THE AMERICAN MINERALOGIST, VOL. 48, NOVEMBER-DECEMBER, 1963

POLYMORPHISM OF THE KAOLIN MINERALS

S. W. Barrnu, Department of Geol'ogy, University oJ Wisconsi'n,Mad,ison 6, Wisconsin.

ABSTRACT

The polymorphism of the kaolin minerals is analyzed in terms of two factors: 1) direc-

tion and amount of interlayer shift, and 2) location of the vacant octahedral site in suc-

cessive layers. Kaolinite and dickite have the same interlayer shift of -|or when referred

to a standard layer orientation. The two structures difier only in regard to the distribution

of the vacant cation site in successive octahedral sheets and the consequences of this

distribution in terms of symmetry, layer distortion, and Z-axis periodicity. In well-

crystallized kaolinite the vacant site is the same in each layer. The structure is triclinic

and may be right-h-anded or left-handed, because either of two octahedral positions, related

by a mirror plane in an undistorted monoclinic structure, may be vacant. In dickite the

vacant site alternates between these two positions in successive layers, creating a two-layer

monoclinic superstructure that is a regular alternation of right- and left-handed kaolinite

layers. Comparison of the two structures is facilitated by changing the standard orientation

of the dickite unit cell to correspond to that of kaolinite. The interlayer shift in nacrite is

]b1 relative to the same axes as for kaolinite and dickite. The layer sequence is that of a

6R polytype, but the pattern of vacant sites reduces the symmetry to Cc and allows selec-

tion of a smaller two-layer unit cell. A modified system of polytype notation is suggested

for describing the kaolins and other minerals in which the actual Z-axis periodicity or

crystal system difiers from that of the ideal trioctahedreal polytype as a result of cation

or vacancy ordering.

INrnooucrIoN

The descriptions in the literature of the individual kaolin structures donot attempt to analyze the relationship of the sequence of interlayer shiftsin one mineral to those in the other two. An understanding of the relativeIayer sequences is necessary, however, if one is to appreciate fully thestructural reasons for the polymorphism of the kaolins.

The polymorphism is most easily analyzed in terms of two factors: 1)

the direction and amount of the interlayer shift, and 2) the location ofthe vacant octahedral site in successive layers. The similarities and differ-ences in the structures of the three minerals caused by these two factorscan be summarized by reference to their f-ray powder patterns. Kaoliniteand dickite have identical interlayer shifts and give powder patterns thatare similar with respect to most of the stronger reflections. The difieringlocation of the vacant octahedral site in the two structures governs thesymmetry and the Z-axis periodicity of each mineral and accounts for theobserved differences in the powder reflections of medium to weak inten-sity. Nacrite has a sequence of interlayer shif ts that is entirely differentfrom that in kaolinite and dickite, and its powder pattern is also markedlydifferent.

Hendricks (1938) first showed the importance of interlayer hydrogen

tt96

KAOLIN POLYMORPHISM 1197

bonds in governing the permissible positions of successive kaolin layers.Figure 1 i l lustrates an ideal trioctahedral 7 A layer in which octahedralhydroxyl ions at the top ot the layer are paired with tetrahedral oxygensat the base of the overlying layer. This OH-O pairing results in the for-mation of long hydrogen bonds, approximately 3.0 A between the anioncenters, that hold the neutral layers together. Hydrogen bond arrange-ments similar to that shown in Fig. 1, although differing in detail, can beformed by several different positions of the layers relative to one an-

o l

a l

| . t i t l/l\

\-il

( ) o H o t z = 4 2 4

I a t o t z = r r o

6 o x tX l o r z = z z o( r 0 JO S i o t Z = O 5 8

r - - t o o t Z . o O o

: | . @ o o t Z = 7 1 6 i

Frc. 1. Normal projection onto (001) of an undistorted 7 A layer of space grorp Cm.The three possible octahedral sites, only two of which are occupied in kaolins, are labeledA, B and C. The second layer has been shifted by -ior, as in kaolinite and dichite, toprovide long hydrogen bonds between the paired OH and O atoms at the layer interface.

other. These wii l be described here in terms of shif ts and rotations of theupper layer relative to a fixed initial layer.

If the initial layer is oriented as in the first layer of Fig. 1, making noassumption as to the distribution of the two aluminum cations and onevacancy over the three available octahedral sites, interlayer hydrogenbonds will result from the following three layer sequences.

1) No shift of the succeeding layer. The hexagonal, or ditrigonal, SioO:r rings in ad-jacent layers may be either exactly superimposed or rotated by +z(60") relative to oneanother.

2) Shi f tof thesecondlayerby -$aalongoneof thethreepseudohexagonalX-axesofthe initial layer, with or without a rotation of the second layer by +n(60o). A shift along

1198 S W. BAILEY

the positive X direction does not lead to pairing of oH-o anions for the orientation

defined in Fig. 1.3) Shift of the second iayer by *]b along one of the three pseudohexagonal Y-axes of

the initial layer, with or v'ithout a rotation of the second layer by +n(60').

If the three sorts of stacking-sequences above are not intermixed, it can

be shown for trioctahedral 7 A layer sil icates that there are twelve pos-

sible polytypes having regular Z-axis periodicit ies between one and six

layers, plus four enantiomorphs (Bailey, in preparation). For diocta-

hedral 7 A layers there are many more regular polytype possibil i t ies

(Zvyagin, 1962), but Newnham (1961) has concluded that certain layer

sequences are favored as a result of the effects of cation-cation superim-

position and of the distortion of the layer due to the position of the vacant

site. The most stable dioctahedral structures turn out to be those of

kaolinite, dickite and nacrite. Radoslovich (1963) has reached a similar

conclusion based on consideration of the directed nature of the interlayer

hydrogen bond for dioctahedral compositions.

Keor,rwnr AND DrcKr[E

The structure of kaolinite was established originally by powder meth-

ods (Brindley and Robinson, 1946; Brindley and Nakahira, 1958) and

Iater refined by single crystal x-ray and electron diffraction techniques(Drits and Kashaev, 1960;Zvyagin, 1960). The structure of dickite, for

which larger and more perfect crystals can be obtained, has been deter-

mined in considerable detail by Newnham and Brindley (1956) and by

Newnham (1961).

Analysis of the structures shows that kaolinite and dickite have iden-

tical layer sequences in which each layer is shifted by -lav as defined in

Fig. 1, relative to the layer below. The two structures differ only in re-

gard to the distribution of the vacant cation site in successive octahedral

sheets and the consequences of this distribution in terms of symmetry,

layer distortion and Z-axis periodicity. This means that if the two min-

erals were trioctahedral, rather than dioctahedral, they would be iden-

tical. They would have a one-layer monoclinic (1M) structure with B:104' and space group Cm, as in the monoclinic form of 7 A chamosite

descr ibed by Br indley (1951).The kaolinite unit cell is similar in shape to that of its trioctahedral

analogue, 1M chamosite, but is distorted slightly to tricl inic geometry.

The three possible octahedral sites, only two of which are fi l led in the

kaolin minerals, are labeled in Fig. 1 as A, lying on the mirror plane of the

id,ealCm structure, and B and C, lying on opposite sides of the mirror

plane. Brindley and Robinson (1946) chose c as the vacant site in kao-

Iinite, whereaszvyagin (1960) and Drits and Kashaev (1960) chose site

KAOLIN POLYMORPHI.SM 1199

B. The two structures so defined are mirror images of one another and arenot distinguishabie by the methods used. Choice of either B or C as thevacant site imposes tricl inic symmetry on the structure due to loss of thesymmetry pianes. The distortion of the unit cell to tricl inic geometry is aseparate anorthic effect, although also dependent upon the location of thevacant site at B or C.

Figure 2 i i lustrates the pattern of vacant octahedral sites in successivelayers in kaolinite and dickite. In well-crystall ized kaolinite each layer isidentical and has octahedral site C (or B) vacant. In dickite the vacant

o@

l o y . r 3

l 6 t . t ?

l 0 t a r I

l--'

I' I,I

o ) X A O L I t T T E b ) 0 t c x t T E

Frc 2. Normal projection onto (001) of the octahedral portions of three layers (labeled

1,2, 3) ol the kaolinite and dickite structures, showing distribution of cations and vacanciesover the A, B and C octahedral sites. In both structures each layer is shifted by -*ot

relative to the layer below. The projected Z-axis vector is shown as a solid line arrow.l'or dickite the cation distribution may be interpreted as related by an z-glide plane orby a c-glide plane, depending on definition of the Z-axis vector. Two sets of octahedralpositions, separated by ab are shown in layer 3 to illustrate the two choices for theZ-axisvector in dickite.

site alternates between C and B in successive layers to create a two-layerstructure. The alternation of vacant sites in dickite tends to balance thestress distribution in the two layers so that the cell 'shape remains mono-clinic. The pattern of vacant sites also creates c a\dn giide pianes paral-lel to (010) and changes the space group to Cc. Thus, dickite can be con-sidered as a regular alternation of right- and left-handed kaolinite layers,in one sense, or as a superstructure of the ideal 1M poiytype due to aparticular ordering pattern of octahedral cations and vacancies.

A/ , . . \

I

I A - r i r . . tA - s i l . . C - ! i t a !

r200 S. W. BAILEY

fn poorly-crystallized kaolins it is quite conceivable that the vacancydoes not aiways occur in the same octahedral site in each layer, as inkaolinite, or alternate regularly between two sites in successive layers,as in dickite. Newnham (1961) has suggested that the diffraction effectsshown by certain fire clays could be interpreted as a random interleavingof right- and left-handed kaolinite crystals, i.e. random choice of C or Bas the vacant site in different layers.

The description presented in this paper emphasizing the similarity inthe layer sequence and the difference in the vacant site distribution in

. - ' - . - . - ' k o o l i n i t c l o l t i c a

d i c k i t c c . l l o f

N c Y n h o m ( 1 9 6 1 )

P r o P o e e d d i c k i t .

c s l l

o r l h o g o n o l 5 - l o Y t r

c c l l

i\i

I

i

t--l+ X t o o t .

Frc. 3. Two alternative 2Jayer cells for dickite shown in relation to a group of sixkaolinite unit cells in the (010) plane. The values of the B angles are only approximate.An orthogonal 3-layer cell, as described for "monoclinic kaolinite," is also shown.

the kaolinite and dickite structures has not appeared elsewhere in theliterature, to the writer's knowledge, and does not seem to be generallyrecognized. One reason may be that the dickite unit cell selected byNewnham and Brindley (1956) and by Newnham (1961) is not easilycomparable with that of kaolinite. Figures 2 and 3 show that the Z-axis ofdickite is inciined in the opposite direction to that of kaolinite, i.e. along*Xr as defined in Fig. 1, and opposite to the direction of interlayer shift.This gives a p angle of approximately 97", corresponding to a unit cell

KAOLIN POLYMORPEISM I2OI

slope of f I o1 per layer in terms of the axes of Fig. 1, but measured in adirection opposite to that of the actual structural shift, which is -]o1

per layer.It is proposed here that the standard orientation of the dickite unit cell

be changed to correspond to that of kaolinite. This proposal has the ad-vantages that it wil l facil i tate comparison of the two structures and that,by having Z inclined in the direction of interlayer shift, i t provides a morerealistic relationship between the unit cell and the layer sequence. The Bangle is then a direct function of the amount of interlayer shift. Thewriter believes that these advantages outweigh the disadvantages ofusing a larger B value and of deviating from the morphological cell chosenby Miers (inDick, 1888). Figure 3 compares the kaolinite unit cell withthe present and the proposed unit cells for dickite.

The new unit cell would change B from 96" 44' (Newnham, 1961) to103o 35'and c f rom 14.424 Fr to t+.750 A. F igo. .26 i l lust rates the geo-metric fact that redefinit ion of the Z-axis direction and repeat distancechanges the n-glide plane in the present axial orientation to a c-glide planein the proposed orientation. Similarly, the c-glide plane changes to an n-glide plane. Convention requires that the origin in the new cell be shiftedby f,futo keep it on a c-glide plane, and it is convenient to move the originin the direction that will place it approximately at the center of a ditri-gonal ring, as in Fig. 1 and in the Brindley and Robinson structure ofkaolinite.

The matrix for the direct transformation from the old unit cell to theproposed cell for axes and Miller indices is given below (InternationalTables, Vol . I , p . 15-17) .

at2 y2 22

T k z b z

4 l z c z

0 1 0

1 0 1

The matrix of the inverse transformation II is the same as I so that thenew atomic coordinates can be obtained by reading matrix I top tobottom as indicated, taking care to subtract { from the y coordinate totake into account the origin shift. The new coordinates are listed in Table1. A powder pattern indexed on the new cell is given in Table 2.

I2O2 S. W. BAILEY

' Tarr-n 1. Drcrrre Arourc Coonnrxarps Rrr,lrrvr ro Nnw Crr,r,r

SrrSrzA1tn \ 2

Oro:Oao{Os(oH)'(oH),(oH)'(oH),

.0288

.5400

.3166

. 8 1 1 6

.0385

.7355

.2412

.0720

.6416

.5756

.0509

.5509

.9635

.3499

.1768

.4966

.3330

.5127

. 2 7 7 6

.2395

.3624

.1687i a n o

.4767

.3551

. 1()()6

.0405

.0402

.2320

.2306- .0061-.0058

.0058

. I J Z J

.1526

. l . l / +

.2948

.2981

.2963

1 Data from Nervnham (1961) transformed to 12: -xt|_ztl yz: -yrl/4; z2:zr. Newcellhas o:5 150 A, 6:8.940 L, c:14.736A, and B: 103o35'.

MoNocrrNrc Keor,rNrrs

A one-layer structure analogous to kaolinite but with octahedral site Avacant would belong to the monoclinic space group Cm, as noted byGruner (1932), and it is tempting to speculate that this may be thestructure of the "monoclinic kaolinite" from Yugoslavia described byKrstanovic and Radoievic (1961). Alth-ough originally described relativeto a large orthogonal cell with c:21.5 A and Bczggo, this is probably notthe smallest cell possible. The only two permissable monoclinic spacegroups for 7 A layer sil icates are Cm and Cc, and the smallest unit cellsfor these two space groups contain one and two layers respectively. Itshould be possible, therefore, to select a smaller monoclinic-shaped cellwith c:7 .37 A and Bczlg4o for the Yugoslavia material, as in Fig. 3, if theproposed structure is correct. Approximate structure amplitudes, d-values, and powder intensities to be expected for this structure are givenin Table 3. A layer distorted to a degree intermediate between that re-ported for dickite (7.3" tetrahedral rotations) by Newnham (1961) andthat found for kaoiinite (11.1o rotations) by Drits and Kashaev (1960)

has been used as a model for the intensity calculations. Unfortunately, nopublished data are available for comparison purposes. An alternativewould be an average structure in which the vacant site is distributed atrandom over the three possible octahedral positions.

NecnrrB

The structure of nacrite that is presently accepted was determined byHendricks (1938). It is described as a six-layer structure with 0 close to

T.lsr.n 2. Drcxrm Pomnn Parrnnx h.prxno on Nnw Crr,r-

7 . 1 5 2

4 442

4 . 3 6 2+ . 2 6 44 . 1 1 93 9543 . 7 9 0

3 . 5 7 8

3 4283 2613 0952 .9362 7942 657

2 558

2 .505

2.4122 s86

2 324

2 211

100

30

oo2!ozo| 111-

1 1 002l1t21 1 1o22

l.004l 1 2o23\141 1 30241 1 5

J tsoI 2oz( t t t

I t t 2lzoo025006

\204[ 134l2o2220o42026224043

Izo6| 1 a L

t 116044

lzo+l t c z226

foos\r tB045

7 .1624 4704.4394 3684 2674 . 1 2 +3 . 9 5 83 . 7 9 23.6 t23 5813 4303 . 2 6 33 0982 9372 . 1 9 52 6582 . 5 6 12 . 5 5 72 . 5 2 72 .5092 .5032 4122.3872 . 3 2 42 . 3 2 42 2122.2072 1842 . 1 3 42 . 1 0 62.0622.O241 . 9 7 5t . 9 7 41 . 9 3 91 . 9 3 51 . E 9 61 .8611 . 8 5 71 .8481 .806| . 7 9 01 .790| 7621 . 7 1 51 . 6 8 71 6861 . 6 8 31 . 6 8 31 .6691.6671 . 6 5 71 . 6 5 11 . 6 5 01 .6421.6401.6271 . 6 1 11 . 6 1 11 5911 . 5 8 7

3020651 05 5

90

208

108

l 02

3 5

.50

2l 5

95

l . )

40

5

10

l 5

8

1 0

33

50

5

3

1 0

10

100a)\ r c

341""3 82860

45q)

, , i86

1956681

1 6 \ r ,r { - '8 l

28 t50A)2I

6 o i ^ _^ - / d o

"ol7 l- l 1 25l

2 .183 22 .132 22 .10+ 52062 22 0 2 6 3

t . 9 7 4

t . 9 3 7

1 .896I .860.l

I| 8s2)I .805

| .790

| 764t . 7 1 7

I .685

I 668

1.6so ltIr .643J

1.626

1 . 6 1 1

1 589

2242413121502421522401 5 1208136

31024r

;;),',\.1 ) '

fr+lut).5,!,

rzi+o24)

:),

il';l'

) L J Z

\zr4forBl r t \

Specimen from Geneva mine, Michigan. fo6u measured visually from 114.6 mm film, CuKc radiation.Iccl=mlpF2 computed from parameters of Table 1 and adjusted to relatiqe scale.

I 457

1 . 4 3 2

1.428

I . J / J

r . sszlI(

r s s r I

1 . 5 2 61 . 5 0 7

| 489

1 469

1 . 5 7 2I . J J 6

t . J . t J

1 5501 .5491 . 5 2 41 5091 .4901 . 4 8 81 . 4 6 81 4591 4571 .456| .4341 . 4 3 2I 430t .428I .4041 . 3 9 7| 392I . 3 9 11 . 3 7 6| 3761 . 3 7 5

I 3611 . 3 2 91 . 3 1 91 . 3 1 7| 296r .289t .287r . 2 8 51 2801.279| 2641 . 2 6 3| 262| . 2 5 41.2521 2391 2381 2351 2341 2341 226r 2241.22rt .2201 .2081 2061 2051 2051 1991 194t . r 9 21 1 9 11 . 1 8 91 . 1 8 9I t821 . t 7 9

1 { , .

i { ";],,,

;?),'sl

Ir ln i loz)

t'il^

ll,.sJ,r\^

I

N:

ii',.1tr'i'

i 54

1r$\206/ 1s3I22B3t2o47

loooI33t226

looz1 334l.r:o{zzal .0 ,0 , 10

\332227

3

20

3

33

40

5

l . )

t0

3

I .3e81fB 10

r sez)

1 5

1 . 3 6 2 31 .329 .3

1 . 3 1 8 3 5

1 . 2 9 6 2

I . 2 8 8 1 0

r . 236 I J

2 , 2 , t O

! t l f iI208o49261262402260404066338334264400t l l1'to2623534063 1 6421354350

0 , 4 , 1 0159420J J O

0, 0 , 122 , 0 , 1 21 , 3 , 1 02664262, 4 , IO421

1 . 2 6 3

1 . 2 5 3 l . )

10

T,tsrr 3, Trrnonnrrc.nr- X-Rev P,s,TTnnN oF "MoNoclrNrc KAoLrNrrE '

d"ot" I.,rcF""t"hk1hkl Iu c a l c Icu l"r

00102011011102l0021 1 1r1202220I130t t2131200003202t J l

1180402271 1 '

201220041o23222203221132ll3o42133202223r14004222311247150151240o24204133312310or3

46

49t92782+ l

M29A t 7

361856

4l91672058

294310332 122543 15928

442825t9t 1

3044

472427l 166I 3

t61 145

28

30

8

19

l l

Z J

10

7. tsr4 .4554.3524.1242 .78 r3 . 5 7 63.4123 .1032 .789

t q q r l

2 . 5 1 52.5032.4942.3842 . 3 2 62 . 3 1 42.3062 . 2 2 82.216\2 .2r0)2 .1952 . 1 7 62 . 1 2 72.1022.062r . 9 7 91.969\1 .96sJ1 9261 .8911 .8611 .8461 .8081 . 7 9 2 \r .7881t .706t .682t . 6 7 9r . 6 7 8t .6641 . 6 6 11 .6591 .6551.6431.64rJ1 .634\r .628)

242 32151 22152 39241 29r34 50224 34203 6ltt4 1313 34311 35060 8333T 78243 2152 rr223 29332 36061 27115 22330 23005 34153 2r242 18205 35044 t4134 273r4 18333 16062 40312 1r331 49025 23225 39244 30153 18135 83204 78261 3740I 54115 t9402 38260 38154 32243 163s4 25063 23224 37332 16262 63400 31

1 6091 .6051 . 5 6 91 . 5 6 41 . 5 5 81 . 5 5 11 .5461.5431 .5281 . 5 1 71.485\r .4841r .479r .474

| .4281.4221 .395r .3941 .3861 . 3 8 1r . 3 7 51 .3711 .368r . 3671.3621 .3311.3281 .3221 .319l . 3 l l1 .284| .2821.2811 . 2 7 71 276r . 2 7 61 .270I .265r .2f f i1 . 2 5 7I .2551 .252r .247

44

12\ ,o) ' ) 1 "

-

3l4 l21132 l2)2 l

if"7l1 l41, ll1zere)8,l4 l

1 Itz;l

10059841 119783633t2r J ti ( i

39l.')

93943

4

r )" t7j

18J

410,\

, L I

L

1 123c)

O,J

+

o)

1 l(1 lo l

1 Powder pattern intensity I:mLpF2, adjusted to relative scale, where z is multi-plicity and 12 is the Lorentz-polarization factor.

KAOLIN POLYMORPHISM 1205

90o. The space group is Cc, although there are only a few deviations froma pseudo-space group of R3c. Relative to the orientation of Fig. 1, eachlayer is shifted by $ Dr and alternate layers are also rotated by 180'. Thisis an entirely different layer sequence from that in kaolinite and dickite.Another significant difference is the interchange of the conventional X-and Y-axes of layer silicates. The glide planes of space group Cc in thiscase are found to be normal to the 5.1 A axis, which is normally X but nowmust be labeled Y. Hendricks accounted for the similarity to the pseudo-space group R3c by fixing the coordinates of all atoms except the octa-hedral Al according to R3c requirements. The AI atoms were positionedaccording to requirements of space group cc. This amounts to a partialfilling of the general position of space group R3c, as Cc is a subgroup ofR3c.

Several years ago the writer made a single crystal study of nacrite inconnection with the discovery of the mineral in iron ore at the Tracymine in upper Michigan (Bailey and Tyler, 1960). It was determined thatthe space group is Cc and that the conventional X- and y-axes are inter-changed, in agreement with Hendricks, findings. Ilowever, the smallestunit cell is of monoclinic shape and contains only two kaolin layers. Twoalternative two-layer unit cells exist, exactly as in the case of dickite,with B values of approximately 100" and 114". The Z-axis is inclined inthe direction of structural interlayer shift in the unit cell with p:II "and is inclined in the opposite direction in the cell with 0ry100.. Theformer unit cell is preferred, therefore, and has been used for indexing thepowder pattern listed in Table 4. This pattern supersedes that given byBailey and Tyler (1960), in which small amounts of dickite impurity wereIater detected.

Discovery of the two-layer nature of nacrite does not mean that Hen-dricks' six-layer structure is incorrect, only that it can be described on thebasis of a smaller unit cell, a fact also recognized by Newnham (1961).The calculated powder intensities given in Table 4, computed on thebasis of layers distorted in the same manner as in dickite but stackedin the sequence postulated by Hendricks, agree closely with the observedintensities. This means that the layer sequence is that of a true six-layerstructure, which would have rhombohedral symmetry R3c if triocta-hedral (6R). The pattern of vacant octahedral sites causes the loss of thethree-fold axes, reducing the symrnetry to Cc, and allows selection of analternate z-axis that has true two-layer periodicity. Another interestingfeature of the structure is that B deviates from the relation cos-r [- 2a/ 3c]by 1*", so that the interlayer shift is slightly, but significantly, greaterthan }o per layer (where a=8.9 A;. Oickite, on the other hand, has aninterlayer shift almost exactly ]o per layer (where oszS.1 A).

S W. BAILEY

Taeln, 4. Necnrrr Poman Plrrntx INlnxno oN 2-L.tvnn Cnr.r-

Icalc

7 t784.441

4 . 3 6 1

4 . 1 3 0 7 0

3 9+3 20J .6291 53. s88.J 803 . 4 7 6 2 03.413 203 1 1 7 73 061 202 929 102 . 6 7 5 2

2 57 t 10

10030

80

026

Iaosi314224

(zz1|l423I rsII s14I s13j t r o\ 122I L1i

I t:2

3 1 5

427

' i s rz

loosi 336

5

5

5

4

5

I 6821

I,rrrj

1003446\ -^,rlun-lrool

108

7620i o

61 8

J

2

5r'J

I1 i 8s)6t\,1 ) -267t ln1 )

z \ ^) J

1 )

24 \ "

I, I1 4 l l 8r)

I J

l 13)o

r lI

1 f 97l

0021 1 1

Jzo2I t ro) 1 2\2001 1 1113004204r121 1,12021131 1 5

\oo6116

I J I DI

12041.0233 1 1

( . i i i

\+ot220316024

! zrz\224I rooI t r i

406( c t c| - '1

!2091 2 2 5l : r zI r ro0253 1 3223226008

(+ozI zoal :rs

7 1864 . 4 3 84 .3834 . 3 5 24 . r 3 74.O793 . 9 3 83 6333 . 5 9 33 4773 4133 . 1 1 93 0592 . 9 2 42 6712 . 5 7 32 . 5 7 r2 . 5 6 42 54r2 5332 5222 . 5 1 82 4382.4222 . 4 0 42 3952 3t92 .2872.2802 2672 . 2 4 52.2192.2092 t762 tt42 .O922.0692 0682.O392 0331 . 9 9 91 9691 . 9 6 01 . 9 4 81 . 9 3 81 . 9 3 1r . 9 1 71 .8961 . 8 3 8I . 8 1 6r . 7 9 7t -783

t . 7 7 3

I 755

1 . 7 3 6

I 7 1 0

1 6271I

1 .61eJ

1 .605

1 589

1 . 5 7 4

1 . 5 5 8

1 . 5 3 1

1 5 1 4

1 486

1 473

1 460

8 1 5

1 7531 . 7 3 91 . 7 3 41 . 7 0 6I 685I 6841 .6831 . 6 8 11 680| 6791 . 6 7 61 6681 . 6 6 51 .660l . 0 5 /

| 6521 6271.6221 . 6 1 81 . 6 1 41 6051 .6051 5871 5801 . 5 7 9| . 5 7 21 . 5 5 91 5 5 51 . 5 4 81 . 5 3 61 5291 . 5 1 3r .492

10 1 .49010 1 .488

1 . 4 8 51 .484t . 4 7 3I 465t .462I .4611 4581 . 4 5 81 . 4 5 11.446| 444| 4+1

10 t .4371 .4191 .4121 3851 3'19

Itf8

:l1 l

i t4

1 l" r23

l 4

2 . 5 3 3

2 5 1 82 432

8 2 0 .l'

1

n

,r)

zl

11,z)

l).;\;i '

specimen from Tracy mine, Michigan. Iob" measured visually from 114.6 mm film, cuKa radiation.

I6b:mLpF2 computed for dickite-tpye layer in unit cell with a-8 9094, 6:5.146A, c:15.6974, and

B:113"12 ' .

2 . 4 0 4

2 3212.287)

IBI

2 . 273 )2 247

2 2t3

2 . t 792 rrr2.092

2 07r

2 039

1 996

1 . 9 7 0

1 937

1. .9171 8q8I a42I 8 1 81 . 7 9 7

1 . 7 7 4

60

40

5

5

2

255

7

2

2

J

l . )

l 5

25.5

.5

1 . 4 4 t B 2

I 4 1 5

1 . 3 8 1

KAOLIN POLYMORPHISM

Trst-n 4-(continued,)

1207

IcalcT .

1 3 4 2 8 3

1 . 3 1 7 B 3

il'.

'j,

J 3 , 1 , 1 1\$zIooo)ozeI 137l z r tlm| .2, 2, 10

1 041

r . 3 7 01 . 3 6 61 .3601 3571 . 3 5 61 . 3 5 5r . 3 4 3| .3391 . 3 2 41 . 3 1 41.2871 2861 . 2 8 3t .2821 2811.2811 2781 2 7 71 .27 I1 . 2 7 01 . 2 6 71 . 2 6 61 2611 2591 . 2 5 5I . 2 5 5

B -

6023 1 8043l l D

s34

7167t+240s352435362445310 , 0 , 1 27182426295303, 1 , 134267rg0 , 2 , 1 16212 , 2 , 1 26, O, t2

t . 2 4 71 . 2 4 31 . 2 4 21 . 2 3 5L . 2 3 51 . 2 3 5r . 2 3 1t . 2 3 Lt 2 2 71 . 2 2 6t . 2 2 5t .2091.2071.2071 1981 . 1 9 51 . 1 8 61 . 1 8 4r . I82t . t 7 61 . 1 7 01 . t 6 7l . l o 5

1 . 1 6 41 . 1 6 01 . 1 5 9

,,,J"

At the present time a three-dimensional refinement of the nacritestructure is being carried out in collaboration with I. M. Threadgold.Further comment on nacrite is reserved unti l completion of this refine-ment.

PolyrypB Svuror,s

Smith and Yoder (1956) have used structural symbols, such as 1M,2M1, and 3T, to designate the Z-axis periodicit ies and the crystal systemsof mica polytypes having different stacking sequences of 10 A layers.Their polytypes were derived for trioctahedral compositions, but haveproved valid for the dioctahedral micas as well. No example is yet knownof a mica for which the layer periodicity or the space group is changedfrom that of the ideal polytype as a result either of location of the vacantoctahedral site, if dioctahedral, or of cation ordering. In the kaolinminerals, however, the location of the vacant octahedral site has causeda reduction of the ideal symmetry in all three minerals and a change inthe layer periodicity in two of the three.

The writer believes that a modification of the standard polytype nota-tion is desirable for describing structures in which the actual Z-axisperiodicity or crystal system differs from that of the ideal trioctahedralpolytype as a result of cation or vacancy ordering. A considerable gain

1208 S. W. BAILEY

in clarity and in structural detail can be obtained in such cases by a nota-

tion that relates the actual structure to that of the ideal polytype' The

symbol of the ideal polytype has special structural significance because it

refers to the specific sequence of interlayer shifts used in the derivation of

the polytype, thus defining the positions of all atoms in the structure to a

first approximation. Cation or vacancy ordering taking place within the

structural framework provided by a given layer sequence is a secondarystructural effect, although it may have a very drastic effect on the result-ing size, shape and symmetry of the unit cell.

A suggested procedure is to list two symbols, first the symbol appro-priate to the actual structure and then, in brackets, the symbol of the

ideal trioctahedral polytype with a subscript attached, such as [1M1] or

[2Mr-".u1, to indicate that it is ordered. Only one symbol is necessary if

the ordered structure retains the periodicity and symmetry of the idealpolytype. The notation for the three kaolin minerals under this procedurewould be:

kaolinitedickitenacrite

trc [ tMo]2M [ lMo]2M [6Rol

Zvyagin (1962) has made a systematic study showing that it is possible

to derive a total of 52 regular dioctahedral 7 A polytypes with Z-axisperiodicit ies between one and six layers. Zvyagin's system of notation is

inconvenient because it gives the same symbol to all unit cells of the samesize, shape and symmetry without regard to their different structures.Thesystem of notation suggested in this paper takes the difiering structuresinto account by incorporating the symbol of the ideal polytype, as illus-trated below for Zvyagin's eight two-layer monoclinic polytypes of spacegroup Cc.

Zvyagin

NotmberI,3I,4IV,3IX,1v,4tI,2rrl,lV,1

Symbol

2Mr2Mt2Mt'2Mt'2Mz2Mz2M,,2Mz'

Suggesteil Symbol2Mr2M [ lMo]2M [2Tol2M [1To]2Mz2M [6R012M [zHr-.a]2M [20.]

The suggested notation restricts the use of number subscripts to sym-bols of the ideal polytypes. This keeps subscripts at a minimum and, inparticular, reserves the symbols 2Mr and 2Mz for the theoretical 2Mrand 2Mr 7 A polytypes, whose interlayer shifts are analogous to the

KAOLIN POLYMORPHISM r2w

shifts within the octahedral sheet in the mica polytypes of the samedesignation.

It is essential in the suggested notation that the symbols of the ideal7 A polytypes, some of which are used above, be defined relative to a se-quence of interlayer shifts. A systematic derivation and the proposednomenclature will be given in a separate paper (Bailey, in preparation).

AcrNowr,roGMENTS

Acknowledgment is made to the donors of The Petroleum ResearchFund, administered by the American Chemical Society, for support ofthis research. The calculations were performed on a CDC 1604 com-puter, made availatrle by support of the National Science Foundationand of the Wisconsin Alumni Research Foundation through the Univer-sity Research Committee. E. W. Radoslovich kindly furnished a copy ofhis paper prior to publication.

RrnrnrrcrsBerr.tv, S. W. ,lNo S. A. Tvrrn (1960) Clay minerals associated with the Lake Superior

iron ores. Econ. Geol.55, 150-175.Buxntnv, G. w. (1951) The crystal structure of some chamosite minerals. Mi,nerol,.Mog.

29,502-525.M. N,trlnrnn (1958) Further consideration of the crystal structure of kao-

linite. Miner al. M ag. 3t, 7 81-7 86.- AND K. RosrNsoN (1946) The structure of kaolinite, Mineral. Mag.27r 242-253.Drcx, A. (1888) On kaolinite Mineral,. Mag.8, 15-27.Dntrs, v. A. erqo A. A. Kasne,nv (1960) An r-ray study of a single crystal of kaolinite.

Kristall o grafiya, 5, 224-227 (Eng. transl. pp. 207 -210).

GnuNtn, J. W. (1932) The crystal structure of kaolinite. Zeit. Kri,st.83, 75-88.Hnxonrcrs, S. B. (1938) The crystal structure of nacrite Alros.2SiOr.2HzO and the poly-

morphism of the kaolin minerals. Zei.t. Krist.lO0. 509-518.Knsr,lwovr6, r. mqo S. RaooSslai (1961) Monoclinic kaolinite from Kodevje mine, yugo-

slaia. Am. Mineral.46, 1198.NrwNnaar, R. E. (1961) A refinement of the dickite structure and some remarks on poly-

morphism in kaolin minerals. Mineral. Mag.32,683-704.G. W. BnrNor-rv (1956) The crystal structure of dickite. Acta Cr1st.9,7S9-764.

R.a.noslovrcn, E. w. (1963) The cell dimensions and symmetry of layer-lattice silicates.IV. Interatomic forces. Am. Mineral,. 4a, 7 6-99.

Surrn, J. v. arvo H. s. Yoopn (1956) Experimental and theoretical studies of the micapolymorphs. Mineral. M ag. 31, 2W-235.

ZvvacrN, B. B' (1960) Electron-difiraction determination of the structure of kaolinite.Kristal,lograf.ya, 5, 40-50 (Eng. transl. pp.3242).

--- (1962) Polymorphism of double-layer minerals of the kaolinite type. Kristallo-graf,ya,7,51-65 (Eng. transl. pp. 38-51).

Manuscri.pl, receiaed., April, 24, 1963 ; accepteil for publication, August 5, 1g63.