Journal of Nuclear Materials 468 (2016) 84e96

Contents lists available at ScienceDirect

Journal of Nuclear Materials

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

Americium and plutonium association with magnesium hydroxide

colloids in alkaline nuclear industry process environments
Zoe Maher a, Peter Ivanov b, Luke O'Brien c, Howard Sims d, Robin J. Taylor a,
Sarah L. Heath e, Francis R. Livens e, David Goddard f, Simon Kellet g, Peter Rand g,
Nick D. Bryan c, *
a
National Nuclear Laboratory, Sellaeld, Cumbria, CA20 1PG, United Kingdom
b
The National Physical Laboratory, Hampton Road, Teddington, Middlesex, TW11 0LW, United Kingdom
c
National Nuclear Laboratory, Chadwick House, Birchwood, Warrington, WA3 6AE, United Kingdom
d
National Nuclear Laboratory, Harwell Business Centre, Didcot OX11 0QJ, United Kingdom
e
The Centre for Radiochemistry Research, School of Chemistry, The University of Manchester, Oxford Road, Manchester, M13 9PL, United Kingdom
f
National Nuclear Laboratory, Preston, Lancashire, PR4 0XJ, United Kingdom
g
Sellaeld Ltd, Seascale, Cumbria CA20 1PG, United Kingdom

h i g h l i g h t s

Am and Pu in Mg-rich alkaline efuents from nuclear fuel storage ponds are mainly associated with colloids.

In true solution, i.e. material passing through a 1 nm lter, the plutonium appears to be in oxidation state V.

In model brucite (magnesium hydroxide) colloid systems, Am(III) is preferentially associated with larger colloids.

At pH 10.4, carbonate concentrations >102 M are able to prevent the binding of Am(III) to colloids.

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

Article history: The behaviours of Pu, Am and colloids in feed solutions to the Site Ion-exchange Efuent Plant (SIXEP) at
Received 27 May 2015 the Sellaeld nuclear reprocessing site in the U.K. have been studied. For both Pu and Am, fractions were
Received in revised form found to be associated with material in the colloidal size range, with ~50% of the Pu in the range 1
31 October 2015
e200 nm. The concentration of soluble Pu (<1 nm) was ~1 nM, which is very similar to the solubility
Accepted 6 November 2015
Available online 17 November 2015
limit for Pu(V). The soluble Am concentration was of the order of 1011 M, which was below the solubility
limit of americium hydroxide. The size, morphology and elemental composition of the particulates and
colloids in the feed solutions were investigated. Magnesium is homogeneously distributed throughout
Keywords:
Brucite
the particles, whereas U, Si, Fe, and Ca were present in localised areas only. Amongst some heterogeneous
Colloids material, particles were identied that were consistent with hydrotalcite. The distribution of 241Am(III)
Americium on brucite (magnesium hydroxide) colloids of different sizes was studied under alkaline conditions
Plutonium representative of nuclear fuel storage pond and efuent feed solution conditions. The morphology of the
Efuent treatment brucite particles in the bulk material observed by ESEM was predominantly hexagonal, while that of the
carbonated brucite consisted of hexagonal species mixed with platelets. The association of 241Am(III)
with the brucite colloids was studied by ultraltration coupled with gamma ray-spectrometry. For car-
bonate concentrations up to 103 M, the 241Am(III) was mainly associated with larger colloids
(>300 kDa), and there was a shift from the smaller size fractions to the larger over a period of 6 months.
At higher carbonate concentrations (102 M), the Am was predominantly detected in the true solution
fraction (<3 kDa) and in smaller size colloidal fractions, in the range 3e100 kDa.
2015 Elsevier B.V. All rights reserved.

1. Introduction

* Corresponding author. The retrieval and processing of signicant volumes of legacy

E-mail address: nick.bryan@nnl.co.uk (N.D. Bryan).

http://dx.doi.org/10.1016/j.jnucmat.2015.11.010
0022-3115/ 2015 Elsevier B.V. All rights reserved.
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96
waste into forms suitable for storage and disposal are among the
major challenges in the UK's nuclear clean-up and decom-
missioning programme. These wastes include large amounts of
sludge that have arisen from the corrosion of spent nuclear fuel
cladding as well as associated nuclear fuel storage pond liquors. It
has been estimated that at Sellaeld alone several thousand cubic
metres of sludges containing 104 TBq of a-particle emitters have to
be processed [1,2]. The Magnox spent fuel elements, which were
used in the UK's rst generation nuclear reactors, comprise a nat-
ural uranium metal core, clad by a magnesium-rich alloy. Spent
nuclear fuel has been wet stored in dedicated ponds for several
decades, and during this period, corrosion of the cladding and fuel
has occurred. Several studies on the corrosion of magnesium and
magnesium-rich alloys in water have been published, and these
indicate that the initial corrosion product is primarily brucite
(Mg(OH)2), formed by the hydration of the MgO layer on the surface
of the magnesium metal in the presence of moisture and air [3e5].
Further, it has been shown that colloidal species form in a solution
in contact with bulk brucite [6].
Liquid wastes arising from fuel storage ponds and decanning
operations typically contain signicant amounts of 137Cs, 90Sr and
magnesium hydroxide (from the corrosion of fuel cladding in the
ponds). At Sellaeld, such wastes are treated in the Site Ion-
exchange Efuent Plant (SIXEP). Feed solutions to SIXEP that may
contain solids are allowed to settle in Bulk Storage Tanks (BST) for
times between 10 h and a week, after which the liquor is decanted
off and fed into SIXEP. In the plant, the radionuclide contaminants
are removed by passing the solution through sand bed lters, fol-
lowed by a carbonation step that reduces the pH from approxi-
mately 11 to neutral. The carbonated solution then passes through
ion-exchange clinoptilolite (zeolite) beds, which are selected to
remove 137Cs and 90Sr. Gregson et al. [1,7] studied the chemical
form of sludges and suspended particles and colloids from an
alkaline storage pond at Sellaeld. They found that the morphology
of the magnesium-containing phases was consistent with brucite,
but perhaps with some incorporation of Al to form hydrotalcite-like
phases.
Colloids are known as potential controllers of radionuclide
speciation and transport due to their small size and high surface
area to mass ratio [8,9]. They are capable of binding radionuclides,
particularly the actinides, in forms that prevent treatment by the
efuent treatment technologies routinely used in the nuclear in-
dustry, such as sand bed ltration and ion exchange. Magnesium
hydroxide particles have been reported to retain various radionu-
clides efciently [10e12]. Therefore, the physical and chemical
nature of the solution-borne colloids and their interactions with
radionuclides need to be better understood, in order to evaluate the
likely challenge to the efuent treatment process and to design
improvements [1].
Spent Magnox fuel has numerous radioactive components.
However, following irradiation, the majority of fuel by mass will
still be uranium. There will also be a range of ssion products and
minor actinides formed by neutron capture reactions. In the efu-
ents from the storage ponds and silos, the most signicant ssion
products are 137Cs and 90Sr, because of their relatively high solu-
bilities and the high radiotoxicity of 90Sr (they are found exclusively
in oxidation states (I) and (II), respectively). Plutonium and amer-
icium are signicant species in spent nuclear fuel, because they are
formed in signicant quantities and have high radiotoxicity. In the
storage ponds and during efuent treatment, Am would be ex-
pected to exist in oxidation state III only, whilst Pu is expected to
exist in oxidation states IV and V.
The sorption of Am(III) on various colloids has been investigated
in several previous studies [13e16], and it has been shown to
associate strongly with colloids. For example, Degueldre et al.
85
studied the sorption of 241Am (3  1011 e 3  1010 M) onto marl
groundwater colloids, with a distribution ratio in the range of
1e3  106 ml g1 [15]. The distribution of Am soluble species was
calculated, and approximately 90% of the species were found to be
negatively charged. The strong sorption observed onto the nega-
tively charged clay colloids was considered to be due to the for-
mation of a ternary surface complex. Pitois et al. carried out a study
of the interaction of 152Eu(III), which was considered as a model for
Am(III) behaviour in nuclear fuel storage pond conditions, with
brucite bulk solid and colloids at pH 10.4 [6]. In the absence of
carbonate, the main 152Eu(III) activity was found to be associated
with the bulk solid phase and with colloidal material when the bulk
phase was removed. Sorption was found to decrease signicantly at
a carbonate concentration of 102 M.
The redox chemistry of Pu is much more complex than that of
Am, with several different oxidation states that may coexist,
although in the systems studied here, Pu(V) and Pu(IV) would be
expected to dominate the speciation [1]. Since Pu(V) exists in
aqueous solution as the PuO 2 ion, it is less prone to hydrolysis than
Pu(IV). Pu(IV) sorption on brucite bulk and colloids has been shown
to occur rapidly and completely in alkaline conditions from pH 8 to
pH 13, and the interaction is sufciently strong that it is not affected
by the presence of 103 M citrate (a competing complexing ligand).
Additionally, X-ray photoelectron spectroscopy (XPS) results indi-
cate that Pu is incorporated into the brucite subsurface at sub-
stantial concentrations and to a depth of at least 50 nm [10,11].
Strontium uptake by naturally occurring and thermally modied
brucite has been investigated and high sorption capacity and
relatively fast kinetics have been reported, particularly at alkaline
pH [12]. Parry et al. studied plutonium behaviour in a model
Magnox storage pond liquor containing Pu(IV), corroded Magnox
sludge (CMS) simulant, sodium carbonate, polyelectrolyte, and
silica [17]. The CMS particles were primarily composed of platy
brucite crystals, with a smaller quantity of acicular artinite crystals.
In the absence of CMS, 100% of Pu was able to pass a 100 nm lter,
but as CMS was added, the Pu was increasingly found in the
>200 nm fraction, and for CMS concentrations >10 ppm, virtually
all of the Pu was removed by a 200 nm lter. At lower CMS con-
centrations, the results depended on pH. For example, at 2 ppm of
CMS, approximately 25%, 60% and 90% of Pu was in the >200 nm
fraction at pH 10.8, 11 and 11.5, respectively. The presence of
carbonate ions (3  102 M) gave the greatest effect on the system,
allowing the majority of Pu to remain in the solution.
In the past, Pu(IV) was presumed as the dominant form of
plutonium in the ponds and silos at Sellaeld [17]. However,
Gregson et al. [1] used analysis of storage pond water samples to
show that in certain parts of the pond, the concentrations are better
described assuming that Pu(V) dominates, but in other parts, the
data were consistent with either Pu(IV) or Pu(V). Although for most
samples, the amount in true solution (<1 nm) was less than 1%, for
some samples up to 6.8% was observed, which strongly suggested
that the Pu was in oxidation state V. For all samples and locations in
the pond, they found that most of the Pu (>90%) was associated
with large particulates (>200 nm), but there was relatively little Pu
in the size range of 1e200 nm, which was interpreted as an absence
of intrinsic Pu(IV) oxyhydroxide polymeric or colloidal species.
The rst aim of this study was to characterise the colloids in the
feed solutions to SIXEP and to study the distribution of Pu and Am
in the solutions. The second aim was to study actinide behaviour in
a model system (brucite) that is relevant to the SIXEP feed solutions
(and the Magnox nuclear fuel storage ponds and silos). Although
the SIXEP plant was designed primarily to remove Cs and Sr from
efuents, it is also quite effective at removing a-activity in partic-
ulate form. Hence, another aim was to attempt to explain this
behaviour better. The association of 241Am species with various size
86 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96
brucite colloids under alkaline conditions has been studied as a
model system of the pond and efuent stream chemistry. The effect
of carbonate on Am sorption and distribution were also charac-
terised. The results contribute to a better understanding of actinide
chemistry in spent nuclear fuel storage ponds and efuent treat-
ment. They also provide data relevant to nuclear legacy waste
treatment (e.g., sludge retrieval and processing).
2. Experimental
All reagents used in the experiments were analytical grade,
apart from the SIXEP samples themselves.
2.1. Storage tank samples
Plant liquor samples from Sellaeld site were characterised to
understand the challenge facing SIXEP, both in terms of the pres-
ence and nature of colloidal material and the extent of actinide
association with them. Two samples of plant liquor were obtained
from the SIXEP Bulk Storage Tanks (BST) in which liquor is stored
prior to being treated. The feeds to these tanks include efuents
from the Magnox fuel storage ponds and silos. These samples
represent a typical feed to SIXEP and have a pH around 11e11.5. The
volume of liquor that could be obtained for study was limited, as
were the procedures that could be applied to it, due to dose re-
strictions and safety case requirements for both the transport and
handling of these materials. In particular, it was not possible to
measure Mg concentrations in these samples. Because of these
problems, this is the rst time that the characterisation of the
colloids in the BST liquor or SIXEP feed solutions has been reported
in the literature. Approximately 200 ml of liquor were obtained.
The pH of the samples prior to analysis was in the range 9.6e9.9: a
reduction in sample pH had occurred during sample bulking and
storage. This behaviour has been observed previously. For example,
Gregson et al. [1,7] also reported a reduction in pH to 8.5e10. This
may be attributed to the dissolution of carbon dioxide into the li-
quor during both sampling (air-lift method) and subsequent
handling of the sample in the open atmosphere at the plant during
bulking and analysis. In the air lift sampling method, a sample tube
is inserted into the plant liquor. The tube is attached to a vessel that
has a pressure lower than atmospheric. When the valve between
the vessel and the sampling tube is opened, the liquor ows into
the vessel. There will still be residual air in the vessel, and so there
is the potential for dissolution of CO2 into the sample as it is taken.
Following collection, the sample had to be kept under conditions
where the atmosphere could not be controlled before it was
transferred to the laboratory. It is impossible to obtain plant sam-
ples that have not been compromised in this way due to handling
restrictions. Currently, these are the only samples available for
studying the properties of plant liquors in detail. Given the pH
range of the feed solutions to SIXEP (11e11.5) and the nal pH range
of the samples, speciation calculations assuming that the change in
pH was caused by the dissolution of CO2 and using the thermo-
dynamic speciation code PHREEQC [18] (ThermoChimie (SIT)
database V8.0 September 2011) predict that the carbonate con-
centration that could have been introduced to the samples during
collection was in the range 7.3  104 e 2.8  103 M. Gregson et al.
[1,7] estimated that the carbonate in their pond samples due to
ingress of CO2 was in the range 2e11  103 M. They also measured
total inorganic carbon concentrations, which were found to be of
the order of 50 ppm (8.3  104 M) for samples taken directly from
the pond and in the range 40e100 ppm (6.7  104 e 1.7  103 M)
following pH drift. The SIXEP storage tanks are different to the
ponds, since they take feeds from other parts of the plant, and so it
is possible that their carbonate concentrations are different.
However, given the PHREEQC calculations based on the pH change
and previous work, it seems likely that the carbonate concentra-
tions in the samples studied here are in the range 5  104 e
102 M.
The plutonium and americium isotopes that are present in the
plant samples were separated from the mixture using solvent
extraction prior to analysis. This was necessary because the other
isotopes would interfere in the alpha-spectrometer. The method
uses tri-iso-octylamine (TIOA) to extract the Pu, (oxidised to VI
with hydrogen peroxide) into the organic phase. The Pu fraction
was back extracted from the organic using NH4I/8 M HCl. The Am is
extracted from the original aqueous phase using di-n-butyl-di-
ethyl-carbamoyl-phosphonate (DDCP) in 12 M nitric acid and
then back extracted using toluene/2 M nitric acid. The separated
fractions were evaporated to dryness and then dissolved in
concentrated sulphuric acid. The resulting solution was electro-
deposited at 10 V, 1.1 A to x the actinides on a steel planchet,
which acts as the cathode. The planchets were counted using an
Ortec Octete Plus alpha spectrometer with ion-implanted silicon
detectors. The data were collected and analysed using Gammavi-
sion software. The typical count times were 200,000 s.
2.2. Brucite model colloid systems
The surface area of the solid brucite sample used in the gener-
ation of colloids was determined as 6.47 m2 g1 by nitrogen gas
adsorption using the BET method on a Tristar 3000 Micrometrics
Surface Area Analyser. Brucite suspensions were prepared in a
glove box under an inert atmosphere (CO2-free nitrogen) to prevent
the dissolution of atmospheric carbon dioxide. The solutions con-
taining colloidal brucite were prepared in screw cap polypropylene
containers using magnesium hydroxide powder with a purity of
99% (Fluka Chemicals). To prepare the samples, an amount of
magnesium hydroxide was added to 18 MU deionised water (Milli-
Q system) to give a concentration of 1 g l1. The pH was set to 10.4
by CO2-free 0.1 M NaOH solution. Sufcient NaOH was added to
compensate for the addition of the tracer (see below). This pH was
adopted for the experiments, since it represents a point in the
middle of the pH range between the pH in the BSTs and that of the
collected samples. Samples of the colloidal phase were separated
from the bulk brucite when steady state was achieved. Previous
work has shown that this takes 30 days [6]. In some samples,
amounts of sodium carbonate (Sigma Aldrich) were added to the
samples to give concentrations of either 103 or 102 M. Sodium
perchlorate (Sigma Aldrich) was used to keep the ionic strength
constant and equal to 0.1 M in all of the pure brucite experiments.
Sodium perchlorate was selected for use here, because it is non-
complexing, and so will not compete for Am(III). It can be oxidis-
ing under acidic conditions. However, under the alkaline conditions
studied here it is relatively inert. Further, neither Am(III) nor Mg(II)
are susceptible to oxidation under these conditions. This ensured
that any changes in behaviour were due to the action of the car-
bonate, rather than the associated change in ionic strength. 241Am
tracer (T1/2 432.2 y, Ey 59.5 keV; 35.9%) was added to the
brucite colloid samples to give an initial concentration of
1.31  109 M, which is below the solubility limit for amorphous
americium hydroxide in these alkaline media: calculations using
PHREEQC [18] and the ThermoChimie database (V8, September
2011) predict that the solubility of amorphous americium hydrox-
ide is approximately 3  109 M. Aliquots of the samples were
removed and sequentially ultraltered using a Millipore Amicon
model 8010 stirred cell, with a capacity of 10 ml and membranes of
various pore sizes (3, 10, 100 and 300 kDa, equivalent to 1, 2.5, 6.5
and 12.6 nm) [19,20]. In a previous study, signicant sorption of
152
Eu(III) on these ultraltration membranes was observed,
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96
particularly at high pH. It was found that membrane pre-treatment
with stable Eu(III) solution prevented sorption and greatly
improved the europium recovery [6]. For this reason, and taking
into account that Eu(III) and Am(III) are commonly considered as
chemical analogues, the ultraltration cell and the membranes
were repeatedly washed before use with deionised water and then
pre-treated with 104 M stable Eu(NO3)3 solution in order to
saturate the sorption sites and prevent 241Am(III) binding to the
membrane surface, which was checked prior to the start of the
experiments. The liquor was continuously stirred during ltration
with a magnetic stirrer to reduce the build-up of concentrated
solutes on the membrane. The resulting ve size fractions (<3 kDa,
3e10 kDa, 10e100 kDa, 100e300 kDa and >300 kDa; i.e. <1 nm;
1e2.5 nm, 2.5e6.5 nm, 6.5e12.6 nm) were used to determine the
association of 241Am with different sized colloids. The SIXEP sam-
ples were also subjected to ultraltration, although a smaller range
of lters was used, due to the small sample volume and because of
restrictions in the handling of the samples. The typical un-
certainties in these experiments are in the region of 2e5%.
Samples were characterized by electron microscopy using a FEI
Quanta 200 FEG (Field Emission Gun) Environmental SEM and
energy dispersive X-ray analysis using an Oxford Instruments INCA
EDX (Energy Dispersive X-ray analyser). Americium concentrations
in the pure brucite experiments were measured by gamma ray
spectrometry, counting a 1 ml aliquot on a high-purity Ge semi-
conductor gamma-detector (50% efciency), attached to an EG&G
model 919 ADC and MCA.
3. Results and discussion
3.1. Storage tank samples
The Pu activity of the BST liquor was determined for the as-
received liquor and after the liquor was ltered through 200 nm
and 1 nm lters. The results are presented in Table 1, where the data
from the duplicate experiments are listed separately. The differ-
ences between the duplicate samples give an indication of the
heterogeneity of the samples.
For Pu, there is reasonable agreement between the duplicate
experiments. 37% of the Pu is associated with particulates or col-
loids greater than 0.2 mm in diameter. Approximately 14% of the Pu
is in true solution (<1 nm). Note, this denition assumes that any
species that passes through a 3 kDa (approximately 1 nm) lter
may be regarded as a simple dissolved species (i.e., free ion, simple
complex or small cluster). This denition is consistent with the
previous work of Gregson et al. [1] and Knopp et al. [21]. Therefore,
approximately 50% is associated with colloids 1e200 nm in diam-
eter. It would be expected that colloids would be more likely to pass
through the SIXEP sand bed lter as their size decreased. The
plutonium in the BST liquor associates more with smaller colloids
and particles, than was observed in the nuclear fuel storage pond
samples studied by Gregson et al. [1], who found that 93e99% of
the Pu was associated with colloids greater than 0.2 microns, with
very little Pu in the 1e200 nm range. However, their samples were
Table 1
Pu and Am fractionation in the Sellaeld plant samples.
Unltered e total solution loading (sample#1)
(sample#2)
1e200 nm fraction (sample#1)
(sample#2)
<1 nm: true solution (sample#1)
(sample#2)
87
taken from a storage pond after the sludge had been disturbed, and
so their samples probably contained more large particles that had
not yet settled, which explains the difference. The samples studied
here are probably more representative of routine operations and
the normal input to SIXEP.
Gregson et al. [1] found that the Pu concentration in true solu-
tion for storage pond liquors depended upon the sampling time and
location. They sampled from two locations, the storage pond itself
and a downstream holding tank that receives transfers of solution
from the pond. The Pu concentrations for the pond were in the
range, 1011 e 1010 M, compared with approximately 1010 M for
the downstream tank. However, following sludge transfer opera-
tions in the pond, the concentrations were in the range 109 e
108 M. The Pu true solution concentrations here (z109 M) most
closely match those recorded by Gregson et al. for the holding tank
following transfers of liquor. This is probably to be expected,
because the bulk storage tanks studied here receive regular trans-
fers from other parts of the plant, and so the BST samples have also
been subjected to disturbance and hence exposure to atmospheric
oxygen. Using preferential extraction, Gregson et al. showed that
the majority of the Pu in their transfer sample solutions was oxi-
dised to Pu(V). Therefore, the results here are consistent with the
presence of Pu(V) predominantly in the solution phase.
Predictions of Pu(IV) and Pu(V) concentrations in the true so-
lution and of Pu(IV) dissolved colloid concentration may be made:
Table 2 gives these predictions and ranges for the pH of the plant
samples (9.6e9.9) [1,20,22e24]. The data in Tables 1 and 2 strongly
suggest that the true solution is dominated by Pu(V): most signif-
icantly, the measured Pu concentration is above the upper bound
limit for Pu(IV) solubility. The amount of Pu in the size range
1e200 nm is approximately 3.8  109 M. This is in the range of
values predicted for Pu(IV) oxyhydroxide intrinsic colloid concen-
tration (Table 2), which would be expected to be in this size fraction
[1]. Hence, some of this material may be intrinsic colloidal Pu(IV)
oxyhydroxide, or it could be Pu sorbed to other colloidal species: it
is not possible to tell from these data. A signicant proportion of the
Pu is associated with particulates and colloids (86% >1 nm), and
this might be present in an oxidation state that is more likely to
bind to surfaces. Pu(V) forms a plutonyl ion, PuO 2 , in aqueous
systems that has a much lower charge density (effective
charge 2.2 compared to 4.0 for Pu(IV)), and so it is less likely to
bind to particle surfaces and more likely to remain in solution
[25,26]. Hence, although it seems that Pu(V) dominates in true
solution, the colloidal fractions could still be dominated by Pu(IV).
The results obtained for 241Am concentration in the plant liquor
before and after ltration showed the same general trends as those
observed for Pu (see Table 1). The 241Am concentration in the un-
ltered plant liquor was in the range 1.3  1011 e 2.8  1011 M,
and that in true solution is an order of magnitude lower. The sol-
ubility limit for amorphous Am(III) hydroxide at pH 9.9 is
8.5  109 M. The 241Am concentrations in the plant samples are
therefore well below the solubility limit, and so unlike Pu, the Am
behaviour in these systems does not seem to be inuenced by
solubility. The relative proportion of Am bound to colloids and
Pu (nM) Am (nM)
7.7 2.3 1.3  102 1.8  103
7.4 1.3 2.8  102 7.8  103
3.9 0.5 1.4  103 1.8  103
3.7 0.6 8.5  103 1.1  103
1.2 0.6 2.2  103 1.3  103
0.9 0.4 2.6  103 1.6  103
88 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96

Table 2
Predicted Pu(IV) colloid and Pu(IV) and Pu(V) true solution concentrations [1,20e23].

Predicted (M) Upper limit (M) Lower limit (M)

Pu(IV) colloid (pH 9.6e9.9) should be in 1 e 200 nm fraction [1] 5.1  109 5.0  108 5.2  1010
Pu(IV) true solution (pH 9.6e9.9) should appear in <1 nm fraction 4.2  1011 1.2  1010 1.3  1011
Pu(V) true solution (pH 9.6) should appear in <1 nm fraction 2.7  1010 1.7  109 4.2  1011
Pu(V) true solution (pH 9.9) should appear in <1 nm fraction 1.1  1010 5.9  1010 1.8  1011

particulates (>1 nm) is approximately 90%, which is similar to the
value for Pu. The amount in the 1e200 nm fraction is 24%. It seems
likely, given that the system is far from saturated with respect to
Am(OH)3, that this fraction is due to Am sorption to colloids not
derived from Am, i.e. pseudo-colloid formation. The variability in
the absolute concentrations between the duplicate samples is more
pronounced for the Am data than for Pu, although there is only a
small difference for the true solution fraction. This may indicate
that there is a homogeneous true solution in the BST, but with more
heterogeneity in the Am associated with the colloidal and partic-
ulate fractions.
Examination of the particles and colloids retained on the lters
by electron microscopy showed a considerable variation in
coverage between the samples. Although there was evidence for a
signicant number of small colloids distributed across the lters,
some images showed large, irregular agglomerates. This reects the
heterogeneous nature of these samples. Occasional ring structures
containing crystalline solids were observed, indicative of the drying
pattern of a liquid droplet. Energy dispersive X-ray (EDX) analysis
showed these crystalline solids to be rich in sodium, consistent
with the liquor composition (sodium hydroxide), and so no further
analysis of these areas was undertaken.
Fig. 1 shows a low magnication image of material isolated from
the BST samples. Quantities of MgeAleO rich material with a
distinctive interlocking platelet morphology were observed. Such a
morphology was previously observed for sludge samples from the
storage ponds by Gregson et al. [7] It has been suggested previously
that Al may substitute for Mg in the brucite structure, forming a
double layered hydroxide, hydrotalcite [7]. Not only do the samples
in this study have a similar morphology, but they also show an EDX
elemental distribution of O, Mg and Al similar to that of the samples
of Gregson et al. [7] and the synthesised hydrotalcite spectrum that
they presented. Hence, some of the material in the BST seems to be
consistent with hydrotalcite. Hence, it seems that sludge samples
from different parts of Sellaeld site may have some common
properties.
Examples of individual particles and agglomerates identied in
the samples are shown in Figs. 2 and 3. The magnesium was found

Fig. 1. Low magnication secondary electron images of particles isolated from BST liquor, with EDX spectrum.
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96 89

Fig. 2. Particulates retained on 200 nm lter; (A) ESEM image of a particulate and corresponding EDX spectrum. (B) Another particulate retained on the 0.2 mm lter; ESEM image of
a particulate and corresponding EDX spectrum and map (C) showing the elemental distribution across the particulate in (B).
90 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96

to be homogeneously distributed throughout the particles (EDX
elemental distribution maps in Figs. 2 and 3), showing that there is
a signicant magnesium contribution to all of the particles. The
maps also show that some of the particles are highly heteroge-
neous, with many components. In Fig. 3, where there are fewer
large agglomerates present, it is possible to see that the magnesium
is distributed as very small particles across the entire lter. Given
that Al is not distributed in the same manner, although some of the
larger particles are hydrotalcite-like, there are also Mg-only parti-
cles present too, particularly in the smaller range. There is a strong
magnesium signal throughout the agglomerates, indicating that the
larger particles are also magnesium-rich. Other elements (such as
iron, manganese, lanthanum, silicon) appear to be localised at
specic locations within the particles. Fe, Mn and La are probably
due to corrosion products from the metals within the storage
ponds. The presence of silicon is also expected, since the storage
tanks sometimes receive backwashes from the sand bed lters in
SIXEP, and the ponds and silos are concrete. Al, which is present as a
component in the Magnox fuel cladding alloy, shows intermediate
behaviour: it was present throughout the agglomerates with Mg,
but there were also bright spots, suggesting that some areas are
enriched in Al. Note, the mass concentrations of Am and Pu are too
low to be mapped: however, a small number of particles were
identied which contained uranium-rich particles, an example is
shown in Fig. 4. Uranium is present in much higher amounts than
americium or plutonium in the Magnox storage ponds at Sellaeld,
because spent nuclear fuel is largely 238U by mass. It seems likely
that the uranium in this particle represents a fragment of corroded
uranium that has become associated with other material. Gregson
et al. [7] also found that some particles isolated from the storage
ponds contained localised regions of uranium (as well as Ca, Fe and
Cr). The data presented here suggest that U-rich particles are

Fig. 3. Particles retained on 0.02 mm lter; (A) EDX maps showing variations in the elemental distribution of the different particles. (B) ESEM and EDX of another particle from the
same sample.
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96
Fig. 4. Backscattered image, showing a Mg-rich particle with a uranium inclusion. A
small bright particle was also analysed (spectrum 21,865 s) and found to be rich in U
(probably uranium oxide). A spectrum of the background lter also shown (bottom)
containing C, F and O.
transported to the efuent treatment plant. Gregson et al. [7] also
observed three different morphologies in their pond samples:
interlocking platelet; rod-like and ovoid. The interlocking platelet
91
motif (similar to that observed here; Fig. 1) was more enriched in Al
than the other two forms. There was also a suggestion that the rod-
like and ovoid particles might be more carbonated than the inter-
locking platelets, although there was some uncertainty, due to the
carbon content of their lters. In this work, although some particles
were observed that showed a common morphology with those
isolated from the storage ponds [7] (e.g., Fig. 1), there were also
other, heterogeneous irregular agglomerates that were different to
those observed previously (Figs. 2e4). The amount of particulate
material observed in samples of pond water was higher than
observed here (based on coverage of the lters) [7], but those were
taken after the sludge had been disturbed in the storage ponds. A
large amount of particulate material would have been mobilised
during that operation and this may not have settled fully prior to
sampling. The BST samples used in this work were taken during
routine operations. Therefore, larger particulate material should
have settled at the time of sampling, so that the sample only con-
tained non-settling particles, which would be smaller. Hence, the
particles observed here are more representative of those that enter
SIXEP routinely.
In order to begin to understand the chemistry and interactions
of colloids and actinides in SIXEP and the storage ponds and silos, a
much simpler system was studied in detail, which was not possible
with the plant sample. The particles and colloids in the BST liquor
are magnesium-rich. Apart from the larger, hydrotalcite-like par-
ticles, the particle morphology in the samples was indistinct,
although previous characterisation of pond samples has indicated
that they are likely to be mostly brucite, with some magnesium
hydroxy-carbonate and substituted brucite [7]. In this case, it seems
likely that some of the particles are brucite, since there are Mg-rich
particles that appear to contain little carbon (e.g. see EDX spectra in
Figs. 3 and 4). Others may be more carbonated (e.g. EDX spectrum
in Fig. 2a), but these seem to contain less magnesium (Fig. 2).
Hence, it seems likely that the bulk of the material is brucite, but
with inclusions of other phases. It is not possible to tell whether
these inclusions grew on the magnesium particles or whether they
formed separately before binding to them. Given the importance of
brucite in the ponds and efuent streams, brucite was chosen for
the model system.
3.2. Brucite colloid systems
The (Fluka) brucite was analysed using SEM. EDX conrmed that
magnesium and oxygen were the only elements present in the
material. High magnication SEM images of the powder in the size
range 100 nme1 mm show a blocky hexagonal morphology (Fig. 5),
which is characteristic for brucite [6,27]. Carbonated brucite sam-
ples, produced by sparging with carbon dioxide, were also studied
and mixed crystal morphology was observed. The images of the
carbonated brucite showed that hexagonal crystallites still
accounted for the majority of the samples, but there were occa-
sional regions of platelets. Although platelets were formed, the
morphology of the products was distinctly different to that of the
samples from the plant.
The distribution of 241Am(III) between the supernatant and the
brucite colloids, as well as the contributions of the various colloid
size fractions were studied at pH 10.4 in the absence of carbonate
and in the presence of 103 M and 102 M carbonate, as a function
of the contact time.
Surprisingly, the results for the carbonate-free system (Fig. 6)
show Am(III) mainly associated with larger colloids (see below,
Fig. 7). This observation is consistent with previously reported
characterisation of storage pond liquors, in which the actinides
have been predominantly associated with large size fractions [1].
Initially, 13% of Am remained in true solution, while 66% was
92 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96

Fig. 5. Clockwise from bottom left: SEM image of brucite and carbonated brucite samples; EDX analysis of carbonated colloids; SEM image of carbonated brucite; Examples of sub-
micron spherical brucite colloids (occasional bacteria were also noted).

241
Fig. 6. Association of Am(III) with various brucite colloidal size fractions in the absence of carbonate. Clockwise from top left: 1, 10, 70 and 160 days from the beginning of the
experiment.
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96 93

associated with colloids larger than 300 kDa (12.6 nm), 12% with
100e300 kDa (6.5e12.6 nm) colloids, 3 and 6% with 10e100 kDa > MgOH Mg2 2H2 O/MgOH2solid > MgOH2 H
(2.5e6.5 nm) and 3e10 kDa (1e2.5 nm), respectively. The ageing of
the system (over 6 months) results in a complete loss of Am from where >MgOH and >MgOH 2 are brucite surface species. Am could
true solution and a sharp increase in the contribution of the largest sorb onto the brucite surface in a similar process, potentially
colloids (96%, >12.6 nm). The variation with time in the size dis- becoming incorporated into the brucite,
tribution is shown in Fig. 7, along with the distribution of the col-
loids themselves. > MgOH Am3 3H2 O/MgOH2solid > AmOHOH2
The concentration of Mg present as colloids (>3 kDa; 1 nm) in
the absence of carbonate (pH 10.4 and I 0.1) was 2H
2.8 0.8 ppm (13%). The colloid size fractionation data for the
brucite colloids are shown in Fig. 7. A transition from hexagonal Addition of 103 M carbonate to the system leads to only very
prismatic morphology, which is characteristic for the mineral, to minor changes in the Am distribution pattern, particularly in the
the more energetically favourable spherical, as the size of colloids initial period (Fig. 8). One day after the beginning of the experi-
decreases, has been previously observed for brucite [6]. This ment, 50% of Am activity is detected in the largest colloidal fraction,
phenomenon may be attributed to a reduction in surface energy as compared to 66% in the carbonate-free system. Once again, as the
the small particles reduce their surface area to volume ratio. To contact time increases, the proportion of Am in true solution and in
calculate the number of colloids and the surface area associated
with the colloid size fractions for brucite, an approximation has
been made by assuming that the particles are spherical: such an
approach has been used previously [6]. The colloid size distribu-
tions by mass, total surface area and number are shown in Fig. 7.
The smallest colloids (3e10 kDa; 1e2 nm) are present in the
largest number, meaning that the biggest colloids, even though
they are signicant in terms of mass, do not contribute greatly to
the overall colloid population number distribution. The smallest
colloids also dominate the available surface area, and so they
might be expected to have the most potential surface binding
sites. In fact, considering the surface areas, the americium has a
disproportionately low association with the small colloids. The
data in Fig. 7 suggest that when americium is introduced into the
system, it should be most likely to bind to small colloids, because
it is most likely to encounter a binding site on a small colloid rst.
However, it is clear that the larger colloids represent a more stable
host even at short contact times, and over time there is further
transfer to the larger fractions. The slow nature of this redistri-
bution process might be explained by the slow rate of interaction,
due to the relatively low concentration of the species in solution,
or it could be that dissociation from the smaller colloids, which is
necessary to allow redistribution, is intrinsically slow.
Compared to their surface area, smaller colloids represent less
interior volume, which is proportional to mass, but even taking into
account colloid volume, the small colloids are unable to compete
for americium. It may be that an Ostwald Ripening effect is
responsible for the distribution. Mass for mass, it would be ex-
pected that the smaller particles are more prone to dissolution than
the larger species, which would be expected to grow. Hence, Am
that binds to the surface of the smaller colloids will be more likely
to be removed as the particle dissolves and made available for
binding to a different colloid, whereas any Am bound to a growing
colloid would be likely to be incorporated within the structure, as
the host particle grows and could become unavailable for interac-
tion with other particles. The net effect would be a progressive
transfer of Am to the larger fractions. Beyond any differential sta-
bility of small and large brucite colloids, the larger colloids will be
able to isolate Am further from the surface, deep in the brucite
structure, making it unavailable for reaction. Previous work has
shown that actinides can be incorporated to a depth of 50 nm in the
brucite structure [10,11]. It would be expected that sorbed radio-
nuclides would become less available as their depth of deposition
increases, and so larger particles with diameters greater than
100 nm would be expected to have an advantage over smaller
particles, particularly those only a few nanometres in size. Fig. 7. Size fraction data for brucite colloids expressed as mass, surface area and
It has been suggested that the reaction responsible for the number (top); Distribution of Am between true solution (<1 nm) and different size
growth of brucite is [26], fractions for carbonate-free system (middle) and with 102 M carbonate (bottom).
94 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96

Fig. 8. Association of 241

Am(III) with various brucite colloidal size fractions in the presence of 103 M carbonate. Clockwise: 1, 10, 70 and 160 days from the beginning of the
experiment.

the small colloidal fractions decreases, and after 160 days, the Am
distribution is analogous to that in the carbonate-free system, with
Am completely removed from true solution and the small colloidal
fractions and 94% associated with the largest colloids (>300 kDa;
>12.6 nm).
A very different distribution pattern was observed in the system
containing 102 M carbonate (Fig. 9). The vast majority of the
americium was found in true solution and in the smaller colloidal
fractions. Initially, 67% of the 241Am activity is in true solution, 23%
is sorbed on the smallest colloids (3e10 kDa), while the larger
colloidal fractions play a negligible role in the binding of Am. In
contrast to the systems with lower carbonate concentrations,
ageing up to 160 days had little effect on the Am distribution
pattern. The distributions in the presence of 102 M carbonate are
shown in Fig. 7 for comparison. This result is consistent with the
study of Parry et al. [17], who found that a similar concentration of

Fig. 9. Association of 241

Am(III) with various brucite colloidal size fractions in the presence of 102 M carbonate. Clockwise: 1, 10, 70 and 160 days from the beginning of the
experiment.
 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96 95

carbonate (3  102 M) was able to prevent the sorption of Pu to carbonate, americium speciation in the alkaline pH range is more
corroded Magnox sludge. complex and consists of charged Am carbonato complexes in
The speciation of Am in alkaline conditions was calculated with addition to the hydroxo-complexes. At pH 10.4, the speciation is
PHREEQC [18], using the ThermoChimie (SIT) database (V8, dominated by both the negatively charged carbonato complex
September 2011) and the results are presented in Fig. 10. In the Am(CO3) 2 and the positively charged hydroxo-complex Am(OH)2 .

absence of carbonate, the Am speciation is dominated by Am(OH) 2,

2
At 10 M carbonate, the speciation is dominated by charged car-
with smaller amounts of Am(OH)3. In the presence of 103 M bonato complexes, and at pH 10.4 the major species are Am(CO3) 2
and Am(CO3)3 3 .
100
The speciation calculations help to explain the changes in Am
90 behaviour as carbonate concentration increases. In the absence of
+
Am(OH)2 carbonate, the proportion of unbound americium (Am3(aq)) is
80
1.85.104%, whilst in the presence of 103 and 102 M carbonate, it
70 is 8.68.105% and 2.15.106%, respectively. The amount of free
Am3(aq) is a measure of the availability of the Am for surface
Percentage (%)

60
complexation (and incorporation). The carbonate complexation
50
competes with the surface binding sites, preventing sorption, but
40 Am(OH)3 concentrations of carbonate greater than 103 M are required for it
30 to have a signicant effect.
The carbonate concentration in the SIXEP samples is probably in
20
the range 5  104 e 102 M (see above), taking into account the
2+
Am(OH)
10 carbonate already present and that introduced during sample
0
handling. Therefore, it is possible that the carbonate concentration
8.5 9.0 9.5 10.0 10.5 11.0 11.5 is in the region where there is a signicant effect in the pure brucite
pH system. This is consistent with the results, since the SIXEP samples
show a signicant fraction of Am in true solution (12%). At lower
100
carbonate concentrations, in the pure brucite system, the amount
of Am in true solution is below the limit of detection at equilibrium.
90 However, the majority of the Am (and Pu) is associated with par-
-
80
Am(CO3)2 ticles in the SIXEP solution, and so either the carbonate concen-
Am(OH)3 tration is too low to remove all of the Am from the particles or the
70 Am dissociation kinetics from the particles is slow, which would be
the case if it had been incorporated within the particles.
Percentage (%)

60 +
Am(OH)2 The actinide concentrations in the storage ponds themselves are
50 predominantly associated with particulate material, and the true
40 solution concentrations are relatively low [1]. The pond carbonate
concentration has been reported as approximately 8  104 M [1].
+
Am(CO3)
30 This is consistent with the Am/brucite experiments, which show
20 that such a carbonate concentration should not compete, for Am at
least.
10 The data from the brucite system help to explain the association
0 with larger particles in the SIXEP and pond samples [1]. The
8.5 9.0 9.5 10.0 10.5 11.0 11.5 colloidal surface area in a solution in equilibrium with brucite will
pH be dominated by smaller colloids, and there is evidence for a large
concentration of small, Mg-rich colloids in the SEM images of the
100 SIXEP particles. The pure brucite systems show that even in such a
system, there can be a greater than expected association of Am with
90
the larger particles. Of course, it is possible that in the SIXEP
80 -
samples the radionuclides are associating with the Fe-rich in-
Am(CO3)2
70 clusions that have been identied in the SEM images, but our re-
3-
Am(CO3)3 sults show that, even in the absence of the inclusions, we would
Percentage (%)

60
still expect the observed behaviour.
50

40 4. Conclusions
30
In a model system of brucite colloids and in the absence of
20
Am(CO3)
+ Am(OH)3 carbonate, 241Am is progressively associated with the larger col-
10
Am(OH)2
+ loids (>300 kDa; >12.6 nm) over a period of 6 months, even though
0
the small colloids dominate the population. This phenomenon may
8.5 9.0 9.5 10.0 10.5 11.0 11.5 be due to the dynamic dissolution and re-precipitation of the
pH brucite colloids, leading to the americium being incorporated
within the larger colloids. The presence of carbonate at a concen-
Fig. 10. Speciation of 241Am(III), calculated with PHREEQC using the SIT database [18]. tration of 103 M does not signicantly change the Am distribution
The calculations were based on 241Am concentration of 1.31.109 M: top e no car-
bonate; middle e 103 M carbonate; bottom e 102 M carbonate. The region of in-
pattern, compared to the carbonate-free system. However, if car-
terest for the SIXEP samples is pH 9.6e9.9, and that for the brucite model systems is bonate is present at a concentration of 102 M, the Am(III) speci-
pH 10.4. ation is dominated by carbonate complexes, and this results in an
96 Z. Maher et al. / Journal of Nuclear Materials 468 (2016) 84e96
increase in the 241Am concentration in the true solution and the
smaller colloidal fractions. The results suggest that radionuclide
behaviour in these systems may be very sensitive to carbonate
concentration, and that high carbonate concentrations may inhibit
radionuclide-colloid interactions (at least for Am).
Analysis of SIXEP feed solution has found that signicant frac-
tions of Pu and Am are found in colloidal and particulate fractions
(>1 nm), although some is still found in the true solution phase. The
colloids and particles in the SIXEP sample are dominated by mag-
nesium: the smaller numerous colloids appear to be largely mag-
nesium phases, whilst the larger particles are magnesium phases,
with inclusions containing U, Fe, Mn, Si and Al. In addition, there is
evidence for the presence of larger hydrotalcite-like particles. There
is a signicant colloid loading of radionuclides in the input solu-
tions to SIXEP. Analysis of the data shows that the Am behaviour in
the SIXEP samples is consistent with radionuclide association with
a population of magnesium hydroxide colloids alone. The presence
of other particulate material would only increase the importance of
colloidal interactions.
These results may help to explain the success of SIXEP in
removing radionuclides from the efuent. Any radionuclides
associated with larger particles should be most easily removed by
the initial sand ltration column, and this is probably responsible
for most of the removal of actinide activity from the feed solution.
However, the sand bed will not remove all colloids quantitatively,
and small colloids in particular could make it through. It would be
expected that a simple ion exchange column would be relatively
ineffective at removing any radionuclides that were incorporated
within colloids. However, SIXEP is effective at removing actinides
from the efuent streams. In the plant, the efuent is treated with
CO2 gas immediately before injection into the zeolite column to
adjust the pH to 7. Any brucite colloids will be destroyed during this
process, releasing any associated radionuclides. Hence, the behav-
iour of SIXEP is consistent with radionuclides being associated with
larger magnesium phase based particles to a signicant extent,
which are removed either by settling in the Bulk Storage Tanks or
by ltration in the sand bed lters.
If possible in the future, it would be useful to analyse samples
from other parts of the Sellaeld plant for comparison with the data
presented here, but with greater protection from the atmosphere,
so that the effect of carbonation could be reduced. It would also be
useful to repeat the work with the model system, but using
plutonium. These data could then be compared with the plant data
reported here.
Acknowledgements
The authors would like to thank Sellaeld Ltd., The National
Nuclear Laboratory, The Nuclear Decommissioning Authority and
The Royal Commission for the Exhibition of 1851 for funding this
work.
References
[1] C.R. Gregson, J.J. Hastings, H.E. Sims, H.M. Steele, R.J. Taylor, Anal. Methods 3
(2011) 1957e1968.
[2] Nuclear Decommissioning Authority, Waste Stream Data Sheets, 2007. http://
www.nda.gov.uk/ukinventory/documents/.
[3] P.M. Bradford, B. Case, G. Dearnaley, J.F. Turner, I.S. Woolsey, Corr. Sci. 16 (10)
(1976) 747e766.
[4] J.H. Nordlien, S. Ono, N. Masuko, K. Nisancioglu, Corr. Sci. 39 (8) (1997)
1397e1414.
[5] R. Lindstrom, L.G. Johansson, J.-E. Svensson, Mater. Corros. 54 (8) (2003)
587e594.
[6] A. Pitois, P.I. Ivanov, L.G. Abrahamsen, N.D. Bryan, R.J. Taylor, H.E. Sims,
J. Environ. Monit. 10 (2008) 315e324.
[7] C.R. Gregson, D.T. Goddard, M.J. Sarseld, R.J. Taylor, J. Nucl. Mater 412 (2011)
145e156.
[8] H. Geckeis, R. Thomas, J. Contam. Hydrol. 102 (3e4) (2008) 187e195.
[9] A.P. Novikov, S.N. Kalmykov, S. Utsunomiya, R.C. Ewing, F. Horreard,
A. Merkulov, S.B. Clark, V.V. Tkachev, B.F. Myasoedov, Science 314 (2006)
638e641.
[10] J.D. Farr, P.K. Schulze, B.D. Honeyman, Radiochim. Acta 88 (9e11) (2000)
675e679.
[11] J.D. Farr, M.P. Neu, P.K. Schulze, B.D. Honeyman, J. Alloys Compd. 444 (2006)
533e539.
[12] G.R. Bochkarev, G.I. Pushkareva, J. Min. Sci 45 (3) (2009) 290e294.
[13] M.A. Kim, P.J. Panak, J.I. Yun, J.I. Kim, R. Klenze, K. Kohler, Colloids Surfaces A
Physicochem. Eng. Aspects 216 (1e3) (2003) 97e108.
[14] M.A. Kim, P.J. Panak, J.I. Yun, A. Priemyshev, J.I. Kim, Colloids Surfaces A
Physicochem. Eng. Aspects 254 (1e3) (2005) 137e145.
[15] C. Degueldre, A. Bilewicz, W. Hummel, J.L. Loizeau, J. Environ. Radioact. 55 (3)
(2001) 241e253.
[16] U. Alonso, C. Degueldre, Colloids Surfaces A Physicochem. Eng. Aspects 217
(1e3) (2003) 55e62.
[17] S.A. Parry, L. O'Brien, A.S. Fellerman, C.J. Eaves, N.B. Milestone, N.D. Bryan,
F.R. Livens, Energy Environ. Sci. 4 (2011) 1457e1464.
[18] D.L. Parkhurst, C.A.J. Appelo, User's Guide to PHREEQC, A Computer
Program for Speciation, Batch-Reaction, One-Dimensional Transport
and Inverse Geochemical Calculations, 1999 (U.S. Geological Survey).
[19] K. Nath, Membrane Separation Processes, Prentice Hall, 2008.
[20] K. Wilkinson, J. Lead, Environmental Colloids and Particles: Behaviour, Sepa-
ration and Characterisation, John Wiley and Sons, 2007.
[21] R. Knopp, V. Neck, J.I. Kim, Radiochim. Acta 86 (1999) 101e108.
[22] V. Neck, M. Altmaier, T. Fanghanel, C. R. Chim. 10 (2007) 959e977.
[23] V. Neck, J.I. Kim, Radiochim. Acta 89 (2001) 1e16.
[24] D. Rai, Radiochim. Acta 35 (1984) 97e106.
[25] G. Choppin, J. Rydburg, J.O. Liljenzin, Radiochemistry and Nuclear Chemistry:
Theory and Applications, second ed., Butter-Heinemann, 1995.
[26] B.A. Powell, Plutonium Oxidation State Geochemitsry in the SRS Subsurface
Environment (WSRC-TR-2003-00035), December 2002.
[27] O. Pokrovsky, J. Schott, Geochim. Cosmochim. Acta 68 (1) (2004) 31e45.