CRYSTAL STRUCTURE OF MAGNESIUM-VERMICULITE
A. McL. M.q.rurBsoNaxn G. F. W,lr-ren' Dittision of Ind'ustriol
C hemistry, CommonweotthScienilrtc and'I nd.ustrial Research
Organization, M elbour ne, A ustr alia.
Assrn,{cr
A single-crystal r-ray analysis of Mg-vermiculite has located the interlamellar water
molecules and exchangeable cations in definite positions with respect to the adjacent
molecules tend to form hydration shelis. The importance of the hydration behaviour of
the cations rather than direct electrostatic interaction between cations and silicate layer
surfaces in iocating the cations is emphasized.
A regular distortion of the surface oxygen network of the siiicate layers from the ideal
hexagonal form is reveaied by the difiraction data.
Vermiculite must be regarded as a true clay mineral since it is formed
in the clay fractions of certain soils (walker, 1950o). The characteristic
properties of this mineral, such as high cation exchangecapacity (Walker,
1O[], tO+O;Barshad, 1948), the ability to form complexeswith organic
substances(Walker, 19506;Barshad, 1952) and a variable interlamellar
distance depending on the exchangeablecation present and the humidity
of the sample (Barshad, 1949; Milne and Walker, 1950; Walker, 1951),
bear a striking resemblanceto those of montmorillonite. This similarity,
moreover, extends to their structures which, in so far as they are known'
consist of complex silicate layers interleaved with layers of water molecules carrying exchangeablecations.
The detailed structure analysis of montmorillonite has been prevented
mainly by the small grain size of the particles, and the configuration
of the interlamellar water molecules and exchangeable cations in this
mineral is not well understood. Mg-vermiculite, on the other hand, is
structure of the silicate lavers and of the relationships between adjacent
231
232
A. MCL. MATHIESON AND G. F. WALKER
silicate layers have appearedwhich are not in accord with previous views.
Earlier workers on the crystal structure of vermiculite (Gruner, 1934,
1939; Hendricks and Jefferson, 1938o) were in general agreement regarding the structure of the silicate layers and their relative disposition when
viewed along the D axis, but neither obtained experimental evidence of
the structure of the interlamellar water layers. Hendricks and
Jefferson
(19380) proposed a hypothetical structure for the interlameilar water
which was based partly on Bernal and Fowler's (1933) concept of the
tetrahedral charge distribution of a water molecule in liquid water.
Neither Gruner nor Hendricks and Jefferson were aware, at the time of
their work, of the high cation exchange capacity of vermiculite, and
more complete information on the dehydration characteristics of the
mineral has since becomeavailable.
ExpnnruBr.trAr,
The vermiculite used came from Kenya and was obtained through the
courtesy of Mr. G. E. Howling of the rmperial rnstitute, London. The
sample has a cation exchange capacity of 130 milliequivalents per 100
grams, all the exchange positions being occupied by Mg2+ ions. The
chemical analysis (Table 1) shows no alkalis, indicating that no interleaved layers of mica are present. small sections,free from cracks, were
cut from cleavageflakes and gently pressedflat to minimize any distortion arising from the cutting operation. The approximate dimensions of
the crystalsusedin obtaining the diffraction data were0.5X0.2 X0.2 mm.
Rotation films indicated that, for the three 9.18 A a*es disposedat
Taerr 1. CsnurcarO"
"trii;iTonuura
or Knxva Vnnurcur.mr
wt.7o
34.04
sio2
Al203
FezO:
1J.J/
8.01
22.58
0.00
0.00
0.00
19.93
Mso
CaO
NurO
KzO
HrO
99.93
Mg2+o::
1
(Mg: mFe+30raAlo ro) (Alr zesizzz) Oro(OH)2. 4.32 lF'zO
Or
st
[CrD4Or0(OH)2]-o
ft.32H2O. 0.32 Mgz+l+o.er
CRYSTAL STRUCTURE OF MAGNESIUM-VERMICU LITE
120o with respect to each other in the cleavage plane, the sharp 3nlayer line spectra coincide in position and intensity, but that marked
differencesoccur in the intensity distribution of the (3iz* 1)-layer spectra. Oscillation films disclosedonly one axis with equivalenceof both
sharp and diffuse *n- and -n-layer spectra. The crystal system is
therefore monoclinic. Moving films of the zero, first, second and third
Iayersabout the D-axiswere recordedon an equiinclination Weissenberg
goniometer (Mathieson, 1951).The cell constants obtained are in good
agreement with those of Gruner (1934), but difier from those of Hendricks and Jefferson(1938a)in the selectionof the B-angle.
Hendricks and
Jefferson
a
b
e
B
Space Group
5.3
9.2
28.57-28.77
97"09',
Cc or C21c
5 .3 3
9 .1 8
2 8. 8 5
93'15',
Cn or C2fn*
Mathieson and
Walker
A
s.33
A
e.18
2 8. 9 0A
97"
Cc
* Reported as Cc or C2/c.
Comparison oI the h)l a\d h3l moving-film photographs shows that,
for B:97", the unit cell correspondingto the sharp fr:32 spectra is
centered on the C Iace and contains one silicate Iayer (c:14.45 A).
Inclusion of the diffuse kI3n spectra requires a doubling of the c axis
@
-@
Frc. 1. (a) Electron-density distribution on (010) for Mg-vermiculite. Contours for
octahedral (C) atoms at intervals of 4e.A-,, for tetrahedral (D) atoms and oxygens
2e' A-2, for HzO and exchangeableMg2+ at 1e.A-2. The dotted line representsthe le'A-'
Ievel. Distances parallel to the s-axis are indicated at the right-hand side of the diagram.
234
A. MCL. MATHIESON AND G. F. WALKER
II
I
R
I
l
Mg"
r@
H.o
D-
c-
t
I
Frc. 1. (b) The corresponding crystal structure of Mg-vermiculite projected on (010).
P, Q, R and. Z refer to silicate half-layers; Q' and R', sheets of water molecules; C, (Mg, Fe,
Al) octahedrally-coordinated atoms; D, (A1, Si) tetrahedrally-coordinated atoms; 01,
oxygens and hydroxyls at z:0.037; Ozt, Oza, 03, oxygens at z:0.I14; HzO, interlamellar
water moleculesl Mg2+, interlamellar cations. An area, (a/2)XGl4), of the unit cell
selected by Hendricks and Jefferson (1938a) is indicated by the broken lines.
CRYSTALS:TRUCTUREOFMAGNESIUM-VERMICULITE 235
(i.e., c:28.90 A). The generalreflexions,hhl, occur only for hlk:2n.
The space group for the structure, based on this value of B, is Cc or
C 2 / c . I f t h e v a l u e 0 : 9 3 . 5 ' i s u s e d , t h e k : 3 r a s p e c t r aa r e a c c o u n t e df o r
by a cell centered on all faces and containing two silicate layers. The correct space group for this unit cell is either Cn or CZf n, not Cc or C2f c
as reported by Hendricks and Jefferson(1938o). The relation of their
unit cell to the one selectedin the present study is shown in Fig. 16.
Intensity data for the 001,h)l and h3l, rcflexions were recorded on the
Weissenberg goniometer using filtered Mo radiation. For each set of
reflexions, two packs, each of four films interleaved with tin-foil to increasethe inter-film ratio, were exposedfor 100 and 5 hours respectively.
Intensity estimationswere made by comparisonwith a set of standards
obtained by timed exposuresof a selectedindividual reflexion oscillated
over a small angular range. For the difiuse reflexions, a crystal was
chosenwhich gave some indication of diffraction maxima rather than the
very diffuse streaks produced by most specimens.Filtered Cu radiation
was utilised to provide adequateresolutionof the peaks.The distribution
ol the 021reflexionswas measuredby means of a Leeds-Northrop recording microphotometer and corrected for film background (Fig. 2).
|
2
3
0
.aL
4
5
6
7
A
9
tO
ll
12 13 14 ls
+
Frc. 2. Photometer curve oI the 02tr diffuse reflexions, corrected for film background.
The calculated values for the mean-positional arrangement of structures q and r are superimposed for comparison,
236
A. MCL. MATHIESONAND G. F. WALKER
The F2 values of the 001,h\l and h3l reflexions were obtained after
correction for polarization and Lorentz factors by graphical means
(Kaan and Cole, 1949). The structure amplitudes were later placed on
an approximately absolute scale by correlation with the caiculated
values.For the calculationof structure amplitudes, the scatteringfactors
for Mgz+ and Sia+ (fnternationale Tabellen, 1935) were loaded by the
respectiveisomorphousreplacementsindicated by the formula basedon
a chemical analysis of the sample (Table 1). The temperature factor,
B : 1 . 2 X 1 0 - 1 6c m . 2 , w a s d e t e r m i n e d b y p l o t t i n g l o e > j F c j / > I F o I
against sin2d in six equal rangesof sin2d from 0.0 to 0.6 (Wilson, 1942).
Fourier summations were carried out on three-figure Beevers-Lipson
strips, the a and c axes being subdivided into 30 and 120 parts respectively. The electron-density map was drawn from ccintour values derived from interpolated sections.
Because of the frequent associationof chlorite and vermiculite as
mixed-layerstructures,it was necessaryto be certain that the specimen
under investigation contained no interleaved chloritic material, although
previous examination of the total sample had indicated its purity. 'Ihe
crystal from which the h)l and.h3l,data had been obtained was therefore
converted to Li-vermiculite, and the 001 reflexions rephotographed. A
prolonged exposure showed the characteistic 12.2 A basal reflexion of
Li-vermiculite with a regular seriesof higher orders (Milne and Waiker,
loc. cit.). No sign oI a 14 A reflexion,such as would be produced by the
presenceof chlorite, could be found. On this basis,it is estimatedthat the
specimencontainedless than 0.5 per cent of chlorite layers.
Srnucrunr Alrer-ysrs
Initial attempts at a unidimensional synthesis of this vermiculite were
reported by Milne and Walker (1951), and an interim statement of
results in the present investigation was made in 1952 (Mathieson and
Walker).
l. Electron-densitydistribution normal to (001)
For the preliminary synthesis,the signsof the 001structure amplitudes
of Mg-vermiculite recorded up to the 62nd order were computed from
the contributions of the atoms of the silicate layer alone. Omission of
the water molecules and exchangeable cations from the calculations
ensuredthat no assumptionswould be involved in their location. Apart
from the distribution corresponding to the silicate layer, the synthesis
disclosedtwo additional peaks, one at z:O.214 and a smaller one at
z:0.250.
A unidimensional synthesis of the same crystal after conversion to
CRYS:TAL
S:IRUCTUREOFMAGNESIUM.TTERMICULITE
237
Sr-vermiculite reproduced the electron-density curve of the Mg-vermicuiite except for the peak at z:0.25, rvhich was substantially increasedin
size. The exchangeablecations are therefore located midway between
the silicate layers. The water moleculesare arranged on either side of
the central cations in sheetswhich lie at a distance of 2.83 A from the
surfaceoxygensof the adjacent silicate Iayers.
Recalculation of the structure amplitudes, taking into account the
contributions of the water molecules and exchangeablecations, gave
reliability factors,
Fo-Fct
*: -2 t*r
of 0.14 and,0.22for Mg- and Sr-vermiculite respectively (Table 2).
2. Project'ionalong the b-oxi.s
Since the approximate configuration of the silicate layer is known,
this portion of the structure can be disposedin the unit cell by consideration of the structure amplitudes for the reflexions 20,12 to 204 (see
Table 2) . With the approximate f parametersderived by this meansand
taking the a parameters from the line synthesis,a preliminary set of
structure amplitudes for the Z0l reflexions was calculated- Again the
contributions of the water molecules and exchangeablecations were
omitted initially so that no assumptions would be involved in their
location. The resultant electron-density contour map provided more
accurate parametersfor the atoms of the silicate layer, and located the
w a t e r m o l e c u l e sa t r : 0 . 1 4 2 , z : 0 . 2 1 3 a n d t h e e x c h a n g e a b l cea t i o n sa t
r:0.0, z:0.250. Recalculationof the structure amplitudes for all constituents of the unit cell necessitatedchangesin sign for a few minor terms
only, and a subsequentsynthesis produced the electron-density distribution shown in Fig. 1o. The correspondingdiagram of the crystal structure
is given in Fig. 10. The final reliability factor for h\l reflexionswas 0.15'
3. The Crystal Structure in Three Dimensions
Initially, deductionsarising from the sharp fr:32 spectrawill be considered and the results referred to a sub-cell containing a single silicate
l a y e r( c : 1 4 . 4 5 A ) .
From the equivalenceof the [010], [310] and [310] axes, it is evident
that the projectionsof the crystal structure along [310] and [310] will be
identical with that along [010] (Fig. 1b). The three projections can
therefore be combined to give the complete structure if the coordination
values of the atoms are also taken into account. The crystal structure
of half a siiicatelayer (e.g.,Q, Fig. 16) viewed normal to (001) is shown
238
A. MCL. MATHIESON AND G. F. WALKER
Tael-n 2' CoupamsoNor OesnnvnnaNo C,{r,culernn Srnucrunc Auprrrunrs r,or (o) Sr-vanlrrcur,rrE AND (6) Mg-venurcurrro. NorB rrlAr F(201):Ft13, t,l2): Srrcnr Drr.nnnENcES
rN rHE
Cer,curerno Var,uas lon Tnrso Snrs or, RrllrxroNs Ann Duo ro AppnoxruarrclNs
rN THEAroMrc Pene,ltntrns eNn SclttnnrNc FAcroRS
\a)
002
004
006
008
00,10
00,t2
00,14
00,16
00,18
00,20
o0,22
oo,24
o0,26
00,28
00,30
0o,32
00,34
00,36
00,38
00,40
00,42
00,44
219
344
55
185
63
<20
209
66
150
52
54
65
76
94
64
10
<26
/41
6l
298
78
-93
168
336
2l
-t74
-76
4
224
69
150
41
88
85
87
95
44
-30
-9
-tJ
76
002
004
006
008
00,10
00,12
00,14
0 0 ,1 6
00,18
00,20
00,22
00,24
00,26
00,28
00,30
o0,32
00,34
00,36
00,3g
00,40
oo,42
00,44
00,46
00,48
00,50
00,52
00,54
00,56
00,58
00,60
oo,62
282 305
62
39
110 -72
2t1
ztJ
342 J T J
60 - 6 7
105 -79
-95
86
124
lo/
161
84
9l
12l t2l
42
34
83 1 1 0
68
69
to7 108
6T
61
2
<20
3 8 -19
<22 - 11
+l
59
39
33
<24 -7
<24 -11
20
<25
(/
76
39
33
o
<26
<26
<27
o
23
2 0 ,5 8
20,56
20,54
20,52
2 0 ,5 0
20,49
20,46
20,44
20,42
20,40
20,39
20,36
20,34
20,32
20,3O
20,29
20,26
20,24
20,22
20,2O
20,18
20,16
20,14
20,t2
20,lo
208
206
204
202
200
202
204
206
208
20,\0
20,12
20,T4
20,T6
20,L8
20,n
20,22
<25
<24
<23
<22
<22
<21
<20
37
32
<19
<18
68
100
105
16
t6
76
105
79
JI
63
37
173
263
105
178
116
304
205
136
t6
1+7
32
100
r47
58
+t
152
236
294
63
o
2l
20
6
20,56
20,58
20,60
20,62
<23
<24
<25
<27
-
t.t
5
1
11
1
-12
14
28
39
40,52
40,50
40,48
40,46
1a
40,44
-9
40,42
70 40,40
92 4 0 , 3 8
122 40,36
6 40,34
-20
40,32
3 1 40,30
8 7 40,28
109 40,26
-.),/ 40,24
-48
u\ ),)
z3
40,20
l/J
40,18
257 40,16
87 + o , 7 4
189 4 0 , 1 2
81 40,10
300 408
2 1 2 406
1 2 7 4.04
20 402
l/J
400
JI
402
72 404
I/J
406
4l
408
-40
40,10
163 40,T2
203 Ln 7-tr
281 40,T6
3 7 40,18
<26 -18
-8
<26
29
.)J
46
45
23
38
-9
<24
27
<23
46
44
63
59
35
18
<20
0
<20
8
<19
52
42
52
29
35
29
42
46
26
150 t42
t62
t43
t44
152
-8
<16
16 -33
63
81
lo/
13+
58
94
110 -126
87 _69
1 2 7 122
225 223
ro/
178
< 1 6 -JU
58
IJ
1 2 7 132
196
206
77
239
CRVSTAL STRUCTURE OF MAGNESIUM.V ERMICA LITE
Trl.ts
2-(continued)
(b)
00r
\
F"b" F".h
]
I
I n* 1n"""
1n tl
20,26
20,28
20,34
20,32
20,54
20,36
20,38
20,4
F,a*
<16
16
105
89
<t7
52
<18
37
73
)n 4t.
20,T4
20,46
20,4
20,54
20,52
20,54
26
<20
47
47
37
<22
40,20 < 1 8 - 2 3
1
30 40,22 29
9
100 +o,24 < 1 8
6l
98 40,26 J Z
aA
JI
40,28 58
-JU
26
4t0,30 1 l
18
40,32 < 2 0
IJ
40,74 < 2 0
68
70 4 0 , 3 6 o.J
66
81
25 40,38
76
81
N,N
JI
8
1 1 Li 4r. < 2 2
<23 -5
54 N , M
11
36 40,46 < 2 3
32
40
28 40,48
17
7
40,50 < 2 4
-
I I
h3l
F*,".
40,52
40,54
40,56
40,58
40,60
60,42
60,40
60,38
60,36
60,34
60,32
60,30
60,28
ffi,26
60,24
60,22
ffi,2o
60,18
60,16
60,14
60,12
60,10
<25
nA
-18
8
<26
29
23
22
<27
/a<
<23
<23
30
<22
t2
<22
<21 -10
<,20
19
20
na
76
50
51
27
<19
<19 -5
50
77
81
oo
<18
55
-2
-Jt
<17
14
80,20
8 0 ,1 8
80,16
80,14
80,r2
80,10
808
806
804
802
800
802
804
806
808
80,10
80,T2
go,T4
80,16
80,1-8
80,20
80,22
80,'4
<23
<23
<22
<22
<22
42
79
J/
<2t
<21
63
5
t9
4
-9
-2
30
70
42
20
4
54
74
JI
32
26
16 -32
)7
<20
48
51
L9
<20
<20 -6
2
<21
JI
63
58
47
t2,08
1 2 , o , T n< 3 0
72,0,r2 <30
12,0,14 < 3 0
12,o,16 < 3 0
1 2 , 01, 8 < 3 0
F"ou.
2+
13,46
t4
I
11 LL
L
11
10
1 F"a".
13,42
13,40
1 3, 3 8
1116
13,34
13,32
1 3, 3 0
13,28
47
/11
2l
28
68
46
26
r7
48
88
98
r J ,z o
IJ
13,24
73,22 < 2 3
62
13,20
1 3 , 1 8 209
1 3 , 1 6 204
11 1L
56
60
13,12
1 3 ,1 0
138
136
t34
AJ
lJz
F"ot"
53
11
30
23
o/
55
1a
-28
.A
97
96
)7
2
38
27r
t96
l.),/
t39
45 -40
40
49
1 7 r 180
74
120
49
32
206 r69
240
A. MCL. MATHIESON AND G. F. WALKER
Trnr-n 2-(continu,ed)
Fot"
608
45
606
60
604
J.)
/
1a
602
600
35
a1
602
604
1+1
606
96
608
JJ
60,T0 20
60,12 < 1 8
60,14 76
60,T6 66
60,1-g < 1 9
60,2A 50
ffi,22 < 1 9
()A,24 86
60,26 8 1
60,28 30
60,30 < 2 0
ffi,32
30
60,34 60
60,36 55
60,38 < 2 1
60,40 < 2 1
60 4'
<22
<22
60,46 < 2 3
35
56
JO
10
43
62
1+2
91
+J
-18
2l
90
54
-11
-J/
1)
94
34
-9
52
67
7
-15
7
ffirM
<2+
60,43
60,50
60,52
60,54
<24
<25
<26
80,32
80,30
80,2g
80,26
80,24
80,22
<27
<26
<25
<25
<24
<24
t4
3
0
10
11
34
30
27
11
80,26
80,28
80,30
80,82
80,34
90,36
80,38
80,40
80,42
80,!m
80,46
80,48
<22
<22
<22
<22
<23
<23
<23
<24
<2+
<25
<25
<26
10,0,14<26
l0,o,l2
47
1 0 , 0, 1 0
26
10,09 <25
10,06 <24
10,04 <24
10,02
<23
10,00 <23
10,02 <23
10,04
<22
10,06 <22
10,08 <22
10,0,T0 36
1 0 , 0 , 1 2 42
1 0, 0 , 1 4 < 2 3
10,0,m <23
10,0,18 <24
1 0 , 0 , 2 0 36
10,0,22 2 6
10,o,24 < 2 5
1 0, 0 , 2 6 < 2 5
10,0,28 <26
10,0,30<26
10,0,32<27
12,06
<30
F,"u.
1a
3
18
IJ
1a
0
-7
-7
7
?,,
19
TJ
2l
37
24
IJ
t4
0
l6
6
o
0
-8
10
28
+3
15
5
18
+l
32
-7
-2r
0
19
f.l
19
)
,*r
)
n*,
It3l.
130
Fa^.
F"alc,
<20
111
234
18
130
134
210
,rJo
JIJ
298
138
l4l
82
1 3, 1 0 174 r87
1 3 , 1 2 I lJ
92
13,14 227 259
1 3 , 1 6 179 1 7 l
13,18 ?)
24
13,20 49 -45
13,22 45 - 5 8
13,24 / o
109
73,26 95
89
13,28 2 l
30
1 3 , 3 0 < 2 3 -19
r32
13,32 < 2 4
13,34 101
lJ,JO
rt2
1 3 , 3 8 63
13,re < 2 5
13,42 < 2 6
13,44 29
1 3 , 4 6 30
13,48 < 2 8
o
123
92
69
-8
39
28
13
33,40 <28
36
1t ?q
t2
<27
33,36 < 2 7 - 3
.A
33,34 <26
3 3 , 3 2 23
2l
33,30
78
9l
33,28 68
78
33,26 J I
60
33,24 < 2 4 -16
7\ ))
12
33,2Q tJz
3 3 ,1 8
IJO
33,16 <23
75
I IJ
119
t9
24r
CRYSTAL STRUCTURE OT' M AGNESI UM -VERM I CULIT E
T rr;,x, 2- (conti,naed')
F2 calc.
D.
33,t+
33,t2
33,10
338
336
334
332
330
332
JJ4
336
JJ6
33,m
1r 1-
33,T4
33,T6
33,T8
33,m
,t
"7
33,24
33,16
33,28
33,80
s3,82
33,34
33,36
33,38
33,m
33,42
33,44
33,46
(1
??
s3,30
5 3, 2 8
53,26
53,24
ql tt
5 3, 2 0
53,18
53,16
53,14
(? 1t
5 3 ,1 0
70
<22
90
139
65
<22
107
207
J l.)
175
82
<22
J6
tr6
84
-23
62
<23
130
tt7
62
<24
36
68
78
LJ
<25
<26
<26
<27
<27
100
82
7
60
Is4
50
-8
104
140
304
140
t02
-5
53
t43
59
4
-88
24
135
I22
79
-15
I L
89
99
24
-30
0
lo
33
55
58
J+
a1
<26
<25
<25
43
23
-15
13
39
72
o/
<24
40
<24
109
135
nl
-35
-4
ro2
133
53,30
s3,32
JJ , ,i+
53,36
53,40
53,42
73,20
73,19
73,t6
73,t+
73,12
73,r0
738
736
734
732
730
732
734
iJo
/J6
73,m
1\ T,
/J,r+
73,T6
73,18
73,n
73,22
73,'4
73,%
73,28
73,m
73,32
73,94
936
934
932
930
932
934
vJt)
M
46
<28
<29
<29
<30
<30
<28
<27
<26
/o
76
34
<25
<25
45
56
<25
<25
<25
\II
78
62
<25
<25
<26
o4
4l
70
-9
-25
-t9
45
58
t.)
1
26
79
,J
1
o
50
38
20
-35
-4
4l
72
68
18
27
+5
72
56
<26
<27
<28
34
34
<30
<32
55
-2
-13
-t2
36
29
<28
<28
<27
<27
<26
22
30
8
-5
2l
47
56
53
11
-6
021
022
o23
02+
025
026
027
028
o29
02,10
i)
11
02,t2
02,13
02,14
nt 1(
<A
7r1,
-25
56
-ol
-11
-81
+3
10
30
o/
-15
29
-31
29
t23
o
31
JI
45
1
oo
18
I
9
45
a
8
10
242
A. MCL. MATHIESON AND G. F. WALKER
Tl^nr-n2-(continued)
538
536
534
532
s30
532
534
536
s38
JJ, IU
\? ]t-
53,14[
53,m
s3,18
53,20
s3,24
53,26
53,2-8
52
<24
43
143
r+2
49
<24
43
43
68
58
<24
<25
96
111
50
<26
<27
JI
27
40
134
130
3+
-4
938
93,m
ol
Tt
93,14
93,16
93,1893,n
o? te
58
53
70
l6
4
o+
t02
93,t4
<25
<25
<25
<25
<25
39
<26
<27
<28
31
_T
12
-3
7
25
JI
12
10
114
-2
13
in Fig. 3a. On the basis of the diffraction data considered so far, the
arrangement of the water molecule sites and the exchangeablecation
sites (my mz, ms) can be represented diagrammatically by the view
(normal to (001) and projected on the base of the cell) of the layer
p'R'(Fig.3D). The chemicaldata indicate that the water moleculeand
exchangeablecation sites are approximately two-thirds and one-ninth
filled respectively. Calculation of the h3l, strtctwe amplitudes for this
structure gives satisfactory agreementwith the observedvalues (Table 2)
establishingits essentialcorrectness(R:0.17). Table 3 lists the atomic
coordinates;the alternative sets of values (o) and (6), given for the r
and y coordinatesof the water molecule sites, will be discussedat a
later stage.
The structure of the silicate layer which emerges is very similar to
that put forward by previous workers, although greater detail is revealed
than has hitherto been obtained for layer silicate structures, largely
becauseof increasedprecision of intensity estimation combined with refinement by successive Fourier syntheses. The bond lengths are as
CRYSTAL STRUCTURE OF MAGNESIUM-VERMICULITE
a)
o
C: (M9,F",A|)
Ool
o p = (si,Al)
2+3
b)
@
M g'*
a\
H,O(a)
i)
H,O(R')
O Q,,
Fro.3. (a) Section of the crystal structure of Mg-vermiculite from z:0.0 to 0.114
(i.e. half-layer Q) projected on the base of the cell normal to (001). (b) Section of the crystal structure from z:0.213 to 0.287. i.e.. sites of the water molecules in sheets O' and R'
and sites of exchangeable Mg2+ cations.
f o l i o w s :C * - O t , 2 . 0 6 F r ; O * - O r , 1 . 6 9A ; a n c LD - O z e , D - O r " , p - O u
1.624 each (probable error, +0.02 A) . The oxygen atom Or is linked
tetrahedrally with one D (or hydrogen) and three C atoms, the angle
COD being 109o, whereas O2a, Ory and Oa are each linked to two D
atoms with a DOD angle of I42o. If r(Sia+) is taken as 0.41 A, th.rt
r(Or) is |.28 L and,r(O2, OB)
, 121 A. llfte relationship between ionic radius
and coordinationnumber (Wells, 1950,p. 71) gives the ratio of two- to
* For convenience, C will be used to represent the (Mg, Fe, Al) atoms in octahedral
positions at the centre of the silicate layer; and D, the (Si, AI) tetrahedrally-coordinated
atoms near the surface. The key to the oxygen atoms is given in Figs. 1b and 3a.
244
A. MCL. MATHIESONAND G. F. WALKER
four-coordinatedradii as 12/4)t:0.92 and,the observed ratio, 0.95, is
therefore of the correct order.
The contour map clearly indicates a distortion in the surfaceoxygen
sheet of the silicate layer. The hexagonof oxygen atoms XIzXtYtXttYtl
(Fig. 3a) is not regular but consists of two interpenetrating isosceles
triangles of oxygens XX'X" and YY'Y" with sides 4.35 and 4.84 A
respectiveiy.The correspondingdistancein a regular hexagonwould be
4.59 A. The distortion is accountedfor by a rotation through 5$o of the
triad of oxygens above each D a.tom. Since the change from the ideal
hexagonal to the distorted configuration wouid involve differencesin
D-O bond lengths of less than 0.01 A, it is improbable that the distortion arisesfrom inability to accommodatea regular hexagonaloxygen
n e t w o r k i n t h e c r o s s - s e c t i o n a rl e a o f t h e u n i t c e l l ( i . e . , 9 . 1 8 X 5 . 3 3A ) .
The distortion is more readily explained on the basis of electrostatic
forces within the silicatelayer, if it is assumedthat a residual negative
charge on each surfaceoxygen interacts with a residual positive charge
on the octahedral C atom lying below and to one side (viewed normal to
(001)). The electrostaticforces thus invoked are correctly disposedto
produce a torque in the required direction (Fig. 3o), the angular movement being limited by the over-riding factor of the D-O distance.
At this juncture, it is appropriate to considerthe approximate spatial
relationshipsof the interlamellar water and exchangeablecation sites to
an adjacent silicate layer. It will be assumedfor the moment that the
water moleculesall lie at the centre of the peak of the electron-density
distribution (Fig. 1o), although the possibility that this is merely a mean
position will be suggestedlater. The sites for the water moleculeslie
directly over C atoms and those for the exchangeablecations over D
atoms (or hydroxyls) of the silicate layer. This latter observationindicatesan important point with regard to the relationshipof adjacent silicate layers, namely the collinearity of (i) atoms 01 and D (or f1) of the
half-layer 0, (ii) the exchangeabiecation site, and (iii) atoms D (or H)
and Or of the halflayer R (Fig. 1D).
The structure of complete silicate layers and the method of stacking
of the Iayers can be deducedfrom a considerationof the ft*.32 spectra.
The water moleculesand exchangeablecations were omitted from the
calculations since they can only have a minor influence on the structure
ampiitudes. The 021reflexions were employed in testing the various possible structures.
In discussingthe silicatelayers, it is convenientto focus attention on
the site at the centre of the distorted hexagonof surfaceoxygen atoms.
Let this site be S(Z) where Z refers to the half-layer P, Q, R or ?. Then
the half-layers P and Q combine to form a single layer and ,S(Q) can be
associatedwith S(P1), S(P2) or S(Pr) (Fig.3o). These three possible
CRYSTAL STRUCTURE OF MAGNESIUM-V ERMICULITE
at<
s-------------+
Sr--------------{
+
s)
I',)
\
r)
Frc. 4. Schematic presentation of the possible stacking sequencesinMg-vermiculite.
0) (q) and (r) /[-type silicate layer displaced with respect to an M-type layer by O, bl3
and Zbl3 respectively. (s) Z-type layers displaced by b/3 (ot 2b/3). (t) Positional mean of
structures q and r.
layer structures correspond to the L, M and 1[ types of layers referred
to by Brindley, Oughton and Robinson (1950) in their study of the
chlorites. The three types of silicate layer have individual symmetry
C2/mand are indistinguishable until the 6 axis of the crystal is defined
by the mutual relations of the stacked layers. When a second complete
silicate layer RT is added, there are twelve possible ways of grouping
P, Q, R and ?. The existenceof a glideoperationc reducesthe possibilities to four, illustrated diagrammaticallyin Fig. 4 P, Q,z and s' As Hendricks and Jefferson (1938o) have shown) none of these arrangements
yields structure amplitudes in agreement with the observed intensity
distribution. However, structures q and r give some measure of agreement for reflections021where l:2n, and a structure with the unit RI
Iocated at the centre of the projection area (Fig. a l) gives a very close
agreement with the observed distribution of intensities for the 021
spectra. Table 2 lists the amplitudes calculated for this arrangement,
and Fig. 2 shows the F2 values superimposedon the photometer curve.
The solution obtained in this way is completely contrary to all the
evidence of the sub-cell analysis, according to which the site S(R) should
246
A. MCL. MATHIESON AND G. F. WALKER
lie over the site S(Q) or displacedtherefrom by +b/6 in this projection
(i.e., S'or S", Fig. a). Furthermore, a unique structure for vermiculite
would be implied by structure I and therefore no explanation for the
occurrenceof the diffuse spectra would be possible.The deductions from
both the sharp and diffuse spectra can be reconciled only if it is assumed
that structures g and r are equally probable and that structure p does
not occur. In other words, the stacking of the silicate layers in vermiculite corresponds to M-type iayers aiternating with V-type layers. The
Iayers, however, can only be arranged so that site S(R) is not in line with
site S(Q) nor S(P) with S(7). The resultant crystal structure can be
represented for the purposes of calculation by the average structure I
(Fig. a). The mutual relationships of the surface oxygens of the silicate
half-layers Q and R are illustrated in Fig. 5.
R
BlsoB:.8:BoB
iii
iil
iii
3.t:3
3.3::
a
3.3:3
OO
oTa
a)
OO
ao
.iii .-o
o.]o
b)
( = n ' r ,e9;
Fro. 5. Plan and elevation of thg relationship between the surface oxygen network of
halfJayers Q and R. ?:ttrr,
r:n2,ms are sites for exchangeable cations as deduced from the subcell analysis, and are defined by their relationship to the halfJayer Q. Projections are
normal to (001).
4. Location oJ the Si,tesoJ the Water Molecules
In the 6-axisprojection (Fig. 1o), the contours of the water moiecules
tend to be drawn out parallel to the a axis and there is evidence of a tail
in the negative *-direction. As a result, the peak height is lower than
would be expectedlor 2.16 water molecules(Table 1).
At first sight, the shapeof the water peak might seemto be due to high
thermal energy in the water molecules, producing movement mainly
parallel to (001) because of restrictions imposed by the environment.
CRYSTAL STRUCTURE OF MAGN ESIU M-VERMICULITE
247
\i
o{
---l|
t,'s:
o
L
oo)
.
(P'i
M9r.
. o(al
\l
.--Y,
+
O H,o(Q)
c)
i..: H.O(R)
Frc.6. (a) Diagram of the reiation of the distorted water network Q'to the surface
oxygen network of the halfJayer Q. Cation sites za1,m4 n s are also shown. This diagram
illustrates the tendency of each water molecule to approach more closely to its associated
surface oxygen. (b), (c) , (d) The three ways of combining two sheets of water sites, the relative shifts being b/3, 2b/3 and 0 respectively. (d) is excluded by the data. The sites nzr
and mz with suitable undistorted octahedral environment for the Mg2+ cations are shown.
This explanation,however,doesnot account for the "tail" on the water
moleculepeak and, in order to do so, it must be assumedthat the asymmetric electron-density distribution correspondsto regular displacements
of the water moleculesfrom the mean position (Table 3, column (a)). In
this event, one would expect such displacements to be reiated to the
adjacent oxygen network and, in a triad of water molecules(HrO(l),
HzO(2), HzO(3), Fig. 6o) over a surface oxygen hexagon, each member
of the triad would be displaced equaily towards or away from its related
oxygen, i.e., the triad would expand or contract. If it expands, there
must be concomitant rotations of unexpanded triads in opposite directions around ?mrand m2 (Fig. 6a), and the correspondingelectron-density
distribution viewed along the 6-axis would develop a tail in the negative
r-direction such as is indicated in the contour map (Fig. 1a)*. The distortion of the water sheet produced in this way results in a closer approach of each water molecule to its related oxygen atom. The relevant
atomic parametersare listed in column (D),Table 3.
Since each water sheet is located by its relation to the adjacent silicate
layer surface, a complete water layer (consisting of two sheets of water
molecules) can only be formed by superimposing the water sheetsin the
two ways permitted by the stacking of the silicate layers. These two
arrangements correspondto relative shifts of b/3 and 2b/3, and are illustrated in Fig. 6 b and c respectively. Inspection shows that the two arrangements are equivaient and, in both, the relative displacements of
the water molecules maintain normal H2O-H2O approach distances bex The displacement of the rn'ater molecules from their ideal hexagonal positions cannot
be determined accurately from the shape of the electron-density distribution but is probably of the order of 0.2 A.
A. MCL. MATHIESON AND G, F, WALKER
Testa 3. Arourc P.rlnnunrrns lon Mc-vnnurculrrn
HrO
site (1)
site (2)
site (3)
0
0
0
0. 3 5 8
0. 3 5 8
0. 3 5 8
0.397
0.397
o.r47
0.r47
0 434
(o)
(b)
0.142 0.160
0.142 0.160
o-t42 0.105
Iv_tg,
srte ,fir
site m.z
site mz
0.500
0.500
0.500
L
C
Ot
Ot
Or
U
D
Ozt
Orn
Ot
0
0. 3 3 3
0.667
0
0.333
0.667
0
0.333
o.404
0.929
0.167
(a)
0
0
0
0. 0 3 7
0 .037
0. 0 3 7
0.096
0.096
0 .1 1 4
0.114
0.114
(b)
-0.019
0
0.333 0.352
0.667 0.667
0.213
0.2r3
o.213
0
0.333
o.667
0. 2 5 0
0.250
0.250
tween the sheets.The method of stacking the water sheetsillustrated in
Fig. 6 d is excluded, since it could only exist if structure f (FiS.4) were
permissible.The nonoccurrenceof structure I is perhaps connectedwith
the fact that such a water-layer arrangement would involve very close
approach of the water moleculesbetween the sheets.
5. The Structure of the Interl,amellarRegion
From the previous sections,it is evident that the arrangement of the
sites of the interiamellar cations and water molecules closely resembles
that of the octahedral C atoms and 01 atoms, respectively, of the silicate
layers although, in the former, the water sites are displaced somewhat
from the regular octahedral arrangement. According to the chemical
data, however, only a proportion of the interlamellar water and cation
sitesare occupiedat one time.
Since each water site is equivalent with respect to the associated
oxygen of the siiicate layer surface, any grouping of the water molecules
will be due to the influence of the cations. The importance of the cations
in this respect is suggestedby the variation of the interlamellar spacing produced when other cations are substituted for the Mg (Milne and
Walker, 1950).A certain amount of evidencehas been adducedby others
to show that Me ions tend to attract octahedral hvdration shells in
CRYST:AL STRUCTURE OF MAGNESIUM-V ERMICULITE
249
water solution (e.g. Bernal and Fowler, 1933), and such octahedral
grouping is known to occur in crystal hydrates such as magnesium
chlorate hexahydrate (West, 1935).It is reasonableto suppose,therefore,
that there will be a tendency for the exchangeableMg ions in vermiculite
to form hydration shells.
Examination of the two equivalent water-layer structures in Fig. 6 6
and c reveals that in each case only one of the three m cation sites provides und,istorted.octahedral coordination for a cation (m1 in Fig. 6 b or
m2inFig.6 c). The Mgr+-HrO distance of 2.06 A involved, moreover,
is in agreement with that observed in magnesium benzene sulphonate
hexahydrate (Broomhead and Nicol, 1948) which is the most accurate
value available. All other sites have an asymmetric environment of water
molecules, consisting of an unexpanded triad in one sheet and an expanded triad in the other. Equal Mg2+-HzO distances at these sites
could only be maintained by displacing the cations 0.06 A from the
central plane, which is beyond the limit of error of measurementfrom the
contour map (Fig. 1o). The resultant equilibrium Mg,+-HrO distance,
moreover, would be 2.20 A.It is almost certainly true, therefore, that the
cations will occupy rnror rnzsites depending on the stacking of the adjacent silicate layers.
The degree of ordering of the water molecules and cations within the
available sites can now be considered further. The possibility of order
within an 'ind'itidual, water-cati,onlayer is not excluded by the diffraction
data since mutual ordering of these layers would be necessaryfor their
structure to be revealed by the data. A case for the existence of longrange order of this nature, not directly deducible from the diffraction
evidence, has been convincingly put forward in recent studies of hollandite (Bystrcim, 1951) and psilomelane (Wadsley, 1952). So far, we
have shown that the exchangeablecations in vermiculite lie in a plane
midway between adjacent silicate layers, and that their lateral disposition is related to the adjacent water sheets. There are, however, only
enough cations to fill one in three of the available m1 (or mq) sites.
If it is assumedthat the cations, due to mutual repulsion effects, tend
to distribute themselves uniformly throughout the available sites, a
hexagonal distribution of the type shown in Fig. 7 is obtained. Six water
molecules can be arranged around and in contact with a Mg ion at a u
site, a triad in the upper and a triad in the lower sheet, forming the
single hydration shell already postulated. In the lower sheet, additional
waters can be added in groups of three around unoccupied sites a or a.,.
The water molecule sites around zir in this sheet will be closer to the
cations than those around z, so that the former sites will be preferred
if it is assumedthat the tendency of the cations to form single hydration
A, MCL. MATHIESON AND G. F. WALKER
250
,o-
-_- -o\
o
,o'---O-.
Mg''
o
Hp (to* ot
o
Hp (.**r
t,
.,e-rq
Al
I
i l-_ry" i Y
o
-/
t-o-
I
- - -a'
d
'--o-
I
^
't.
...e'-'9.
A''A
O
Y
'Y
a....6r"'
to-
-- .- -o'
,'
...*'.o.
\
l*-ol
+ ?'i
'or-
/
- '-/
I
tto
- - -t/
z,z,2,,
Sites
region.
r,n"L,""o.ii:?,H:.},i?;,ii:iffi:ffi1]",
Frc.i. Diagram
shells indicates a similar although less-marked tendency to form second
shells.*In the upper sheet,sites around r will be preferred to those around
?, since the former are in this casecloser to the cations. Such an arrangement allows each Mg ion to have around it an almost complete double
shell of water molecules, and the linking-up of these double shells produces layers in which two out of every three water sites are occupied.
In this arrangement of the water layers, there are holes correspondingto
triads of vacant water sites, each triad being equidistant from three cations with their associateddouble hydration shells. These sites may tend
to be vacant (at normal temperature and pressure) because a water
molecule in such a site would be in the third hydration shells ol lwo
cations and hence less stable than those associatedwith a single cation.
The types of bonding operating within the interlamellar region can
now be consideredwith the aid of measurementstaken from the contour
map (Fig. 1o). If each water molecule(e.g.HzO(-4), Fig- 8) is attached
to its related oxygen by means of a hydrogen bond (HzO-O, 2.84 L),
the second hydrogen can be directed towards another water molecule
angle is 104fo in
(HrO(B)) in the same sheet, so that the H-O-H
agreement with the spectroscopically measured value (Darling and
* Support for the assumption is provided by evidence of the occurrence of Mg2+.12HzO
groups in salt solutions (Sidgwick, 1950, p. 237), and by the high energy of hydration of
Mg cations.
CRYSTAL STRUCTURE OF MAGNESIUM-VERMICULITE
H,O(A)
H,o(c)
251
Hp(a)
\i
t.j
Frc. 8. Suggested system of weak hydrogen bonding illustrated for a single water sheet
assuming all water molecule sites filled. oxygens of water molecules represented by full
circles, packing radius (Rees, 1949) to scale; hydrogens.linked to silicate surface oxygens
not shown; hydrogens in the plane of the water sheet represented by broken semi-circlesl
negative poles represented by black triangles and rectangles, triangles indicate greater
displacement of charge from the piane of the sheet than rectangles.
Dennison, 1940). The second hydrogen lies in the plane of the water
sheet but the two negative poles of the water molecule are slightly displaced therefrom, one directed towards the Mg ion site and the other pointing in the general direction of a third water molecule site (HrO(C)).
Since the intermolecular distances within a water sheet are either
3.06 A or 3.40 A, it is clear that normal hydrogen bonding does not
operate at these points, and that the interactions are of a less specific
nature. They may be regarded as weah hydrogen bonds and, as a consequence, a rigid application of tetrahedral linkages need not apply. In
the vicinity of a cation site, the observed intermolecular distance within
a water sheet of 3.06 A is greater than the value calculated for a perfectly regular octahedronof waters around a Mg ion (vi2.2.92 A;. fne
triads of water moleculesassociatedwith a cation are expanded slightly
in the plane of the sheetsso as to allow each water to approach its associated oxygen. As a consequence,the cations sink slightly into the water
triads in both sheetsand the intersheet distance of the water molecules
is reduced to 2.88 A so as to preservethe Mg-H2O distanceof 2.06 A.
The intersheet distance of 2.88 A is uniform throughout the water
252
A. MCL. MATHIESON AND G. F. WALKER
layers, which provides a possible further reason for the occurrence of
triads of vacant water sites associated with the unoccupied a and w
cation sites where the binding effect of the cations is absent. The observed intermolecular distance in an expanded triad of waters, viz.
3.40 A, is consistentwith the model if packing radii are assumed(Rees,
re4e).
The idealised arrangement proposed above (Fig. 7) gives two out of
three water sites occupied, which is approximately the required ratio,
and the ratio of exchangeablecations to waters in the network is one in
twelve which is close to the value indicated by the chemical formula
(1:13.5). This model may thereforebe closelyapproximatedin the Mgvermiculite under examination. A highly-ordered arrangement of the
interlameilar region is not inconsistent with the observed easeof cation
replacement (Milne and Walker, 1950), since normal diffusion processes
can be expectedto operate through vacant a sites and "interstitial" positions ? and w, with concomitant adjustments in the water network. It is
evident that the interlamellar region must be regarded as a dynamic system in which constant rearrangement and adjustment is taking place.
DrscussroN
Detailed analysis has revealed a distortion in the surface oxygen network of the silicate layer in vermiculite. The explanation advanced is
based on residual electrostatic forces acting between the central octahedrally-coordinated atoms and the surface oxygens of the silicate layers.
Evidence for the existenceof a distortion of the silicate layer in muscovite
has been noted prev,iously by Hendricks and Jefferson (1939), although
in this case the data did not allow the nature of the distortion to be
defined.'slight departures from regularity are therefore possible in silicate layers of this type.
The solution advanced for the stacking of the silicate layers, if valid,
would impose certain limitations on similar minerals, and may therefore
be applicable to the stacking sequencesin talc and pyrophyiiite' No
satisfactoryexplanationfor the \kt, h!3n spectrahas so far beendeduced
for theseminerals. The observedstructure amplitudes of the02l reflexions
in vermiculite have been shown to correspond closely to the values
calculated not for g or r (Fig. 4) separately, but for the positional mean
of these two possiblearrangements.It is necessaryto point out, however,
that the results of the interpretation here placed on the diffuse reflexions
cannot be reconciledwith the theoretical treatment of disorder for cobalt
in the regiona-] (Wilson, !949, p. 70), which is analogousto the present
case and predicts the almost complete disappearanceof the diffuse regions. The discrepancy may arise from the failure of the theoretical
CRYSTAL STRUCTURE OF MAGNESIAM-V ERMICULITE
253
treatment to take into account the apparent increasein symmetry of the
structure in the region a-f.
The main interest attached to the structure analysis lies in the light
which it throws on the organization of the interlamellar region. The
water-cation network consists of two sheets of distorted hexagonallylinked water molecule sites with the cations occupying definite positions
in the midway plane. Although the chemical data require that only onethird of cation sites (zz1or rm2),and two-thirds of water sites should be occupied, it has been shown that a random filling of the available sitesis not
a necessary consequenceof the data and, in the vermiculite under
examination, a highly-ordered arrangement of the water molecules and
cations is probable. In vermiculites which carry a higher population of
exchangeable cations (Barshad, 1950, 1952), the idealised structure
would be realised to a lesser extent, since u, a and w sites would be
expected to be partly occupied by cations and the distribution of the
water molecules in the available sites would be altered accordingly.
Nevertheless,grouping of the water molecules around the cations and a
tendency towards a uniform distribution of the cations are to be expected in all similar minerals if the assumptions made above are valid.
ft has usually been assumed, in considering the interaction of exchangeable cations with the planar surfaces of layer silicates, that the
former are located by reactive spots of negative charge on the silicate
layer surfaces arising from individual isomorphous substitutions. The
present structure analysis has located the interlamellar Mg2+ ions in a
plane midway between silicate layers. In this plane, the potential energy
due to electrostatic forces can be shown by calculation to be efiectively
uniform, if the high frequency of isomorphous substitution is taken into
account.* The distribution of the cations in the midway plane is therefore
not related to any directiaeefiect of the electrostatic forces. The importance of the hydration behaviour of the cations in this respect is attested
by the relationship between the cations and the water molecules as
revealed by the difiraction data, and also by the observed spontaneous
rehydration of partially-dehydrated Mg-vermiculite ( Walker, 1949). In
the interlamellar region, the water molecules within a sheet are linked
by weak hydrogen bonds and the individual water sheets are held together by the cations. The hydrogen bonds binding the water molecules
to the oxygen atoms of the silicate surfaces are probably stabilised to
some extent by the electrostatic forces acting through the water molecules. We may conclude that, in Mg-vermiculite,trthe Iocation of the
* In vermiculite, which is a trioctahedral mineral, the substitution of about one in
three tetrahedral and about one in five octahedral atoms gives rise to negative and positive
charges respectively. The result is a net negative charge acting at the silicate layer surfaces.
254
A, IICL, MATHIESON AND G, F. VI.ALKER
water molecule sites is determined by the surface configuration of the
silicate layers, and that the mutual relationship of the water sheetsin a
single water-cation layer is determined by the requirement for octahedral coordination of the water molecules around the cations. The unimportance of direct electrostatic interaction between the cations and surface oxygens is consistent with the conclusion of Guggenheim and
McGlashan (1951) that the influenceof next-nearestneighboursis entirely negligible. Their result, however, does not invaiidate the proposed
tendency to the formation of second hydration shells around the interlamellar cations in Mg-vermiculite, since each water molecule is equivalent in position with respect to the adjacent silicate layer surface and
dipole interaction between water molecules can therefore exert a significant influence.
The structure derived for the interlamellar region in Mg-vermiculite,
even in its less specific form, excludes the hypothetical water-layer
structure proposed by Hendricks and Jefierson (1938b) for vermiculite,
and later modifications made to take account of the exchangeablecations
(e.g. Hendricks, Nelson and Alexander, 1940). Furthermore, the structure observed does not appear to have been considered in any of the
models proposed for the water-cation network of layer silicate minerals.
AcrwowrooGMENT
The authors gratefully acknowledgehelpful criticism by Dr. A. L. G.
Rees,particularly with regard to the nature of the forces operating within
the crystal lattice.
RrrnnnNcrs
(1948),
BrnsnAo,I.
Am.Mineral.,33,
655.
Bansv.ll, I. (1949),Am. Mineral.,34,675.
Bansnen,I. (1950),Am. Mineral,.,35,225.
Brnsnao, I. (1952),Proc.Soil. Sci. Soc.Am., 16, 176.
BrRNAr,J. D., axo Fowr,nn,R. H. (1933),J. Chem.Phys.,1,515.
Bmnnr.rv, G. W. (1951),Chap. VI in X-nry IonNrlrcarrox eno Srnucrunrs ol Cr,ev
MrNrt,l.rs, edit. G. W. Brindley, Min. Soc.,Lonilon.
BnrNnr.nv,G. W., Oucnror, B. M., aNo RonrnsoN,K. (1950),Acta Cryst.,3, 408.
Bvsrnciu,A., exn A. M. (1951),Acta Cryst.,4,469.
Danr,rNo,B. T., auo DnNNrsoN,D. M. (1940),Phys.Rn.,57, t28.
GntrNen,J. W. (1934),Am. Mineral., 19, 557.
Gnutnn, J. W. (1939),Am. Mineral.,24, 428.
GuGcnxunru,E. A., lNn McGr.asnax,M. L. (1951),Trans.Far.Soc.,47,929.
Ifrxontcrs, S. B., eun JelrnnsoN,M. F. (1938a),Am. Mineral.,23r 85l.
Hnxontcrs, S. 8., aNnJrlrrnsox, M. F. (1938b),Am. Mineral.,23r863.
IfrrnRrcrs, S.8., alto Jnlrenson, M. F. (1939),Am. Mi.neral,.,24r729.
Hnnonrcrs, S. B., NnlsoN, R. A., ewo Ar-rxeNonr,L. T. (1940),J. Am. Chem.Soc.,62,
1457.
CRYSTAL STRUCTUREOF MAGNESIT]M.VERMICALITE
255
Tanrr-r.rN zun Bnsrruv uNGvoN Knrsrer.r,srnurrunnN (1935),Berlin :
INrnp1q.trroNAl-r
Borntraeger.
KreN, G., rNn Corn, W. F. (1949),Acla Cryst.,2,38.
MnrnrnsoN,A. McL. (1951),J. Scimt. Instr.,28, ltz.
G. F. (1952),ClayMi,ns.Bull. l' (8),272.
Mrrurrsox, A. McL., ANDWAT,KER,
Mrr.rw, A. A.,.lNn WAnnn, G. F. (1950),Trans.FourthInt. Cong.Soil'Sci'.,2,62.
MrlNn, A. A., eNo Wlr,rnt, G. F. (1951),ClayMi.ns.Bull., | (6),17l.
Rens, A. L. G. (1948),J. Chem.Phys.,16,995.
Smcwrcr, N. V. (1950),Tne Crrurcer. Er.BMsNrsANDTHErRCoupounns: Or/ord.
A. D. (1952),Natwe, l7O,973.
WAnsr-Ev,
Wa.r-ren,G. F. (1947),ClayMins. Bxtll.,| (l), 5.
Warxon, G. F. (1949),Natwe, 163,726.
War.xrn, G. F. (1950o),Mineral.Mag.,29,72.
W.al.r<nn,G. F. (1950b),Natwe, 166,695.
W,tt-Kln, G. F. (1951),Chap. VII inlx-nav Iorrurtc.qlron eNr Srnucrunes ol Cr..a.v
Mrurnnr,s, edit. G. W. Brindley, Min. Soc.,London.
Wrsr, C. D. (1935),Z. Krist.,9l, 480.
Wnrr-s, A. F. (1950),Srnucrun.qr,INoncllrrc CunMEtnY, 2nd ed., OrJoril'
Wnsorv,A.J.C. (1942),Nature,150,152.
Wrr-som,A. J. C. (1949).X-nev Oprrcs, Lond,on.
Manuscript receioed
Feb. 18,1953.