1997, Formation of silica oncoids around geysers and hot springs at El Tatio, northern Chile: Sedimentology, v. 44, p. 287–304.

Jones, B., and Renaut, R.W., 2007, Microstructural changes accompanying the opal-A to opal-CT transition: New evidence from the siliceous sinters of Geysir, Haukada-lur, Iceland: Sedimentology, v. 54, p. 921–948, doi: 10 .1111 /j .1365 -3091 .2007 .00866 .x .

Jones, B., Renaut, R.W., and Rosen, M.R., 1999, Actively growing siliceous oncoids in the Waiotapu geothermal area, North Island, New Zealand: Journal of the Geo-logical Society of London, v. 156, no. 1, p. 89–103, doi: 10 .1144 /gsjgs .156 .1 .0089 .

Jones, B., Rosen, M.R., and Renaut, R.W., 2001, “Geyser Eggs” from Te Whakarewarewatangaoteopetauaawa-hiao, North Island, New Zealand: Journal of Sedimen-tary Research, v. 71, no. 1, p. 190–204, doi: 10 .1306 /060100710190 .

Klein, C., 2005, Some Precambrian banded iron-formations (BIFs) from around the world: Their age, geologic setting, mineralogy, metamorphism, geochemistry, and origin: The American Mineralogist, v. 90, no. 10, p. 1473–1499, doi: 10 .2138 /am .2005 .1871 .

Knauth, L.P., 2005, Temperature and salinity history of the Precambrian ocean: Implications for the course of mi-crobial evolution: Palaeogeography, Palaeoclimatol-ogy, Palaeoecology, v. 219, p. 53–69, doi: 10 .1016 /j .palaeo .2004 .10 .014 .

Knauth, L.P., and Lowe, D.R., 1978, Oxygen isotope geo-chemistry of cherts from the Onverwacht Group (3.4 billion years), Transvaal, South Africa, with implica-tions for secular variations in the isotopic composition of cherts: Earth and Planetary Science Letters, v. 41, p. 209–222, doi: 10 .1016 /0012 -821X (78)90011 -0 .

Knauth, L.P., and Lowe, D.R., 2003, High Archean climatic temperature inferred from oxygen isotope geochem-istry of cherts in the 3.5 Ga Swaziland Supergroup, South Africa: Geological Society of America Bulle-tin, v. 115, no. 5, p. 566–580, doi: 10 .1130 /0016 -7606 (2003)115 <0566: HACTIF>2 .0 .CO;2 .

Köhler, I., Konhauser, K.O., Papineau, D., Bekker, A., and Kappler, A., 2013, Biological carbon precursor to dia-genetic siderite with spherical structures in iron forma-tions: Nature Communications, v. 4, p. 1741, doi: 10 .1038 /ncomms2770 .

Krüner, A., Byerly, G.R., and Lowe, D.R., 1991, Chronol-ogy of Early Archaean granite-greenstone evolution in the Barberton Mountain Land, South Africa, based on precise dating by single zircon evaporation: Earth and Planetary Science Letters, v. 103, p. 41–54, doi: 10 .1016 /0012 -821X (91)90148 -B .

Lowe, D.R., 1999a, Petrology and sedimentology of cherts and related silicifi ed sedimentary rocks in the Swa-ziland Supergroup, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 83–114.

Lowe, D.R., 1999b, Shallow-water sedimentation of ac-cretionary lapilli–bearing strata of the Msauli Chert: Evidence of explosive hydromagmatic komatiitic volcanism, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 213–232.

Lowe, D.R., and Braunstein, D., 2003, Microstructure of high-temperature (>73°C) siliceous sinter deposited around hot springs and geysers, Yellowstone National

Park: The role of biological and abiological processes in sedimentation: Canadian Journal of Earth Sciences, v. 40, no. 11, p. 1611–1642, doi: 10 .1139 /e03 -066 .

Lowe, D.R., and Byerly, G.R., 1986, Early Archean silicate spherules of probable impact origin, South Africa and Western Australia: Geology, v. 14, no. 1, p. 83–86, doi: 10 .1130 /0091 -7613 (1986)14 <83: EASSOP>2 .0 .CO;2 .

Lowe, D.R., and Byerly, G.R., 1999, Stratigraphy of the west-central part of the Barberton greenstone belt, South Africa, in Lowe, D.R., and Byerly, G.R., eds., Geological Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 1–36.

Lowe, D.R., and Fisher Worrell, G., 1999, Sedimentology, mineralogy, and implications of silicifi ed evaporites in the Kromberg Formation, Barberton greenstone belt, South Africa, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 167–188.

Lowe, D.R., and Nocita, B.W., 1999, Foreland basin sedi-mentation in the Mapepe Formation, southern-facies Fig Tree Group, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 233–258.

Lynne, B.Y., and Campbell, K.A., 2004, Morphologic and mineralogic transitions from opal-A to opal-CT in low-temperature siliceous sinter diagenesis, Taupo volcanic zone, New Zealand: Journal of Sedimentary Research, v. 74, no. 4, p. 561–579, doi: 10 .1306 /011704740561 .

Lynne, B.Y., Campbell, K.A., Moore, J.N., and Browne, P.R.L., 2005, Diagenesis of 1900-year-old siliceous sinter (opal-A to quartz) at Opal Mound, Roosevelt Hot Springs, Utah, U.S.A.: Sedimentary Geology, v. 179, no. 3–4, p. 249–278, doi: 10 .1016 /j .sedgeo .2005 .05 .012 .

Lynne, B.Y., Campbell, K.A., Moore, J., and Browne, P.R.L., 2008, Origin and evolution of the Steamboat Springs siliceous sinter deposit, Nevada, U.S.A.: Sedi-mentary Geology, v. 210, no. 3–4, p. 111–131, doi: 10 .1016 /j .sedgeo .2008 .07 .006 .

Maliva, R.G., 2001, Silicifi cation in the Belt Supergroup (Mesoproterozoic), Glacier National Park, Montana, USA: Sedimentology, v. 48, p. 887–896, doi: 10 .1046 /j .1365 -3091 .2001 .00399 .x .

Maliva, R.G., Knoll, A.H., and Siever, R., 1989, Secular change in chert distribution: A refl ection of evolving biological participation in the silica cycle: Palaios, v. 4, no. 6, p. 519–532, doi: 10 .2307 /3514743 .

Maliva, R.G., Knoll, A.H., and Simonson, B.M., 2005, Sec-ular change in the Precambrian silica cycle: Insights from chert petrology: Geological Society of America Bulletin, v. 117, no. 7–8, p. 835–845, doi: 10 .1130 /B25555 .1 .

Marin-Carbonne, J., Chaussidon, M., and Robert, F., 2012, Micrometer-scale chemical and isotopic criteria (O and Si) on the origin and history of Precambrian cherts: Im-plications for paleo-temperature reconstructions: Geo-chimica et Cosmochimica Acta, v. 92, p. 129–147, doi: 10 .1016 /j .gca .2012 .05 .040 .

Nakanishi, T., Izawa, E., and Watanabe, K., 2003, Miner-alogical and geochemical characteristics of siliceous sinter in Japan, in Proceedings of the 25th New Zea-land Geothermal Workshop, University of Aukland, p. 155–160.

Opfergelt, S., Cardinal, D., Andre, L., Delvigne, C., Bremond , L., and Delvaux, B., 2010, Variations of δ30Si and Ge/Si with weathering and biogenic input in tropi-cal basaltic ash soils under monoculture: Geochimica et Cosmochimica Acta, v. 74, no. 1, p. 225–240, doi: 10 .1016 /j .gca .2009 .09 .025 .

Opfergelt, S., Georg, R.B., Delvaux, B., Cabidoche, Y.-M., Burton, K.W., and Halliday, A.N., 2012, Silicon iso-topes and the tracing of desilication in volcanic soil weathering sequences, Guadeloupe: Chemical Geol-ogy, v. 326–327, p. 113–122, doi: 10 .1016 /j .chemgeo .2012 .07 .032 .

Rasmussen, B., Meier, D.B., Krapez, B., and Muhling, J.R., 2013, Iron silicate microgranules as precursor sedi-ments to 2.5-billion-year-old banded iron formations: Geology, v. 41, p. 435–438, doi: 10 .1130 /G33828 .1 .

Renaut, R.W., Jones, B., and Rosen, M.R., 1996, Primary silica oncoids from Orakeikorako Hot Springs, North Island, New Zealand: Palaios, v. 11, no. 5, p. 446–458, doi: 10 .2307 /3515212 .

Rimstidt, J.D., and Barnes, H.L., 1980, The kinetics of silica-water reactions: Geochimica et Cosmochimica Acta, v. 44, no. 11, p. 1683–1699, doi: 10 .1016 /0016 -7037 (80)90220 -3 .

Rodgers, K.A., Browne, P.R.L., Buddle, T.F., Cook, K.L., Greatrex, R.A., Hampton, W.A., Herdianita, N.R., Holland, G.R., Lynne, B.Y., Martin, R., Newton, Z., Pastars, D., Sannazarro, K.L., and Teece, C.I.A., 2004, Silica phases in sinters and residues from geothermal fi elds of New Zealand: Earth-Science Reviews, v. 66, p. 1–61, doi: 10 .1016 /j .earscirev .2003 .10 .001 .

Schäfer, A., and Teyssen, T., 1987, Size, shape and orienta-tion of grains in sands and sandstones—Image analysis applied to rock thin-sections: Sedimentary Geology, v. 52, p. 251–271, doi: 10 .1016 /0037 -0738 (87)90064 -9 .

Siever, R., 1957, The silica budget in the sedimentary cycle: The American Mineralogist, v. 42, p. 821–841.

Siever, R., 1992, The silica cycle in the Precambrian: Geo-chimica et Cosmochimica Acta, v. 56, p. 3265–3272, doi: 10 .1016 /0016 -7037 (92)90303 -Z .

Simonson, B.M., 1985, Sedimentological constraints on the origins of Precambrian iron-formations: Geological Society of America Bulletin, v. 96, no. 2, p. 244–252, doi: 10 .1130 /0016 -7606 (1985)96 <244: SCOTOO>2 .0 .CO;2 .

Simonson, B.M., 1987, Early silica cementation and subse-quent diagenesis in arenites from four early Protero-zoic iron formations of North America: Journal of Sedimentary Research, v. 57, no. 3, p. 494–511.

Simonson, B.M., 2003, Origin and evolution of large Pre-cambrian iron formations, in Chan, M.A., and Archer, A.W., eds., Extreme Depositional Environments: Mega End Members in Geologic Time: Geological Society of America Special Paper 370, p. 231–244.

Smith, B.Y., Rodgers, K.A., and Browne, P.R.L., 2001, Opal-A—The fi rst 100 days, in Proceedings of the 23rd New Zealand Geothermal Workshop: Auckland, New Zealand, Geothermal Institute, University of Auckland, p. 131–136.

Smith, B.Y., Turner, S.J., and Rodgers, K.A., 2003, Opal-A and associated microbes from Wairakei, New Zealand: The fi rst 300 days: Mineralogical Magazine, v. 67, no. 3, p. 563–579, doi: 10 .1180 /0026461036730118 .

Stefurak, E.J.T., Lowe, D.R., Zentner, D., and Fischer, W.W., 2014, Primary silica granules—A new mode of early Archean sedimentation: Geology, v. 42, no. 4, p. 283–286, doi: 10 .1130 /G35187 .1 .

Stefurak, E.J.T., Fischer, W.W., and Lowe, D.R., 2015, Tex-ture-specifi c Si isotope variations in Barberton green-stone belt cherts record low temperature fractionations in early Archean seawater: Geochimica et Cosmo-chimica Acta, v. 150, p. 26–52.

Steinhoefel, G., Horn, I., and von Blanckenburg, F., 2009, Micro-scale tracing of Fe and Si isotope signatures in banded iron formation using femtosecond laser abla-tion: Geochimica et Cosmochimica Acta, v. 73, no. 18, p. 5343–5360, doi: 10 .1016 /j .gca .2009 .05 .037 .

Steinhoefel, G., von Blanckenburg, F., Horn, I., Konhauser, K.O., Beukes, N.J., and Gutzmer, J., 2010, Decipher-ing formation processes of banded iron formations from the Transvaal and the Hamersley successions by combined Si and Fe isotope analysis using UV femto-second laser ablation: Geochimica et Cosmochimica Acta, v. 74, p. 2677–2696, doi: 10 .1016 /j .gca .2010 .01 .028 .

Thompson Stiegler, M., Lowe, D.R., and Byerly, G.R., 2010, The petrogenesis of volcaniclastic komatiites in the Barberton greenstone belt, South Africa: A textural and geochemical study: Journal of Petrology, v. 51, p. 947–972, doi: 10 .1093 /petrology /egq008 .

Tice, M.M., and Lowe, D.R., 2004, Photosynthetic micro-bial mats in the 3,416-Myr-old ocean: Nature, v. 431, p. 549–552, doi: 10 .1038 /nature02888 .

Tice, M.M., and Lowe, D.R., 2006, The origin of carbona-ceous matter in pre–3.0 Ga greenstone terrains: A re-view and new evidence from the 3.42 Ga Buck Reef Chert: Earth-Science Reviews, v. 76, p. 259–300, doi: 10 .1016 /j .earscirev .2006 .03 .003 .

Stefurak et al.


Tobler, D.J., Stefansson, A., and Benning, L.G., 2008, In-situ grown silica sinters in Icelandic geothermal areas: Geobiology, v. 6, p. 481–502, doi: 10 .1111 /j .1472 -4669 .2008 .00179 .x .

Treguer, P., Nelson, D.M., Van Bennekom, A.J., DeMaster, D.J., Leynaert, A., and Queguiner, B., 1995, The silica balance in the world ocean: A reestimate: Science, v. 268, p. 375–379, doi: 10 .1126 /science .268 .5209 .375 .

Trendall, A.F., 2002, The signifi cance of iron-formation in the Precambrian stratigraphic record, in Altermann, W., and Corcoran, P.L., eds., Precambrian Sedimentary Environments: A Modern Approach to Ancient Depo-sitional Systems: International Association of Sedi-mentologists Special Publication 33, p. 33–66.

van den Boorn, S.H.J.M., van Bergen, M.J., Nijman, W., and Vroon, P.Z., 2007, Dual role of seawater and hydro-thermal fl uids in early Archean chert formation: Evi-dence from silicon isotopes: Geology, v. 35, no. 10, p. 939–942, doi: 10 .1130 /G24096A .1 .

van den Boorn, S.H.J.M., van Bergen, M.J., Vroon, P.Z., de Vries, S.T., and Nijman, W., 2010, Silicon isotopic and trace element constraints on the origin of ~3.5 Ga cherts: Implications for Early Archaean marine envi-

ronments: Geochimica et Cosmochimica Acta, v. 74, p. 1077–1103, doi: 10 .1016 /j .gca .2009 .09 .009 .

Van Kranendonk, M.J., 2011, Pilbara craton, in Gargaud, M., Amils, R., Quintanilla, J.C., Cleaves, H.J., II, Irvine, W.M., Pinti, D.L., and Viso, M., eds., Encyclopedia of Astrobiology: Berlin, Springer-Verlag, p. 1257–1260.

Van Kranendonk, M.J., Hickman, A.H., Smithies, R.H., Nel-son, D.M., and Pike, G., 2002, Geology and tectonic evolution of the Archean North Pilbara terrain, Pilbara craton, Western Australia: Economic Geology and the Bulletin of the Society of Economic Geologists, v. 97, p. 695–732.

Walsh, M.M., and Lowe, D.R., 1999, Modes of accumula-tion of carbonaceous matter in the early Archean: A petrographic and geochemical study of the carbona-ceous cherts of the Swaziland Supergroup, in Lowe, D.R., and Byerly, G.R., eds., Geologic Evolution of the Barberton Greenstone Belt, South Africa: Geological Society of America Special Paper 329, p. 115–132.

Walter, M.R., 1976, Geyserites of Yellowstone National Park: An example of abiogenic “stromatolites,” in Walter , M.R., ed., Stromatolites: Amsterdam, Nether-lands, Elsevier, p. 87–112.

Walter, M.R., Desmarais, D., Farmer, J.T., and Hinman, N.W., 1996, Lithofacies and biofacies of mid-Paleo-zoic thermal spring deposits in the Drummond Basin, Queensland, Australia: Palaios, v. 11, no. 6, p. 497–518, doi: 10 .2307 /3515187 .

Williams, L.A., and Crerar, D.A., 1985, Silica diagenesis: II. General mechanisms: Journal of Sedimentary Petrol-ogy, v. 55, no. 3, p. 312–321.

Williams, L.A., Parks, G.A., and Crerar, D.A., 1985, Silica diagenesis: I. Solubility controls: Journal of Sedimen-tary Petrology, v. 55, no. 3, p. 301–311.

Ziegler, K., Chadwick, O.A., White, A.F., and Brzezinski, M.A., 2005, δ30Si systematics in a granitic saprolite, Puerto Rico: Geology, v. 33, no. 10, p. 817–820, doi: 10 .1130 /G21707 .1 .

SCIENCE EDITOR: CHRISTIAN KOEBERL

ASSOCIATE EDITOR: HENNING DYPVIK

MANUSCRIPT RECEIVED 1 AUGUST 2014REVISED MANUSCRIPT RECEIVED 29 DECEMBER 2014MANUSCRIPT ACCEPTED 5 FEBRUARY 2015

Printed in the USA

sedimentology and geochemistry of archean silica...

Documents