THE AMERICAN MINERALOGIST, VOL. 50, MAY_JUNE, 1965

MINERALOGICAL STUDIES OF KAOLINITE- CLAYS: PART III. A FIBROUS KAOLIN FROM PIEDADE, SAO PAULO, PE,nsroDE SouzA SeNros, Instituto ile PesquisasTecnologicas, University of Sd.oPaulo, Sdo Paulo, Brazil, G. W. Bnrnlr.nv, Departmentof Geochemistryand. Minerology, The Pennsylaani,a Slate (Jniaers'ity,Uniaersity Park, Po., aNn HBTBNI oB Souze SaNros, Centro de Microscopia Electronico, (Jn'iversityof Sd,oPaulo. Sao Paulo. Brazil.

Arsrn,rcr An unusual form of kaolin mineral, filling cracks in weathered prophyritic in the countl, of Piedade, Sdo Paulo, Brazil, consists of visibly fibrous material which, in the electron microscope, appears as thin tubular fibers of considerable length; platy particles are almost entirely absent. X-ray rotation diagrams of dried fiber bundles and selected area electron diffraction patterns of single fibers show an appreciable degree of structural regu- larity in the kaolin layer structure, with 6 as the fiber axis. The results show generar agree- ment with previous data by Honjo, Kitamura and Mihama. Prior to drying, the material as collected can be described as consisting of f endellite and $ halloysite. The endellite com- ponent has the swelling and shrinking behavior characteristic of this mineral with respect to ethylene glycol, water, and gentle heat treatments. This may be the first observation of a mineral of endellite-halloysite type existing in a directly visible fibrous form.

INrnolucrror.l An unusual form of kaolin has been found in the county of Piedadel, Sao Paulo, BraziI. It is fibrous in character at all degreesof magnification and behaves with respect to water, ethylene glycol, and gentle heat treatments, as a mixture of endellite and halloysite in a roughly 2:l ratio.2 The material occurs in layers, about 1-3 cm thick, within cracks of weatheredporphyritic granite (for a general description of the geology oI of the area, see Knecht, 1949) and is easily visible in road cuts as white Iayers running in various directions within the bulk material. Where the surface of the altered granite is eroded, the white layers protrude and eventually break off in fragments which on drying have a remarkable fi.brousappearance. Samples were collected from below the exposedsur- face and were kept in water in closed containers. They were crushed, ground, or separatedby hand in the wet state. Chemicalanalysis of material dried at 110'C. gave the resultslisted

1 The pronunciation is approximately as follows: The word has four syllables of equal 'pee' emphasis,Pi (as in peel), e (as'ay'in say), da (asin dark), and de (as 'dir'in dirt). 2 Endellite and halloysite are used to signify kaolin clay having basal spacings of about 10.0and 7.2 A respectively.

619 620 P. DE S. S,41TTOS.G. W. BRINDL]'Y AND IT. DE S. SANTOS

Tasr-n 1. Cnturcal ANer,vsrsol Prtoale Crnv Dnrro ar 110' C.

SiOz 45.0 Al2o3 40.1 HrO (+110) r5.4 Iron as FezO: 0.72 CaO trace Mgo trace NarO trace KrO 002

Total lo.r24

in Table 1. Differential thermal analysis of dried material gave the usual endothermicand exothermicpeaks of a halloysiteor kaolinite mineral.

Oprrcer, AND ELECTRoN-OprrcAlExaurN,r.rroN The fibrous character of the clay material can be seen directly in fractured surfaces.With the optical microscope,at 80X magnification,

Frc. 1. (a) Fiber bundles, about 0.3 mm long, photographed at 80X magnification. (b) Thin fiber bundie mounted on glass fiber for:r-ray study. (c) X-ray fiber diagram of bundle shown in (b). P-IBROUSKAOLIN MINDRAL 621

the material appearsas shownin Figs. 1, a,b. The fibersform thick bun- dles parallel to their length and have considerablecoherence so that dry material is difficult to grind. A lightly Pt-shadowed carbon replica of a surface appears in the electron microscope as in Fig. 2a, where the individual fibers are about 0.05-0.1 micron wide and many microns long in nearly parallel

Frc. 2. (a) Lightll' Pt-shadowed, carbon replica of a fracture surface of piedade clay. (b) Carbon replica of dispersed fibers. (c) Dispersed material on a collodion film. (d) Dry- ground fragments of Piedade clay fibers. aggregates.Individual fibers are seenmore ciearly in Fig. 2b, obtained from a carbonreplica of wet material crushedand smearedbetween glass plates, and in Fig. 2c, obtained by placing a drop of water-dispersed material on a collodion film. In Fig. 2a, the fibers have a flat, or flattened, lathlike appearance. In severalplaces, they appear to open out into almost platy forms. fn Fig. 2b, they appear as rolled tubes, sometimesonly partially rolled; thin, fractured walls, with holes possibly formed by escapingwater or vapor, can be seen. Some fibers are split lengthwise, and others are flattenedand twisted into ribbon-likeforms. Figure 2c showsfibers which 622 P. DE S. SANTOS,G. W. BRINDLIJYAND I]. DE S. S,41TTOS are thin and long, and others which are much thicker and have a frac- tured appearance. Attempts have been made to obtain direct evidencefor the cross- sectional form of the fi.bers by preparing thin transverse sections of methacrylate-embedded fibers with a Porter Blum glass-knife ultra- microtome,but the material 'smeared'under the cutting action and little direct evidencewas obtained.However, longitudinal sectionsprepared in this way showedvery clearlythe hollow of the elongatedfibers. Figure 2d shows fragments of tubes obtained after dry grinding the material twelve times and each time setting it in "Lakeside" cement as describedby Brindley and Kurtossy (1961). This wasthe only method by which the long fibers could be broken into short lengths which permitted r-ray powder diffraction data to be obtainedwithout marked orientation effects.

DrnlnactroN SrUDY

X-ray f.ber d,iagram.The fiber bundle shown in Fig. 1b, about 0.5 mm long, was obtained by drawing out a thin wisp of material from the dry clay. It was attached to a glass fiber as shown in the figure, and was mounted in a rotation-type #-ray camera. The resulting diffraction diagram, taken with filtered CuKo radiation and a stationary fiber bundle, is reproducedin Fig. 1c. It has the appearanceof a rotation diagram due to the large number of individual fibers within the bundle, but with considering'arcing' of the reflectionsalong powder lines. The repeat distancealong the fiber axis, about 9 A, correspondsto the b-axis of the kaolinite and halloysite structures. Basal reflections along the zero level, up to 005, correspondto d(001):7.10 A in agreementwith the values for kaolinite and halloysite within the accuracy of the measure- ment. On either sideof the 003 reflection,are strong and sharp reflections which can be indexed 200 and 202.

Single f,ber electron difraction. Electron difiraction patterns of single fibers taken by selectedarea diffraction yield results (Fig. 3) somewhat simiiar to the c-ray fiber pattern in that the fiber axis has a repeat dis- tance of 8.93 A, measuredon the 040, 060, and 080 reflections,and d(001):7.t+ A, measuredon 001, 002, and 003 reflections.The material was lightly shadowedwith Al metal to obtain the calibrationrings seenin Fig. 3. The diffraction pattern clearly doesnot correspondwith a single net plane in reciprocal space.With b* as the fiber axis and c* perpendicular to the axis,such a net plane would contain reflectionswith indices0kl, and with k an even number since(h*k) is even for the base-centeredkaolin- FIBROUSKAOLIN MIN]]RAL 623 ite structure. In Fig. 3, many diffractionsappear with k an odd number, and the diagrarn as a whole has the appearanceof a rotation, or partial rotation diagram. This result is broadly consistent with the rolled form of the particles. Sharp reflections on either side of 003 appear to be the 200 and 202 reflections, which were seen also in the r-ray diagram, and the closedoublet with k:l appearsto be the 110,111 pair.

Frc. 3. Selected area, electron diffraction pattern of the single fiber shown in the inset. 'Ihe detailed interpretation of electron difiraction patterns of single halloysite particles has been consideredalready by Honjo and Mihama (1954),by Honjo et al. (1954),and by Waser (1955).The closelyanalo- gous problem of difiraction by has been consideredin consider- able detail by Whittaker (1953-57) and by other investigators. In the caseof halloysite, the discussionshave by no means reachedfinality. The information contained in Fig. 3 appears to be very similar to the results obtained by Honjo and his collaborators and probably is similar to the diagram given by Waser so far as one can judge from a very poor repro- duction. In particular, it is seenin Fig. 3 that the strong 020 reflection is 624 P. DE S. S,4ffTOS,G. W. BRINDLLY AND IT. DE S. S,41rlOS drawn out into a with broad maxima on either side which corre- spond in position with I indices *]. Honjo et al. indexed thesemaxima as 021 reflectionsfrom a two-layerunit cell with a doubledc-parameter, and some weaker spots also were given odd-numberedI indiceswith respect to the larger cell. Waser analysed,along much the same lines as Whit- taker, the diffraction effectsto be expectedfrom a tubular halloysite particle, but made no detailed comparison between the analytical and the experimentalresults. However, he observeddetails in the diffrac- tion patterns which required "that the layers making up a curved sheet have not becomedisordered through the curving" of the layers; such a model would seemto imply cracksor flaws in the overall tubular struc- ture so that appreciableradial alignment of the layers could be main- tained. Similar modelshave beenshown schematically b1' Hofmann et al. 1962\. The evidencefrom these sourcesand from the present study points, therefore,to somedegree of three dimensionalstructural order in halloy- site greater than that which is normally assumedon the basisof r-ray powder diffraction data. It may be noted that in the supposedlyanalo- gous caseof chrysotile, ihe diffuse scattering along layer linesl appearsto be considerablymore marked than one finds for halloysite,a result which points to greaterorder in the halloysitestructure.

X-ray powder d,ifraction Diffractometer data obtained from samples under a variety of conditionslead to the conclusionthat the initial mate- rial consistsof a mixture of endelliteand halloysitein a ratio of about 2: 1, see Fig. 4a. After gentle heating ( < 100' C.) or after a few hours in vacuo,the 10.05A reflectionof the endellitecomponent is Iargely lost and the 7.2 A reflectionof the halloysiteconsiderably increased (Fig. 4b). No expansionof the 7.2 A spacingoccurs in water vapor, but when the mate- rial is wetted with water for an hour or more, the 7 A reflectionis greatly reducedwithout restoringthe 10 A reflection;a broad band of scattering indicatesa partial but very irregularrehydration. With repeatedvacuum and water treatments the previous historf is repeated. Addition of ethylene glycol to vacuum dried material gives a clear 10.6A reflection which, however, is considerablyless intense than the original 10.0 A reflection(Fig. 4c); evidently the ethyleneglycol re-expandsoniy part of the vacuum collapsedmaterial. Treatment with water doesnot restorea clear 10 A spacing. The most notableaspect of the diffractometerdata is the high degreeof I See, for example, fiber diffraction patterns for chrysotile given by Warren and Bragg (1930,p. 209) and by Hey and Bannister (1948). FIBROUSKAOLIN MINERAL 625

basal plane orientation achievedwhen the dispersedclay is allowed to dry in open air conditions on a glass slide or a porous tile (Fig. 4b). Ordinarily the tubular morphology of halloy-site appears to eliminate such marked basal plane orientation (see,for example,Fig. 1, patterns C and D, given by Brindley and de SouzaSantos, 1963, p. 899), but with the presentmaterial, the basal001 and 002 reflectionsgive intensitiesof the order of ten times that of the 02 band maximum. This behavior is consistentwith the flattenedappearance of the fi.bersseen in Fig. 2a. The

002r I 003rcxl \|

8050+0.2030nn

Frc. 4. Difiractometer traces of Piedade clay with filtered CuKa radiation; H:halloy- site, E: endellite, EGH: ethylene glycol complex of halloysite. (a) Wet clay without prior drying in cavity-type holder; 0019 more intense than 001n. (b) Air-dried material on glass slide; H reflections only. (c) Material of (b) saturated with ethylene glycol; 001n approxi- mately the same as in (a); 001ea;sis not to be confused with 0018 in (a). marked orientation was eliminated only after grinding the dry material twelve times in "Lakeside" cement and drastically reducing the fiber length(Fig.2d). The 20, 13 band between 20:35" and 40o in the powder diffraction ('type pattern of halloysites previously called D" (Brindley and de SouzaSantos, 1963) is a broad band of scatteringwith no marked maxima or minima. The halloysitespreviously called "type C," which showeda small enhancementof basal reflectionswith respectto the 02 band, aiso showedstructural featureswithin the 20,13 band. The presentmaterial, with still more marked orientationalability, likewiseyields a 20, 13 band with maxima and minima of the type C kind. 626 P DL. S. SANTOS.G. W. BRINDLNY AND II, DE S. S,4ITZOS

DrscussroNeNn CoNcrusroNs The experimentaldata for the Piedade clay give rise to interesting questions of classification. Its fibrous and tubular morphology, the ab- senceof platy particles, the existenceof a 10 A spacing in about f of the material prior to dehydration, and the partial re-expansionin ethylene gl1'col,lead one to classify it as a mixture of endellite and halloy- site. The marked tendency of the material to give basal plane orientation, and the evidencefor some degreeof structural regularity indicated by both r-ray and electron diffraction data Iead one to consider whether kaolinite is a more appropriate name, at least for the material with the 7.2 Abasalspacing,with the 10 A material as an expandedform. If one admits that a mineral with tubular or rolled form can exhibit an appreciabledegree of structural order, then the question ariseswhether it should be called a "tubular kaolinite." The present writers have dis- cussedalread.v (P. and H. de SouzaSantos and Brindley,1964) someof the materials previously called "tubular kaolins" and have shown that they consist of mixtures of platy and rolled forms. The Piedade clay, however,is not such a mixture and the observedstructural characteris- tics must be attributed to the tubular particles. Therefore, there is a better justification for calling this clay a "tubular kaolinite," if by "kaolinite" one understands a material with some degree of three- dimensional structural order but not necessarilya platy morphology. The description of the material depends, therefore, on nomenclature. The present writers consider that the terms "kaolinite" and "halloysite" should appll' primarily to platy, and rolled or tubular forms of composi- tion AlrSiaOro(OH)rrespectively; they also considerthat both forms may possessvarying degrees of structural order. The evidence presently available indicates that generally the platy forms tend to be structurally ordered and the rolled or tubular forms to be structurally disordered, so that classificationsbased on morphology or based on structural order Iead generally to the same result. Neverthelessthere are now indications that such classificationsmay not always lead to the same result, and it becomesnecessary to definethe basison which the terms are used. or the It may be argued that -the presenceof a 10 A basal spacing ability to expand to a 10 A spacing is clear evidencefor the presenceof endellite and halloysite. However, this view implies that a platy kaolin mineral cannot exist with a 10 A spacing or be made to expand to a 10 A spacing. Under appropriate conditions, however, kaolinites have been causedto expand(see for exampleWada, 1961;Weiss et al.,1963) though not simply by addition of lvater or ethylene glycol. In conclusion, the Piedade clay can be described as a fibrous kaolin FIBROUS KAOLIN MINERAL 627 mineral, exhibiting 10.0 and 7.2 A basal spacings,having particles of tubular form many micronsin length, with cross-sectibnsof the order of 0.05-01 micron, which normally are aggregatedto give a visibly fibrous character, and showing by r-ray and electron diffraction an appreciable degreeof structural regularity. In the expanded10 A form, the material is an endelliteand in the contracted7 A form it is a halloysite.The halloy- site has an appreciable degreeof structural regularity and can be placed in the type C category of kaolin minerals, as previously described. It is consideredthat the terms "kaolinite" and "halloysite" should be used primarily for materials with platy and tubular (or rolled) particles re- spectively, and that (for the present at least) one cannot exclude con- siderablevariations in the extent of structural order, or disorder,in these diff erent morphological f orms.

AcxNowr,BnGMENTS This work has been carried out partly in the Instituto de Pesquisas Tecnologicas,University of Sao Paulo, Brazil, and partly in the Mineral Industries CoIIege of The Pennsylvania State University. Thanks are tenderedto the RockefellerFoundation, to the Organizationof American States,and to the FundagSode Amparo d. Pesquisado Estado de Sao Paulo, for grants which have made this work possibleand facilitated exchangeof scientific knowledge and personnel.

RnlnnBncns

BnrNor.nv, G. W. aNp Sanr S. Kunrossv (1961) Quantitative determination of kaolinite by r-ray diffraction . Am. M ineral. 46, 1205-1215. -- P. on Souze Serros eNn H. nr Souze Sexros (1963) Mineralogical studies of kaolinite-halloysite clays, I. Identification problems. Am. Mineral.48, 897-910. Hov, M. H. aNo F. A. Bexxrsrnn (1948) Thermal decomposition of chrysotile. Mineral. Mag.28,333-337. Hornaanr, U., S. Moncos axo F. W. ScnnMsne (1962) Das sonderbarste Tonmineral, der Halloysit. B er . d.euts chen K eram. Ges etr|,. 39, 47H82. HoN;o, G., N. Krreuune aNo K. Mrneua (195a) A study of clay minerals by means of single-crystal electron diffraction diagrams-the structure of tubular kaolin. Clay Min. Bul,l,. 2, 133-141. K. Mrneue (1954) Study of clay minerals by electron-diffraction diagrams due to individual cry stallites. A cta Cr y st.7, 5 11- 5 13. KNrcnt, T. (1949) Ocorr6ncias de minerais m6talicos na Serra de Sdo Francisco, 56o Paulo. M ineraEdoe M eta.lurgia, 14, 37 40. Souza Sewros, P. or, Hnlnr.r. or Souza Saxros ewo G. W. BnrNor,nv O96a) Mineralog- ical studies of kaolinite-halloysite clays, II. Some Brazilian kaolins. Am. Minerd.49, 1543-1548. Wloa, K. (1961) Lattice expansion of kaolin minerals by treatment with potassium acetate. Am. MineraJ.6,78-97. Wannnx, B. E. exn W. L. Brecc (1930) Structure of chrysotile, Zei.t. Krist 76,201-210. 628 P. DL, S. S,4NiT"OS.G. W. BRINDLEV AND H, DE S. SANTOS

Sesnn, J. (1955) Fourier transforms and scattering intensitiesof tubular objects.Acla Cryst.8, 142-150. Wnrss, A., W, Trrrr.npepn, G. GcinrNc,W. Rrrren aNo H. ScnArnn (1963) Kaolinit- Einlagerungs-Verbindungen.-Iz t er, Clay.C onJ ., 7963, 287 -305, Wnrlresnt, E. J, W. (1953-1957)Structure of chrysotile,I-Y. Acl'a Cryst.6, 747 748l'9, 855-867;10,145-156. -- (1954-1955)Diffraction of r-rays by a cylindrical lattice, I-IY . Acta Cryst.7, 827- 832; 8, 2,61-27| ; a, 726-1 29.

Manuscriplrecei.ved,, Septcmber 21, 1964; accepted. Jor publ'iraLion,Noztem.ber 1, 1961.