Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Applied Clay Science 26 (2004) 249 – 257

www.elsevier.com/locate/clay

Modelling bentonite–water interactions at high solid/liquid ratios:


swelling and diffuse double layer effects
P. Wersin a,*, E. Curti b, C.A.J. Appelo c
a
National Cooperative for the Disposal of Radioactive Waste (Nagra), Hardstrasse 73, Wettingen, CH-5430, Switzerland
b
Paul Scherrer Institut (PSI), Waste Management Laboratory, CH-5232 Villigen, Switzerland
c
Hydrochemical Consultant, Amsterdam, The Netherlands
Received 15 March 2003; received in revised form 19 September 2003; accepted 23 September 2003
Available online 21 February 2004

Abstract

Previous experimental studies on bentonite – water interactions have demonstrated the importance of ion exchange and
surface complexation reactions occurring at the clay surface as well as dissolution/precipitation reactions of mineral impurities.
Based on these findings, thermodynamic models have been widely used to derive porewater compositions for the compacted
bentonite used as backfill material for nuclear repositories. Conventional models typically neglect phenomena important in
compacted clays such as anion exclusion induced by swelling of the expandable clay fraction and by the formation of electrical
double layers on charged edge surfaces. In this study, we evaluate such phenomena by applying a refined diffuse double layer
(DDL) approach to model porewater composition in a compacted bentonite backfill surrounded by argillaceous host rock, as
foreseen for the Swiss high-level waste repository. Model calculations also include the effect of water incorporation in the
structural interlayers.
The results indicate that the conventional model and the refined DDL model without distinction between interlayer and
external water only differ slightly. The main buffering reactions include ion exchange of Ca for Na, calcite and gypsum
dissolution and deprotonation of surface hydroxyl groups. On the other hand, the calculation accounting for the distinction of
external and interlayer water indicates significant anion exclusion effects on the external water composition. Most notably, this
leads to an increased salinity and drop in pH.
From a performance assessment perspective, however, the differences induced by the inclusion of swelling and diffuse
double layer effects are not very significant relative to uncertainties related to system variables, such as the pCO2 of the host
rock. Finally, it should be emphasised that significant uncertainties related to the thermodynamic properties of water in
compacted clays, e.g., dielectric constant, are still unresolved and deserve further investigations.
D 2004 Elsevier B.V. All rights reserved.

Keywords: Bentonite; Thermodynamic modelling; Swelling

1. Introduction

* Corresponding author. Tel.: +41-56-437-12-80; fax: +41-56-


Bentonite clay is envisaged as backfill material for
437-13-17. nuclear waste repositories because of its favourable
E-mail address: paul.wersin@nagra.ch (P. Wersin). physical and chemical properties. The migration of

0169-1317/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.clay.2003.12.010
250 P. Wersin et al. / Applied Clay Science 26 (2004) 249–257

2001), in order to derive the porewater chemistry. A


precise knowledge of porewater composition is nec-
essary to predict solubility and sorption parameters for
radionuclides.
The conventional approach used to derive the
bentonite porewater chemistry is not satisfying be-
cause it does not take into account specific effects
induced by swelling at constant volume and high S/L
ratios. Through the swelling process, a large fraction
of the water comes to reside in the interlayer, from
which anionic species are excluded by virtue of its
permanent negative charge. Moreover, because of the
Fig. 1. Sketch showing the bentonite – water system as modelled in
this report. The main chemical reactions are: (1) protonation/
very small distances between the negatively charged
deprotonation of surface hydroxyls, (2) cation exchange, (3) edge surfaces, a significant fraction of the external
precipitation/dissolution of minor solids, and (4) gas exchange. water will be strongly influenced by the diffuse
double layer (DDL, see Fig. 2) where ion exclusion
most contaminants is significantly retarded via sorp- occurs as well. Depending on the sign of the pH-
tion on the reactive clay surfaces and colloids are dependent charge, either cations or anions will be
efficiently filtered. In addition, the capability of ben- partially excluded from the DDL. The remaining
tonite to swell upon water addition makes this mate- fraction of water (uncharged external solution) is
rial a very efficient diffusion barrier to contaminant accessible to all negatively and positively charged
migration. This latter property, on the other hand, species. Obviously, ion exclusion in interlayer and
makes it difficult to study the aqueous phase experi- DDL will affect the composition of such water. In this
mentally under compacted conditions, i.e., at the high study, we evaluate the effects of both ion exclusion
solid to water (S/L) ratios relevant to repositories. processes on the chemistry of the uncharged external
The thermodynamic description of the bentonite – solution by means of a refined DDL model for
water interactions has largely been based on experi- bentonite – water interaction at high S/L ratios. The
mental studies performed at low S/L ratios. In these model is applied to the compacted bentonite backfill
studies, the importance of the ion exchange and acid – surrounded by argillaceous host rock, as foreseen for
base properties of the montmorillonite surface could
be shown (Wanner et al., 1994; Bradbury and Baey-
ens, 1997; Kraepiel et al., 1998). Two different types
of surface processes describe the clay surface chem-
istry, namely ion exchange occurring on the perma-
nently charged basal surfaces of expandable clay
minerals (interlayer) and surface complexation occur-
ring at so-called edge sites, where charge depends on
pH. Besides the clay surface, mineral impurities in the
bentonite, such as calcite, gypsum and NaCl, signif-
icantly affect water composition. As illustrated in Fig.
1, the thermodynamic modelling of bentonite –water
interactions is commonly based on ion exchange and
surface complexation reactions occurring at the clay –
water interface and dissolution/precipitation of ben-
tonite impurities. This approach has also been widely
applied to compacted bentonite systems relevant to
repository conditions, i.e., at high S/L ratios (Wieland
et al., 1994; Bruno et al., 1999; Fernández et al., Fig. 2. Water types in bentonite.
P. Wersin et al. / Applied Clay Science 26 (2004) 249–257 251

the Swiss high-level waste repository. The modelling surface potential depends on the protonation/depro-
results are compared with those obtained from the tonation reactions on the external surface of mont-
conventional model (Curti and Wersin, 2002). morillonite, and the latter depends on the chemistry
of the bulk solution, there is no trivial solution to
this problem. The compositions of the two waters
2. Model development can only be obtained by solving simultaneously all
relevant chemical reactions and an appropriate model
2.1. Aims of model calculations relating surface potential to spatial charge distribu-
tion (see Section 2.3).
As indicated above, conventionally no distinction In case (b), we account—in addition to the effects
is usually made between the three different types of of the double layer—for the incorporation of water in
water illustrated in Fig. 2 when modelling of bentonite the structural interlayer of montmorillonite. This water
porewater. All the water admixed to the clay is is considered to be accessible only to exchangeable
regarded as accessible to all solutes, implying that cations (Na+, K+, H+, Mg2 +, Ca2 +) but not to anions
anion exclusion effects due to swelling of the mont- and other cationic or neutral species. Therefore, one
morillonite interlayer and formation of a diffuse expects higher solute concentrations due to reduction
double layer are disregarded. of the external porosity (i.e., of the amount of water
Here, we develop an alternative model to show accessible to mobile solutes).
how the composition of the external porewater
changes as the mentioned processes are ‘‘switched 2.2. Generic model
on’’ sequentially. We will therefore compare calcula-
tions based on the conventional method with equiva- Our generic model computes chemical reactions to
lent calculations in which we assume: achieve thermodynamic equilibrium between reactive
solids in the bentonite backfill and the groundwater
(a) the formation of a diffuse double layer, with from the surrounding host rock. The model is applied
consequent redistribution of charged ions between to the Swiss high-level waste repository concept in
double layer water and the remaining aqueous which compacted Na-rich bentonite is surrounded by
solution; a consolidated shale, the Opalinus Clay. The chemical
(b) incorporation of water in the montmorillonite processes considered in the model include gas ex-
interlayer through swelling under constant volume change (CO2), complexation reactions in solution,
conditions. cation exchange for Na+, K+, H+, Mg2 + and Ca2 +,
protonation/deprotonation of surface hydroxo groups,
In case (a), all the groundwater admixed to the precipitation/dissolution of reactive minerals (calcite,
clay reacts with montmorillonite and other minor gypsum, quartz, kaolinite) and dissolution of residual
solids in bentonite, as in the conventional model. soluble salts (NaCl). Fig. 1 shows schematically the
However, as a result of the electrostatic field exerted modelled bentonite –water system with the main reac-
by the pH-dependent charge of external clay surfa- tions considered.
ces, charged species are distributed between water MX-80 Na-bentonite from Wyoming, USA (sup-
bound to the double layer and the remaining ‘‘free plied by Bentonite International, D-41 Duisburg-Mei-
water’’, where the mobile solutes reside. In other derich) was selected as a reference material for our
words, two solution compositions can be differenti- model calculations. This material has been thoroughly
ated. Such effects can be usually neglected in non- characterised in preceding studies, so that mineralog-
compacted clay – water mixtures because the ical composition, cation exchange properties and
proportion of DDL water is very small. In highly surface site densities are well known. We relied on
compacted clays, as in the case of high-level waste recent characterisation work by Baeyens and Brad-
repositories, the fraction of DDL water increases to bury (1997) and Bradbury and Baeyens (2002), who
the extent that significant effects on the composition largely confirm the previous findings of Müller-Von-
of the ‘‘free’’ water are expected. Because the moos and Kahr (1983).
252 P. Wersin et al. / Applied Clay Science 26 (2004) 249–257

In the model, we assume that the bentonite reacts Table 1


with reference Opalinus Clay water, as defined by Overview of model parameters, properties of MX-80 bentonite and
Opalinus Clay water composition used for the bentonite porewater
Pearson (2002). The relevant material properties of calculations
MX-80 bentonite and the composition of the refer-
Value Reference
ence Opalinus Clay water are listed in Table 1 along
System parameters
with other model parameters, reflecting the condi-
S/L ratio (kg l 1) 4.5 [1]
tions expected in the planned Swiss high-level waste Temperature (jC) 25 [1]
repository. log pCO2 (bar)  2.2 [2]
In a previous study (Curti and Wersin, 2002;
Wersin, 2003), this generic model was tested by Surface complexation
BET surface area (m2 kg 1) 31,500 [3]
applying two different modelling approaches, the
Surface site concentrations 0.0284 [4]
one-site diffuse double layer model of Wieland et al. (eq kg 1)
(1994) and the three-site nonelectrostatic model of Surface complexation constants, log Kint:
Bradbury and Baeyens (1997) to simulate recent QSOH + H+ = QSOH+2 + 5.4 [4]
bentonite –water interaction experiments of Muurinen QSOH = QSO + H+  6.7 [4]
and Lehikoinen (1999) at variable S/L ratios. An
Ion exchange
important outcome of that ‘‘conventional’’ modelling Cation exchange capacity 0.787 [5]
study was that differences between results of the two (CEC; eq kg 1)
models were small in spite of their significantly Initial occupancies (equivalent fraction):
different underlying concepts. In this work, we base Na+ 0.848, Ca2 + 0.084, Mg2 + 0.051, K+ 0.017
Selectivity coefficients, Kc:
our refined model on the diffuse double layer model
2NaX + Ca2 + = CaX2 + 2 Na+ 2.6
of Wieland et al. (1994). NaX + K+ = KX + Na+ 4.0
2NaX + Mg2 + = MgX2 + 2 Na+ 2.2
2.3. DDL effects
Saturated solids, log Ks0
Quartz SiO2(s) + 2H2O = H4SiO4  3.98 [6]
In the first step of our chemical model refinement,
Calcite CaCO3(s) = Ca2 + + CO23   8.48
we account for the effects of double layer formation Kaolinite + 7.44
on the composition of the external water. This is Al2Si2O5(OH)4(s) + 6H+ =2 Al3 + + 2 H4SiO4 + H2O
achieved by applying the electrostatic model of Gypsum  4.58
Borkovec and Westall (1983), which is built-in in CaSO4 . 2 H2O(s) = 2H2O +Ca2 + + SO24 
the PHREEQC-2 v. 2.4 code (Parkhurst and Appelo,
Soluble solids, inventory (mol kg  1)
1999). The model relies on the Poisson –Boltzmann’s Halite NaCl 1.35  10 3[5]
equation, which relates electrostatic surface potential
to charge density in the DDL, and results in the Reacting groundwater (Opalinus Clay water)
simultaneous calculation of both diffuse layer and pH 7.24 [2]
Ionic strength (eq l 1) 0.228
external solution compositions. The basic equation
Solution composition (mmol l 1):
states that the second derivative of the surface Na 168.6, Ca 10.5, Mg 7.5, K 5.7, Cl 160, S(VI) 24.0, C
potential must be proportional to the ratio of local (inorganic) 2.7
charge density and dielectric constant of the aqueous References: [1] Nagra, unpublished data; [2] Pearson (2002); [3]
medium: Baeyens and Bradbury (1997); [4] Wieland et al. (1994); [5]
Bradbury and Baeyens (2002); [6] Hummel et al. (2002).
d2 w qðxÞ
2
¼ ð1Þ
dx e

where x is the distance from the surface (m), w(x) is simultaneously with the chemical equilibria consid-
the surface potential (V), q(x) is charge density (C ered in the system. For every charged ion i the
m 3), and e is the dielectric constant of the aqueous ‘‘surface excess’’ Ci,s (mol m 2) is calculated as
medium (C m 1 V 1). This equation is solved the difference between local concentration ci(x) in
P. Wersin et al. / Applied Clay Science 26 (2004) 249–257 253

the DDL and concentration coi in the bulk aqueous (1952) indicate that the dielectric constant of adsorbed
solution (see Fig. 3): water in clays is considerably smaller than in the bulk
Z l water, implying that ionic association reactions would
 
Ci;s ¼ ci ðxÞ  coi dx : ð2Þ be favoured in the DDL water (i.e., dissociation
xd constants for ion pairs would be smaller than in the
bulk water, see Fuoss, 1958).
Although the integration is carried out from the
position of the Stern layer, xd, to infinity, an oper-
2.4. Water incorporation in the interlayer
ational thickness of the DDL can be defined in terms
of Debye length, 1/j, which is by definition the
Kahr et al. (1986, Fig. 11) determined experimen-
distance at which the surface potential has decreased
tally the lattice expansion (c*-axis) for montmorillon-
by a factor e. The Debye length is not fixed and
ite separated from MX-80 bentonite, as a function of
decreases with increasing ionic strength of the me-
the gravimetric water content. Such measurements can
dium (Horseman et al., 1996). Ci,s can be converted
be used to evaluate the volumetric expansion of clay
to average concentrations in the double layer when
minerals following incorporation of water in the
specific surface area r (m2 kg 1) and S/L ratio (kg
structural interlayer (swelling). Smectites can host
l 1) are known:
exchanged cations and up to four sheets of water
molecules in the interlayer (Grauer, 1986). These
Ci;DDL ¼ rðS=LÞCi;s : ð3Þ
exchanged cations neutralise the negatively charged,
unexpandable tetrahedral – octahedral – tetrahedral
There are nevertheless two limitations in the
units of montmorillonite particles.
PHREEQC implementation. First, the Debye length
Assuming constant bulk volume conditions, a
1/j is fixed during the calculation and cannot be
residual porosity can be calculated, representing
adjusted automatically to ionic strength and electro-
the ratio of external water volume (i.e., volume
lyte composition. Attempts to do this by manual
occupied by water not incorporated in the structural
iterations failed due to program crashes at 1/j < 1
interlayer of montmorillonite) to bulk bentonite
nm. Second, the PHREEQC model assumes that both
volume after swelling has occurred. If the bulk
speciation and dielectric constant are the same in the
densities of dry and water-saturated bentonite
DDL water as in the bulk solution, which is probably
(qdry,qwet) are known, the external water content
not correct. Measurements by Cownie and Palmer
after swelling can be determined. One starts with
the definitions:

ms
qdry ¼ ð4Þ
Vb

and

ms þ mw
qwet ¼ ; ð5Þ
Vb

where ms is the mass of dry minerals, mw is the


total mass of incorporated water and Vb is the bulk
bentonite volume. At constant volume, the follow-
ing relation applies:

ms ms þ mw
Fig. 3. Sketch of surface potential vs. distance for diffuse double ¼ ð6Þ
layer model. Symbol explanation is discussed in text. qdry qwet
254 P. Wersin et al. / Applied Clay Science 26 (2004) 249–257

from which one obtains [noting that w u (mw/ms)]: bentonite. The volume of solid after the expansion
due to the water saturation process is then
qwet 0.62  (1 + 0.535) = 0.952 m3, leaving a residual
w¼ 1: ð7Þ
qdry porosity of 0.048. Thus, our calculation predicts
that about 85% of the initial physical porosity is
Substitution of the reference values for the dry and destroyed by the swelling process, leaving a resid-
wet bentonite densities (1711.2 and 2091.2 kg m 3, ual external porosity of about 5%. This residual
see Curti and Wersin, 2002) yields w = 0.222. Taking porosity is taken to represent the volume of exter-
into account the presence of 25% accessory minerals nal solution, i.e., the sum of DDL and uncharged
in MX-80 bentonite, we end up with the water content external solution.
of w* = 0.296, normalised to the weight of pure
montmorillonite. 2.5. Model implementation
From Fig. 11 in Kahr et al. (1986), one extracts
d0 c 9.66 Å and d c 16.5 Å for the d-spacing before The implementation of the refined model with
and after swelling, respectively (see Fig. 4). The PHREEQC follows this general computation
relative volumetric expansion due to swelling is then: scheme. In a first step, a very small amount of
pure water reacts with the initial clay and minor
DVs d  d0 minerals, yielding a tiny amount of solution in
¼ xM c53:5% ð8Þ
Vs0 d0 equilibrium with the exchanged cations in the
bentonite. Because the amount of water is negli-
where xM = 0.755 is the volume fraction of montmo- gibly small, the initial inventories of minor solids
rillonite in MX-80. The volume of dry mineral per (calcite, gypsum, quartz, kaolinite) and exchanged
unit volume of bentonite is given by: cations are not modified by this calculation. This
solution, which represents a native ‘‘formation
qdry Vb 1711:2  1 water’’ in equilibrium with the initial clay, is then
Vs0 ¼ ¼ c0:62 m3 ð9Þ
qs 2760 rescaled to 1 kg of water. Note that this pristine
solution composition is not unique because the
which yields an initial porosity (edry) of 0.38. This original partial pressure of CO2 in equilibrium
represents the air-filled volume in 1 m3 of dry with the bentonite is not known. Here we arbi-
trarily assume a CO2 partial pressure of 10 2.2
bar, the same partial pressure assumed for the
conditions expected in a repository. This arbitrary
assumption has however negligible effects on the
final results because of the effective buffering
mechanism operating in this system (see below
and Curti and Wersin, 2002 for details). In a
second step, the surface site speciation and DDL
composition in equilibrium with 1 kg of pristine
‘‘formation water’’ are calculated. The inferred
amount of chloride released from the MX-80
bentonite upon wetting (1.35 mmol Cl/kg clay)
is modelled by specifying an appropriate NaCl
concentration in the Opalinus Clay water. In the
third step, the reactive minor solids (calcite, gyp-
Fig. 4. Structure of dry montmorillonite projected on the (a,c) plane. sum, quartz and kaolinite) and the initial clay with
Note that d0 is not identical to the c0 lattice parameter because
montmorillonite is monoclinic (c0 is represented through the long the inferred surface species are equilibrated with
segment of the unit cell). The separation between basal planes (h) Opalinus Clay water, yielding the required benton-
considerably increases upon swelling. ite porewater.
P. Wersin et al. / Applied Clay Science 26 (2004) 249–257 255

3. Results Table 2
Model calculations for the reaction between Opalinus Clay water
and MX-80 bentonite at S/L = 4.5 kg l 1 and pCO2 = 10 2.2 bar
Calculations with and without the explicit deter-
Solution Initial Model
mination of the DDL composition were performed
(mmol/kg conditions calculations
for a system composed of Opalinus Clay water H2O) (reacting (postreaction)
reacting with MX-80 bentonite at S/L = 4.5 kg l 1. system)
All calculations were carried out with the selectivity Opalinus No DDL DDL
and surface complexation parameters reported in clay DDL assumed, assumed +
Table 1. A summary of the results is given in Table water assumed all water interlayer
2. The first model calculation was carried out with- external water
out consideration of the DDL, assuming that all the Na 169 236 251 482
water present in the system is available to every K 5.7 2.8 3.0 5.6
Mg 7.5 8.8 9.6 25.9
dissolved species. The second calculation differs
Ca 10.5 13.7 14.1 27.6
from the first one only inasmuch as formation of a Cl 160 168 181 524
DDL is considered, but without taking into account CO3 2.7 2.8 2.7 2.0
water displacement in the interlayer of montmoril- SO4 24.0 60.0 59.0 34.5
lonite. The results indicate only small anion exclu- Ionic strength 0.228 0.308 0.320 0.618
pH 7.24 7.25 7.24 7.03
sion effects compared to the case without DDL, the
Cl molality increases from 0.168 to 0.181 in spite of D Solids/gases (mmol/kg H2O), + precipitation,  dissolution
the large fraction of water assigned to the DDL (25% Calcite + 3.4 + 2.5 + 18.4
of the total external water). Gypsum  35.8  31.2 + 24.2
In the third calculation, we studied the additional CO2(g)  3.5  2.4  15.8
effect of water incorporation in the interlayer. It was
Exchange ions (equivalent fraction), MX-80
assumed that water is removed until 11 –12% of the NaX 0.848 0.832 0.839 0.866
total incorporated water is left as external solution, CaX2 0.084 0.100 0.094 0.066
corresponding to the calculated residual external po- MgX2 0.051 0.050 0.050 0.050
rosity of 5%. Our results now indicate significant KX 0.017 0.018 0.018 0.018
CEC/eq kg 1 0.787 0.787 0.787 0.787
effects; with respect to the preceding calculation, pH
decreases by 0.2 units, while the Na, Cl, K, Ca EDGE SITES (equivalent fraction)
elemental concentrations in the uncharged external QSOH+2 0.017 0.023 0.022 0.051
solution increase by a factor of 2 – 3. In contrast, QSOH 0.433 0.476 0.475 0.597
carbonate and sulphate concentrations are reduced. QSO 0.550 0.501 0.503 0.352
Net charge  68.1  61.2  79.1  49.7
These effects can be explained by a reversal of the
[meq/kg H2O]
Na – Ca exchange reaction. Due to the incorporation
of water in the interlayer, Na is concentrated in the DDL (mmol/kg H2O)
external water to a level sufficient for exchanging Ca Na – 376 559
from the clay (in all previous cases, the opposite Ca – 28.5 39.0
Mg – 18.6 36.2
occurred). The displacement of exchanged Ca into
SO4 – 37.2 31.9
the aqueous phase entrains gypsum and calcite pre- CO3 – 2.1 1.9
cipitation with consequent acidification of the solu- Cl – 127 479
tion. The relevant reactions are: % Interlayer water 0 0 66.0
% External water 100 77.5 11.6
% Water in DDL 0 22.5 22.5
CaSO4 ðsÞ þ 2NaX ! 2Naþ þ CaX2 þ SO2
4 ð10Þ

In the first two calculations (no water assigned to


Ca2þ þ CO2 ðgÞ þ H2 O ! CaCO3 ðsÞ þ 2Hþ ð11Þ
interlayer), Ca is exchanged for Na in montmorillon-
ite, causing the dissolution of more than 30 mmol of
ZSO þ Hþ ! ZSOH0 : ð12Þ gypsum per kg water. All the three reactions above
256 P. Wersin et al. / Applied Clay Science 26 (2004) 249–257

proceed to the right; the pH is buffered through parameter which is not well constrained in the argilla-
reactions (11) and (12). If the displacement of water ceous repository environment, was determined to be
in the interlayer is considered, the Na concentration more significant.
rises to the point that the exchange reaction is re- Finally, it should be emphasised that other uncer-
versed. Na is displaced onto the clay exchanger and tainties, such as changes of the dielectric constant of
Ca is released to the solution, causing the precipitation water, could not be considered in our model due to
of considerable amounts of calcite and gypsum. The current lack of understanding. Further experimental
acidity released through calcite precipitation is only and theoretical investigations are required to improve
partially compensated by the protonation of surface our understanding on the thermodynamics of com-
hydroxyl groups (reaction (12)), thus pH decreases pacted clays.
significantly.
An uncertainty regards the influence of surface
complexes of major ions (e.g., Ca2 +, Mg2 +, K+) at Acknowledgements
edge sites, for which reliable constants are lacking.
However, scoping calculations performed with esti- This work has been part of Nagra’s high-level waste
mated formation constants for Mg and Ca surface disposal programme. We thank Lawrence Johnson,
complexes (Bradbury and Baeyens, 1995) suggest that Bernhard Schwyn, Bart Baeyens and Mike Bradbury
significant fractions of the edge site capacity would be for fruitful discussions, as well as an anonymous
occupied by Mg and Ca surface complexes, leading to a reviewer for the constructive criticism.
decrease of the negative surface charge. However, the
effects on the composition of the external solution
proved to be minor. References

Baeyens, B., Bradbury, M.H., 1997. A mechanistic description of


4. Discussion and conclusions Ni and Zn sorption on Na-montmorillonite: Part I. Titration and
sorption measurements. J. Contam. Hydrol. 27, 199 – 222.
Borkovec, M., Westall, J., 1983. Solution of the Poisson – Boltz-
The refined models presented in this section clearly mann equation for surface excesses of ions in the diffuse layer at
indicate that the compositional effects induced by the the oxide – electrolyte interface. J. Electroanal. Chem. 150,
explicit consideration of cation surface excesses in the 325 – 337.
Bradbury, M.H., Baeyens, B., 1995. A quantitative description of
diffuse double layer are generally negligible. However,
Ni, Zn and Ca sorption on Na-Montmorillonite: Part III. Mod-
significant changes in the porewater chemistry are elling. PSI Bericht, vol. 95-12. Paul Scherrer Institut, Villigen,
predicted if double layer effects are assumed in con- Switzerland. ISSN 1019-0643.
junction with the incorporation of water in the inter- Bradbury, M.H., Baeyens, B., 1997. A mechanistic description of
layer of montmorillonite at very high S/L ratios. Then, Ni and Zn sorption on Na-montmorillonite: Part II. Modelling.
our model predicts a more concentrated bentonite J. Contam. Hydrol. 27, 223 – 248.
Bradbury, M.H., Baeyens, B., 2002. Porewater chemistry in com-
porewater (I f 0.6 vs. f 0.3 m) and a lower pH pacted re-saturated MX-80 bentonite. Nagra Technical Report
(7.03 vs. 7.25) compared to the composition calculated NTB 01-08, Wettingen, Switzerland.
ignoring double layer and swelling effects. Bruno, J., Arcos, D., Duro, L., 1999. Processes and features affect-
From the perspective of performance assessment, ing the near field hydrochemistry. Groundwater-bentonite inter-
these changes are not very significant compared to actions. SKB Technical Report TR-99-29, Stockholm, Sweden.
Cownie, A., Palmer, L.S., 1952. The effect of moisture on the
other system uncertainties. In fact, a sensitivity analysis electrical properties of soils. Proc. Phys. Soc. B 65, 295 – 301.
for the Swiss high-level waste repository case (Curti Curti, E., Wersin, P., 2002. Assessment of porewater chemistry in
and Wersin, 2002) has indicated that anion exclusion the bentonite backfill for the Swiss SF/HLW repository. Nagra
effects (predicted with aid of the refined DDL model) Technical Report NTB 02-09, Wettingen, Switzerland.
has only a minor effect on solubility and speciation of Fernández, A.M., Cuevas, J., Rivas, P., 2001. Pore water chemistry
of the FEBEX bentonite. Mater. Res. Soc. Symp. Proc. 663,
safety-relevant radionuclides. Other system parameters 573 – 595.
were found to be more critical. Specifically, the effect Fuoss, R.M., 1958. Ionic association: III. The equilibrium between
of the CO2 partial pressure on bentonite porewater, a ion pairs and free ions. J. Am. Chem. Soc. 80, 5059 – 5061.
P. Wersin et al. / Applied Clay Science 26 (2004) 249–257 257

Grauer, R., 1986. Bentonit als Verfüllmaterial im Endlager für Muurinen, A., Lehikoinen, J., 1999. Porewater chemistry in com-
hochaktiven Abfall: Chemische Aspekte. Nagra Technical Re- pacted bentonite. Posiva Report, vol. 99-20. Posiva, Helsinki,
port NTB 86-12, Wettingen, Switzerland. Finland.
Horseman, S.T., Higgo, J.J.W., Alexander, J., Harrington, J.F., Parkhurst, D.L., Appelo, C.A.J., 1999. User’s guide to PHREEQC
1996. Water, gas and solute movement through argillaceous (Version 2) - A computer program for speciation, batch reaction,
media. Report CC-96/1 Nuclear Enegy Agency (NEA). Organi- one-dimensional transport, and inverse geochemical calcula-
sation for Economic Co-operation and Development (OECD), tions. U.S. Geological Survey Water-Resources Investigations
Paris, France. Report 99 – 4259.
Hummel, W., Berner, U., Curti, E., Pearson, J.F., Thoenen, T., 2002. Pearson, J.F., 2002. Benken Reference Water Chemistry. Nagra
Nagra/PSI Thermochemical Data Base 01/01. Nagra Technical unpubl. Report, Wettingen, Switzerland.
Report NTB 02-16, Wettingen, Switzerland. Wanner, H., Albinsson, Y., Karnland, O., Wieland, E., Wersin, P.,
Kahr, G., Krähenbühl, F., Müller-Vonmoos, M., Stöckli, H.F., Charlet, L., 1994. The acid/base chemistry of montmorillonite.
1986. Wasseraufnahme und Wasserbewegung in hochverdich- Radiochim. Acta 66/67, 157 – 162.
tetem Bentonit. Nagra Technical Report NTB 86-14, Wettin- Wersin, P., 2003. Geochemical modelling of bentonite porewater in
gen, Switzerland. high-level waste repositories. J. Contam. Hydrol. 61, 405 – 422.
Kraepiel, A.M.L., Keller, K., Morel, F.M.M., 1998. On the acid – Wieland, E., Wanner, H., Albinsson, Y., Wersin, P., Karnland, O.,
base chemistry of permanently charged minerals. Environ. Sci. 1994. A surface chemical model of the bentonite – water inter-
Technol. 32, 2829 – 2838. face and its implications for modelling the near field chemistry
Müller-Vonmoos, M., Kahr, G., 1983. Mineralogische Untersu- in a repository for spent fuel. SKB Technical Report TR-94-26,
chungen von Wyoming Bentonit MX-80 und Montigel. Nagra Stockholm, Sweden.
Technical Report NTB 83-12, Wettingen, Switzerland.

You might also like