Tải bản đầy đủ - 0trang
III. Structure of Allophane Soils
T. MAEDA ET AL.
The structure of imogolite in the 10- to 100-A range is known from electron
micrographs supplemented by X-ray diffraction, electron diffraction, and salt
absorption measurements (Wada et aZ., 1970; Wada and Henmi, 1972). The unit
is a hollow tube with inside and outside diameters of 7-10 and 17-21 A,
respectively. A number of tubes lying parallel form threads of 100 to 300 A in
diameter. Three different classes of pores would be present in such a material:
intraunit pores of about 10 A in diameter, intrathread (interunit) pores of the
same or slightly smaller diameter, and interthread pores of several hundred
Angstrom (A) units diameter.
The “unit particle” of allophane has been defined (Kitagawa, 1971) as a
spherical particle with a diameter of about 55 A (Birrell and Fieldes, 1952). The
density of this particle is about 1.9 g ~ m - Gtagawa
(1971) has assumed that
these unit particle spheres exist in close packing in the air-dry state. This forms
the “microaggregates” commonly found. On heating, the layer of adsorbed
water is lost, and the spheres come closer together. Grinding distorts the unit
particle spheres allowing closer packing, which, in the electron micrographs,
gives the appearance of sheets rather than the individual spherical particles of the
The close packing of these 55-A diameter units would again result in voids about
10 A in diameter, so the water absorption properties would be simdar t o
imogolite. Irregular packing would produce some larger pores.
These models, therefore, explain many observed properties. The decrease in
surface area on grinding is due to collapse of some of the small pores. Kitagawa
(1971) has shown that calculation of particle size from surface area and density
also gives values of about 55 A diameter. He found that phosphate adsorption
was not decreased by grinding on drying, which argues against any chemical
change in the surface, specifically in the number of hydroxyls at the surface.
Interparticle bonding is then by physical forces that would be weaker than if
chemical bonding were involved.
These measurements were made on a series of allophane soils from Japan. It is
not known whether all allophanes have this same “unit particle.” Measurements
by Rousseaux and Warkentin (1976) of water vapor absorption show a maximum in pore volume at diameters of 7-10 A for allophanes from the Caribbean
and from Japan. The Caribbean samples show a narrower size distribution that
could result from closer and more regular packing of unit particles. Fujiwara and
Baba (1973) calculated effective pore size from nitrogen absorption measurements, and found a maximum at 25 A. The volume of these 25-A pores was
about one-twentieth of the volume of 10-A pores found from water absorption
by Rousseaux and Warkentin (1976). These measurements confirm the unit
particle in allophane soils.
Measurements of water vapor absorption are of necessity made on dried
samples. It is reasonable to assume that undried samples consist of the same unit
PHYSICAL PROPERTIES OF ALLOPHANE SOILS
particles, but that they are further apart and more irregular. Such a model would
explain water retention, plasticity, etc. It is more difficult to relate differences in
these properties among soils, wet or dry, to different structures in the model.
Some allophane soils become nonplastic on air drying, others do not. This could
be due to differences in arrangement of unit particles, or to different nonallophane components in the soils. Also, it is not known how differences in chemical
composition such as the AI/Si ratio affect the structure, although these differences are correlated to differences in physical properties (Rousseaux and Warkentin, 1976). The Caribbean samples, for example, have lower Si02/A1203ratios
than the Japanese samples.
Ito (1 964) postulated that allophane structure consisted of individual particles
and massive particles (structure units), with only the latter affected by drying.
The individual particles retain their undried properties, and different phases of
shrinkage are due to different effects of individual particles and structure units.
The irreversible changes in physical properties of allophane soils on drying set
them apart as a separate group of soils. In the extreme, allophane soils change
from highly plastic when wet, to sandy when dry. This has been described by
many workers, e.g., Sherman et al. (1964), who suggest research on changing the
colloidal properties of allophane soils by promoting dehydration. The particles
become cemented together to form units of sand size that are sufficiently strong
to withstand the usual manipulation in field or laboratory. Neither the forces
involved in this cementing nor the fabric of the units have been adequately
described. A tentative model for structure is described in Section 111, B.
The observed pore structure of allophane soils has been described by a number
of workers in relation to permeability. Tabuchi et al. (1963) described the
channels for water flow that they observed in thin sections of surface and
subsoils. Interaggregate pores conducted water in surface soils, tubelike pores
were found to conduct water in subsoils. Takenaka et al. (1963) report similar
results. Nagata (1963) found that plots of log air permeability against air
porosity were linear functions which could be related to the kind of structure in
the soil sample. For some soils log K, versus va consisted of two straight-line
portions, depending upon changes in structure with void size.
A more detailed description is available for the pores in pumice (Borchardt et
al., 1968; Maeda et al., 1970; Tsujinaka et aZ., 1970). On the basis of waterretention measurements carried out in different ways, Maeda et al. (1970)
distinguished dead, active, semiactive, secondary active, and semidead pores. The
active and semiactive pores usually occupy the largest volume, but semidead and
dead pore space may be large in some samples of pumice. The pumice materials
have been described by Sasalu (1 957).
Some information on structure, especially on the fabric component, can be
obtained from measurements of rheotropy or thixotropy of allophane soils. The
behavior of allophane on mechanical manipuiation is usually not true thixo-
T. MAEDA ET AL.
tropy, i.e., a reversible sol/gel transformation. The changes on remolding described by Takenaka and Yasutomi (1965) are discussed in Section V, B.
A number of papers have reported experiments with soil conditioning chemicals; the general impression is that these chemicals are only marginally effective
in promoting aggregation in allophane soils. This may be partly due t o mixing
the chemicals with soils at water contents which are not optimum for aggregation. Sudo and Suzuki (1963) found that sodium alginate increased stable
aggregates in an allophane soil but the polyelectrolyte CMC did not. They
report, in their literature review, that additives are generally considered to have
little effect on aggregation of allophane soils. Kawaguchi et al. (1963) report
that bentonite added to polyvinyl alcohol is effective in aggregation. Fujioka et
d. (1965) found an effect of soil conditioners on allophane soils, but Terasawa
(1 967) found very little effect of synthetic polymers on aggregate formation.
B. MODEL FOR PHYSICAL PROPERTIES O F ALLOPHANE
What is required is a model for structure, including fabric, of allophane soils.
This model must explain the known physical properties of allophane and their
changes on drying. The model should predict other physical properties. It should
also be related t o important differences in chemical composition and properties,
for example the Si/Al ratio. The model must describe structure in the 0.01- to
100-pm range, where physical properties can be explained.
Measurements to date on allophane soils allow only a general specification of
this model. The unit particle of about 50 A is well established and can be taken
as the starting point. This particle has an “internal” water content. The unit
particle has been established in dried samples and is assumed to be present also
in wet samples. These unit particles are then weakly bonded together to form
domains in the diameter range of 0.01 to 1 pm. The modal size may be 0.05 to
0.1 pm. The voids between particles in these domains account for the large water
retention above 15-bar suction. Drying brings the particles closer together,
increasing the bulk density of the domains and decreasing water retention above
15 bar. Irreversible changes occur when the unit particles come sufficiently close
together to allow strong bonding between unit particles. These domain units
may be the clay-size grains measured in a grain size determination. They are not
broken up on remolding and account for the high water content at the plastic
limit. They are the units that move on plastic readjustment. Organic molecules
are held within the domains.
The domains are arranged in clusters in the size range of 1-100 pm. These
units have weaker bonding than the domains. Remolding breaks up the clusters,
releasing water held within the cluster. Drying causes shrinkage of the domains
and rearrangement into clusters of highei bulk density. In some allophane soils
PHYSICAL PROPERTIES OF ALLOPHANE SOILS
the bonds holding the clusters together after drying are strong, and the clusters
become the units of dried soils. Bonding within clusters would then aIso
contribute to irreversibility. This may be true only in soils that are composed
entirely of allophane and do not have crystalline minerals or oxides mixed in.
Water in the plant-available range is held withm the clusters.
The water-retention curves for allophane soils show an approximately linear
change of water content between 0.03 and 1 bar, with decreasing amounts of
water retained below and above these values. The break at 1 bar indicates a
change at void diameters of 30 pm; the upper boundary of cluster size could be
drawn there. This would put much of the plant available water between clusters.
Kubota (1971) has studied the formation of units in the sand- and silt-size
range. Drying is necessary for pedogenic formation of these units. These units
fall within the range defined here as “clusters.” These units are stable against
dispersion if the allophane soil has been dried. Dehydration alone can cause
irreversible binding. The formation of these clusters is best studied on subsoil
samples of allophane soils that have been continually moist. Presumably the
clusters are already present in surface soil horizons that have been previously
Kubota (1971) divides the clay-size grains into three types on the basis of their
potential for forming sand- and silt-size grains on drying. The clay content was
measured on sonic-dispersed samples of subsoils of wet allophanes. Type I is
active free clay which can form large grains on air drying. This is measured as the
difference between clay content of moist and air-dry soils on shaking. Type I1 is
the aggregated clay that is released from moist soils by sonic dispersion. These
are pedogenically formed aggregates. Type 111 is inactive free clay not affected
by air-drying. Ttus is the measured clay content of an air-dry soil. Type I clay
content increases from the Ap to the B2 horizon, while Type 111 decreases. Type
I1 remains approximately constant with depth.
Monolayer adsorption of water vapor on allophane is strong but multilayer
adsorption of more than two water layers is weak compared with layer silicate
minerals. Therefore unit particles of allophane can come in close contact.
Kubota (1971) also states that hydroxyaluminum groups are the site for adsorption of water molecules; they are also the sites for bonding of unit particles in an
The details remain to be fdled into this model. The size ranges of the fabric
units, domains, and clusters, have been chosen mostly for convenience of
description. They will undoubtedly need refinement.
Neither the size boundaries nor the terminology of domains and clusters are
suggested as being definitive. They are both used here for convenience. The
clusters may be better described as microaggregates. However, Kitagawa (1971)
refers to the unit particles of 55 A as microaggregates, and Kubota (1971) refers
to sand- and silt-size units as aggregates.
T.MAEDA ET AL.
IV. Physical Characteristics of Allophane Soils
A. VOLUME CHANGE
Wet allophane soils show a large volume decrease on air-drying, and a limited
volume increase on rewetting. Most of the volume change is irreversible. This
distinguishes allophane soils from swelling mineral soils, where the volume
changes are more nearly reversible. The amount of shrinkage of an allophane soil
depends upon the initial water content and the fabric changes during drying,
both of which depend upon the allophane content. Many physical properties
change in association with this volume decrease, e.g., permeability increases.
However, volume change in the field cannot be predicted quantitatively from
shrinkage measured on small samples because the boundary conditions are
Wet allophane samples dried to intermediate water contents, suctions below
10-20 bars, will regain most of the volume on rewetting. Volume change is
approximately reversible over this range (Takenaka, 1961).
Volume change curves can be plotted as changes in measured volume, linear
dimensions, bulk density, or void ratio with change in water content. The
measurements are not precise because it is difficult to measure volume repeatedly on a sample as the water content changes. The precision is sufficient t o
characterize shrinkage on drying, but greater precision is desirable in describing
volume change for the water-retention curve (Section IV, B).
The general nature of the shrinkage curves is shown in Fig. 3 for a highly
allophanic soil (Cl) from Dominica, West Indies, and for a soil (Nl) from
Hokkaido, Japan, with low allophanic properties (Warkentin and Maeda, 1974).
There is a break in the volume change curve which can be called a shrinkage
limit, but the volume change at higher water contents is not “normal” shrinkage
where volume decrease equals water content decrease (Takenaka, 1961; Ito,
The shrinkage limit is less pronounced and occurs at a higher water content as
allophane content increases. The shrinkage limit occurs between 50 and 100%
water for allophane soils, a much higher value than for crystalline clays (Takenaka, 1965; Warkentin and Maeda, 1974; Soma and Maeda, 1974). Takenaka
(1965) found that the shrinkage limit occurred near pF 6 for a sample of Kanto
loam. He related shrinkage to soil suction values and showed that the amount of
shrinkage depended upon the rate of drying. The measured water content at the
shrinkage limit also increases with increasing organic matter content of the soil
(Takenaka, 1973). He measured a 15% increase in shrinkage limit for 10%
increase in organic matter up to 25%. Remolding does not change the value of
the shrinkage limit (Takenaka, 1965), but drying increases the shrinkage limit.
The usefulness of the shrinkage limit is in the information it gives about fabric
and structure of allophane soils. The high water content at which the slope of
PHYSICAL PROPERTIES OF ALLOPHANE SOILS
Water Conieni, %
FIG. 3. Shrinkage curves for two allophane soils. Reproduced from Soil Science Society
of America Proceedings, Volume 38, Page 375, 1974 by permission of the Soil Science
Society of America. X , Field moisture; @, air dry; 0,oven dry.
the shrinkage curve changes and the high value of residual shrinkage (between
the shrinkage limit and zero water content) indicates a random arrangement of
units. Measured shrinkage is isotropic, again indicating random arrangement. For
crystalline minerals, the shrinkage limit is lowest for high swelling soils. The
shrinkage limit is, therefore, diagnostic for allophane soils (Warkentin and
Despite the large amount of shrinkage on drying, one would predict little
visible cracking for allophane soils in the field because the shrinkage is taken up
in small spaces between clusters (Section 111, B). Cohesion of allaphane is low
and it decreases if the samples are dried (Soma and Maeda, 1974). AUophane
soils do not form dry clods with dimensions of tenths of a meter.
The effect of remolding a soil depends upon its degree of consolidation
(Croney and Coleman, 1954). Remolding an overconsolidated soil exposes new
surfaces for water retention. AUophane soils are underconsolidated. Remolding
breaks some of the fabric bonds, decreases the soil suction, and increases the
amount of shrinkage (Takenaka, 1965).
B. WATER RETENTION
This section will deal with soil water characteristics measured on allophane soil
samples in the laboratory; the mechanisms of water retention will be discussed.
The field water regime is described in Section IV, D.