Anoxic Continental Surface Weathering Recorded By The 2.95 Ga Denny .

A.W. Heard et al. / Geochimica et Cosmochimica Acta 295 (2021) 1–233Fig. 1. Geological map showing the distribution of the Nsuze and Mozaan groups of the Pongola Supergroup (Hofmann et al., 2019).Borehole locations of TSB07-26 (this study) and PMH-24 (Crowe et al., 2013) are shown.is a comparatively undeformed volcano-sedimentary succession that was deposited on early Archean basementgranitoids and gneisses from 2.985 Ga (Hegner et al.,1994; Hicks and Hofmann, 2012; Mukasa et al., 2013). Itoutcrops in a 250 km, roughly north-south trending linear belt that crosses the northern KwaZulu-Natal andMpumalanga provinces, and southern Swaziland. The Pongola Supergroup is divided into the underlying NsuzeGroup and the overlying Mozaan Group (Figs. 1 and 2).The Nsuze Group is a compositionally diverse volcanic succession dominated by basaltic andesites dated by U-Pb zircon SIMS method between 2.980 Ga at its base and 2.954Ga at its top (Mukasa et al., 2013). The Mozaan Group is asedimentary succession of conglomerate, shales, and locallydeveloped iron formation (IF), deposited in a marine setting (Beukes and Cairncross, 1991).The Denny Dalton paleosol sampled in this study comesfrom the White Mfolozi Inlier and occurs at the very top ofthe Nsuze Group (Fig. 2), developed on a basaltic andesitelava flow that was tilted 10 prior to paleoweathering, sothat the paleosol lies above an angular unconformity (Hicksand Hofmann, 2012). The basaltic andesite protolith onwhich the Denny Dalton paleosol was developed is anamygdaloidal lava flow with a fine-grained groundmass ofamphibole, chlorite and albite, with chlorite, in additionto quartz and calcite, concentrically filling the amygdales0mDrill core TSB07-26 LogDoleriteIron formation50ShaleSandstonePebbly sandstoneBasaltic andesite100150200Nsuze/Mozaan unconformityPaleosol samplesFig. 2. Stratigraphic log of the TSB07-26 drill core, modified fromDelvigne et al. (2016). The hashed area marks the location of thepaleosol at the top of the basaltic andesite unit.(Nhleko, 2003). The present-day mineral assemblage ofthe Nsuze Group basaltic andesites corresponds to greenschist facies metamorphic conditions (Crow et al., 1989;Nhleko, 2003). Rutile is directly observed in the basalticandesites of the White Mfolozi Inlier, and the presence of

4A.W. Heard et al. / Geochimica et Cosmochimica Acta 295 (2021) 1–23zircon and monazite in the overlying paleosols (Nhleko,2003) indicates that these accessory minerals were also present in the protolith. Zircon grains are nevertheless smalland rarely occur in the Nsuze basaltic andesites (Mukasaet al., 2013).Sample powders are aliquots of the same samples thatwere analyzed for major elemental concentrations (Fig. 2,Table 1), Si isotope composition, and Ge/Si ratios byDelvigne et al. (2016), and also for Cr isotope values byColwyn et al. (2019), who found no significant Cr isotopicvariation. These samples come from the TSB07-26 drill core(Figs. 1 and 2), which is 10 km south of the PMH-24 drillcore that provided samples for an earlier Cr isotope studythat inferred the oxidative mobilization of Cr under an oxygenated atmosphere (Crowe et al., 2013). Crowe et al.(2013) referred to this horizon as the Nsuze paleosol, whileDelvigne et al. (2016) named it the Denny Dalton paleosolto avoid confusion with other paleosols known in the NsuzeGroup; but the paleosols are stratigraphically equivalentand would have formed under the same atmospheric redoxconditions. The Denny Dalton paleosol in the TSB07-26drill core consists of a 1.9 m thick upper zone mostly composed of sericite with minor chlorite and rutile (sericitedominated zone, SDZ), overlying a lower zone dominantlycomposed of chlorite with minor quartz, epidote, and rutile(chlorite-dominated zone, CDZ). The SDZ experiencedmetasomatism related to percolation of K (and Rb)-richfluids during diagenesis and low-grade metamorphism(Delvigne et al., 2016), which is a typical feature observedin many Precambrian paleosols (e.g. Macfarlane et al.,1994; Rye and Holland, 1998; Yang and Holland, 2003).The lowermost paleosol sample (WM15) and the underlying unweathered basaltic andesite (WM16) are outcropsamples taken close to the drill core location and correspond to depths of 10 and 15 m below the top of the paleosol, respectively. Despite being taken from outcrop, theparent basaltic andesite sample (WM16) has a chemicalindex of alteration (CIA) in the same range as fresh unaltered andesites (Colman, 1982; Delvigne et al., 2016;Elwood Madden et al., 2020). This CIA value is also comparable to or lower than CIA values of the parent rockexamined at a different location by Crowe et al. (2013).The paleosol is terminated at the top by the unconformablyoverlying Mozaan Group, suggesting that the very top ofthe paleosol profile may have been eroded away.2.2. Analytical methods2.2.1. Trace element analysisTrace element analyses were conducted at BostonUniversity. About 25 mg of sample powder was weighedinto 15 mL acid-cleaned Teflon beakers. For digestion,3 mL of concentrated HNO3, 1 mL of concentrated HCl,and 1 mL of concentrated HF were then added to the samples, the beakers were sealed, and placed on hot plate at 120 C for at least 24 h. The samples were sonicated forabout 30 min and 1 mL of hydrogen peroxide was addedto oxidize any organic material that may have been present.The vessels were left loosely capped until the solutionsstopped bubbling and were then re-sealed and placed on ahot plate for an additional 12–24 h. The samples weregently evaporated to dryness, and then re-dissolved in1 mL of concentrated HNO3. A small amount of H2O2(0.5–1 mL) was added to each sample to ensure completedigestion. Samples were then diluted with MQ water and2% HNO3 to a final dilution factor of 5000.All acid digestions were analyzed on a VG PlasmaQuadExCell inductively coupled plasma mass-spectrometer(ICP-MS) at Boston University. Samples were introducedto the instrument in 2% HNO3, via a Meinhardt-C nebulizer with a flow rate of about 1 mL/min, and an impactbead, Peltier cooled spray chamber. The instrument wasoptimized by obtaining the maximum signal intensity ona 1 ng/g solution of In, while concurrently minimizing oxideproduction. Oxide levels were monitored with the CeO /Ce ratio, which was typically 1%.Instrumental drift was monitored and corrected for byanalyzing an in-house drift solution several times duringan analytical session. A blank solution, which was preparedin an identical manner to our samples, was also measured atthe beginning and end of each run, and all data were blankcorrected. Several geological standard reference materialswere interspersed throughout an analytical run, and a calibration curve was generated by comparing the signal intensity to that of the established concentrations. Theconcentrations of the samples were then calculated basedon these calibration curves. The method precision, estimated from the 1 r.s.d. of the mean of repeat analyses ofthe USGS geostandard BHVO-2, was better than 3% forthe majority of elements. The accuracy of geostandard measurements, estimated from with the literature values for thismaterial, was within 5% for a large number of elements,including many elements (V, Co, Ni, Cu, Zn, Th, and U)discussed in detail in this study. Accuracy was 5–15% forthe remaining elements. The supplementary material contains the repeat analyses for BHVO-2 that were used forprecision and accuracy estimates.2.2.2. Iron isotopic analysisAnalytical procedures for iron purification and isotopicmeasurements followed standard procedures used in theOrigins Laboratory at the University of Chicago(Dauphas et al., 2009). Between 2.5 and 50 mg of powderedrock samples, calculated from published Fe concentrationdata to give a few 100 mg of Fe for analysis, were digestedin acid-cleaned Savillex Teflon beakers using a HF HNO3 HClO4 mixture, followed by digestion in aHCl HNO3 HClO4 mixture. After full digestion anddissolution, samples were evaporated and redissolved in0.5 ml of 6 M HCl for Fe purification. Iron was purifiedthrough a column chemistry procedure routinely used inthe Origins Laboratory (Dauphas et al., 2004, 2009). Aftertwo identical purifications, samples were dried down andredissolved in 5 ml 0.3 M HNO3 for mass spectrometry.Iron isotopic compositions were measured on a Neptunemulti-collector inductively coupled plasma mass spectrometer (MC-ICP-MS) at the University of Chicago inmedium-resolution mode. Sample solutions containing1 ppm Fe in 0.3 M HNO3 were introduced using a quartzcyclonic spray chamber, which gave a signal of 7 V. The

A.W. Heard et al. / Geochimica et Cosmochimica Acta 295 (2021) 1–235Table 1Major element, Fe, Ti, and Cr isotopic, and trace element composition of the Denny Dalton paleosol.WM 08WM 09WM 10WM 11WM 12WM 13WM 14WM 15WM 16ZoneDepth from paleosol top [m]SDZ 0.2SDZ 0.4SDZ 0.9INT. 2.2CDZ 3.4CDZ 4.3CDZ 5.4CDZ 10BAS. AND. 00.0470.027 0.1990.030 0.2400.0300.5090.0300.0670.028 0.2520.030 0.2500.0200.4040.0840.040.045 0.2500.030 0.3000.020 0.0900.0840.0760.013 0.2510.030 0.3000.020 0.0230.0840.0550.016 0.2650.030 0.2900.030 0.0570.0320.0390.023 0.2680.030 0.2400.020 0.0480.0320.070.026 0.2760.026 0.2400.0200.0530.0320.0750.029 0.2480.030 0.2600.0200.0830.0300.0540.023 0.2610.030 11.670.263.230.412.062.370.60‰d56Fe2r (95%d49Ti2r (95%d53Cr (a)2r (95%d53Cr (b)2r a from Delvigne et al. (2016).(a)Indicates data from this study.(b)Indicates data from Colwyn et al. (2019).

6A.W. Heard et al. / Geochimica et Cosmochimica Acta 295 (2021) 1–23isotopes 54Fe , 56Fe , 57Fe , and 58Fe were measuredsimultaneously along with 53Cr and 60Ni for correctionof 54Cr and 58Ni interferences on 54Fe and 58Fe respectively. The 53Cr and 60Ni interferences were corrected forusing the exponential law. Flat-topped Fe peak shoulderswere measured to overcome molecular interference fromargide ions (40Ar14N , 40Ar16O , 40Ar16O1H , and40Ar18O ). Standard-sample bracketing was used to correctFe isotopic ratio measurements for instrumental mass fractionation. Sample and standard 56Fe concentrations werematched to within 5% before analysis. Iron isotopic ratiosof samples are reported in d56Fe notation, which is theper mil deviation (see Teng et al., 2017 for a discussion ofd notations in non-traditional stable isotope systematics)of the 56Fe/54Fe ratio of the sample relative to the averageisotopic ratio of the IRMM-524 standard that has an identical isotopic composition to IRMM-014 and is very closein its isotope composition to the bulk silicate Earth value(Craddock and Dauphas, 2011):d56 Fe (‰) [(56 Fe/54 Fe)sample /(56 Fe/54 Fe)IRMM 014 -1] 1000Here and elsewhere unless otherwise stated, uncertainties are given at the 95% confidence interval (Dauphaset al., 2009). The IF-G and BHVO-2 geostandards wereprocessed through the same digestion and purification protocol as the paleosol samples and were analyzed along withthe paleosol samples in each measurement sequence. The Feisotopic composition for these geostandards was withinerror of their recommended values (Craddock andDauphas, 2011).2.2.3. Titanium isotopic analysisSamples were prepared for Ti isotopic analysis followingroutine procedures used at the Origins Lab at the University of Chicago (Millet and Dauphas, 2014; Millet et al.,2016; Greber et al., 2017). Samples were homogenized viaflux fusion using a lithium metaborate flux in a 1:6 sampleto flux ratio. This step also enables complete Ti digestion inacid. Aliquots of flux fusion sample pellets, containing atleast 20 lg of Ti were then digested in HNO3, and the47Ti-49Ti double spike solution was added according toknown Ti concentrations to give the optimal sample-spikemixing ratio for precise determination of Ti isotope composition, to correct for any potential mass dependent fractionation during column chemistry, and instrumental mass biasduring mass spectrometry (Millet and Dauphas, 2014). Thedissolved samples were purified using a column chemistryprocedure that followed the method of Zhang et al. (2011).The Ti isotope composition of each sample was analyzedon the Neptune MC-ICP-MS in medium-resolution modeat the University of Chicago. The samples were introducedin a 0.3 M HNO3 and 0.005 M HF mixture using a CetacAridus 1 desolvating nebulizer. Standard-sample bracketing was used, with 48Ti concentrations matched within 10% for the sample and standard, with typical 48Ti beamintensities of 20 V. Following a measurement block of fivesamples, a Ti-free solution of 0.3 M HNO3 and 0.005 MHF was measured to re-determine and correct for the onpeak baseline. The Ti isotope composition of each sampleis reported in d49Ti notation, which is the per mil deviationof the 49Ti/47Ti ratio of the sample relative to the OriginsLaboratory Ti reference material (OL-Ti), whose composition is close to that of chondrites and the bulk silicate Earth(Millet et al., 2016):d49 Ti (‰) [(49 Ti/47 Ti)sample /(49 Ti/47 Ti)OL Ti 1] 1000.The uncertainty on each measurement encompasses themeasurement session and the long-term external reproducibility and was evaluated according to Dauphas et al.(2009). The d49Ti isotope composition of each sample wascalculated by double-spike data reduction using a Mathematica script (Millet and Dauphas, 2014). The BHVO-2geostandard was processed through flux fusion, digestion,and Ti purification and was analyzed along with theunknown samples, giving a Ti isotopic composition withinerror of the recommended value.2.2.4. Chromium isotopic analysisAbout 50 mg sample powder was digested in a Microwave Digestion System (WX-8000) with 3 ml HNO3 and1 ml HF. The samples were heated at 220 C until completedigestion. Chromium concentrations were measured on aPerkin Elmer Elan DRCII ICP-MS at University of Scienceand Technology of China (USTC). Prior to Cr purificationusing column chromatography, sample solutions containing 1 lg Cr were mixed with 50Cr-54Cr double-spike solution and dried on a hot plate. Samples were re-dissolved in0.2 ml 6 N HCl and heated for 3 h at 120–130 C. Chromium was purified using a two-step cation exchange chromatography procedure following Zhang et al. (2019).Both column chemistry protocols made use of Bio-Rad200–400 mesh AG50-X8 resin, and the full proceduralblank was typically 10 ng, whilst Cr yields were 70–90%.Chromium isotopic compositions were measured on aTriton Plus multi-collector thermal-ionization mass spectrometer (TIMS) at USTC. Before analysis, about 1 lg ofpurified Cr was loaded in 3 N HNO3 on outgassed Re filaments with silica gel, saturated boric acid, and aluminumoxide. Sample and standard measurements were conductedat ionization temperatures between 1270 and 1390 C tominimize interferences present at lower and higher ionization temperatures. Chromium isotope signals (50Cr ,52Cr , 53Cr , and 54Cr ) were collected on L3, L1, axial,and H1 Faraday cups, respectively. All four cups were connected to 1011 X amplifiers. The typical beam intensity of52Cr was between 3 and 7 V. To monitor isobaric interferences from 50Ti and 50V on 50Cr and 54Fe on 54Cr ,intensities of 49Ti , 51V , and 56Fe were measured simultaneously on L4, L2, and H2 Faraday cups, respectively.L4 and L2 were connected to 1012 X amplifiers, and H2was connected to a 1011 X amplifier. Each analysis consisted of 200 cycles, each integrating isotope ratio measurement over 8.389 seconds. Double-spike data reduction wasperformed off-line to correct for mass fractionation duringcolumn chemistry and TIMS isotopic analysis.The Cr isotope ratios are expressed as the per mil variation relative to National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 979:

A.W. Heard et al. / Geochimica et Cosmochimica Acta 295 (2021) 1–23d53 Cr (‰) [(53 Cr/52 Cr)sample /(53 Cr/52 Cr)NIST979 1] 1000,relative to which the bulk silicate Earth value lies at 0.124 0.101‰ (Schoenberg et al., 2008). Each analytical sessionbegan with measurements of the spiked internal Cr isotopestandard SCP (Science, ON, Canada) and NIST SRM3112a, and SCP was analyzed between every five to six samples, which were each measured once or twice. The uncertainty reported here is the largest among 2 standard error(2SE) of a single sample measurement, 2 standard deviations (2SD) of repeat sample measurement, and the longterm reproducibility for the standard solution (SCP andNIST SRM 3112a; 0.03‰).3. RESULTS3.1. Elemental compositionsTrace element compositions are presented along withmajor element composition data from Delvigne et al.(2016) in Table 1. Elemental concentrations in paleosolsare best expressed in terms of their relative enrichmentsor depletions compared to the unaltered parent rock, andtherefore we employ the s notation after Brimhall andDietrich (1987), which is a measure of the mass fractionof an element i that is lost or added to the soil horizon relative to an immobile element (in this case Ti),si;w ¼½i w ½Ti w 1½i p ½Ti pwhere the subscripts w and p refer to weathered and unaltered parent rocks, respectively. When s is positive, it indicates that mass of element i has been added to the soilhorizon at the depth at which the value has been calculated,and vice versa. Calculated s values are reported in Table 2.We normalized to Ti to maintain convention with previousstudies, and we demonstrate below that Ti isotopes and thesTi value normalized to Al show that Ti was indeed immobile in this paleosol.As reported previously by Delvigne et al. (2016), thepaleosol is strongly depleted in Mg, Ca, Mn, and Fe(s 0.98), and, to a lesser extent, Ge (s 0.77), Na(s 0.64), and Si (s 0.61) in the upper SDZ (Fig. 3,Table 2). These data were taken to indicate intense chemicalweathering and desilication i

Anoxic continental surface weathering recorded by the 2.95 . gCAS Center for Excellence in Comparative Planetology, School of Earth and Space Sciences, University of Science and Technology of China, Hefei 230026, China . should be used when applying these ratios in shale studies of the ancient upper continental crust composition.