635

The Canadian MineralogistVol. 37, pp. 635-651 (1999)

THE MOBILITY OF ZIRCONIUM AND IDENTIFICATION OF SECONDARYZr-BEARING PHASES lN BAUXITE FROM POSOS DE CALDAS, MINAS GERAIS,

BRAZIL: A MASS-BALANCE AND X-RAY ABSORPTION SPECTROSCOPIC STUDY

LAURE DUVALLET$

Laboratoire de Miniralogie-Cristallographie, UMR 5563 CNRS, Universiti Paul Sabatier,39, Alldes Jules-Guesde, F-31000Toulouse, France et Instituto de Geoci€ncias, Universidade de Sdo Paulo,

Rua do Ingo 562, Cidade Universitaria, Caixa Postal I I 348, Cep 05422-970, Sdo Paulo, SP, Brasil

FRANqOIS MARTIN!

Laboratoire de Mindralogie-Cristallographie, UMR 5563 CNRS, Universitd PauI Sabatier,39, All4es Jules-Guesde, F-31000 Toulouse, France

FRANqOIS SOUBIES$

ORSTOM, URl2 Gdosciences de I'Environnement Tropical, 32, Avenue Henri Varagnat, F-93143 Bondy, France

STEFANO SALVI

Laboratoire de Mindralogie-Cristallographie, UMR 5563 CNRS, Universitd Paul Sabatier,39, All4es Jules-Guesde. F-31000 Toulouse. France

ADOLFO JOSE MELFI9

Departamento de Geofisica, Instituto Astronbmica e Geofisica, Universidade de Sdo Paulo,Caixa Postal 30627, Cep 01051, SAo Paulo, SP, Brasil

JEAN-POL FORTUNE

Labortttoire de Mindralogie-Cristallographie, UMR 5563 CNRS, Universitd Paul Sabatier,39, All4es Jules-Guesde, F-31000 Toulouse, France

ABSTRACT

The geochemistry and mineralogy of Zr have been studied in a bauxitic profile derived from a nepheline microsyenite atPogosdeCaldas,MinasGerais,southeastemBrazil TheZrintheparentrockishostedbythesorosilicatehainite(54VoZrO2)and a Ca-Na amphibole (0 14.2Vo ZtO2\. Comparison of the variations in Zr, Ti, Nb and Th contents as function of the apparentdensity of different zones in the top few meters of the saprolite shows Th to be the least mobile element of the profile Therefore,it was chosen as an invariant constituent in order to evaluate the mobility of other elements, especially the Zr. Mass-balancecalculations demonstrate the leaching of Zr all along the profile, with a rate of removal of up to 407o at the weathering front. Inthe residual bauxitic materials, Zr is essentially localized in fem.rginous products such as goethite-rich plasma filling the pores,and granules within the gibbsite-dominant maffix. Its mineralogical expression appears to be within polyhedra with an average of6.6Oatoms, andaverageZr-Odistancesof 2.l6A,probablyinadsorptiononthesurfaceof goethiteintheporesandingranulesof the gibbsite-rich matrix.

Keywords:batxite, thorium, mass balance, zirconium, mobilization, X-ray absorption spectroscopy, baddeleyite, Pogos de Caldas,Braztl.

Sowarnr

La g6ochimie et la min6ralogie du zirconium ont 6t6 6tudi6es dans un profil lat6ritique provenant d'une sy6nite du complexede Pogos de Caldas, 6tat du Minas Gerais, dans le sud-est du Br6sil. La teneur enZr de la roche mbre provient de deux silicates:

e E-mail addresses: lorduval@usp.br, martinf@cict.fr, soubies@orstom.rio.net, ajmelfi@usp.br

636 THE CANADIAN MINERALoGIST

lahainite (5-6Vo ZrO2) et une amphibole Ca-Na (0 .lO.2Vo ZrO2). Des comparaisons effectu6es sur les variations des teneurs enTi, Nb et Th en fonction des densit6s apparentes des diff6rentes zones sur les premiers mdtres de la saprolite montrent que lethorium est l'6l6ment le plus immobile du profil. La concentration en Th est invariante par rapport d celle des autres 6l6ments etpermet de calculer des balances de masse, sp6cialement poutrZr. Les observations mon[ent qteleZr est lessiv6 dans tous leshorizons du profil, avec un taux de lixiviation de 407o enzx02atfront d'alt6ration. Dans la bauxite, le Zr-se trouve dans despolybdres de coordination ayant 6.6 atomes d'oxygbne, en moyenne, avec une distance moyenne de 2.16 A, probablement enadsorption h la surface des plasmas riches en goethite en cutanes dans les pores et dans les granules de la matrice riche en gibbsite.

Mots-cMs: bauxite, thorium, mass balance, zirconium, mobilisation, spectroscopie d'absorption des rayons X, baddeleyite, Pogosde Caldas, Brazil.

lurnooucrron

Zirconium is commonly considered an immobile el-ement during weathering processes. The accumulationof residual Zr in materials formed under supergene al-teration has been widely documented over the years(e. 9., Goldschmi dt l93T, 1 954, Gordon & Mtrata 19 52,Degenhardt 1957, Adams & Richardson 1960, Bene-lavskii 1963, Zeissink 1971, Erlank et al. 1978). Asexpected, the greatest concentrations reported are fromstrongly leached environments, such as in lateritic baux-ites, where Zr can easily become concentrated by a fac-tor of 3 to 4 (Adams & Richardson 1960).

The immobility of this element can be easily ex-plained by (i) the fact that Zr generally occurs in therocks as zircon, which is known to be a mineral highlyresistant to dissolution (e.9., Milnes & Fitzpatrick 1989),and (ii) the extremely low solubility of Zr in aqueoussolutions (Levinson 1980, Brookins 1988). As a result,zirconium has often been used as an immobile elementto evaluate mass transfer in weathered profiles (e.9.,Brimhall et al. 1991, Freyssinet 1994).

However, the high Zr content of many bauxites can-not be accounted for solely by the amount of zirconpresent. Degenhardt (1957), Benelavskii (1963) andZakrutkin & Shvetsiva (1974)have suggested that Zr inbauxites can occur also as colloidal oxides or hydrox-ides (ZrO2.4ll2O), more or less sorbed on secondaryminerals, and can isomorphically replace iron and alu-minum in the structure of oxyhydroxides or kaolinite. Ithas also been suggested that zircon stability may be low-ered and Zr mobllity enhanced in some strongly acidand organic-matter-r ich media, such as podzols(Swindale & Jackson 1956, Ronov et al. 1961, Kimura& Swindale 1967,Berrow et al. l97S,TejanKellaet al.1991). A review of this subject can be found in Milnes& Fitzpatrick (1989). Lastly, Colin et al. (1993) and,Braun et al. (1993) have questioned, with stronger ar-guments than Carroll (1953) produced in the past, the"axiomatic immobility" of zircon in soils and the use ofzirconium as an invariable element in mass-balance cal-culations. This brief overview points to the likelihoodthat some mobilization of Zr and dissolution of zfuconcan occur in specific surficial environments. However,the neoformed secondary zirconiferous phases have notbeen well characterized yet.

This paper deals with a Brazilian occurrence ofbaux-ite developped on nepheline microsyenite in the Pogosde Caldas area, in the state of Minas Gerais. We hereprovide evidence for the mobilization of Zr duing alateritization process along a profile down to the bed-rock, and thus confirm the inference of Soubibs er al.(1991) on the basis of field observations. We bring newchemical, mass-balance and spectroscopic data to bearon the question of Zt mobilization. Furthermore, wereport on the crystal chemistry of the secondaryzirconiferous phases.

GBor-ocrcer- Sprrntc

The Pogos de Caldas complex is one of the largestalkaline bodies in the world, forming a large circularbody of almost 800 km2 (diameter 30-35 km, 1300 maltitude) at the boundary of the Minas Gerais and SdoPaulo states, southeastern Brasil (Fig. 1). Nephelinesyenites or microsyenites (locally known as "tinguaites")and phonolites, along with ankaratritic lavas and pyro-clastic rocks, are the main components of this UpperCretaceous intrusion.

The aluminous syenitic rocks have been intenselytransformed to bauxite as a consequence of the tropicalclimate. The Morro do Cristo deposit, where the profilePCMC-Z chosen for study is to be found, is locatedalong the northern border ofthe complex. The very gooddrainage, the high and irregular relief (1500-1700 maltitude) and the steep slopes of this area have led tointense and rapid weathering of the parent rock(tinguaite), resulting in the formation of a bauxiticmantle in abrupt contact with the fresh rock (Fig. 2).

DnscnprroN oF THE PRoFILE

Parent rock

The tinguaite of Morro do Cristo, a dark grey and vir-tually aphanitic rock, consists of a mosaic of fine-grainednepheline and alkali feldspar in which small prisms ofaegirine-augite (200 x 100 pm) are dispersed. Largercrystals (0.5-1 mm) of a complex fluorine-bearing Ti, Zsorosilicate of Ca and Na identified as hainite ( Atencio etal. 1999) and a Na-Ca amphibole (ferro-edenite) also aredispersed in the felsic groundmass. Hainite forms lonq

IHE MOBILITY OF Zr, POQOS DE CALDAS, BRAZIL 637

\ * *

: r y

\ Caldas

*$. JcfloBoa Vista

sP tT rt'

5

N

l0kn€\' *A*dtu* 0

Clastic sedimentary rocks

Foyaite, lujavrite and khibinites

poikilitic crystals, with high relief and birefringence toblue colors of the second order. Aegirine-augite com-prises about 207o of the rock by volume, whereas othersmafic minerals account for about 5Vo. Aegirine-augite ispresent in fine prisms of dark green color, and ferro-edenite forms larger, poikilitic crystals of dark brown

Tinguaites

Alkaline lavas and alkline basic lavas,associated pyroclast rocks

Diabase

color. Zircon forms minute euhedral crystals, but is ex-tremely rare. With an agpaitic index [(Na + K)/Al] ofabou/' 1.2, and a high content of Z, Nb and the rare-earthelements (REQ (Tables 1,2), this nepheline microsyeniteplots close to the limit between the agpaitic and miaskiticgroups (cl Sprensen 1974).

l-1

&I

E& Gneiss Complex and Fenite

mww

Frc. 1. General geology and location of the PoEos de Caldas alkaline igneous complex (after Ellert 1959, Bjiirnberg 1959)Diamond: study area.

THE MOBILITY OF Zr, POQOS DE CALDAS, BRAZIL 639

TABLE 1 MAIOR-ELEMENT COMPOSMIONS OF SELECTED SAMPLES TAKEN FROM TI{E BATDOTIC PROFILE (PCMC-2),POqOS DE CALDAS, MINAS GERAIS, BRAZIL

4 6 5 4 8 1 5

0 t 4 5 8 7 3

3 2 4 5 1 5 0

0 3 8 5 8 6 2

224 5231

0 4 1 5 ' 7 U

2 0 9 5 t 6 3

027 51 29

0 2 7 5 1 1 2

0 5 3 6 0 4 1

l f4 4465

2 2 6 5 3 2 5

0 4'7 58 54

0 3 5 5 1 M

0 3 0 5 5 2 9

0 3 5 5 5 9 2

0 3 5 5 6 5 6

I 1 9 5 s E 4

0,t0 56 20

55 E3 2090

7 6 0 0 0 5

6 4 1 0 0 5

8 5 2 0 0 6

1 5 0 0 0 3

1 0 0 9 0 0 7

7 2 9 0 0 4

8 7 8 0 0 6

7 8 5 0 0 4

8 4 5 0 0 s

5 1 1 0 0 1

1 9 9 5 0 5 4

l l 5 5 0 1 4

6 4 9 0 0 4

1 9 5 0 0 5

9 7 9 0 0 7

9 4 5 0 0 4

9 7 9 0 1 6

9't0 0 07

9 35 025

3 9 4 0 2 5

< 0 0 1 < 0 0 1 < 0 0 1 < 0 0 1 t 2 9

< 0 0 1 0 0 5 < 0 0 1 < 0 0 1 0 8 8

002 <0 0t <0 0t <0 0l l 38

< 0 0 1 0 0 6 < 0 0 1 < 0 0 1 0 9 8< 0 0 1 0 0 2 < 0 0 1 < 0 0 1 t 4 8

< 0 0 1 < 0 0 t < 0 0 1 < 0 0 t I 0 l< 0 0 1 < 0 0 1 < 0 0 1 < 0 0 1 1 3 1

< 0 0 1 < 0 0 t 0 0 2 0 0 3 t t 2

< 0 0 1 0 0 6 < 0 0 r 0 0 t l 1 9

<0 0t <0 0t <0 0l <0 0t 069<0 0l <0 0l <0 0l 0 33 295

< 0 0 1 < 0 0 t < 0 0 1 < 0 0 1 1 6 8

< 0 0 1 < 0 0 1 < 0 0 1 < 0 0 1 0 9 1< 0 0 1 < 0 0 1 < 0 0 t < 0 0 1 1 0 8

< 0 0 1 < 0 0 1 < 0 0 1 < 0 0 1 1 3 1

< 0 0 t < 0 0 t < 0 0 1 < 0 0 1 1 2 4

< 0 0 1 < 0 0 1 < 0 0 t < 0 0 1 1 3 7

< 0 0 1 < 0 0 1 < 0 0 1 < 0 0 1 1 3 3

< 0 0 t 0 0 3 < 0 0 1 < 0 0 1 t 2 4

Q 2 t t 4 s ' 1 6 1 ' t 5 4 0 5 1

0 t4 33 56 259 98 030 13 3 t 32 0 39 98320 13 31 39 233 98570 t f 3 1 8 7 0 3 9 9 9 9 40 17 3097 1 92 99290 t 5 3 t 7 2 0 5 9 9 t 5 l0 t 3 3 t i l 1 3 7 9 6 4 80 13 31 60 0 42 98'77008 3r 16 050 9949o l t 3215 043 9944026 2589 09'7 96680u 2903 091 99 t20 t6 3t 91 025 98 84022 31 43 0 32 9t 840 20 3013 0 44 9E 190 20 30'77 0 45 9t 4202s 3064 056 996t022 3046 067 9948023 3091 043 99040 18 091 004 9949

Compositions are quoted in wt% oxide Shown in parentheses in the column labeled "Sanrple olass ' is the mmb€r of sarnple analyzed (meanreported). * Pebbly bauxite: cosrse gibbsite-rich gravels (G) with clayey matrix (m) Massive barxite: mor€ coher€nt horizon in which theunderlying nodules are less sharply defined Nodular bauxite: concentric spheres (N) (drr) with highly porous rnstrix (M); see Figure 4. Isalt€riticbauxite: fnely porous gibbsite-rich mass @) cut by some fine (cm) and wggy veins (V); Br and B2 are two diferent sanrples collected d thecotrtact ofth€ fr€6h rock

TABLE 2- TRACE-ELEMENT CONCENIRATIONS Gpnr) IN SELECTED SAMPLES ($) TArGN FROM TIIE

($) Data are reported here for the same samples as in Table 1(r) ELREE: I light trFearth elenmts (La + Ce + Nd + Sm + ng;(++) EHREE: E heavy rare-earth el€ments (Dy + Er + Yb + r4

BATD(TIC PROFILE (PCMC-2), POCOS DE CALDAS, MINAS GERAIS, BRAAL

HorLotr DeDth (cml Semolcclrss Zr Nb Th Lr C€ Nd Sm Eu Dv Er Yb Y :LREEC) THNEC')

Pebbty b3uxite

(0 - 190 cn)

20 a (3) ' t3s 518 61 2 189 2 4 04 <05 3

575 629 91 8 686 6 2 04 <05 5

9a2 508 55 4 221 2 | 03 <05 3

104 649 l to t0 941 1 2 04 I 5

6 1 8 5 5 2 6 9 1 0 3 5 1 4 2 0 2 1 3

t@3 102 109 l i 876 I 4 03 I 5

649 594 76 a 514 5 1 03 I l4 3

979 686 i lo r0 969 8 13 03 I 5

1389 610 10 5 382 4 5 02 36 1

2502 643 85 5 492 4 2 03 I 4

535 37t 43 8 222 3 3 <01 I 3

5718 t421 586 45 6962 32 l0 2 5 12

J232 937 179 19 1866 9 3 I 3 ',l

1222 534 48 5 4E 2 I 02 <05 3

2320 696 16 5 342 4 | 03 | 4

2510 824 14 2 138 2 2 0l I 5

2609 190 89 4 430 2 <X5 04 <05 5

l5t3 841 82 3 146 3 3 03 <05 5

1292 at5 74 3 163 4 3 03 <05 5

1799 750 81 5 498 4 3 02 <05 5

t051 296 16 234 344 109 l8 4 9 7

I

2

I

2

1

2

I

2

I

I

7

3

I

I

I

I

I

4

I

1

I

6

I

6

3

1

2

2

I

20

l 0

I

2

I

z

2

2

2

397

102

234

960

361

899

548

I 000

396

503

236

70sl

1 8 9 8

56

352

144

436

1 5 5

1 7 3

5 1 0

109

5

l 4

5

t 4

8

l l

l 2 r

l 5

42

8

6

44

23

5

8

8

I

8

8

8

50 a (2)

t 0 0

m

1 5 0 c

m

210 (3)

rxtrstfioml lev€l O30 -330 cm) 100 N

Noduhr bruxite

(350 - 500 cn)

t50 N

M

400 M

450 N (2)

M

Isalteritic bruxih

(500 - 600 cm)

500 B

600 B I

B2

FBh rock > 600

642 THE CANADIAN MINERALOGIST

dilution for major elements and Zr, and after a mixedHF-HCI-HNO3 digestion for the other trace elements.Bulk densities were estimated with the method usingparaffin as coating solution. Means of three determina-tions of three different samples were used in the mass-balance calculations (Table 3).

Mineral analyses were carried out using a CAMECASX-50 electron microprobe fitted with three wave-length-dispersion spectrometers (Universit6 PaulSabatier, Toulouse) on carbon-coated thin sections.Operating conditions were: accelerating voltage 15 kV,beam current 10 nA, electron beam 10 pm across, withl0 s counting time for each element, and 5 s for back-ground.

Powder X-ray-diffraction patterns for each samplewere collected in reflection mode using a Philips PW1130 diffractometer with Ni-filtered CuKct radiation(30 kV, 30 mA). Recording conditions were of 3.5 s bystep of 0.0005' (20) from 2" to 80" (20). The percent-age of each mineral constituent was evaluated using themethod of Rietveld (1969) from X-ray powder data (4 sby step of 0.02"). This method also permitted to refinethe _structure of the goethite present.

)/Fe Mcjssbauer absorption spectra (Laboratoire deChimie de Coordination, Toulouse) were recorded overthe range +14 mm/s in 512 channels to characterize theFe-bearing phases. The M<issbauer spectrometer is com-posed of a compact "y-detector system for high count-rates and a conventional constant-acceleration Mtiss-bauer Wisel device. A sTCo (in Rh) source with nominalactivity of 50 mCi was used. Spectra were collected attemperatures of 298 and 70 K, and recorded on aCanberra multichannel analyzer,linked to a computer.The isomer shift was recorded with respect to ct-Femetal. The absorbing sample was prepared with lessthan 10 mg Fe/cm2, to avoid the effect of absorber thick-ness. Lorentzian l ine-shapes were assumed fordeconvolution, using least-square fitting procedures.The 1' and misfit values were calculated in order toevaluate the goodness of the computer fit. For this analy-sis, samples were finely ground in acetone (to minimizepossible oxidation of Fe).

To obtain some information on the crystallographicsite of Zr in the ferriferous phases of the bauxite, we rana number of absorption spectra at the LURE synchro-tron facility (Orsay, France), on the D44 station. TheDCI storage ring was operated at 1.85 GeV and246mA.A Si (311) dguble crygtal was used to monochromatizethe X-ray beam. EXAFS (X-ray absorption fine struc-ture) spectra at the Zr K edge were collected in trans-mission mode at room temperature with energy steps of2 eV over the energy range 17 .9-18.9 keV, and with anacquisition time of 2 seconds per step. XANES spectraat the Zr K edge were collected with energy steps of 1eV over the energy runge 17.95-18.1 keV, and with anacquisition time of I second by step. Spectra were cali-brated by reference to metallic Zt (Zr K edge at 17998eV). Samples were finely ground, and the resulting pow-

ders were sandwiched between two pieces of kaptontape. Sample absorbance was controlled by adjustingsample thickness.

Rnsulrs

M a s s - balanc e c alc ulati ons

Two independent methods are generally used to es-timate element mobility during weathering processes:the "isovolume" and the "isoelement" methods (Braunet al. 1993). In the first case, if the volume of a horizoncan be considered unaffected by the weathering pro-cesses, relative gains and losses in the concentration ofany element, compared with its concentration in theparent rock, are given by: Vo change = [(x.p / x6.ps) - 1]* 100, where x and x6 are the element concentrations,and p and p6 the bulk densities, of the weathered mate-rial and of the parent rock, respectively (Millot &Bonifas 1955).

If. on the other hand. evidence exists that the con-centration of any given element did not change, i.e., theelement was immobile during weathering, the percent-age increase or decrease of any other element in theweathered rock relative to the fresh rock can be calcu-lated (Nesbitt 1979): Vo change = lt6ll) I (xo/io)l - 1l x

100, where x and xe refer to the same parameters asabove, and i and io are the concentrations of the immo-bile element in the weathered and in the fresh rock, re-spectively. Cramer & Nesbitt (1983) pointed out thatowing to uncertainties in sample representativity, den-sity measurements and results of chemical analyses,only changes in concentration greater than +20Vo canbe considered significant in these calculations.

In the present study, both methods of calculationwere used to monitor chemical changes in the PCMC-2 profile, with the particular objective to characterizethe behavior of Zr during bauxite formation. The con-centrations of the major and selected trace elements usedfor these calculations are reported in Tables I and2.

The isovolume method (Table 3) was only appliedto the lowermost meter of the bauxite (isalteritic hori-zon, Fig. 3), where large- and small-scale structural andtextural features (e.9., mineral distribution and shape,organization of fractures) of the parent rock are verywell preserved (Fig. 2). With these calculations, welooked for an immobile element that could be used withthe isoelement method to estimatg Zr rnobility duringthe first stage of weathering, as well as in those parts ofthe profile where textural evidence for volume conser-vation were not found. Data for vuggy veins of theisalteritic bauxite were not taken into consideration be-cause their parent rock (hydrothermally transformedtinguaite) was not analyzed, and because they representonly a small percentage of the profile.

Focusing onZr,Ti, Nb and Th, commonly consid-ered as immobile in secondary environments, Table 3shows that: i) Zr experiences a surprisingly large

643THE MOBILITY OF Zr, POQOS DE CALDAS, BRAZIL

q g q

O O @q n ' ?

@ : =

\ o 6 -

Y A T

€ r6 N

€ ( { 9€ r €

N N

6 6 O

*

'.1

z

N

F

Fza&

A

( j r I J' I

z=

cts+ N

9 AX E I

E rrt

& '

a

$

rl

F

@o

, k

()g

ooe.l

c.l

q

6' ro

o

@

o

{)

E()

r s r

i:

v

z

U

6

r-1Fza

OZ s )

S F

' |

' l r r

v t A17. ai ; <

(r)

mrrl

F

o

N - N

6 i l +

" Y T

N

d

T i

.1H

F

z

N

r t l

t >a '

x t rE e =. 2 Eo =

? 9

X E

q =

= h

644 THE CANADIAN MINERALOGIST

decrease at the contact with the fresh rock; ii) as ex-pected, Ti and Nb accumulate in the weathered materialwith respect to the fresh rock; iii) Th seems to be theleast mobile element in the saprolite, as already notedby Braun et al. (1993) in similar environments. Obvi-ously, these results are only valid insofar as the distri-bution of these elements was homogeneous in the parentrock. Our observations indicate that this is the case forthe fresh tinguaite sampled within a few hundred metersof the profile.

Using Th as an invariable element, we can apply thesecond method of mass-balance calculation to the wholeprofile, in particular, to the upper parts, where the origi-nal structure and texture of the tinguaite are least pre-served (Tables 4, 5). Considering the structuralheterogeneity of the upper levels and the limited num-ber of analyses done, only the most obvious trends inTables 3 to 5 will be considered.

Si, Na, Ca, K and Mg

As is always observed in cases of strong ferralliticweathering, these elements are almost entirely leachedin the first centimeters away from the fresh rock.

Zr

In the lower part of the profile (below 500 cm), bothmethods of mass-balance calculation showed a 30--40%loss in Zr (Tables 3, 4). Therefore, there is a little doubt

that at Morro do Cristo, an important proportion (up to40Vo) of the Zr content of the parent rock is leachedduring the early stages ofweathering. Zr leaching is stillmore effective in the upper (and older) parts of the pro-file (above 190 cm), withZr losses up to70-80%o fromthe massive bauxite to the surface. A loss also is evi-dent in the uppermost part of the nodular bauxitic hori-zon (400-350 cm). Sandwiched between these threelevels of high degree of Zr leaching, two levels ofweaker Zr mobllization occur Oable 5). These can beexplained as 1) levels ofabsolute accumulation, wherebyZr was supplied by the leaching of the over$ing hori-zons,2) the expression of some degree of parent-rockheterogeneity or, alternatively 3) the consequence of achange that could have occurred in the drainage condi-tions of the area represented by the PCMC-2 profile.

The mobility of Zt in the Morro do Cristo profile isamong the highest ever reported in the literature. Thishigh degree of mobility could be related to the fact thatthe principal Zr-bearing primary mineral is not zirconbut the Na{a silicate hainite, in which Zr is octahe-drally coordinated and, as a result, more easily removedthan in the zircon structure. Furthermore, there is anabundance of fluorine in this mineral (-7-8%), fluorinebeing known to be a very efficient ligand for Zr in solu-tion. Nevertheless, we have still no clear evidence thatZr is transported in a dissolved form in the area studied;a colloidal form, which could be a suspension ofzirconiferous particles neoformed in the bauxite (seebelow), should be also considered.

TABLE 5 ISOTHORIUMMASS-BALANCE e/o) AI-TONGTHE ENTIREPCMC-2 PROFILE,FO9OS DE CALDAS, MINAS GERAIS, BRAZIL

Horizon Depth {cm) Semple cl&ss Si Tt Al Fe Mn Mg Ca Nr K Ti P Zr 'HREE

Pebbty be[xite

(0 - 190 cn)

20 c (3) -70 -96

-6t -91

-18 -95

-56 -93

-11 -93

-58 -93

-63 -8

-54 -92

-71 -57

-70 -95

-72 -92

-39 -96

-46 .92

-94 -94

-76 -94

-90 -94

-90 -94

-88 -94

t00 -2 49 -3 -96 -100 -97 -100 -100 -2 -58 {3 -6

-100 0 -9 -24 -92 -100 -100 -100 -100 0 -69 -78 -16

-t00 t3 84 7 -87 .r00 -98 -100 -100 13 -53 -39 12

-98 - l | -19 -29 -92 -100 -100 -100 -100 - l I -16 -18 -28

-r00 o 46 -l -94 -100 -98 -100 100 0 -68 -69 -3

-100 4 -11 .15 -91 -100 -99 -100 -100 4 -69 -69 -22

. l0o -6 30 -t2 -92 -100 -100 -100 -100 -6 -60 -71 -5

-99 -16 -19 -27 -92 -100 -100 -100 -100 -16 -16 -70 -24

-l0o 13 41 2 -92 -tO() -99 -100 -100 13 -63 -32 6

Joo -t7 16 -9 -91 -100 -98 -100 -100 - l -81 0 -8

-99 t3 t42 9 -9? -100 -100 -100 -100 13 49 -58 7

-100 -64 -87 49 -87 -100 -100 -100 -100 {4 -91 -66 -10

-gg -34 49 41 -89 -100 -100 -100 -100 -34 - '71 -38 -36

-gg 34 l l0 23 -88 -100 -100 -100 -100 34 -33 -13 35

-100 0 30 -4 -90 -100 -100 -100 -100 0 -42 4 l l

-100 31 29 2l -86 -100 -100 n00 -100 1I 46 18 15

-100 l8 19 9 .?2 .100 -100 -100 -100 18 -39 -34 25

.99 Z7 30 20 -86 "100 -100 -100 -100 21 40 40 34

50 c (2)

t00 G

h

t 5 0 c

Massivc 230 210 (3

t00 N

Noduhr bruxite

(350 - 500 cm)

150 N

M

400 M

450 N (2)

M

ls*lteritic bauxite

(500 - 600 cn)

500 B

600 B 1

B2

Data are reported here for the me samples as fur Table 1,(*) ELREE: X light rare-earth elunents (L + Ce + Nd + Sm + Eu)(*i) XHREE: X heavy rare-earth elements (Dy + Er + Yb + Y)

THE MOBILITY OF Zr, POQOS DE CALDAS, BRAZIL 645

Ti. Nb. Fe

These elements can broadly be considered as moreor less immobile along the profile (weak leaching orabsolute accumulation). An exception is the highly po-rous matrix material found in the intermediate part ofthe nodular bauxitic horizon (400-350 cm), where theseelements are strongly leached, along with Zr and Al (seebelow). It is likely that part of the leached elements aretrapped in the nodules, for which calculations invari-ably show a gain in mass.

AI

Al seems to be more mobile than Fe, Ti, and Nb, atleast in the median and upper horizons, where elevatedabsolute accumulations occur in some gibbsite-richnodular blocks (balance up to +100 Vo) and where ma-trices are systematically depleted in this element.

The rare earths

The rare earths are strongly leached along the wholeprofile, in particular the heavy rare earths. This processof leaching and fractionation is by and large similar tothat observed by Braun et al. (1993) in their study of alateritic profile derived from a syenite in Cameroon, andwill not be discussed further in this paper.

Mineralogy of the bauxite

The bauxite of Morro do Cristo appears to be essen-tially composed of gibbsite and goethite (Fig. 3). Ana-tase, kaolinite, quartz and rutile were found as accessoryrninerals in some horizons.

Zr distribution in the bauxite

In order to assess the distribution ofZr rn the baux-ite, in the absence of a macroscopic Zr-bearing phase(Fig. 3), electron-microprobe analyses were performedon secondary phases and relics in each horizon of theprofrle. The analyses revealed the presence of l-2%o Zrin pore-filling products and in the surrounding orangegranules. The orange plasma, the most ferriferous ofthese pore-filling products, can contain tp to 8% ZrO2.All analyses of these ferriferous products yielded lowtotals (about 707o), indicating that the material is highlyporous or hydrated (or both). Results of semiquantitativeelectron-microprobe analyses show that Zr is well cor-related with Fe and Ti (Fig. 6), suggesting that Zr isassociated with one or both of these elements. A similarcorrelation was found in chemical analyses of bulksamples of bauxite, with correlation coefficients of0.917 with Fe and 0.899 with Ti (SoubiEs et al. 1991).As Fe is a major component of these materials and theonly one of the two elements to be present along the

fcr$* {W*}

taorcdp*amun

nra,.|g;glw,h*

*f!*

0{s r t "5z tsstrr$! $e.)

Frc. 6. Zr-Fe correlation, microprobe data.

whole profile (Fig. 3), we focussed on goethite, the onlyFe-bearing phase found in the profile.

INvnsrrceloNs oF THE Cnvsrar CuelrlsrnvOF ZT IN THE SECONDARY PHASES

Mri s sbaue r sp e ctro s c opy of goethite

Goethite is a magnetically ordered mineral. The dif-ferences in Mtissbauer spectra of natural samples ofgoethite depend on their chemical composition and thedimensions of the crystalline domains (Johnston &Norrish 1981). Femrginous material from each bauxitichorizon of the Morro do Cristo profile has been ana-lyzed by Mcissbauer spectroscopy. The resulting spec-tra show a supelparamagnetic couple at 298 K, and asextuplet at 70 K with field value of 44 T and broad-ened lines. These data indicate that the Fe in the goe-thite is commonly substituted by a diamagnetic element.Al is the most likely candidate, as it commonly substi-tutes for Fe in goethite (Schulze 1984). Therefore, thevariation of the cell parameters determined by Rietveldrefinement of X-ray diffractograms collected from thesame materials allowed us to calculate the extent oftrAl-for-Fe substitution on the basis of Schwertzmann's

$N

646 THE CANADIAN MINERALOGIST

model (1984). About 30 mole Vo of the componentAIOOH has been found in all samples of goethite; thisvalue had been confirmed by Mcissbauer spectroscopyanalyses (Stucki 1985). Nevertheless, since Zr also is adiamagnetic element, it too could be incorporated in thestructure, as suggested by Soubibs et al. (1991),becausethe difference in ionic radius between Fe3* and both Aland Zr in octahedral coordination is similar. To shedsome light on the crystal chemistry of Zr in these mate-rials, an X-ray absorption spectroscopy study was car-ried out.

XAFS spectroscopy on Zr K-edge

Some ferriferous pore-filling material taken from thebauxite of Morro do Cristo (PCMC-2) was analyzed byXAFS transmission spectroscopy.

Model compounds.for XAFS data collection

Zr-bearing compounds having well-refined struc-tures were chosen as models of characteristic environ-ments for Zr in minerals. Nonmetamict zicon,ZrSiOa,is a model for atslZr environment; t71Zr environmentsoccur in baddeleyite, ZrO2, zirkelite, (Ca,Th)ZrTizOt,and in aqueous complexes ofZr.Zr in sixfold coordina-tion is most common inZr-beaing alkali silicates suchas catapleiite, NazZrSi:On.3H2O. All these referencesspectra were provided courtesy ofFrangois Farges. Thebaddeleyite used in the present study is synthetic re-agent-grade monoclinic zirconia, 99Vo pve (Farges eral. 1994). The zircon used is synthetic also, but its spec-trum is like that of natural non-metamict zircon (Farges& Calas 1990). The zirkelite comes from Sri Lanka; itwas originally metamict, but was recrystallized by heat-ing (Farges et al.1993).

Resr,r-rs

Zr X-ray absorption spectoscopy near-edgestructure (XANES)

Zr K-edge XANES spectra (Fig. 7) for the selectedmodel compounds and PCMC-2 show slight but sig-nificant differences, depending on the coordination en-vironment of Zr. For the PCMC-2 soectrum. theabsorption edge is in the same location and quite simi-lar to the one of baddeleyite and zirkelite, which corre-sponds to a UlZr coordination. The similarities in theXANES spectra of PCMC-2, baddeleyite, and zirkelitealso suggest that the individual Zr-O distances are simi-lar.

Zr X-ray absorption spectroscopy.fine structure (XAFS)

All XAFS spectra have been fitted between 0.9 andl0 A-l since spectrum PCMC-2 is too noisy over this

179s0 18000 1 8 0 s 0 1 8 1 0 0

ENERGY (eV)

FIG. 7 Normalized XANES spectra on Zr K-threshold ofMorro do Cristo sample (PCMC-2), andZr-containing ref-erence compounds (baddeleyite, catapleiite, zirkelite, zir-con, aqueous Zr)

range. XAFS spectra for PCMC-2, baddeleyite andzirkelite are presented in Figure 8. Fourier transforms(FT) of the k3-weighted XAFS spectrum at the Zr K-edge (Fig. 9) for the reference compounds zircon,baddeleyite, aqueous Zr, catapleiite, and zirkelite are ingood agreement with previous crystal-structure refine-ments (Farges & Calas 1991, Farges et al. 1993,1994).The spectra depend essentially on the coordination num-ber, but some disorder interaction in the structure mayexist. Two groups of FT are distinguished dependingon the intensity of their first peak. One group of FT iscomposed of zircon, catapleiite and aqueous Zr becauseoftheir high coordination number, or the high degree ofcrystallinity. The first peak of the FT of the zircon spec-trum corresponds to t81Zr, four O atoms at 2.13 A andfour others at 2.27 A. The first peak of the FT of theaqueous Zr spectrum corresponds bulZr, which is verywell organized at short-range order, and the first peakof the FT of the catapleiite spectrum correspon ds bl6lZrin which the Z106 polyhedra are regular (all Zr-O dis-

Uk

n<Ca

-N-

q

E

THE MOBILITY OF zr. POEOS DE CALDAS. BRAZIL 647

XzAE

FUzJhOFl

Xr-'l-

baddeleyite

O PCMC.2

o zirkelite

r {4"'}

tances are equal). The other group ofFT is composed ofbaddeleyite (a seven-coordinated Zr mineral with weakorganization at short range) and zirkelite, in which thefirst peak corresponds to seven atoms of oxygen.

The FT of PCMC-2 is in good agreement with thesecond group of FT, containing baddeleyite andzirkelite. The frst peak for these two compounds corre-sponds to the first shell ofZr neighbors, which are sevenatoms of oxygen (Figs 8, 9). The second peak corre-sponds to the second shell ofneighbors. For baddeleyite,it corresponds to only Zr, and for zirkelite, to Ca, Ti andFe neighbors.

O neighbors

The extracted edge-energy (Es) values for PCMC-2, baddeleyite, and zirkelite are in good agreement withthe one measured in catapleiite. The low absolute value

b

R (A)1 2l o

Frc. 8. Fourier transforms (FT) of the k3-weighted XAFS spectra at the Zr K-edge forPCMC-2, zirkelite and baddeleyite. Peaks on the Zr-X pair correlation functions arelabeled to identify X.

for AE6 (lAEgl < 1.6 eV) suggests that the empiricalamplitude and phase-shift functions extracted fromcatapleiite should provide a good calibration of theXAFS structural parameters. Because of its very well-determined structure, we used catapleiite as referenceto fit XAFS spectra. Zr in cataplelite is known-to be ina regular polyhedron of six O atoms at 2.073 A.

Inverse Fourier transforms and least-squares fits ofthe O first-neighbor contribution to the Zr XAFS of thebaddeleyite with two sheets show that the first peak inthe FT corresponds to Zr-O conelations of an averageof T.3Oatomsatanaveragedistance of 2.17 A (Farges& Calas 1991, Farges et al. 1993,1994). This result isin excellent agrgement with structure-refinement data(Smith & Newkirk 1965) and previous XANES data.

Inverse Fourier transforms and least-squares fits ofthe O first-neighbor contribution to the Zr XAFS ofPCMC-2 with two sheets show that the first peak in the

648

r-'l-l

A-JH-zr l r

I

Er-

0.7

0 .6

0 .5

0 .4

0 .3

0 .2

0 .1

FT corresponds to Zr-O correlations of an average of6.6 O atoms at an averase distance of 2.16 A (Debve-Waller-type factor Ao2 =Z.A x 10-3 A). This resuli inexcellent agreement with previous XANES results, isquite similar to that of baddeleyite. Zr in PCMC-2seems to have an atomic environment similar to that forbaddeleyite.

Second neighbors

The second shell of coordination in baddeleyite iscomposed of seven atoms of Zr centered at 3.47 A,whereas in zirkelite, it consists of one Ca atom at 3.24A, three Ti atoms at3.25,3.30 and 3.35 A, two-Fe at-oms at 3.40 and 3.4'7 A, one Ti atom at 3.53 A, andthree Ca atoms at 3.67 L. Least-squares fit of the FTXAFS spectrum of baddeleyite and zirkelite are pre-sented Figure 10. It seems that where the second neigh-bor is Zr (as in baddeleyite), the FT of the XAFS signalof its contribution shows a maximum around 5 A-l;where the second neighbor ofZr is "light" (as in zirkel-ite), the maximum of the FT is near the lower k(A-l)range.

The FT of PCMC-2 shows a second-^neighbor con-tribution at a unrefined distance of 2.95 A, but becauseof the low signal-to-noise ratio, information about themiddle distance (10 to 15 A-1) is lost, and so is impos-sible to analyze quantitatively. However, qualitativeanalysis of the inverse FT for this RDF (radial distribu-tion function) feature suggests the presence of a "lightelement", as in the case for zirkelite. The PCMC-2sample being essentially composed of Fe (goethite), wecan suppose that the zirconium's second-neighbor is Fe.The distance Zr-Fe of 2.95 A must be corrected byphase pair (Zr-Fe) factor Q7r-p". But we already canestimate Qzr-F" = 2.3 A, and the distance Zr-Fe wouldbe about 3.3 A (+0.1 Al. Wittr the distances Zr-O

Q.l6 b and the distance vIFe3*-O in goethite (2.04 A),we calculated the value of the angle Zr-O-Fe to beabout 105o. This value is typical of shared edges.

Our XANES and XAFS results show that Zr is lo-cated in sevenfold-coordinated sites in the ferriferousphase of pore-filling in bauxite (PCMC-2) from Pogosde Caldas. This result leads us to discard the hypothesisof Zr incorporation in the goethite structure, where itwould be in the sixfold-coordinated site of Fe. Zr-O

0 l0

R (A)

FIG. 9. Superposition of the Fourier transforms (FT) of the k3-weighted XAFS spectra at the Zr K- edge for PCMC-2, andreference compounds.

a

baddeleyitezirconcatapleiitezirkeliteaqueous zrPCMC-2

THE MOBILITY OF Zr, POQOS DE CALDAS' BRAZIL

-0.0256

k (A-')

Frc. 10. Superposition of least-squares fit of the Fourier-filtered XAFS spectrum ofbaddeleyite, zirkelite and PCMC-2 sample at the Zr K edge. The fit for these com-oounds includes conffibutions from O nearest neishbors and second neishbors

649

X

distances of 2.16 A are similar to those in baddeleyite,but the second neighbors are probably Fe atoms, withZr-Fe distances estimated at 3.3 A. If the Fe is fromgoethite, it is possible to calculate the angle Zr-O-Fe tobe about 105', which is typical of an environment ofedge- sharing. These results suggest that Zr is present inthe goethitic plasma, and is located within a polyhedronof seven O atoms (withZr-O distances about 2.16 A)edge-sharing with Fe-bearing octahedra ofthe goethite.

CoNcLusroNs

In the Pogos de Caldas area, Minas Gerais, Brazil, aprofile across a thick section of bauxite developed on anepheline microsyenite was examined in detail in orderto investigate the behavior of Zr during supergeneweathering. Field and microscopic observations suggestthat this element was mobile. Evidence includes the to-tal disappearance, during the first stages of weathering,of the primary Zr-bearing minerals (hainite and amphib-ole), and the occunence of Zr in neoformed femrginousproducts (goethite-rich plasma and granules) that do notreplace primary minerals. In addition, mass-balancecalculations demonstrate that Zr was leached from allhorizons ofthe profile. The extent ofleaching was cal-culated to reach 40Vo ZrO2 at the weathering front.

A crystal-chemical study of the Zr-bearing femrgi-nous material revealedthatZr does not substitute for Fein the goethite (the only ubiquitous Fe-bearing phase),as proposed by Soubids et al. (1991), but rather that it ispresent within polyhedra of 6.6 O atoms on average,with average Zr-O distances of 2.16 A. This baddel-eyite-like first shell of neighbors probably shares edgeswith Fe3*-bearing octahedra. This arrangement impliesa radical change in the structure of theZr-beaing phase,which in tum is yet another line of evidence that thiscation was transported during the weathering process.Some improvements are needed to identify thisZr-bear-ing phase produced by the supergene weathering ofhainite, and more geochemical data are needed to un-derstand the role of F released from the hainite in thetransport of Zr.

ACKNOWLEDGEMENTS

The staff of LURE is acknowledged for its assistancein the XAS measurements. We also thank PhilippeIldefonse for his XAS assistance, and Philippe deParseval for his electron-microprobe analyses. Refer-ences were graciously provided by F. Farges. We thanksincerely F. Farges for his help, his constructive reviewand many discussions. We also are grateful to J.

650 THE CANADIAN MINERALOGIST

Delvigne and R.F. Martin for their constructive com-ments. Field work and analytical costs were underwrit-ten by a grant from FAPESP.

Rerensxces

Annts, J.A.S. & RTCHARDSoN, F A. (1960): Thorium, uraniumand zirconium concentration in bauxite. Econ. Geol.55,1653-16'.75.

ArENcro, D., CourrNHo, J.M V , ULBRrcr, M.N C , Vlecn,S.R.F, RasrsvBTAEVA, R.K. & PusncnanovsKy, D.Yu.(i999): Hainite from Pogos de Caldas, Minas Gerais, Bra-zll. Can. Mineral. 37 ,9I-98.

BrNrr-nvsrrr , S. I . (1963): The Mineralogy of Bauxi tes.Gosgeol. Tekhizdat, Moscow, Russia (in Russ.)

BBnnow, M.L., WrsoN, M.J. & REAvEs, G.A. (1978): Origrnof extractable titanium and vanadium in the A horizon ofScottish podzols Geoderma 21, 89-103.

BronNnsnc, A.J.S. (1959): Rochas cldsticas do planalto dePogos de Caldas Bol. Fac. Filos. Ci€ncias Letras (Jruv.Sdo Paulo 237 , Geologia 18, 64-123.

BneuN, J J, Pe,crr-, M., HEnuu-oN, A & RosrN, C. (1993):Mobilization and redistribution of REEs and thorium in asyenitic lateritic profile: a mass balance sttdy. Geochim.Cosmochim. Acta 57,4419- 4434.

Bmuuelr-, G H., LBwrs, C.J , FoRD, C., Burr, J., TeyLoR, G& W.qnIN, O. (1991): Quantitative geochemical approachto pedogenesis: importance of parent material reduction,volumetric expansion, and eolian influx in lateritizationGeodetma 51,51-91

BRooKrNS, D.G. (1988): Eh-pH Diagrams for Geochemistry.Springer-Verlag, New York, N.Y.

CennorL, D. (1953): Weatherability of zircon J. Sed. Petrol.23,106-116.

Cor-nv, F., ALARqoN, C. & Vmlrann, P. (1993): Zircon: animmobile index in soils? Chem. Geol 107,273-2'76.

Cn-c.]!rpn, J.J. & Npssnr, H.W (1983): Mass balance relarionsand trace element mobility during continental weatheringof various igneous rocks. Sci G6ol. (Strasbourg), Mdm.73,63-'.73.

DEGENHARDT, H. (1957): Untersuchungen zur geochemischenVerteilung des Zirkoniums in der LithosphZire. GeochimCo smochim. Ac ta Il. 2'7 9-309

ELLERT, R (1959): ContribuigSo h geologia do macigo alcalinode Pogos de Cald,as. Bol Fac. Filos. CiAncias Letras Univ.Sdo Paulo 237, Geologia 18, 5-63.

EnLeNr, A J., Sunn, H S., MencslNr, J.W., Cnnnoso, M.P.& AHRENS, L.H. (1978): Zirconium. In Handbook ofGeochemistry, vol. II-4, sect. 40 (K.H. Wedepohl, ed.).Springer-Verlag, Berlin, Germany.

FencBs, F., BnowN, G.E., Jn. & Vsr-on, D (1994): Structuralenvironment of Zr in two inosilicates from Cameroon: min-eralogical and geochemical implications. Az. Mineral 79,838-847

& C,qles, G. (1991): Structural analysis ofradiationdamage in zircon and thorite: an X- ray absorption studyAnr. Mineral 76.60-73.

Ewnc, R.C & BnowN, G.E., Jn. (1993): The structure of metamict, aperiodic (Ca,Th)2Zr2Ti2Ou, J. Mater.Res. 8, 1983-1995

Fnevssr.{Er, P. (1994): Gold mass balance in lateritic profilesfrom savanna and rain forest zones. Catena 21. 159-172.

Gor-lscnaror, V.M. (1937): The principles of the distributionof chemical elements in minerals and rocks. J. Chem. Soc.London 1,655-673

(1954): Geochemistry. Clarcr'don Press, Oxford,U.K.

GoRDoN, M., JR. & MuRArA, K.J. (1952): Minor elements inArkansas ba.uxite. Econ. Geol.47. 169-119.

JoHlN, Z. & ircn, F. (1989): Nouvelles donn6es sur la harnite,INa2Caa(Ti,Zr,Mn,Fe,Nb,Ta)1.5s[16,5s](Si207)2Fa et ses re-lations cristallochimiques avec la g<itzenite, Na2CasTi(Si2O?)2F4. C.R. Acad. Sci. Paris 308, slr ll,1237-1242.

JoHNsroN, J.H & Nonnrsu, K. (1981): A siFe Mijssbauerspectroscopic study of a selection of Australian and othergoethites Aust. J. Soil Res.19,23I-237.

Krrraune, H.S. & Swr.tleI-r,L.D. (196'7): A discriminant func-tion using zirconium and nickel for parent rocks of stronglyweathered Hawaiian solls. Soil Science 104.69-76.

LnvnsoN, A A. (1980): Introduction to Exploration Geochem-ist,) (second ed.). Applied Publ. Ltd , Calgary, Alberta.

MILLor, G. & BoNnc.s, M. (1955): Transformations isovolu-m6triques dans les ph6nombnes de lat6ritisation etbauxitisation. Bull. Sen Carte Gdol Alsace-Lorraine 8.3-20.

Mrlxss, A.R. & FrrzpernlcK, R.W. (1989): Titanium and zir-conium minerals. In Minerals in Soils Environments (J.B.

Dixon & S.B. Weed, eds.; second edition) Soil Sci. Am.,Madison, Wisconsin (1 13 1-1 205).

NESBITT, H.W. (1979): Mobility and fractionation of rare-earthelements during weathering of a granodiorite Nature 279,206-2t0.

Rmrvnn, H.M. (1969): A profile refinement method for nu-clear and magnetic structures. J. Appl. Crystallogr.2,65-71.

RoNov, A.B., VlrNsnrBrN, EE. & TuzovA, A.M. (1961):Geochemistry of hafnium, zirconium and some otherhydrolyzate elements in clays. Geochemistry 41343-355.

THE MOBILITY OF Zr, POQOS DE CALDAS' BRAZIL 651

ScHULzE, D.G. (1984): The influence of aluminium on ironoxides. VIII. Unit-cell dimensions of Al-substitutedgoethite and estimation of Al from them - Clays CIay Min-erals 32,36-44.

Scnwnnrzrrremt, U. (1984): The influence of aluminium oniron oxides. IX. Dissolution of Al-soethites in 6M HCl.CIay Mineral. 19,9-19,

Surrn, D.K. & Npwxm, H.W. (1965): The crystal sfiuctureof baddeleyite (monoclinic ZrO) and its relations to thepolymorphism of ZrO2. Acta Crystallogr. 18,983-991.

SonBNsBN, H. (1974): The Alkaline Rocks John Wiley & Sons,New York, N.Y.

Sousms, F. , Mer-Fr, A.J., DILVIGNE, J & SARDU-I, I.A. (1991):Mobilidade do zirc6nio na bauxitisagdo de rochas alcalinasdo maciqo de PoEos de Caldas, Minas Gerais. RevislaBrasileira de Geoci4ncias 2l(L), 17-22.

Srucrr, J.W. (1985): hon in soils and clay minerals. NATO Adv.Study Inst., Ser. 2I7. Reidel, Dordrecht, The Netherlands

SwrNo*l, L.D. & JecrsoN, M.L. (1956): Genetic processes

in some residual podzolised soils ofNew Zealand.Trans.Sixth Int. Congress SoiI Science (Paris) 37,233-239.

TBreN-Krlr-e, M.S. Fnzp.lrnrcK, R.W. & CHnrLssonoucH,D.J. (1991): Scanning electron microscope study ofzircons

and rutiles from podzol chronosequence at Cooloola,

Queensland, Australia. Catena t8, ll-3o.

ZAKRUTKTN, V.E. & Snvprswl, I.V. (1974): Zircon in the

bauxites ofthe southem Timan region. Litologya i Polznye

Iskopaemye 5,32-40.

ZErssINK, H.E. (1971): Trace element behaviour in two

nickeliferous lateritic profiles . Chem. Geol. 7 ,25-36.

Received July 29, 1998, revised manuscript accepted March

15,1999.