322 Eart h a n d Pl ane t ar y Sc i e nc e Letto:~.
78 (1986) 322 326
Elsevier Science Publ i sher s B.V., Ams t e r d a m - Pr i nt ed in The Ne t he r l a nds [31 The magnetic soils of Brazil Ma u r o Re s e n d e ~, J a c que l i ne Al l an 2 a nd J . M. D. Co e y 2 J De par t me nt o f S o i l Sci ence, Federal UniversiO, o f Vi~osa, Vigosa. MG ( Brazi l ) -" De par t me nt o f Pur e a n d Ap p l i e d PIwsi cs, Tr i ni t y College, Dubl i n 2 ( I rel and) Recei ved Augus t 29, 1985; revi sed ver si on accept ed Febr uar y 25, 1986 Cer t ai n Brazi l i an l at osol s show a s pont a ne ous magnet i zat i on of t he cl ay size fract i on in excess of 1 J / T kg: t hey can be pi cked up by a ha nd magnet . Magnet i zat i on is a good i ndi cat or of soil type, bei ng gr eat est in t hos e soils deri ved f r om maf i c or itabiritic rocks whi ch have been little i nf l uenced by xant hi zat i on (t he soil yel l owi ng process). The magnet i c phas e is ful l y-oxi di zed t i t anomaghemi t e whi ch is i nheri t ed di rect l y f r om weat her ed bedrock. Si mpl e magnet i c me a s ur e me nt s are r e c omme nde d as an ai d in cl assi f yi ng t ropi cal l at osol s. Tropi cal latosols (also known as Oxisols or Ferralsols) are composed mostly of quartz, kaolin- ite and oxides [1] of iron or aluminium. Iron oxides are typically 10- 30 wt.% of the clay f r ac- tion, al t hough somet i mes they can be much more. The main iron phases, aluminous hemat i t e and al umi nous goethite, impart, respectively, the char- acteristic red or yellow col our to the soils [2]. Maghemi t e has somet i mes been suggested as a possible const i t uent . Accordi ng to Mullins [3] and Fi t zpat ri ck [4], it coul d be formed (a) from the ot her iron oxides in a surface l ayer in the presence of burni ng organi c mat t er [5], (b) by pedogeni c t ransformat i on under hydr omor phi c conditions, formi ng localized concret i ons [6], and (c) by in- heri t ance and alteration of t i t anomagnet i t e. Following observat i ons in the field that the entire soil col umn at cert ai n locations in Brazil is easily at t ract ed by a hand magnet [7], we investi- gat ed the origin and significance of these unusual magnetic propert i es. Here we discuss results for the following materials: (1) A series of 15 represent at i ve latosols selected by the Brazilian Soil Survey (samples B0- B14) which includes two except i onal l y iron-rich materi- als derived from itabirite [8,9]. (2) A vertical sequence (samples V1- V4) in- cl udi ng soil from the B and C hori zons at a site near Vicosa in the Zona da Mat a of Mi nas Gerais. (3) A hori zont al t oposequence (samples G1 G6) from the B2 hori zon of a shallow hill slope (1%) about 200 m long in I t umbi ar a in sout hern Goi as [9]. It represents in mi crocosm the xanthiza- tion or yellowing process in tropical soil i nduced by climatic and pedogeni c factors (e.g. moist soil climate, high organi c matter), which favour trans- f or mat i on of hemat i t e into goethite [10,11]. On passing f r om summi t (G6) to foot sl ope (G1) the soil col our changes from red to yellow as it be- comes richer in goethite and kaolinite, but poor er in hemat i t e and gibbsite. Our measurement s were made on the clay fract i on whose magnet i zat i on is especially sensitive to these factors. Magnet i zat i on curves for some represent at i ve samples are shown in Fig. l a. Two paramet ers are derived: sat urat i on magnet i zat i on o~ obt ai ned by ext rapol at i on to zero appl i ed field, and the high- field slope X= d ~ / d B0 defi ned in the interval 0 . 3 < B, , < 1 . 5 T. Values of o~ and )t for all 24 samples are listed in Tabl e 1. Variations in o~ for the soils span more t han three orders of magni- t ude whereas X is relatively const ant . Magnetiza- tion is expressed here per kg of Fe20~, since Mossbauer spectra show t hat pract i cal l y all the iron in the soils is oxidized to Fe 3+ (1 J / T kg = 1 emu/ g) , o~ reflects quant i t at i vel y the presence of ferri magnet i c oxides, whereas ), arises mainly f r om ant i ferromagnet i c oxides or paramagnet i c iron. These data, t oget her with extensive observat i ons in the field, establish that magnet i zat i on is sub- 0012- 821X/ 86/ $03. 50 ~:' 1986 Elsevier Science Publ i sher s B.V. b 0 30 Slope X / T- O6 f j a i , i , l J , , , 20 I 0 b i . . . . I . . . . . . . 0 O5 GI v 2 m v l m v1 10 B,~ r r ) Z5 Fig. 1. Room t emper at ur e magnet i zat i on curves for (a) t he clay fract i on sampl es GI and G6 and the whol e soil V4, and (b) t he whol e rock V1 and t he magnet i c ext ract s of t he rock and yel l ow r i nd Vl m and V2m. ( Not e the di f f er ent vertical scale.) st ant i al onl y when the soil derives f r om maf i c rock or f r om t he except i onal l y i ron-ri ch itabirites. Al- t hough t here can be l arge vari at i ons of os rel at ed to local fact ors such as relief and dr ai nage ( G1 - G6 : the t wo ext remes are i ncl uded in the figure), as well as br oad di fferences in cl i mose- quence in di fferent regions of the count r y ( B9- B12), the magnet i c soils t end to be those t hat are high in phos phor us and t race el ement s [12,13]. They are among the mos t fertile soils in Brazil. Magnet i zat i on is t herefore a val uabl e i ndi cat or of soil type, and one t hat is mor e sensitive t han col our to i nci pi ent xant hi zat i on. B9 and B10, for exampl e, are qui t e alike in col our and f or m f r om si mi l ar rock but the l ower magnet i zat i on of the l at t er reflects a cooler, mor e moi st regime. I n or der to est abl i sh the origin of the magnet i - zat i on, we focussed our at t ent i on on the vert i cal 323 sequence, V1- V4. The magni t ude of o s for these, and ot her st rongl y magnet i c sampl es, is t oo large to be expl ai ned by any moder at el y- pur e cryst al l i ne i ron oxi de except f er r i magnet i c magnet i t e or mag- hemi t e. Ther e are some report s, f r om X- r ay dif- fract i on, of maghemi t e in the cl ay fract i on of i Brazilian soils [12-16], i ncl udi ng B0 and G6 [9]. Because of the l i kel i hood t hat i ron in the cl ay fract i on will be oxidized, it is general l y t aken for gr ant ed t hat the phase giving the di agnost i c peak near 2.95 A is maghemi t e r at her t han magnet i t e [7]. However, large magnet i t e and i l meni t e grai ns i nheri t ed di rect l y f r om the par ent rock can persi st in the coarse fract i on of t ropi cal soils, where t hey are f ound to be except i onal l y st abl e (e.g. t here is about 10% of the i ron present in i l meni t e bot h in V1 and V4). Magnet i zat i on of the whole soil mat eri al in the vert i cal sequence V2- V4 is of si mi l ar magni t ude to t hat of t he r ock V1 f r om which it is derived. V1 is a maf i c rock compos ed of quart z, plagioclase, pyr oxene and opaque oxides. The yel l ow rind, V2, 1- 10 mm thick, which f or ms di rect l y on the weat her ed rock provi des the clue needed to under - st and the magnet i zat i on of the soil. An at t empt at magnet i c separ at i on was i neffect i ve for sampl es f r om the B hori zons but we found t hat coarse, bl ack, sand-si zed grai ns (V2m) coul d easily be separ at ed f r om the yel l ow ri nd wi t h a hand mag- net ; X- r ay di ffract i on i ndi cat ed t hat maghemi t e ( a = 8.33(1) A) was present in these grai ns to- get her wi t h some ilmenite. X- r ay fl uorescence in- di cat ed t hat the ot her maj or met al l i c el ement pres- ent in the magnet i c ext ract was Ti ( Ti Oz/ Fe203 = 0.43). Magnet i zat i on of V2m was o 2 = 45 J / T kg (25 J / T kg of sampl e), while t hat of the clay fract i on of the yel l ow ri nd was onl y 0.1 J / T kg. The M0ssbauer spect r um of the ext ract (Fig. 2a) shows a ferric magnet i c pat t er n but there is no sign of the charact eri st i c Fe 2+/3+ hyper f i ne sextet f r om B-site i ron in magnet i t e [17]; a mi nor amount of i l meni t e is present . We t herefore i dent i fy the magnet i c const i t uent in V2 as coarse grai ns of fully oxi di zed t i t anomaghemi t e. The phase, whose composi t i on is est i mat ed to be F e l . 9 Ti 0 . 6 0 4 (aft er correct i on for the i ron in ilmenite), decomposes t her mal l y at 290 + 10C. The magnet i zat i on of ful l y-oxi di zed synt het i c t i t anomaghemi t e of this composi t i on is about 10 J / T kg [18], whereas the 324 TABLE 1 Magnetic properties of Brazilian latosols Sample Description Colour, Elevation (m) Fe203 o~ k No. bedrock (wt.%) ( J / T kg ( J / T 2 kg) of Fe203 ) (a) Representatiue Brazilian latosols B0 B 2 clay < 2 btm itabirite 63.3 6.43 0,52 B1 B z clay 10R3/ 6 1200 itabirite 47.4 2.91 0.30 B2 B 2 clay 7. 5YR5/8 850 gneiss 11.2 0.08 0.46 B3 B 2 clay 7.5YR4/5 950 charnockite 18.3 0,09 0.59 B4 B 2 clay 10YR5/6 480 Tertiary sediments 8.6 0.12 0.52 B5 B 2 clay 2. 5YR3/7 siltstone 16.6 2.08 0.35 B6 B 2 clay 4YR4/ 4 645 mafic 23.3 2.38 0.60 B7 B 2 clay 1. 5YR3/4 760 marie 26.4 8.05 0.30 B8 B 2 clay 1.5YR3.5/6 sandstone 10.4 0.89 0.16 B9 B 2 clay 1YR3/ 4 640 mafic 23.8 14.39 0.51 BIO B 2 clay 1YR3/ 5 760 marie 20,1 5.62 0,43 BI 1 B 2 clay 3.5YR3.5/5 920 mafic 20.6 1.22 0.35 B12 B 2 clay 5YR3,5/5 1100 mafic 19.9 0.25 0.49 B13 B 2 clay 1YR4/ 8 910 Tertiary sediments 10.8 0.56 0.31 BI4 B 2 clay 10YR6/ 8 860 migmatite 13.5 0.07 0.53 (b) Soi l profile VI C rock mafic 19.6 6.00 1.04 VI m C rock mafic 40.2 44.60 1.36 magnetic extract V2 C yellow rind mafic 31.9 8.60 0.78 V2c C yellow rind mafic 37.1 0.10 0.20 clay V2m C yellow rind mafic 55.4 45.20 1.26 magnetic extract V4 B 2 soil mafic 22.0 3.83 0.66 (c) Toposequence G1 B 2 clay (bottom) mafic 17.9 0.29 0.30 G2 B 2 clay mafic 16.6 0.48 0.45 G3 B 2 clay mafic 20.2 0.72 0.37 G4 B 2 clay mafic 20.9 1.01 0.38 G5 B 2 clay mafic 23.5 1.49 0,30 G6 B 2 clay (top) mafic 26.3 4.45 0,36 -. , 0 o ~ 6 C) t~ - Q 0 a i V2 m I t I I ' ' ' ' t I l 1 I I I -I0 -6 -2 2 6 I0 Veloc/ty (mm/s) Fig. 2. Room t emper at ur e M~Sssbauer spect r a of t he magnet i c ext ract s of (a) t he yel l ow r i nd V2m, and (b) t he basal t i c rock f r om whi ch it is deri ved Vl m as well as of t he whol e r ed magnet i c soil V4 (c). magnet i zat i on of t he nat ural mi neral in the mag- net i c ext ract s is about 50 J / T kg (Tabl e 1). This suggests t hat t here are significant di fferences in the cat i on di st ri but i on in nat ural and synt het i c samples. Knowi ng o S of the nat ural t i t anomag- hemite, it is possible to i nfer the quant i t y present in soils f r om magnet i zat i on measurement s. The magnet i cal l y separat ed fract i on of the rock Vl m gives a M~Sssbauer spect rum (Fig. 2b) similar to t hat of the magnet i c ext ract f r om the yellow rind (Fig. 2a). Fi t paramet ers of the magnet i c sextet are i somer shift relative to iron 0.32 r a m/ s , quadru- pol e shift 0.00 mm/ s , hyperfi ne field 49.7 T. The lattice par amet er for the spi nel -st ruct ure phase is a = 8.31(1) A. It t herefore appears t hat the t i t anomaghemi t e is i nheri t ed f r om the par ent rock and it is br oken down to finer part i cl es and in- t i mat el y mi xed with ot her soil material in the upper hori zons (Fig. 2c). The r emar kabl e decrease in magnet i zat i on in the r e d- ~ yellow sequence G6 - G1 , as well as the di fferences in B9- B12, 325 i ndi cat e t hat soil maghemi t e in the clay fract i on is mor e likely to be el i mi nat ed t han soil hemat i t e duri ng xant hi zat i on. In summar y we concl ude: (1) Magnet i c latosols formi ng on mafi c rocks are wi despread in Brazil; t hey cover perhaps 5% of its l and area. (2) Ti t anomaghemi t e in these soils is i nheri t ed f r om rock cont ai ni ng al ready-al t ered t i t anomag- netite. (3) Ti t anomaghemi t e in the clay fract i on is sensitive to pedocl i mat i c envi ronment s favouri ng i nci pi ent xant hi zat i on. (4) Magnet i zat i on measurement s can be useful in Oxisol t axonomy. Since the original t i t anomag- net i t e associates with trace elements, the soil mag- net i zat i on correl at es with a number of desirable soil characteristics. (5) Ti t anomaghemi t e can be det er mi ned quant i - tatively f r om soil sat urat i on magnet i zat i on, pro- vi ded the magnet i zat i on of the pure mi neral is known. A c k n o w l e d g e m e n t s We are grat eful to Dr. I. Sanders for help and advi ce on the pet rographi c analysis and to Mr. D. Si mpson for the el ect ron mi cr opr obe analysis. Thi s work was suppor t ed by the Uni t ed Na- tions Devel opment Project, under cont r act UNDP BRA 82/ 023. R e f e r e n c e s 1 " Oxi de s " are t aken t o i ncl ude hydr oxi des and oxyhydr ox- ides. 2 U. Schwer t mann and R. M. Taylor, I r on oxides, in: Mi ner- als in Soil Envi r onment s, J.B. Di xon and S.B. Weed, eds., 145 pp., Soil Science Society of Ameri ca, Madi son, Wisc., 1977. 3 C.E. Mullins, Magnet i c suscept i bi l i t y of t he soil and its si gni fi cance in soil s c i e n c e - - a review, J. Soil Sci. 28, 223-246, 1977. 4 R. W. Fi t zpat ri ck, The occur r ence and pedogeni c signifi- cance of Fe-oxi des in soils of t he sout her n hemi sphere, in: I r on in Soils and Clay Mi neral s, J.W. Stucki and B.A. Goodman, eds., Reidel, in press. 5 U. Schwer t mann and B. Hei nemann, Ober das Vor kommen und die Ent s t ehung von Maghemi t in nor ddeut s chen B6den, Neues Jahrb. Mi neral . Monat sh. 8, 174-181, 1959. 6 R. M. Tayl or and U. Schwer t mann, Maghemi t e in soils and its origins, Clay Mi ner. 10, 289-310, 1974. 7 M. Resende, Mi neral ogy, chemi st ry, mor phol ogy, and geo- mor phol ogy of some soils of t he Cent ral Pl at eau of Brazil, Ph. D. Thesis, Pur due Uni versi t y, 1976. 326 8 Sampl es B1- B14 bel ong to a set s t udi ed by Brazi l i an soil sci ent i st s in connect i on wi t h t he devel opment of a soil cl assi fi cat i on s ys t em unde r t he l eader shi p of S NL C S / EMBRAPA. 9 N. Curi and D. P. Fr anzmei er , Topos equence of Oxi sol s f r om t he Cent r al Pl at eau of Brazil, Soil Sci. Soc. Am. J. 48, 341- 346, 1984. 10 U. Schwer t mann, Tr a ns f or ma t i on of hemat i t e to goet hi t e in soils, Na t ur e 232, 624- 625, 1971. 11 U. Schwer t mann, The effect of pedogeni c e nvi r onme nt s on i ron oxi de mi ner al s, Adv, Soil Sci. 1, 171- 200, 1985. 12 N. Curl , Li t hos equence and t opos equence of Oxi sol s f r om Goi as a nd Mi na s Ger ai s St at es, Brazil, Ph. D. Thesi s, Pur due Uni ver si t y, 1983. 13 M. Resende, unpubl i s he d dat a. 14 J. Esson and L.C. Sur can dos Sant os. Che mi s t r y and mi ner al ogy of a sect i on t hr ough t he l at eri t i c nickel deposi t at Li ber dade, Brazil, Tr ans . Inst . Mi n. Met al l . 87, B53- B60, 1978. 15 L. C. A. Pombo, E. Kl amt , I. Kus r a t h and D. Gi anl uppi , Identificaca~o de oxi dos de ferro na f r acao argi l a de Lat os - sol o Roxo, Rev. Bras. Ci en. Solo. 6. 13 17, 1982. 16 D. P. Sant ana, Ph. D. Thesi s, Pur due Uni ver si t y, 1984. 17 J. M. D. Coey, A. H. Mor r i s h and G. A. Sawat zky, A M6s s ba ue r s t udy of conduct i on in magnet i t e, J. Phys. 32, C1. 271- 273, 1971. 18 P. W. Re a d ma n and W, O' Rei l l y, Magnet i c pr oper t i es of oxi di zed t i t anomagnet i t es , J. Geomag. Geoel ect r. 24, 69 90, 1972.