Tải bản đầy đủ - 0trang
V. Swelling of Calcium Clays
Permeability and X-Ray Spacing for Na- and
Ca-Montmorillonite (Wyoming) in NaCl and CaCI, Solutions
0. I M CaCI,
< 10-5 M CaCI,
I .O M NaCl
0.5 M NaCl
0.3 M NaCl
(105 cm sec-1)
0.3 M NaCl (Quirk, 1952; see also Quirk, 1968). Mering (1946) considered that
the large macroscopic swelling of Na-montmorillonite was not due to the expansion of the crystals. However, Quirk (1952) concluded that the dramatic decrease
in permeability was due to the onset of extensive crystalline swelling as a result
of the development of diffuse double layers on all the opposing surfaces. This
interpretation was confirmed by X-ray measurements (Table VI) which showed
that the basal spacing jumped from 18.9 in 0.5 M NaCl to 43 A in 0.3 M NaCl.
Furthermore, Norrish and Quirk (1954) also reported that Ca-montmorillonite
maintained a basal spacing of 19 8, even in distilled water; the lamellae of the
montmorillonite remain in a primary potential minimum and are separated by 9 A
of water or about three molecular “layers.”
The swelling (solution content) of compressed cores of Na- and Camontmorillonite (Wyoming) and Ca-illite (Willalooka) in relation to electrolyte
concentration at 1 kPa suction (Aylmore and Quirk, 1962) is shown in Table VII.
Since the material is saturated, the volume of the solution taken up is a measure
of the swelling since the initial porosities were similar. It may be noted that the
solution content of Ca-montmorillonite changes to some extent over a wide
concentration range, whereas those of the Na-montmorillonite increase markedly
as a result of the development of diffuse double layers on opposing lamellae in
dilute solution. The characteristics of the swelling of Na-illite are essentially
similar to Na-montmorillonite in that the solution content attained at a given
suction is greater the more dilute the NaCl solution used to wet the clay (Fig. 5).
Although the swelling of Na-illite and Na-montmorillonite is qualitatively similar
the solution contents attained by the illite are less than those for montmorillonite
because the surface area of the illite is much less, being about 160 m2gg1
compared with 750 m2g-1 for the montmorillonite (Table VII, Fig. 5).
The Ca-illite solution contents (Table VII) are almost constant, differing little
between a concentration of 1 M CaCl, (K-I = 1.76
and distilled water.
Clearly diffuse double-layer theory cannot be used to describe the swelling
J. P. QUIRK
Solution Content (g . g-1 Clay) at 1 kPa Suction
for Compressed Cores of Na- and Ca-Montmorillonite
(Wyoming) and Na- and Ca-Illite (Willalooka) in Relation
to the Concentration of NaCl and CaCI, Solutions
between crystals of Ca-illite and the intracrystalline swelling of Camontmorillonite. The similarity in behavior of the two systems led Aylmore and
Quirk (1960, 1962) to conclude that contiguous Ca-illite crystals at regions of
overlap reside in a potential minimum similar to that which gives rise to the fixed
19 8, basal spacing for Ca-montmorillonite.
These observations, together with electron microscopy and pore size distribution measurements, are the basis on which Aylmore and Quirk (1959, 1960,
1962, 1967), Quirk (1968), and Quirk and Aylmore (1971) have described the
assemblages of Ca-illite crystals and the assemblage of montmorillonite quasicrystals as clay domains. The principal basis for the stability of these compound
particles in water is the operation of ion-ion correlation forces at areas of overlap
between crystals of illite and quasicrystals of Ca-montmorillonite; as a result the
swelling of Ca clays is almost insensitive to electrolyte concentration. It is
noteworthy that Terzaghi (1956) referred to “clusters of clay particles” as consisting of a “great number of crystals.” He also drew attention to the need to
establish “the seat and nature of the molecular forces which prevent the clusters
or piles floating in a suspension from being dispersed by the osmotic pressure
acting on their elements.” Bolt and Koenigs (1972) have also discussed the
significance of potential barriers at regions of particle overlap.
Olson and Mitronovas (1962) studied the compression of Ca-Fithian illite and
Electrolyte concentration had almost no effect on the position of the rebound
swelling curves. The most important effect of electrolyte concentration
seemed to be its effect on the geometric arrangement of particles that were
sedimented from dilute suspensions. The geometric arrangement of particles
seems to be a more significant variable than the osmotic repulsion between the
Ben Rhaiem ef al. (1987) examined the structure of Ca-montmorillonite (Wyoming) in equilibrium with a range of suctions by means of high intensity, lowangle X-ray diffraction (see Table VIII). It may be seen that, over the pressure
range 100 MPa to 3 kPa on the rewetting cycle, the average number of lamellae
(Ei) in a quasicrystal changes from 1 1 to about 8, indicating that these entities are
relatively stable. The average number of lamellae in an assemblage of quasicrystals (M)vanes from 400 to 90, indicating that the clay domains break up to form
smaller domains on the wetting cycle; this is part of the swelling process. If it is
assumed, for the sake of illustration, that it is permissible to divide by E then
the number of quasicrystals per clay domain would vary from 36 to 11. The
average thickness of the clay domain, obtained by multiplying the 4001) value
varies from 0.76 to 0.17 pm.
Tessier ( 1990) concluded, from electron microscope observations, that the
pores between the quasicrystals within a clay domain of montmorillonite are 3040 A across and that during wetting the clay domains split at some of these sites
to give several domains as is indicated by the decrease in (Table VIII). Murray
et al. ( 1985) examined the microstructure of the natural aggregates of 12 Queensland vertisols containing smectite by applying the general B.E.T. equation to the
low temperature nitrogen adsorption isotherms. They concluded that their results
were consistent with the presence of pores with widths of 14 to 28 A in the dried
materials. The surface areas of these materials varied from 100 to 200 m2g-I
when expressed on a clay basis (<2 pm). Using 750 m 2 g 1as a reference area
for montmorillonite, then a value of ii of 11 (Table VIII) would indicate an area
of 68 m2g-1. Aylmore and Quirk (1962) reported a nitrogen surface area of 38
m2g- 1 for Ca-montmorillorite (Wyoming). The surface area obtained using E,
an average, would include areas of overlap, between quasicrystals, which would
not be accessible to nitrogen.
Characteristics of Quasicrystals and Domains of Ca-Montmorillonite(Wyoming)
Influenced by Applied Suctionu
Low-angle X-ray diffraction results are from Ben Rhaiem ef al. (1987).
The average number of elementary layers in a clay domain of montmorillonite
The average number of elementary layers in a quasicrystal.
The average number of quasicrystals in a flay domain ( Z / E ) .
The average particle thickness [d(001) x MI.
J. P. QUIRK
As well as order in the c axis direction there is also a long-range order in the ab plane (Laffer et d.,
1969; Fitzsimmons et d., 1970; Greene e t a / ., 1973); these
authors prepared single lamellae of Ca-montmorillonite in a virtually electrolytefree solution. Optical birefringence developed after shaking a 0.3% suspension
for a period of I month, and electron microscope observation of the particles,
adsorbed onto a positively charged grid, indicated that single lamellae of about
0.3 pm across had condensed to give quasicrystals of more than 5 pm in the a-b
direction with a thickness of about 50 8,. The kinetics of the condensation was
hastened if the single lamellae were flocculated with 0.1 M CaCl,; birefringence
developed after a few days shaking and the quasicrystals were much thicker.
The stability of quasicrystals with respect to the degree of sodium saturation
has been investigated by Shainberg and colleagues (Shainberg and Otoh, 1968;
Shainberg and Kaiserman, 1969; Bar-On et al., 1970). They found that as the
percentage of exchangeable sodium increased up to 20% the electrokinetic velocity also increased but that the light scattering showed that the particle size was
unaffected. They therefore concluded that the Na+ ions were adsorbed on the
external surface of the quasicrystals. Quasicrystals progressively disintegrated as
increasing amounts of Na+ ions, between 20 and 60% exchangeable sodium,
occupied the interlamellar exchange sites.
Within a swollen mass of Ca-montmorillonite there are pores between the
individual lamellae within a quasicrystal, pores between the quasicrystals assembled in a clay domain, and pores between the clay domains themselves. The
definition of the clay domains and quasicrystal entities is primarily energetic in
that these entities represent relatively stable groups of particles connected by
overlapping areas in an adhesive potential minimum and have a near parallel
alignment because of their plate-shaped character. Within a clay soil matrix these
assemblages must be in random array to be consistent with the observed isotropic
behavior of the swelling of clay cores and clay soil aggregates.
Because two lamellae in contiguous quasicrystals of montmorillonite can conform more closely than two crystals of illite, perhaps 50 8, thick, the stability of a
quasicrystal of montmorillonite and an assemblage of quasicrystals would be
greater than an illite domain with the additional possibility that individual
lamellae could be shared between contiguous quasicrystals. Where there is a high
degree of conformity between illite surfaces the greater surface density of charge
would give rise to a strong attraction because of ion-ion correlation forces. The
distribution of particle thicknesses in illite may be an important feature contributing to the stability of a domain as the thin particles would interleave with coarser
The less than 0.2-pm fraction of Willalooka illite has a nitrogen surface area
of 227 m2g-' (Alymore and Quirk, 1967), indicating that there would be some
very thin particles since the average thickness would be less than four elementary
silicate sheets thick. Because of their flexibility these thin particles could act as
an aggregating agent and could play a significant role in determining long-range
order in the a and b directions in a clay domain. The stability of an illite domain
within a soil would be considerably enhanced by the large suctions developed
during drying. Aylmore and Quirk (1962) have shown that clay domains of
Willalooka illite, composed of individual crystals of about 50 8, thick and 700 8,
in lateral extent, have a lateral extent of about 5 km.
Lebron et al. ( 1993), using photon correlation spectroscopy, investigated the
nature of clay domains of Silver Hill illite which has 1.5 units of charge per unit
cell formula (Sposito and LeVesque, 1985); this is a similar surface density of
charge to that of Willalooka illite. They found that the Ca-saturated illite had a
much larger particle size than the Na-saturated illite. They also reported that the
Ca clay domains broke down to give smaller particles at SAR values in the range
of 10 to 15 and also that the particle diameter decreased by a factor of 2 to 3 ,
indicating a large decrease in particle size since this would vary as the cube of the
1. Crystalline Swelling of Calcium Smectites
SIade and Quirk (1991) reported d(OO1) spacings in distilled water for the
following smectites for which the total charge and tetrahedral contribution to the
charge are shown in parentheses: hectorite, 20.7 8, (0.70,0.04);Wyoming, 19.1
8, (0.74, 0.08); Otay, 18.5 8, (1.04, 0.06); Nibost, 17.8 8, (1.14, 0.74); and
Drayton 16.7 A (1.06, 1.02). Thus the extent of the crystalline swelling of Casmectites is influenced by the magnitude of the charge and also by its structural
Figure 6 shows the effect of CaCl, concentration on the transition from the
spacing in water to about 15.5 8,. Posner and Quirk (1964a,b) reported that this
transition is reversible with respect to concentration and took place between 1
and 2 molal for Li, Na, Mg, and Ca chloride solutions and that the spacing did
not decrease appreciably at higher concentrations. Since the electrolyte did not
penetrate the interlamellar space until the transition had taken place, the transition is an osmotic phenomenon. The osmotic pressure of the solutions is indicated in Fig. 6. Figure 6 shows that the 15.5 8,spacing for Wyoming montmorillonite is attained at a pressure of I3 MPa, the transition for Otay montmorillonite
occurs in two stages with the first at 3.4 MPa and the second at about 13 MPa,
the transition for Nibost beidellite is from 17.8 8, to about 15 8, and also occurs
in two stages but at lower pressures. Each 3 8, increase in spacing contributes
about 0.10 g H,Og-l clay if 650 m2 g-I is taken as the internal area.
The different behavior of the smectites shown in Fig. 6 for pressures less than
J. P. QUIRK
5 4 1
Molality of CaCI,
Osmotic Pressure (MPa)
Figure 6 The d(001) spacings for Ca smectites as intluenced by the total charge, the structural
origin of the charge, and CaCI, concentration. Since salt does not enter the interlamellar spaces until
the 15.5 8, spacing is attained, the transition is in response to the osmotic pressure of the solution.
(After Slade and Quirk, 1991 .)
3.4 MPa (0.5 molal), the range of principal interest for agriculture, is such that
Wyoming montomorillonite cannot be regarded as representing the smectites
generally. It should be noted that Drayton, a subsoil clay is classed as subplastic
(Nomsh and Tiller, 1976).
2. Clay Swelling
The swelling of Ca clay cores in an overconsolidated state has been investigated with respect to applied suction and electrolyte concentration by Aylmore
(1960) and Aylmore and Quirk (1962, 1966). The swelling to equilibrium with 1
kPa suction and 1 M CaCI, for Ca-kaolinite (Rocky Gully), Ca-illite (Willalooka), and Ca-montmorillonite (Wyoming) expressed as a percentage increase
in the original volume of the core is, respectively, 3 1,45, and 85%. However, if
the solution uptake by montmorillonite is corrected for the interlamellar water
(0.30 cm3 g-'), then this last figure becomes 43%. The original porosity and
surface area of these clays are, respectively, 0.185, 0.188, and 0.135 cm3 g-'
and 36, 160, and 38 m* g-1. This information reveals that swelling cannot be
interpreted simply in terms of the nitrogen surface area involved. The work of
Ben Rhaiem et af. (1987) shows that quasicrystals separate to give a larger
number of quasicrystals and thereby increase the external surface area of Wyoming montmorillonite during swelling.
The swelling in a I M CaCl, solution ( K - ' = 1.76 A) and other evidence
indicates that diffuse double-layer forces make little contribution so that swelling
must result from a balance between attractive forces (ion-ion correlation and van
der Waals) and repulsive forces (water structural). The suction in the soil water
acts to restrain swelling, and as the suction is reduced the balance of the forces
favor swelling. A simple three plate model can be used to illustrate the interplay
of the forces; in this, two plates in parallel alignment are separated by a third
plate in such a way that a pore is formed with a surface separation similar to the
thickness of the middle plate and in the areas of overlap the surfaces are virtually
in contact. With the adsorption of water these contacting surfaces may be separated by distances up to 10 A. Water structural forces are operative in the slitshaped pore between the separated plates, and in the pores, generated by overlap,
ion-ion correlation forces, together with van der Waals forces, act to restrain
swelling. The available evidence indicates that the magnitude of the water structural forces is influenced by surface density of charge and that the ion-ion
correlation forces are a function of both the surface density of charge and the
electrolyte concentration. The surface density of charge of the kaolinite, illite,
and montmorillonite, referred to earlier, are, respectively, 0.106, 0.260, and
0.119 Cm-2; 1 unit of charge per 152, 62, and 135 A 2 of surface.
Figure 3 indicates the large repulsive pressures which can be generated by
water structural forces in relation to the distance of separation of surfaces for slitshaped pores. For surfaces with a smaller surface density of charge these pressures would be lower. To quantitatively appreciate the operation of these forces it
would, ideally, be appropriate to have pore size distribution measurements before
swelling takes place and at various stages of the swelling process. For a sample
of Urbrae loam B horizon (80% clay, surface area of 116 m2g-l), Murray and
Quirk (1980a) measured the pore size distribution of the dry clay material and in
equilibrium with a suction of 10 kPa; the modal pore sizes were 36 and 63 A. The
swollen porosity was preserved by the displacement of water by increasing the
concentration of dioxane in dioxane-water mixtures, followed by the displacement of dioxane by liquid C 0 2 which was then removed at its critical point.
Features of the pore size distributions could reasonably be interpreted as arising
from variable crystal thicknesses. The nitrogen desorption isotherm enabled pore
sizes less than 300 A to be measured. Pore sizes larger than this value could not
be measured because the pressure involved in Hg injection porosimetry caused
collapse of the pores and this, as well as the preservation of the swollen porosity,
indicates that the original swollen structure was, at least partially, maintained.
J. P. QUIRK
Although the initial porosity was preserved by freeze-drying techniques the pore
size in the < I 0 0 A range was identical with the oven-dry material because the
water migrated and was frozen elsewhere in the clay matrix.
In considering the application of information such as that given in Fig. 3 to the
swelling of a clay-water system, the range of surface separations (pore sizes) in
the system arising from the range of crystal thicknesses, needs to be taken into
account. The clay mass is thus more complex than the three plate model used
earlier, as can be seen from Fig. 8 in which the presence of wedge-shaped pores
is evident; however, the description of near parallel alignment is still apposite.
Another contribution to the swelling process is the interaction between the clay
domains. Since the domains are in random array and swell unidimensionally they
will compete for space and thereby enlarge the porosity as the suction is decreased.
The swelling results of Aylmore and Quirk (1962) given in Table VII show
that the effect of CaCI, concentration on swelling is greater for montmorillonite
than for illite; the surface charge of illite ( 1 unit of charge per 62 A) is in the
The variation of ion-ion correlation attractive pressure with CaCI, concentration has been considered by Kjellander et al. (1990) who for a charge density of 1
unit charge per I35 A 2 calculated that the pressure increased from I .7 to 5.5 MPa
over the concentration range from zero (counterions only) to 2 M CaCI,. For 1
unit of charge per 60 A2 the pressure increased from about 15 to 30 MPa over the
same concentration range.
The hysteresis between the wetting and drying curves for these materials is
accentuated with increasing CaCI, concentration; this would be consistent with
the operation of ion-ion correlation attractive forces on the wetting cycle. On the
drying cycle the solution content decreases only slowly to a suction of 0.1 MPa
and then falls more rapidly at greater suctions. This is thought to be associated
with the strength of the structure formed on the wetting cycle. On the drying
cycle the water structural forces between surfaces removed from the potential
minimum would be operative and, furthermore, considerable work would be
required to bring opposing surfaces into the potential minimum because the two
sets of ions associated with each surface have to be shared at the midplane in the
Calcium ions at clay surfaces have a dual role; they are responsible for the
attractive ion-ion correlation pressures at regions of close approach of surfaces,
and in larger pores the Ca ions, because of their large concentration at the clay
solution interface, perturb the normal structure of water which is the basis for
the repulsive pressures described as the structural component of disjoining pressure or secondary hydration forces. It is also noteworthy that these countervailing forces each increase considerably in magnitude with increasing surface density of charge. The balance between the two sets of forces is obviously a delicate
one since 10 to 20% Na+ ions, or even less, can give rise to adverse physical
behavior at low electrolyte levels. If it were not for the presence of ion-ion
correlation forces within a clay matrix or soil aggregate it would be virtually
impossible to use soils since water structural pores would dominate.
Aylmore and Quirk (1966) have reported on the swelling of four Ca-kaolinites:
Mercks I1 ( 1 1 m2g-1, 0.18 Cm-z), Malone (17 m2g-1, 0.24 Cm-,), Rocky
Gully (36 m2g-', 0.11 Crn-,), and New Zealand (40m2g-', 0.08 Crn-,). The
initial porosity of these materials was similar and the percentage increases in
volume, from the dry state, for the clay cores in equilibrium with 1 M CaCI, and
I kPa were, respectively, 13, 22, 3 I , and 50; when these percentages are expressed in relation to the surface area of each clay the following results emerge:
1.18, 1.29, 0.86, and 1.25. These ratios may be compared with that for Willalooka illite of 0.3 which suggests that the kaolinites behave as single crystals.
The basic swelling mechanism must be water structural forces accompanied by a
general relaxation of the clay as the suction is reduced.
VI. SURFACE AREA AND PORE SIZE
The B.E.T. equation and the Kelvin relationship [Eq. (2)] have been applied to
low temperature nitrogen adsorption-desorption isotherms to obtain, respectively, surface area and pore size distribution of clays and soils (Aylmore and
Quirk, 1967; Murray and Quirk, 1980a,b, 1990a,b; Murray et al., 1985). Murray et al. (1985) have critically assessed the assumptions used in the B.E.T. and
Kelvin equations as applied to clay systems. In applying the Kelvin equation,
allowance is made for the thinning of multilayers on the surface of pores already
emptied by capillary evaporation on the desorption isotherm (Aylmore and
Quirk, 1967). The procedure adopted is to obtain the volume of capillary evaporate removed over a series of intervals @lpo values) on the desorption isotherm.
Since the Kelvin equation provides the average slit width for each vapor pressure
interval, this value, together with the change in volume in the vapor pressure
interval, is used to yield a surface area. The sum of the surface areas for all of the
intervals provides the desorption isotherm surface area. The pore size distribution is also obtained and the procedures involved are set out by Aylmore and
Quirk (1967) and Murray el al. (1985).
Figure 7 shows the N, sorption isotherms for Ca-Willalooka illite (a separate
sample from that quoted earlier with an area of 160 m2g-') and the natural
J. P. QUIRK
0 . 2 . 4 .6 .8 1
pore width (nm)
pore width (nm)
Figure 7 Nitrogen sorption isotherm and derived cumulative pore size distribution (a) natural
aggregates of a vertisol (total porosity, 0.13 cm3g-I) and (b) Ca-Willalooka illite-compressed cores
(total porosity, 0.23 cm3g-I). (After Murray er al.. 1985.)
aggregates of a Queensland vertisol (No. 10). It can be noted that the plot of
cumulative volume versus pore size indicates modal values of about 4 and 3 nm
respectively. The slope of the desorption isotherm for Willalooka illite indicates a
mixture of slit- and wedge-shaped pores. The isothern for the vertisol shows a
large hysteresis and its shape indicates a predominance of slit-shaped pores.
There is a considerable variation in the isotherms for the other vertisols studied
with some having a sloping desorption isotherm.
Figure 7 also shows that the pore size in these materials is predominantly < 10
nm. For the vertisol, 50% of the pore space is in pores less than < 3 nm and 95%