WEATHERING OF BIOTITE INTO DIOCTAHEDRAL CLAY …minersoc.org/pages/Archive-CM/Volume_25/25-1-51.pdf · Weathering of biotite to dioctahedral clays 53 RESULTS AND DISCUSSION Apart

Clay Minerals (1990) 25, 51-63

W E A T H E R I N G OF BIOTITE I N T O D I O C T A H E D R A L CLAY M I N E R A L S

A. W. F O R D H A M

CSIRO Division of Soils, Private Bag No. 2, Glen Osmond, South Australia 5064

(Received 6 March 1989; revised 1 September 1989)

A B S T R A C T : Changes which occur during the natural weathering of biotite in granite gneiss and associated soils are measured by microanalysis and illustrated by SEM. Biotite weathers through a series of interstratified minerals to vermiculite and/or smectite phases which decompose rapidly to kaolinite. Both vermiculite and smectite phases appear to be dioctahedral, on the basis of chemical compositions derived from microprobe data. Weathering products are first apparent on the edges of laminae, where interstratified minerals are formed at right angles to both the edge face and the cleavage. Weathering soon develops along cleavage planes, initially most strongly near the edges of flakes, but then permeating extensively into the body of flakes and subdivided segments. The orientation of interstratified minerals and kaolinite within cleavages is parallel to the cleavage.

Oxidation of Fe in biotite causes internal stresses which are relieved by physical deformation of the crystals. This accelerates chemical decomposition, particularly along cleavage planes. At an advanced stage of weathering when decomposition is active at many cleavages, biotite deteriorates to very finely divided, wafery remnants consisting of thin laminae separated by more open layers of particulate clay. Parts of the thin laminae remain relatively unweathered and have the same chemical composition as the original biotite immediately after oxidation. These relatively unweathered layers within the laminae have X-ray diffraction (XRD) characteristics of trioctahedral illite and they persist when the biotite remnants are broken up into clay (Fordham, 1989b).

Other decomposition products are also formed. Some occur in sufficient bulk (at the microanalytical scale) to be identified directly by microprobe, but the chemical composition of others must be estimated by graphical procedures (Fordham, 1989a). These additional decomposition products of biotite are the subject of the present paper. Their formation in different parts of biotite flakes taken from different locations within the soil profile is described, and the weathering sequence as a whole is discussed. Microanalyses were performed on both thin sections and whole specimens, and the products are illustrated by scanning electron microscopy (SEM).

E X P E R I M E N T A L

Mater&& and methods

The sampling site at Mt Crawford in South Australia has a Mediterranean-type climate with hot, dry summers and cool, wet winters. Mainly Palexeralf (Soil Survey Staff, 1975) soils are developed on granite gneiss. The A horizon has a sandy-loam texture with the lower part often bleached, and the B horizon is a medium-heavy clay, merging into the weathered rind of subsurface boulders of granite gneiss.

�9 1990 The Mineralogical Society

52 A. W. Fordham

Using SEM, decomposition products are clearly visible (Fig. 2) at the edges of ftakes hand- and cut into thin sections. After optical examination, biotite flakes at different stages of weathering were analysed by electron probe microanalyser (EPMA) and, less frequently, by energy dispersive X-ray (EDX) microanalyser, attached to a scanning electron microscope. Over 400 points were analysed, each point having been selected after optical examination to ensure, as far as possible, that it was a weathering phase of biotite not contaminated with extraneous material.

In addition, whole specimens were examined by SEM. When decomposition products at the edges of biotite laminae were to be analysed by EDX, flakes were glued end-on in shallow slits scratched in carbon stubs. This allowed the edge faces to be set in optimum position relative to both electron source and detector. Such geometrical effects on EDX were minimal if the baseline of the elemental spectrum on visual display was level, and the count rates as well as the sum of all oxides were reasonably high. Microanalyses by SEM-EDX on whole specimens are designated as such in the text, while those by EPMA on thin sections are referred to as microprobe analyses.

Except in samples taken from exposed boulders of granite gneiss, most of the Fe within the crystal structure of biotite is in the ferric condition (Fordham, 1989b) and consequently Fe content is expressed as Fe203.

XRD data referred to in the present paper are illustrated in Fordham, 1989b, which also gives further details of materials and methods.

Treatment of data

Each microanalysis is first processed by computer to remove any contribution to the analysis by free iron oxides (Fordham, 1989a). The K20 content of the processed analysis is then plotted against Si/A1 ratio, such as in Fig. 1. From the position of the analytical point relative to calculated boundary curves, some idea of the mineralogical composition at the spot analysed can be obtained.

The boundary curves were calculated (Fordham, 1989a) after consideration of a large number of microprobe data, on the probability that at least some of the microanalyses would consist of only a single phase (or almost so). Such points would lie close to the intersection of boundary curves. One of the components, kaolinite, was available in sufficient bulk for its chemical composition to be measured directly by microprobe, but the composition of other components had to be derived from the amassed data.

Ifiterpretation of the results for micas is complicated because XRD indicates the presence of interstratified minerals as well as small amounts of more weathered products (see Fordham, 1989b and later in the present paper). For example, an analytical point on the boundary curve between oxidized biotite and vermiculite-like mineral can be thought of as a physical mixture of these two distinct phases spatially separated within the small volume of material penetrated by the microprobe. However, in view of the XRD results, it is more likely to represent an interstratified mineral with layers of one phase intimately interspersed with layers of the other phase or, in a more advanced weathering condition, with small domains of one phase between layers of the other phase.

Consequently, the results which follow are expressed in terms of interstratified minerals with mixed layers and/or small domains. The alternative of mixed but discrete phases undoubtedly occurs in some instances, especially when colour differences under the petrological microscope indicate a defined boundary.

Weathering of biotite to dioctahedral clays 53

R E S U L T S A N D D I S C U S S I O N

Apart from oxidized biotite or its clay-size analogue trioctahedral illite, other products of biotite weathering are a vermiculite-like mineral, kaolinite and free iron oxides (Fordham, 1989a). As explained above, the vermiculite-like mineral is taken to be part of a series of interstratified minerals. Each of these products is now described in more detail.

Interstratified minerals at edge faces of laminae

When viewed in thin sections taken from the interior of a submerged boulder, biotite shows the first signs of physical alteration at the boundaries of flakes. The original euhedral contours become ragged in parts, and colour changes occur. These colour changes are closely confined to thin bands about 5 #m wide near the edges, in contrast to the much wider aureoles observed by Wilson (1970). Microprobe analyses (Fig. 1) at edges of flakes show that the chemical composition at the edges is different from that in the body of the flakes where the composition approaches that of oxidized biotite (Table 1). In Fig. 1, the positions of the analytical points relative to the boundary curves indicate that most of the mixtures are composed of kaolinite and interstratified minerals together, in some instances, with oxidized biotite (see section on treatment of data).

10

K20 1

0 I 10 15 20 25 Si/AI FIG. 1. Microprobe analyses near the edges of biotite flakes in thin sections from the interior of a submerged boulder. Data processed to eliminate free iron oxides. O - - A represents mixtures of oxidized biotite O and kaolinite A. O- -D represents interstratified minerals with layers of oxidized biotite O and layers of vermiculite-like mineral []. Iq--A represents mixtures of

vermiculite-like mineral [] and kaolinite A.

TABLE 1. Microprobe analyses and derived compositions.

Description SiO 2 A1203 TiO 2 Fe203 MgO K20 Na20 CaO Total Si/A1

Products at edge faces 44-6 17.7 1.8 19.1 3.6 5.0 0-0 0.1 91.9 2.22 Oxidized biotite 34.4 14.8 2.9 24-0 8.7 9.2 0.2 0-0 94.2 2.05 Vermiculite-like

mineral 48.6 18.6 1.7 14.0 2.6 4.3 0.2 0.0 90.0 2.30 Vermiculite 50.6 20.8 2-0 16.0 2.4 0.0 0.2 0.0 92-0 2.15 Smectite-like

mineral 54.7 20.2 1-2 9.7 0.0 2.2 0.2 0.0 88-2 2.39 Kaolinite 43.2 36.9 0-1 2.0 0.1 0.1 0.0 0.1 82.5 1-03

54 A. IV. Fordham

Using SEM, decomposition products are clearly visible (Fig. 2) at the edges of flakes hand- picked from the same source as the thin sections. A composite SEM-EDX analysis encompassing most of the particles on the edge face of Fig. 2 is given in Table 1, after processing to remove a very small contribution from free iron oxides. This analysis, as well as four others from similar particles on the same flake, is plotted in Fig. 3. The point in Fig. 3 corresponding to the particles in Fig. 2 is labelled A. The location of the five points relative to the boundary curves indicates that the edge products are interstratified minerals mixed with small amounts of kaolinite. To test the validity of SEM-EDX analysis, two measurements were made on clean parts of laminae close to the decomposition products in Fig. 2. The corresponding points (Fig. 3) are located close to the oxidized biotite apex. Their location on the graph, together with the other element oxide contents in the total analyses, are in good agreement with microprobe analyses given for unblemished parts of biotite flakes in thin sections of samples taken from the same source as the whole specimens (Fordham, 1989b).

FIG. 2. SEM micrograph of interstratified minerals at the edge face of biotite laminae. Sample was taken from the interior of a submerged boulder.

10

�9 A

I

K20

f.Q t,5 2.0 25 Si/AI

FIG. 3. SEM-EDX measurements, processed to remove free iron oxides, on biotite flakes taken from the interior of a submerged boulder. Boundary curves are as for Fig. 1. �9 Weathering products at edge faces. �9 Clean laminae, x Products at edge faces at a later stage of weathering.

Weathering of biotite to dioctahedral clays 55

At this stage of weathering, there is very little change in the XRD patterns. The same hand- picked flakes from which the results of Figs 1 and 3 are obtained show a sharp 10 A peak with only slight broadening on the low-angle side to indicate interstratified minerals, while peaks due to kaolinite are very small.

The particles in Fig. 2 are 3-4/zm in diameter and lie with one axis at about right angles to the cleavage planes of biotite and another axis at about right angles to the edge face of the laminae. This particular orientation must result from partial dissolution followed by reconstruction of the crystal. The particles have a similar appearance and orientation to slightly smaller (2-3/tin) ones illustrated by Curmi & Fayolle (1981) and discussed further by Bisdom et al. (1982). Curmi & Fayolle decided the particles were kaolinite, apparently not from direct measurement of whole specimens but by inference from work on thin sections, where kaolinite was seen aligned at right angles to the cleavage at wide exfoliations within biotite. In the present work, the SEM-EDX analyses (Table 1) are clearly quite different from the analytical composition of kaolinite. The difference could be due to a contribution from underlying biotite, but if that were the case and the particles were kaolinite, then the data points in Fig. 3 would lie on the oxidized biotite to kaolinite boundary curve. Their actual location near the interstratified minerals curve confirms that they consist predominantly of interstratified minerals.

Interstratified minerals along cleavage planes

As weathering progresses, decomposition along cleavage planes becomes more important than that at edges. At first, it is most active at the ends of laminae (see later). In several instances, biotite flakes are weakened near the edges by fractures running at high angles to the cleavage, and this causes accelerated decomposition of angular blocks. Microprobe analyses (not shown) within such blocks indicate high contents of interstratified minerals and kaolinite. However, in due course, weathering along cleavage planes pervades the whole flake, which breaks up into smaller segments along the wider exfoliations. The more weathered segments consist of numerous wafers of relatively compact material separated by equally thin spaces containing aggregates of particulate clay. Many of the microprobe analyses reported previously (Fordham, 1989a) for thin sections of these weathered segments indicate mixtures with variable proportions of oxidized biotite, interstratified minerals and kaolinite. The degree of subdivision at this stage is so fine that pure phases larger in volume than that analysed by microprobe are not encountered. The thin wafers are believed to consist of remnants of oxidized biotite arranged in parallel with, and merging into, interstratified oxidized biotite-vermiculite minerals, with the proportion of the latter increasing as weathering increases. Such a parallel orientation fits the appearance of the wafers and is consistent with X-ray and electron diffraction work (Wilson, 1966; Gilkes & Suddhiprakarn, 1979b) and with lattice imaging in the transmission electron microscope (Banfield & Eggleton, 1988).

Vermiculite-like mineral

The chemical composition (Table 1) of the vermiculite-like mineral was derived previously from microanalytical data (Fordham, 1989a). Its structural formula (Table 2) is calculated by assuming that enough Mg goes into octahedral sites to give a total occupancy of 4 and that the remainder is in the interlayer region. The resulting clay mineral has a charge of 1-3 per unit- cell Ozo(OH),. However, K comprises 74~o of the interlayer cations and this is not consistent with the expanded nature of vermiculite (see later).

56 A, W. Fordham

TABLE 2. Structural formulae based on 44 charges.

Sample No.* 1 2 3 4

Tetrahedral Si 5.23 6.89 6.84 7.54 A1 2,65 1.11 1.16 0.46 Fe 0.12 - - - - - - Sum 8.00 8,00 8.00 8-00

Octahedral A1 - - 2.00 2-15 2.83 Ti 0-33 0.18 0,20 0.12 Fe 2.57 1.49 1-63 1.01 Mg 1.96 0-33 - - - - Sum 4.86 4.00 3-98 3.96

Interlayer K 1.78 0.78 - - 0.39 Na 0.06 0-05 0,05 0,05 Mg - - 0.22 0.48 - - Sum 1.84 1-05 0,53 0,44

Charge Tetrahedral - 2-77 - 1.11 - 1,16 - 0.46 Octahedral + 0.93 - 0.15 + 0,14 + 0.01 Interlayer + 1.84 + 1-27 + 1.01 + 0,44

*1 = Oxidized biotite; 2 = Vermiculite-like mineral; 3 = Vermiculite; 4 = Smectite-like mineral.

There is XRD evidence of small amounts of vermiculite in the clay fractions of some samples in the B horizon. A peak at 14/k is not displaced by Mg-saturation and glycerol- solvation. It disappears subsequently after K-saturation or heating at 400~ and the 10 A peak becomes more intense. These results indicate the absence of interlayer hydroxy sheets within the vermiculite.

The vermiculite identified by XRD in bulk clay fractions is not necessarily derived from biotite. Muscovite is also present in the samples, although it is at least 10 times less numerous (by point counting of thin sections) and is less reactive (by appearances in thin section) than

biotite.

Hydrobiotite

XRD of the clay fraction separated from a narrow clay band located below one submerged boulder and above another shows the presence of hydrobiotite. A peak at 10.2/k is broadened on the low-angle side, develops a broad peak at 11-9 A, and another at 14.0/k which is not

displaced by Mg-saturation and glycerol solvation. Microprobe data obtained from biotite remnants in a thin section of the narrow clay band

are shown in Fig. 4. They fall reasonably close to a line (not shown) which represents mixtures of kaolinite with a 1 : 1 interstratification of oxidized biotite and the vermiculite-like mineral, but the high K20 content (6.8 ~ ) of the 1 : 1 mineral is again abnormal. Furthermore, the ratio of MgO to K20 in the data of Fig. 4 is significantly higher than that in the postulated 1:1 mineral. Based on those data, another 1:1 mineral composed of vermiculite layers which contain no K (i.e., the 1:1 mineral has a total K 20 content of 4-6~) would have a MgO

Weathering of biotite to dioctahedral clays

K20

1o

9

s t \

2 .... Z ................. ~S

1 ....................... I ............... i

o i 10 15 20 25

Si/AI

FIG. 4. Microprobe analyses on weathered biotite fragments in thin section from a narrow clay band between two submerged boulders. Data processed to eliminate free iron oxides. O - - A represents mixtures of oxidized biotite O and kaolinite A. O - - A represents interstratified minerals with layers of oxidized biotiteO and layers of vermiculite-like mineral I--]. I--]--/~ represents mixtures of vermiculite-like mineral [] and kaolinite A. HB = hydrobiotite.

S = smectite-like mineral. V = vermiculite.

57

content of 5.5%. Using the methods already described (Fordham, 1989a), the chemical composition of this hydrobiotite can be derived, and from it, the chemical composition of vermiculite (Table 1) in the layers.

The boundary curves for mixtures of oxidized biotite and vermiculite, and for mixtures of the corresponding hydrobiotite and kaolinite, are drawn as dashed lines in Fig. 4. Points in the lower part of the diagram in Fig. 4 can now be interpreted as mixtures of hydrobiotite and kaolinite, while the other points contain additional amounts of oxidized biotite from less altered parts of the biotite remnants.

The possibility of a vermiculite with the composition listed in Table 1 exposes one of the limitations (Fordham, 1989a) of the method used to treat microprobe data. This occurs when none of the data for mixtures containing a particular component falls outside the diagram occupied by the other data for mixtures without that component.

The structural formula of the vermiculite without K is given in Table 2. The mineral is dioctahedral with a charge of 1.0 per unit-cell O20(OH)4, most of the charge being contained within the tetrahedral sheet. There is a relatively high proportion of Fe in the octahedral sheet and all the Mg is in the interlayer region. The mineral could be regarded as a low-charge vermiculite or a high-charge beidellite-nontronite (Borchardt, 1977; Bailey, 1980).

It should be noted that a similar XRD pattern to that described for clay from the narrow clay band is also given by clay separated from the A2 horizon. In the latter case, microprobe data (not shown) are located close to the oxidized biotite to kaolinite boundary curve. The reason for this discrepancy is not yet clear.

Smectite-like mineral

If a vermiculite without K is present, then the vermiculite-like mineral discussed previously still retains interstratified layers of oxidized biotite. The chemical composition of the vermiculite-like mineral was derived by giving considerable weight to the microprobe data located close to the boundary curve subsequently drawn from oxidized biotite to the vermiculite-like mineral. The MgO contents of these data do not increase relative to K 20 as

58 A. W. Fordham

the K20 contents decrease: the reverse is true. For these data, there is a relationship (not shown) between K20 and MgO contents which, if it remains valid at very low contents, gives a K20 content of ~ 2.2~o when the MgO content is zero. In chemical terms, no oxidized biotite could remain at this stage. In Fig. 4, the boundary curve from oxidized biotite to the vermiculite-like mineral is extended by a dotted line to a K20 content of 2"2~o, and the chemical composition corresponding to this point is listed as a smectite-like mineral in Table i. By simple arithmetic, the vermiculite-like mineral can now be interpreted as an interstratified mineral consisting of 30~ oxidized biotite layers and 70~ smectite-like layers.

The structural formula of the smectite-like mineral (Table 2) is that of a very low-charge smectite with K in the interlayer positions. It must be emphasized that the chemical compositions derived for the smectite-like mineral and for vermiculite are hypothetical. They were not determined by direct microprobe analysis, but this may simply mean that smectite and vermiculite do not accumulate in situ to an extent where they occupy a volume greater than that penetrated by the microprobe (~ 5 #m3). In this context, vermiculite without A1- hydroxy interlayer sheets is known to be unstable (Douglas, 1977).

Smectite is found by XRD in the bulk fraction from only one location in the soil profile, namely, the lower B horizon. The XRD pattern has a small broad peak at 14 A which develops and shifts to about 18 A when the sample is Mg-saturated and glycerol-solvated. Both the 14 A and 18 ./( peaks subsequently disappear after K-saturation or heating at 400 ~ confirming the presence of a smectite and the absence of interlayer hydroxy sheets. As before, there is no certainty that the smectite detected by XRD is derived specifically from biotite.

Chemical changes during weathering

From the evidence presented, oxidized biotite weathers through a series of interstratified minerals to vermiculite and smectite phases which, under the prevailing conditions, decompose quite rapidly to kaolinite. If this sequence is represented by the structural changes shown in Table 2, then there is displacement of A1 by Si from the tetrahedral sheet, together with transfer of A1 to the octahedral sheet to compensate for losses of Fe and Mg. Curmi & Fayolle (1981) reported similar results. The movement of A1 from one sheet to replace ions expelled from an adjacent sheet is consistent with the concept that some cations in biotite are relatively mobile within the total framework of oxygen atoms (Gilkes & Suddhiprakarn, 1979b). As weathering increases, the charge on each sheet of the structure decreases, with that on the octahedral sheet tending towards zero. The changes during natural weathering of Mt Crawford biotite into clay-size particles thus appear to be much more than the idealized exchange of interlayer K by other cations to form expanded vermiculite (Fanning & Keramidas, 1977).

Weathering sequence

The proposed sequence of oxidized biotite-vermiculite-smectite via a series of interstrati- fled intermediates is a mechanism by which high-charged mica alters to low-charged smectite and, at the same time, a large proportion of the total charge is always present within the tetrahedral sheet. The evidence presented mainly in the form of microanalytical data complements other work based on XRD. For example, Nettleton et al. (1970) reported hydrobiotite, vermiculite, montmorillonite and kaolinite as weathering products of biotite in a toposequence over quartz-diorite. The relative development of each product depended on the chemical environment during weathering. The formation of smectite was favoured by the

Weathering o f biotite to dioctahedral clays 59

presence of high concentrations of Na and Mg in the soil solution under conditions of restricted drainage. April et al. (1986) found a similar suite of clay minerals, together with vermiculite containing Al-hydroxy interlayer sheets, in a large number of soils developed on a variety of parent materials. They suggested that the vermiculite was derived from biotite. Smectite was found in poorly drained conditions (Borchardt, 1977) and, in the present work, it was detected only in the clay band where strong mottling indicates periodic waterlogging and reduction. By contrast, Kapoor (1972) found smectite in freely draining soil profiles in Norway and, although hydrobiotite, vermiculite and interstratified smectite minerals were also detected, kaolinite was absent.

Trioctahedral to dioctahedral forms

According to Table 2, the trioctahedral structure of oxidized biotite is converted to the dioctahedral structures of vermiculite, smectite and then kaolinite. This is contrary to the general idea (Douglas, 1977) that trioctahedral vermiculite is an alteration product of biotite whereas dioctahedral vermiculite is derived from muscovite, but the results are not without precedent. Stoch & Sikora (1976), April et al. (1986) and Fritz (1988) all observed changes from trioctahedral biotite to dioctahedral minerals. Kerns & Mankin (1967) interpreted their results on different size fractions of a vermiculite as a conversion from trioctahedral to dioctahedral forms. On the other hand, Nettleton et al. (1970), Kapoor (1972), Wilson (1973) and Rebertus et al. (1986) reported that the trioctahedral structure is preserved during decomposition of biotite.

Kaolinite from biotite

Kaolinite and halloysite have been recognized as natural weathering products of biotite on numerous occasions, the most recent reports being those of Harris et al. (1985a,b), Rebertus et al. (1986) and Banfield & Eggleton (1988). In the present work, kaolinite is one of the end- products of the weathering sequence observed at Mt Crawford. Its widespread occurrence within weathered biotite from Mt Crawford is clearly expressed in the microprobe analyses already reported (Fordham, 1989a), where a large number of data are located near the kaolinite apex of the boundary diagram in the graph of K20 content against Si/A1 ratio.

In the SEM, kaolinite is first observed on the edge faces of laminae. The interstratified minerals illustrated in Fig. 2 are altered to kaolinite in some of the hand-picked flakes from the interior of a submerged boulder, and the edges then appear as shown in Fig. 5. Smaller (0.5-1.0 #m) particles of kaolinite and tubes of halloysite can be seen mixed with larger angular particles of interstratified minerals. Three SEM-EDX analyses within the frame of Fig. 5 are plotted in Fig. 3 as crosses. The location of these points relative to the boundary curves confirms that the particles consist of approximately equal amounts of kaolinite/halloy- site and interstratified minerals.

Tubular halloysite is always a minor weathering product of biotite at Mt Crawford, the tubes showing no particular orientation relative to other particles. Nothing of the kind reported by E swaran & Heng (1976) and Penven et al. (1981), where halloysite protrudes from the edges of laminae in the same plane as the cleavage, is observed.

The formation of well-defined particles at edge faces is only a temporary condition. I f the particles are present at all on more weathered flakes, they are less oriented and have smoothed contours as if affected by dissolution. They become replaced by much larger, nondescript scales which adhere in a random fashion to the side walls. Scales such as this are

60 A. W. Fordham

FIG. 5. SEM micrograph of a mixture of interstratified minerals, kaolinite and halloysite at the edge face of biotite laminae. Sample was taken from the interior of a submerged boulder.

illustrated in papers by Tarzi & Protz (1978), Shoba & Sokolova (1981) and Ghabru et al.

(1987). At higher magnification (Fig. 6), the scales are seen to be aggregates of tightly knit, almost fused particles. The particular aggregate depicted in Fig. 6 is found by SEM-EDX to consist almost wholly of kaolinite, but usually the aggregates also contain some interstratified minerals. The scales are believed to be remnants of weathering within laminae rather than outside them. They are formed when decomposition along cleavage planes is initially concentrated at the ends of laminae as a result of free access of external solutions. Intensely weathered products accumulate there, but they break off from the laminae and some are retained as aggregates or scales on the side walls of the flake.

When weathering along cleavage planes becomes dominant within the body rather than along the edges of flakes, kaolinite is frequently detected by microprobe in thin sections of weathered flakes, particularly along wider (> 1 #m) exfoliations. In some cases, the microprobe analyses correspond closely to the composition of pure kaolinite (Table 1), but usually free iron oxides and minor amounts of interstratified minerals are present as well. Kaolinite in wider exfoliations has the same appearance in thin sections as the particulate material between wafers in more strongly weathered fragments. Although this material between wafers cannot be identified directly because of the fine scale of subdivision, there is little doubt that it is predominantly kaolinite. Transported clay sometimes occurs within wider exfoliations, but it is easily distinguished by the tightly packed and uniformly fine texture resulting from the very small size and dispersed nature of the clay particles. When whole specimens of slightly weathered biotite are cleaved with a sharpened needle, the weathering products on the exposed surface lie parallel to the cleavage, but become more disoriented as weathering progresses. They are identified by SEM-EDX as predominantly kaolinite.

Kaolinite appears to be formed relatively rapidly from the expanded clay minerals in the weathering sequence, because there is no appreciable accumulation of vermiculite and smectite in the profiles at Mt Crawford. A similar situation was found in deeply weathered

Weathering of biotite to dioctahedral clays 61

FIG. 6. SEM micrograph of an aggregate of kaolinite particles adhering to the side wall of a biotite flake taken from the weathering rind around a submerged boulder.

granitic rocks in Western Australia by Gilkes & Suddhiprakarn (1979a), with differences that smectite was not detected, and gibbsite and crystalline iron oxides were abundant. Mechanisms for the transformation of 2:1 into 1:1 clay minerals were discussed by Harris et al. (1985b) and Rebertus et al. (1986). Those models which include intermediates containing an Al-hydroxy interlayer sheet are not applicable to the present study, because there is no evidence for such minerals, nor in the work of several others including Eswaran & Heng (1976) and Penven et al. (1981). The unusual orientation of kaolinite at edge faces of biotite suggests that some form of partial dissolution and reconstruction (Wilson, 1966) is involved at these sites and probably also within the cleavages. Such a mechanism requires release of Si, and amorphous looking particles of pure SiO2 are sometimes detected within exfoliations of biotite in thin sections. A silicon-rich environment favours the weathering of oxidized biotite, according to the scheme in Table 2.

Iron oxides from biotite

Biotite is a source of abundant Fe and, according to the weathering sequence inferred in Table 2, Fe is most readily available from the time oxidized biotite layers begin to decompose into vermiculite-like layers. Free iron oxides are detected in the processed microprobe data long before they can be seen in thin sections or whole specimens. After they accumulate into microaggregates, they are visible optically as stains in thin sections, and as small (0.1/~m) pellets by SEM where they have a similar appearance to those illustrated by Sousa & Eswaran (1975), Eswaran & Heng (1976) and Penven et al. (1981). In the present work, most of the iron oxides are found to be poorly crystalline, with only small amounts (< 5~) of goethite and/or hematite being identified by XRD in a few of the samples. Finely divided iron oxide pellets are seen by SEM to cement clay particles together and to form part of clay skins around larger particles.

62 A. W. Fordham

C O N C L U S I O N S

Weathering products of biotite are first observed on the edge faces of laminae, where they occur as well-defined particles in perpendicular orientation both to the face and to the cleavage plane. Microanalysis indicates that they are interstratified minerals containing trioctahedral illite layers and dioctahedral vermiculite and/or smectite layers, together with kaolinite and halloysite. However, this particular phase of weathering is shortlived, because such well-defined particles are absent from the edges of more weathered flakes. Instead, there are now much larger, irregular scales of aggregated material adhering to the sides of the flakes. These scales, which consist predominantly of kaolinite, are believed to be products of weathering along cleavage planes close to the edges of laminae. Weathering along cleavage planes, as opposed to edge faces, is at first most active near the ends of laminae, and is displayed at a less magnified scale in thin sections by fan-shaped tips of some segments. Products at the edges apparently peel off quite readily, because in most cases there is no accumulation beyond that of a thin band.

As weathering progresses, decomposition along cleavage planes pervades the body of flakes and dissociated segments, leading eventually to a very finely stratified structure. Thin wafers of relatively compact material are separated by more open layers of particulate clay. Part of the wafers still consists of a single phase of trioctahedral illite, but this merges into interstratified minerals in which layers of trioctahedral illite are interspersed with layers of vermiculite and, less commonly, smectite. The clay particles between wafers are predominantly kaolinite and free iron oxides.

The conversion of trioctahedral illite to dioctahedral vermiculite and smectite via interstratified intermediates seems to occur by displacement of A1 by Si from the tetrahedral sheet and transfer of A1 to the octahedral sheet to replace some of the Fe and Mg lost from the structure. No hydroxy interlayer sheets are formed and no excess AI in the form of gibbsite is found. Alteration of the expanded 2:1 minerals to kaolinite occurs relatively rapidly in the prevailing weathering conditions.

ACKNOWLEDGMENTS

The author is grateful to Mr R. Merry for his major contribution in the collection and impregnation of samples, to Mr S. McClure for SEM micrographs and EDX analyses, and to Mr H. Rosser for assistance with the EPMA.

REFERENCES

APRIL R.H., HLUCHY M.M. & NEWTON R.M (1986) The nature of vermiculite in Adirondack soils and till. Clays Clay Miner. 34, 549-556.

BAILEY S.W. (1980) Summary of recommendations of AIPEA nomenclature committee. Clay Miner. 15, 85-93.

BANFIELD J.F. & EGGLETON R.A. (1988) Transmission electron microscope study of biotite weathering. Clays Clay Miner. 36, 47-60.

BISDOM E.B.A., STOOPS G,, DELVIGNE J., CURMI P. & ATLEMULLER H-J. (1982) Micromorphology of weathering biotite and its secondary products. Pedologie 32, 225-252.

BORCHARDT G.A. (1977) Montmorillonite and other smectite minerals. Pp. 293-330 in: Minerals in Soil Environments (J. B. Dixon & S. B. Weed, editors). Soil Sci. Soc. Am., Madison, Wisconsin.

CURMI P. & FAYOLLE M. (1981) Microscopic characterization of weathering in a granitic saprolite. Pp. 249- 270 in: Submicroscopy of Soils and Weathered Rocks (E. B. A. Bisdom, editor). Centre for Agricultural Publishing and Documentation (Pudoc), Wageningen.

Weathering o f biotite to dioctahedral clays 63

DOUGLAS L.A. (1977) Vermiculites. Pp. 259-292 in: Minerals in Soil Environments (J. B. Dixon & S. B. Weed, editors). Soil Sci. Soc. Am., Madison, Wisconsin.

ESW~d~AN H. & HENG Y.Y. (1976) The weathering of biotite in a profile on gneiss in Malaysia. Geoderma 16, 9- 20.

FANNING D.S. & KERAblIDAS V.Z. (1977) Micas. Pp. 195-258 in: Minerals in Soil Environments (J. B. Dixon & S. B. Weed, editors). Soil Sci. Soc. Am., Madison, Wisconsin.

FOrtDI-IAM A.W. (1989a) Treatment of microanalyses of intimately mixed products of mica weathering. Clays Clay Miner. (in press).

FORDHAM A.W. (1989b) Formation of trioctahedral illite from biotite in a soil profile over granite gneiss. Clays and Clay Miner. (in press).

FRIa'Z S.J. (1988) A comparative study of gabbro and granite weathering. Chem. Geol. 68, 275-290. GHABRU S.K., MERMUT A.R. & ST. ARNAUD R.J. (1987) The nature of weathered biotite in sand-sized

fractions of gray luvisols (boralfs) in Saskatchewan, Canada. Geoderma 40, 65-82. GILKES R.J. & SUDDHII'RAKARN A. (1979a) Biotite alteration in deeply weathered granite. I. Morphological,

mineralogical and chemical properties. Clays Clay Miner. 27, 349-360. GILKES R.J. & SUDDHIPRAKARN m. (1979b) Biotite alteration in deeply weathered granite. II. The oriented

growth of secondary minerals. Clays Clay Miner. 27, 361-367. HARRIS W.G., ZELAZNY L.W., BAI~ER J.C. & MARTENS D.C. (1985a) Biotite kaolinization in Virginia

Piedmont soils. 1. Extent, profile trends and grain morphological effects. Soil Sci. Soc. Am. J. 49, 1290- 1297.

H~a',RIS W.G., ZELAZl'~' L.W. & BI.OSS FD. (1985b) Biotite kaolinization in Virginia Piedmont soils. II. Zonation in single grains. Soil Sci. Soc. Am. J. 49, 1297-1302.

KAPOOR B.S. (1972) Weathering in micaceous clays in some Norwegian podzols. Clay Miner. 9, 383-394. KERNS R.L. & MANKIN C.J. (1967) Compositional variation of a vermiculite as related to particle size. Clays

Clay Miner. 15, 163-177. NETTLETON W.D., FLACH K.W. & NELSON R.E. (1970) Pedogenic weathering of tonalite in southern

California. Geoderma 4, 387-402. PENVEN M-J., FEDOROrF N. & ROBERT M. (1981) Weathering of biotites in Algeria. Geoderma 26, 287-309. REBERTUS R.A., WEED S.B. & BUOL S.W. (1986) Transformations of biotite to kaolinite during saprolite-soil

weathering. Soil Sci. Soc. Am. J. 50, 810-819. SHOBA S.A. & SOKOLOVA T.A. (1981) Weathering products of biotite in sod-podzolic soil. Soviet Soil Sci. 6, 91-

97. SOIL SURVEY STAFF (1975) Soil Taxonomy. Handbook US Dept Agric. No. 436, Gov. Printer, Washington,

DC. SOUSA E.C. & ESW~,AN H. (1975) Alteration of micas in the saprolite of a profile from Angola: a

morphological study. Pedologie 25, 71-79. Srocn L. & SIKORA W. (1976) Transformations of micas in the process of kaolinization of granites and

gneisses. Clays Clay Miner. 24, 156-162. TARZI J.G. & PROTZ R. (1978) Characterization of morphological f~atures of soil micas using scanning

electron microscopy. Clays Clay Miner. 26, 352-360. WILSON M.J. (1966) The weathering of biotite in some Aberdeenshire soils. Mineral. Mag, 35, 1080-1093. WILSON M.J. (1970) A study of weathering in a soil derived from a biotite-hornblende rock. 1. Weathering of

biotite. Clay Miner. 8, 291-303. WILSON M.J. (1973) Clay minerals in soils derived from Lower Old Red Sandstone till: effects of inheritance

and pedogenesis. J. Soil Sci. 24, 26-41.