MINERALOGY OF KAOLIN CLAYS FROMPUGU, TANGANYIKA

R. H. S. RonnnrsoN, 16 Kirklee Road., Glasgow, W.Z, Scotland.,G. W. BnrwDLEy,* Department. oJ Phltsics, Llniaersity oJ

Leed.s, Leed,s, England,, aNn R. C. MacrBwzrn, TheMacaulay Insl.itute Jor Soil Research, Aberdeen,

Scotland..

ABsrRAcr

Kaolin clays from Pugu, Tanganyika, have been studied by r-ray, thermal, chemicaland electron-optical methods. They include deposits of weli-crystallized kaolinite and alsoof the 6-axis disordered kaolin mineral. The latter give exceptionally clear r-ray diagramswhich have been studied in greater detail than has previously been possible. Morphologi-cally the disordered kaolin shows small but well-formed hexagonal plates. The cation ex-change capacities have been mee.sured and appear to be related to substitutions in bothtetrahedral and octahedral lattice positions. The main chemical difierence between theordered and disordered clays appears to lie in the amount of tetrahedral substitution.

INrnopucrrorv

The work described in this paper is the result of a combination ofstudies which present an unexpected and, in some respects, unusualpicture of the mineralogy of kaolinite. Briefly, from a study of a numberof kaolin-bearing sediments from Pugu, Tanganyika, two were selectedfor detailed study. Electron micrographs showed, in one specimen,slightly elongated hexagonal flakes, some showing a re-entrant angle

Frc. 1. Electronmicrograph ol PuguK Kaolin, X5000 enlarged to 20,000,showing one crystal of anatase.

* Now Research Professor of Mineral Sciences, The Pennsylvania State University,State College, Pa.

Frc. 2. Electronmicrograph of. PuguD Kaolin, X8000, enlarged to 20,000.

1 1 8

MINERAI.OGY OF KAOLIN CLAYS FROM TANGANYIKA 119

(Fig. 1), and in the other specimen, smaller but extremeiy regular anduniform hexagons (Fig.2). X-ray analysis showed the larger flakes tobe well crystallized kaolinite, but the smaller flakes gave the type ofr-ray powder diagram previously obtained with the disordered kaolinstructure found in many fireclays (Brindley and Robinson, 1947a;Brindley, 1951). Relatively pure specimens of both types of kaolin werereadily obtained and their chemical composition, cation-exchangecapacity, thermal and mechanical properties studied. The kaolinmineral in fireclays is usually associated with considerable quantit ies offine-grained qlrartz and mica, together with other impurities, which haverendered diff icult its detailed examination. ft was thought to be worth-while, therefore, to study in some detail these ordered and disorderedforms of kaolin.

GBor,ocv ol rrrE DBposrt

During the recent exploitation of the massive kaolin-sandstone de-posits in the Pugu Hil ls, west of Dar-es-Salaam, Tanganyika, samplesof the rock were taken from Adit "D" west of the present factory andfrom Adit "K" west of opencast workings previously developed. Thesamples described came mainly from 40-50 ft. from the portal of Adit

"D" anTd from 120 ft. from the portai of Adit ( 'K"-and for brevity wil lbe referred to hereafter as "Pugu D" and "Pugu K" respectively.tBecause both samples originated from the same series of strata, a briefaccount of their geology may be given.

The source rocks of the deposits are the Basement Complex (Archaean)

rocks, which extend to within 100 miles of Pugu and which consist of aseries of migmatites, biotite-gneisses and gneissose granites (Stockley,

1948). Rapid erosion and transportation in the Upper Cretaceous andvariations in land level produced a rhythmic succession of sediments,each cycle beginning with micaceous arkose, slightly kaolinized, passing

up into deltaic deposits of mica-free arkose, contemporaneously kaolin-ized in situ to a considerable extenl (e.g., Pugu K), and ending with adeeper, marine facies consisting of kaolin deposited with finer qvartzand feldspar sand (e.g., Pugu D).

Pugu K is a white permeable kaolin-sandstone containing, in the speci-men tested, 29.47o kaolin, the remainder being sand grains of a narrowparticle size range with a median diameter of 0.28 mm. The sand grains,

although highly polished, are not so completely rounded as in some desertsands. Judging from photographs of Davies and Rees (1945), their ru-gosity or angularity (area divided by area of a sphere of the same mass

t By strange coincidence, the K material turns out to be good kaolinite and the D ma-

terial the distorted structure.

120 R. 11. S. ROBERTSON. G. W. BRINDLEY. AND R. C, MACKENZIE

(Robertson and Emcidi, 1943) ) appears to be about 1.08. Absence ofvery small sand grains, the narrow particle size range of the grains, therarity of mica and the presence of sand-blasted pebbles demonstratethat the sand was subjected to aeolian influence by passing throughdesert conditions on its way to the sea. Accessory minerals include ilmen-ite and zircon. Hematite and goethite occur as "iron-staining," espe-cially where the overburden is shallow.

Pugu D is less white, almost impermeable and very uniform in qualitythroughout; the kaolin content is 52.8/6 and the sand grains, with amedian diameter of 0.18 mm., have a much wider particle size range.

Similar rocks are the kaolinized fine-grained Keuper arkoses at Hir-schau and Schnaittenbach in Bavaria and the coarser-grained arkosesfound near Steinfels, Weiherhammer and elsewhere (Zwetsch, 1950). InGeorgia, U.S.A., kaolin was also formed in the Upper Cretaceous, buthere the sand grains are more angular and there is a great variation inclay content and quality over this area (Mitchell and Henry, 1943;Kesler , 1951).

X-Rav ANlr.ysrs ol Pucu D nmr K Cravs

X-ray analyses of these materials have been carried out with (o) a20 crr;^. diameter, semi-focusing camera, using flat powder specimens, (D)

a 19 cm. diameter Unicam instrument, using thin coatings of powder onthin-walled sil ica tubes of external diameter about 0.025 cm., and (c)the same camera using powders packed in thin silica tubes of internaldiameter about 0.03 cm. Methods (o) and (c) require no binding agent;gum-tragacanth was used in (D). Filtered CoKa radiation (I: 1.78659kX)was employed.

X-ray powder diagrams are reproduced in Fig.3;diagrams I and2 ofPugu K clay and 3 oI Pugu D clay were taken by method (b) and arereproduced without change of scale. Diagrams 4 and 5, of K and D claysrespectively, were taken by method (c) and are shown with two-foldenlargements in order to make clear the details of the lower orderreflections.

Since the K and D clays differ in crystal size, it was considered impor-tant to determine whether the difierences shown in their r-ray diagramscould be attributed to the crystal sizes of the materials. Two fractionsof the K material were therefore prepared, KI and K2, with crystal sizesrespectively 1.4-0.2 pr and iess than0.2,r.r equivalent spherical diameter(e.s.d..); the corresponding r-ray diagtams are numbers I and2 in Fig.3'The reflections from the two samples are practically identical. Impuritylines from quartz, indicated in the key by dotted lines marked Q, aremore prominent in 1 than in 2. The reflections in 2 are somewhat less

MINERALOGY OF KAOLIN CLAVS FROM TANGANYIKA 121

e\ !q

t P '

v 6

d i l

3v|<

v o

. 3 6s q N

9 v , ;

Y x ir i i . X- ^ X: \ ;b 6 ' g

l o r ra E \

I N : :

. N - =- ' a o

6 N

I22 R. 11. .'. ROBERTSON, G. W, BRINDLEY, AND R. C. MACKENZIE

sharp than in 1, in accordance with the smaller crystal size of K2, and

the weakest reflections ftorr K2 are less easily seen, but in al'l essenlial

respects lhe coarse and f'ne Jractions are crystallographically id'entical.

The full pattern of iines from the K clay is indicated in the key in

diagram 1, with asterisks marking the basal (001) reflections, and Q and

,4 indicating impurity lines from quartz and anatase' This diagram

agreed in all respects with the data of Brindley and Robinson (1946) for

well crystallized kaolinite from which the triclinic structure of the mineral

was evaluated. The very close tricl inic doublet, (111) and (ttt;, with

spacings 4.170 and 4.120kX, which is resolved by the best crystall ine kao-

Iinites, is imperfectly resolved by the Pugu K clay, but the doublet char-

acter is undoubtedly present and can be seen in the enlarged diagram,

number 4.It is concluded that the Pugu K clay, both coarse and fine fractions,

consists of a well-crystallized kaolinite together with some fine quartz

and anatase. The presence of anatase, shown by one or two lines only

in diagrams 1-3, was confirmed by heating the clays to about 600" C. to

destroy the kaolin mineral, when the full sequence of anatase lines to-

gether with quartz lines is observed.Crystallographic interest centers mainly in the Pugu D clay, the r-ray

diagram of which shows that it is a very good example of the 6-axis dis-

ordered kaolin mineral f irst discussed by Brindley and Robinson (1947o).

The sample prepared Ior r-ray analysis is of a high degree of purity and

contains no more than a trace of quartz and anatase. The clarity of the

diffraction diagram, Nos. 3 and 5 in Fig. 3, is greatly superior to that

obtained from the fireclays previously studied and justifies a full de-

scription of the results obtained.The diagrams in Fig. 3 are arranged so as to facilitate the comparison

of the difiraction patterns of the well crystallized kaolinite and the D-axis

disordered mineral. The outstanding feature of the latter is the broadly

spreading band of intensity with a sharp low-angle termination at about

4.47 kX, the nature of which is entirely consistent with two-dimensional

diffraction from randomly displaced layers. The well-ordered mineral

gives a series of lines in this angular range. This difference is clearly

shown in figures 4 and 5.The lattice spacings, estimated reflection intensities, and character-

istics of line profiles, such as sharp, broad or diffused, are tabulated in

Table 1 together with the calculated spacings and indices. The following

Iattice parameters were used:

o 5.14(7) kXb 8.91(s) B:104.s'o 7.38(0)

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA r23

Tanln 1. X-nev Da.r.q. lon Pucu D Cr,av. e b-exrs Drson-onnro Keolrl MrNtur.

Characteristics of line profileshkt

001

"02"

002

20r130

131200

003

202131

132201

203132

133202

004

204133

r34203

060331

332061330

005

sharp

band, strongly diffused to higher angles

sharp

sharp

sharp

sharp

difiused to higher angles

very broad

diffused to higher angles

not observed

sharp

difiused to higher angles

difiused to lower angles

broad

124 R. I1. S. ROBERTSON, G. W, BRINDLEY, AND R. C. MACKENZIE

Trsrn l-Continued

1 (est.) I

Characteristics of line profiles

not observed

broad

very broad, spreading towards lower angles

moderately sharp

diffused to higher angles

205t34

333062331

I J O

204

261401

334063332

262400

40326r

Calculated

r .1908

t .2624

r .2476

| . z54J1 .2385\r .22e1J

Additional observed lines, ali very weak, of uncertain indices:tt:7.112, 1.042, 1.032, 1.022, 0.972,0.958 kX.

On the basis of the earlier work the unit cell is assumed monoclinic with

6:104.5". A mean value of 6 sin B is obtained from the 001 spacings

( l ranging f rom 1 to 6) , v iz . ,7 '145 kX. The D parameter is ca lculated

fro..r alooo) : 1.485(8) kX and the o parameter is taken to be b/t/3'

The results in Table 1 confirm and amplify the earlier data and the

conclusions drawn therefrom. Altogether 23 reflections are now indexed

and 6 higher orders of uncertain indices are recorded, as compared, in the

earlier work, with 14 indexed reflections and two others which probably

did not arise from the kaolin mineral. Of the 23 indexed reflections, 22

are of the type (hkl) with ft: 3n,whete n:0,3 or 6' The only reflection

observed with kl3n is indexed as "02" and is the 'two-dimensionzl' hk

band with its characteristic intensity distribution. The results support

the earlier conclusion that the distortion consists of largely random dis-

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA 125

placements of the structural layers parallel to 6 and of integral multiples

ol b/3.One aspect of the data given in Table 1 calls for comment, namely the

fact that most of the observed reflections correspond to two or three cal-

culated reflections in close proximity such as doublets of the type (201)'

(13r+1). The question arises whether a more appropriate choice of axes

or parameters would bring these calculated groups of lines into coinci-

dence. It should be noted, in the first place, that the B-angle, 104'5",

corresponds very closely with a relative shift of successive structurai

layers by -af3 along the a-axis. If the displacement is exactly -a/3,

then B is given by cos F: -a/3c, which leads to a value for B of 103'5' '

With this value of B and with the axial ratio bf a:V3, the groups of

closely spaced lines are brought into coincidence. Under these conditions

a hexagonal cell embracing three structural layers exists which is related

to the monoclinic cell by the following transformation matrix

so that rI HKL are the hexagonal indices, then H:h, K: -h/2+k/2,

and L:h*31', and, the hexagonal parameters are an:5.14(7)kX and

cn:21.43(5)kX. The available evidence suggests that B:104.5" is in

better agreement with the observations than is the ideal value 103.5"

and that the structure is, therefore, more appropriately regarded as being

monoclinic, or at least pseudo-monoclinic, than hexagonal. Although it

has not been found possible to photograph the calculated doublets and

triplets, even at the higher angles where greater resolving power obtains,

the diffusion of many of the reflections shown in diagram 3 of figure 3 sug-

gests that the lattice is not based on a clearly defined hexagonal cell. In-

deed, this would not be expected since in kaolinite there is a clear evidence

that the lattice is triclinic, the translation from layer to layer being only

approximately -a/3 along the o-axis together with a small translation

along the 6-axis making the angle a:91.8o.

Thus, the r-ray evidence shows that the Pugu D clay is a D-axis dis-

ordered mineral of the type described by Brindley and Robinson (1947 a);

the previous data are confirmed and amplified by the exceptionally clear

diffraction diagrams obtained with this clay.

DrrlpnnwrrAr- THERMAL ANnr,vsrs ol PuGU Cr'avs

AT

126 J?. 1I. S. ROBERTSON, G. W, BRINDLEY, AND R, C. MACKENZIE

I

\

VV\

VT\

\

I

/

/

/

/

/

/

/

I

)

J)

)

J

\2

\ 5

\ 6

f,7\ ,

J

\

\

J

J)

I

\

\

o 200 400 600 800 toooTEMPERATURE CC

Frc. 4. Thermograms for colored samples of Pugu Kaolin. Data are given in Table 2 underthefollowingcurveNos.:1-253;2-256;3-252;4-251;5-259;6-257;7-314;8-254.

:

f

5

I

I

2

3

4

5

6

7

I

I

\

\

\

)

)

I

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA 127

AT

TEMPERATT RE oc

Frc. 5. Thermograms for some kaolin and halloysite samples. Data are given in Table 2

under the following curve Nos.: l-249 2-213; T379; 4-282; 5416; 6414; 7418'

128 R. .B. S. ROBERTSON, G. W, BRINDLEY. AND R. C. MACKENZIE

Fr

!

F

a

E

!

o

J

(.) () o (.)

6 o o

F F FE

FF F B

o o :

! U J M

<gE

r

d v

FT

d =

T E

? J U'l* = aoo N - o

+ d vJ V G

k k 6

d . EZ " o

c a t

6

l-l-

r D o b l 6 r €s o $ s t 6 6 6

o - 3

r 9 e> . : 9 . ?

a 5 3> . c a

g M M g

F F F

o ? 6'7 )J r2= - -

7 O C )' i s s

. 4 2

t4Fr=U>zllo

3

E

FN

HA

tr

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA 129

cluded use of a constant amount for ali determinations, although 0.2O g.samples were used wherever possible.

fn order to characterize the materials responsible for the red and yellowstaining on some of the clay samples, the first specimens examined werea range of colored clays from Adits C, D and K, (Fig. 4). Yellow-stainedsamples apparently contained goethite (peak at about 350o C.), with, inthe Adit C sample (curve 2), possibly some boehmite (peak at 595" C.),and in the Adit K samples (curves 5 and 6) a small arnount of gibbsite(peak at about 320o C.). The concentrations of these minerals were inall cases low-never more than about 57o. Since red-stained material(curves 3 and 4) did not show the presence of any hydrous oxide, theiron must presumably be present as hematite. At about 665o C., coloredclays from Adits C and K (curves 1-7) showed a small peak, possibly dueto a trace of illite. Quartz, if present, would not be observable on any ofthese thermograms.

It is well recognized that the thermal effects given by members of thekaolin group serve to distinguish them reasonably distinctly-e.g. in-creasing disorder in the lattice ('i..e. from kaolinite to halloysite) causesa decrease in the temperature of the principal endothermic peak and anincrease in its asymmetry (Grimshaw et a1.,7945 Brindley and Robin-son, 1947 a) . A simple method of numerically evaluating this latter effecthas recently been proposed by Bramao et al. (L952).In interpreting suchthermal data, however, other factors afiecting peak temperature, a{rdshape, e.g. the amount and the particle size of the reactant, must alsobe considered and, if necessary, allowed for. In the present set of experi-ments the sample weight was not always constant and consequentiycorrection for the effect of this factor must be made before a rigorouscomparison of curves in Figs. 4 and 5 can be attempted.

From the sample-weight data in Table 2, it will be observed that thepeak temperatures in column 6 will be slightly too low for curves 256,259 and 379 and too high for 314. Although the conditions prevaiiinghere are not identical with those normally obtaining when the sample isdiluted with foreign material, and the sample weight kept consrant,it is considered sufficiently accurate, in view of the relatively small dif-ferences in sample weight, to employ the weight of reactant as determinedby the ratio oJ the area und,er lhe peak on each curoe to the peak areaJor thepure material (the usual relationship for determination of amount ofreactant: see MacKenzie, 1952) for correlation with peak temperature(Fig.6). As, from all available data, the material which gave curve 379would appear to be practically pure, the amount of reactant here cannotbe calculated as an area percentage and it was considered little errorwould arise by considering the amount of reactant to be 0.175g.-the

130 R. 11. S. ROBERTSON, G, W. BRINDLEY, AND R. C. I,IACKENZIE

weight derived from other measurements (Table 5). The correlation of

peak temperature with amount of reactant (Fig. 6) indicates that all

the Pugu C and K samples are reasonably similar to Cornish kaolinite.

The slightly lower peak temperature iot Pugu K than for Cornish kao-

linite tends to suggest that it is rather less well-crystaliized, while the

slightly higher peak temperatures for the sample labelled Lilhomarge-

a compact segregation of kaolinite-and for the red surface clay from

sample 6, Adit C, seems to indicate, if anything, rather better crystal-

lized material. The difierences are slight, however, and broken lines have

been inserted to cover what might well be the kaolinite region. The

point for Pugu D falls well outside this region and, in fact, Iies intermedi-

ate between the kaolinite and halloysite regions-i.e. in the position

which should correspond to that for the disordered form of kaolinite

found in many fireclays.fn view of the difierence in particle size oI Pugu K and D clays it is

necessary, however, to examine closely the efiects of such variation be-

fore drawing any definite conclusions. For this purpose thermograms

were obtained for the samples Kl (l '4-0.2 p,, e.s d'.) and KZ (10'2p,

e.s.d.)-see r-ray section-and ior K6, a 10.4p', e.s.il' fuaction of a

Cornish kaolinite (curves 5, 6 and 7, Fig. 5). To avoid confusion, the data

WTIGHT OF REACTII IG MATERIAL

Frc. 6. Relationship between peak temperature and amount of reacting material

for various samples : X : Comish kaolinite; O : PuguK; E : Pugu C; V : "Lithomarge";

A:Pueu D; *:Iowa halloYsite.

c3F

eUc

UF

!

UA

I'IINERALOGY OF KAOLIN CLAYS FROM TANGANI'IKA 13I

Tlsr,n 3. Slopa Reuos non KaoltN Seuplps

Curve Number Sample Number or Description

Cornish KaoliniteCornish Kaolinite 50%Cornish Kaolinite 25Vo

Pugu 1(

Pugu C

Slope Ratio

249309192

254

313314259257

0 . 9 6

0 .950 .930 . 9 1

0 . 8 10.640 . 6 00 . 6 4

0 . 8 20 . 8 60 . 7 60 . 7 3

1l

,ilirJ

l J . t

256252251

6l- lo l'.1oJ

"Lithomarge"

416414418

Pugu K1 (l .4-0.2p)Pugu KZ (<0.2p)Cornish K 6 (<0.4p)

5 Pugu D 1 . 2 6

Halloysite (Iowa)

Halloysite (Iowa) 50/6Halloysite (Iowa) 25/s

for these samples are not plotted on Fig. 6, but from Table 2 it willbe appreciated that while both KI and K6, fall within the kaoliniteregion, K2 falls outside. However, although the particle size oI K2 isalmost identical with that of Pwgu D (Meldau and Robertson, 1952),the peak temperature for K2 is not sufficiently low (by about 5") tobring it into line with the value for Pugu D and, consequently, one seemsjustified in definitely distinguishing Pugu K and Pugu D clays.

The "slope ratio" of the thermogram (i.e. tan af tan B, where a is theangle between the perpendicular to the peak and the tangent to thedescending side and 6 the corresponding angle on the ascending side)has recently been proposed (Bramao et at. ,1952) as a convenient means ofdistinguishing members of the kaolin series, provided the particle sizesare reasonably uniform, as kaolinite normally has a smaller slope ratiothan halloysite. It has been pointed out, however, that very finelydivided kaolinite might give a slope ratio in the halloysite region, al-though there is little possibility of the reverse happening. The presentseries of minerals gives values (Table 3) which are of little use in sepa-

r . 9 7r . 8 71 . 9 5

282383385

132 .R. 11. S. ROBERTSON, G. W. BRINDLEIT, AND R. C. MACKBNZIE

rating kaolinite and the disordered form of kaolinite since K2, K6 and

Pugu D (curve 379), all of a similar particle-size, have slope ratios

which are virtually identical. It would appear, therefore, that particle-

size is relatively more important in determining the asymmetry of the

peak than order or disorder in the lattice-except insofar as order or

disorder may encourage or discourage the growth of larger particles. In

halloysite, with a very high slope ratio, for example, the sheets are con-

sidered to be randomly stacked and here the "crystal" determining the

asymmetry of the peak may indeed be the single sheet' From Table 3,

Tasr-o 4. Cnnurcnr. Axar-vsrs ol Snpanarnn Pucu Keor,rns

SiozTiOsAlzo:FezOgFeOMgoCaONa:OKzOH,O+105'c.HrO-105'c.SOrPzOrL I

Fezoa in residue after extraction

Total exchange caPacity

Exchangeable II+ ions

100.59

l . l J

7.4 m.e./ lOO g.2.46 m.e./100 s

M . 5 9I . 3 8

38.121 . 4 3nil0 . 0 60 . 1 00 . 1 20.08

1 3 . 9 10 . 7 r0 . 0 9

tracenil

48.140 .76

. J J . / +

1 . 4 5nil0 . 1 10 . 1 90.040 . 1 4

13.140 . 4 80 . 0 5nilnil

1 .0s4.6 m.e. /100 g.0.0 m.e. /100 g.

l. Pugu D Kaolin.

2. Pugu K Kaolin.

Analyst: A. H. Clarke, M'Sc. (Geochemical Laboratories Ltd.)

however, it would appear that all samples from Adits c and K are some-

what coarse in particle-size, while lithomarge is of the same particle size

order as the Cornish kaolinite.fn conclusion, therefore, the thermal analysis results agree with the

r-ray ones in placing Pugu K in the kaolinite group and Pugu D in the

disordered lattice group and, in addition, demonstrate that, of all the

samples from this deposit studied, this seems to be the only one with a

disordered lattice. These samples have also provided additional data re-

garding the efiect of particle size upon thermograms of kaolin minerals'

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIRA 133

Cunltcar Dar.q

Samples of. Pugu K and Pugu D of less than 5p e.s.d.. were separatedfrom the crude rock by the usual sedimentation procedure using ammoniaas the dispersing agent for Pugu K and ammonia together with sometetrasodium pyrophosphate for Pugu D, and were subjected to chemicalanalysis (Table 4). The free iron oxides in these samples were determinedby the sodium hydrosulphite method (Galabutskaya and Govorova,1934; Mackenzie and Mitchell), and the cation-exchange capacity wasdetermined by the micro-method of Mackenzie (1951). The exchange-able H+ was determined from direct titration measurements on the orig-inal sandstone.

fn calculating the mineralogical analysis, titania was assumed to beanatase, since this mineral was detected by r-ray difiraction; the ironoxide made soluble by sodium hydrosulphite was assumed to be goethite,as shown by difierential thermal anaiysis of yellow-stained material;sulfur dioxide may reasonably be ascribed to gypsum. As mica is veryrare, and feldspar occurs in the sand, K+ is probably present in feldspar.The small amount of Na+ added as pyrophosphate was subtracted fromthe Pugu D analysis, and a correction made for the NHa+ (0.04%) re-placing H+. The sum of the analyzed constituents exceeds 100/6 by0.5570 (after correcting for ammonia) in Pugu D kaolin; it was con-sidered probable that an error of this size had arisen in the determinationof one rather than of a number of elements. A preliminary calculationthrew suspicion on alumina, so the excess was subtracted from the per-centage of alumina in the analysis. No correction was made with Pugu K.

With these assumptions, the chemical analysis oI Pugu D kaolin was

Tasrr 5. MrNpner,ocrcar, Arllvsns or Pucu Keor-rxs

Pugu DKaolin

o//o

Pugu KKaolin

ol/o

Kaolin mineral

Q:uartzAnataseOrthoclaseMoisture (70-105' C.)GoethiteGypsum

Mineral impuritiesMineral impurities without quartz

96.93

1 . 3 80.470 . 7 r0 . 3 10 . 1 9

92.405 . 1 30 . 8 3u . / o0 . 4 80 . M0 . 1 0

99.99

2 . 3 52 . 3 5

134 R. 11. .'. ROBERTSON. G. W, BRINDLEY, AND R. C. MACKENZIE

Taeln 6. foNrc ComsrrturroN oF rnr Cr,nv Mrrnner-s,oN Besrs or O:5.000

Theoretical forKaolinite

Pugu DKaolin

Pugu KKaolin

Ms/2Ca/2NaH

?:33?), ,'

i 33!)' ,,

0.018\? ooo1 . 9 8 2 J - ' " - -

1 . es7l0.037l1.eee0.00sJ

5 . 04 . 0

employed to calculate the mineralogical composition and the ionic con-

stitution of the clay minerals (Tables 5 and 6). The cation-exchange

capacity calculated from the ionic constitution is 7.53 m.e'/100 g. of dry

clay, which is in excellent agreement with the measured value of 7.4

m.e./I00 g., a value which, since it was determined by the micro-method,

is probably l-2/6 low. With no assumptions as to the excess alumina,

however, the calculated cation-exchange capacity is 10.9 m.e'/100 g.,

which is not in accord with the measured value, and which suggests that

the above assumptions are valid. If i t is assumed that there is the same

correspondence between measured and calculated values fot Pugu K,

it is possible to calculate the amount of free siiica Ieft by imperfect

dispersion. In this connection, it may be remarked that the samples

submitted to thermal and x-ray analysis, although from the same rock,

had been better dispersed before separating.AIso from the ionic constitution of the clay minerals (Table 6), it

would appear that there is greater Al+3 for Si+a substitution in Pugu D

than in Pugu K kaolin, that Pugu D contains more exchangeable Na+

(reflecting perhaps its more decidedly marine origin) and also H+ ions,

and that Pugu D appears to have 0.122 morc hydroxyls than is required

by the formula AI2(SLO|(OH)a and Pugu K 0'092 hydroxyls in excess.

ErBcrnoN-oPTrcAL Sruorns

Electron micrographs of Pugu K and D kaolins, of the paper-coating

kaolin Hyd,ratex No.2 from Georgia, U.S.A., and of Cornish Low Vis-

cosity S.P.S. china clay were obtained, and the D-clay crystals were also

AIFe+3Mg

A1Si

oOH

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA I35

T,csr.n 7. Mrcnounnrrrc Me.lsunrurNrs on SouB Kaor,rlr Mrrener,s

Pugu DKaolin

512348

196152

Pugu, KKaolin

GeorgiaHyd.rate*

No. 2 Kaolin

CornwallLow Viscosil,yChina Clay

Length, average, msBreadth, average, mp

(at 90' to length)Thickness, average, mpLength/breadthElongationRugositl,Specific surface, m2/g.

518360

247180

2 2 . 31 . 2 91 1 21 . 9

44.9

ca. 67 .21 . 4 71 . 2 7

c a . 1 . 81 5 . 3

palladium-shadowed for thickness measurement. The results of mrcro-metric measurements on these electron-micrographs have already beenpubiished (Meldau and Robertson, 1952) and, together with the valuesfor more recently determined parameters and with measurements forother kaolins, are collected in Table 7. It should be noted the "elonga-tion" is expressed as the ratio of the crystal l,ength to the crystal breadthd.iaided by 1.155 (the ratio for a true hexagon). No values for the specificsurface of Hydrater No. 2 or of Cornish Lou' Viscosill S.P.S. china clayare available, but for comparative purposes may be cited the valuesfound by Vassiliou and White (1948) for Cornish Supreme china cJayusing a moisture capillarity/temperature technique, and those of Car-man and Malherbe (1950) for an unnamed Cornish china clay usingnitrogen adsorption and air permeabil ity-namely, 23 m.2/g. and 11.4m.'/ 9., respectively.

Electron-optical studies show that a"bout98/6of Pugu D crystals areeuhedral. In Georgia kaolin the smallest crystals (<0.5 p) are euhedral,but large crystals approaching 1 p in diameter almost all have one re-entrant angle. In Cornish Low Viscosill S.P.S. china ciay, euhedralcrystals are rare, and large crystals often have more than one re-entrantangle or even a jagged, broken edge. These observations amplify those ofClark (1950), who showed that crystals of plastic china clay from Corn-wall are thinner and more irregular than in the low viscosity types; andthat Zettlitz kaolin is so irregular in outline that crystal angles arerarely seen. Meldau and Robertson (1952) gave some evidence that theviscosity of deflocculated kaolin slips was closely related to rugosity,but not to exchange capacity.

Gold, sol ad.sorption

Electron micrographs showed that unprotected gold sol particles areparticularly strongly attracted to the corners and edges of Pugu D clay,

136 R. 1{. S. ROBERTSON, G. W. BRINDLEY, AND R. C, MACKENZIE

whilst relatively fewer are attracted to the Georgia kaohn crystals, manyremaining near the crystals without being influenced by thern. LoutViscosity .S.P.S. china clay is more weakly attracting still, except atjagged broken crystal edges where the gold particles form a thickly-studded fringe. Gold particles are also attracted preferentially to re-entrant angles.

These observations confirm that Pugu D kaolin is very well crystal-Iized, the electrostatic charge being mainly concentrated at the 6 corners.fn less perfect crystals the charge is distributed among many pointsalong a longer perimeter and is therefore weaker, while broken crystalshave many points of high charge because of unsatisfied valence bondsand consequently attract many gold particles.

Impurities

fn view of observations on the occurrence of titania as anatase in fire-clays and in sedimentary kaolin (Brindley and Robinson, 1947b; Nagel-schmidt et aL, 1949) the electron-micrographs were scanned for thismineral. Opaque isometric crystals showing hexagonal, and occasionallysquare and triangular outl ines were considered to be anatase (Fig. 1);and this was confirmed by c-ray examination in the case of Pugu D. It wasestimated by counting that anatase crystals are about twice as commonin the Georgi,akaolin as in the Pugukaolins.

Large clouds of 1 ;r diameter and extreme thinness, seen in both D andK clays are believed to be a hydrated iron compound: a coral-like frag-ment was unidentified, but an opaque sphere may possibly be hematite.Otherwise the Pugu and Georgia clays appear to be practically mono-mineralic, whereas the Cornish clay shows considerable numbers of wavymica crystals.

GBNnnal Drscusstor.t AND CoNCLUSToNS

Both *-ray diffraction and differential thermal analysis show thatPugu D kaolin is the disordered type of kaolin mineral found in manyfireclays, and that Pugu K kaolin is a well-crystallized kaolinite inwhich the structural layers are much more regularly superposed, butpossibly not so regularly as in the best specimens of kaolinite. AlthoughPugu D shows a high degree of structural irregularity, it is morphologi-cally very perfect; indeed, the crystals are more nearly hexagonal thanin any other clay known to us. On the other hand, Pugu K is more per-fect structurally but less perfect in crystal shape. The two types are re-garded as having been formed under different conditions; it seems rea-sonable to suppose that fine-grainedeuhedral crystals of D-axis disordered

MINERALOGY OF KAOLIN CLAYS FROM TANGANYIKA 137

kaolin will not develop into larger grained crystals of well-ordered kaolin-ite having re-entrant angles. fndeed, if one accepts the dislocationhypothesis of crystal growth as recently discussed by Dekeyser and Ame-linckx (1951, 1952) it is quite inconceivable that a disordered systemcould, with further growth, develop into an ordered one.

The Pugu D clay provided an excelient r-ray diagram of a 6-axis dis-ordered mineral of a high degree of purity which confirmed and amplifiedearlier data and the conclusions drawn therefrom. On the other hand, thedifferential thermal results amply demonstrated the difficulties whichmay arise in attempting to distinguish members of a series when particle-size effects are superimposed. Although it has proved possible, even whenthe samples were relatively impure, to distinguish kaolinite from D-axis disordered kaolin, further study of the thermal efiects in the kaolinseries arising from order and disorder in the lattice and from particle-size differences would appear very desirable.

As the separated clays were remarkably free from impurities, evalua-tion of their chemical composition was possible. From this, and otherevidence, it seems probable that the cation-exchange capacities of theseclays arise from substitutions in both the tetrahedral and octahedralpositions in the lattice rather than from valences at the edges of the crys-tals-e.g., observed and calculated exchange capacities for Pugu Dagree almost exactly if certain reasonable assumptions are made, andalthough the area of "edge" faces per gram for Pugu D is 4.0 times asgreat as for Pugu K the exchange capacity is only 1.61 times as great.

The chemical calculations show that, while the degree of substitutionin the silica layer is about twice as great in the 6-axis disordered as in theordered minerai, there is relatively little difference in substitution in theoctahedral layer, viz., about 1.93/6 of these positions are occupied byFe+3 and 0.1570 by Mg*' in Pugu D compared with about 1.85/6 and0.2870 respectively, in Pugu K. The main difference, chemically, betweenthe clays, therefore, appears to lie in the amount of tetrahedral substitu-tion.

AcrxowlrIGMENTS

The authors desire to thank Dr.-Ing. Robert Meldau, Laboratoriumfiir Staubtechnik, Harsewinkel, Westphalia, for conducting that part ofof the electron-optical work carried out at the Rheinisch-WestfdlischesInstitut fiir Ubermikroskopie, Diisseldorf (Director: Prof . B. vonBorries), and Dr. fan M. Dawson for that part carried out at GlasgowUniversity; Kenneth P. Wright for preparing specimens of the clay;N. Tunstall and R. F. Youell who assisted in taking the r-ray photo-

138 R. 11. S. ROBERTSON, G. W. BRINDLEY, AND R. C. MACKENZI]1

graphs; Miss E. S. Murdoch who assisted in the thermal and cation-exchange measurements; and the Pugu China Clay Co. Ltd. for the gift

of samples.RrlnnnNcBs

Bomrns, B. voN, aNn Kauscnn, G. A. (1940): Kolloid. 2.,90, 132.Bnaueo, L., Cenv, J. G., Hnnnnrcxs, S. 8., axn Swrnnrow, M. (1952): Soil' Sci.,73,273.BnrNor-nv, G. W., exr Ronrusox, K. (1946) : Min. Mag., 27, 2+2. (1947a), Trons. Brit.

C a' atn. S oc., 46, 49. (1947 b), M in. M og., 28, 244.BmNor-nv, G. W. (Editor) (1951): X-Ray Identification and Crystal Structures of Cla,v

Minerais. London: Mineralogical Society. Chapter II.C.r.nuan, P. C., aNo M.nr-Ernnr, P. r.n R. (1950): -r. Soc. Cltem. Ind'.,69, 134.Crrnr, N. O. (1950): Trons. Brit. Ceram. Soc.,49,4O9.Devtns, W., lNo Rnns. W. J. (19a5) : J. Iron StreeJ Inst., p. 7 | P.Drrevsrn, W., aNn AunlrNcxx, S. (1951): Compt. rend.233, t297; (1952), Compt. rend',

234,44.6.Gar..nsursraya, E., aNo Govorov.t, R. (1934): Min. Suire,91 27.Gnnrsnrw, R. W., Halrox, E., ANo Rolrnrs, A. L. (1945): Trans. Bri.t. Ceratn Soc.,44,

/o.Knsr.nn, T. L. (1951) : M,ining Engng.,3,879.MacrcNzrn, R. C. (1951): f . Colloi.il, 9ei,6,219. (1952), An. Ed.aJoL Fisiol zteg., ll, 159.Macx.nrzre, R. C., nNo Mrrcunr,r,, B. D, In preparation.Mnr.nru, R., .lNn RonnnrsoN, R. H. S. (1952) : B er. ileut. keratn. Ges., 29, 27 .Mrtcunrr, L., aNl HnNnv E. C. (1943) : J. Am. Ceram.!oc.,26, 105 and 113.Nacrr.scnuror, G., DolrNrr.r.v, H. F., lNo Moncolr, A. J. (19a9): Mi.n. Mag.,28,492.RonrRtsoN, R. H. S., aNo Eucinr, B. S. (1943): Natwe, 152, 539.Srocrr-nv, G. M. (1948): Tanganyika Territory Geol. Surt;. Dept. Bull'., No. 20, 1-37.Vassuou, B., exo Wrrrrr, J. Q948): Clay Min. BulI.,1,80.Zwrrscu, A. (1950): Tonind.ustr. Ztg.,74, 766, 196,283 and 313.

Manuscript receiod. Jan. 31, 1953.