HOSTED BY Available online at www.sciencedirect.

com

ScienceDirect
Soils and Foundations xxx (2017) xxxxxx
www.elsevier.com/locate/sandf

Evaluation of specic surface area of bentonite-engineered barriers

for Kozeny-Carman law q
I. Kobayashi a,, H. Owada b, T. Ishii b, A. Iizuka c
a
Kajima Corporation, Tokyo, Japan
b
Radioactive Waste Management Funding and Research Center, Tokyo, Japan
c
Kobe University, Kobe, Japan

Received 11 July 2016; received in revised form 25 April 2017; accepted 16 May 2017

Abstract

The impermeability of bentonite is dependent on montmorillonite microstructures and varies with chemical alteration caused by
related changes. The Kozeny-Carman equation is viewed as a rational hydraulic model for bentonite because it incorporates microstruc-
ture parameters such as specic surface area in addition to macrostructure parameters such as the void ratio. In the study reported here, a
technique for measuring the specic surface area of compacted bentonite under constant volume conditions using X-ray diraction was
developed, and the hydraulic conductivity of various bentonite materials was evaluated based on the measurements thus obtained. The
applicability of the Kozeny-Carman hydraulic model for evaluation of impermeability of bentonite was claried.
2017 Production and hosting by Elsevier B.V. on behalf of The Japanese Geotechnical Society. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Bentonite; Montmorillonite; Hydraulic conductivity; Specic surface area; Kozeny-Carman law

1. Introduction the groundwater and the radionuclides and to slow the

Finding methods for the nal disposal of radioactive
waste is a pressing issue in Japan, and ongoing studies have
been conducted on the feasibility of disposal projects from
a scientic perspective. The countrys planned disposal
methods include encasing trans-uranium (TRU) and other
high-level radioactive waste in facilities built under stable
bedrock at least 300 m below the ground surface. This tech-
nique is referred to as geological disposal; it involves the
use of a system comprised of both natural and engineered
barriers. The system is expected to retard contact between
q
This manuscript was submitted to the Special Issue on the
International Symposium on Geomechanics from Micro to Macro
IS-Cambridge 2014 (Vol. 56 No. 5).
Peer review under responsibility of The Japanese Geotechnical Society.
Corresponding author.
E-mail address: koba13@kajima.com (I. Kobayashi).
migration of the radionuclides into the biosphere even if
the groundwater becomes contaminated (JAEA, 2000).
Fig. 1 shows a conceptual model of a geological disposal
facility for TRU waste in Japan. The facility has engineered
barriers, including cementitious types to reduce diusion
and bentonite types to minimize hydraulic conductivity
(FEPC and JAEA, 2005).
An engineered bentonite barrier is a compacted soil mix-
ture of bentonite and silica sand. Its low hydraulic conduc-
tivity comes from montmorillonite (the principal mineral in
bentonite clay), and therefore, depends on the type and
content of the montmorillonite used.
The primary particles of montmorillonite have a sheet-
like form, a permanent electric charge and a capacity for
cation exchange. Various interlayer cations are adsorbed
between the primary particles. Montmorillonite with Na
interlayer cations is known as Na-montmorillonite, and
bentonite whose principal mineral is Na-montmorillonite

https://doi.org/10.1016/j.sandf.2017.08.001
0038-0806/ 2017 Production and hosting by Elsevier B.V. on behalf of The Japanese Geotechnical Society.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
2 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
Fig. 1. Basic concept of TRU geological disposal (FEPC and JAEA,
2005).
is known as Na-bentonite. The hydraulic conductivity of
montmorillonite depends on the type of cations, as its
microstructure also diers with the interlayer cations. Na-
montmorillonite exhibits lower hydraulic conductivity than
other types of montmorillonite because it retains many
water molecules among its primary particles (known as
hydration or swelling), resulting in the absence of large
voids under saturated conditions. Accordingly, Na-
bentonite is used as a reference material for engineered
bentonite barriers in Japan.
However, interlayer cations are readily displaced
depending on the type and the ionic strength of the cations
contained in the pore water due to cation exchange reac-
tions. As a result, the use of Na-bentonite for building engi-
neered barriers in geological disposal facilities may increase
Fig. 2. Modeling based on tting experiment results (adapted from Kikuchi et al., 2003).
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
the hydraulic conductivity due to these cation exchange
reactions depending on the subsequent compositional
changes in the pore water. For example, after passing
through an engineered cementitious barrier, the groundwa-
ter (cement leachate) becomes highly alkaline and Ca ion-
rich. As cement leachate permeates bentonite, the inter-
layer cations in the montmorillonite (in this case, Na ions)
are displaced by the Ca ions. This results in the alteration
of the bentonite to Ca-bentonite, which swells less than
Na-bentonite. Thus, large voids can be expected in Ca-
bentonite even under saturated conditions unless the mate-
rial is compacted to a high dry density, which could
increase the hydraulic conductivity. Additionally, the reac-
tion of highly alkaline pore water (such as cement leachate)
with montmorillonite over an extended period may cause
the montmorillonite to dissolve and create a non-swelling
mineral. Similarly, the hydraulic conductivity of engineered
bentonite barriers may increase due to the reduced mont-
morillonite content caused by such an alteration and the
resultant increased presence of large voids.
Against such a background, an assessment to determine
the feasibility of radioactive-waste geological disposal facil-
ities with a lifespan of tens of thousands of years must
involve a prediction of the temporal changes in the perfor-
mance of the engineered barriers in consideration of the
related chemical interaction. Therefore, this study was con-
ducted to examine a hydrologic model for engineered bar-
riers that enables an analysis of the microstructural
changes caused by the chemical interaction.
A typical approach to the development of hydrologic
models for engineered bentonite barriers involves the
experimental determination of the relationship between
the dry density and the hydraulic conductivity of the
barrier material as well as the establishment of a tting
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
function (see Fig. 2). Methods are currently being studied
that will enable the determination of the changes in
hydraulic performance resulting from the chemical alter-
ation using an engineered geochemically altered bentonite
barrier followed by an experimental clarication of the
relationship between the barriers dry density and hydrau-
lic conductivity. From this information, a tting function is
established and the coecients for both pre- and post-
alteration functions are interpolated according to the
degree of alteration. This approach enables the expression
of forecasted events for the experiment, but is not applicable
to the prediction of non-forecasted events. By way of example,
as shown in Fig. 2, it enables representation of the experi-
mental results to some degree within the test range with dry
densities between 1.0 and 1.8 Mg/m3, but is limited at
lower levels due to the reduced hydraulic conductivity.
Meanwhile, the hydrologic models proposed by Hazen
(1892), Creager et al. (1945), Terzaghi and Peck (1967)
and others take the microstructures of earth materials into
account via consideration of the representative particle
size. However, when a cation exchange reaction causes
the interlayer Na-bentonite cations to be displaced by Ca
ions, no changes in the primary particle size occur, but
the hydraulic conductivity increases (e.g., Sasakura et al.,
2004, Pusch and Yong, 2006; Morodome, 2008). Conse-
quently, particle size is not a rational indicator of the
microstructural changes resulting from the bentonite chem-
ical reactions.
In contrast, the Kozeny-Carman hydrologic model
(1956) considers microstructures based on the specic sur-
face area of the material in question. In previous studies
(Fig. 3), changes in the hydraulic performance of bentonite,
Fig. 3. Conceptual microstructural model for bentonite (Kikuchi et al., 2003) and structural changes caused by pore water (adapted from Kikuchi et al.,
2003).
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
3
resulting from cation exchange reactions, have been associ-
ated with alterations in the contact area between the pore
water and the soil particles caused by changes in the swel-
ling properties. Hence, the Kozeny-Carman relationship
appears suitable for evaluating the hydraulic performance
of engineered bentonite barriers. However, its application
as a hydrologic model requires knowledge of the specic
surface area of bentonite under the prescribed dry density
and moisture conditions, as this depends on the barriers
pore water composition and moisture conditions.
This implies that conventional specic surface area mea-
surement methods (such as the BET approach used for dry
powder samples and the mercury intrusion technique, in
which pore water composition is not considered) are not
applicable.
In this study, a method for determining the specic sur-
face area of compacted bentonite using the X-ray dirac-
tion (XRD) method under constrained volume and
saturation conditions was developed and used to evaluate
the hydraulic conductivity of various bentonite types.
This paper presents a revision of the research by
Kobayashi et al. (2011) including their recent ndings.
2. Microscopic analysis of smectitic minerals
The reference material for engineered barriers, designed
for radioactive waste disposal, is the Na-bentonite pro-
duced in Yamagata Prefecture (see Fig. 4 for the related
mineral composition). It is made of montmorillonite (a
swelling mineral) and non-swelling minerals, making its
constituents limited. However, the constituent percentages
vary with each production lot, with the principal mineral
4 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
Fig. 4. Mineralogical composition of sodium bentonite mined in Yam-
agata Prefecture, Japan (Ito et al., 1994).
10-5
Hydraulic conductivity k (m/s)
-7
10
10-9
10-11
10-13
10-15
0.0 0.5
Dry density (Mg/m3)
Fig. 5. Comparison of hydraulic conductivities of bentonite and kaolinite
(adapted from Sasakura et al., 2004).
(montmorillonite) generally varying from 50 to 60%. The
material lot used in this study had a montmorillonite con-
tent of 59%.
Fig. 5 compares kaolinite (a non-swelling mineral) and
Na-bentonite, showing the relationships between dry den-
sity and hydraulic conductivity. The graphs reveal that at
the same dry density, the hydraulic conductivity of Na-
bentonite is much lower than that of kaolinite. This implies
that consideration of the dry density and other parameters
representing the macrostructures alone is ineective for
expressing the dierences in hydraulic conductivity. Table 1
shows the parameters (particle shape, particle size and
specic surface area) used to represent the microstructures
of kaolinite and montmorillonite. The smectite family
(which includes montmorillonite) has smaller particles
and a far greater specic surface area than kaolinite.
Assuming that the dierences in hydraulic conductivity
shown in Fig. 5 are attributable to these microstructures,
it appears necessary that, for evaluation, hydraulic models
of bentonite must at least be comprised of both micro and
macrostructural parameters.
Table 1
Particle shape, particle size and specic surface area of various clays (CSSJ, 1987).
Clay mineral
Smectite (montmorillonite)
Kaolinite
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
Fig. 6 shows a schematic microstructural model of
montmorillonite. The primary particles of montmorillonite
have a sheet-like structure, as also shown in Fig. 3, with the
major and minor axes ranging from 100 nm to 50 nm. The
sheet thickness is approximately 1 nm. Secondary particles
of montmorillonite are considered to be a lamination of
1020 primary particles (Shibutani et al., 1999).
Fig. 7 shows the microstructural changes in montmoril-
lonite caused by water permeation. The primary particles
form the laminar structure of secondary particles, with
the spaces between the primary particles known as internal
voids and the others known as external voids. As the cation
concentration of the internal voids is higher than that of
the external voids, water molecules migrate from the latter
Na-bentonite to the former to compensate for the dierences in concen-
(Sasakura et al., 2004) tration upon water penetration. Water selectively perme-
Kaolinite ates the internal voids, causing greater interlayer
distances in the phenomena known as permeation and
swelling. As the interlayer distance generally increases with
a greater water content, the external voids become smaller
to balance the expanding internal voids when the water
permeates the montmorillonite under constrained volume
conditions. Since internal voids have much smaller pores
1.0 1.5 2.0 2.5 3.0 than external voids, the hydraulic performance of the ben-
tonite improves under fully hydrated conditions, with
internal voids accounting for an increased portion of the
total. Such an improvement can be interpreted as an eect
of the increased specic surface area due to these factors.
These successive phenomena were observed in studies by
Tomioka et al. (2007) using microfocus X-ray CT, as
shown in Photo 1.
3. Hydrologic model of bentonite
In this study, the Kozeny-Carman relationship was
employed as a hydrologic model for bentonite allowing
for consideration of the water permeation-related changes
in the montmorillonite microstructures. This relationship
is expressed as
1 qg 1 e3
k 1
C l S 2v 1 e
where k is the hydraulic conductivity, C is the shape factor
of the soil particles (usually constant at 5), q is the pore liq-
uid density, g is the gravity-related acceleration, l is the
coecient of viscosity, S v is the surface area per unit vol-
ume (specic surface area) and e is the void ratio. The
specic surface area shown in Table 1 is the mass per unit.
For conversion to the surface area per unit volume, as
required for the Kozeny-Carman relationship, multiplication
Particle form Particle size (lm) Specic surface area (m2/g)
Sheet From 0.1  0.1  0.001 to 1  1  0.001 From 760 to 810
Panel sheet From 1  10.02 to 1  10.1 From 9 to 40
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx 5

Fig. 6. Montmorillonite microstructure.

area observed under compacted conditions with a certain

Fig. 7. Bentonite swelling.
by dry density and other calculations are needed. It should
also be noted that the specic surface areas in the table
were observed as part of a system of soil particles fully
diused in a large amount of water, and that the
Kozeny-Carman relationship requires a specic surface
level of dry density.
The Kozeny-Carman relationship involves the assump-
tion of spherical soil particles and the Hagen-Poiseulle
ow, making it somewhat incompatible with the subject
of the study in some points. However, being based on a
theoretically derived equation, the relationship is advanta-
geous in that physical phenomena are expressed coherently,
as this is a monotonically decreasing function with a range
in dry density of 0 < qd < qs within which the hydraulic
conductivity can be dened. It also comprises microstruc-
tural and macrostructural parameters, as shown by Eq.
(1). More specically, the void ratio (i.e., that of the vol-
ume of all voids to the entire volume of soil particles)
can be viewed as a macrostructural parameter because,
by itself, it expresses neither the neness of the soil particles
nor changes in the soil particle structure. Meanwhile, the
specic surface area varies with the soil particle structures
a characteristic that depends on particle size and the
degree of hydration, making it a microstructural parame-
ter. By way of example, the use of Eq. (1) enables the
expression of dierences in hydraulic conductivity between
kaolinite and montmorillonite at the same void ratios
(Fig. 5) as dierences in the specic surface area (Table 1).
Moreover, with a microstructure formed by water perme-
ation and swelling under constrained volume conditions
(Fig. 7), the improvement in hydraulic performance
(reduced hydraulic conductivity) can be represented as an
increase in the specic surface area.

Before saturation After saturation

Photo 1. Saturation-related changes in microstructure of montmorillonite under microfocus X-ray CT (Tomioka et al., 2007).

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
6 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
Thus, in contrast to the conventional mainstream
approach of experimentally determining the relationships
between dry density and hydraulic conductivity, the use
of the Kozeny-Carman relationship as a hydrologic model
to evaluate the impermeability of engineered bentonite bar-
riers allows for the expression of more events such as
changes in hydraulic conductivity caused by variations in
the microstructure stemming from the chemical-
mechanical-hydraulic interaction. It is also highly probable
that, being based on a theoretically derived equation, the
relationship supports a high prediction capacity. It should
be noted that the use of the Kozeny-Carman relationship
requires the determination of the specic surface area of
compacted swelling cohesive clay permeated with water
under constrained volume conditions.
4. Determining specific surface area of bentonite blocks with
constant volume
Kozaki (2003) and other researchers have conducted
studies to determine the specic surface area of bentonite,
assessing the eects of microstructures on the diusion
behavior of related radioactive nuclides via surface obser-
vation based on SEM and AFM, a particle size distribution
analysis and an internal observation using microfocus
X-ray CT and other methods. Kozaki notably determined
the specic surface area using the BET method with nitro-
gen gas and a surface adsorption approach involving the
use of ethylene glycol monoethyl ether (known as the
EGME method) to calculate the surface area of internal
voids from the dierences in measurement determined from
both approaches (see Table 2). However, both the BET
method and the EGME method require testing with com-
pletely dry powder samples. The surface area data from
such tests do not always fully match the data on the contact
areas between the pore uid and the minerals, as required
for the Kozeny-Carman relationship.
As stated previously, bentonite used for engineered bar-
riers is in the form of compacted swelling cohesive soil. It is
highly likely that its microstructure depends on the type of
liquid permeating its voids and on its dry density. Hence,
the use of the Kozeny-Carman relationship to evaluate
the impermeability of such barriers requires a method for
determining the specic surface area of the bentonite per-
meated with water under constrained volume conditions.
Accordingly, focus in this study was placed on a rational
method for determining the specic surface area.
Table 2
Specic surface area of montmorillonite samples based on BET and
EGME methods (Kozaki, 2003).
Particle size distribution BET method EGME method
(mesh) (m2/g) (m2/g)
100200 45 7.0  102
<330 62 7.0  102
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
4.1. XRD test on bentonite with various levels of water
content
4.1.1. Two-layer hydration water content
Morodome (2008) reported that the number of mole-
cules of adsorbed water depends on the relative humidity
and temperature of the void spaces in a bentonite speci-
men, and that the hydration state can be ascertained via
an XRD observation. In research on montmorillonite
hydration states, the layer concept (Fig. 8) is often used.
This concept was applied to the determination of the speci-
c surface area of montmorillonite, as outlined below.
On the surface of a mineral, the area occupied by one
water molecule is conceived as the projection area of the
water molecule at 10.8 (A2) (Sudo, 1967). Consequently,
on the surface of a mineral, the area occupied per unit mass
of water, S g , is expressed by
2
S g 10:8 A  6:02  1023 =18
3:6  103 m2 =g-water 2
where 6:02  1023 is Avogadros number and 18 g/mole is
the molecular weight of water.
If the monomolecular adsorption of water molecules
occurs without excess or deciency on the surfaces of the
primary particles of montmorillonite, two primary particles
have two layers of water molecules between them, as shown
in Fig. 8. This condition is dened as a two-layer hydration
state, and the water in it is known as the two-layer hydra-
tion water content, w . Using w , the specic surface area
per unit mass, sv, is expressed as
sv 3:6  103  w m2 =g-soil 3
[E1] In this paper, the state at which the N water mole-
cules hydrate between two primary particles is hereinafter
referred to as the N-layer hydration state, while the dis-
tance between two primary particles in the N-layer hydra-
tion state is hereinafter referred to as the N-layer hydration
distance.
The specic surface area required for the Kozeny-
Carman relationship is the surface area per unit volume.
The two sides of Eq. (3) can be multiplied by the dry den-
sity, qd (Mg/m3), to yield the following value:
S v 3:6  103  w  qd
3:6  109 w qd m2 =m3 -soil 4
Eq. (4) enables the calculation of the surface area per unit
volume, Sv, of the montmorillonite compacted to a certain
dry density based on two-layer hydration water content w .
Next, the two-layer hydration water content was exam-
ined. As a rst step, the primary particles of the montmo-
rillonite were placed in a two-layer hydration state in a
constrained volume cell, as shown in Fig. 8. If there are
considerably fewer primary particles in such a state, com-
pared with cell volume V, the dry density of the entire cell
is very low. However, as the particles are in a two-layer
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx 7

Volume: V
Two-hydration state particle:
Total mass of particles : M s
Total mass of Water : M w Two-hydration state water content : w* Mass of
Mass of
Ms
Dry density : d e 1 s
V Water content in saturation state : wsat 1
Mw Gs Gs d
Water content : w Specific gravity : Gs Void ratio : e
Ms
Soil particle density : s
(A) Unsaturation state (B) Unsaturation state

Volume constant cell Add the two-

hydration state
particles

Ms : increase , w w*
d ,w w* d
V
(C) Saturation state (D) Saturation state

d : increase , w wsat w* d : increase , w wsat w*

When number of particles with two-hydration state will increase, bulk dry density will
increase but water content will be constant until saturation.
In order to increase number of particles after saturation, water content has to decrease. In
this case, the two-hydration state can not maintain.

Fig. 8. Two-layer hydration water content.

hydration state, the water in the cell is the two-layer hydra-
tion water content (Fig. 8(A)). Adding the primary parti-
cles in a two-layer hydration state to the cell also merely
increases the dry density, with the water remaining as the
two-layer hydration content until saturation is reached
(Fig. 8(B) and (C)]. If the cell is saturated, in terms of
the water content, it is not possible to add the primary par-
ticles in a two-layer hydration state to the cell. To intro-
duce more primary particles (or to raise the dry density),
the cell must be drained (Fig. 8(D)). Thus, if the water con-
tent in the cell is below saturation, the two-layer hydration
water content remains constant irrespective of the dry den-
sity. After saturation, however, this content depends on the
dry density. At high dry density, under which a two-layer
hydration state cannot occur, it should be noted that the
two-layer hydration water content is estimated to be lower
than the actual level. At such high dry density, the hydrau-
lic conductivity calculated from the Kozeny-Carman rela-
tionship will monotonically decrease with the increasing
dry density due to the substantially reduced void ratio,
which is a macroscopic parameter.
To apply the above arguments to bentonite, it was
assumed in this study that water adsorbed onto the sur-
faces of the non-swelling minerals in the material could
be ignored because the surface area of such minerals is
much smaller than that of montmorillonite (Table 1).
Based on this assumption, the two-layer hydration water
content of bentonite with the desired montmorillonite con-
tent can be ascertained by determining the two-layer hydra-
tion water content of the montmorillonite alone.
4.1.2. Determination of two-layer hydration water content
This study involved the use of X-ray diraction (XRD)
to clarify the state of two-layer hydration. XRD is usually
applied to determine the interatomic distances in crystals
and other materials with regularly aligned atoms. Particu-
lar focus was placed on the spacing of the montmorillonite
primary particles, which is uniform after permeation and
swelling. The spacing of the multiple particles of this type
was determined in line with the peak positions in the
XRD patterns (Fig. 9) to enable the calculation of the
number of water molecules between the primary particles

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
8 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
Fig. 9. XRD cell for evaluation of two-layer hydration state of montmorillonite based on measurement of interlayer distance.
based on the distances determined. The XRD method was
used to identify the two-layer hydration state in this study
based on the in situ approach of Kawamura et al. (1999),
which enables observation of the changes between the pri-
mary particles of montmorillonite during the processes of
water penetration and saturation. In this technique, the
montmorillonite is compacted to a prescribed dry density
and placed in an XRD cell that allows water penetration
at a constrained volume, which was ideal for the purposes
of this study. The XRD equipment used here was the ordi-
nary h-h type (Rint Ultima III, Rigaku Co., Ltd.; MoKa,
k = 0.7093 A) with a goniometer radius of 285 mm. The
XRD cell shown in Fig. 9 was set at the center of the
goniometer. The tube current was 30 mA, the voltage was
50 kV and the scanning range was 1.05.0 2h. The surface
of the specimen was irradiated with X-rays at steps of 0.05
and a scan rate of 0.1 per minute.
This study involved the determination of the two-layer
hydration water content of Na-montmorillonite, Ca-
montmorillonite, Na-bentonite, Ca-bentonite and engi-
neered Na-bentonite. The Na-bentonite was a natural
variety produced in Yamagata Prefecture. The
Na-montmorillonite was exclusively extracted from the
Na-bentonite (Kudo and Kawanishi, 2005), and the Ca-
montmorillonite and Ca-bentonite were produced by
rening the Na-montmorillonite and the Na-bentonite via
a forced cation exchange reaction using a CaCl2 aqueous
solution (Sasakura et al., 2004). The engineered bentonite
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
was produced by blending the Na-montmorillonite and
nely crushed silica sand to achieve the same level of mont-
morillonite content as natural Na-bentonite.
(1) The two-layer hydration water content w of the
materials was determined as follows:
(2) Each bentonite sample was cooled with ice to approx-
imately 20 C.
(3) The ice was then crushed to create a ne powder.
(4) The powdered ice and cooled bentonite were mixed to
achieve the prescribed water content.
(5) The resulting mixture was placed in an XRD cell
(Fig. 9) for compaction to the prescribed dry density
(Fig. 8), and the cell was rmly lidded. Steps 14 were
carried out in a 20 C environment to prevent the
powdered ice from melting.
(6) The water content preparation was considered com-
plete when the mixture of the powdered ice and the
cooled bentonite sealed in the XRD cell was thawed
at room temperature. The above steps were followed
to create bentonite samples with dierent water con-
tents in increments of 1%. As shown in Fig. 9, the
XRD cell consisted of a 1.5-mm-thick polycarbonate
pipe and a stainless steel sheath (with windows to
allow for X-ray passage) to constrain the volume.
During the XRD process, X-rays entered a window
and passed through the polycarbonate pipe to the
water content-controlled sample. After diraction,
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx 9

X-rays passing through the other window were eval-

uated. Molybdenum (MoKa) was used as the radia-
tion source for the XRD lamp.
(7) XRD was conducted on each sample to establish the
distance between the primary particles for each level
of water content, and the diraction patterns were
observed to identify the peaks. The state of hydration
was determined by identifying the highest diraction
intensity among the one- to four-layer hydration
states. By way of example, for the XRD performed Na-montmorillonite
on the lower-water-content samples followed by the
higher-water-content samples, the diraction patterns
and identied peaks revealed that the hydration state
of the highest diraction intensity shifted from a one-
layer hydration state to a four-layer hydration state.
For this reason, the water content immediately pre-
ceding that at which the three-layer hydration state
exhibited the highest intensity was dened as the
two-layer hydration water content, w .
Ca-montmorillonite

4.2. XRD test results for bentonite with various levels of

water content

Fig. 10 shows the XRD patterns from all the test cases.
The red vertical lines represent the diraction angles corre-
sponding, respectively, to the zero interlayer distances and
the one- to four-water-molecule interlayer distances.
Accordingly, the position of the central red line represents
the two-layer hydration state [E2]. The values in the paren- Na-bentonite
theses in the legend of Fig. 10 represent the water content
of the montmorillonite portion of bentonite, dened by
the mass of the pore water relative to the dry mass of the
material.
The graphs reveal that, in all the test cases, diraction
peaks shifted towards the left with the increasing sample
water content. This implies that at higher water contents,
increasing numbers of interlayer water molecules con-
tributed to increased interlayer distances. Consequently, Man-made Na-bentonite
the water content at which the diraction peak was posi-
tioned at 2.68 (as determined via XRD using the water
content as a parameter) was equal to the two-layer hydra-
tion water content. However, the baseline needs to be
removed because the diraction patterns shown in Fig. 10
were aected by the polycarbonate pipe. The numbers of
water molecules between the primary particles also show
some variation. Accordingly, the two-layer hydration state
was identied via a peak separation process.
Fig. 11 shows examples of the baseline removal and the
peak separation for the Na-montmorillonite and the Ca- Ca-bentonite
bentonite. At a water content of 19%, as shown in the
Fig. 10. XRD patterns for all materials tested.
upper-left gure, the two-layer hydration state exhibited
high diraction intensity at 2.68. However, as the water
content reached 20%, as shown in the upper-right gure, XRD was conducted following these steps with water
the diraction intensity of the two-layer hydration state content increments of 1%. The two-layer hydration water
diminished sharply and that of the three-layer hydration content was dened as the content observed immediately
state (at 2.26) increased toward similar levels. before the sharp decrease in diraction intensity for the

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
10 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx

Na montmorillonite
3
d
=1.6Mg/m , w=20%
XRD pattern
Peak separation results
Peak separation results
Peak separation results
Resynthesised XRD pattern

Ca bentonite
Ca bentonite
d
=1.6Mg/m3, w=10%
d
=1.6Mg/m3, w=9% XRD pattern
XRD pattern Peak separation results
Peak separation results Peak separation results
Peak separation results Peak separation results
Peak separation results Resynthesised XRD pattern
Resynthesised XRD pattern

Fig. 11. Peak separation for XRD patterns.

two-layer hydration state. Based on this denition, the two-
layer hydration water content for the Na-montmorillonite
was 19%. Similarly, for the Ca-bentonite (shown in the lower
gures), the two-layer hydration water content was 9%.
The test results also revealed that the water content of
the montmorillonite portion of bentonite in a two-layer
hydration state was roughly constant for the same types
of montmorillonite.
Montmorillonite water content wmnt is expressed as
Mw Mw w
wmnt
M smnt vmnt M s vmnt
where M w is the mass of the pore water, M smnt is the mont-
morillonite dry mass, M s is the total mass of soil particles,
w M w =M s is the bentonite water content and vmnt is the
montmorillonite content (0.59 in this study). As Eq. (5)
holds true for the two-layer hydration water content, the
two-layer hydration water content w of bentonite is
expressed as wmnt vmnt w based on the montmorillonite
content.
More specically, the two-layer hydration water content
of the bentonite samples, with the same type of montmoril-
lonite at dierent contents, can be determined by nding
the two-layer hydration water content of the montmoril-
lonite itself and multiplying it by a factor representing
the montmorillonite content.
These outcomes indicate that most of the water added to
the bentonite was adsorbed by the montmorillonite, with
the water-retention capacity of the accessory minerals
being minimal.
4.3. Relationship between dry density and two-layer
hydration water content
Table 3 summarizes the test results. The natural Na-
bentonite and engineered bentonite exhibited similar levels
of two-layer hydration water content. Consequently, it can
be considered that the accessory minerals had very little
5 eect on the content in these tests. This, in turn, implies
a minimal eect from the accessory minerals on the
hydraulic performance of the engineered bentonite barriers
when these minerals are not electrolytes. These results were
used to determine the relationship between the dry density
and the two-layer hydration water content.
Fig. 8 suggests that the two-layer hydration water con-
tent, below saturated water content wsat , was independent
of the dry density. Hence, the following equation must hold
true:
E3 w const: when w < wsat 6
Meanwhile, at high dry densities, with low pore volume
and insucient space for a two-layer hydration state, the
saturated water content fell below the two-layer hydration
water content. In this case, the two-layer hydration water
content was regarded as being equal to the saturated water
content.

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx 11

Table 3
Relationship between two-layer hydration water content and eective bentonite dry density for dierent materials.
Material w Relationship between eective bentonite dry density and two-layer hydration water content

w 0:01vmnt =0:052 3:8  106 vmnt qd mnt

12:1v
Na-montmorillonite 0.19
w 0:01vmnt =0:066 2:6  107 vmnt qd
15:16vmnt
Ca-montmorillonite 0.15
w 0:01=0:091 2:2  10 qd  0:01vmnt =0:052 3:8  106 vmnt qd mnt
8 17:57 12:1v
Na-bentonite 0.11
w 0:01=0:091 2:2  108 q17:57 6 12:1vmnt
Man-made Na-bentonite 0.11 d  0:01v mnt =0:052 3:8  10 v q
mnt d
11 24:5 7
w
15:16vmnt
Ca-bentonite 0.09 0:01=0:11 3:9  10 qd  0:01vmnt =0:066 2:6  10 vmnt qd

Previous research (Sato, 2008) has indicated that this
condition occurs when the dry density of bentonite is
1.7 Mg/m3 or more. In such a high dry density range, direct
contact among the soil particles is highly likely. Conse-
quently, once a certain dry density is reached, the specic
surface area appears to decrease with the increasing dry
density. Hence, it was assumed that the following condition
must be fullled:
 
 def e 1 qs
E4 w wsat  1 when w > wsat 7
Gs Gs qd
In this study, the relationship between dry density qd
and two-layer hydration water content w , fullling the
above condition, was dened as
0:01vmnt
w b vmnt
8 inition, CR is 0 for natural bentonite 1 when all the Na ions
cmnt amnt vmnt qdmnt
where amnt , bmnt and cmnt are tting parameters, as shown in
Table 3, in which the bentonite-heading elds show rela-
tionships between the dry density and the two-layer hydra-
tion water content, as determined by multiplying the test
results for the montmorillonite itself by the montmoril-
lonite content. Table 3 also highlights the relationship
between the dry density and the two-layer hydration water
content, as determined from the test results.
The relationship between the dry density and the two-
layer hydration water content depends on the types of
interlayer cation, as shown in Table 3. An indicator known
as the Ca-bentonite development rate (CR) can be used to
express the eects of the interlayer cations on the hydraulic
performance (Sasakura et al., 2004). The percentage R of
Ca ion equivalence Ca2  to the sum of Na ion equivalence
Na  (meq/100 g) and Ca2  is determined as
Ca2 
R 9
Ca2  Na 
Given this, Ca-bentonite development rate CR is
expressed by
R  R0
CR 10
R100  R0
where subscript 0 is used to indicate R under natural con-
ditions and 100 indicates R after the complete displacement
of the interlayer Na ions by the Ca ions. Based on this def-
have been displaced by the Ca ions. In the test cases of this
study, CR was either 0 or 1. Accordingly, linear approxima-
tion is needed to set amnt , bmnt and cmnt as a function of CR.
A similar analysis using CR as a parameter allows for the
development of a detailed model to determine the eects
of CR on amnt , bmnt and cmnt .
4.4. Kozeny-Carman law with specific surface area
calculated from two-layer hydration water content
Fig. 12 shows the relationships linking dry density qd
and two-layer hydration water content w (left), dry density
and the surface area per unit mass sv (center), and dry

Fig. 12. Relationship between eective bentonite dry density and two-layer hydration water content/specic surface area (Na-montmorillonite).

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
12 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
density and surface area per unit volume S v (right) for
Na-montmorillonite. The graphs reveal a constant surface
area per unit mass up to saturation water content and a
subsequent decrease with a reduced saturation water
content to fulll the condition of Eq. (7). Meanwhile, the
surface area per unit mass (center) exhibits values slightly
lower than those for montmorillonite in Table 1, which
shows the specic surface area of montmorillonite in a dif-
fused system, while the graph in the center shows corre-
sponding values obtained in a compressed system.
The specic surface area per unit volume required for
the Kozeny-Carman relationship reaches its maximum at
a dry density of around 1.8 Mg/m3, as shown in the graph
Fig. 13. Comparison of relationships between dry density and hydraulic
conductivity from various studies.
sv 36w* (m2 / g )
Sv 36w* d (m2 / m3 )
w* 1/ 0.052 3.8E 6
Fig. 14. Hydraulic conductivity calculation.
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
on the right. This dry density can be calculated as a specic
value (1.813 Mg/m3) at which the Na-montmorillonite
two-layer hydration water content of 19% at a soil particle
density of 2.766 Mg/m3 is equal to the saturated water
content.
Fig. 13 shows the relationships linking the dry density
and the hydraulic conductivity for all materials based on
the steps shown in Fig. 14, along with the relationships
linking dry density and hydraulic conductivity proposed
in previous studies (Pusch, 1994; Ichikawa et al., 2002)
and in the experimental results for Na-bentonite from pre-
vious studies (JAEA Database; Sasakura et al., 2004).
Compared with existing studies, the results from the study
reported here for the Na-bentonite conform relatively well
to the Kozeny-Carman relationship, as shown in Fig. 13.
However, it should be noted that the results based on this
relationship were slightly higher than the experimental
values.
Fig. 13 also shows the results for Na-montmorillonite,
Ca-montmorillonite and Ca-bentonite, indicating that dif-
ferent hydraulic conductivities resulting from dierences
in the interlayer cations and the montmorillonite content
(even at the same dry density) were successfully
represented.
4.5. Separation of internal and external voids using XRD
The relationship between dry density and hydraulic
conductivity, as calculated in accordance with the
Kozeny-Carman relationship using the specic surface area
determined from two-layer hydration water contents,
d
12.1
Kozeny-Carman law
1 g 1 e3
k k ( , , e, Sv )
5 Sv 2 1 e
Sv: Specific Surface Area (m2/m3)
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx 13

corresponds more closely to the experimental results than eext wsat w* Gs

those presented in previous studies, as shown in Fig. 13. s
However, the relationship values are slightly higher than e 1
d
the experimental values. In the next step of the study, inter-
nal and external voids were analyzed to determine their e
respective hydraulic conductivities. eext

With Na-montmorillonite, permeating water is
adsorbed into the interlayer spaces and the primary parti-
cles eventually diuse into the water. The hydration prop-
erty of Ca-montmorillonite is lower than that of Na-
montmorillonite. Water permeating Ca-montmorillonite
results in a three- to four-layer hydration state at best,
and the permeation of more water eventually leads to mate-
rial disintegration (e.g., Fukushima 1984; Morodome,
2008). Consequently, the proportions of internal to exter-
nal voids depend on the type of interlayer cation and rep-
resent the parameters for the determination of
microstructures. As such, it is important to individually
examine these proportions in a hydrologic performance
evaluation. To separate the internal and external voids,
the internal voids were dened in this study as those exist-
ing up to a two-layer hydration state, while all other voids
were considered external voids.
Based on this denition, internal void ratio eint is
expressed by the following equation with two-layer hydra-
tion water content w :
M w qw V int eint
w
Ms qs V s Gs
eint Gs w
Here, Gs is the specic gravity of the soil particles, qw and
qd are the water density and the dry density, and V int and
V s are the volumes of internal voids and soil particles,
respectively.
External void ratio eext is expressed by the following
equation with total void ratio e and internal void ratio eint :
wsat Gs
eext e  eint  w Gs wsat  w Gs
1
Fig. 15 shows the relationships of dry density to the
total, internal and external void ratios. The external void
e
eint
s
*
d
w*Gs 1
*
eint w Gs
Fig. 15. Separation of external and internal voids in Na-montmorillonite.
state. Accordingly, a denition of internal voids that more
accurately represents the experimental results should be
established in practice. The above denition was established
in this study because it was considered important to present a
procedure for separating the internal voids from the external
voids. Incidentally, in order to use a dierent hydration state
for the denition of internal voids, two-layer hydration
water content w can be changed to the water content of a
hydration state conforming to the new denition.
4.6. Hydrologic model incorporating consideration of
11 internal and external voids
The internal and external void ratios can be quantita-
tively separated by dening the internal voids as being in
a state up to two-layer hydration, as discussed above.
Using the separated ratios, the hydraulic conductivity
was calculated in accordance with the Kozeny-Carman
relationship. Firstly, the hydraulic conductivities of inter-
12 nal voids k int and external voids k ext were dened as
1 qg 1 e3int
k int 14
C l S vint 1 eint
2

ratio is below the internal void ratio when the dry density 1 qg 1 e3ext
is around 1.3 Mg/m3 or above, and reaches zero at a dry k ext 15
C l S vext 1 eext
2
density of 1.813 Mg/m3. This is roughly consistent with
the outcome of a previous research (Sato, 2008). Dry den- where S vint and S vext are the surface area of the internal and
sity qd , at which the external voids disappear, is external voids per unit volume, respectively. However, as it
q is not possible to separate S vint from S vext logically, propor-
qd  s 13 tions of internal and external void ratios to the total void
w Gs 1
ratio were used to make the following assumption:
When dry density qd is exceeded, the total void ratio is e 23 e 23
int ext
equal to the internal void ratio. S vint S v ; S vext S v 16
Internal voids were dened as those in a state of up to two- e e
layer hydration because the water between the primary par- Next, it was assumed that water permeates the internal
ticles is considered free rather than adsorbed in a hydration and external voids in parallel in montmorillonite in an
state with three or more layers. However, it is highly likely engineered bentonite barrier, with the same hydraulic gra-
that the water between the primary particles is in fact inu- dient acting on both types of voids. Then, the total hydrau-
enced by the adsorbed water even in a three-layer hydration lic conductivity was expressed as follows (parallel model):

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
14 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx

Fig. 16. Evaluation of hydraulic conductivity considering internal and external voids.

k k int k ext 17 Eq. (1), which expresses the ordinary Kozeny-Carman

In contrast, when it was assumed that water permeates
the internal and external voids in a series (at the same ow
rate for both voids), the following equation expressed the
total hydraulic conductivity (series model):
1 1 1 determinant of the total hydraulic conductivity. This sug-
18
k k int k ext
With the parallel model, the greater of the two hydraulic
conductivities is a determinant of the total hydraulic con-
ductivity; with the series model, the lower hydraulic con-
ductivity is a determinant.
Fig. 16 shows the relationships between dry density and
hydraulic conductivity for Na-bentonite as calculated in
accordance with the Kozeny-Carman relationship and rep-
resented by Eqs. (1), (14), (15), (17) and (18). It also shows
the experimental results for the previous studies displayed
in Fig. 13.
The hydraulic conductivity of the internal voids, repre-
sented by Eq. (14), and the hydraulic conductivity of the
series model, represented by Eq. (18), are more consistent
with the experimental results than those derived from
relationship. The hydraulic conductivity of the external
voids falls below that of the internal voids at a certain level
of dry density. At this level, the external voids are no longer
present (or are indistinguishable from the internal voids).
Consequently, the internal voids are the predominant
gests that the engineered bentonite barriers have discrete
(rather than interconnected) external voids at a certain
level of dry density. These results indicate that the internal
voids may be the dominant determinant of the total
hydraulic conductivity and that the external voids are not
interconnected within the range in dry density required of
engineered bentonite barriers.
5. Conclusion
In this study, the Kozeny-Carman relationship was used
to evaluate the hydraulic performance of engineered ben-
tonite barriers, and a method was developed for measuring
the specic surface area required for this relationship under
constrained volume conditions with water permeation. A

Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
 I. Kobayashi et al. / Soils and Foundations xxx (2017) xxxxxx
technique for separating the internal and external voids,
based on a microstructure analysis for specic surface area
determination, was also proposed.
As the studys results and related discussion were based
on various assumptions, the quantitative argument of the
outcomes requires careful consideration. Nevertheless, the
application of the Kozeny-Carman relationship, in consid-
eration of the microstructures, appears to be important for
evaluating the hydraulic performance of engineered ben-
tonite barriers. The studys results can be summarized as
follows:
(1) A method was developed for determining the specic
surface area of various types of montmorillonite and
bentonite based on an XRD analysis of the two-layer
hydration states. The relationship between the speci-
c surface area and the dry density was determined.
(2) Based on the specic surface area, as calculated from
the two-layer hydration water content, the link
between the dry density and the hydraulic conductiv-
ity was determined in accordance with the Kozeny-
Carman relationship. The results were more consis-
tent with the experimentally ascertained data than
with the data of previous studies.
(3) A method was proposed for separating the internal
and external voids based on the assumption that the
former are present in a state of up to two-layer hydra-
tion, and the hydraulic conductivities of both were
calculated in accordance with the Kozeny-Carman
relationship. The results suggest that, within the dry
density range required of engineered bentonite barri-
ers, the internal voids are the dominant determinant
of the total hydraulic conductivity and the external
voids are not interconnected.
Acknowledgement
This research was part of a study entitled Development
of a Technique for the Evaluation of Long-term EBS Perfor-
mance (FY 2007 2012) conducted with a grant from the
Agency for Natural Resources and Energy in Japans Min-
istry of Economy, Trade and Industry.
References
Carman, P.C., 1956. Flow of Gases through Porous Media. Academic
Press, New York, 182pp.
Clay Science Society of Japan (CSSJ), 1987. Clay Handbook, second ed.
Creager, W.P., Justin, J.D., Hinds, J., 1945. Engineering for Dams. John
Wiley, New York, London.
Federation of Electric Power Companies and Japan Nuclear Fuel Cycle
Development Institute, 2005. Second Progress Report on Research and
Development for TRU Waste Disposal in Japan -Science and
Please cite this article in press as: Kobayashi, I. et al., Evaluation of specic surface area of bentonite-engineered barriers for Kozeny-Carman law, Soils
Found. (2017), https://doi.org/10.1016/j.sandf.2017.08.001
15
Technology, Safety Assessment and Means of Implementation-, JNC
TY1400 2005-013, FEPC TRU-TR2-2005-02.
Fukushima, Y., 1984. X-ray diraction study of aqueous montmorillonite
emulsions. Clays Clay Miner. 32 (4), 320326.
Hazen, A., 1892. Some Physical Properties of Sands and Gravels Annual
Report. State Board of Health, Mass., pp. 539556.
Ichikawa, Y., Kawamura, K., Fujii, N., Theramast, N., 2002. Molecular
dynamics and multiscale homogenization analysis of seepage/diusion
problem in bentonite clay. Int. J. Numer. Meth. Eng. 54, 17171749.
Ito, M., Okamoto, Y., Suzuki, K., Shibata, M., Sasaki, Y., 1994. Mineral
composition analysis of bentonite. J. Atomic Energy Soc. Jpn.,
ATOMOS 36 (11), 6366.
JAEA, 2000. H12 report Supporting Report 2 Repository Design and
Engineering Technology, JNC TN1400 99-022.
JAEA, Buer Material Database <http://buerdb.jaea.go.jp/bmdb/>.
Kawamura, K., Ichikawa, Y., Nakano, M., Kitayama, K., Kawamura,
H., 1999. Swelling properties of smectite up to 90 C: in situ x-ray
diraction experimental and molecular dynamics simulation. Eng.
Geol. 54, 7579.
Kikuchi, H., Tanai, H., Matsumoto, K., Sato, H., Ueno, K., Tetsu, T.,
2003 Hydraulic characteristics of Buer Material-II -The inuence
which saline water exert on hydraulic properties of bentonite buer
material. Testing Document, JNC TN8430 2003-002.
Kobayashi, I., Owada, H., Ishii, T., 2011. Hydraulic/mechanical modeling
of smectitic materials for HMC analytical evaluation of the long term
performance of TRU geological repository. In: Proceedings of the 14th
International Conference on Environmental Remediation and
Radioactive Waste Management, ICEM2011.
Kozaki, T., 2003. Eects of microstructure of clay on diusion behavior of
radionuclides in buer materials (III), JNC TJ8400 2002-053.
Morodome, S., 2008. On swelling behavior of montmorillonite with
various exchangeable cations and structure of montmorillonitewater
systems, -In situ observation and numerical simulation of X-ray
diraction and small angle X-ray scattering method, Doctoral theses of
Tokyo Institute of Technology.
Kudo, K., Kawanishi, H., 2005 Study on estimation method of montmo-
rillonite content in bentonite. In: Proceedings of Annual Conference of
Japan Society of Civil Engineering, pp. 3515 (in Japanese).
Pusch, R., 1994. Waste Disposal in Rock. Elsevier, p. 490.
Pusch, R., Yong, N.R., 2006. Microstructure of Smectitic Clays and
Engineering Performance. Taylor and Francis.
Sato, H., 2008. Thermodynamic model on swelling of bentonite buer and
backll materials. Phys. Chem. Earth 33, S538S543.
Sasakura, T., Kobayashi, I., Sahara, F., Murakami, T., Ohi, T., Mihara,
M., Itoh, H., 2004. Studies on the mechanical behavior of bentonite for
development of an elasto-plastic constitutive model. In: DisTec 2004,
International Conference on Radioactive Waste Disposal, pp. 498
507.
Shibutani, T., Kohara, Y., Oda, C., Kuno, Y., Shibata, M., 1999
Physico/chemical characteristics of puried Na-smectite and protona-
tion/deprotonation behavior of smectite surface in NaCl media. JNC
TN8400 9-066.
Sudo, S., 1967. Specic surface area of soil and measurement by BET
method. J. Jpn. Soc. Soil Phys., JSSP 16, 3942.
Terzaghi, K., Peck, R.B., 1967. Soil Mechanics in Engineering Practice.
John Wiley & Sons.
Tomioka, S., Kozaki, T., Takamatsu, H., Noda, N., Nishiyama, S.,
Kozaki, N., Suzuki, S., Sato, S., 2007. Analysis of microstructural
images of dry and water saturated compacted bentonite samples
observed with X-ray micro CT. In: Proceedings of Workshop on Long-
term Performance of Smectitic Clays Embedding Canisters with
Highly Radioactive Waste, pp. 179191.