Science of the Total Environment 610–611 (2018) 511–520

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Cesium sorption reversibility and kinetics on illite, montmorillonite,

and kaolinite
Chad B. Durrant a,b,⁎, James D. Begg a, Annie B. Kersting a, Mavrik Zavarin a
a
Glenn T. Seaborg Institute, Physical & Life Sciences, Lawrence Livermore National Laboratory, Livermore, CA 94550, United States
b
The Pennsylvania State University, 101 Brezeale Nuclear Reactor, University Park, PA 16802, United States

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Permeable membrane used to examine

desorption of 137Cs in the presence of
two minerals.
• Ternary experiments carried out of over
N 500 days.
• Cs is reversibly sorbed to kaolinite and
montmorillonite.
• Ternary experiments indicate a slow ki-
netic effect on Cs desorption from illite.

a r t i c l e i n f o a b s t r a c t

Article history: Understanding sorption and desorption processes is essential to predicting the mobility of radionuclides in the en-
Received 19 June 2017 vironment. We investigate adsorption/desorption of cesium in both binary (Cs + one mineral) and ternary (Cs
Received in revised form 12 August 2017 + two minerals) experiments to study component additivity and sorption reversibility over long time periods
Accepted 12 August 2017
(500 days). Binary Cs sorption experiments were performed with illite, montmorillonite, and kaolinite in a
Available online 17 August 2017
5 mM NaCl/0.7 mM NaHCO3 solution (pH 8) and Cs concentration range of 10–3 to 10–11 M. The binary sorption
Editor: Jay Gan experiments were followed by batch desorption experiments. The sorption behavior was modeled with the
FIT4FD code and the results used to predict desorption behavior. Sorption to montmorillonite and kaolinite was
Keywords: linear over the entire concentration range but sorption to illite was non-linear, indicating the presence of multiple
Ternary system sorption sites. Based on the 14 day batch desorption data, cesium sorption appeared irreversible at high surface
Desorption loadings in the case of illite but reversible at all concentrations for montmorillonite and kaolinite. A novel experi-
Irreversibility mental approach, using a dialysis membrane, was adopted in the ternary experiments, allowing investigation of
Permeable membrane the effect of a second mineral on Cs desorption from the original mineral. Cs was first sorbed to illite, montmoril-
lonite or kaolinite, then a 3.5–5 kDalton Float-A-Lyzer® dialysis bag with 0.3 g of illite was introduced to each ex-
periment inducing desorption. Nearly complete Cs desorption from kaolinite and montmorillonite was observed
over the experiment, consistent with our equilibrium model, indicating complete Cs desorption from these min-
erals. Results from the long-term ternary experiments show significantly greater Cs desorption compared to the
binary desorption experiments. Approximately ~45% of Cs desorbed from illite. However, our equilibrium model
predicted ~65% desorption. Importantly, the data imply that in some cases, slow desorption kinetics rather than
permanent fixation may play an important role in apparent irreversible Cs sorption.
© 2017 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Glenn T. Seaborg Institute, Physical & Life Sciences, Lawrence Livermore National Laboratory, Livermore, CA 94550, United States.
E-mail address: durrant1@llnl.gov (C.B. Durrant).

http://dx.doi.org/10.1016/j.scitotenv.2017.08.122
0048-9697/© 2017 Elsevier B.V. All rights reserved.
512 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520
1. Introduction
Significant quantities of radioactive cesium (Cs) have been released
to the environment. These releases have occurred as a result of nuclear
weapons testing (Klement, 1965), nuclear accidents at Chernobyl,
Russia and Fukushima, Japan (Castrillejo et al., 2016; Steinhauser et al.,
2015), and leaks from high level waste storage sites such as Hanford,
USA (Zachara et al., 2002) and Mayak, Russia (Novikov et al., 1998).
The persistence of 137Cs (t1/2 30.1 y) in the environment can pose sub-
stantial health concerns. The U.S. EPA regulates the amount of beta
emitting radionuclides in drinking water such that the equivalent dose
is b4 mrem per year (United States Environment Protection Agency,
2009). This dose corresponds to a maximum 137Cs concentration of
7.4 Bq/L or 1.7 × 10−14 M. Thus, 137Cs is considered a drinking water
hazard even at trace concentrations. The environmental contamination
and potential health risks associated with 137Cs necessitate understand-
ing its behavior in the environment. Understanding this behavior may
also guide remediation strategies for Cs clean-up in the environment
and help to design strategies for the safe storage of radionuclides pres-
ent in spent nuclear fuel.
Sorption and desorption processes play a large role in controlling the
fate of Cs in the environment. Sorption to immobile soils or sediments
will limit the transport of Cs in solution, while sorption to colloids (de-
fined as b1 μm particulates) may enhance mobility. If Cs is sorbed to
minerals, then its desorption behavior will affect its bioavailability and
in the case of mineral-colloid association, the extent of transport.
Due to their high sorption capacity and their ubiquity in nature, alu-
minosilicate clays are particularly important in controlling Cs behavior
in natural systems. Cs sorption reversibility on clay minerals with
high, intermediate, and low Cs affinities (illite, montmorillonite, and ka-
olinite, respectively) was the focus of this study. Illite is a non-
expanding 2:1 (2 tetrahedral silica layers sandwiching an octahedral
alumina layer, TOT) aluminosilicate clay. Cesium sorption to illite has
been described as a cation exchange process on either the basal/planar,
edge, or interlayer sites (Staunton and Roubaud, 1997). Weathering
near the edges of the illite expands the interlayer producing wedge or
frayed edge sites (FES) (Jackson, 1963). These FES sites are highly selec-
tive for Cs (Rich and Black, 1964; Bolt et al., 1963; Brouwer et al., 1983).
As a result, while illite has much lower cation exchange capacity (CEC)
than montmorillonite, at Cs concentrations below 10− 7 M, illite will
sorb more Cs than montmorillonite per unit mass (Sawhney, 1970;
Komarneni and Roy, 1980).
Montmorillonite is a 2:1 smectite clay with an expanding interlayer
(Jackson, 1963) that allows for increased interaction between cations in
solution and those in the clay interlayer. The high CEC of montmorillon-
ite leads to its high affinity for Cs (Sawhney, 1970; Sawhney, 1972; Hsu
and Chang, 1995; Komarneni and Roy, 1979).
Kaolinite is a 1:1 aluminosilicate non-expanding clay. Sorption of Cs
to kaolinite occurs by cation exchange at permanently charged sites on
the basal surfaces of the mineral (Swartzen-Allen and Matijevic, 1974;
Kim et al., 1996; Kim and Cygan, 1996; Ohnuki and Kozai, 2013;
Reinoso-Maset and Ly, 2014). Due to its low CEC and absence of FES
sites, kaolinite sorbs Cs less effectively than montmorillonite or illite.
In terms of desorption behavior, research has focused on whether Cs
is irreversibly sorbed by clay minerals. In this context we mean that
sorption is considered to be reversible if Kdsorption ≈ Kddesorption
(Missana et al., 2004). However, determining true irreversibility can
be difficult because sorption and desorption behavior can be affected
by the solution composition, kinetics, Cs concentration; and the pres-
ence of multiple sorption sites with differing Cs affinities (Comans
et al., 1991; Bostick et al., 2002; Missana et al., 2004; Iijima et al., 2010).
In the case of montmorillonite and montmorillonite dominated
clays, sorption is believed to be reversible (Missana et al., 2004; Iijima
et al., 2010), however recent work has shown that some fixation of in-
trinsic Cs may be achieved by the addition of divalent cations before de-
sorption (Fukushi et al., 2014; Fukushi and Fukiage, 2015). Similarly, Cs
is readily desorbed from the surface of kaolinite (Shahwan et al., 1999;
Erten et al., 1988). However, the behavior with illite is more complex
with apparent irreversible behavior observed in some cases. This irre-
versibility has been attributed to sorption to high affinity FES, and
their subsequent collapse that prevents access to the sites (Sawhney,
1972). High concentrations of Cs have been shown to exacerbate inter-
layer collapse leading to apparent irreversible sorption behavior. A time
dependence on the amount of reversibly sorbed Cs has been attributed
to diffusion of Cs from FES to the interlayer sites of illite (Comans and
Hockley, 1992). However, more recently a time dependent increase in
sorption irreversibility has been observed at timescales too short to be
explained by diffusion processes (De Koning and Comans, 2004).
Most sorption/desorption studies are conducted as single mineral
binary batch experiments over short time periods (days to weeks).
However, in the environment, Cs will interact with multiple minerals si-
multaneously and over long time scales. This presence of multiple min-
erals is especially important to consider in the case of colloid-facilitated
transport of Cs and previous column experiments have demonstrated
that Cs can be desorbed from mobile colloids and adsorbed on the im-
mobile host minerals (Zhuang et al., 2003; Turner et al., 2006). Previous-
ly, De Koning and Comans (2004) examined Cs sorption reversibility on
illite with ammonium hexacyanoferrate (HCF) which has a high selec-
tivity and capacity for Cs. The high sorption affinity of HCF ensures
that solution Cs concentrations are low and drives desorption of Cs
from illite. In our current work, we employ a similar approach but use
illite to drive desorption. This novel approach has the advantage of
being more representative of natural environments.
In this work, we evaluate Cs sorption to illite, montmorillonite, and
kaolinite over a wide range of initial Cs concentration that extends to
low, environmentally relevant concentrations. We perform binary Cs
sorption and desorption experiments with illite, montmorillonite, and
kaolinite across an initial Cs concentration range of 10−3 to 10−11 M.
We then examine long-term Cs sorption/desorption behavior in two-
mineral ternary experiments. By keeping the two mineral phases phys-
ically separated using a Specta/Por® Float-A-Lyzer® dialysis bag, we are
able to examine the contribution of each mineral to the overall sorption
and evaluate the competitive sorption and desorption of Cs in ternary
experiments over a period of nearly 17 months. This method provides
a means to evaluate component additivity (Davis et al., 1998), revers-
ibility, and desorption kinetics in the presence of clay minerals over en-
vironmentally relevant timescales.
2. Materials and methods
2.1. Materials
All solutions were prepared using ACS grade chemicals without fur-
ther purification and ultrapure water (Milli-Q Gradient System,
N18 MΩ·cm). Fithian illite from Rochester, NY (Ward's Science), SWy-
1 montmorillonite (Clay Minerals Repository), and KGa-1b kaolinite
(Clay Minerals Repository) were used in this study. Preparation of
homoionic SWy-1 Na-montmorillonite has been previously reported
(Zavarin et al., 2012) and the same method was also used for the prep-
aration of illite and kaolinite. Briefly, the minerals were prepared by
mixing 100 g/L of the mineral in 0.001 M HCl for 30 min in order to re-
move soluble salts and impurities. Next, 0.03 M H2O2 was added for
30 min to minimize the oxidative and reductive potential of any remain-
ing impurities as well as removing trace organic material. The clay sus-
pensions were then centrifuged for 6 h at 2500g and the supernatant
(calculated to contain particles b50 nm diameter) was discarded. The
wet solids were transferred to 6000–8000 Da molecular weight cut off
(MWCO) dialysis bags and suspended in a 0.01 M NaCl electrolyte solu-
tion and dialyzed for seven days to homoionize the minerals. It is possi-
ble that there still exists some intrinsic Cs and this may have an effect on
the sorption processes. However, the effect is likely to be small and no
further consideration is given in this work. The minerals were extruded
 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520
from the dialysis bags into Milli-Q H2O. The suspension was then centri-
fuged (sedimentation for 1 h was used for illite and kaolinite) at 100g
for 5 min and the supernatant containing particles b 2 μm was discarded.
The remaining solids (50 nm b − b 2 μm) were dried at 60 °C. A portion
of the dried minerals was resuspended in a 5 mM NaCl and 0.7 mM
NaHCO3 solution (pH 8) with an ionic strength of 0.0057 M as a synthet-
ic groundwater. The solid:solution ratio of the mineral suspension was
1 g/L unless otherwise noted. The mineral suspensions were allowed
to equilibrate over several days.
2.2. Binary sorption experiments
Ten mL of each mineral suspension was added to 15 mL polyethyl-
ene centrifuge tubes. A total of 17 different initial Cs concentrations
ranging from 10−3 to 10−11 M were prepared. A combination of stable
CsCl and 137Cs was added to samples with Cs concentrations of 10−9 M
and higher and only 137Cs for concentrations lower than 10−9 M. The
centrifuge tubes were then placed on an orbital shaker at 0.4 g at
room temperature and kept in the dark for the duration of the experi-
ment. The samples were shaken continuously for at least two weeks
to allow sufficient time for Cs to sorb to the mineral (Comans et al.,
1991). Mineral blanks were also performed to test for any sorption to
container walls. At the end of the sorption period, each tube was centri-
fuged for two hours at 3600g in order to obtain a b 50 nm particle size
cut off in the supernatant. An aliquot of the supernatant was taken
from each sample and the 137Cs measured by liquid scintillation
counting (LSC, Packard Tri-Carb TR2900 LSA and Ultima Gold cocktail).
The distribution coefficient, Kd, for each sample was calculated as fol-
lows:
½Csi −½Cseq 1 and rinsed several times with Milli-Q water to remove any residue
Kd ¼  ð1Þ
½Cseq ½mineral
where [Cs]i is the initial concentration of Cs, [Cs]eq is the equilibrium
concentration of Cs in the aqueous phase, and [mineral] is the
solid:solution ratio in g/mL. The units of Kd are mL/g.
2.3. Binary desorption experiments
Binary desorption experiments were performed at the conclusion of
the sorption experiments. At the end of the 14-day sorption period and
after centrifugation, a total of 9.5 mL of the supernatant was quantita-
tively removed and replaced with 9.5 mL of Cs-free buffer solution.
The centrifuge tubes were placed in the dark on the orbital shaker and
shaken continuously for at least two weeks. The centrifuge tubes were
then spun for 2 h at 3600 g (b50 nm particle size cut off). An aliquot
of the supernatant was removed and 137Cs measured by LSC. The mea-
sured 137Cs was used to determine the total fraction of Cs desorbed
from each of the samples. At least one sequential batch exchange was
carried out for each mineral by repeating the above steps.
2.4. Ternary sorption/desorption experiments
In our two-mineral ternary Cs sorption/desorption experiments,
150 mL of each clay suspension (dry clay mass of 0.15 g) was added to
250 mL centrifuge tubes. 137Cs was added as a tracer to stable Cs before
being added to each of the mineral suspensions. Cesium concentrations
were chosen such that Cs sorption to illite is dominated by the high af-
finity FES sites (low Cs) and edge or planar site begin to contribute to
total Cs sorption (high Cs). Three illite suspensions with Cs concentra-
tions of 10−6, 5 × 10−7 and 10−7 M were prepared. Three montmoril-
lonite suspensions with Cs concentrations of 5 × 10−7, 10− 7, and 5
× 10−8 M were prepared and two kaolinite suspensions with Cs con-
centrations of 5 × 10−7 and 10−7 M were prepared. Lastly, solutions
with Cs concentrations of 10−6, 5 × 10−7, 10−7, and 5 × 10−8 M (and
no minerals) were used as controls. All samples were placed on an
513
orbital shaker in the dark and allowed to equilibrate for two weeks.
After the equilibration period, 5 mL of a 60 g/L illite suspension (0.3 g
total) was loaded into 3500–5000 MWCO Da Float-A-Lyzers (see graph-
ical abstract). The Float-A-Lyzers with the illite suspensions were then
inserted into each of the centrifuge tubes. The centrifuge tubes were
then returned to the dark on the orbital shaker. The suspensions outside
the Float-A-Lyzers were periodically sampled over the duration of the
experiment, a period of nearly 17 months. Samples were taken as fol-
lows: a 1 mL aliquot of the suspension was taken and counted by LSC.
Then a 1.2 mL aliquot of the suspension was removed and centrifuged
for 2 h at 3600 g to achieve a particle size cutoff of 50 nm. A 1 mL aliquot
of the supernatant was removed and counted by LSC to determine the
aqueous Cs concentration. Separate experiments were performed to
verify that the suspensions were sufficiently stable to allow for quanti-
tative and reproducible measurement of the total Cs activity in the sus-
pensions. Characterization of sample blanks indicated that Cs sorption
to the membrane or walls of the centrifuge tube was negligible and
the Float-A-Lyzer pore size allowed for exchange of aqueous Cs (SI Fig.
A1). Therefore, it can be assumed that the decrease in total Cs outside
the Float-A-Lyzer corresponds to a proportional increase of the total
Cs sorbed to illite inside the Float-A-Lyzer. The amount of Cs sorbed to
the illite in the Float-A-Lyzer was calculated as follows:
Csid;n ¼ Cs0 −Csbulk;n ð2Þ
where Csid,n is the amount of Cs (moles) on the illite inside the Float-A-
Lyzer after n days, Cs0 is the amount of Cs in the suspension before the
Float-A-Lyzer was added and Csbulk,n is the amount of Cs in the bulk sus-
pension outside the Float-A-Lyzer after n days. After the last sampling
time point, the Float-A-Lyzer was removed from the centrifuge tube
from the outside of the Float-A-Lyzer. The Cs activity in the Float-A-
Lyzer was then measured by gamma spectroscopy to determine the
total amount of Cs sorbed to the illite and compared to the values calcu-
lated from Eq. (2).
2.5. Modeling
The ion exchange reaction between Cs and each of the Na-
homoionized clay minerals was written as follows:
Na−Clay þ Csþ ↔Cs−Clay þ Naþ ð3Þ
Based on this reaction an ion exchange model was developed to de-
scribe Cs sorption to each mineral using the FIT4FD code (Zavarin et al.,
2005). The FIT4FD code was allowed to adjust the selectivity coefficient
and CEC parameters to reproduce Cs sorption for each of the Cs mineral
sorption isotherms. The Vanselow convention, which is used in the
model for this work, defines the selectivity coefficient (KSEL) for the re-
action as follows:
Cs ðM Cs ÞZ Na ðaNa ÞZ Cs
Na K SEL ¼ ð4Þ
ðM Na ÞZ Cs ðaCs ÞZ Na
where MCs and MNa are the mole fractions of Cs and Na defined as the
moles of Cs or Na sorbed divided by the total number of moles sorbed,
Z is the cation charge (+ 1 in the case of Na and Cs), and aCs and aNa
are the activities of Cs and Na is solution, respectively (Vanselow, 1932).
3. Results and discussion
3.1. Binary sorption
Binary Cs sorption experiments were conducted with illite, kaolinite,
and montmorillonite across an initial Cs concentration range of 10−3 to
10−11 M. Fig. 1 presents these sorption data as the logarithm of the
514 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520

Fig. 1. Cs sorption Kd values (mL/g) for ( ) Na-illite, ( ) Na-kaolinite, and ( ) Na-montmorillonite as a function of Cs concentration in solution (mol/L). All experiments performed in
0.7 mM NaHCO3, 5 mM NaCl buffer solution at pH 8. The solid solution ratio is 1 g/L for each mineral and 2 σ error bars are propagated from 2% LSC uncertainties. The solid lines
represent the sorption models with parameters developed using the FIT4FD code.

distribution coefficient (log(Kd)) versus the logarithm of the equilibri- At Cs concentrations below 5 × 10−9 M, there is a pronounced differ-
um aqueous Cs concentration. Cs sorption to illite is linear for Cs concen- ence in the extent of Cs sorption between illite, montmorillonite, and
trations below 5 × 10−9 M with an average log(Kd) value of 4.7 ± 0.4 kaolinite. The equilibrium Kd for illite is approximately two orders of
(uncertainty reported at 2 standard deviations throughout the text). magnitude greater than the Kd values for montmorillonite or kaolinite.
At Cs concentrations N 5 × 10− 9 M, the Kd decreases with increasing This highlights the importance of Cs concentration and the mineralogy
concentration. At a Cs concentration of 4 × 10− 4 M, the log(Kd) was present when predicting Cs environmental behavior. The Kd values for
2.4. The transition from linear sorption to non-linear sorption has previ- each of the three minerals are consistent with Kd values reported in re-
ously been explained by the presence of multiple sorption sites on illite cent literature. However, the specific Kd values are affected by the ionic
(Brouwer et al., 1983; Comans et al., 1991; Poinssot et al., 1999). The Cs strength (SI Fig. A2), electrolyte composition, and the mineral source
initially sorbs to the high affinity FES sites until filled and then the Cs be- used by the various authors (Ohnuki and Kozai, 2013; Benedicto et al.,
gins to sorb to the edge and planar sites. Hence, at low Cs concentrations 2014; Fuller et al., 2014; Missana et al., 2014; Reinoso-Maset and Ly,
where the FES sites are not saturated, the Kd values are high and con- 2014).
stant. As the FES sites become saturated and the Cs begins to sorb to The FIT4FD code was used to model the sorption of Cs to each of the
the less selective edge and planar sites, the Kd begins to decrease. Our minerals. Selectivity coefficients and cation exchange capacities, includ-
experiments are consistent with behavior reported in the literature ing the site concentrations which comprise the CEC of each mineral,
where the high affinity FES sites are saturated around 5 × 10−9 M were initially based on values from the literature for each clay
(Missana et al., 2014). (Missana et al., 2014). However, the selectivity coefficients were adjust-
In contrast to illite, Cs sorption to montmorillonite is linear across ed to improve the fits to the data. Model fits to the data are shown as
the entire concentration range investigated with an average log(Kd) solid lines in Fig. 1 and the fitted selectivity coefficients and cation ex-
value of 2.9 ± 0.2. Montmorillonite has previously been described as change capacities are listed in Table 1. Also listed are the selectivity co-
having only a single sorption site (Staunton and Roubaud, 1997) and efficients and cation exchange capacities from Missana et al., 2014 for
the linearity of our sorption results are consistent with that conceptual comparison. Although the source of the minerals used by Missana
model. In the case of kaolinite, sorption is linear for equilibrium Cs con- et al. (2014) is different from those used here, e.g. FEBEX bentonite vs.
centration of 10−11–10−5 M with an average log(Kd) value of 2.5 ± 0.2. SWy-1 montmorillonite, the total site concentrations presented by
Above 10−5 M Cs, sorption to kaolinite becomes non-linear.
Both 1-site and 2-site models have previously been used to describe
Table 1
the sorption of Cs to kaolinite (Erten et al., 1988; Cornell, 1993; Ohnuki,
Selectivity coefficient and site concentration parameters used in model development of Cs
1994; Missana et al., 2014). The 2-site models usually do not represent sorption onto Na-clay minerals. The uncertainties represent two standard deviations pro-
2-sites on kaolinite but rather reflect the presence of high affinity, low duced by the model.
capacity sites attributed to trace amounts of other clay minerals
log (Cs
NaKSEL) Sites (mol/g)
such as illite and vermiculite (Reinoso-Maset and Ly, 2014; Ohnuki
and Kozai, 2013; Ohnuki, 1994; Missana et al., 2014). In our work, Missana This Work Missana This Work
et al.a et al.a
X-ray diffraction (XRD) analysis showed no evidence of additional
trace minerals in montmorillonite or kaolinite (SI Figs. A3, A4). Fur- Illite FES sites 6.90 ± 0.15 6.39 ± 0.1 4.62 × 10−7 5.92 × 10−7 ± 0.08
Illite site 2 3.10 ± 0.15 3.02 ± 0.04 3.88 × 10−5 1.76 × 10−5 ± 0.13
thermore, the presence of high affinity, low capacity sites attributed
Illite site 3 1.75 ± 0.15 1.31 ± 0.04 1.55 × 10−4 1.81 × 10−4 ± 0.16
to trace phases should lead to non-linearities at low Cs concentration Smectite site 1 7.59 ± 0.15 NA 2.31 × 10−8 NA
that were not observed here. Therefore, Cs sorption to kaolinite ob- Smectite site 2 1.68 ± 0.15 0.77 ± 0.1 1.02 × 10−3 1.03 × 10−3 ± 0.15
served here can be attributed to a single cation exchange site. Kaolinite site 1 5.50 ± 0.15 NA 5.00 × 10−9 NA
Above 4 × 10− 5 M Cs, the decrease in Kd is likely caused by a satura- Kaolinite site 2 2.10 ± 0.15 2.05 ± 0.1 1.99 × 10−5 2.01 × 10−5 ± 0.07

tion of planar sorption sites. a

(Missana et al., 2014).
 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520 515

Missana et al. (2014) are similar to our fitted CEC values. Moreover, our
values are consistent with previously published values for the minerals
used in this work: 19.1, 94.0, and 2.0 meq/100 g for Fithian illite
(Comans et al., 1991), SWy-1 montmorillonite (Staunton and
Roubaud, 1997) and KGa-1 kaolinite (van Olphen and Fripiat, 1979)
respectively.
Previous evaluations of Cs sorption to illite over a wide range of Cs
concentrations have used both 2-site and 3-site models (Poinssot
et al., 1999; Bradbury and Baeyens, 2000; Zachara et al., 2002;
Benedicto et al., 2014; Missana et al., 2014). The third site has been
shown to contribute to sorption only at Cs concentrations N10−4 M
(Bradbury and Baeyens, 2000; Benedicto et al., 2014). Since the highest
initial Cs concentration in our binary isotherm experiments was
10−3 M, a 3-site model was used here. The fitted parameters are in
very good agreement with the recently published values of Missana
et al. (2014).
Cs sorption to montmorillonite has been simulated with both 1-site
and 2-site models. Staunton and Roubaud (1997) and Iijima et al.
(2010) showed that a 1-site model can describe the sorption of Cs to
montmorillonite (Staunton and Roubaud, 1997; Iijima et al., 2010).
When 2-site models have been used, the second site is often attributed
to minerals such illite or vermiculite that are interstratified in the mont-
morillonite (Zachara et al., 2002; Missana et al., 2004; Missana et al.,
2014). In our case, the presence of a second site was not inferred from
the data nor warranted in our modeling (see SI Fig. A5). The fitted selec-
tivity coefficient for our data is an order of magnitude smaller than that
reported by Missana et al. (2014). This is likely the result of significant
illite inclusions in the smectite used by Missana et al. (2014), however,
our fitted selectivity coefficient is in good agreement with that reported
by Gast et al. (1969) for the same montmorillonite used in this work.
Gast et al. (1969) reported a log(Cs NaKSEL) of 0.79 compared to a log(Na
- At trace concentrations, irreversible sorption to illite has been attrib-
Cs
KSEL) of 0.69 reported in this work (Gast et al., 1969).
For Cs sorption to kaolinite, the presence of a very low concentration
but high affinity site has been speculated (Cornell, 1993) and 2-site
models have been used to describe Cs sorption (Missana et al., 2014;
Reinoso-Maset and Ly, 2014). However, this second site may also result
from a trace amount of high affinity Cs sorbing mineral interstratified in
the kaolinite (Cornell, 1993; Kim and Cygan, 1996). The absence of
mineral impurities in the KGa-1 kaolinite used here has been discussed
in previous work (Pruett and Webb, 1993) and a 1-site model was suf-
ficient to simulate the sorption behavior of Cs. The Cs-kaolinite sorption
data (Fig. 1) suggest site saturation at Cs concentrations N 10−4 M. Sim-
ilar behavior was also observed by Missana et al. (2014). A second low
affinity site could be used to better fit the 2 highest concentration data
points. However, the CEC of this additional site would need to be
more than an order of magnitude higher and unreasonable when com-
pared to CEC values reported in the literature for kaolinite. Furthermore,
given the large errors associated with the high Cs concentration kaolin-
ite data, the additional modeling complexity was not warranted in our
case.
3.2. Binary desorption
Batch desorption experiments were conducted to determine Cs
sorption reversibility on illite, montmorillonite, and kaolinite. In Fig. 2,
the fraction of sorbed Cs that had desorbed from each mineral after
two weeks is plotted against the logarithm of the equilibrium aqueous
Cs concentration after desorption. In the case of illite, at equilibrium
Cs concentrations below 10− 7 M, b2% of the Cs desorbed after two
weeks. In contrast, at higher Cs concentrations, the percent Cs desorbed
increased until it reached a maximum of 33% for equilibrium Cs concen-
trations of 4 × 10−4 to 6 × 10−5 M. It is apparent that both the sorption
and desorption behavior of Cs is non-linear. The minimal desorption ob-
served at low Cs concentrations may be suggestive of an irreversible
sorption phenomenon. However, the reversibility can only be evaluated
by comparing the observed desorption behavior to predicted desorption
using the equilibrium sorption constants determined in the sorption
experiments.
uted to binding of Cs to FES sites that are susceptible to interlayer col-
lapse causing the fixation of Cs (Hird and Rimmer, 1995). Schulz et al.
(1960) showed that Cs sorbed on illite at low initial Cs concentrations
(b10− 7 M) exhibited “irreversible” sorption and that 10% or less of
the Cs was desorbed from illite even with multiple extractions (Schulz
et al., 1960). In other studies, b 5% of the Cs was desorbed from illite at
concentrations below 10−7 M or less (Ohnuki, 1994; Hsu and Chang,

Fig. 2. Cs batch desorption experiments showing the fraction desorbed for ( ) Na-illite, ( ) Na-kaolinite, and ( ) Na-montmorillonite as a function of Cs concentration in solution (mol/L).
All experiments performed in 0.7 mM NaHCO3, 5 mM NaCl buffer solution at pH 8. The solid solution ratio is 1 g/L for each mineral and 2 σ error bars are propagated from 2% LSC
uncertainties. The solid lines represent the predicted fraction of Cs to desorb by the sorption models.
516 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520

1995; McKinley et al., 2001). However, these observations of low de-
sorption should not be attributed to irreversible sorption since the
high affinity of Cs for FES sites would limit the overall quantities of
desorbed Cs even if these reactions are completely reversible. To test
sorption reversibility, the amount of Cs to desorb from illite was simu-
lated using the same model parameters developed to describe sorption
(Fig. 2). With the exception of the highest Cs concentration samples, the
predicted Cs desorption is in agreement with the data. The ability of the
adsorption parameters to describe the desorption behavior is indicative
of a reversible sorption process at these timescales.
As the equilibrium Cs concentration increases and the FES sites are
filled, subsequent adsorption occurs to the edge and planar sites for
which Cs has a lower affinity. Previous work has shown that at increas-
ing Cs equilibrium concentrations, there is a concomitant increase in the
amount of Cs desorbed from illite ranging from 10 to 50% depending on
factors including; Cs equilibrium concentration, ionic strength and solid
solution ratio (Comans et al., 1991; Bostick et al., 2002). Our sorption
model also predicts that the fraction of Cs to desorb will increase with
equilibrium Cs concentration.
Interestingly, our model prediction diverges from the experimental
data at concentrations N10−5 M. This difference between the model,
which assumes reversible sorption, and experimental data is indicative
of some irreversible sorption behavior at high Cs concentrations. These
results are consistent with the observations of De Koning and Comans
(2004) in which high concentrations of Cs were shown to exacerbate in-
terlayer collapse and, thus, lead to irreversible sorption behavior at high
Cs concentrations. However, it should be noted that our desorption ex-
periments were performed over two weeks. Thus, slow desorption ki-
netics could also play a role in the apparent irreversible behavior at
high Cs concentrations.
In contrast to illite, Cs desorption from montmorillonite is relatively
constant over the entire equilibrium concentration range (Fig. 2).
Approximately 40% of Cs desorbs from montmorillonite under these
conditions. The amount of Cs desorbed from montmorillonite as pre-
dicted by the 1-site sorption model (Fig. 2) is in good agreement with
these desorption data. Thus, sorption reversibility is clearly demonstrat-
ed under these conditions. Approximately 50% of Cs desorbs from kao-
linite across the equilibrium concentration range 10−12–10−5 M and
the 1-site sorption model effectively predicts this behavior. Above
10−5 M, the experimental data and predicted desorption behavior di-
verge with the experimental fraction desorbed decreasing while the
predicted fraction desorbed increases. However, given the large errors
in the experimental fraction desorbed, potential irreversibility at high
Cs concentrations cannot be confirmed with these data.
Sequential batch desorption experiments were conducted to further
examine reversibility of Cs sorption to each mineral. Fig. 3 shows the
distributions coefficients for illite, montmorillonite and kaolinite in
our sequential batch desorption experiments. In Fig. 3-A the Kd values
for Cs desorption from illite are constant at low Cs concentrations
(b10−8 M). The Kd values then decrease until ~10−5 M where they ap-
pear to plateau with increasing Cs concentration. For Cs concentrations
below ~10−5 M the Kd predicted by the model follow the measured de-
sorption Kd values. Above ~10−5 M the predicted desorption Kd values
underestimate the observed values. Furthermore, the second extraction
leads to a larger deviation between predicted and measured Kd values.
The results suggest that Cs sorption to illite at high Cs concentrations
may be partially irreversible, consistent with our Fig. 2 observations
and the findings of De Koning and Comans (2004) who attributed irre-
versible sorption to the collapse of FES sites caused by the presence of
excess Cs cations. However, we cannot exclude kinetics effects in
these relative short-term batch desorption experiments.
Fig. 3-B and C show the Kd values for montmorillonite and kaolinite
after each batch desorption, respectively (aqueous desorbed fraction re-
ported in SI Figs. A6 and A7). The solid lines represent the corresponding

Fig. 3. Cs sequential batch desorption Kd values for (A) Na-illite after ( ) 14 days, ( ) 197 days, (B) Na-montmorillonite after ( ) 14 days, ( ) 142 days, ( ) 209 days, and (C) Na-kaolinite
after ( ) 14 days, ( ) 307 days. All experiments performed in 0.7 mM NaHCO3, 5 mM NaCl buffer solution at pH 8. The solid solution ratio is 1 g/L for each mineral and error bars
propagated from 2% LSC uncertainties are removed for clarity. The sold lines represent predicted Kd values from sorption models.
 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520 517

Kd values predicted by the sorption models for montmorillonite and ka- montmorillonite, which is consistent with its weaker affinity for Cs. Im-
olinite. The montmorillonite desorption Kd values are linear across the portantly, the near complete desorption of Cs from both montmorillon-
range of Cs equilibrium concentrations, are the same for each batch de- ite and kaolinite in these ternary experiments suggests that Cs sorption
sorption period, and correspond well with both the model and experi- to these minerals is not only reversible but that complete desorption
mental sorption behavior from Fig. 1. The similarity in Kd values over will occur in the presence of a competing sorbent and given sufficiently
the three extractions is consistent with reversible sorption of Cs to long timescales (months to years).
montmorillonite at all Cs concentrations. Fig. 3-C shows that kaolinite For the illite experiments (Fig. 4A), the three Cs concentrations used
desorption Kd values are also linear across the range of equilibrium con- were 1 × 10−6, 5 × 10−7, and 1 × 10−7 M and correspond to equilibri-
centrations studied. Moreover, the similarity in Kd values obtained for um Cs concentrations around the non-linear inflection point of the Kd
the two extractions also supports reversible Cs sorption to kaolinite values from Fig. 1. Overall a smaller fraction of the initially sorbed Cs
across all Cs concentrations. is desorbed from illite than from montmorillonite and kaolinite and
the desorption kinetics are exceedingly slow. Also, the amount of Cs
3.3. Ternary desorption desorbed from illite appears to be dependent on the initial Cs concentra-
tion. 52, 44 and 40% of the Cs desorbed from illite by 500 days in the ex-
Long-term two-mineral “ternary” Cs desorption experiments were periments where 1 × 10−6, 5 × 10−7, and 1 × 10−7 M Cs respectively,
carried out to investigate the reversibility and kinetics of Cs desorption was initially added. While equilibrium appears to have been reached
in the presence of an additional clay mineral To ensure that Cs sorption after one year at the highest Cs concentration, it is not clear whether
to illite inside a Float-A-Lyzer was not hindered, four control experi- equilibrium was achieved after 500 days at the two lower
ments were performed. In the control experiments, Cs was added to concentrations.
the background electrolyte solution inside the centrifuge tube and illite Compared to the batch desorption data presented in Fig. 2, the frac-
suspensions were placed in the Float-A-Lyzers. Nearly all of the Cs in so- tion of Cs desorbed is much greater in the ternary experiments. After a
lution was sorbed to the illite in the Float-A-Lyzers (SI Fig. A8) consis- single batch desorption, 50% of the Cs desorbed from the kaolinite sur-
tent with results from batch adsorption experiments. face whereas in the ternary experiments near complete desorption is
For the ternary experiments, mineral suspensions in the centrifuge observed (Fig. 4C). For montmorillonite, 40% of the Cs was desorbed in
tubes were sampled prior to adding the Float-A-Lyzers so that the initial batch experiments compared with 85% in the ternary experiments. Se-
Cs partitioning between the aqueous and mineral phase was known. quential batch desorption can approach the amounts desorbed in the
These initial conditions are reported in Table 2. Of the initial Cs added ternary experiments for low Kd minerals such as montmorillonite and
to each of the samples, 60, 30, and 5% was sorbed to illite, montmoril- kaolinite (70% total Cs desorbed, see appendix Figs. A6 and A7). Howev-
lonite, kaolinite, respectively, prior to the addition of the illite- er, that is not the case for high Kd minerals such as illite.
containing Float-A-Lyzers. Cesium desorption from illite in the ternary experiments was also
After adding the illite-containing Float-A-Lyzers, small aliquots, so as greater than that measured in binary batch desorption experiments.
not to significantly change the total amount of mineral, were taken pe- After a single binary desorption 6, 2 and b 1% of the Cs desorbed from
riodically from the outer mineral suspensions and used to measure the the illite surface for initial Cs concentrations of 10− 6, 5 × 10− 7 and
total amount of Cs in the suspension and the supernatant. The amount 10−7 M respectively. Sequential binary batch desorption resulted in
of Cs desorbed from the clays in the suspension was calculated from similarly low Cs desorption (total Cs desorbed of 10, 4 and 2%, see Ap-
the measured suspension and supernatant Cs concentrations. The pendix Figs. A6 and A7). For the same concentrations in the ternary ex-
amount of Cs sorbed to the illite in the Float-A-Lyzer was then calculated periments, 52, 44 and 40%, respectively, of the Cs desorbed.
according to Eq. (2) and assuming no loss of Cs to container walls. Fig. 4 A higher desorbed fraction of Cs in our ternary experiments com-
is a plot of the Cs fraction desorbed as a function of time for the suspen- pared to binary desorption experiments is consistent with previous re-
sions of illite, montmorillonite, and kaolinite (A, B and C, respectively). sults using ammonium hexacyanoferrate inside a dialysis membrane
For all experiments with montmorillonite and kaolinite (Fig. 4B and C, (De Koning and Comans, 2004). De Koning and Comans (2004) posit
respectively), the majority of the initially sorbed Cs was desorbed that Cs is reversibly sorbed to illite in specific conditions including low
from the mineral surface by the end of the experimental period regard- cation concentrations and a short equilibration time of 5 min. However,
less of the initial Cs concentration. However, equilibrium was achieved if cation concentrations surpass a threshold then even a 5-min equili-
only after several months (kaolinite) or over a year (montmorillonite). bration time will result in partial irreversibility of Cs (De Koning and
491 days after the Float-A-Lyzers were added, an average of 85% of the Comans, 2004). The most likely environmental scenario is a longer
Cs desorbed from the surface of montmorillonite and an average of equilibration time where some Cs diffuses into the interlayer and be-
97% of the Cs desorbed from the surface of kaolinite. Desorption of Cs comes fixed (Fuller et al., 2015).
appears to have occurred more quickly from kaolinite than from To quantitatively test for sorption reversibility in the ternary exper-
iments, Kd values for the ternary experiments were predicted from the
binary sorption models. The predicted Kd values for the ternary experi-
Table 2 ments were determined by combining the binary sorption models for il-
Initial conditions for each of the ternary experiments prior to Float-A-Lyzer addition.
lite and the mineral outside the Float-A-Lyzer. The model determined
Mineral Suspension [Cs]i [Cs]eq [Cs]solid Cs sorbed the equilibrium solution Cs concentration and the total Cs concentration
percent expected on each mineral surface at equilibrium, from which a total
mol/L
mineral Kd value was determined. The predicted amount of Cs desorbed
montmorillonite 5.0 × 10−7 3.5 × 10−7 1.4 × 10−7 28
montmorillonite 1.0 × 10−7 6.9 × 10−8 2.9 × 10−8 29
from the mineral outside the Float-A-Lyzer was determined by compar-
montmorillonite 5.0 × 10−8 3.5 × 10−8 1.4 × 10−8 28 ing the final predicted Cs concentration on that mineral to the initial
kaolinite 5.0 × 10−7 4.7 × 10−7 2.2 × 10−7 4.4 predicted Cs concentration on that mineral (i.e. Table 2). The predicted
kaolinite 1.0 × 10−7 9.4 × 10−8 5.0 × 10−9 5.0 log(Kd) values, percent desorbed and equilibrium aqueous Cs concen-
illite 1.0 × 10−6 4.6 × 10−7 5.5 × 10−7 55
trations for each case are presented in Table 3 along with the experi-
illite 5.0 × 10−7 1.9 × 10−7 3.0 × 10−7 60
illite 1.0 × 10−7 3.6 × 10−8 6.1 × 10−8 61 mental values.
none 1.0 × 10−6 9.8 × 10−7 0 0 For the ternary montmorillonite experiments, the predicted Kd and
none 5.0 × 10−7 5.0 × 10−7 0 0 percent desorption values were higher than the experimentally deter-
none 1.0 × 10−7 1.0 × 10−7 0 0 mined values. However, the differences are not statistically significant.
none 5.0 × 10−8 5.0 × 10−8 0 0
Nearly all of the Cs was predicted to desorb from the montmorillonite
518 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520

Fig. 4. The fraction of Cs desorbed from the (A) illite, (B) montmorillonite and (C) kaolinite mineral suspensions outside the dialysis membrane over time is shown. Red, blue, orange and
purple colors represent initial [Cs] of 10−6, 5 × 10−7, 10−7 and 5 × 10−8 M, respectively. Error bars based on propagation of 2 s % LSC uncertainties are removed for clarity.

and the experimental and modeling results confirm sorption
reversibility.
For the ternary kaolinite experiments, the predicted Kd and percent
desorption values were in agreement with the experimentally deter-
mined values. The large errors associated with the kaolinite values re-
sult from the small amount of Cs initially sorbed to kaolinite before
the Float-A-Lyzer was added. Thus, even small variations in the fraction
of Cs to desorb from the kaolinite can lead to large errors when propa-
gated to the calculated Kd and percent desorbed values. Despite the
large variance, the predicted and experimental values indicate that Cs
is completely desorbed from the kaolinite and this is consistent with re-
versible sorption of Cs on kaolinite.
Unlike montmorillonite and kaolinite, the predicted Kd and percent
desorption values for the illite ternary experiments do not agree with
the experimental values for each concentration. At the highest initial
Cs concentration (10−6 M Cs) the predicted and experimental values
are statistically equivalent. This is in contrast to the batch high concen-
tration sorption/desorption experiments and suggests that slow de-
sorption kinetics can lead to misidentification of irreversibility at these
concentrations.
As the initial Cs concentration decreases, the predicted and experi-
mental percent Cs desorbed values diverge and the difference between
the predicted and experimental Kd becomes statistically significant.
However, the data in Fig. 4 also suggest that equilibrium may not have
been reach even after 500 days. Thus, Cs desorption from illite is simply
kinetically limited. This apparent irreversibility appears to contradict
the earlier results of the batch desorption experiments that suggested
reversible sorption at these Cs concentration (Fig. 3). However, in the
batch desorption experiments, a much smaller fraction of Cs desorbed
from illite. Thus, batch desorption experiments only accessed a small

Table 3
The measured and predicted values for aqueous Cs, Kd, and percent Cs desorbed from the mineral outside the Float-A-Lyzer. The uncertainties represent two standard deviations. For the
experimental results this error is propagated from the counting error. For the predicted values, the uncertainty was determined by accounting for the uncertainty in the fitted selectivity
coefficients (Table 1).

Outside mineral Experimental Predicted

[Cs]i log[Cs]eq log(Kd) Desorbed log[Cs]eq log(Kd) Desorbed

mol/L mol/L mL/g percent mol/L mL/g percent

Montmorillonite 5.0 × 10−7 −7.5 ± 0.01 3.7 ± 0.1 84 ± 7 −8.3 ± 0.1 4.0 ± 0.1 98 ± 10
Montmorillonite 1.0 × 10−7 −8.3 ± 0.01 3.8 ± 0.1 86 ± 7 −9.1 ± 0.1 4.2 ± 0.1 99 ± 8
Montmorillonite 5.0 × 10−8 −8.8 ± 0.01 4.0 ± 0.1 88 ± 7 −9.4 ± 0.1 4.2 ± 0.1 99 ± 7
Kaolinite 5.0 × 10−7 −8.0 ± 0.01 4.2 ± 0.1 99 ± 23 −8.3 ± 0.1 4.0 ± 0.1 99 ± 10
Kaolinite 1.0 × 10−7 −8.5 ± 0.01 4.0 ± 0.1 94 ± 20 −9.1 ± 0.1 4.2 ± 0.1 99 ± 10
Illite 1.0 × 10−6 −7.5 ± 0.01 3.9 ± 0.1 52 ± 4 −8.0 ± 0.1 4.1 ± 0.1 63 ± 16
Illite 5.0 × 10−7 −7.9 ± 0.01 4.1 ± 0.1 44 ± 4 −8.5 ± 0.1 4.2 ± 0.1 65 ± 13
Illite 1.0 × 10−7 −8.7 ± 0.01 4.2 ± 0.1 40 ± 4 −9.3 ± 0.1 4.3 ± 0.1 66 ± 12
 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520
fraction of the most labile Cs component on the surface. These experi-
mental results indicate that under these experimental conditions and
over long time periods significantly more Cs can desorb from illite
than the few percent that was initially determined from batch experi-
ments over timeframes of weeks. While the ternary experiments sug-
gest the possible existence of irreversible Cs sorption to illite, the more
significant results of these experiments is that desorption kinetics on il-
lite are extremely slow (months to years).
4. Conclusions
Batch sorption experiments demonstrate nonlinear sorption of Cs to
illite and linear sorption of Cs to montmorillonite and kaolinite.
Fourteen-day batch desorption experiments, when combined with
equilibrium sorption models, show that Cs desorption from these min-
erals appears reversible except in the case of high concentration Cs on
illite. These batch experiments provided the foundation to analyze the
results of ternary experiments.
A methodology was developed using permeable dialysis bags to
investigate the reversibility of Cs sorption to the minerals illite, mont-
morillonite and kaolinite. Predictions of the ternary experiment equilib-
rium were made using the sorption models developed from the binary
experiments and then compared to the measured data from the ternary
experiments. This comparison shows that Cs is reversibly sorbed to
montmorillonite and kaolinite. The addition of a mineral with strong
affinity for Cs such as illite was sufficient to drive the nearly complete
desorption of Cs from montmorillonite and kaolinite over a timescale
of months. However, the amount of Cs desorbed from illite was less
that predicted by our reversible sorption model. This appears to suggest
some irreversible sorption of Cs to illite, likely caused by fixation of Cs in
the interlayer due to the collapse of the FES sites. However, in some ex-
periments, equilibrium was not achieved in the ternary experiments
even after 500 days. Thus, desorption kinetics on illite are extremely
slow (months to years) and investigations into sorption reversibility
should be performed on the scale of years to distinguish between
kinetically-limited and irreversible sorption behavior.
Acknowledgements
This work was supported by the Subsurface Biogeochemical Re-
search Program of the U.S. Department of Energy's Office of Biological
and Environmental Research. Prepared by LLNL under Contract DE-
AC52-07NA27344. This material is based upon work supported by the
U.S. Department of Homeland Security under Grant Award Number,
2012-DN-130-NF0001. The views and conclusions contained in this
document are those of the authors and should not be interpreted as
representing the official policies, either expressed or implied, of the
U.S. Department of Homeland Security.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.scitotenv.2017.08.122.
References
Benedicto, A., Missana, T., Fernandez, A.M., 2014. Inter layer collapse affects on caesium
adsorption onto illite. Environ. Sci. Technol. 48 (9), 4909–4915.
Bolt, G.H., Sumner, M.E., Kamphorst, A., 1963. A study of equilibria between three catego-
ries of potassium in an illitic soil. Soil Sci. Soc. Am. Proc. 27 (3), 294–299.
Bostick, B.C., Vairavamurthy, M.A., Karthikeyan, K.G., Chorover, J., 2002. Cesium adsorp-
tion on clay minerals: an EXAFS spectroscopic investigation. Environ. Sci. Technol.
36 (12), 2670–2676.
Bradbury, M.H., Baeyens, B., 2000. A generalised sorption model for the concentration de-
pendent uptake of caesium by argillaceous rocks. J. Contam. Hydrol. 42, 141–163.
Brouwer, E., Baeyens, B., Maes, A., Cremers, A., 1983. Cesium and rubidium ion equilibria
in Illite clay. J. Phys. Chem. 87, 1213–1219.
519
Castrillejo, M., Casacuberta, N., Breier, C.F., Pike, S.M., Masque, P., Buesseler, K.O., 2016. Re-
assessment of Sr, Cs and Cs in the coast off Japan derived from the Fukushima Dai-ichi
nuclear accident. Environ. Sci. Technol. 50 (1), 173–180.
Comans, R.N., Hockley, D.E., 1992. Kinetics of cesium sorption on illite. Geochim.
Cosmochim. Acta 56 (3), 1157–1164.
Comans, R.N., Haller, M., DePreter, P., 1991. Sorption of cesium on Illite: non-equilibrium
behaviour and reversibility. Geochim. Cosmochim. Acta 55, 433–440.
Cornell, R.M., 1993. Adsorption of cesium on minerals: a review. J. Radioanal. Nucl. Chem.
Artic. 171 (2), 483–500.
Davis, J.A., Coston, J.A., Kent, D.B., Fuller, C.C., 1998. Application of the surface complexa-
tion concept to complex mineral assemblages. Environ. Sci. Technol. 32, 2820–2828.
De Koning, A., Comans, R.N., 2004. Reversibility of radiocaesium sorption on illite.
Geochim. Cosmochim. Acta 68 (13), 2815–2823.
Erten, H.N., Aksoyoglu, S., Hatipoglu, S., 1988. Sorption of cesium and strontium on mont-
morillonite and kaolinite. Radiochim. Acta 44 (5), 147–151.
Fukushi, K., Fukiage, T., 2015. Prediction of intrinsic cesium desorption from Na-smectite
in mixed cation solutions. Environ. Sci. Technol. 49, 10398–10404.
Fukushi, K., Sakai, H., Itono, T., Tamura, A., Arai, S., 2014. Desorption of intrinsic cesium
from smectite: inhibitive effects of clay partical organization on cesium desorption.
Environ. Sci. Technol. 48, 10743–10749.
Fuller, A.J., Shaw, S., Peacock, C.L., Trivedi, D., Small, J.S., Abrahamsen, L.G., Burke, I.T., 2014.
Ionic strength and pH dependent multi-site sorption of Cs onto a micaceous aquifer
sediment. Appl. Geochem. 40, 32–42.
Fuller, A.J., Shaw, S., Ward, M.B., Haigh, S.J., Mosselmans, J.F., Peacock, C.L., ... Burke, I.T.,
2015. Caesium incorporation and retention in illite interlayers. Appl. Clay Sci. 108,
128–134.
Gast, R.G., Van Bladel, R., Deshpande, K.B., 1969. Standard heats and entropies of exchange
for alkali metal cations on Wyoming bentonite. America Proceedings]–>Soil Sci. Soc.
Am. Proc. 33 (5), 661–664.
Hird, A.B., Rimmer, D.L., 1995. Total caesium-fixing potentials of acid organic soils.
J. Environ. Radioact. 26, 103–118.
Hsu, C.N., Chang, K.P., 1995. Study of factors dominating sorption-desorption of Cs in soil.
Radiochim. Acta 68 (2), 129–133.
Iijima, K., Tomura, T., Shoji, Y., 2010. Reversibility and modeling of adsorption behavior of
cesium ions on colloidal montmorillonite particles. Appl. Clay Sci. 49, 262–268.
Jackson, M.L., 1963. Interlayering of expansible layer silicates in soils by chemical
weathering. Clay Clay Miner. 11, 29–46.
Kim, Y.K., Cygan, R.T., 1996. Cs-133 NMR study of cesium on the surfaces of kaolinite and
illite. Geochim. Cosmochim. Acta 60 (21), 4059–4074.
Kim, Y., Cygan, R.T., Kirkpatrick, R.J., 1996. 133Cs NMR and XPS investigation of cesium
adsorbed on clay minerals and related phases. Geochim. Cosmochim. Acta 60 (6),
1041–1052.
Klement, A.W., 1965. Radioactive fallout phenomena and mechanisms. Health Phys. 11,
1265–1274.
Komarneni, S., Roy, D.M., 1979. Shale as a radioactive-waste respoitory-importance of
vermiculite. J. Inorg. Nucl. Chem. 41 (12), 1793–1796.
Komarneni, S., Roy, D.M., 1980. Hydrothermal effects of cesium sorption and fixation by
clay-minerals and shales. Clay Clay Miner. 28 (2), 142–148.
McKinley, J.P., Zeissler, C.J., Zachara, J.M., Serne, R.J., Lindstrom, R.M., Schaef, H.T., Orr, R.D.,
2001. Distribution and retention of Cs-137 in sediments at the Hanford site,
Washington. Environ. Sci. Technol. 35 (17), 3433–3441.
Missana, T., Garica-Gutierrez, M., Alonso, U., 2004. Kinetics and irreversibility of cesium
and uranium sorption onto bentonite colloids in a deep granitic environment. Appl.
Clay Sci. 26 (1–4), 137–150.
Missana, T., Garcia-Gutierrez, M., Benedicto, A., Ayora, C., De-Pourcq, K., 2014. Modelling
of Cs sorption in natural mixed-clays and the effects of ion competition. Appl.
Geochem. 49, 95–102.
Novikov, A.P., Pavlotskaya, F.I., Goryachenkova, T.A., Posokhov, A.K., Kazinskaya, I.E.,
Emel'yanov, V.V., ... Myasoedov, B.F., 1998. Radionuclide content in underground wa-
ters and rock from observation wells around Lake Karachai. Radiochemistry 40 (5),
484–490.
Ohnuki, T., 1994. Sorption characteristics of cesium on sandy soils and their components.
Radiochim. Acta 65, 75–80.
Ohnuki, T., Kozai, N., 2013. Adsorption behavior of radioactive cesium by non-mica min-
erals. J. Nucl. Sci. Technol. 50 (4), 369–375.
van Olphen, H., Fripiat, J.J. (Eds.), 1979. Data Handbook for Clay Minerals and Other Non-
metallic Minerals. Pergamon Press, New York.
Poinssot, C., Baeyens, B., Bradbury, M.H., 1999. Experimental and modelling studies of cae-
sium sorption on illite. Geochim. Cosmochim. Acta 63 (19–20), 3217–3227.
Pruett, R.J., Webb, H.L., 1993. Sampling and analysis of KGa-1B well-crystallized kaolin
source clay. Clay Clay Miner. 41 (4), 514–519.
Reinoso-Maset, E., Ly, J., 2014. Study of major ions sorption equilibria to characterize the
ion exchange properties of kaolinite. J. Chem. Eng. Data 59, 4000–4009.
Rich, C.I., Black, W.R., 1964. Potassium exchange as affected by cation size, pH, and min-
eral structure. Soil Sci. 97 (6), 384–390.
Sawhney, B.L., 1970. Potassium and cesium ion selectivity in relation to clay mineral
structure. Clay Clay Miner. 18, 47–52.
Sawhney, B.L., 1972. Selective sorption and fixation of cations by clay minerals: a review.
Clay Clay Miner. 20, 93–100.
Schulz, R.K., Overstreet, R., Barshad, I., 1960. On the soil chemistry of cesium 137. Soil Sci.
89 (1), 16–27.
Shahwan, T., Erten, H.N., Black, L., Allen, G.C., 1999. TO-SIMS study of Cs+ sorption on
natural kaolinite. Sci. Total Environ. 226 (2–3), 255–260.
Staunton, S., Roubaud, M., 1997. Adsorption of Cs-137 on montmorillonite and illite: ef-
fect of charge compensating cation, ionic strength, concentration of Cs, K and fulvic
acid. Clay Clay Miner. 45 (2), 251–260.
520 C.B. Durrant et al. / Science of the Total Environment 610–611 (2018) 511–520
Steinhauser, G., Niisoe, T., Harada, K.H., Shozugawa, K., Schnieder, S., Synal, H.-A., ...
Koizumi, A., 2015. Post-accident sporadic releases of airborne radionuclides from
the Fukushima Daiichi nuclear power plant site. Environ. Sci. Technol. 49 (24),
14028–14035.
Swartzen-Allen, S.L., Matijevic, E., 1974. Surface and colloid chemistry of clays. Chem. Rev.
74 (3), 385–400.
Turner, N.B., Ryan, J.N., Saiers, J.E., 2006. Effect of desorption kinetics on colloid-facilitated
transport of contaminants: cesium, strontium, and illite colloids. Water Resour. Res.
42 (12), W12S09.
United States Environment Protection Agency, 2009. National Primary Drinking Water
Regulations EPA 816-F-09-004. US EPA.
Vanselow, A.P., 1932. Equilibria of the base-exchange reaction of bentonites, permutites,
soil coloids and zeolites. Soil Sci. 33 (2), 95–114.
Zachara, J.M., Smith, S.C., Liu, C., Mckinley, J.P., Serne, R.J., Gassman, P.L., 2002. Sorption of
Cs+ to micaceous subsurface sediments from the Hanford site, USA. Geochim.
Cosmochim. Acta 66 (2), 193–211.
Zavarin, M., Roberts, S.K., Hakem, N., Sawvel, A.M., Kersting, A.B., 2005. Eu(III), Sm(III),
Np(V), Pu(V), and Pu(IV) sorption to calcite. Radiochim. Acta 93, 93–102.
Zavarin, M., Powell, B.A., Bourbin, M., Zhao, P.H., Kersting, A.B., 2012. Np(V) and Pu(V) ion
exchange and surface-mediated reduction mechanisms on montmorillonite. Environ.
Sci. Technol. 46 (5), 2692–2698.
Zhuang, J., Flury, M., Jin, Y., 2003. Colloid-facilitated Cs transport through water-saturated
Hanford sediment and Ottawa sand. Environ. Sci. Technol. 37 (21), 4905–4911.