Stability And Sorption Capacity Of Montmorillonite Colloids

Transcript

Stability and sorption capacity of montmorillonite colloids Investigation of size fractional differences and effects of γ-irradiation KNAPP KARIN NORRFORS KTH Royal Institute of Technology School of Chemical Science and Engineering Department of Chemistry Applied Physical Chemistry SE-100 44 Stockholm, Sweden Copyright © Knapp Karin Norrfors, 2015. All rights reserved. No parts of this thesis may be reproduced without permission from the author. Paper V © 2011 Elsevier B.V. TRITA-CHE Report 2015:8 ISSN 1654-1081 ISBN 978-91-7595-444-8 Akademisk avhandling som med tillstånd av KTH Kungliga Tekniska Högskolan i Stockholm framlägges till offentlig granskning för avläggande av teknologie doktorsexamen fredagen den 6 mars kl 10:00 i sal F3, KTH, Lindstedtsvägen 26, Stockholm. Avhandlingen försvaras på engelska. Fakultetsopponent: Dr. Daniel Kaplan, Savannah River National Laboratory, Aiken, SC, USA ii To my family, The difference between stupidity and genius is that genius has its limits. Albert Einstein iii iv Abstract Bentonite clay is intended to form one of the barriers in most repositories of spent nuclear fuel located in granite. One important function of the bentonite barrier is to retard transport of radionuclides in the event of waste canister failure. Bentonite has a high sorption capacity of cations and its main constituent is montmorillonite. In contact with groundwater of low ionic strength, montmorillonite colloids can be released from bentonite and thereby control transport of radionuclides sorbed onto the colloids. In colloid transport in bedrock fractures, size separation of clay colloids may occur due to physical and chemical interactions with the bedrock fracture surface. This may enhance or retard the overall transport of radionuclides, depending on the sorption capacities and stability of the differently sized clay colloids. The bentonite barrier will be exposed to γ-radiation from the spent nuclear fuel. Irradiation affects surface-related properties of bentonite. If an average sorption capacity value cannot be used for all colloid sizes or if sorption is affected by exposure to γ-irradiation, corrected sorption capacity values would give higher resolution in current reactive transport models. In order to study the size separation process, a protocol was developed and successfully applied to fractionate montmorillonite into different-sized colloid suspensions by means of sequential or direct centrifugation. The stability and sorption capacity were studied using these fractions. Both stability and sorption capacity were found to be similar for all colloid sizes. Bentonite exposed to γ-radiation sorbed less divalent cations with increasing radiation dose. The effect was not large enough to have any impact on diffusion. The presence of bentonite enhanced irradiationinduced corrosion of copper under anaerobic atmosphere. An average sorption capacity value for montmorillonite can be used for all colloid sizes in reactive transport models. The effect of γ-irradiation on sorption capacity is sufficiently large to require consideration in transport modelling. Keywords: montmorillonite, colloids, size effect, γ-irradiation, sorption, colloid stability, radionuclides v Sammanfattning Bentonite är planerad som en av barriärerna i de flesta slutförvar av använt kärnbränsle. Bentonite har en hög sorptionskapacitet för katjoner. Den huvudsakliga beståndsdelen av bentonit är montmorillonit. Montmorillonitkolloider kommer att frigöras från bentonitbufferten i kontakt med grundvatten av låg jonstyrka och på så vis styra transporten av sorberade radionuklider. Under den kolloidala transporten i bergsprickorna kan en separation med avseende på storlek uppstå genom interaktioner mellan kolloiderna och bergytan. Detta kan få till följd att den genomsnittliga transporten av radionuklider bromsas eller tilltar beroende på sorptionskapaciteten och stabiliteten av de olika kolloidstorlekarna. Bentonitbarriären kommer även att utsättas för γ-bestrålning från det använda kärnbränslet, vilket påverkar dess ytrelaterade egenskaper. Om inte ett medeltal för sorptionskapaciteten är giltigt för alla kolloidstorlekar eller om sorptionen påverkas av γbestrålning, behövs nya sorptionskapaciteter bestämmas och impliceras för noggrannare transportmodeller. En metod för att separera montmorillonitkolloider med avseende på storlek via direkt och stegvis centrifugering har utvecklats. Stabiliteten och sorptionskapaciteten för dessa fraktioner har studerats. Både stabilitet och sorptionskapacitet visade sig vara lika för alla kolloidstorlekar. Bestrålad bentonit sorberar mindre andel divalenta katjoner med ökad dos bestrålning. Effekten är dock inte stor nog för att slå igenom i diffusionsexperimenten. Förekomst av bentonit ökar även den strålningsinducerade korrosionen av koppar under anaeroba förhållanden. Ett medelvärde för sorptionskapaciteten kan användas för alla kolloidstorlekar i transportmodeller. Effekten av γ-bestrålning är dock stor nog för att implementeras i modellerna. Nyckelord: montmorillonit, kolloider, sorption, kollodial stabilitet, radionuklider vi storlekseffekt, γ-bestrålning, List of publications I. Smectite and latex colloid agglomeration at high ionic strength – Is there any effect of particle size? Knapp Karin Norrfors, Muriel Bouby, Johannes Luetzenkirchen, Mats Jonsson and Susanna Wold Submitted to Colloids and Surfaces A: Physicochemical and Engineering Aspects II. Montmorillonite colloids. I: Characterization and stability of suspensions with different size fractions Knapp Karin Norrfors, Muriel Bouby, Stephanie Heck, Nicolas Finck, Rémi Marsac, Thorsten Schäfer, Horst Geckeis and Susanna Wold Submitted to Applied Clay Science III. Montmorillonite colloids. II: Dependency of Colloidal size on Sorption of Radionuclides Knapp Karin Norrfors, Muriel Bouby, Rémi Marsac, Stephanie Heck, Thorsten Schäfer, Horst Geckeis and Susanna Wold Manuscript to be submitted IV. Montmorillonite colloids. III: Influence of Colloidal Size on the Sorption Reversibility of Radionuclides Knapp Karin Norrfors, Muriel Bouby, Rémi Marsac, Stephanie Heck, Thorsten Schäfer, Horst Geckeis and Susanna Wold Manuscript in preparation V. Effect of γ-radiation on radionuclide retention in compacted bentonite Michael Holmboe, Knapp Karin Norrfors, Mats Jonsson and Susanna Wold Radiation Physics and Chemistry, 80 (2011) 1371-1377 VI. Radiation induced corrosion of copper in bentonite-water systems under anaerobic conditions Knapp Karin Norrfors, Amanda Kessler, Åsa Björkbacka, Susanna Wold and Mats Jonsson Manuscript in preparation vii Contributions to the papers The author’s contributions to the appended papers were as follows: I. Principal author. Took part in planning the experiments and performed all the experimental work. Major part in writing. II. Principal author. Planned the experimental work together with Muriel Bouby and performed all experimental work, except the ICP-OES/MS, IC, XRD and AsFlFFF-analysis. Major part in writing. III. Principal author. Planned the experimental work together with Muriel Bouby and performed all experimental work, except the ICP-OES/MS-analysis. Major part in writing. IV. Principal author. Planned the experimental work together with Muriel Bouby and performed all experimental work, except the ICP-OES/MS-analysis. Major part in writing. V. Took part in planning the experiments and performed the batch sorption experiments. Minor contribution to writing. VI. Principal author. Planned and performed all experimental work. Minor contribution to writing. viii Table of contents Abstract....................................................................................................... v Sammanfattning ...................................................................................... vi List of publications ................................................................................ vii Contributions to the papers ............................................................... viii Table of contents ..................................................................................... ix 1. Introduction ........................................................................................... 1 1.1 Background – repository of spent nuclear fuel in Sweden ........ 1 1.2 The bentonite barrier ....................................................................... 2 1.3 Worst case scenario ........................................................................ 3 1.4 Transport of radionuclides by bentonite colloids ......................... 4 1.4.1 Stability of clay colloids ............................................................ 5 1.4.2 Size exclusion of colloids ........................................................ 7 1.4.3 Effect of γ-irradiation ................................................................ 8 1.4.4 Sorption of radionuclides onto clay colloids ......................... 9 1.4.5 Sorption reversibility of radionuclides .................................. 12 1.5 Objectives of this thesis ................................................................ 14 2. Materials ............................................................................................... 15 2.1 Unpurified MX80 ............................................................................ 15 2.2 Homo-ionic montmorillonite .......................................................... 15 2.3 Organic matter ................................................................................ 16 2.3.1 Humic substances .................................................................. 16 2.3.2 Fulvic acid ................................................................................ 16 2.4 Nano-sized polystyrene (latex) particles .................................... 17 2.5 Synthetic groundwater .................................................................. 17 2.6 Copper cubes ................................................................................. 18 3. Methods ................................................................................................ 19 3.1 Size fractionation of MX80............................................................ 19 3.1.1 Fractionation method ............................................................. 19 3.1.2 Characterisation of size fractions ......................................... 21 ix 3.2 Stability studies............................................................................... 22 3.3 Sorption of radionuclides onto size-fractionated MX80 ............ 24 3.3.1 Radionuclide (RN) cocktail .................................................... 24 3.3.2 Preparation of sorption samples ........................................... 24 3.3.3 Determination of sorption ...................................................... 24 3.4 Sorption reversibility studies on radionuclides........................... 25 3.4.1 Change in chemical composition.......................................... 26 3.4.2 Addition of competing ligand ................................................. 27 3.5 γ-irradiation experiments ............................................................... 27 3.5.1 Batch sorption experiments ................................................... 27 3.5.2 Radionuclide retardation in compacted bentonite ............. 28 3.5.3 Copper cube corrosion in the presence of bentonite ........ 28 3.5.4 Effect of total dose on the bentonite sorption capacity of Cu2+ ..................................................................................................... 29 4. Results and Discussion.................................................................... 31 4.1 Size fractionation of MX80 ............................................................ 31 4.1.1 Size distribution ....................................................................... 31 4.1.2 Characterisation of the colloids............................................. 35 4.1.3 Estimations of edge site densities ........................................ 37 4.2 Size effect on colloid stability ....................................................... 39 4.2.1 Latex particles ......................................................................... 39 4.2.2 Montmorillonite colloids.......................................................... 43 4.3 Sorption properties of RNs and analogues onto montmorillonite colloids .................................................................................................... 45 4.3.1 Effect of clay colloid size........................................................ 45 4.3.2 Effect of γ-irradiation .............................................................. 48 4.4 Sorption reversibility ...................................................................... 51 4.4.1 Introducing organic matter as competing ligand ................ 51 4.4.2 Decrease in pH........................................................................ 53 4.4.3 Increase in ionic strength ....................................................... 54 4.4.4 Addition of crushed bedrock material (CBM) ...................... 54 4.5 γ-irradiation-induced copper corrosion in the presence of bentonite................................................................................................. 56 4.6 General discussion ........................................................................ 57 Conclusions ............................................................................................. 59 x Further work ............................................................................................ 61 List of abbreviations and symbols .................................................... 63 Acknowledgements ............................................................................... 65 References ............................................................................................... 67 xi xii 1. Introduction 1. Introduction 1.1 Background – repository of spent nuclear fuel in Sweden Nuclear power has been one of the major energy sources for production of electricity in Sweden during the last four decades. Every year, each nuclear reactor in Sweden produces 25-30 tons of highly radioactive spent nuclear fuel, so hundreds of tons of spent fuel that have to be disposed of safely [1]. Actinides are present in the spent nuclear fuel, making it highly radioactive and toxic. Therefore, it has to be isolated from the biosphere for a long time, approximately 100 000 years, before the activity has decayed to the same levels as natural uranium minerals in the bedrock. A concept for storing spent nuclear fuel has been proposed and developed by the Swedish Nuclear Fuel and Waste management company (SKB). This multi-barrier concept is named KBS-3 [2], and the design includes three barriers preventing the release of radionuclides (RNs) to the environment (Figure 1). The spent nuclear fuel is planned to be stored encapsulated in copper canisters at 500 m depth in the crystalline granite bedrock. Each canister will be surrounded by bentonite clay to keep it in place. Figure 1: The KBS-3 concept designed for storage of spent nuclear fuel in Sweden [2]. 1 1. Introduction The outer layer of copper on the canister will provide corrosion resistance, while the cast iron insert will provide mechanical support. The outermost barrier, the granite bedrock, will provide an anoxic atmosphere during longterm storage. A bedrock with the lowest fracture frequency possible will be chosen. In the event of canister failure, most RNs will adsorb to the bedrock due to its negatively charged surfaces and its ability to reduce redoxsensitive RNs. Transport of RNs with the groundwater will thereby be delayed. The last barrier, the compacted bentonite clay, will form an approximately 35 cm thick layer around the copper canister. This clay barrier is the main subject of this thesis. 1.2 The bentonite barrier Bentonite has been chosen as a barrier in the KBS-3 repository for spent nuclear fuel due to its mechanical and chemical properties. The clay swells in contact with water. Therefore, it can keep the copper canister in place following any movements in the bedrock and at the same time seal small fractures present in the adjacent bedrock. In addition to its mechanical properties, bentonite will prevent corrosive elements from the surroundings, such as sulphide, thiosulphate and polythionates [3], from making contact with the copper canister. In the event of canister failure, bentonite has a high capacity for surface complexation and cation exchange, allowing potential leaching RNs to sorb onto the bentonite. This will lead to retention of any RNs migrating towards the biosphere. The main component of bentonite is montmorillonite, an aluminium (Al)rich smectite mineral. Montmorillonite layers of approximately 1 nm are stacked onto each other to form one particle, one stack. These layers are negatively charged at pH > 6. As water is absorbed between the layers, the clay swells [4]. Due to the negatively charged montmorillonite surfaces, bentonite has a high sorption capacity for cations. In cases of canister failure, most RNs will be present as multivalent cations and thereby have a high affinity for montmorillonite. Therefore, retention of cationic RNs will be significant in compacted bentonite [5]. 2 1. Introduction Bentonite releases stable colloids (i.e. particles of 1 nm to 1 µm diameter in at least one dimension in suspension [6]) when in contact with groundwater of low ionic strength, i.e. < 10-2 M [7-9]. During the intended lifetime of spent nuclear fuel repositories (~100 000 years), ice age cycles are predicted at northern latitudes [2]. At the end of an ice age, glacial meltwater of low ionic strength may press away old, saline groundwater equilibrated with the bedrock over a long time. In addition, with the pressure release from melting glaciers, bedrock fractures will be formed. Therefore, the bentonite barrier may come into contact with water of low ionic strength. From inventories of existing glaciers, glacial melt water is known to have a pH of 8-9 and low ionic strength, i.e. 5·10-4 M or less [10]. The released montmorillonite colloids can be transported in bedrock fractures with the groundwater flow. In the event of large mass loss, the buffer functionality will be endangered. In the case of canister failure, mobile montmorillonite colloids will control transport of RNs [5]. 1.3 Worst case scenario To perform safety assessments for the storage of spent nuclear fuel, the probability of possible events and their consequences have to be studied. A complete safety assessment must be based on rather extreme extrapolations in time. In this thesis, the following worst case scenario was considered (see Figure 2). • The copper canister is broken, the fuel pellets start to dissolve and release RNs. • Released RNs are sorbed onto montmorillonite colloids. • Groundwater of low ionic strength has pressed away the old, saline groundwater, opened up fractures in the bedrock and reached repository depth with high flow. 3 1. Introduction Figure 2: Schematic diagram of the worst case scenario studied in this thesis. In addition to this worst case scenario, γ-irradiation from the spent nuclear fuel will penetrate through the copper canister, exposing the bentonite to γirradiation. This exposure to γ-irradiation could change the transport properties and the sorption capacity of the clay colloids. 1.4 Transport of radionuclides by bentonite colloids Radionuclide transport through bedrock fractures is a highly complex process with many parameters that have to be considered. The transport is dependent on the interactions between the RNs and the clay colloids and also on the interactions between the clay colloids themselves and the bedrock surface. Furthermore, chemical conditions (e.g. water chemistry and chemical composition of clay colloids and bedrock) and physical properties (e.g. structure of bedrock fracture and amount of fracture filling material) have to be considered. The near-field concentration of clay 4 1. Introduction colloids will affect the stability of the colloids and thereby the transport of sorbed RNs. 1.4.1 Stability of clay colloids One of the most important aspects in transport of nano-particles is colloid stability. Unstable colloid suspensions will sediment and the particles will be excluded from solution and further transport. However, stability is an equilibrium with both physical and chemical parts, which means that new equilibria will be established in new groundwater conditions, such as water flow or chemical composition. Sedimented colloids can be resuspended back to the water phase if the surrounding conditions change. Stability can be expressed both in kinetic and thermodynamic terms. From a practical point of view, the stability of a colloid suspension is seen as the life-time of the suspension. The DLVO-theory predicts the stability of particles in a suspension in terms of energetics, where the electrostatic ). repulsion (V ) is compared to the van der Waals attraction (V V =V +V (Equation 1) Van der Waals forces are always attractive between two particles of the same material, expressed by the positive Hamaker constant (A) [11, 12]. Assuming two identical spherical particles in the suspension, it can be seen that attraction between the particles increases with increasing particle size and decreasing separation distance (Equation 2) [13-15]. V =− + + ln (Equation 2) where r is the radius of the spherical particles and s is the centre-to-centre separation. The stabilising force in suspension is mainly the electrical double layer repulsion, which can be seen as the activation energy for the agglomeration process. This repulsion is highly dependent on the ionic strength of the suspension and the charge of the particle surface. For a symmetric electrolyte it can be expressed as a function of the surface-to-surface 5 1. Introduction separation, h, the number concentration of electrolyte, n , the Boltzmann constant, k , and the absolute temperature, T [11, 16, 17]: V = n k Tγ ∙ e (Equation 3) where γ is calculated from the valence of the electrolyte, z, the charge of the electron, e, and the diffuse layer potential, ψ : γ = tanh (Equation 4) The measured ζ-potential can be used to approximate ψ due to the correlation between the charge at the slipping plane and the surface charge. The inverse of Debye-Hückel screening length, κ, is given by: κ= / (Equation 5) where I is the ionic strength, ϵ the permittivity of vacuum and ϵ the permittivity of the material. At high ionic strength, where the repulsion between the particles decreases towards zero, the amount of agglomerates formed is proportional to the probability of two particles colliding with each other. Therefore, the agglomeration kinetics is dependent on the particle concentration in suspension. There are many methods used for determining colloid stability. Dynamic light scattering (DLS) can monitor the in situ development of mean particle size in the systems. However, this technique only monitors the increase in mean agglomerate size with time. Besides, agglomerates are most often not spherical and their shape may depend on the kinetics of agglomeration. Therefore, the treatment of data from DLS-measurements is not straightforward [18-21]. To summarise, particle stability in suspension is dependent on particle characteristics, such as surface charge and particle composition (giving the Hamaker constant), and on the background electrolyte and initial particle concentration. The colloid stability of montmorillonite will determine the 6 1. Introduction concentration of migrating bentonite colloids in bedrock fractures. Montmorillonite colloids are stable at low ionic strength [7-9], and thus the colloid-facilitated transport of RNs will be more likely under these conditions. Holmboe et al. [22] have found that bentonite colloids are more stable when γ-irradiated, with the potential to be transported further in the bedrock. Finally, the relationship between particle stability and particle size has not been clarified and is thus investigated in this thesis. 1.4.2 Size exclusion of colloids Montmorillonite colloids may be separated by size during transport due to interactions between the bedrock and the colloids. In a ‘clean’ fracture with a low amount of fracture filling material and high water velocity, laminar water flow is expected. Here, transport of smaller colloids will be enhanced due to their lower mass and surface area (Figure 3). In contrast, in bedrock fractures containing a high amount of fracture filling material, the colloids may be separated by size exclusion effects, where larger colloids are transported faster than smaller colloids in a process comparable to size exclusion chromatography. This phenomenon occurs due to chemical and electrostatic interactions between the clay colloids and the fracture filling material or bedrock. The smaller colloids are transported more slowly since they are able to interact with surfaces where large colloids are restricted due to their size. Figure 3: Schematic diagram of a plausible clay colloid size separation effect occuring in a bedrock fracture containing a low (left) or high (right) amount of fracture filling material. 7 1. Introduction The overall bulk properties of the total clay distribution may differ if its properties are connected to the exposed surface area and specific sizes of colloids are transported further. Intuitively, one would expect larger amounts of RNs to sorb onto smaller montmorillonite colloids per unit mass because of the excess of lateral surface area, where the sorption active edge sites are expected to be present [23]. In addition, in laminar water flow, smaller colloids will be transported further in the bedrock fractures than large clay colloids (Figure 3). In this case, by only considering bulk properties in reactive transport models, the RN transport would be underestimated. Colloid size effects on the stability or sorption capacity of bentonite colloids have not previously been reported in these types of systems. If a colloid size effect exist, these parameters would have to be included in reactive transport models, where size separation of montmorillonite colloids is not considered at the present time [24]. 1.4.3 Effect of γ-irradiation The bentonite barrier will be exposed to ionising radiation, mainly γirradiation from 137Cs, under anoxic conditions. Initially, the bentonite barrier will have low water content and no great effect of γ-irradiation on the properties of bentonite has been found in such conditions [25-27]. However, the bentonite will be saturated with groundwater over time. Therefore, γ-irradiation effects on bentonite suspensions with high water content were studied in this work. At a high water content, several γirradiation effects on clay colloids have been reported, e.g. the stability of montmorillonite colloids increases upon γ-irradiation [22]. Furthermore, the structural Fe2+/Fe3+-ratio in montmorillonite colloids increases after exposure to γ-irradiation [28]. This indicates that the negative surface charge on the montmorillonite colloids increases accordingly. Sorption is clearly related to surface charge, which raises the question of whether exposure to γ-irradiation can also affect the sorption of RNs onto montmorillonite. Since sorption of RNs onto montmorillonite is described both by electrostatic interactions such as cation exchange (macroscopic properties) and surface complexation (microscopic properties), sorption of different RNs may be affected differently upon γ-irradiation. If the effect on 8 1. Introduction sorption capacity is sufficiently large, this may also be reflected in the retention properties of RNs in clays. 1.4.4 Sorption of radionuclides onto clay colloids Sorption of inorganic cations onto clays has been studied for decades [29, 30]. However, the mechanisms for interactions between RNs and the clay surface are not fully understood. The sorption mechanisms vary with the valency and thermodynamic properties of the RN, so no general sorption model can be used for all RNs. The most important elements to be studied in the storage of spent high and low level nuclear fuel, and their analogues, are [31]: • Actinides: plutonium (Pu), thorium (Th), americium (Am), protactinium (Pa), neptunium (Np), curium (Cm), actinium (Ac) and uranium (U) • Other elements: lead (Pb), radium (Ra), technetium (Tc), caesium (Cs), zirconium (Zr), nickel (Ni), tin (Sn), selenium (Se), palladium (Pd) and niobium (Nb) Among these elements, there is great diversity in speciation, ionic radius and thermodynamic properties. Sorption capacity is most often given as an equilibrium constant (KD value) where the amount of RNs sorbed onto the clay surface is normalised to the amount of clay (mass). These values are useful when comparing sorption of different RNs, or one element in different oxidation states, onto one type of clay with a given size distribution. The drawback with KD values is that they assume a sorption site density per unit mass for the material studied and not a site density normalised to its exposed surface. Therefore, the question is whether sorption models for RNs can be developed and applied to different sizes of clay colloids. In order to do this, sorption properties, such as specific sorption sites or cation exchange properties, have to be characterised for each type of clay and its colloid size [23, 29]. Several sorption models have been developed where sorption of RNs onto clay surfaces is explained by interactions at specific edge sites. Most often, no physical explanation of how these sites are constructed is given [32, 33]. The only information given is a general edge site density per unit area for a 9 1. Introduction given clay mineral, which is calculated from the theoretical formula of the clay [34]. These models have recently been further developed using metal oxides, such as Al-OH and Si-OH, as the specific sites on the clay colloids [35]. Interactions between the surface and RNs are explained as: ≡ MeOH + RN →≡ MeO − RN ( ) + H where ≡ Me corresponds to a metal on the clay surface. Even though Al and silicon (Si) oxides are the most dominant species in the interactions between RNs and clays [36-38], many other possible sites of metal oxides at the clay surface may explain the complexity of the sorption mechanism. Up to 27 different metal oxides and oxides formed from two different metals are suggested for determining the interactions between sorption of protons onto the clay structure. This development is performed using a non-electrostatic model based on the MUSIC approach [38, 39]. The question is whether all these sites are active in sorption of RNs and whether association constants can be stated for all these sorption mechanisms. As mentioned above, a first modelling approach, using Al and Si oxides as sorption sites, has been performed by Gaskova et al. [35] for sorption of U(VI) and Eu(III) onto bentonite. Even if these results correspond well to the experimental results, this is most probably not the complete description of the sorption mechanisms. If possible, all sites which participate in proton exchange [38] should be included in sorption models of heavier elements, such as RNs. In addition to the specific sorption sites, cation exchange of the clay particles has to be considered in the sorption modelling of RNs. Knowledge about the amount of interlayers and basal planes for the clay stacks is needed since ion exchange occurs faster at basal planes than on interlayers. There, the ions have to diffuse into the interlayers before participating in ion exchange [40, 41]. In all models available for estimation of sorption of RNs onto clay particles, a cation exchange capacity and an edge site density for the clays are given as general properties normalised to the surface area of the clays, irrespective of clay colloid size [23, 32, 35, 41]. In most studies, an average particle size of the clay is assumed after short-term sedimentation to remove particles larger than 2 µm. From the estimated area, the edge sorption capacity can 10 1. Introduction be calculated from structural formula of the clay particles [30]. This assumes a symmetrical amount of edge site density to mean particle size. There has been no previous discussion about whether the smaller particles are miniatures of the larger ones and whether edge sites are distributed in the same way on the surface. The number of available sorption sites would be expected to increase with decreasing particle size since both the surface area and the number of broken structures in the crystalline clay structure are increased (Figure 4). In this thesis, changes in sorption with changes in clay colloid size were studied. If present, a dependency of sorption on colloid size should be implemented in reactive transport models. If there is no effect of colloid size on the sorption, an approach is needed to estimate sorption from the composition and conformation of the clay particles. In addition to sorption capacity studies at chemical equilibrium, the kinetics of the sorption process has to be studied over a long time. Sorption of RNs onto montmorillonite present in bedrock fractures is a dynamic process that aims to reach the thermodynamically most favourable equilibrium. Figure 4: Schematic diagram of how the sorption of RNs may differ with size of the clay colloids. The question is whether the sorption remains unchanged with a change in clay particle size. If metal oxides in the clay structure correspond to the specific sorption sites used in sorption models [38, 42], this information can be used when 11 1. Introduction comparing the sorption properties of γ-irradiated and unirradiated bentonite colloids. The specific sorption sites may be affected upon exposure to ionising irradiation, since the surface characteristics of clay colloids are known to change [22, 28, 43]. Moreover, organic degradation products are affected by ionising radiation [44] and are known to be present in bentonite (approximately 0.2% by mass) [45]. If sorption of RNs onto bentonite colloids is affected by the amount of organic matter present at the clay surface, sorption is affected by γ-irradiation. Transport of RNs is highly dependent on the sorption capacity of RNs onto clay colloid surfaces. When the sorption mechanisms are fully understood, risk assessment models with finer resolution can be constructed. 1.4.5 Sorption reversibility of radionuclides During transport of RNs in bedrock fractures, sorption of RNs is driven to be in equilibrium with the surrounding chemical conditions. The sorption equilibrium will be affected if the chemical conditions change over time or distance or if competing ligands are added to the system (Figure 5). During the estimated lifetime of a repository for spent nuclear fuel, i.e. at least 100 000 years, the chemical conditions may vary significantly. Figure 5: Schematic diagram of different sorption reversibility cases. 12 1. Introduction Two factors with a large impact on RN sorption are pH and ionic strength [30, 32, 46]. Both the surface charge of the particles and the RN valency are dependent on pH. The chemical composition of the background electrolyte is also affected by a change in pH if, for example, carbonate is present in solution. An increase in ionic strength affects sorption since clay colloids are unstable at higher ionic strength [47, 48]. Ions in the background electrolyte may compete in sorption onto the clay by cation exchange. Furthermore, the charge of the clay particles will be screened at higher ionic strength. Organic matter present in groundwater (up to 5 mg/L [49]) is known to form strong and highly soluble complexes with RNs under these conditions [50-53]. Small organic molecules are also known to interact with clay particle surfaces by electrostatic interactions [19, 54]. Therefore, it is highly important to study the sorption behaviour of RNs onto clays in the presence of organic matter as a competing ligand. Moreover, RNs are known to sorb onto bedrock surfaces and fracture filling material, surrounding the bentonite clay [55-57]. This increases the complexity of the sorption system. By adding samples of fracture filling material or crushed bedrock material to sorption samples of RNs and bentonite colloids, these ternary systems can be studied. 13 1. Introduction 1.5 Objectives of this thesis The aim of this thesis was to determine whether smaller- and larger-sized montmorillonite colloids have the same stability and sorption properties. In addition, the effect of γ-irradiation on sorption capacities of montmorillonite colloids was studied. The main objectives investigated in the thesis were: • Do smaller and larger montmorillonite colloids have the same composition and surface-related physical properties? • Can montmorillonite colloids be separated by size in suspension? If so, • Is there an effect of colloid size on colloid stability? • Can the sorption capacity of RNs onto montmorillonite colloids be normalised to exposed lateral surfaces, related to the size of the colloids? • Is the sorption reversibility of RNs affected by the exposed area of clay colloids in suspension? In addition, • How are the sorption properties of metal cations onto clay colloids affected by γ-irradiation? 14 2. Materials 2. Materials 2.1 Unpurified MX80 One of the bentonite clays used as a reference material for the storage of spent nuclear fuel in Sweden is unpurified MX80 bentonite [58]. Wyoming Volclay bentonite MX-80 from American Colloid Co., consisting of approximately 82 % montmorillonite, was used as the test material in this thesis. The structural formula of this montmorillonite is [45]: Na . (Al . Fe . Mg . )(Si . Al . )O (OH) The molar weight of the structural unit is MW = 372.6 g/mol [45] and its cation exchange capacity (CEC) is 0.75 meq/g (measured according to [59]). In addition to montmorillonite, accessory minerals (e.g. cristobalite, quartz, gypsum) are present in the clay [45, 60], in separate phases to montmorillonite. The montmorillonite was purified during colloid size separation, with the accessory minerals only present in the larger colloid size fractions [60]. Furthermore, a small amount (approximately 0.2% by mass) of organic matter is present in unpurified MX80, which may affect its properties [45, 61]. Due to the high heterogeneity of this material in mineralogical composition and colloid size, additional purified clays and latex standards were used in the experiments to better understand the reactive mechanisms of MX80 bentonite. 2.2 Homo-ionic montmorillonite In order to estimate how the impurities in MX80 bentonite affect clay properties on exposure to γ-irradiation, the sorption capacity of purified clays was compared. Apart from Mx80 bentonite, two Na-montmorillonites and one Ca-montmorillonite were used in the study [61]. Purified Namontmorillonite originating from Wyoming montmorillonite was supplied from the Clay Mineral Society. This clay is free from both free iron (Fe) (citrate-borate treatments) and organic matter (H2O2 treatment), and is 15 2. Materials hereafter denoted SWyNaw. In addition, Na- and Ca-montmorillonite originating from MX80 (WyNa and WyCa) where the clays are cationexchanged by NaCl or CaCl2, respectively, were included in the analyses. A detailed description of preparation of these clays is provided elsewhere [45]. 2.3 Organic matter 2.3.1 Humic substances The humic substances (HS) used in the γ-irradiation experiments comprised a commercial humic standard (1S102) obtained from the International Humic Substances Society. The HS was first dissolved in NaOH at pH 14 and later neutralised to pH ~7 by addition of H2SO4 under stirring. The ionic strength of the solution in this thesis is given after adjustment of pH. 2.3.2 Fulvic acid The organic matter used in all experiments performed on the size-separated montmorillonite was fulvic acid (FA-573) originating from natural water (Gohy573) at the Gorleben site, Germany [62]. The element composition of the FA, given by the manufacturer, is presented in Table 1. Fulvic acid most often has smaller colloids than HS in suspension and has more functional groups, resulting in a higher total change. Therefore, FA is more soluble in acid solution [63]. Table 1: Element composition of fulvic acid (FA). Element C H O N S Proton exchange capacity Amount 54.1 ± 0.1 % 4.23 ± 0.08 % 38.94 ± 0.04 % 1.38 ± 0.02 % 1.32 ± 0.01 % 6.82 ± 0.04 meq/g 16 2. Materials 2.4 Nano-sized polystyrene (latex) particles In the stability studies, well-defined, spherical and homogeneous carboxylated polystyrene (latex) nanoparticle standards (Magsphere Inc., Pasadena, USA) were used to gain a better fundamental understanding of agglomeration processes. These latex particles represent a large size range of nano-particles, from 24 nm up to 495 nm. Their dimensions correlate well with independent measurements by Scanning Electron Microscopy (SEM; FEI QUANTA 650 FEG). The sizes and their corresponding charge densities are presented in Table 2. Table 2: Characteristics of the standard carboxylated latex particles used in the stability studies and their corresponding charge density. Particle diameter and surface charge density as given by the manufacturer. Sample LC24 LC40 LC60 LC81 LC97 LC217 LC420 LC495 Particle diameter [nm] 24 ± 8 40 ± 13 60 ± 20 81 ± 6 97 ± 1.5 217 ± 24 420 ± 29 495 ± 50 SEM measurements [nm] 30 ± 5 47 ± 6 64 ± 10 88 ± 15 104 ± 19 214 ± 28 419 ± 9 511 ± 79 Surface charge density [mC/m2] 72.8 94.8 121 45.1 34.1 182 220 209 2.5 Synthetic groundwater To simulate carbonated groundwater with low ionic strength, representing glacial meltwater, a synthetic groundwater (SGW) was prepared. This SGW was used as the background electrolyte in all experiments that included sizeseparated montmorillonite colloids. The SGW was chosen over a natural groundwater to avoid the influence of natural colloids present in the groundwater and to provide a homogeneous background throughout all experiments. The final composition of the SGW is given in Table 3. The groundwater had a total ionic strength of 1.57 mM and pH 8.4 ± 0.1. 17 2. Materials Table 3: Chemical composition of the synthetic groundwater (SGW) of Grimsel type used as background electrolyte in all experiments performed. Element Na+ Ca2+ FClSO42Si HCO3- Concentration [mM] 1.2 0.05 0.1 0.074 0.04 0.5·10-3 1.4 2.6 Copper cubes The copper cubes used in the experiments had dimensions 1 × 1 × 1 cm3 and consisted of 99.992% Cu. They were cut from a SKB copper canister wall and polished on all sides, using SiC abrasive papers of 1200 grit in 99.5% ethanol, prior to addition to the bentonite slurry to avoid effects from impurities on the copper surfaces. After polishing, the Cu cubes were ultrasonicated during 5 min and later dried under N2 in a glovebox where they were stored until preparation of γ-irradiation samples (section 3.5.3). 18 3. Methods 3. Methods 3.1 Size fractionation of MX80 3.1.1 Fractionation method Size fractionation of MX80 bentonite in SGW was performed prior to stability, sorption and sorption reversibility experiments. All montmorillonite colloids were separated in size by a sedimentation step, followed by sequential or direct centrifugation according to a specially designed protocol for the system. In the sedimentation step, 10 g/L MX80 was diluted in SGW, stirred for one day and left to sediment for three days. The colloid suspension collected after sedimentation is hereafter referred to as S0 (Figure 6). The collected suspension S0 was the starting material for all centrifugation steps. S0 S0 R0 Figure 6: Colloid suspensions of MX80 bentonite before and after sedimentation for three days in synthetic ground water (SGW). The suspension S0 and the residual R0 are marked in the right-hand beaker. A sequential centrifugation procedure was performed on the S0 suspension using the centrifugation times and speeds presented in Table 4 and Figure 7. The ith suspension, Si, was centrifuged according the conditions given in 19 3. Methods Table 4 and, after centrifugation, resulted in the suspension S(i+1) and the corresponding residual R(i+1) (Figure 7). Table 4: Conditions used for fractionation of clay suspensions, and colloid sizes (equivalent sphere diameters) expected in the residuals (Ri) and the supernatants (Si). Suspension Conditions of separation (C: centrifugation, UC: ultracentrifugation) Particle size expected in the ith residual clay fraction (Ri) in nm Mean colloid size expected in the ith supernatant (Si) in nm S0 3 days sedimentation 1000 ≤ R0 0 ≤ S0 ≤ 1000 S1 C: 30 min (S0) at 313 x g 450 ≤ R1 ≤ 1000 0 ≤ S1 ≤ 450 S2 C: 1 h (S1) at 700 x g 200 ≤ R2 ≤ 450 0 ≤ S2 ≤ 200 S3 C: 4 h (S2) at 1.200 x g 70 ≤ R3 ≤ 200 0 ≤ S3 ≤ 70 S3.5 UC: 30 min (S3) at 26.000 x g 50 ≤ R3.5 ≤ 70 0 ≤ S3.5 ≤ 50 S3.5UC UC: 30 min (S0) at 26.000 x ga 50 ≤ R3.5UC 0 ≤ S3.5UC ≤ 50 S3.5UC, FA UC: 30 min (S0) at 26.000 x ga 50 ≤ R3.5UC, FA 0 ≤ S3.5UC, FA ≤ 50 a : One-step ultra-centrifugation from a colloid suspension, S0, obtained after stirring and sedimentation of an MX80 suspension at 10 g/L in the presence or absence of 11.8 mg/L fulvic acid (FA) . Figure 7: Separation procedure used to obtain suspension Si during the sequential centrifugation separation. For comparison to sequential centrifugation, the colloids in S0 were separated by one direct centrifugation with the same centrifugation speed and time as for the smallest-sized colloids in the sequential centrifugation 20 3. Methods process (Table 4). This direct centrifugation was performed both in the absence and presence of FA. 3.1.2 Characterisation of size fractions All size-fractionated montmorillonite colloids were characterised to study whether the composition of the colloids remained the same throughout all separation steps. The element concentrations and size distributions were determined. 3.1.2.1 Ion and colloid concentrations The element concentrations in the suspensions presented in Table 4 were determined by inductively-coupled plasma-optical emission spectrometry (ICP-OES; Optima 2000 DV, PerkinElmer), inductively-coupled plasmamass spectrometer (ICP-MS; X-Series2, Thermo Scientific, Germany) and ion chromatography (IC; ICS-3000). The concentrations were then compared to the structural formula of the montmorillonite colloids to obtain the clay colloid concentration. Background concentrations for all elements were taken from fast centrifuged samples where no colloids were detected [60]. 3.1.2.2 Mineral characterisation and content of organic matter The mineral composition was determined by X-ray diffraction (XRD) where all suspensions and residuals were dried on low background Si wafers. The composition of the raw material MX80 was also monitored as a reference. The powder diffractograms were recorded with a D8 Advance (Bruker) diffractometer (Cu Kα radiation) equipped with an energy dispersive detector (Sol-X). The phases were identified with the DIFFRAC.EVA version 2.0 software (Bruker) by comparison with the JCPDS 2 database. The amount of FA as organic matter in the suspensions was measured using a total organic carbon analyser (TOC-5000, Shimadzu) before and after centrifugation. These results indicated that one-third of the FA was associated to the montmorillonite colloids after size separation. 3.1.2.3 Size distribution measurements To determine the efficiency of the size fractionation, size distribution measurements were performed using two methods: i) Photon correlation 21 3. Methods spectroscopy (PCS; homodyne single beam ZetaPlus System equipped with a 50mW solid-state laser emitting at 632 nm, Brookhaven Inc, USA, using the Non-Negatively constrained Least Squares (NNLS) algorithm [64]) and ii) an asymmetric flow field-flow fractionation system (AsFlFFF; HRFFF 10.000 AF4, Postnova Analytics, Landsberg, Germany) coupled to a UV-Vis. detector (LambdaMax LC Modell 481, Waters, Milford, USA) and an ICPMS unit (X–Series2, Thermo Scientific, Germany). These techniques complement each other since PCS gives the mean intensity-weighted and volume weighted colloid size of the measured sample, whereas the AsFlFFF/UV-Vis./ICP-MS system gives the element concentration as a function of particle size. A detailed description of the system and method used for AsFlFFF can be found in the Supporting Information (SI) to [60]. For the conditions used in this thesis, the AsFlFFF/UV-Vis./ICP-MS system does not detect the largest sized clay colloids, which were detected with PCS. 3.2 Stability studies Mono-dispersed nano-sized latex particles with well-defined particle sizes (section 2.4) were used as reference material in the stability studies (Table 5). Agglomeration in the suspensions was taken as the increase in mean hydrodynamic diameter for the particles as a function of time, monitored by PCS (same instrument as described in section 3.1.2.3). The conditions studied for the different particles at pH 7 are summarised in Table 5. Table 5: Electrolytes and their concentrations expressed in ionic strength used in the stability studies. Electrolyte NaCl CaCl2 NaCl CaCl2 MgCl2 Particles Latex particles Latex particles Montmorillonite colloids Montmorillonite colloids Montmorillonite colloids Ionic strength [M] 0.5 and 2 0.5 and 2 0.01, 0.1 and 1 0.03, 0.3 and 3 0.03, 0.3 and 3 The agglomeration rate for the systems was determined by linear approximations of the increase in hydrodynamic diameter, using the first 10-25 data points for each set of data. This agglomeration rate was then 22 3. Methods used to estimate the rate constant, k, of the system using the expression [1821, 65]: k= ∙[ ] 1− ∙ ∙ ( ) (Equation 6) where r is the hydrodynamic radius of a singlet, r the radius of a doublet, I the intensity of a singlet and I the intensity of a doublet, d the hydrodynamic diameter obtained from PCS, d the initial diameter of the particles, t the time and [X] the initial concentration of particles present in the samples, valid for low particle concentrations. Agglomerate rate constants could not be determined for montmorillonite since data on the initial concentration of particles were needed but were unavailable due to the polydispersivity. The stability of the bentonite suspensions was characterised by determining the stability ratio [18, 19, 66, 67] and the master curve [68-72] of the suspensions. The stability ratio, W, was defined as the ratio between the fast agglomeration rate (suffix f) and the measured agglomeration rate in the present sample: W= → → /[ ] () (Equation 7) /[ ] The master curve of a system was constructed by plotting the increase in hydrodynamic diameter, normalised to the initial particle size (d /d ) against the time of the measurement (t), normalised to the characteristic time of the Brownian aggregation process (t ). From the slope of this plot, the fractal dimension (d ), i.e. the packing density of the agglomerates formed, was obtained. It gives general information on the mechanisms of the agglomeration process. Master curves were constructed for the latex particles too, for comparison. 23 3. Methods 3.3 Sorption of radionuclides onto size-fractionated MX80 3.3.1 Radionuclide (RN) cocktail The radionuclide cocktail used in the sorption studies was prepared prior to addition of RNs to the colloid suspensions. It was prepared in an argon glove box and kept acidified until addition to the pre-size-fractionated clay colloids. The oxidation states and final concentration of RNs in the sorption samples are presented in Table 6. Table 6: Radionuclides, their oxidation state and final concentration in the sorption samples. Radionuclide Th(IV) 99Tc(VII) 233U(VI) 237Np(V) 242Pu(IV) 232 Concentration [M] 10-8 5·10-9 10-8 10-8 2·10-9 3.3.2 Preparation of sorption samples All suspensions in Table 4 were used in the sorption study performed on size fractionated montmorillonite colloids. This study was performed in a glovebox under argon atmosphere (<1 ppm O2) at pH 8.9 ± 0.3 and Eh(SHE) +232 ± 50 mV. Each sorption sample was prepared in duplicate with 20 mg/L montmorillonite colloids and a total sample volume of 50 mL. To determine the time dependence of the RN sorption, samples were taken after 72 h (3 days), 260 h (2 weeks), 580 h (1 month) and 3800 h (6 months). The amount of RNs was measured by ICP-MS before and after ultracentrifugation and compared against the initial amount of RNs present in the samples to determine the sorption. 3.3.3 Determination of sorption The RN concentrations were measured with ICP-MS, where the distribution of RNs is given as particulate (%Part), colloid associated (%Colloid) or free (%Free) RNs in the sample (Figure 8). Particulates correspond to the sedimented clay colloids and eigen-colloids formed by RNs. The amount of particulates was determined from element concentrations in the sorption 24 3. Methods samples at a given sorption time without any shaking, compared with the initial concentration of the elements. The change in concentration before and after ultracentrifugation at a given sorption time corresponded to the montmorillonite-associated RNs, since ultracentrifugation under the conditions applied is known to effectively remove clay colloids of nano-size from solution [60, 73]. Lastly, the remaining concentration of RNs in the sorption samples was taken to be free RNs in the samples. Note that no distinction between free RNs and FA-RN-complexes could be made in the presence of FA in the samples. Figure 8: Schematic diagram of the radionuclide (RN) distribution in the batch sorption samples, as detected by ICP-MS. 3.4 Sorption reversibility studies on radionuclides To study the sorption reversibility processes for size-separated montmorillonite colloids, the sorption samples described in section 3.3.2 were used. After a given sorption time, the chemical conditions in the samples were changed or a competing ligand was added. Four different sorption reversibility studies were performed for the sorption samples where the sorption reversibility samples were prepared from the same sorption sample at each sorption time as shown in Figure 9. From each sorption sample, five sorption reversibility samples were prepared. The 25 3. Methods sorption reversibility was studied after one week and one year. The distribution of RNs in the sorption reversibility samples was detected as described for the sorption samples (see section 3.3.3). Figure 9: Schematic diagram of preparation of the sorption reversibility samples from the sorption samples (described in section 3.3.2) at each given sorption time. 3.4.1 Change in chemical composition 3.4.1.1 Decrease in pH A 0.1 mL aliquot of o.1 M HCL was added to 10 mL sorption sample to lower the pH. After addition of acid, the pH was decreased from 8.9 to 5.6, and then buffered back to pH 7.5 after one week. Therefore, an additional 0.15 mL 0.1 M HCl was added to the samples before storing them during one year. After one year, the pH was decreased to 2.7. 3.4.1.2 Increase in ionic strength By addition of 5 mL CaCl2 to 5 mL sorption sample, 0.5 M ionic strength CaCl2 was obtained. The montmorillonite colloids are known to be unstable at this ionic strength. These samples were prepared in duplicate. 26 3. Methods 3.4.2 Addition of competing ligand 3.4.2.1 FA as organic matter Addition of FA was studied, since FA is known to form strong complexes with RNs. A final concentration of 5.1 mg/L FA was obtained in the sorption reversibility samples. The distribution of RNs in the samples was analysed after one week and one year. 3.4.2.2 Crushed Bedrock Material (CBM) Grandiorite from Äspö was used as crushed bedrock material (CBM). The grandiorite had a grain size of 0.5-1 mm after crushing and was added to the samples in a 1:4 solid to liquid ratio. The specific surface area of the CBM grains, measured by BET (Quantachrome Autosorb Automated Gas Sorption), was approximately 0.1 m2/g. Information on preparation of the size-fractionated CBM and its composition is presented in [74]. 3.5 γ-irradiation experiments 3.5.1 Batch sorption experiments Two analogues to RNs, relevant for the storage of spent nuclear fuel, Cs+ and Co2+, were used to study the effect of γ-irradiation on sorption onto bentonite and montmorillonite. All clays presented in sections 2.1 and 2.2 were used in the study. The solid to liquid ratio for the clays was 1:200, with trace amounts of 137Cs+ and 60Co2+ and 1 µM inactive carrier isotopes. The experimental set-up used for the batch sorption samples is summarised in Table 7. All irradiated clays were irradiated prior to addition of the analogues in N2-atmosphere or air. For comparison, samples with added humic substances (HS) prior to γ-irradiation were prepared to detect whether presence of organic matter during irradiation affects the sorption. The irradiated samples were irradiated with a γ-dose of approximately 60 kGy using a Gammacell 1000 Elite 137Cs-source. The pH in the samples was controlled to pH 9 by a borate buffer. After five days of sorption, the RNs were separated from the clay by centrifugation for 30 min at 6.000 rpm and analysed with γ-spectroscopy. 27 3. Methods Table 7: Experimental set-up used for batch sorption experiments studying the effect of γirradiation. Clay MX80 Na/Camontmorillonite Na/Camontmorillonite Atm. N2 N2 dry dry Air dry γ-irradiated wet wet + HS wet wet + HS wet wet + HS Unirradiated wet wet + HS wet wet + HS wet wet + HS 3.5.2 Radionuclide retardation in compacted bentonite As a complement to the batch sorption samples described in section 3.5.1, diffusion cells with the clays MX80 and SWyNaw at a dry density of 1.6 and 1.5 g/cm3, respectively, were prepared. The diffusivity of Co2+ and Cs+ was studied for both γ-irradiated and unirradiated samples, as shown in Table 7. During the experimental time (80 days), the outlet solution was collected on a weekly basis and the γ-activity was analysed with γ-spectroscopy. After the experiments were terminated, the clay plugs were sliced into thin sections and their respective γ-activity was analysed. 3.5.3 Copper cube corrosion in the presence of bentonite The Wyoming Volclay bentonite MX-80 used in this study was unpurified prior to sample preparation. The bentonite was first dried at 80°C for 12 hours before being suspended at 4 wt% in suprapure MilliQ-water. The clay slurry was degassed by bubbling N2 for 12 hours and thereafter transferred to a glovebox. Six irradiation samples were prepared in a glovebox by adding one copper cube to 10 mL bentonite slurry. Two additional samples were prepared and stored in the glovebox as unirradiated references. The irradiated samples were irradiated for 7 days using a Gammacell 1000 Elite 137Cs-source. The dose rates and the total doses for all samples are given in Table 8. 28 3. Methods Table 8: Dose rates and total doses used for studying γ-irradiation induced copper corrosion in bentonite clay. Sample 1 2 3 4 5 6 Dose rate (Gy/s) 0.1 0.135 0.134 0.114 0.155 0.168 Total dose (kGy) 60 81 81 69 93 101 The amount of dissolved Cu from the copper cubes entering the suspension was analysed by ICP-OES (Thermo Scientific iCAP 6000 series ICP spectrometer) using a multi-element standard (Merck). The Cu concentration in the suspensions was measured both before and after centrifugation (30 min at 24.000 x g, Thermo Scientific Heraeus Megafuge 16, Microliter 30x2 mL-sealed rotor). The ICP-OES samples including bentonite colloids were acidified by suprapure HNO3 (Merck) prior to analysis. 3.5.4 Effect of total dose on the bentonite sorption capacity of Cu2+ The samples used to study the effect of total dose of γ-irradiation on Cu2+ adsorption onto MX80 were prepared similarly to those in the copper corrosion experiments. A suspension of 4 wt% MX80 bentonite was degassed by N2 for 12 hours prior to transfer into a glovebox. Nine samples of 10 mL bentonite slurry were sealed under anoxic atmosphere and seven of these were irradiated by the dose rates and total doses presented in Table 9. After irradiation, 3 mM CuCl2 were added to all samples and they were allowed to equilibrate for 24 hours. Thereafter, the Cu concentration was measured by ICP-OES before and after centrifugation (30 min at 24.000 x g, Thermo Scientific Heraeus Megafuge 16, Microliter 30x2 mL-sealed rotor). The ICP-OES samples including bentonite colloids were acidified prior to analysis. 29 3. Methods Table 9: Dose rates and total doses used for studying the sorption capacity of Cu bentonite. Sample 1 2 3 4 5 6 7 Dose rate (Gy/s) 0.155 0.297 0.168 0.100 0.135 0.249 0.134 Irradiation (hours) 262 186 245 138 208 217 208 30 time Total dose (kGy) 146 199 148 50 101 195 100 2+ onto MX80 4. Results and Discussion 4. Results and Discussion In the following, the results from all experiments are discussed in terms of whether the surface characteristics of montmorillonite colloids differ between the smaller and larger colloids in the size distribution and whether colloid properties are modified by exposure to γ-irradiation. If there is a difference in surface properties, this may affect the overall macroscopic properties of the bentonite barrier. 4.1 Size fractionation of MX80 4.1.1 Size distribution The colloid size distribution of the isolated, size-fractionated clay suspensions was monitored by both PCS and AsFlFFF, as described in section 3.1.2.3. These measurements complement each other, since they separate the colloid sizes by different physical properties. In addition, the PCS-measurements are based on scattered laser intensity for the whole samples, whereas the largest colloids were not eluted in the AsFlFFFmeasurements. As the intensity is highly dependent on particle size, i.e. large particles result in higher scattered intensities, the mean hydrodynamic diameters from the PCS-measurements will be larger than those sizes given by AsFlFFF-measurements, which are dependent on the recovery of the measurements as mentioned above. The size distributions from the AsFlFFF-measurements were obtained from the elution time, calibrated to latex standards of varying sizes. For the conditions used in this work [60], the largest particles in the suspensions were moving too slowly in the channel to be detected. The loss of signal for the larger colloids in the AsFlFFF-measurements was reflected in the recovery for the measurements. This means that care should be taken when determining the mean size in colloid clay suspensions by these techniques. 4.1.1.1 Average particle sizes monitored by PCS The intensity-weighted and volume-weighted mean sizes of the clay suspensions obtained from PCS-measurements are presented in Table 10. Since the mean colloid size decreased with increasing number of 31 4. Results and Discussion centrifugation steps and force, this indicates that colloid size separation succeeded, even though the mean sizes were larger than expected (Table 4). In addition, the average count rates obtained from the measurements decreased with decreasing particle size when all suspensions were monitored at 10 mg/L clay colloids. Since larger particles result in higher scattered light than smaller particles, the higher count rates for the suspensions S0→S2 indicated a higher amount of large particles in these fractions. The sizes obtained from direct (S3.5UC) and sequential (S3.5) centrifuged suspensions under the same final centrifugation conditions were similar. Therefore, the preparatory centrifugation steps did not affect the final colloid size distribution in the suspensions. Table 10: Size distributions for the size-fractionated clay colloid suspensions obtained from PCS-measurements and the respective count rate measured in 10 mg/L colloid suspensions. PCSI - intensity weighted mean size; PCSV - volume weighted mean size Mean diameter (nm) PCSI PCSV Average count rate (kcps) S0 1452 ± 632 962 ± 225 118 ± 7 S1 513 ± 60 610 ± 57 87 ± 2 S2 404 ± 95 337 ± 29 53 ± 1 S3 S3.5 S3.5UC S3.5UC, FA 248 ± 28 181 ± 28 180 ± 39 185 ± 47 186 ± 64 172 ± 49 167 ± 50 143 ± 43 37 ± 1 20 ± 1 25 ± 1 29 ± 1 Sample 4.1.1.2 AsFlFFF Aluminium is one of the main elements in montmorillonite, so the Al signal was used as a fingerprint for the clay colloids in the AsFlFFF-measurements. In the Al fractograms, broad size distributions were present in all clay fractions ranging from 10 to 275 nm. However, the mean colloid size clearly decrease with increasing fractionation steps, Figure 10. A more defined separation of the different colloidal sizes was not obtained due to the conditions used during the measurements [60], supporting information. The recovery from the AsFlFFF-measurements is also an indication of the size distribution, since the measurement time was too short for elution of larger particles and therefore resulted in lower recovery. As shown in the 32 4. Results and Discussion legend to Figure 10, the recovery of smaller-sized colloids (i.e. S3→S3.5UC) was higher than that of larger-sized colloids. Consequently, the total size distributions of the smaller-sized particles are presented in Figure 10, whereas the larger-sized clay colloids were not apparent in the fractograms of the largest-sized clay suspensions. S0 (35 ± 4 %) 3 Al concentration (ng/s) S1 (62 ± 5 %) 2.5 S2 (77 ± 3 %) S3 (88 ± 5 %) 2 S3.5 (>90 %) S3.5UC (> 95 %) 1.5 S3.5UCFA (87 ± 10 %) 1 0.5 0 0 25 50 75 100 125 150 175 200 225 250 275 Hydrodynamic diameter (nm) Figure 10: Fractograms of monitored size distributions for the size-separated clay suspensions, as measured by AsFlFFF. Recovery of the sample is indicated in brackets. 4.1.1.3 Comparison of the two techniques The size distribution of particle suspensions monitored by different techniques may differ if the particles are non-spherical and the techniques are based on particle size separation by different physical properties. Most often, the equivalent spherical diameter (ESD) is given from particle size measurements. The ESD is defined either as the diameter of a sphere with the same volume as the monitored particle [75] or as “the diameter of a spherical particle which will give identical geometric, optical, electrical or aerodynamic behaviour to that of the particle (non-spherical) being examined” [76]. Clay colloids are known to be non-spherical [4], while their size distribution has to be further investigated. 33 4. Results and Discussion Particle sizes obtained from the volume-weighted PCS-measurements were frictional translator diffusion data and are denoted dT hereafter [75]. In contrast, the particle diameter given by AsFlFFF-measurements is the Stokes diameter, dS [77]. By assuming a geometry of the clay particles and comparing the PCS- and AsFlFFF-results, estimates of ‘real’ colloid sizes were obtained from the measured ESD values. Here, clay particles were assumed to be disc-shaped, for which previous works have described a mathematical relationship between the ESD for a translating particle (dT) and its corresponding ESD Stokes diameter (dS) [75]: δ δ = = ρ ρ ∙ ρ ρ ρ (Equation 8) (Equation 9) where ρ is the axial (or aspect) ratio (ρ = δ/t ), with δ as the diameter of the disc-shaped particle and t the thickness of the disc. From Equation 8 and 9, the axial ratio (ρ) can be obtained by the dT/dS-ratio [60]. If ρ, dT and dS are known, the diameter of the discs (δ) can be calculated, as can their thickness (t ). By assuming a given thickness of one basal layer in the colloid stack (t ), the amount of clay sheets can be calculated. In this thesis, t was assumed to be 1.3 nm, as given by XRDmeasurements (section 4.1.2) and the literature [78]. The ESD values from the PCS- and AsFlFFF-measurements are presented in Table 10 and Figure 10. The calculated sizes of the disc-shaped clay colloids in all suspensions are presented in Table 11. For details, see [60]. 34 4. Results and Discussion Table 11: Mean diameters (δ) and thickness (t ) of disc-shaped clay colloids and number of sheets in a clay stack calculated from PCS and AsFlFFF mean equivalent sphere diameter (ESD) and mathematical Equations 8 and 9, according to [75]. The number of layers was = 1.3 nm as the thickness of one clay sheet. calculated by taking t S0 66.3 Disc diameter δ (nm) 1496 S1 34.6 940 27 21 S2 10.3 500 48 37 S3 3.7 250 68 52 S3.5 14.4 258 18 14 S3.5UC 10.0 246 25 19 S3.5UC,FA 3.0 187 63 48 Sample Axial ratio (ρ) Disc thickness (nm) 23 Number of sheets 17 4.1.2 Characterisation of the colloids The element composition of the clay suspensions was measured with ICPOES, IC and XRD, as described in section 3.1.2. If the composition of the montmorillonite colloids was found to remain similar throughout the colloid size distribution, this could be a first indication that smaller clay colloids are miniatures of the larger ones. The concentration of the main elements in the montmorillonite colloids (i.e. Al, Mg, Si and Fe) decreased proportionally with increasing number separation steps. Therefore, the element composition of all colloids, independent of size, was assumed to be similar. The colloid concentration calculated from these element concentrations decreased from 1100 mg/L for S0 down to 90 mg/L clay colloids for S3.5UC. Interestingly, the montmorillonite suspension which was fractionated in the presence of FA had a higher clay colloid concentration (125 mg/L). The increase in clay colloid stability is explained by minor sorption of highly negatively charged FA, increasing the total surface charge of the colloids and thereby their stability [19, 51, 79, 80]. 35 4. Results and Discussion There was a direct release of cations and anions from the unpurified MX80 bentonite to the background electrolyte in all suspensions. The largest release was found for Na+ and SO42-, where the concentrations increased by a factor of 3 and 10, respectively, compared with the SGW. These elements were not associated to the clay colloids, since their concentrations remained unchanged throughout all suspensions. There was an increase in Ca concentration in the suspensions, which was associated with the amount of clay colloids in the suspension. Calcium is not generally one of the structural elements in montmorillonite. However, calcite is present in unpurified bentonite clay [45, 81, 82]. Therefore, Ca may be dissolved and, by ion exchange with Na at the basal plane, associated to the clay colloids. The ion exchange between Ca and Na also explains the large release of Na to the background electrolyte. To summarise, all size-separated montmorillonite suspensions had similar element composition and no changes in the composition were found between the colloid fractions (Table 12). Table 12: Element composition of the size-fractionated montmorillonite suspensions. The concentration of clay colloids was calculated from the Al-, Fe-, Si- and Mg-concentrations and compared to the structural formula. The amount of recovered colloids in the suspensions was defined as the amount of colloids in suspension relative to the initial suspension, S0. Suspension S0 S1 S2 S3 S3.5 S3.5UC S3.5UC, FA Suspension S0 S1 S2 S3 S3.5 S3.5UC S3.5UC, FA SGW [Al] mg/L [Mg] mg/L [Si] mg/L [Fe] mg/L [Coll.] mg/L 133 ± 20 88 ± 8 65 ± 4 33 ± 3 11.3 ± 0.6 10.6 ± 0.6 14.7 ± 0.3 [Ca] mg/L 8.6 ± 0.5 5.7 ± 0.1 4.0 ± 0.2 2.5 1.1 1.0 1.6 1.6 20 ± 2 13.6 ± 0.4 9.9 ± 0.6 5.5 ± 0.2 1.9 ± 0.1 1.8 ± 0.1 2.5 ± 0.1 [Na] mg/L 87 ± 1 82 ± 2 76 ± 2 77 73 74 75 28 382 ± 110 252 ± 70 178 ± 46 99 ± 25 40 ± 18 40 ± 18 51 ± 18 [F] mg/L 2.9 ± 0.5 2.9 ± 0.5 2.9 ± 0.5 3.0 ± 0.4 2.9 2.7 3.0 2.5 32 ± 2 21.4 ± 0.6 15 ± 2 8±1 2.7 ± 0.1 2.8 ± 0.3 3.5 ± 0.1 [SO4] mg/L 37 ± 3 37 ± 3 37 ± 4 37 ± 4 36 36 36 3.2 1127 ± 170 746 ± 68 551 ± 34 280 ± 26 96 ± 6 90 ± 6 125 ± 3 Recovered colloids (%) 100 ± 0 67 ± 4 48 ± 5 25 ± 2 9.4 ± 0.1 9.0 ± 0.2 12.1 ± 0.1 36 4. Results and Discussion XRD was used to monitor the mineral phase composition of the clay colloids. MX80, prior to any dilution into suspensions, was used as a reference for the bentonite. Minerals, such as cristobalite, halite and quartz, were present in the raw material and the largest-sized colloid suspensions (S0, S1 and S2), as well as their corresponding residuals. In all clay suspensions, intense reflections corresponding to montmorillonite were observed, confirming presence of montmorillonite (Figure 11). The XRD results confirmed that the structural composition of all montmorillonite colloids was similar, independent of their size. Figure 11: X-ray diffractograms obtained for (left) all residuals and (right) the respective supernatants following size fractionation of MX80 bentonite. 4.1.3 Estimations of edge site densities By approximating the clay colloids to discs with a given diameter and thickness, the amount of edge sites for each mean particle size could be estimated. Calculations of number of edge sites performed in this thesis are based on previous work by White and Zelazny [34] and Tournassat et al. 37 4. Results and Discussion [23]. Here, the number of aluminol (Al-OH) and silanol (Si-OH) groups was given by the perimeter to area ratio of the particles, together with the number of montmorillonite sheets stacked in one colloid and the structural formula of bentonite. From the particle sizes presented in Table 11, the number of sheets in one colloid was assumed not to affect the access to lateral surfaces (where the edge sites were assumed to be present). By assuming a clay density of 2.7 g/cm3, the number of edge sites was calculated by equations given by Tournassat et al. [23] and is presented in Table 13. Table 13: Number of accessible edge sites calculated by assuming disc-shaped clay colloids and the particle sizes presented in Table 11. The concentration of aluminol (nAl) and silanol (nSi) groups was determined by taking into account the total number of layers in the colloids, considering one clay layer. Number of edge sites (nTot) sum of aluminol + silanol groups. Sample S0 S1 S2 S3 S3.5 S3.5UC S3.5UC,FA nAl mmol/kg 4.9 7.8 15 29 28 30 39 nSi mmol/kg 6.3 10 19 38 36 38 50 nTot mmol/kg 11 18 34 67 64 68 89 A six-fold difference in number of edge sites was found between the smallest (S3.5UC) and largest (S0) sized clay fractions (Table 13). However, it should be borne in mind that the recovery from the AsFlFFF-measurements was low for the larger-sized clay colloids (i.e. S0 and S1), so the edge site densities were underestimated for these clay fractions. In addition, the approximation of the montmorillonite colloids to discs probably cause underestimation of the exposed surface area of the colloids and thereby the number of edge sites. These amounts of edge sites were used in interpreting the results of the sorption and sorption reversibility studies (sections 4.3 and 4.4). 38 4. Results and Discussion 4.2 Size effect on colloid stability 4.2.1 Latex particles To study a possible particle size effect on the agglomeration process, homogeneous, spherical and well-characterised polystyrene (latex) particles in the size range 20-500 nm were first used. Since the latex particles are monodispersed, the initial particle concentration in the agglomeration studies was known. The initial concentration is of great importance when studying the agglomeration kinetics in a system. In addition, latex particles with carboxylic functional groups have a surface charge close to the charge of montmorillonite colloids at low ionic strength and pH < 7. As mentioned in section 1.4.1, the DLVO-theory can be used for predictions of the total interaction energy between two spherical particles. At high ionic strength, the electrostatic repulsion between the two particles decreases towards zero due to screening of the electrical double layer from high amount of ions in solution. Therefore, the electrostatic repulsion is negligible at high ionic strength and the total interaction between the particles is dominated by van der Waals-attraction forces. This results in agglomerates being formed upon collisions between two particles. However, diffusion-controlled agglomeration is dependent on the kinetic energy of the particles, since this controls the collision frequency in the suspension. By comparing the total interaction energy and the kinetic energy for two spherical particles as a function of particle size, it was found that the kinetic energy dominated over the interaction energy, except at very short (approximately 10% of the particle size) separation distances (Figure 12). Hence, no difference in agglomeration rate due to initial particle size was expected. 39 4. Results and Discussion 0 100 200 300 400 Interaction energy /kT 0 -5 24nm Vtot 40nm Vtot -10 60mn Vtot 81nm Vtot -15 97nm Vtot 217nm Vtot -20 420nm Vtot 495nm Vtot -25 Separation (Å) Kinetic energy Figure 12: Total interaction energy (Vtot) calculated between two particles of equal size as a fuction of separation distance and particle size from DLVO-theory, and compared to the kinetic energy k T of the particles. The calculations were made at high ionic strength where no electrostatic repulsion between the particles is expected. Agglomeration rate constants were calculated from the increase in average hydrodynamic diameter as a function of time (described in section 3.2 and Equation 6). From Figure 12, the agglomeration rate constant was expected to be independent of initial particle size. By comparing the agglomeration rate constants to the initial particle concentration, it was found that the size dependence was an artefact and instead the agglomeration rate constants were dependent on the initial particle concentration (Figure 13). The rate constants obtained for the suspensions with low particle concentration (i.e. < 1016 particles/m3) were higher than the theoretical Smoluchowski rate constant, k = 1.22 ∙ 10 m3/s (Figure 13). These higher rate constants are probably related to the fact that PCS is more sensitive to larger agglomerates. Here, the agglomeration rates were taken as agglomeration of multiples and not aggregation of monomers, since the agglomeration was very fast. 40 4. Results and Discussion 1E-15 Agglomeration rate constant (m3/s) 24 nm 40 nm 60 nm 1E-16 81 nm 97 nm 217 nm 1E-17 420 nm 495 nm 1E-18 1E-19 1E+14 1E+15 1E+16 1E+17 1E+18 1E+19 Particle concentration (particles/m3) Figure 13: Agglomeration rate constants in diffusion controlled agglomeration as a function of initial particle concentration for different sizes of latex particles. Another way of studying the agglomeration process is to determine the fractal dimension, d , i.e. the porosity of the agglomerates formed, from a master curve, as described in section 3.2. In a previous study, Wu et al. [72] reported dependence of initial particle size on the d -value, which would imply that the structures of the agglomerates differ with particle size. On comparing our results with d -values from the literature, summarised by Wu et al. [72], it was found that they were in line with previous findings (Figure 14a). At d > 2, the agglomeration was assumed to be in the reaction-limited controlled agglomeration (RLCA) range and therefore negligible. However, when the d -values were instead plotted as a function of initial particle size, no correlation to particle size was found. Instead, the d -values increased with increasing initial particle concentration (Figure 14b). To conclude, no differences in the agglomeration behaviour between different sizes of particles were found with regard to agglomeration rate constant or in cluster density (d -value), as expected from theory. However, there was a large 41 4. Results and Discussion dependence on initial particle concentration. This indicates that the agglomeration mechanism was the same for all particle sizes. 3.0 a) 2.8 literature data this study (averaged over all results) 2.6 2.4 df 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0 100 200 300 400 500 600 Particle diameter (nm) 3.0 2.8 b) 2.6 2.4 df 2.2 2.0 1.8 1.6 1.4 1.2 1.0 1E+13 1E+15 1E+17 1E+19 Initial particle concentration (particles/m3) 1E+21 Figure 14: Fractal dimension (d ) -values obtained from master curves for experimental data (open squares) and literature data from Wu et al. [72] (filled diamonds) as a function of a) initial particle size and b) initial particle concentration. 42 4. Results and Discussion 4.2.2 Montmorillonite colloids From the agglomeration study of latex particles, no differences in stability properties of montmorillonite colloids by size were expected, since the chemical composition of the clay colloids was the same for all particle sizes (section 4.1.2). As the particle concentration in the montmorillonite suspensions was unknown, no agglomeration rate constants could be determined. When constructing a master curve for montmorillonite colloids, the clay colloids disservice the universality of agglomeration, as found for illite colloids [83]. From the slope of the master curve the z/d -ratio was 0.28, indicating that z ≠ 1 and no d -values could be determined. Therefore, no conclusions regarding particle size effects on cluster density could be drawn for montmorillonite colloids. Hence, the assumption of non-sphericity of the clay colloids in section 4.1.1 is supported by the finding that z ≠ 1, since this indicates that clay colloid agglomerates have different conformations, such as card-house structures. However, relative stability, W, was determined by comparing agglomeration at an ionic strength to the fastest agglomeration rate obtained in that electrolyte (Equation 7). Due to the polydispersivity of the clay suspensions, the agglomeration data were scattered. There were no differences in stability ratios between different sizes of clay suspensions (Figure 15). Calcium ions in the suspensions decreased the stability ratio more than Na ions at the same concentrations. This is explained by the fact that divalent ions affect colloids more efficiently than monovalent ions, from the Schulze-Hardy-rule [84]. The similarities in stability ratios for all clay suspensions are in line with the theory in section 4.2.1 where it was concluded that the initial particle size does not affect agglomeration behaviour. No effect of colloid size on the stability properties of the particles has been found in previous studies [66, 85], but here this was explained by limitations in the instruments. 43 4. Results and Discussion a) 20 25 S1 S2 20 S3 15 W S3.5 S3.5,UC 10 S3.5,UC-FA W 25 S1 b) S2 S3 S3.5 15 S3.5,UC 10 S3.5,UC-FA 5 5 0 0 0.01 0.1 1 Ionic strength [M] 0.03 3 0.3 Ionic strength [M] 3 Figure 15: Relative stability ratio, W, for all size-fractionated clay suspensions as a function of ionic strength in a) NaCl and b) CaCl2. The agglomeration was not measured for suspension S3.5 at 0.01 M NaCl. The long-term stability of the clay colloids was monitored in all clay suspensions by determining the concentration of suspended colloids after two months without shaking. Larger particles, present in the suspensions S0 and S1 (Table 10) sedimented over time [60], whereas the smaller-sized colloid fractions remained stable. To summarise, no differences in agglomeration behaviour induced by addition of electrolyte were observed for the size-separated clay colloids. Even though universal agglomeration was not obtained for the clay colloids [86], this is in line with the agglomeration behaviour of latex particles of varying sizes. The lack of size effect is consistent with the composition of the clay colloids remaining the same in all suspensions (Table 12 and Figure 11) and with the overall physical properties remaining the same for all colloid sizes. As expected, larger clay particles continued to sediment with time, since complete sedimentation was not achieved before the first fractionation. In the case of transport of clay colloids in bedrock fractures, similar stability properties of all particle sizes would allow the particles to be transported under the same chemical conditions present in the groundwater. Consequently, the particles could be separated by size, as described in section 1.4.2. If the stability differed between particle sizes, the size 44 4. Results and Discussion separation would be suppressed by the more stable particles being transported further, even though their conformation is not the most favourable in the prevailing water flow. Sedimentation of larger clay colloids would be present, but these could be re-dispersed if the conditions changed, e.g. higher water velocity, during transport. 4.3 Sorption properties of RNs and analogues onto montmorillonite colloids 4.3.1 Effect of clay colloid size 4.3.1.1 RNs in the higher oxidation states, Np(V), U(VI) and Tc(VII) The three RNs Np, U and Tc were mainly present as NpO2CO3-, UO2(CO3)34and TcO4- in the sorption studies performed [87-93]. The negatively changed complexes and the high affinity to carbonate for Np and U predict low sorption of these RNs onto montmorillonite. By comparing the concentrations of the RNs in suspension before and after ultracentrifugation, it was found that their concentrations remained constant, indicating no sorption of the RNs onto the montmorillonite colloids. The absence of sorption of the RNs is expected from previous studies, due to the low mass to liquid ratio of clay colloids. The low concentration of montmorillonite predicts sorption of 1% for U [89] and even less for Np since KD values < 102 dm3/kg have been reported [56, 94, 95]. This low RN uptake was not detectible within the analytical uncertainty in this thesis. Tc(VII) is known not to sorb onto minerals [87, 93]. In addition, no reduction to Tc(IV) was found since TcO2(s) was not stable in the SGW [96]. The absence of sorption of Np, U and Tc onto montmorillonite was found for all suspensions investigated (Table 10). For the suspension which was isolated in the presence of FA during centrifugation (S3.5UC,FA), FA is negatively charged at pH 9 and therefore no sorption of FA onto montmorillonite was expected [51, 80, 97]. Since FA will remain soluble in the suspensions, no effect on non-sorbing RNs was expected or monitored. 45 4. Results and Discussion 4.3.1.2 Tetravalent RNs Th(IV) and Pu(IV) The tetravalent RNs studied, i.e. Th and Pu, sorbed to a high extent onto all size fractions of montmorillonite. As expected from the previous long term stability measurements of the suspensions (Table 10), the larger sized clay colloids sedimented over time. The sorbed Th and Pu remained sorbed onto montmorillonite. This was seen by comparing the amount of sedimented concentrations of Al to the Th or Pu concentrations. From the large decrease in concentration before and after ultracentrifugation, approximately 98% of the initial Th concentration was associated to montmorillonite. Thorium was probably sorbed as ternary Thcarbonate-montmorillonite complexes (Figure 16). This corresponds to a log KD value of 6.4 (dm3/kg). No differences in sorption capacity between the different sized montmorillonite suspensions were observed. The amount of sorbed Th remained constant over the total sorption time investigated. Plutonium sorbed initially (after three days of sorption) to 87%, which increased to 98% sorbed Pu by the end of the sorption experiments (Figure 16). This corresponded to an increase in log KD value (dm3/kg) from 5.5 ±0.5 up to 6.3 ±0.5, which was not significant. The higher amount of sorbed Pu was similar to the amount of sorbed Th, which indicates that several tetravalent RNs have similar sorption affinities, even if the dynamics in the sorption process differ. As mentioned in section 4.3.1.1, FA present in the colloid suspension S3.5UC,FA will not be sorbed onto the montmorillonite surfaces, and will instead act as a competing ligand. In the presence of FA, the sorption of Pu and Th onto montmorillonite was significantly less. The associated Th was 15% less and Pu 10% less, compared with the sorption samples in the absence of FA. No increase in sorbed Pu was observed for S3.5UC,FA. The log KD values (dm3/kg) obtained for Th and Pu were 5.3 ±0.5 and 5.1 ±0.5, respectively. From the concentration of FA and montmorillonite added in the samples, these KD values are similar to those obtained in previous studies [51, 93, 97, 98]. 46 4. Results and Discussion S0 Th-232 S0 Pu-242 7 6 5 4 3 2 1 0 120 100 80 60 40 20 0 3 days 7 6 5 4 3 2 1 0 120 100 80 60 40 20 0 2 weeks 1 month 6 months 3 days S3.5 Th-232 7 6 5 4 3 2 1 0 120 80 60 40 20 0 2 week 80 60 40 20 0 1 month 6 months 3 days 100 80 60 40 20 0 7 6 5 4 3 2 1 0 100 80 60 40 20 0 1 month 6 months 3 days %Colloid 242 1 month 6 months 120 7 6 5 4 3 2 1 0 %Part 2 week S3.5uc,FA Pu-242 120 2 week 7 6 5 4 3 2 1 0 120 100 S3.5uc,FA Th-232 3 days 1 month 6 months S3.5 Pu-242 100 3 days 2 week %Free 232 2 week 1 month 6 months Kd-values Figure 16: Distribution of Pu(IV) and Th(IV) in the presence (20 mg/L) of different-sized montmorillonite colloids in synthetic ground water (SGW) in the presence and absence of FA. The x-axis represents the sorption time. The left y-axis represents the percentage of particulate (%Part, blue), colloid bound (%Colloid, red) and free (%Free, green) radionuclides in solution. 3 The right y-axis is the corresponding log KD value (dm /kg) (purple line). 47 4. Results and Discussion No differences in sorption capacity for Th and Pu onto montmorillonite colloids of varying sizes were detected. An estimated six-fold difference in the sorption was found in the sorption samples (Table 13). By using the estimated amount of sites for the different size fractions, normalisation of the sorption to number of sites instead of mass was possible. Given a sixfold difference for the smallest- and largest-sized colloid fractions, a difference of log 6 = 0.8 in the log KD values was expected. This small difference in sorption capacity could not be detected within the experimental uncertainty. 4.3.2 Effect of γ-irradiation γ-irradiation influence the properties of bentonite and montmorillonite colloids, by affecting colloid stability and Fe2+/Fe3+-ratio [22, 28]. If the structure in montmorillonite colloids is changed upon radiation, this may affect the sorption properties of the clays. Sorption of Co2+ and Cs+ onto γirradiated and unirradiated clays (as references) is presented as massnormalised KD values in Figure 17. There was a large difference in sorption capacity between the clays, with the Na-montmorillonites sorbing significantly less than WyCa and MX80. Therefore the irradiation effect was compared for each clay separately. 48 4. Results and Discussion Sorption capacity KD (dm3/kg) 16000 a) Co2+ 14000 12000 SWyNaw 10000 SWyNaw + Ca 8000 WyNa 6000 WyCa 4000 MX80 MX80 + Ca 2000 0 Dry, Dry, Wet, Wet, N2, - O2, - N2, - O2, - Wet, Wet, No, No, N2, O2, N2, - N2, Hum Hum Hum Sorption capacity KD (dm3/kg) 700 b) Cs+ 600 500 SWyNaw 400 SWyNaw + Ca WyNa 300 WyCa 200 MX80 MX80 + Ca 100 0 Dry, N2, - Dry, O2, - Wet, N2, - Wet, O2, - Wet, Wet, No, No, N2, O2, N2, - N2, Hum Hum Hum 2+ + Figure 17: Sorption capacity (KD values) for a) Co and b) Cs onto γ-irradiated or unirradiated clays. The γ-irradiated clays were irradiated under wet or dry conditions, as indicated. The unirradiated clays are denoted by No. The atmosphere in the samples was N2 or air (denoted O2). Hum indicates presence of humic substances (HS). ‘+ Ca’ indicates addition of calcium to the samples. 49 4. Results and Discussion There was a significant effect of γ-irradiation on the sorption capacity of Co2+ for all clays, with less sorbing onto the clays after γ-irradiation, except in the washed Na-montmorillonite (SWyNaw). This trend was more pronounced for clays with a high affinity to C02+. No significant effect of γirradiation on sorption of Cs+ was found. It is known that Co2+ is mostly sorbed onto montmorillonite by surface complexation onto specific surface sites, while Cs+ interacts with the clay mainly by cation exchange. Therefore, the effect on C02+ sorption may indicate that γ-irradiation affects the specific sites, whereas the overall CEC of the clay colloids remains unchanged. The SWyNaw clay was pretreated with H2O2, which is one of the main products in radiolysis of water. This may be one reason why the sorption capacity for this clay was not affected by γ-irradiation. It is known from the literature that Si is dissolved from the montmorillonite structure at the edges of the sheets during exposure to γ-irradiation [26, 99]. Assuming that Al-OH and Si-OH are the main sorption sites for higher valence cations [36, 38], this is supported by the present results. Interestingly, the stability of montmorillonite colloids, and thereby their charge, was previously found to increase upon γ-irradiation. A higher colloid charge could be expected to lead to higher sorption of cations, but the opposite was found in this study. The change in sorption capacity between irradiated and reference clays did not affect the diffusivity experiments [61]. This means that the change in sorption has to be larger to affect the overall diffusion of RNs through bentonite clay. Addition of Ca2+, to compare the Na- and Ca-montmorillonites, decreased sorption of Co2+ onto MX80, as expected due to competition between the two divalent cations. However, the sorption increased for Co2+ onto SWyNaw, which cannot be explained. A second study of sorption capacity of divalent cations onto γ-irradiated bentonite was performed using Cu2+. Here, the effect of γ-irradiation on the sorption was studied as a function of total dose (Figure 18). As for sorption of Co2+, sorption of Cu2+ decreased upon irradiation, although the effect was not very dramatic [100]. The decrease in sorption capacity was most pronounced up to a dose of 100 kGy. This result confirms that a change in sorption capacity upon γ-irradiation occurs for bentonite and has to be considered when including sorption capacity in reactive transport models. 50 4. Results and Discussion 700 600 KD (dm3/kg) 500 400 300 200 100 0 0 50 100 150 200 250 Total dose (kGy) 2+ Figure 18: Sorption capacity of Cu as a function of total dose of γ-radiation for bentonite prior to sorption studies. 4.4 Sorption reversibility 4.4.1 Introducing organic matter as competing ligand 4.4.1.1 Effect for size-fractionated clay colloids Fulvic acid acted as a competing ligand to RN sorption onto montmorillonite, decreasing the sorption of both Pu and Th, since the majority of FA was not sorbed onto montmorillonite under the conditions used in this thesis [51, 80, 97]. When FA was added to the size-fractionated sorption samples, the amount of reversible Th was approximately 20% (Figure 19), reported previously [50, 51]. No effect of particle size was found [101]. For Pu, the amount of FA-complexed Pu was dependent on the sorption time prior to addition of FA. The reversibility was greater at shorter sorption time (Figure 19). One explanation may be that Pu sorbs to less assessable sorption sites over time, decreasing the reversibility with longer sorption times. The interactions between FA and RNs have previously been categorised as reversible, pseudo-irreversible and irreversible depending on the equilibrium time needed [52, 53]. The reversible cation interactions occur rapidly, whereas the site-specific interactions, which occur after a 51 4. Results and Discussion longer contact time between the RNs and FA, are classified as irreversible. Interactions of Th and FA have faster kinetics in the first 10 days, followed by a slower sorption process over several months [53]. This may be the case for Pu as well. Comparing the CEC values for RNs to montmorillonite colloids and FA, assuming a CEC of 0.75 meq/g for montmorillonite [45] and 6.82 meq/g for FA [62] gave a two-fold higher CEC per unit volume for FA than for montmorillonite (0.035 meq/L compared with 0.015 meq/L). This would result in most RNs which interact via cation exchange being favoured to bind to the FA. Previous studies of sorption competition between other minerals and organic matter have found that the CEC has to be three-fold larger than that of the mineral to affect Th significantly [50], which may explain the large association of Th to the montmorillonite colloids in this thesis [101]. Thorium 120 Plutonium 120 100 100 80 80 60 60 40 40 20 20 0 3 days 2 week 1 month Sorption time 0 6 months 3 days 120 120 100 100 80 80 60 60 40 40 20 20 2 week 1 month Sorption time 6 months 0 0 3 days 2 week 1 month 6 months Sorption time prior to addition of FA 232 242 3 days 2 week 1 month 6 months Sorption time prior to addition of FA Figure 19: Amount of Th (left) and Pu (right) in their respective phase where the y-axis represents the percentage of particulate (%Part, blue), colloid bound (%Colloid, red) and free (%Free, green) radionuclides in solution in the sorption samples (upper figures [102]) and after one year in contact with 5.1 mg/L fulvic acid (FA) (lower figures [101]). The five bars for each UC sorption time are (left to right): S0, S1, S2, S3 and S3.5 . 52 4. Results and Discussion 4.4.1.2 Presence of HS in γ-irradiated montmorillonite suspensions On adding organic matter as a competing ligand to the sorbing Co2+ and Cs+ in the samples before γ-irradiation, the interactions between the cations, HS and clay may be different from those in unirradiated samples due to decomposition of HS by γ-irradiation [103, 104]. The experimental results obtained in this thesis showed that presence of HS in the samples decreased the sorption capacity of Co2+ more than in the γ-irradiated clays (Figure 17). However, sorption of Cs+ was not significantly affected by the presence of HS, which can be explained by its lower sorption capacity. An effect of γirradiation on the Co-HS complexes was not found. This indicated that even if HS decomposes under γ-irradiation, its lower molecular units are still charged and interact with cations in a similar way to the original HS. This suggests that the main functional groups in HS, which are active in cation interactions, remain unaffected by γ-irradiation. 4.4.2 Decrease in pH All samples were buffered back from pH 5.6 to 7.5 after one week due to the presence of carbonate in the SGW, which has good buffering capacity [105]. Uranium-carbonate complexes were less soluble at lower pH, whereas approximately 80% U was sorbed onto the montmorillonite after one week at pH 7.5. No differences in sorption capacity for the different-sized montmorillonite colloids were found, as in the sorption samples. No sorption of U onto montmorillonite occurred for the suspension S3.5UC,FA, since FA strongly interacts with uranyl under the chemical conditions provided [93]. When additional acid was added to lower the pH further, the final pH obtained was pH 2.7. In these samples, U was once again free in solution, since montmorillonite surfaces are positively charged under these conditions [80]. Thorium, which was associated to montmorillonite to 98% prior to the decrease in pH, was dissolved to 80% at pH 2.7. The amount of free Th is in line with the literature where thorium is predicted to be present as Th(OH)3+, which will not sorb to the positively charged montmorillonite [106, 107]. Reversibility of Th sorption was found for all clay suspensions, independent of colloid size. In contrast to Th, all Pu was in particulate form at pH 2.7 after one year. Montmorillonite colloids were unstable at pH 2.7, 53 4. Results and Discussion whereas all sorbed Pu remained sorbed onto the sedimented clay. In addition, the amount of free Pu prior to the decrease in pH was found among the particles. This can be explained by slow reduction of Pu(V), present initially, to Pu(IV). Another explanation may be that a small amount of Pu eigen-colloids may be formed under the conditions applied [108, 109]. 4.4.3 Increase in ionic strength After addition of 0.5 M ionic strength CaCl2, all clay colloids became unstable and sedimented in the samples, as expected from the stability studies of montmorillonite reported in section 4.2.2. Both Th and Pu remained sorbed onto the sedimented montmorillonite. The amount of RNs associated to FA in fraction S3.5UC,FA was also in the particulate phase in the samples. This can be explained by FA being unstable at this ionic strength too [79, 110]. No effect of colloid size was found in these suspensions. 4.4.4 Addition of crushed bedrock material (CBM) On addition of crushed bedrock material to the sorption samples, Np and Tc interacted with the solid phase and 41% of Np and 32% of Tc was in the particulate phase after one week’s equilibrium time (Figure 20). After one year, 97% of Np and all Tc were in the particulate phase (Figure 20). These processes are probably a result of reduction of Np(V) to Np(IV) and Tc(VII) to Tc(IV), where the eigen-colloids NpO2(s) and TcO2(s) are formed. The tetravalent RNs may also sorb onto grandiorite. One explanation for the reduction is the Fe present in grandiorite, which is known to reduce technetium [55, 111]. Interestingly, no, or very little, Tc was reduced in the presence of fracture filling material (FFM) in previous studies [57], whereas fast reduction has been found in the presence of grandiorite [56]. The differences in reactivity between grandiorite and FFM have been attributed to the fact that grandiorite contains more iron [56]. No influence of colloid size of the montmorillonite was found. In the suspension S3.5UC,FA, approximately 11% of Tc was complexed to the FA, instead of interacting with the CBM. Complexation between FA and Tc has been found in previous studies in the absence of grandiorite [112-114], and is therefore a probable explanation why a small amount of Tc remained 54 4. Results and Discussion free in the solutions. This indicates that FA is a strong competing ligand to CBM. Technetium 120 100 100 80 80 60 60 40 40 20 20 0 3 days 120 2 week 1 month Sorption time 6 months 0 3 days 120 1 week CBM 100 100 80 80 60 60 40 40 20 20 0 0 3 days 2 week 1 month 6 months Sorption time prior to addition of CBM 120 Neptunium 120 3 days 2 week 1 month 6 months Sorption time prior to addition of CBM 100 100 80 80 60 60 40 40 20 20 0 0 3 days 2 week 1 month 6 months Sorption time prior to addition of CBM 99 237 6 months 1 week CBM 120 1 year CBM 2 week 1 month Sorption time 1 year CBM 3 days 2 week 1 month 6 months Sorption time prior to addition of CBM Figure 20: Amount of Tc (left) and Np (right) in their respective phase where the y-axis represents the percentage of particulates (%Part, blue), colloid bound (%Colloid, red) and free (%Free, green) radionuclides in solution in the sorption samples (top figures [102]) and after one week or one year in contact with crushed bedrock material (CBM) [101]. The six bars at UC UC,FA . each sorption time are (left to right): S0, S1, S2, S3, S3.5 and S3.5 55 4. Results and Discussion From the literature, it is known that uranium sorbs onto grandiorite, with a log KD value of 1.8 (dm3/kg) for fractions < 1160 µm and one week of sorption time [56]. This KD value corresponds to the amount of sorbed 233U after one week of contact time with CBM in this thesis, where 44% of 233U sorbed. After one year, most 233U was released as free ions in the suspensions. One explanation for this may be that natural 238U was released from the CBM at a higher concentration than that of the added 233U. The excess 238U may have led to cation exchange with the sorbed 233U, resulting in 233U being released in suspension. The large amount of 238U release was because uranium interacts with carbonate to form U-carbonate complexes. Both Th and Pu remained sorbed onto the montmorillonite colloids, which partly sedimented in the presence of CBM. In addition, the remaining, previously free, part of Pu sorbed onto the CBM, resulting in all Pu being in the particulate phase. The additional sorption of Pu may occur due to reduction of a small amount of Pu(V) initially present to Pu(IV) by the Fe known to be present in high amounts in crushed grandiorite [56]. 4.5 γ-irradiation-induced copper corrosion in the presence of bentonite When copper cubes were irradiated in the presence of bentonite, the free Cu concentration in the aqueous phase increased with dose (Figure 21a) as previously shown for copper cubes irradiated in pure anaerobic water [115]. However, the concentrations were considerably lower than in presence of bentonite, owing to the high sorption of Cu onto bentonite as shown in Figure 18. To determine the amount of sorbed Cu, the bentonite slurry was acidified in order to desorb any Cu it had immobilised. The total concentration of Cu released from the cube is presented in Figure 21b. The same trend of increasing Cu concentration with dose was found in both cases (Figure 21a and b). The total concentration of Cu was approximately five-fold higher in the presence of bentonite than earlier found in water [100]. Hence, it would appear that the effect of adsorption of Cu onto bentonite is a significant driving force for the corrosion. 56 4. Results and Discussion 600 a) Concentration Cu (µM) Concentration Cu (µM) 0.40 0.30 0.20 0.10 0.00 0 50 100 150 Total dose (kGy) b) 500 400 300 200 100 0 0 50 100 150 Total dose (kGy) Figure 21: Release of Cu from irradiation of copper cubes under anaerobic conditions in the presence of 4 wt% bentonite. a) Cu in aqueous phase b) Total concentration of Cu in samples. 4.6 General discussion There were no significant differences in element composition, stability or sorption capacity between the montmorillonite colloid sizes studied. The similarities found in stability criteria indicate that smaller colloids are miniatures of larger colloids. If the element composition or surface charge varied by size, this would influence the Hamaker constant and the electrostatic interactions. Thereby, the stability would also be affected. Given the similar chemical composition of all colloids, similar sorption site density per area can be expected. The edge sites are assumed to be present at the lateral surfaces [23]. The number of edge sites for a colloid suspension would then be given by the amount of lateral surface. By assuming discshaped colloids, the lateral surface was estimated here from the mean colloid size. The lateral surface was approximately six-fold higher for the smallest size fraction in this thesis compared with the largest. Within the experimental uncertainty, this is not a sufficiently large difference to be detected for highly sorbing radionuclides. 57 4. Results and Discussion The remaining question is whether there is a small difference in sorption capacity between smaller and larger colloids which cannot be detected, or whether the distribution of sorption sites changes slightly with colloid size. Bentonite is a natural clay and assuming all colloids to be disc-shaped is a rough approximation. The surface roughness and the inner surfaces in the porous colloids may influence sorption so that even smaller differences between larger and smaller colloids would be expected. Even though montmorillonite colloids can be separated by size initially, a change in size distribution over time is expected. Larger colloids will sediment and there may be a slow agglomeration process present in the suspensions for the smaller-sized colloids. A more homogeneous distribution of colloid sizes would also give an even smaller difference in sorption capacity. More studies of the effects of γ-irradiation on clay surface properties have to be performed to confirm whether the experimental sorption capacities reported here are valid. A significant decrease in sorption capacity for bentonite after exposure to γ-irradiation would call for more refined reactive transport models. 58 Conclusions Conclusions From the studies conducted within the framework of this thesis, the following conclusions can be drawn: Dependence on particle size - A protocol was developed and successfully applied to fractionate montmorillonite colloids into different-sized suspensions by centrifugation. This protocol can be applied to other polydisperse systems where the initial colloid size distribution cannot be controlled. All montmorillonite colloid sizes have the same element composition. - No effect of particle size could be detected in agglomeration behaviour of either size-separated montmorillonite colloids or homogeneous and well-defined polystyrene (latex) particles. - Thorium and plutonium sorbed onto all montmorillonite colloids, independent of colloid size, in synthetic groundwater. - No dependency on colloid size was found in the sorption reversibility studies. Effect of γ-irradiation - Irradiated bentonite sorbed significantly less surface-complexed, divalent cations than non-irradiated bentonite. Cation exchange was not affected by irradiation. - Presence of bentonite enhanced corrosion of copper cubes upon γirradiation under anaerobic conditions. 59 Conclusions General conclusions - Average sorption capacities can be used in reactive transport models, as they are valid for the whole size distribution of montmorillonite colloids. No dependency on exposed lateral surfaces of the colloids has to be taken into consideration. This facilitates model use. - The influence of γ-irradiation should be considered in reactive transport models since divalent cations sorb less to irradiated bentonite. 60 Further work Further work To the best of my knowledge, this thesis presents some of the first studies of how the colloid size of bentonite affects transport of radionuclides in the near-field of a planned repository for spent nuclear fuel. Based on the results, the experimental design of sorption studies onto size-fractionated clay colloids can be further developed. Firstly, suspensions including smaller sized montmorillonite colloids with a concentration relevant for sorption studies of radionuclides are desirable. In addition, narrower-range suspensions with respect to colloid size, where smaller colloids are not present in all suspensions, would probably give a clearer trend in sorption capacity, if present. In order to see whether the colloids continue to differ in size, the equilibrium colloid size of montmorillonite in suspension after a longer time (over years) would be interesting to study. These studies of the effects of colloid size on sorption should be complemented with sorption studies of trivalent RNs, which were not included in this thesis. Trivalent RNs are interesting since they can be released from spent nuclear fuel and have different chemistry than tetravalent cations. Furthermore, use of higher concentrations of RNs in experiments would be desirable if radiation protection allows, since the measurement uncertainty would be lower. Field-flow fractionation measurements in the sorption studies would give the RN distribution with colloid size in the suspensions. RNs which sorb to a lower extent onto clay are interesting for use, since these RNs would give less experimental uncertainty. These modifications would also be interesting for sorption reversibility samples. For the sorption reversibility samples, more data points over time would be needed to monitor the reversibility kinetics. Moreover, classical desorption studies of different initial sizes of montmorillonite colloids would give valuable information for reactive transport models. Several complementary techniques, such as static light scattering, can be used to determine the agglomeration rate constants for different-sized colloids. In suspensions with narrower size-fractionated montmorillonite colloids, the initial particle concentration would be better estimated and agglomeration rate constants for these colloids could also be estimated. 61 Further work The effect of γ-irradiation of clays in the presence of metallic copper needs more study, since no studies on γ-irradiation-induced copper corrosion or γirradiation effects on bentonite have previously been reported. In addition, the effect of γ-irradiation on humic substances is currently unknown and may affect the equilibrium between sorption onto clays and organic matter. Many of the main objectives in this thesis are interesting for other similar systems, such as transport of metal ions or organic pollutants in contaminated soil. There the metal concentrations are often higher, the groundwater conditions different and the time scale shorter. In such cases, dependence of sorption capacity related to exposed surface area of the soil particles would be an interesting topic for study. 62 List of abbreviations and symbols List of abbreviations and symbols Symbol %Colloid %Free %Part [X] A AsFlFFF BET CBM CEC δ d d d dT dS DLCA DLS ϵ ϵ e ESD f FA FFM h HS I I I IC ICP-MS ICP-OES κ k k k n n Explanation Amount of RNs associated to clay colloids in sample Amount of RNs free in suspension Amount of RNs associated to particulates in sample Initial concentration of particles X present in sample Hamaker constant Asymmetric flow field-flow fractionation Brunauer-Emmett-Teller surface Crushed bedrock material Cation exchange capacity Diameter of a disc-shaped particle Hydrodynamic diameter obtained from DLS Fractal dimension Initial diameter of a particle ESD for a translating particle ESD Stokes diameter Diffusion limited controlled agglomeration Dynamic light scattering Permittivity of the material Permittivity of vacuum Charge of the electron Equivalent spherical diameter Suffix for fast agglomeration rate Fulvic acids Fracture filling material Surface-to-surface separation between two particles Humic substances Ionic strength Intensity of a singlet Intensity of a doublet Ion chromatography Inductively coupled plasma-mass spectrometry Inductively coupled plasma-optical emission spectroscopy Inverse of Debye-Hückel screening length Agglomeration rate constant Boltzmann constant Smoluchowski’s rate constant of agglomeration Number concentration of electrolyte Number of aluminol groups 63 n n NNLS ρ PCS PCS PCS r r r RLCA RN s SEM SGW SI t T t t t TOC UC UV-Vis. V V V W XRD ψ z Number of silanol groups Total number of aluminol and silanol groups Non-negatively constrained least squares Axial (or aspect) ratio Photon correlation spectroscopy Intensity-weighted mean size from PCS-measurement Volume-weighted mean size from PCS-measurement Radius of a spherical particle Hydrodynamic radius of a singlet Hydrodynamic radius of a doublet Reaction-limited cluster aggregation Radionuclide Centre-to-centre separation between two particles Scanning electron microscopy Synthetic groundwater Supporting information Time Absolute temperature, Thickness of a disc-shaped particle Thickness of one basal layer in the colloid stack Characteristic time of the Brownian aggregation process Total organic carbon Ultra-centrifugation Ultraviolet–visible spectroscopy Electrostatic repulsion van der Waals attraction Interaction energy Stability ratio X-ray diffraction Diffuse layer potential Valence of the electrolyte 64 Acknowledgements Acknowledgements First, I would like to express my deepest gratitude to my supervisors Assoc. Prof. Susanna Wold and Prof. Mats Jonsson for accepting me as a PhD student in your group and giving me the opportunity to perform this work. Your support along the way gave me the possibility to grow as a researcher. I would also like to gratefully thank Dr. Muriel Bouby for giving me the opportunity to work with you at KIT-INE, Karlsruhe, Germany, during a total period of 8 months. I cannot express my gratitude enough for your supervision in the lab, our interesting discussions, your encouragement and your friendship. I wish to thank my co-authors for interesting discussions and fruitful collaborations. I would also like to extend my appreciation to my additional colleagues and friends at INE: Johannes, Nicolas, Stephanie, Thorsten, Horst, Florian, Tom, Daniel, Patric, Aurélie, Cornelia and especially Rémi. Thank you for sharing your expertise, for your warm welcome, nonscientific discussion during lunch time and making Karlsruhe into my second home. To all my present and former colleagues at Applied Physical Chemistry at KTH, namely: Michael, Amanda, Åsa, Inna, Kristina, Yang, Alex, Mats, Johan, Gabor, Björn, Joakim, Camilla, Istvan, Anders, Martin, Veronica, Sandra, Helena, Sara, Claudio and Marianne. Thanks for making me looking forward to my working days and always being available to help out in the lab. Our board game evenings, after-works, trips etc. will remain as good memories for a very long time. Prof. Per Claesson is acknowledged for reviewing the thesis and making valuable comments. Thanks to Joakim, Sara and Björn for giving feedback to my thesis. Lena Skowron, Kenneth Andersson, Susanne De Meyere, Ilona Mozsi, Vera Jovanovic and Dr. Ulla Jacobsson have been very helpful 65 Acknowledgements with all administrative work during my time at Applied Physical Chemistry, KTH. The Swedish Nuclear Fuel and Waste Management Co. (SKB) is gratefully acknowledged for financial support to this project. Moreover, part of the work was funded by the project CP-BELBaR Fission 2010-1.1.1. Additional financial support for conference scholarships and travel grants from Ångpanneföreningens Forskningsstiftelse, T&E network activity in TALISMAN and Klasons stiftelse is gratefully acknowledged. Last, this work was supported by the European FP7 TALISMAN project, under contract with the European Commission. My last stay in Karlsruhe would not have been possible without this support. Without all my friends from KTH, Kårspexet, orchestras etc., this work would never have been finished. You are my never-ending source of joy. To my family, for all the patience and support they have given during this time. To my parents, for always being by my side and believing in me. To Anders, who teaches me every day to never give up. 66 References References [1] S.N.F.a.W.m.c. (SKB), The interim storage facility for spent nuclear fuel, in, 2014. [2] SKB, Design and production of the KBS-3 repository, SKB internal report TR-10-12, (2010). [3] D.D. Macdonald, S. Sharifi-Asl, Is Copper Immune to Corrosion When in Contact With Water and Aqueous Solutions?, Swedish Radiation Safety Authourity, SSM, Report 2011:09, (2011). [4] F. Bergaya, B.K.G. Theng, G. Lagaly, Handbook of Clay Science, 2006. [5] A. Möri, W.R. Alexander, H. Geckeis, W. Hauser, T. Schafer, J. Eikenberg, T. Fierz, C. Degueldre, T. Missana, The colloid and radionuclide retardation experiment at the Grimsel Test Site: influence of bentonite colloids on radionuclide migration in a fractured rock, Colloids and Surfaces a-Physicochemical and Engineering Aspects, 217 (2003) 33-47. [6] W. Stumm, Aquatic colloids as chemical reactants: surface structure and reactivity, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 73 (1993) 1-18. [7] H. Geckeis, T. Schäfer, W. Hauser, T. Rabung, T. Missana, C. Degueldre, A. Möri, J. Eikenberg, T. Fierz, W.R. Alexander, Results of the colloid and radionuclide retention experiment (CRR) at the Grimsel Test Site (GTS), Switzerland – impact of reaction kinetics and speciation on radionuclide migration, Radiochim. Acta, 92 (2004) 765-774. [8] T. Missana, Ú. Alonso, M.J. Turrero, Generation and stability of bentonite colloids at the bentonite/granite interface of a deep geological radioactive waste repository, Journal of Contaminant Hydrology, 61 (2003) 17-31. 67 References [9] S. Garcia-Garcia, C. Degueldre, S. Wold, S. Frick, Determining pseudoequilibrium of montmorillonite colloids in generation and sedimentation experiments as a function of ionic strength, cationic form, and elevation, Journal of Colloid and Interface Science, 335 (2009) 54-61. [10] G.H. Brown, Glacier meltwater hydrochemistry, Applied Geochemistry, 17 (2002) 855-883. [11] J. Israelachvili, Intermolecular and Surface Forces, 2nd ed., Academic Press Limited, San Diego, 1992. [12] J. Lyklema, Fundamentals of Interface and Colloid Science, Elsevier Ltd., 2005. [13] P.S. Shah, J.D. Holmes, K.P. Johnston, B.A. Korgel, Size-selective dispersion of dodecanethiol-coated nanocrystals in liquid and supercritical ethane by density tuning, J. Phys. Chem. B, 106 (2002) 2545-2551. [14] A. Videcoq, M. Han, P. Abélard, C. Pagnoux, F. Rossignol, R. Ferrando, Influence of the potential range on the aggregation of colloidal particles, Physica A: Statistical Mechanics and its Applications, 374 (2007) 507-516. [15] H.C. Hamaker, The London—van der Waals attraction between spherical particles, Physica, 4 (1937) 1058-1072. [16] K.S. Birdi, Surface Tension and Interfacial Tension of Liquids, in: Handbook of Surface and Colloid Chemistry, Second Edition, CRC Press, 2002. [17] J.C. Berg, An introduction to interfaces & colloids : the bridge to nanoscience, World Scientific, Singapore; Hackensack, NJ, 2010. [18] H. Holthoff, S.U. Egelhaaf, M. Borkovec, P. Schurtenberger, H. Sticher, Coagulation Rate Measurements of Colloidal Particles by Simultaneous Static and Dynamic Light Scattering, Langmuir, 12 (1996) 5541-5549. [19] R. Kretzschmar, H. Holthoff, H. Sticher, Influence of pH and Humic Acid on Coagulation Kinetics of Kaolinite: A Dynamic Light Scattering Study, Journal of Colloid and Interface Science, 202 (1998) 95-103. 68 References [20] T. Oncsik, G. Trefalt, Z. Csendes, I. Szilagyi, M. Borkovec, Aggregation of Negatively Charged Colloidal Particles in the Presence of Multivalent Cations, Langmuir, 30 (2014) 733-741. [21] H. Holthoff, M. Borkovec, P. Schurtenberger, Determination of lightscattering form factors of latex particle dimers with simultaneous static and dynamic light scattering in an aggregating suspension, Physical Review E, 56 (1997) 6945-6953. [22] M. Holmboe, S. Wold, M. Jonsson, S. Garcia-Garcia, Effects of gammairradiation on the stability of colloidal Na+-Montmorillonite dispersions, Applied Clay Science, 43 (2009) 86-90. [23] C. Tournassat, A. Neaman, F. Villiéras, D. Bosbach, L. Charlet, Nanomorphology of montmorillonite particles: Estimation of the clay edge sorption site density by low-pressure gas adsorption and AFM observations, Am. Miner., 88 (2003) 1989-1995. [24] F. Vahlund, H. Hermansson, A direct numerical approach to solving the transport equations for radionuclide transport in fractured rock, SKB internal report R-04-50, (2006). [25] M. Plötze, G. Kahr, R. Hermanns Stengele, Alteration of clay minerals— gamma-irradiation effects on physicochemical properties, Applied Clay Science, 23 (2003) 195-202. [26] R. Pushkareva, E. Kalinichenko, A. Lytovchenko, A. Pushkarev, V. Kadochnikov, M. Plastynina, Irradiation effect on physico-chemical properties of clay minerals, Applied Clay Science, 21 (2002) 117-123. [27] A. Negron, S. Ramos, A.L. Blumenfeld, G. Pacheco, J.J. Fripiat, ON THE STRUCTURAL STABILITY OF MONTMORILLONITE SUBMITTED TO HEAVY γ-IRRADIATION, Clay Clay Min., 50 (2002) 35-37. [28] M. Holmboe, M. Jonsson, S. Wold, Influence of γ-radiation on the reactivity of montmorillonite towards H2O2, Radiation Physics and Chemistry, 81 (2012) 190-194. 69 References [29] G. Sposito, The thermodynamics of soil solutions, Oxford University Press, 1981. [30] G. Sposito, The surface chemistry of soils, Oxford University Press, 1984. [31] S. Wold, Sorption of prioritized elements on montmorillonite colloids and their potential to transport radionuclides, SKB internal report TR-1020, (2010). [32] M.H. Bradbury, B. Baeyens, A mechanistic description of Ni and Zn sorption on Na-montmorillonite Part II: modelling, Journal of Contaminant Hydrology, 27 (1997) 223-248. [33] E. Tertre, D. Prêt, E. Ferrage, Influence of the ionic strength and solid/solution ratio on Ca(II)-for-Na+ exchange on montmorillonite. Part 1: Chemical measurements, thermodynamic modeling and potential implications for trace elements geochemistry, Journal of Colloid and Interface Science, 353 (2011) 248-256. [34] G.N. White, L.W. Zelazny, Analysis and implications of the edge structure of dioctahedral phyllosilicates, Clay Clay Min., 36 (1988) 141-146. [35] O.L. Gaskova, M.B. Bukaty, Sorption of different cations onto clay minerals: Modelling approach with ion exchange and surface complexation, Physics and Chemistry of the Earth, Parts A/B/C, 33 (2008) 1050-1055. [36] C. Hennig, T. Reich, R. Dähn, A.M. Scheidegger, Structure of uranium sorption complexes at montmorillonite edge sites, in: Radiochim. Acta, 2002, pp. 653. [37] R. Dähn, A.M. Scheidegger, A. Manceau, M.L. Schlegel, B. Baeyens, M.H. Bradbury, D. Chateigner, Structural evidence for the sorption of Ni(II) atoms on the edges of montmorillonite clay minerals: a polarized X-ray absorption fine structure study, Geochim. Cosmochim. Acta, 67 (2003) 1-15. 70 References [38] I.C. Bourg, G. Sposito, A.C.M. Bourg, Modeling the acid–base surface chemistry of montmorillonite, Journal of Colloid and Interface Science, 312 (2007) 297-310. [39] T. Hiemstra, W.H. Van Riemsdijk, G.H. Bolt, Multisite proton adsorption modeling at the solid/solution interface of (hydr)oxides: A new approach: I. Model description and evaluation of intrinsic reaction constants, Journal of Colloid and Interface Science, 133 (1989) 91-104. [40] M. Ochs, B. Lothenbach, H. Wanner, H. Sato, M. Yui, An integrated sorption–diffusion model for the calculation of consistent distribution and diffusion coefficients in compacted bentonite, Journal of Contaminant Hydrology, 47 (2001) 283-296. [41] C. Tournassat, M. Bizi, G. Braibant, C. Crouzet, Influence of montmorillonite tactoid size on Na–Ca cation exchange reactions, Journal of Colloid and Interface Science, 364 (2011) 443-454. [42] C. Tournassat, E. Ferrage, C. Poinsignon, L. Charlet, The titration of clay minerals: II. Structure-based model and implications for clay reactivity, Journal of Colloid and Interface Science, 273 (2004) 234-246. [43] L.M. Wang, J. Chen, R.C. Ewing, Radiation and thermal effects on porous and layer structured materials as getters of radionuclides, Current Opinion in Solid State and Materials Science, 8 (2004) 405-418. [44] W.J. Cooper, R.D. Curry, K.E. O'Shea, Environmental Applications of Ionizing Radiation, Wiley, 1998. [45] O. Karnland, S. Olsson, U. Nilsson, Mineralogy and sealing properties of various bentonites and smectite-rich clay materials, SKB internal report TR-06-30, (2006). [46] B. Baeyens, M.H. Bradbury, A mechanistic description of Ni and Zn sorption on Na-montmorillonite Part I: Titration and sorption measurements, Journal of Contaminant Hydrology, 27 (1997) 199-222. 71 References [47] L. Liu, L. Moreno, I. Neretnieks, A Novel Approach to Determine the Critical Coagulation Concentration of a Colloidal Dispersion with Plate-like Particles, Langmuir, 25 (2008) 688-697. [48] S. Garcia-Garcia, S. Wold, M. Jonsson, Kinetic determination of critical coagulation concentrations for sodium- and calcium-montmorillonite colloids in NaCl and CaCl2 aqueous solutions, Journal of Colloid and Interface Science, 315 (2007) 512-519. [49] E. Tullborg, J. Smellie, A.-C. Nilsson, M. Gimeno, L. Auqué, V. Bruchert, J. Molinero, SR-Site-sulphide content in the groundwater at Forsmark, Technical Report TR-10-39, in, SKB Swedish Nuclear Fuel and Waste Management Co., Stockholm (Sweden), 2010. [50] P. Reiller, F. Casanova, V. Moulin, Influence of Addition Order and Contact Time on Thorium(IV) Retention by Hematite in the Presence of Humic Acids, Environmental Science & Technology, 39 (2005) 1641-1648. [51] S.M. Yu, C.L. Chen, P.P. Chang, T.T. Wang, S.S. Lu, X.K. Wang, Adsorption of Th(IV) onto Al-pillared rectorite: Effect of pH, ionic strength, temperature, soil humic acid and fulvic acid, Applied Clay Science, 38 (2008) 219-226. [52] N.D. Bryan, D.L.M. Jones, R.E. Keepax, D.H. Farrelly, L.G. Abrahamsen, A. Pitois, P. Ivanov, P. Warwick, N. Evans, The role of humic non-exchangeable binding in the promotion of metal ion transport in groundwaters in the environment, Journal of Environmental Monitoring, 9 (2007) 329-347. [53] H. Geckeis, T. Rabung, T.N. Manh, J.I. Kim, H.P. Beck, Humic ColloidBorne Natural Polyvalent Metal Ions:  Dissociation Experiment, Environmental Science & Technology, 36 (2002) 2946-2952. [54] Y. Furukawa, J.L. Watkins, Effect of Organic Matter on the Flocculation of Colloidal Montmorillonite: A Modeling Approach, J. Coast. Res., 28 (2012) 726-737. 72 References [55] D. Cui, T.E. Eriksen, Reduction of Pertechnetate in Solution by Heterogeneous Electron Transfer from Fe(II)-Containing Geological Material, Environmental Science & Technology, 30 (1996) 2263-2269. [56] T. Missana, H. Geckeis, Grimsel Test Site – Investigation Phase V. The CRR Final Project Report Series II: Supporting Laboratory Experiments with Radionuclides and Bentonite Colloids NAGRA Technical Report Series NTB 03-02, (2006). [57] F. Huber, P. Kunze, H. Geckeis, T. Schäfer, Sorption reversibility kinetics in the ternary system radionuclide–bentonite colloids/nanoparticles–granite fracture filling material, Applied Geochemistry, 26 (2011) 2226-2237. [58] SKB, Äspö Hard Rock Laboratory Annual Report 2011, SKB internal report TR-12-03, (2012). [59] L.P. Meier, G. Kahr, Determination of the cation exchange capacity (CEC) of clay minerals using the complexes of copper(II) ion with triethylenetetramine and tetraethylenepentamine, Clay Clay Min., 47 (1999) 386-388. [60] K.K. Norrfors, M. Bouby, S. Heck, N. Finck, R. Marsac, T. Schäfer, H. Geckeis, S. Wold, Montmorillonite colloids. I: Characterization and stability of suspensions with different size fractions, Submitted to Applied Clay Science, (2014). [61] M. Holmboe, K. Karin Norrfors, M. Jonsson, S. Wold, Effect of γradiation on radionuclide retention in compacted bentonite, Radiation Physics and Chemistry, 80 (2011) 1371-1377. [62] M. Wolf, G. Buckau, S. Geyer, Isolation and characterisation of new batches of GoHy-573 humic and fulvic acids., in: G.B. ed. (Ed.) Humic substances in performance assessment of nuclear waste disposal: Actinide and iodine migration in the far-field, Wissenschaftliche Berichte FZKA 6969, Forschungszentrum Karlsruhe, 2004, pp. 111–124. 73 References [63] F.J. Stevenson, Humus chemistry: genesis, composition, reactions, John Wiley & Sons, 1994. [64] R. Bro, S. De Jong, A fast non-negativity-constrained least squares algorithm, Journal of chemometrics, 11 (1997) 393-401. [65] J.W.T. Lichtenbelt, H.J.M.C. Ras, P.H. Wiersema, Turbidity of coagulating lyophobic sols, Journal of Colloid and Interface Science, 46 (1974) 522-527. [66] S.H. Behrens, D.I. Christl, R. Emmerzael, P. Schurtenberger, M. Borkovec, Charging and Aggregation Properties of Carboxyl Latex Particles:  Experiments versus DLVO Theory, Langmuir, 16 (2000) 2566-2575. [67] S. Czigány, M. Flury, J.B. Harsh, Colloid Stability in Vadose Zone Hanford Sediments, Environmental Science & Technology, 39 (2005) 15061512. [68] M.Y. Lin, H.M. Lindsay, D.A. Weitz, R.C. Ball, R. Klein, P. Meakin, Universality in colloid aggregation, Nature, 339 (1989) 360-362. [69] L.H. Hanus, R.U. Hartzler, N.J. Wagner, Electrolyte-Induced Aggregation of Acrylic Latex. 1. Dilute Particle Concentrations, Langmuir, 17 (2001) 3136-3147. [70] P. Sandkühler, J. Sefcik, M. Morbidelli, Kinetics of gel formation in dilute dispersions with strong attractive particle interactions, Advances in Colloid and Interface Science, 108–109 (2004) 133-143. [71] P. Sandkühler, J. Sefcik, M. Morbidelli, Kinetics of Aggregation and Gel Formation in Concentrated Polystyrene Colloids, The Journal of Physical Chemistry B, 108 (2004) 20105-20121. [72] H. Wu, M. Lattuada, M. Morbidelli, Dependence of fractal dimension of DLCA clusters on size of primary particles, Advances in Colloid and Interface Science, 195–196 (2013) 41-49. [73] M. Altmaier, V. Neck, T. Fanghänel, Solubility and colloid formation of Th(IV) in concentrated NaCl and MgCl2 solution, in: Radiochimica 74 References Acta/International journal for chemical aspects of nuclear science and technology, 2004, pp. 537. [74] T. Rabung, J. Molinero, D. Garcia, V. Montoya, 1st Workshop Proceedings of the Collaborative Project "Crystalline Rock Retention Processes" (7th EC FP CP CROCK) (KIT Scientific Reports ; 7629), KIT Scientific Publishing, 2012. [75] B.R. Jennings, K. Parslow, Particle Size Measurement: The Equivalent Spherical Diameter, Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 419 (1988) 137-149. [76] IUPAC, Compendium of Chemical Terminology, 2nd ed. (the "Gold Book"), Blackwell Scientific Publications, Oxford 1997. [77] M. Schimpf, K. Caldwell, J.C. Giddings, Field flow fractionation handbook, (2000). [78] A. Meunier, Clays, Springer-Verlag Berlin Heidelberg, 2005. [79] S. Wold, T. Eriksen, Diffusion of humic colloids in compacted bentonite, Physics and Chemistry of the Earth, Parts A/B/C, 32 (2007) 477484. [80] A. Stockdale, N.D. Bryan, The influence of natural organic matter on radionuclide mobility under conditions relevant to cementitious disposal of radioactive wastes: A review of direct evidence, Earth-Science Reviews, 121 (2013) 1-17. [81] M.H. Bradbury, B. Baeyens, Porewater chemistry in compacted resaturated MX-80 bentonite: Physico-chemical characterisation and geochemical modelling PSI Bericht Nr. 02-10, (2002) 28. [82] U. Vuorinen, M. Snellman, Finnish reference waters for solubility, sorption and diffusion studies Posiva Working Report 98-61 (1998) 17-19. [83] L. Derrendinger, G. Sposito, Flocculation Kinetics and Cluster Morphology in Illite/NaCl Suspensions, Journal of Colloid and Interface Science, 222 (2000) 1-11. 75 References [84] J.T.G. Overbeek, The Rule of Schulze and Hardy, Pure Appl. Chem., 52, No. 5 (1980) 1151-1161. [85] R.H. Ottewill, J.N. Shaw, Stability of monodisperse polystyrene latex dispersions of various sizes, Discussions of the Faraday Society, 42 (1966) 154-163. [86] K.K. Norrfors, M. Bouby, J. Luetzenkirchen, M. Jonsson, S. Wold, Smectite and latex colloid agglomeration at high ionic strength – Is there any effect of particle size?, in, 2014. [87] R. Guillaumont, F.J. Mompean, Update on the chemical thermodynamics of uranium, neptunium, plutonium, americium and technetium, (2003). [88] M.H. Bradbury, B. Baeyens, Predictive sorption modelling of Ni(II), Co(II), Eu(IIII), Th(IV) and U(VI) on MX-80 bentonite and Opalinus Clay: A “bottom-up” approach, Applied Clay Science, 52 (2011) 27-33. [89] M. Marques Fernandes, B. Baeyens, R. Dähn, A.C. Scheinost, M.H. Bradbury, U(VI) sorption on montmorillonite in the absence and presence of carbonate: A macroscopic and microscopic study, Geochim. Cosmochim. Acta, 93 (2012) 262-277. [90] G. Bernhard, G. Geipel, T. Reich, V. Brendler, S. Amayri, H. Nitsche, Uranyl(VI) carbonate complex formation: Validation of the Ca2UO2(CO3)3(aq.) species, Radiochim. Acta, 89 (2001) 511-518. [91] S. Regenspurg, D. Schild, T. Schäfer, F. Huber, M.E. Malmström, Removal of uranium(VI) from the aqueous phase by iron(II) minerals in presence of bicarbonate, Applied Geochemistry, 24 (2009) 1617-1625. [92] E. Hartmann, H. Geckeis, T. Rabung, J. Lützenkirchen, T. Fanghänel, Sorption of radionuclides onto natural clay rocks, in: Radiochimica Acta International journal for chemical aspects of nuclear science and technology, 2008, pp. 699. 76 References [93] K.V. Ticknor, P. Vilks, T.T. Vandergraaf, The effect of fulvic acid on the sorption of actinides and fission products on granite and selected minerals, Applied Geochemistry, 11 (1996) 555-565. [94] T. Wu, S. Amayri, J. Drebert, L.R.V. Loon, T. Reich, Neptunium(V) Sorption and Diffusion in Opalinus Clay, Environmental Science & Technology, 43 (2009) 6567-6571. [95] D.R. Turner, R.T. Pabalan, F.P. Bertetti, Neptunium(V) sorption on montmorillonite; an experimental and surface complexation modeling study, Clay Clay Min., 46 (1998) 256-269. [96] T. Eriksen, E., P. Ndalamba, J. Bruno, M. Caceci, The Solubility of TcO2 · nH2O in Neutral to Alkaline Solutions under Constant pco2, in: Radiochimca Acta, 1992, pp. 67. [97] C. Chen, X. Wang, Sorption of Th (IV) to silica as a function of pH, humic/fulvic acid, ionic strength, electrolyte type, Applied Radiation and Isotopes, 65 (2007) 155-163. [98] M. Bouby, H. Geckeis, J. Lutzenkirchen, S. Mihai, T. Schafer, Interaction of bentonite colloids with Cs, Eu, Th and U in presence of humic acid: A flow field-flow fractionation study, Geochim. Cosmochim. Acta, 75 (2011) 3866-3880. [99] C. Fourdrin, T. Allard, I. Monnet, N. Menguy, M. Benedetti, G. Calas, Effect of Radiation-Induced Amorphization on Smectite Dissolution, Environmental Science & Technology, 44 (2010) 2509-2514. [100] K.K. Norrfors, A. Kessler, Å. Björkbacka, S. Wold, M. Jonsson, Radiation induced corrosion of copper in bentonite-water systems under anaerobic conditions, Manuscript in preparation, (2015). [101] K.K. Norrfors, M. Bouby, R. Marsac, S. Heck, T. Schäfer, H. Geckeis, S. Wold, Montmorillonite colloids. III: Influence of montmorillonite colloidal size on the sorption reversibility of radionuclides, Manuscript in preparation, (2015). 77 References [102] K.K. Norrfors, M. Bouby, R. Marsac, S. Heck, T. Schäfer, H. Geckeis, S. Wold, Montmorillonite colloids. II: The Dependency of Colloidal size on the Radionuclide Sorption Capacities in, 2015. [103] N. Getoff, Radiation-induced degradation of water pollutants—state of the art, Radiation Physics and Chemistry, 47 (1996) 581-593. [104] W. Goraczko, J. Slawinski, Luminescence from γ-irradiated humic acid, Journal of Luminescence, 128 (2008) 1155-1161. [105] M. Ochs, B. Lothenbach, M. Shibata, M. Yui, Thermodynamic modeling and sensitivity analysis of porewater chemistry in compacted bentonite, Physics and Chemistry of the Earth, Parts A/B/C, 29 (2004) 129136. [106] S. Lu, Z. Guo, C. Zhang, S. Zhang, Sorption of Th(IV) on MX-80 bentonite: effect of pH and modeling, J. Radioanal. Nucl. Chem., 287 (2011) 621-628. [107] D.L. Zhao, S.J. Feng, C.L. Chen, S.H. Chen, D. Xu, X.K. Wang, Adsorption of thorium(IV) on MX-80 bentonite: Effect of pH, ionic strength and temperature, Applied Clay Science, 41 (2008) 17-23. [108] A.B. Kersting, Plutonium Transport in the Environment, Inorganic Chemistry, 52 (2013) 3533-3546. [109] J. Rothe, C. Walther, M.A. Denecke, T. Fanghänel, XAFS and LIBD Investigation of the Formation and Structure of Colloidal Pu(IV) Hydrolysis Products, Inorganic Chemistry, 43 (2004) 4708-4718. [110] R.A. Alvarez-Puebla, C. Valenzuela-Calahorro, J.J. Garrido, Theoretical study on fulvic acid structure, conformation and aggregation: A molecular modelling approach, Science of The Total Environment, 358 (2006) 243-254. [111] D. Cui, T.E. Eriksen, Reduction of Pertechnetate by Ferrous Iron in Solution: Influence of Sorbed and Precipitated Fe(II), Environmental Science & Technology, 30 (1996) 2259-2262. 78 References [112] B. Gu, W. Dong, L. Liang, N.A. Wall, Dissolution of Technetium(IV) Oxide by Natural and Synthetic Organic Ligands under both Reducing and Oxidizing Conditions, Environmental Science & Technology, 45 (2011) 47714777. [113] S. Kumar, N. Rawat, A.S. Kar, B.S. Tomar, V.K. Manchanda, Effect of humic acid on sorption of technetium by alumina, Journal of Hazardous Materials, 192 (2011) 1040-1045. [114] A. Maes, K. Geraedts, C. Bruggeman, J. Vancluysen, A. Rossberg, C. Hennig, Evidence for the Interaction of Technetium Colloids with Humic Substances by X-ray Absorption Spectroscopy, Environmental Science & Technology, 38 (2004) 2044-2051. [115] Å. Björkbacka, S. Hosseinpour, M. Johnson, C. Leygraf, M. Jonsson, Radiation induced corrosion of copper for spent nuclear fuel storage, Radiation Physics and Chemistry, 92 (2013) 80-86. 79