40ar 39ar geochronological constraints on the evolution of lateritic iron deposit qf spier c a et al...

(2006) 79–104www.elsevier.com/locate/chemgeo

Chemical Geology 234

40Ar/39Ar geochronological constraints on the evolution of lateriticiron deposits in the Quadrilátero Ferrífero, Minas Gerais, Brazil

Carlos Alberto Spiera⁎, Paulo M. Vasconcelos, Sonia M.B. Oliviera* Minerações Brasileiras Reunidas-MBR, Av. de Ligação 3580, Nova Lima, MG, 34000-000, Brazil

Received 4 January 2005; received in revised form 29 March 2006; accepted 11 April 2006

Abstract

Weathering profiles overlying the Sapecado, Pico and Andaime iron ore deposits, Quadrilátero Ferrífero (QF), Minas Gerais,Brazil, reach depths of 150–400 m and host world-class supergene iron orebodies. In addition to hosting supergene ore bodies ofglobal economic significance, weathered banded iron-formations at the Quadrilátero Ferrífero and elsewhere (e.g., Carajás,Hamersley) are postulated to underlie some of the most ancient continuously exposed weathering profiles on earth. Laserincremental-heating 40Ar/39Ar results for 69 grains of hollandite-group manganese oxides extracted from 23 samples collected atdepths ranging from 5 to 150 m at the Sapecado, Pico and Andaime deposits reveal ages ranging from ca. 62 to 14 Ma. Older Mn-oxides occur near the surface, while younger Mn-oxides occur at depth. However, many samples collected at the weathering–bedrock interface yield ages in the 51–41 Ma range, suggesting that the weathering profiles in the Quadrilátero Ferrífero hadalready reached their present depth in the Paleogene. The antiquity of the weathering profiles in the Quadrilátero Ferrífero iscomparable to the antiquity of dated weathering profiles on banded iron-formations in the Carajás Region (Brazil) and theHamersley Province, Western Australia. The age versus depth distributions obtained in this study, but not available for otherregions containing similar supergene iron deposits, suggest that little further advance of the weathering front has occurred in theQuadrilátero Ferrífero lateritic profiles during the Neogene. The results suggest that weathering in some of these ancient landscapesis not controlled by the steady-state advance of weathering fronts through time, but may reflect climatic and geomorphologicalconditions prevailing in a remote past. The geochronological results also confirm that the ancient landsurfaces in the QuadriláteroFerrífero probably remained immune to erosion for tens of millions of years. Deep weathering, mostly in the Paleogene, combinedwith low erosion rates, account for the abundance and widespread distribution of supergene iron, manganese, and aluminumorebodies in this region.© 2006 Elsevier B.V. All rights reserved.

Keywords: 40Ar/39Ar dating; Geochronology; Quadrilátero Ferrífero;Weathering; Iron ore; Sapecado; Paleoclimate; Geomorphology; Landscape evolution

1. Introduction

The Quadrilátero Ferrífero (QF) (literally, the IronQuadrangle), Minas Gerais (Fig. 1), hosts one of thelargest concentrations of lateritic iron ore deposits on

⁎ Corresponding author.E-mail address: [email protected] (C.A. Spier).

0009-2541/$ - see front matter © 2006 Elsevier B.V. All rights reserved.doi:10.1016/j.chemgeo.2006.04.006

Earth (Dorr, 1964; Dorr, 1969). It also hosts the mostcomplete and well-developed weathering profiles insoutheastern Brazil, reaching depths of 400–500 m, andaveraging 150 m. These weathering profiles are devel-oped on Paleoproterozoic Minas Supergroup rocks whichoutcrop along plateaus and ridges ranging from 1100 to2100m in elevation, surrounded by dissected areas,whereArchean granite–gneiss basement units and the Rio das

mailto:[email protected]

http://dx.doi.org/10.1016/j.chemgeo.2006.04.006

Fig. 1. Location and geological sketch map of the Quadrilátero Ferrífero region (After Alkmim and Marshak, 1998).

80 C.A. Spier et al. / Chemical Geology 234 (2006) 79–104

Velhas Greenstone Belt are exposed (Fig. 2) (Dorr, 1969).The QF weathering profiles have great economic signi-ficance because they host world-class supergene iron oredeposits as well as economically important manganese,bauxite, and industrial clay deposits. They also havesignificant geomorphologic and paleoclimatic implica-tions because of the extreme depths and antiquity ofweathering processes possibly recorded in these profiles(King, 1956; Dorr, 1969; Carmo and Vasconcelos, 2003;Carmo and Vasconcelos, 2004).

The formation and preservation of deep weatheringprofiles and massive supergene enrichment blanketsis characteristic of mineralized banded iron-formations(BIFs) across the planet (e.g., Quadrilátero Ferrífero,Carajás, Hamersley, Transvaal, etc.) (Dorr, 1964; Beisie-gel et al., 1973; Morris, 1985; Van Schalkwyk andBeukes, 1986; Harmsworth et al., 1990; Taylor et al.,2001; Spier et al., 2003) and suggest effective weatheringunder very low erosion rates. Weathering geochronology

studies on profiles overlying the CarajásMountains, Pará,Brazil, reveal deep (100–500 m deep) complete lateriticprofiles as old as 70Maoverlying the 600–900mplateaus(Vasconcelos et al., 1994a,b; Ruffet et al., 1996a; Shusteret al., 2005), while the 200–300 m surrounding plainsonly host incipient weathering profiles (20–40 m deepsaprolites) dated at ca. 12 Ma (Vasconcelos, unpublishedresults). Similar studies in the Hamersley Ranges,Western Australia (Vasconcelos, unpublished results)yield similar results, suggesting that the supergeneenrichment of banded iron formations in WesternAustralia dates as far back as ca. 80 Ma. These resultssuggest that the formation and preservation of supergeneenrichment blankets in iron ore deposits is highly depen-dent on a protracted weathering history (tens of millionsof years) combined with local conditions favoring lowerosion rates.

The physiographic position of the QF, protected byerosion-resistant steeply dipping quartzites and banded

Fig. 2. Digital elevation model for the Quadrilátero Ferrífero region (A) illustrating location of weathering profiles investigated in this study (B). (C)SW–NE topographic profile from the Serra da Moeda (MO) to the Sapecado mine (SAP) at the Serra do Itabirito.

81C.A. Spier et al. / Chemical Geology 234 (2006) 79–104

iron-formations and located at the elevated drainagedivide separating the Rio Doce, Rio São Francisco, andthe Rio Paraná basins, suggests that erosion rates in thisarea must be relatively low to permit the preservation ofthe deeply enriched weathering profiles (Dorr, 1969).Despite their importance, and some comprehensivegeomorphological studies of the QF (King, 1956; Dorr,1969; Barbosa, 1980; Varajão, 1991), several questionsregarding the genesis of QF weathering profiles remainunanswered: How did they form? How long did it take?When did it start? Under what climatic conditions did

these weathering profiles evolve? Why are they pre-served? Are they still forming today? Answers to thesequestions havemany implications to our understanding ofsurficial processes in general and in quantifying thegenesis of supergene ore deposits in particular.

Active mining operations in the QF provide a uniqueopportunity to answer some of these questions. Openpits and drill cores permit sampling complete weather-ing profiles. Extensive mining exposures and severalthousands of meters of drill-cores maximize the chancesof obtaining suitable samples for geochemical and


geochronological studies. Abundant Mn-oxides in someof these profiles, particularly the supergene ore bodiesformed on manganiferous itabirites at the Sapecado andAndaime MBR mines (Figs. 1 and 2), provide suitablesamples for 40Ar/39Ar dating. 40Ar/39Ar geochronologyof K-bearing manganese oxides formed at differentdepths in the weathering profiles permits measuring therate of propagation of the weathering front and thetiming of formation of the profiles.

The laser incremental-heating 40Ar/39Ar method hasbeen used to date supergene hollandite-group mineralsfrom weathering profiles across the world (Vascon-celos, 1992; Vasconcelos et al., 1992; Vasconcelos et al.,1994a; Dammer et al., 1996; Ruffet et al., 1996b; Hé-nocque et al., 1998b; Dammer et al., 1999; Vasconcelos,1999a,b; Li and Vasconcelos, 2002; Carmo and Vas-concelos, 2004, 2006) providing a useful tool in thestudy of weathering processes. Hollandite-group Mn-oxides are suitable to 40Ar/39Ar dating due to theirrelatively high K contents (1–5 wt.%) and their K and40Ar retentivity. These hollandite-group minerals gen-erally form clusters of oriented crystallites (10–20 μmlong and 1–2 μm wide) (Vasconcelos et al., 1994b).These clusters have a large volume of intercrystallinespace that may absorb and adsorb large quantities ofatmospheric 40Ar (Vasconcelos et al., 1995). On theother hand, hollandite-type crystallites form tunnelstructures that favor the imprisonment of large cationslike K, Ba, and Pb. These tunnel sites may quantitativelyretain 40Ar produced by 40K decay (Vasconcelos et al.,1995). Incremental-heating analyses of Mn-oxide clus-ters liberate atmospheric argon adsorbed or trapped inintercrystalline or damaged intracrystalline sites at lowtemperatures, while radiogenic 40Ar⁎ and nucleogenic39Ar hosted in the intracrystalline tunnel sites are onlyliberated at higher temperature steps. The ability to tapeach argon reservoir separately permits determiningaccurate and precise ages for these minerals, in additionto retrieving useful information on a sample's Ar and Kretentivity, the physical properties of the sites wherethese elements occur, the possible presence of con-taminants, and any significant 39Ar recoil losses themineral may undergo during neutron irradiation (Vas-concelos, 1999b).

We will present below the results of 40Ar/39Ar datingof 69 grains extracted from 23 samples of Mn-oxidesdistributed from ∼ 5 m to ∼ 150 m below the presentsurface in two distinct QF weathering profiles. We willdiscuss the reliability of the results obtained in light ofancillary mineralogical information obtained by petro-graphic observations and scanning electron microscopy(SEM), X-ray diffractometry (XRD), and electron

microprobe (EMP) analyses. We will finally discussthe significance of the geochronological results to theformation of the weathering profiles in the QF region, inparticular, and the genesis of the supergene iron orebodies, in general.

2. Geological setting

2.1. Regional geology and geomorphology

The geology and the lithostratigraphy of the QF weredescribed in detail by the USGS-DNPM team between1946 and 1963, and summarized by Dorr (1969). Sincethen, the subject has been reviewed by several authors(Ladeira and Viveiros, 1984; Marshak and Alkmim,1989; Chemale et al., 1994; Renger et al., 1994; Alkmimand Marshak, 1998). The most important lithostrati-graphic units of the Quadrilátero Ferrífero (Figs. 1 and3) are Archean to Paleoproterozoic granitic–gneissicbasement, Archean greenstone belt sequences (Rio dasVelhas Supergroup), the Paleoproterozoic Minas Super-group and post-Minas intrusive rocks.

The giant iron ore deposits of the QuadriláteroFerrífero are hosted in itabirites of the Minas Super-group, which consist of three Paleoproterozoicsequences (Renger et al., 1994; Babinski et al., 1995;Noce, 1995). The lowest unit is the Caraça Group,which consists of alluvial conglomerates and sandstonesgrading upwards into shallow-water marine siltstonesand mudstones. The intermediate Itabira Group consistsmainly of chemical sedimentary rocks. A metamor-phosed banded iron-formation unit of variable compo-sition occurs at the base of the Itabira Group and,together with hematitic and dolomitic phyllites, marblesand dolomites, constitutes the Cauê Formation. Theoriginal stratigraphic thickness of the iron formationmay have been in the order of 250–300 m, but it nowranges in apparent thickness from a few meters to morethan 1400 m, because of tectonic rock flowage (Dorr,1964). The Cauê Formation grades upward intodolomites and dolomitic/manganiferous itabirites ofthe 2.4 Ga Gandarela Formation (Babinski et al.,1995). Clastic sediments (with subordinate dolomites)of the Piracicaba Group overlie the Itabira Group. Theuppermost sequence of the Minas Supergroup is theSabará Group, a sequence of metamorphosed volcano-clastic sediments, turbidites, BIFs, and conglomerates(Renger et al., 1994) dated as Transamazonian (2.1–1.7 Ga.).

Deep weathering of the Itabira Group BIFs, postu-lated to have occurred during the Mesocenozoic,produced the large supergene iron orebodies spread

Fig. 3. Stratigraphic column of the Quadrilátero Ferrífero region (after Alkmim and Marshak, 1998).


throughout the QF (Dorr, 1969). A total of 29 Gton ofiron ore, mined continuously since the beginning of the40s and having produced more than 3 Gton of high-grade iron ore, are estimated to occur in the region(Wallace, 1965; Walde, 1986; TEX Report, 2002). TheQF total iron ore production in 2000 was 115 Mton andis estimated for 2004 at 160 Mton (TEX Report, 2002).Most of the deep weathering profiles in the region arecomplete stratified weathering profiles, exhibiting anindurated cap (canga) overlying absolute and relativeenriched supergene iron ores, which immediatelyoverlie saprolitic horizons.

Geomorphologically, the Quadrilátero Ferrífero con-sists of a roughly square 7000 km2 region (Fig. 2A)characterized by various plateaus, ranging from 1100 to2100 m elevation, surrounded by dissected plains at ca.500 to 600 m elevation. The quadrangle is delimited bysteeply dipping BIFs and quartzites that constituteelevated and highly resistant barriers to erosion andscarp retreat. The western limit of the QF is the Serra daMoeda, its northern limit the Serra do Curral, the easternlimit the Serra do Caraça, and the southern limit theSerra do Ouro Branco (Figs. 1 and 2). The core of the

Iron Quadrangle is drained by the north-flowing Rio dasVelhas, creating a dissected window into the ArcheanBação Complex (Fig. 2A). This inner depression issurrounded by several deeply weathered plateausdeveloped on BIFs and quartzites. The eastern sectionof the QF hosts a 700 km2 plateau, the Moeda Syncline,delimited by the Serras da Moeda (W), Serra do Curral(N), Serra das Serrinhas (E), and Serra do Ouro Branco(S) (Fig. 2A). This plateau hosts several importantsupergene ore bodies, including the Sapecado andAndaime deposits investigated in this study (Fig. 2B).Weathering profiles along the Serra da Moeda wereinvestigated in another study (Carmo and Vasconcelos,2003).

The QF plateaus have been interpreted as remnants ofan ancient erosion surface developed before the Africa–South America break-up (the Gondwana Surface ofKing (1956). Weathering profiles on the plateaus areconsistently deeper (up to 400 m) and more developedthan weathering profiles on surrounding plains anddissected valleys (b100 m). Landsurfaces on theplateaus are also distinct from the surrounding plainsbecause they are blanketed by strongly iron-hydroxide


cemented covers (cangas), which impart a smoothcharacter to the plateau landsurfaces. When the cangasare breached by recent erosion, they form sharp andangular scarps (Fig. 2c), another distinctive character-istic of this landscape. Finally, internal basins on theplateaus (e.g., Bacias do Fonseca and Gandarela), filledwith argillite, lignite, sandstone, diamictite, and kaolin-itic sediments now partially weathered to bauxite, areassociated with collapse structures (karstic lakes)formed during weathering of underlying banded ironformations (Ribeiro, 2003) or possibly created byextensional tectonic activity during the Paleogene(Sant'Anna et al., 1997; Maizatto, 2001). We chose tostudy the weathering history of two supergene iron oredeposits in the Moeda Syncline because of thelikelihood that these supergene systems host the mostcomplete information about the weathering history ofthe Quadrilátero Ferrífero and of the hypothesizedGondwana landsurface of King (1956). We also chose tostudy these profiles because the intimate relationshipbetween supergene iron ore and datable supergene Mn-oxides permits relating the timing of Mn-oxideprecipitation with the evolution of the chemicalreactions leading to the formation of supergene enrichediron orebodies in the region.

2.2. Local geology and geomorphology

Three adjacent iron mines (Sapecado, Pico, andAndaime) (Figs. 1 and 2) provide access to deeplyweathered and supergene enriched BIFs. Weatheringprofiles at these mines vary from a few meters to ca.400 m below the present surface (Fig. 4A–C). This largevariation in weathering profile depth results on aninterlayering of unweathered itabirites with deeplyweathered itabirites and soft-hematite ores. Spier et al.(2003) provide detailed characterization of the iron oresand their hosts rocks in these deposits. In this study, wedated samples from the Sapecado and Andaimeweathering profiles, briefly described below.

The Sapecado orebody, hosting 400 Mton of highgrade iron ore, represents the southern extension of thePico orebody. It is composed of lenses of soft and hardiron ores interlayered with relatively weathered siliceousitabirite. The major geological units present at theSapecado mine are itabirites and iron ores of the CauêFormation, clastic metasedimentary rocks of the Moedaand Batatal Formations, and predominantly chemicalmetasedimentary rocks of the Gandarela Formation(Fig. 4B–C). The rocks are oriented along a generalstrike N30°–35°E, dipping 45°–90° to SE or NW. Thequartzites of the Moeda Formation crop out continu-

ously along the eastern edge of the mine and show avery consistent sub-vertical dip and, in general, weaklydeveloped foliation. The quartzites exhibit abruptcontacts with the overlying phyllites of the BatatalFormation. The uppermost Gandarela Formation cropsout at the pit's western wall. It consists of dolomiticphyllite and deeply weathered dolomite with a gradualcontact with the itabirites. The main banded ironformation at the Sapecado mine is a siliceous itabirite,the protore of the iron ore bodies. This siliceous itabiriteis a metamorphosed BIF and consists of alternatemicrobands of quartz with martite and hematite. Total orpartial leaching of quartz during weathering promotedthe residual enrichment of iron, generating the soft ironores (Spier et al., 2003). Soft ores are highly porous andconsist of microbands of aggregates of martite andhematite still cemented together with highly porousbands where martite and hematite grains fill cavitiesgenerated by the removal of large quantities of primaryquartz and carbonate. Some remnant quartz, highlyetched and corroded, may occur in varied proportions.

The Andaime orebody (ca. 100 Mton) is located10 km North of the Pico mine (Fig. 1), representing thenorthern continuity of the iron mineralization. Here, therocks of the Minas Supergroup are oriented along strikeN10°–30°W, dipping 40°–80° to NE. The orebodyconsists of soft siliceous itabirite, the product ofweathering of siliceous banded iron formations thatoccur along the strike. Lenses of manganiferous itabiriteare interlayered with the iron ore.

Manganiferous itabirite is a particular type of softiron ore formed at the top of the Cauê Formation, nearthe contact with the Gandarela Formation (Fig. 4A–B)at all three— Pico, Sapecado, and Andaime—mines. Itconsists of irregularly distributed lenses varying fromdecimeters to 20 m width, 5–100 m length, andoccurring at depths varying from a few meters to150 m. The lenses occur parallel to the itabirite units,interbedded with them (Figs. 4A–B and 5A). Themanganiferous itabirite contains on average 48 wt.%(11 wt.%–63 wt.%) Fe and 6 wt.% (3 wt.%–14 wt.%)Mn. Rarely the Mn content exceeds 15 wt.%. In thiscase, the ore is called Fe–Mn ore. Manganese oxides inthe manganiferous itabirite are irregularly distributed.They generally cement the walls of pores in the soft ironores or form cm-size concretions precipitated in cavitiesand cracks (Fig. 5B).

The Pico, Andaime, and Sapecado orebodies arecapped by small basins filled with argillites andsiltstones. These basins, associated with karstic depres-sions created by the collapse of the BIF after leaching ofsilica and carbonate minerals during weathering

Fig. 4. Studied geological sections showing distribution of weathering profile. (A) Andaime Deposit. (B–C) Sapecado Mine.


Fig. 5. (A) Drill hole of the Andaime Deposit showing manganiferous itabirite between siliceous itabirite. (B) Detail of the manganiferous itabiriteshowing concretion formed by cryptomelane (c) and pyrolusite (p).


(Ribeiro and Carvalho, 2002; Ribeiro, 2003), were datedas Paleocene to Eocene based on pollen assemblages(Lima and Salard-Cheboldaeff, 1981; Maizatto, 2001).

3. Sampling and analytical procedure

The visual identification of blackMn-oxides dispersedthroughout dark hematite ores is a major challenge. Toovercome this sampling difficulty, we focused our searchfor datable supergeneMn-oxides on drill cores previouslyanalyzed for Mn. Sections with high Mn contents, wheremanganiferous itabirite and Fe–Mn ore occur, werecarefully examined with a hand lens and any suitableoccurrence of datable Mn-oxide sampled. Datable Mn-oxides, hollandite-groupminerals, are easily recognizablebased on their superior hardness (N6), a steel-bluemetallic luster, and a botryoidal texture. Section 6300Eof the Sapecado mine was chosen for detailed samplingdue to the presence of a prominent manganiferous itabiriteunit, which forms a continuous lens from the surfacedown to ∼ 150 m (Fig. 4B). We sampled each meter ofmanganiferous itabirite that contained suitable Mn-oxideminerals. The samples selected are supposed to representa time–stratigraphic sequence from the oldest samplesnear the surface to the youngest sample at the base of theweathering profile (fresh rock–weathering interface).

Fig. 6. Reflected-light photomicrographs of Mn-oxides from the Sapecado andMn-oxides investigated. (A) Cryptomelane (c) filling voids between hematitealternated layers of cryptomelane and voids (v). Pyrolusite (p) occurs in tcryptomelane and pyrolusite. Hematite and goethite (g) occur in the center of tthe size of the crystals with crystallization. (E) Detail of a coarse layer showpyrolusite. (h) Botryoidal texture showing alternation between pyrolusite an

Complementary sampling at Sections 4700E and3500E, at the Sapecado and Andaime mines, respective-ly (Figs. 2 and 4), provide information on weathering agevariation along strike. At Section 4700E (Sapecado), wesampled Mn-minerals that occur along fracture planes ofdeeply weathered dolomites of the Gandarela Formation(Fig. 4C); at Section 3500E (Andaime), we sampledmanganiferous itabirite (Fig. 4A).

Sample preparation followed the method detailed byVasconcelos (1999b) and Vasconcelos et al. (2002).Samples were crushed to 0.5–2 mm, ultrasonicallycleaned in distilled water and absolute ethanol, anddried under a heat lamp. The clean sample fragments wereexamined under a binocular microscope, and six to eightvisually homogeneous grains were hand-picked andloaded into aluminum disks for neutron irradiation. Thesamples were irradiated for 14 h, along with the FishCanyon Sanidine standard (28.02±0.28 Ma), in the Trigareactor at the B-1 CLICIT Facility, Radiation Centre,Oregon State University Reactor, Oregon, USA. After acooling-off period of three weeks, three grains from eachsample were loaded into a 145-pit copper disk, baked at180 °C for 13 h, and incrementally heated by a continuous10 W Ar-ion laser with a defocused beam at the UQ-AGES laboratory, Australia. Gases released duringincremental-heating were cleaned by a cryocooled trap

Andaime profiles showing textural relationships and mineralogy of thecrystals (h), and forming a thin layer. (B) Botryoidal texture formed byhe center of the concretion. (C) Mn-concretion formed by layers ofhe concretion. (D) Several layers of cryptomelane showing variation oning needle shape crystals between voids. (f–g) Several generations ofd goethite.


(T=−130 °C) and two C-50 SAES Zr–V–Fe getterpumps, and the clean gas fraction expanded into a MAP-215-50 mass spectrometer, equipped with a third C-50SAES Zr–V–Fe getter pump. Air pipettes and blankswere run before and after each sample to correct forbackground andmass discrimination. Air pipettes, blanks,and unknownswere analyzed by the peak hopingmethod,8–10 cycles were measured for each isotope, and 7measurements of 40Ar, 39Ar, 38Ar and 37Ar and 25measurements of 36Ar were made at each peak top.Isotope evolution was fitted with either parabolic or linearfits. Typical blank values and air pipette 40Ar/36Ar ratiosare reported in Appendix 1. The neutron flux for eachsample was estimated from J-factors (Appendix 1)calculated for each disk from the analyses of a minimumof 15 grains of FC sanidine standard placed in the diskfollowing the geometry illustrated in Vasconcelos et al.(2002). All ages are reported using the constants of Steigerand Jager (1977). Corrections for interfering isotopesgenerated during irradiation are as follows: (2.64±0.62)×10−4 for (36Ar/37Ar)Ca, (7.04±0.06)×10−4 for(39Ar/37Ar)Ca, and (8±3)×10−4 for (40Ar/39Ar)K. Datacorrected for blanks, mass discrimination, nucleogenicinterferences, and atmospheric contamination were usedto calculate apparent ages for each degassing step andare reported on Appendix 1 (available in the on lineversion of this manuscript or under request with theauthors).

From the same batch of grains used for isotope ana-lyses, additional aliquots of 3–5 grains were mountedin an epoxy container and polished for petrographic,SEM and EMP analyses. The SEM and EMP analyseswere carried out at the Centre for Microscopy andMicroanalyses at the University of Queensland, Aus-tralia. A JEOL JXA 8800L Superprobe, operating at15 kVand 15 nA current andwith 0.5 μmprobe diameter,was used for EMP analyses. Oxygen was analyzed asunknown and the following standards were used: MnO(Mn and O), Cl-apatite (P), CoO (Co), TiO2 (Ti), PbS(Pb), Al2O3 (Al), Fe2SiO4 (Fe, Si), ZnO (Zn), CaSO4

(Ca), BaSO4 (Ba), NaAlSi3O8 (Na) and KAlSi3O8 (K).Approximately 450 spots were analyzed on ∼ 60samples. Whenever possible, microprobe analyseswere pre-programmed perpendicularly to the growthbands to investigate possible variations in Mn-oxidecompositions during mineral precipitation.

Fig. 7. Backscattered electron micrograph showing several generationscryptomelane and pyrolusite (p). Hematite (h) occurs in the external layer. Nohollandite and pyrolusite cutting two generations of pyrolusite. (C) Veins ocryptomelane, hollandite and pyrolusite. Goethite (g) occurs subordinately. (forming a botryoidal texture.

4. Results and discussion

4.1. Mineralogy and petrography

Reflected-light microscopy and SEM investigationshow that cryptomelane, pyrolusite and hollandite arethe main Mn-oxides present. These minerals occurcementing pore spaces in the iron ore or form mm-sizegrowth bands (Fig. 6A–D). Minor Al-enrichmentgenerally occurs at the contact between Mn-oxidesand the hosting iron ore. Pure lithiophorite was notdetected in any of the grains analyzed. Most Mn-oxidesare mineralogically complex (Fig. 6D–G). Hematite(either as martite or tabular hematite) inclusions arecommon within Mn-oxide grains, particularly at thecontact of void-filling concretions with the iron ore (Fig.6A). Goethite and gibbsite bands alternate with Mn-oxides in some botryoidal samples (Fig. 6H).

Manganese oxide textures are also variable. Whencementing pores in the iron ore, hollandite-group Mn-oxides are generally very pure, cryptocrystalline, andcharacterized by botryoidal textures composed of micro-metric growth bands (Fig. 6B–D). Individual micro-bands display uniform grain size and mineralogicalcomposition. Cryptomelane crystals inmicrobands occuras 10–20 μm needle-like crystals, with 10:1 aspect ratio,and oriented perpendicularly to the growth bands (Fig.6E). When pyrolusite is present, coarse aggregates of0.01–0.20mm grains occur. Pyrolusite may also occur asradiating prismatic crystals aligned perpendicularly tothe growth bands, and it is often partially replaced byhollandite-group phases (Fig. 6F–G). Desiccation cracksare often observed in pyrolusite crystals (Fig. 6F–G).

Textural relationships and mineralogy of the samplesindicate that Mn-oxides were directly precipitated fromweathering solutions. Paragenetic relations among Mnminerals were difficult to assess. Generally, cryptome-lane and hollandite are older than pyrolusite (Fig. 6C),but this is not always the case. Multiple generations ofMn-oxides, revealed by banded and colloform textures,systems of intersecting veinlets, and mutual replace-ments, are observed (Fig. 7A–H). The small size of theMn-oxide growth bands (less than 20 μm in average)does not allow for the selection of fragments from asingle generation. In addition, many grains selected forgeochronology may contain minor supergene pyrolusite,

of Mn-oxides. (A) Layering of cryptomelane (c), hollandite (ho),te the thin vein of cryptomelane cutting the layering. (B) Thin veins off pyrolusite occurring in cryptomelane. (D–F) Several generations ofG) Two generations of pyrolusite. (H) Concentric layers of pyrolusite

Fig. 8. Ternary plots showing the major chemical composition of the Mn-oxides from the Sapecado and Andaime deposits. Note the completeexchanges between Al, Fe, K and Ba.


goethite, or gibbsite inclusions. Some grains also containmartite–hematite inclusions, which may cause furthercomplications during dating (see discussion below).Petrographic and SEM examinations, however, clearlyshow the absence of any hypogene K silicate in thesamples submitted to 40Ar/39Ar geochronology.

4.2. Electron microprobe analysis

Electron microprobe analyses for 448 spots in60 grains of Mn-oxides show that weathering solutionsfrom which they precipitated were rich in Mn, Al, Fe, KandBa (Fig. 8). Based on their K, Ba,Mn, andO contents,compositions of the grains analyzed are consistent withcryptomelane, hollandite, and pyrolusite stoichiometries(Table 1). Total Mn contents vary from 48–63 wt.% (Fig.9A). Variation on Mn concentrations reflects distinctmineralogy and variation in tunnel cation occupancy (Ba,K, etc.). When the contents of K and Ba were below 0.5%the mineral was identified as pyrolusite. Fig. 9B showsthat the potassium concentrations reach 3.9 wt.% incryptomelane (average of 2 wt.%).

4.3. Manganese oxide paragenesis

Field and petrographic observations indicate that Mn-oxides sampled in this study formed during chemicalweathering of the dolomite-bearing itabirite of the CauêFormation and the dolomite of the Gandarela Formation.Manganese and iron are present as minor elements in thedolomite. Manganese and iron are also common incarbonate cements associated with the diagenetic or hy-drothermal alteration of the BIFs. Mn+2 and Fe+2 arereleased into the weathering solutions during thedissolution of the Mn-bearing carbonates. As long as the

weathering solutions remain relatively reducing andneutral to acidic, Mn+2 stays in solution. If the Mn+2-bearing solutions encounter descending oxygenatedmeteoric waters along permeable horizons and faultplanes, Mn+2 readily oxidizes and precipitates as Mn+4-oxides and -hydroxides (pyrolusite, ramsdellite, manga-nite). Alternatively, when Mn+2-bearing weatheringsolutions encounter strongly alkaline conditions, whichoften occurs at the contact between the BIF and the hostcarbonates of the Gandarela Formation, Mn+2 may alsoreadily oxidize and precipitate as Mn+4 oxides/hydro-xides. If, in addition to Mn+2, the weathering solutionsalso contain dissolved K+1, Ba+2, or other cations,complex tunnel structuremanganese oxides (romanéchite,hollandite, cryptomelane, todorokite) may form. Theprocesses above account for the common occurrence ofMn-oxides along faults and along the contact between theBIF and the adjacent carbonate rocks. Manganese oxideparagenesis in the supergene iron orebodies suggests anintimate association of the oxides with weathering of thebanded iron formations. This, in turn, assures that geo-chronological analyses of the Mn-oxides provide infor-mation on the weathering history of the BIFs.

4.4. 40Ar/39Ar geochonological analysis

Laser-heating 40Ar/39Ar dating of 69 grains extractedfrom 23 samples of Mn-oxides yields the resultssummarized in Table 2 and illustrated in Figs. 10 and 11.Complete results, corrected for interfering isotopes andmass discrimination, are listed in Appendix 1.

The incremental heating spectra in Fig. 10 revealsthat 31 grains yield plateau ages, where plateaus aredefined by 2 or more contiguous steps yielding ≥50%of the total 39Ar gas released for the sample whose

Fig. 9. Histograms of the Mn and K contents of 60 Mn-oxide grains byca. 448 spot EMP analyses.

Table 1Average composition of the Mn-oxides investigated

Cryptomelane Hollandite Pyrolusite

(n=253) (n=131) (n=64)

O 34.75 33.66 34.45Mn 58.37 53.65 61.08K 1.99 1.22 0.08Ca 0.12 0.10 0.05Pb 0.02 – 0.02P 0.07 0.15 0.11Ba 0.47 5.11 0.12Fe 0.27 0.70 0.32Co 0.08 0.05 0.02Zn 0.05 0.01 0.03Ti 0.01 – –Na 0.10 0.04 0.01Al 0.51 1.37 0.22Si 0.04 0.10 0.08

Elements with (–) were not detected.


apparent age values are within 2σ from the mean value.Plateau ages are interpreted to represent the age ofprecipitation of the grains analyzed (Vasconcelos,1999a). Twenty-one grains do not yield plateausaccording to the definition above, but these grainsyield flat segments, containing ≥50% of total 39Arreleased, that define “plateau-like” ages. Ages for these“plateau-like” segments, identified on Fig. 10 by anasterisk next to the age, are consistent with “true”plateau ages yielded by other grains from the samesample (e.g., grains 2665-01/02/03 or grains 2659-01/02/03). Therefore, these “plateau-like” segments areinterpreted to also represent the approximate age ofprecipitation of the Mn-oxide grains. Five of the69 grains, identified on Fig. 10 by a cross, yield verylittle 39Ar to permit calculating an age. In addition, thesespectra are inconsistent with spectra displayed by othergrains from the same sample. The low K-contentssuggest that the grain(s) analyzed may contain Mn-minerals other than hollandite, and are not suitable forreliable geochronology. Analytical results from these5 grains are ignored from any further consideration. Theremaining 12 grains (identified on Fig. 10 by an # nextto the age) yield poor but geochronologically validresults that cannot be ignored. These grains eithersuggest the presence of minor contaminants and partialrecrystallization (Grain 2655-02), significant recoil(Grains 2680-02, 2680-03, and 2681-03), some recoilwith the presence of contaminants (Grains 2660-02 and2660-03), or yield relative well-defined “plateau-like”age, but the ages are significantly different among the3 grains from the same sample (Grains 2681-01, 2681-02, and 2681-03). Discussions on the interpretation of

this type of spectra is available in Vasconcelos (1999b)and Vasconcelos and Conroy (2003).

Petrographic and SEM examination indicates thatsamples whose various grains yield distinct plateau orplateau-like ages correspond to samples that containvisually distinct generations of Mn-oxides (Fig. 7D-Sample S15). These distinct generations are identifiedby cross-cutting relationships and may suggest thefollowing sequence of events: Mn-oxide precipitation, ahiatus, partial dissolution of first generation Mn-oxide,and reprecipitation of cross-cutting veins of a youngergeneration (or generations) of Mn-oxides.

Despite some of the complicating factors above, avery encouraging aspect of the results is the fact thatgrains from samples collected in closed proximity toeach other (e.g., samples S11-grains 2658-01/02/03, andS12- grains 2659-01/02/03) yield similar plateau or“plateau-like” ages, again suggesting that the 40Ar/39Arresults represent true mineral precipitation ages (Figs. 9and 10 and Table 2).

Table 2Summary of the 40Ar/39Ar geochronology data of the 23 samplesanalyzed in this study

Sample Location Elevation(m)

GrainID

Age (Ma)

Drill hole Depth (m)

S1 SAP 25/01 45.00 1399.31 2665-1 61.3±1.72665-2 61.5±1.2⁎

2665-3 61.0±5.0⁎

S2 SAP 17/01 22.60 1380.93 2649-1 48.6±0.92649-2 52.0±0.8⁎

2649-3 48.8±0.8S3 SAP 17/02 24.60 1379.20 2650-1 56.4±1.7⁎

2650-2 58.0±0.3S4 SAP 17/01 25.60 1378.33 2652-1 50.7±0.5⁎

2652-2 51.6±0.5⁎

2652-3 48.9±1.6⁎

S5 SAP 17/01 26.60 1377.46 2653-1 52.9±0.7⁎

2653-2 54.1±0.42653-3 53.0±0.5

S6 SAP 17/01 28.50 1375.82 2655-1 46.2±0.22655-2 55.6±0.5#

2655-3 45.2±0.5S7 SAP 19/01 41.00 1357.71 2661-1 49.6±0.6⁎

2661-2 46.9±0.22661-3 49.3±0.4⁎

S8 SAP 19/01 45.00 1354.25 2662-1 47.4±0.5⁎

2662-2 48.3±0.5⁎

2662-3 47.6±0.5S9 SAP 19/01 46.00 1353.38 2663-1 55.9±0.4

2663-2 56.1±0.32663-3 56.5±0.2

S10 SAP 19/01 49.00 1350.79 2664-1 45.1±0.8⁎

2664-2 45.5±0.32664-3 43.6±0.5

S11 SAP 16/01 67.00 1358.74 2658-1 45.3±0.22658-2 44.1±0.22658-3 44.4±0.3

S12 SAP 16/01 68.00 1357.87 2659-1 42.7±0.5⁎

2659-2 42.6±0.12659-3 43.0±0.2

S13 SAP 16/01 69.00 1357.01 2660-1 41.6±0.3⁎

S14 SAP 24/99 83.10 1340.16 2656-1 47.1±0.4⁎

2656-2 46.9±0.22656-3 46.5±0.2⁎

S15 SAP 09/99 75.70 1318.4 2666-1 48.0±3.02666-2 53.6±0.62666-3 46.4±1.0

S16 SAP 37/02 60.55 1277.95 2667-1 37.0±4.0#

2667-3a 33.0±5.0#

2667-3b 30.0±5.0#

S17 SAP 37/02 62.00B 1276.70 2673-1 18.0±0.5⁎

2673-2 19.2±1.52673-3 16.8±0.9

S18 SAP 37/02 62.50 1276.27 2674-1 14.2±0.22674-3 47.9±1.5#

S19 SAP 51/02 120.60 1351.07 2676-1 49.0±4.0#

2676-2 39.0±2.0#

2676-3 50.0±2.0#

S20 AND 15/02 53.30 1310.76 2681-1 33.2±0.5#

2681-2 39.4±0.3#

2681-3 31.8±0.4#

S21 AND 15/02 87.65 1281.01 2682-1 32.0±0.12682-2 32.3±0.4⁎

Table 2 (continued)

Sample Location Elevation(m)

GrainID

Age (Ma)

Drill hole Depth (m)

2682-3 32.5±0.4⁎

S22 AND 08/02 33.20 1310.60 2679-1 46.2±1.4⁎

2679-2 48.2±1.52679-3 45.4±1.3

S23 AND 08/02 84.15 1266.48 2680-1 49.1±1.3⁎

2680-2 36.0±3.0#

2680-3 46.0±3.0#

The uncertainties are given at the 2σ level. Ages with (⁎) are “plateau-like” ages. Ages with (#) are poor “plateau-like” ages (see text forexplanation).


Plateau and plateau-like ages for the QF samples rangefrom 61.3±1.7 to 14.2±0.2 Ma, but are particularly con-centrated at 51–41 Ma interval. All the analytical results,with the exception of the five samples that wereeliminated for the reasons above, are plotted in the ideo-grams in Fig. 12. These ideograms are useful to visualizethe distribution of ages of the supergene minerals formedat different times in a weathering profile (Vasconcelos,1999a). They are also useful in evaluating whether theresults obtained for a single weathering profile reveal acontinuum of ages, or whether the profile contains ageclusters. The ideograms in Fig. 12 reveal a discontinuousdistribution of ages.

A few surface samples from the Sapecado profile hadbeen previously dated in another study (Carmo andVasconcelos, 2003). In order to compare results obtainedfrom that study with our results, we plotted ideogramsfor Sapecado samples dated in each of these studies(Fig. 12A). The large differences in the number of grainsanalyzed, combined with the fact that in one of the studies(Carmo and Vasconcelos, 2003) only surface sampleswere dated, prevents a in-depth comparison between thetwo studies. However, both studies suggest that weath-ering at the Sapecado profile was already on going at thebeginning of the Paleogene.

Another ideogram, showing 40Ar/39Ar ages fromthe Sapecado and Andaime (this study) and Serra daMoeda (Carmo, 2004) profiles separately (Fig. 12B),confirms that the weathering histories preserved inthe Moeda Syncline profiles initiated near the K–Tboundary. Unfortunately, the large variation in samplesizes, the different sampling strategies (some studiesfocused primaily on surface samples while other stu-dies sampled deeper horizons in the weathering pro-files), and the relatively small number of samplesanalyzed for this large region (when compared withresults fromVasconcelos andConroy, 2003) prevents an in-depth interpretation of the weathering history of the three

Fig. 10. Incremental-heating spectra for the 3 grains of the 23 samples from the Sapecado and Andaime deposits. Ages with (⋆) are “plateau-like”ages. Ages with (#) are poor “plateau-like” ages (see text for explanation).


Fig. 10 (continued).




sites.Despite these limitations, the geochronological resultspermit answering some relevant questions regarding theweathering history of the Quadrilátero Ferrífero profiles.

4.5. How old are the Quadrilátero Ferrífero weatheringprofiles?

Well-defined and reproducible plateau and plateau-like ages, ranging from 61.5±1.2 Ma to 14.2±0.2 Ma(2σ), indicate a protracted history of chemical

weathering for the Quadrilátero Ferrífero, which wasalready ongoing in the Paleogene and continued untilthe Neogene. The ages obtained are compatible withages obtained from another study of QF weatheringprofiles (Carmo and Vasconcelos, 2003). They are alsoin agreement with the palynologic studies of theGandarela and Fonseca basins carried out by Lima andSalard-Cheboldaeff (1981) and by Maizatto (2001),which indicate that the sedimentation of these basinsoccurred during the Middle to Late Eocene. Considering

Fig. 11. Geological sections of the Andaime and Sapecado deposits showing distribution of the samples and age plateau for each grain of each sample(S20 33.2/39.4/56.3 means Sample 20 and age plateau for grains 1, 2 and 3). Legend as in Fig. 4.


that the genesis of the basins can be related to karsticcollapse structures formed during weathering of the un-derlying BIFs and dolomites (Ribeiro and Carvalho,2002), the geochronology results of the Sapecado and

Andaime mines, corroborated by the palynologic data,indicate a prolonged weathering history for theQuadrilátero Ferrífero, already ongoing in the earlyPaleocene.

Fig. 12. Probability density plot (ideogram) for all the significant steps (steps yielding negative ages, steps in which the analytical uncertainty was toogreat, and steps from samples showing evidence of major recoil or contamination were eliminated) analyzed for samples from the Moeda Synclineregion. (A) Data for the Sapecado mine from this study compared with those of (Carmo and Vasconcelos, 2003). (B) Data for the eastern limb of theMoeda Syncline (Sapecado and Andaime Deposits, this study) compared with those of the western limb (Serra da Moeda, from (Carmo andVasconcelos, 2003).


The absence of older geochronological results doesnot preclude a much older weathering history for theregion. Older manganese oxides could exist and havenot been sampled; older manganese oxides could havebeen dissolved by more recent weathering events; orolder weathering assemblages could have been eroded

from the upper parts of the profile during the past 60 Ma.At this stage, based on the results presented in this study,we can only affirm that the weathering profiles in theQuadrilátero Ferrífero are at least 61.5±1.2 Ma.

The longevity of the QF supergene blankets, deter-mined with modern geochronological techniques,


confirms long-held hypotheses that this area of SEBrazil preserved weathering profiles formed in anancient past (King, 1956; Dorr, 1969). These resultsconfirm the suitability and usefulness of weatheringgeochronology in testing landscape evolution modelsnot testable by other approaches (Vasconcelos, 1999b;Gunnell, 2003), and beg for the application of similarapproaches to weathering profiles elsewhere.

One surprising and noticeable feature of the distri-bution of ages in the three weathering profiles illustratedin Fig. 11 is the similarity in age vs. depth obtained. Fig.13 illustrates the relationship between ages and eleva-tions for the Moeda Syncline Mn-oxides. At first in-spection, there is no distinct trend. However, closerinspection reveals that for each individual site, Mn-oxideages decrease with increasing depth. A similar trend wasobserved on a weathering profile formed on Mn-bearingmetasediments in the Cachoeira Mine, MG (Carmo,2004); this trend was interpreted as indicating the firstarrival of oxidizing weathering solutions at progressive-ly greater depths (the propagation of the weatheringfront). A surprising feature of the present study is that,although older results are obtained from samples closerto the present surface while younger results occur indeeper horizons in the profile (Fig. 11A–B), several Mn-oxides sampled near the bedrock–weathering interfaceyield ages at ca. 40 Ma (Fig. 11B–C). Younger results(b30 Ma) occur in some deep horizons, but theoccurrence of old manganese oxides at the presentweathering–bedrock interface suggest that the MoedaSyncline weathering profiles had already reached theirpresent depth ca. 40 Ma, indicating that very littleadvance of the weathering front has occurred in theNeogene. Neogene ages found deeper in the profile may

Fig. 13. Correlation of age plateau and elevation for grains of Mn-oxides analyzed at the Moeda Syncline region. Data for the Andaimeand Sapecado weathering profiles are from this study and for the Serrada Moeda are from (Carmo and Vasconcelos, 2003). Note the poorcorrelation (r=0.3).

record recrystallization of older Mn-oxides during morerecent events, but they do not require much advance ofthe weathering front since ca. 40 Ma.

4.6. Continuous or episodic weathering record?

The presence of Paleogene sediments in karstic lakesoverlying the Cauê and Gandarela Formations, and theabsence of other thick sedimentary units in the area,suggest that the QF lithologies have been continuouslyexposed to surficial conditions since the beginning of theTertiary. Ideograms for the weathering profiles at theSapecado and Andaime mines, however, do not yield acontinuous distribution of ages. Instead, the geochrono-logical results strongly cluster in the 50–40 Ma interval,with other prominent peaks at ca. 56 (Sapecado) and 32(Andaime) Ma. The lack of surface samples from theAndaime profile may explain the absence of older resultsfor that profile. Although it is not our intention to comparethe ideograms, peak by peak, it is important to notice thatsome of the significant intervals (e.g., 50–40 Ma) ofmanganese oxide precipitation in the Sapecado profilesare also represented in the Serra daMoeda profiles studiedby Carmo (2004) (Fig. 12B). All of the profiles do showclustering of ages at specific time periods, independentlyof where in the profiles the samples were collected,suggesting an episodic history of mineral precipitation.Given the longevity and the depth of the profiles, thecomplex history of mineral dissolution and reprecipita-tion, and the large area covered by this study, furtherstudies, at different sites and including the analysis of amuch larger number of samples, are necessary if acompleteweathering history for theQuadrilátero Ferríferois to be obtained.

4.7. Paleoclimatic implications

The distribution of lateritic weathering profiles hasbeen used as a paleoclimatic indicator (Frakes, 1979;Nahon, 1986; Crowley and North, 1991). The authorsconsider the presence of those profiles as indicative ofwarm temperatures with seasonal rains, favoring intensechemical weathering and consequently the developmentof deep weathering profiles.

A major limitation in using weathering profiles aspaleoclimatic indicators is the absence of reliable nume-rical ages for these profiles. In addition, the few studieswhere ages of profiles are reported (Ruffet et al., 1996a;Dammer et al., 1996, 1999; Vasconcelos et al., 1994b)suggest that lateritic profiles do not form instantly butrecord a protracted history of mineral dissolution andreprecipitation lasting tens of millions of years. The

Fig. 14. Relative probability plots of 40Ar/39Ar ages from QuadriláteroFerrífero region (Sapecado and Andaime mines, this study), Carajásregion (Vasconcelos et al., 1994b; Ruffet et al., 1996a,b), West Africa(Hénocque et al., 1998a,b) andWestern Australia (Dammer et al., 1999).


present study confirms that lateritic profiles reflect a pro-tracted history of exposure. The ages of theminerals in theprofiles record times in the past more propitious tochemical weathering. While geochronological resultsrevealing that lateritic weathering profiles record aprotracted history of exposure and weathering maypreclude using the presence of these profiles as paleocli-matic indicators, the likelihood that mineral dissolutionand precipitation in these profiles is a function of climatemay permit using the distribution of supergene mineralages as a proxy for paleoclimates (Vasconcelos, 1999a).The majority of the supergene Mn-oxides samplesanalyzed for the QF precipitated between 51 and 41 Ma,suggesting hot and wet climatic conditions at that timeinterval. In addition, the occurrence of ca. 40 Mamanganese oxides at the bottom of the weatheringprofiles suggest that Paleogene greenhouse climaticconditions may have been much more significant for thedevelopment of the weathering profiles in the QF thansubsequent warm and humid climatic events. Palynologicdata for the Gandarela and Fonseca basins reveals thepresence of an exuberant and diversified paleoflora in theEocene, indicative of hot and wet environmental condi-tions (Maizatto, 2001). This period is coincident with themajor periods of chemical weathering registered at theCarajás, Amazon region, Brazil (Vasconcelos et al.,1994b; Ruffet et al., 1996a), in Australia (Dammer etal., 1996, 1999) and in Burkina Faso, Africa (Hénocque etal., 1998b) (Fig. 14). This might indicate that the climaticconditions that favored intense chemical weathering in theEocene had a more global character, consistent with theclimatic record for the Eocene obtained from oceansediments (Zachos, 1994l; Bowen et al., 2004).

4.8. Landscape evolution implications

The deeply stratified weathering profiles in the Quad-rilátero Ferrífero record weathering processes alreadyongoing at the Paleogene, and may suggest even moreancient weathering in the region. The absence of deeplystratified weathering profiles in the adjacent dissectedvalleys (Fig. 2A), and evidence suggesting that weather-ing profiles in these dissected parts of the landscape aremuch younger than the profiles on the plateaus (Carmoand Vasconcelos, 2004), corroborate previous sugges-tions that the landsurfaces blanketing the QF plateaus, the“Gondwana Surface” of King (1956) are much older thanthe adjacent dissected landsurfaces (King, 1956; Dorr,1969). A cycle of erosion (the “South America cycle”,King, 1956) post-dating the formation of this postulated“Gondwana Surface” would have partially eroded it,creating the more dissected surrounding landscape. Our

oldest geochronological results for theweathering profilesin the QF differ from King's estimated age for thepostulated Gondwana surface. And given that our resultspermit only determining the minimum age of weatheringfor the exposed profiles, the presence or absence of aCretaceous Gondwana surface remains unresolved.Furthermore, our study does not have the regional


character needed to address the existence or not of thesepostulated landsurfaces and erosional cycles. However,the geochronological results obtained for the QF profilessuggest that profiles in the elevated landscape associatedwith deeplyweathered banded iron formations, quartzites,and karstic paleolake sediments are distinctly older thanweathering profiles in the surrounded dissected plains andvalleys, consistent with the preservation of a more ancientlandsurface blanketing the QF plateaus.

The ancient landsurface in the Quadrilátero Ferrífero,easily recognizable based on its topographic position,displays additional characteristics clearly setting it apartfrom the dissected landscape that surrounds it: it hostsmuch deeper weathering profiles than the surroundinglandscape; local relief within the QF plateau is lesspronounced than in the surrounding plains; ages ofweathering profiles in the QF (this study) are muchgreater than ages for the surrounding plains (Carmo andVasconcelos, 2003, 2004, 2006); the QF surface isarmored by iron cemented profiles (cangas), which areabsent in the surrounding plains.

Several factors may have contributed to the preser-vation of the QF ancient weathering profiles: (1) the QFis both protected by a series of steeply dipping quartzitesand BIF units that act as barriers for scarp retreat drivenby the river drainage in the dissected plains; (2)weathering driven Fe-cementation has armored thelandscape in the QF, protecting the surface from erosion;(3) the elevated position, porous and permeable natureof the Fe-cemented cangas, and the relatively small arearepresented by the elevated plateau in the QF impedesthe development of an effective internal drainagesystem, slowing the erosive power of surface waters,enhancing chemical erosion by the deeply penetratingweathering solutions while, at the same time, slowingphysical erosion by diffusion or advection in slopes andstreams. The role of iron cementation in slowing erosionsuggests that weathering processes are self-enhancing,and when weathering has advance far enough to armorthe landscape through Fe cementation, erosion process-es slow down, promoting further weathering. This leadsto other questions: What is the minimum threshold ofweathering that initiates armoring of the landscape (asopposed to making material available for erosion)? Wasthe landsurface armored by weathering much larger thanthe plateau surface preserved today? When did armoringand diminished erosion begin to affect the QF landscapeevolution? Answers to these questions are crucial forunderstanding cyclical models of continental scalelandscape evolution, and these answers require numer-ical correlation of landsurfaces at regional and conti-nental scales to determine their possible co-genesis. This

study provides an extensive database of weatheringgeochronology results for a limited section of the Quad-rilátero Ferrífero that will permit correlation with otherparts of the QF and with other plateaus elsewhere in SEBrazil. We are currently also dating, by the (U–Th)/Hemethod (Shuster et al., 2005), goethite cements fromcangas at the Andaime, Pico, and Sapecado mines, toaddress the issue of when iron cementation and armor-ing of the landscape may have started.

The preservation of ancient weathering profiles posessignificant constraints on the erosional history ofcontinental landscapes (Vasconcelos, 1999a). As pointedout in Vasconcelos (1999a), continental areas hostingancient weathering profiles cannot have undergone km-scale erosion since the start of weathering; otherwise theweathering profiles would not be preserved. Therefore,the mere presence of ancient weathering profiles in theCarajás (Brazil) and Mount Isa (Australia) regionssuggests that apatite fission track modeling resultssuggesting 1.5–2 km for these regions during the Ceno-zoic possibly resulted from some artifact of the AFTmodeling approach. More recently, Gunnel (2003) eva-luated the compatibility between AFT modeling resultsfor the West African craton where Hénocque et al. (1998)dated weathering profiles as old as 49.8 Ma. Gunnel(2003) proposed that ages of weathering preserved in theprofiles were not compatible with the proposed erosionalhistory for the region based on AFT results. PublishedAFT results for areas (Serras da Mantiqueira and Mar)east of the QF suggest km-scale erosion, which could onlybe reconciled with the weathering history presented hereif the QF is separated from these areas of active uplift bycrustal-scale faults reactivated during the Cenozoic.

4.9. Comparison with the weathering geochronologyhistories of other regions

The 40Ar/39Ar ages of Mn oxides from Sapecado andAndaime are compared with the results of similarstudies from the Carajás Region (Brazil) (Vasconcelos etal., 1994b; Ruffet et al., 1996a,b); from WesternAustralia (Dammer et al., 1999) and from West Africa(Hénocque et al., 1998a,b) in Fig. 14. This figure showsepisodes of formation of weathering-related Mn-oxidesalternating with periods of relatively slow formation orcessation of these oxides spanning from Late Cretaceousup to Quaternary. It also shows a high degree ofsimilarity in the relative probability curve between 50and 40 Ma for 40Ar/39Ar ages from Southern andNorthern Brazil (Quadrilátero Ferrífero and Carajás,respectively) and West Africa, with a conspicuousoverlap of age clusters around 46–47 Ma.


The overlap in weathering events between 50 and42 Ma for the different weathering profiles in Brazil andWest Africa indicates the widespread occurrenceweathering-prone conditions during this period. Sinceweathering processes are controlled by water/rock inter-actions, and are accelerated with increasing temperatureand the presence of organic exudates, the concentrationof ages between 50 and 42Ma suggests that vast areas ofthe southern hemisphere cratons were under humid tro-pical conditions at that time.

4.10. Implications for the supergene enrichment ofbanded iron formations

The geochronological results in this study permitdrawing three major conclusions: that the supergeneenrichment blankets that make the Quadrilátero Ferrí-fero such a prolific iron-producing region were alreadyforming at more than 60 Ma; that most of the supergeneprocesses had already reached their present extent at ca.40 Ma; and that weathering processes in the Neogenehave contributed little to the further advance of theweathering fronts in the supergene orebodies. Inaddition, we may safely conclude that giant supergeneiron orebodies are likely to be found only in the highelevation plateaus (Gondwana Surface) that wereexposed to the more intense and prolonged weatheringhistory. Dissected plains and valleys surrounding the QFplateaus are likely to host only incipient supergeneblankets. The results for the QF are consistent withconclusions regarding the origin of supergene orebodiesin the Carajás (Vasconcelos et al., 1994a,b) andHamersley (MacLeod, 1966; Morris, 1985; Harmsworthet al., 1990; Morris, 1993) regions, suggesting that acombination of long continuous exposure, effectivedrainage, and subdued erosion rates are basic require-ments for the formation of supergene iron orebodies.

A surprising result from the QF study, and notpreviously measured in weathering geochronologystudies elsewhere, is the fact that the QF supergeneenrichment blankets likely reached their present depthsat ca. 40 Ma suggest that weathering solutionspercolating through the profile were ineffective atpromoting further advances of the weathering frontonce the profile reached a certain depth. This diminishedweathering capacity may be related to diminishingreactivity of the weathering solutions at depth (e.g.,depletion in dissolved oxygen or saturation withdissolved species), or it may reflect the horizontal, asopposed to the vertical, migration of the weatheringsolutions once they reach certain depths. Horizontallymigrating weathering solutions presently discharge at

the sides of the cliffs (e.g., Serra da Serrinhas, Fig. 2), atthe headwaters of local springs, and they contribute littleto the further advance of the weathering fronts.

This surprising result suggests that deep weathering,at least locally, is not necessarily an incremental oriterative process over time but may operate massively atdiscrete intervals of geological history, the rest of thetime being spent recycling and redistributing theproducts of weathering at the landscape scale.

5. Conclusions

Deep weathering profiles developed within Paleo-proterozoic BIFs of the Cauê Formation at theQuadrilátero Ferrífero host Mn-oxides suitable for40Ar/39Ar dating. These Mn-oxides are concentrated insupergene Fe–Mn ore and manganiferous itabirite at thecontact with the Gandarela Formation. They consist ofminerals of the hollandite group (cryptomelane andhollandite) and pyrolusite formed by precipitation fromcomplex (Mn-, Fe-, K-, and Ba-rich) weatheringsolutions. Petrographic and SEM investigations revealthe presence of several generations of Mn-oxides.

Laser-heating 40Ar/39Ar data reveal ages rangingfrom 62 to 14 Ma, indicating a prolonged history ofweathering at the Quadrilátero Ferrífero, already ongo-ing at least in Paleogene and continued until Neogene.The majority of Mn-oxides, however, precipitated from51 to 41 Ma with a peak at 46.7 Ma. This period iscoincident with the development of small basins formedby the collapse of the underlying BIFs during intensechemical weathering, as indicated by palynologic dataobtained in the sediments.

Samples collected near the present surface or near thecontact with the weathering–bedrock interface presentsimilar ages, indicating that the main weathering processwas developed around 47 Ma ago. Further chemicalweathering had limited effect on the weathering of theBIFs, which means that giant supergene iron oredeposits were completely formed at the Paleogene.

Geochronological 40Ar/39Ar results confirmed theintuitive view of low denudation rates at this site of theQuadrilátero Ferrífero, revealing that some sites of theEarth's surface can remain virtually uneroded for longgeological periods. They also suggest that formation ofdeep weathering profiles, at least locally, is not neces-sarily an incremental or iterative process over time butmay operate at discrete intervals of geological history.

The coincidence of the major period of precipitationof the Mn-oxides (and therefore of intense chemicalweathering) at the QF with major periods of precipita-tion in the Brazilian Amazon (Vasconcelos et al., 1994b;


Ruffet et al., 1996a), Australia (Dammer et al., 1996,1999) and West Africa (Hénocque et al., 1998a,b)suggests that climatic conditions which favored intensechemical weathering in the Eocene had a more globalcharacter.

Acknowledgements

This paper is an integral part of the senior author'sPh.D. thesis presented at the Instituto de Geociências ofthe Universidade de São Paulo (USP). This researchproject was made possible through the grant issued bythe Comissão de Aperfeiçoamento de Pessoal de NívelSuperior (CAPES) to C.A.S (grant BEX2189/02-0). C.A.S. thanks Minerações Brasileiras Reunidas (MBR) forliberating him of his usual activities as a mine geologistduring the period of his stay at the University of Queen-sland. The construction of the Ar lab at UQ was partiallyfunded by ARC Large Grant A39531815. 40Ar/39Aranalyses were funded by UQ-AGES. We are grateful tothe staff of the Center of Microscopy and Microanalysesof the University of Queensland (CMM), particularlyRonald Rasch, Graeme Auchterlonie, and Kim Sewell,for their assistance. C.A.S. is indebted to Isabela deOliveira Carmo for the DTM images and availability ofthe 40Ar/39Ar data of the Serra da Moeda region. GillesRuffet and one anonymous reviewer are thanked fortheir detailed and insightful review comments. [LMW]

Appendix 1. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.chemgeo.2006.04.006.

References

Alkmim, F.F., Marshak, S., 1998. Transamazonian orogeny in the SãoFrancisco craton, Minas Gerais, Brazil: evidence for Paleoproter-ozoic collision and collapse in the Quadrilátero Ferrífero.Precambrian Research 90, 29–58.

Babinski, M., Chemale, F., Van Schmus, W.R., 1995. The Pb/Pb agesof the Minas Supergroup carbonate rocks, Quadrilátero Ferrífero,Brazil. Precambrian Research 72, 235–245.

Barbosa, G.V., 1980. Superfícies de erosão no Quadrilátero Ferrífero,Minas Gerais. Revista Brasileira de Geociências 10, 89–101.

Beisiegel, V.R., Bernardelli, A.L., Drummond, N.F., Ruff, A.W.,Tremaine, J.W., 1973. Geologia e recursos minerais da Serra dosCarajás. Revista Brasileira de Geociências 3, 215–242.

Bowen, G.J., Beerling, D.J., Koch, P.L., Zachos, J., Quattlebaum, T.,2004. A humid climate state during the Paleocene/Eocene thermalmaximum. Nature 432, 495–499.

Carmo, I.O., 2004. Geocronologia do Intemperismo Cenozóico noSudeste do Brasil. Ph.D. Thesis, Universidade Federal do Rio deJaneiro, Rio de Janeiro, 134 pp.

Carmo, I.O., Vasconcelos, P., 2004. Geochronological evidence forpervasive Miocene weathering, Minas Gerais, Brazil. EarthSurface Processes and Landforms 29, 1303–1320.

Carmo, I.O., Vasconcelos, P.M., 2003. 40Ar/39Ar geochronology ofweathering profiles in southeastern Brazil. IV South Americansymposium on isotope geology. Extended Abstracts, Salvador,pp. 49–52.

Carmo, I.O., Vasconcelos, P.M., 2006. 40Ar/39Ar geochronology con-straints on Late Miocene weathering rates in Minas Gerais, Brazil.Earth and Planetary Science Letters 241, 80–94.

Chemale, F., Rosière, C.A., Endo, I., 1994. The tectonic evolution ofthe Quadrilátero Ferrífero, Minas Gerais, Brazil. Precambrian Re-search 65, 25–54.

Crowley, T.J., North, G.R., 1991. The Last 100 Million Years,Paleoclimatology. Oxford Univ. Press, London/New York,pp. 183–211.

Dammer, D., Chivas, A.R., McDougall, I., 1996. Isotopic dating ofsupergene manganese oxides from the Groote Eylandt deposit,Northern Territory, Australia. Economic Geology 91 (2), 386–401.

Dammer, D., McDougall, I., Chivas, A.R., 1999. Timing ofweathering-induced alteration of manganese deposits in WesternAustralia: evidence from K/Ar and 40Ar/39Ar dating. EconomicGeology 94, 87–108.

Dorr, J.V.N., 1964. Supergene iron ores of Minas Gerais, Brazil.Economic Geology 59 (7), 1203–1240.

Dorr, J.V.N., 1969. Physiographic, stratigraphic and structuraldevelopment of the Quadrilátero Ferrífero, Minas Gerais, Brazil.U.S.G.S. Professional Paper, vol. 614-A. Washington. 110 pp.

Frakes, L.A., 1979. Climates Throughout Geologic Time. Elsevier,Amsterdam/Lausanne/New York.

Gunnell, Y., 2003. Radiometric ages of laterites and constraints on long-term denudation rates in west Africa. Geology 31 (2), 131–134.

Harmsworth, R.A., Kneeshaw, M., Morris, R.C., Robinson, C.J.,Shrivastava, P.K., 1990. BIF-derived iron ores of the HamersleyProvince. In: Hughes, F.E. (Ed.), Geology of the Mineral Depositsof Australia and Papua New Guinea. The Australasian Institute ofMining and Metallurgy, Melbourne, pp. 617–642.

Hénocque, O., Ruffet, G., Colin, F., 1998a. 40Ar/39Ar dating of Africanmanganese pisolites: implications for Cenozoic climatic changes.Mineralogical Magazine 62A (Part 1), 606–607.

Hénocque, O., Ruffet, G., Colin, F., Feraud, G., 1998b. 40Ar–39Ardating of west African lateritic cryptomelanes. Geochimica etCosmochimica Acta 62 (16), 2739–2756.

King, L.C., 1956. A geomorfologia do Brasil oriental. RevistaBrasileira de Geografia 2, 3–121.

Ladeira, E.A., Viveiros, J.F.M., 1984. Hipótese sobre a estruturação doQuadrilátero Ferrífero com base nos dados disponíveis. Boletim daSociedade Brasileira de Geologia de Minas Gerais 4.

Li, J.W., Vasconcelos, P., 2002. Cenozoic continental weathering andits implications for the palaeoclimate: evidence from 40Ar/39Argeochronology of supergene K–Mn oxides in Mt Tabor, centralQueensland, Australia. Earth and Planetary Science Letters 200(1–2), 223–239.

Lima, M.R., Salard-Cheboldaeff, M., 1981. Palynologie des bassins deGandarela et Fonseca (Eocene de L'etat de Minas Gerais, Bresil).Boletim Instituto de Geociências-Universidade de São Paulo 12,33–54.

MacLeod, W.N., 1966. The geology and iron deposits of theHamersley Range area, Western Australia. Geological Survey ofWestern Australia, Bulletin 117, 1–170.

Maizatto, J.R., 2001. Análise bioestratigráfica, paleoecológica esedimentológica das bacias Terciárias do Gandarela e Fonseca-

http://dx.doi.org/doi:10.1016/j.chemgeo.2006.04.006

http://dx.doi.org/doi:10.1016/j.chemgeo.2006.04.006


Quadrilátero Ferrífero-com base nos aspectos palinológicos esedimentares. PhD Thesis, Universidade Federal de Ouro Preto,Ouro Preto, 249 pp.

Marshak, S., Alkmim, F.F., 1989. Proterozoic contraction/extensiontectonics of the southern São Francisco craton region, MinasGerais, Brazil. Tectonics 8 (3), 551–571.

Morris, R.C., 1985. Genesis of iron ore in banded iron-formation bysupergene and supergene–metamorphic processes— a conceptualmodel. In: Wolf, K.H. (Ed.), Handbook of Strata-Bound andStratiform Ore Deposits. Elsevier, Amsterdam, pp. 73–235.

Morris, R.C., 1993. Genetic modeling for banded iron formation of theHamersley Group, Pilbara Craton, Western Australia. PrecambrianResearch 60, 243–286.

Nahon, D.B., 1986. Evolution of iron crusts in tropical landscapes. In:Cohen, S.M., Dethier, D.P. (Eds.), Rate of Chemical Weathering ofRocks and Minerals. Academic Press, New York, pp. 169–191.

Noce, C.M., 1995. Geocronologia dos eventos magmáticos, sedi-mentares e metamórficos na região do Quadrilátero Ferrífero,Minas Gerais. Ph.D. Thesis, Universidade de São Paulo, SãoPaulo, 128 pp.

Renger, F.E., Noce, C.M., Romano, A.W., Machado, N., 1994.Evolução sedimentar do Supergrupo Minas: 500 Ma. de registrogeológico no Quadrilátero Ferrífero, Minas Gerais, Brasil.Geonomos 2 (1), 1–11.

Ribeiro, D.T., 2003. Enriquecimento Supergênico de FormaçõesFerríferas Bandadas: Estruturas de Colapso e Desordem. Ph.D.Thesis, Universidade Federal do Rio de Janeiro, Rio de Janeiro,123 pp.

Ribeiro, D.T., Carvalho, R.M., 2002. Simulation of weathered iron orefacies: integrating leaching concepts and geostatistical model. In:Armstrong, M., Bettini, C., Champigny, N., Galli, A., Remacre, A.(Eds.), Geostatistics Rio 2000. Kluwer, Dordrecht, pp. 101–115.

Ruffet, G., Innocent, C., Michard, A., Feraud, G., Beauvais, A.,Nahon, D., Hamelin, B., 1996a. A geochronological 40Ar/39Ar and87Rb/87Sr study of K–Mn oxides from the weathering sequence ofAzul, Brazil. Geochimica et Cosmochimica Acta 60 (12),2219–2232.

Ruffet, G., Innocent, C., Nahon, D., Feraud, G., Hamelin, B., 1996b.Geochronological study of weathering profile by 40Ar/39Ar analysisof supergene K–Mn oxides. Terra Abstracts 7 (Suppl. 1), 350.

Sant'Anna, L.G., Schorscher, H.D., Riccomini, C., 1997. Cenozoictectonics of the Fonseca Basin region, eastern QuadriláteroFerrífero, MG, Brazil. Journal of South American Earth Sciences10 (3–4), 275–284.

Shuster, D.L., Vasconcelos, P.M., Heim, J.A., Farley, K.A., 2005.Weathering geochronology by (U–Th)/He dating of goethite.Geochimica et Cosmochimica Acta 69 (3), 659–673.

Spier, C.A., Oliveira, S.M.B., Rosière, C.A., 2003. Geology andgeochemistry of the Águas Claras and Pico iron mines,Quadrilátero Ferrífero, Minas Gerais, Brazil. Mineralium Deposita38, 751–774.

Steiger, R.H., Jager, E., 1977. Subcommission on geochronology:convention on the use of decay constants in geo- and cosmochro-nology. Earth and Planetary Science Letters 36, 359–362.

Taylor, D., Dalstra, H.J., Harding, A.E., Broadbent, G.C., Barley, M.E.,2001. Genesis of high-grade hematite orebodies of the HamerleyProvince, Western Australia. Economic Geology 96, 837–873.

TEX Report, 2002. Iron Ore Manual 2001–2002. The TEX ReportLtd, Tokyo.

Van Schalkwyk, J.F., Beukes, N.J., 1986. The Sishen iron ore deposit.Mineral Deposits of Southern Africa. Geological Society of SouthAfrica, Johannesburg, pp. 157–182.

Varajão, C.A.C., 1991. A questão das superfícies de erosão doQuadrilátero Ferrífero, Minas Gerais. Revista Brasileira deGeociências 21 (2), 138–145.

Vasconcelos, P.M., 1992. Timing and rates of evolution of hydro-chemical systems in semiarid and humid environments byapplication of 40K–40Ar and laser-heating 40Ar–39Ar dating ofK-bearing weathering product minerals. Ph. D. Thesis, Universityof California, Berkeley.

Vasconcelos, P.M., 1999a. 40Ar/39Ar geochronology of supergeneprocesses in ore deposits. Economic Geology 12, 73–113.

Vasconcelos, P.M., 1999b. K–Ar and 40Ar/39Ar geochronology ofweathering processes. Annual Review of Earth and PlanetarySciences 27, 183–229.

Vasconcelos, P., Conroy, M., 2003. Geochronology of weathering andlandscape evolution, Dugald River Valley, NW Queensland,Australia. Geochimica et Cosmochimica Acta 67 (16), 2913–2930.

Vasconcelos, P.M., Becker, T.A., Renne, P.R., Brimhall, G.H., 1992.Age and duration of weathering by 40K–40Ar and 40Ar/39Aranalysis of potassium–manganese oxides. Science 258, 451–455.

Vasconcelos, P.M., Brimhall, G.H., Becker, T.A., Renne, P.R., 1994a.40Ar/39Ar analysis of supergene jarosite and alunite: implicationsto the paleoweathering history of the western USA and WestAfrica. Geochimica et Cosmochimica Acta 58, 401–420.

Vasconcelos, P.M., Renne, P.R., Brimhall, G.H., Becker, T.A., 1994b.Direct dating of weathering phenomena by 40Ar/39Ar and K–Aranalysis of supergene K–Mn oxides. Geochimica et Cosmochi-mica Acta 58 (6), 1635–1665.

Vasconcelos, P.M., Renne, P.R., Becker, T.A., Wenk, H.R., 1995.Mechanisms and kinetics of atmospheric, radiogenic and nucleo-genic argon release from cryptomelanes during 40Ar/39Ar analysis.Geochimica et Cosmochimica Acta 59 (10), 2057–2070.

Vasconcelos, P.M., Onoe, A.T., Kawashita, K., Soares, A.J., Teixeira,W., 2002. 40Ar/39Ar geochronology at the Instituto de Geociências,USP: instrumentation, analytical procedures, and calibration.Anais da Academia Brasileira de Ciências 74 (2), 297–342.

Walde, D.H.G., 1986. Geologia do ferro. In: Schobbenhaus, C.,Coelho, C.E.S. (Eds.), Principais Depósitos Minerais do Brasil.Ferro e metais da indústria do aço. Departamento Nacional daProdução Mineral, Brasília, p. 501.

Wallace, R.M., 1965. Geology and mineral resources of the Pico doItabirito District, Minas Gerais, Brazil. U.S.G.S ProfessionalPaper, vol. 341-F. 68 pp.

Zachos, J.C., 1994. From the greenhouse to the icehouse — aSouthern-Ocean perspective of paleogene climate. Oceanus 36 (4),57–61.

40ar 39ar geochronological constraints on the evolution of lateritic iron deposit qf spier c a et al...

Documents

qf weathering profiles

exposed weathering profiles

incipient weathering

abstract weathering

effective weathering

quadriltero ferrfero

quadriltero ferrfero

profiles king