Boom Clay Natural Organic Matter

Home , Humic substance, Quinone

EXTERNAL REPORT SCK•CEN-ER-206 12/CBr/P-37

Boom Clay natural organic matter

Status Report 2011

Christophe Bruggeman and Mieke De Craen

SCK•CEN Contract: CO-90-08-2214-00 NIRAS/ONDRAF contract: CCHO 2009- 0940000 Research Plan Geosynthesis

June, 2012

SCK•CEN RDD Boeretang 200 BE-2400 Mol Belgium

EXTERNAL REPORT OF THE BELGIAN NUCLEAR RESEARCH CENTRE SCK•CEN-ER-206 12/CBr/P-37

Boom Clay natural organic matter

Status Report 2011

Christophe Bruggeman and Mieke De Craen

SCK•CEN Contract: CO-90-08-2214-00 NIRAS/ONDRAF contract: CCHO 2009- 0940000 Research Plan Geosynthesis

June, 2012 Status: Unclassified ISSN 1782-2335

SCK•CEN Boeretang 200 BE-2400 Mol Belgium

Phone +32 14 33 21 11 Fax +32 14 31 50 21 http://www.sckcen.be

Contact: Knowledge Centre [email protected]

All property rights and copyright are reserved to SCK•CEN. In case of a contractual arrangement with SCK•CEN, the use of this information by a Third Party, or for any purpose other than for which it is intended on the basis of the contract, is not authorized. With respect to any unauthorized use, SCK•CEN makes no representation or warranty, expressed or implied, and assumes no liability as to the completeness, accuracy or usefulness of the information contained in this document, or that its use may not infringe privately owned rights. SCK•CEN, Studiecentrum voor Kernenergie/Centre d'Etude de l'Energie Nucléaire Stichting van Openbaar Nut – Fondation d'Utilité Publique ‐ Foundation of Public Utility Registered Office: Avenue Herrmann Debroux 40 – BE‐1160 BRUSSEL Operational Office: Boeretang 200 – BE‐2400 MOL

Table of Contents

Abstract ...... 7 1. Introduction ...... 9 1.1. Natural organic matter: background and definitions ...... 9 1.2. Natural organic matter in the Boom Clay ...... 10 2. Solid Boom Clay organic matter ...... 11 2.1. Geochemistry of kerogen - general ...... 11 2.1.1. Isolation and analysis of kerogen ...... 12 2.1.2. The classification of kerogens...... 14 2.1.3. Kerogen transformation on burial and its evolution path ...... 15 2.1.4. Kerogen conceptual models ...... 16 2.2 Studies on Boom Clay organic matter and kerogen ...... 20 2.2.1 PhD thesis of Noël Vandenberghe ...... 20 2.2.2 PhD thesis of Ben Laenen ...... 21 2.2.3 PhD thesis and Post-doc of Isabelle Deniau ...... 31 2.2.4 TRANCOM-II studies...... 34 2.2.5 PhD thesis of Pascale Blanchart...... 37 2.3.Comparison of kerogens from Boom Clay, Callovo-Oxfordian argillite and Toarcian shales ...... 38 2.4. Conclusion on the solid Boom Clay organic matter ...... 41 3. Dissolved organic matter ...... 43 3.1. Dissolved organic matter – definitions and background ...... 43 3.2. Functional group content of dissolved organic matter in Boom Clay ...... 47 3.2.1 Introduction ...... 47 3.2.2 Elemental and spectroscopic characterisation of Boom Clay dissolved organic matter ...... 48 3.2.3 Potentiometric characterisation of Boom Clay dissolved organic matter ...... 51 3.2.4 Modelling of potentiometric titrations ...... 58 3.3 Composition and source of Boom Clay dissolved organic matter ...... 61 3.3.1 Introduction ...... 61 3.3.2 Work performed during TRANCOM-II (MORPHEUS piezometer) ...... 61 3.3.3 PhD of Pascale Blanchart ...... 69 3.3.4 Conclusion with respect to the source and composition of dissolved organic matter in Boom Clay ...... 72 3.4 Redox reactivity of dissolved organic matter ...... 73 3.5 Size distribution of dissolved organic matter in Boom Clay ...... 76 3.5.1 Introduction ...... 76 3.5.2 Characterisation of Boom Clay dissolved organic matter size distribution ...... 76 3.5.3 Conclusion: mobile organic matter concentration in Boom Clay ...... 85 3.6 Migration of dissolved Boom Clay organic matter ...... 87 3.6.1 Sorption of dissolved organic matter onto Boom Clay ...... 87 3.6.2 Overview of migration experiments with Boom Clay dissolved organic matter 89 3.6.3 Conclusion on the migration of dissolved organic matter in Boom Clay ...... 111 4. Perturbations and evolving conditions related to Boom Clay organic matter ...... 115 4.1 Thermal perturbation ...... 115 4.1.1. Release of thermolabile components ...... 115 4.1.2. Production of CO2 from the Boom Clay kerogen under thermal stress ...... 117

4.1.3. Comparison to other geological formations...... 123 4.2. Impact of alkaline perturbation ...... 125 4.3. Impact of ionic strength perturbation ...... 127 4.4 Impact of oxidation ...... 130 4.4.1 Introduction ...... 130 4.4.2 Impact of oxidation on the compositions of kerogen and bitumen ...... 130 4.4.3 Impact of oxidation on the fraction of dissolved organic matter ...... 132 4.5 Impact of compaction...... 136 5. General conclusions and recommendations ...... 140 6. References ...... 141

Abstract

Boom Clay contains an appreciable amount of low-maturity natural organic matter. This organic matter may be operationally subdivided into different pools, which have various impacts on, among others, the pore water chemistry and radionuclide geochemistry. Moreover, the creation and operation of a geological repository located in the Boom Clay will cause perturbations to the organic matter, which need to be properly assessed. This state-of- the-art report aims at bringing together the results obtained from various scientific sources and communities involved in the study of Boom Clay organic matter. Both origins, preservation pathways, nowadays in situ situation (Mol-Dessel reference site) and expected future evolutions of Boom Clay natural organic matter (and important subfractions thereof) are evaluated and discussed.

Keywords

Boom clay, geological disposal, natural organic matter, perturbations

1. Introduction

1.1. Natural organic matter: background and definitions

Natural organic matter is all the organic matter in a reservoir or natural ecosystem, other than living organisms and compounds of man-made origin (BUFFLE, 1988). It is ubiquitous in the environment, occurring in all soils, waters and sediments of the ecosphere. The natural organic matter found in soils and sediments possesses a large variety of properties and is composed of an extremely complex mixture of compounds, most of which are not yet identified. Indeed, the number of compounds making up natural organic matter can be considered as "infinite" and there is little hope to separate the different compounds completely. The reason for this being that natural organic matter is of detrital nature and is derived from different major sources (e.g., marine and terrestrial reservoirs). Furthermore, different pathways (degradation, oxidation, polymerisation, etc.) are assumed in the formation of natural organic matter (STEVENSON, 1982). Natural organic matter is therefore heterogeneous by definition, showing a continuous distribution in mass (molecular weight), size, elemental composition, structure and functional entities. Examples of different functional entities are aromatics, aliphatics, phenolics and quinones (GAFFNEY et al., 1996).

Consequently, studies of natural organic matter have mainly been concentrated on groups of compounds separated from the initial mixture by means of operationally defined techniques, rather than on pure compounds. In this report, we will report on the studies conducted thus far on natural organic matter (NOM) present in Boom Clay (Rupelian, Belgium). The total NOM pool is frequently expressed as the total organic carbon (TOC) content (in wt%) of a formation.

"Humic substances" is also used as a synonym for natural organic matter, although humic substances normally refer to that part of NOM which has already undergone severe transformation and decomposition after deposition (therefore, they cannot be classified as any other chemical class of compounds such as polysaccharides, proteins, etc. (GAFFNEY et al., 1996)). Because of its age, assumed formation pathways and diagenetic history, the NOM in Boom Clay can be typified as consisting solely of humic substances, and both terms can be used equally well. In this report however, we will try to keep from using this term as much as possible because of the possible confusion.

Solid organic matter is defined as that part of the NOM pool which is insoluble in aqueous solutions at all pH values. In sediments it forms typically the largest part of all NOM, exceeding 90 % or more of all organic matter present. In soil science, it is referred to also as "humin". Solid organic matter is further subdivided into bitumen and kerogen. Bitumen is the part of solid organic matter which can be dissolved in usual organic solvents while kerogen is the remaining insoluble part (VANDENBROUCKE, 2003). Both terms stem from petroleum science in which the study of petroleum-making capabilities of natural organic matter is of primary concern.

The fraction of the NOM pool which can be solubilised in aqueous solutions is referred to as dissolved organic matter (DOM). Dissolved organic matter is often expressed as the dissolved organic carbon (DOC) content (in mg C/l, mostly measured after filtration over a 0.45 µm membrane). Dissolved organic matter is subdivided into humic and fulvic acids. Humic acids 9

are those materials that are insoluble at acidic pH values (pH < 2) but are soluble at higher pH values and are typically composed of larger macromolecules. Fulvic acids are soluble in water at all pH values and are composed of smaller molecules showing in general more reactivity (higher contribution of functional groups). Humic and fulvic acids are the most studied fractions of NOM in soil science because of their abilities to complex radionuclides and toxic metals (SPARKS, 2002; SPOSITO, 1989; STEVENSON, 1982).

Numerous advances in analytical methods and separation techniques have enabled researchers interested in NOM to begin to chemically characterise these complex molecular structures. These methods have also allowed the interactions of NOM with metals and with organic pollutants in the environment to be explored. This allowed the environmental chemistry and physics of NOM to be better understood, although a lot of open questions still remain to be answered.

1.2. Natural organic matter in the Boom Clay

Compared to other argillaceous formations studied in the European context of geological disposal of radioactive waste, the Boom Clay contains substantial amounts of organic matter of low immaturity (TOC content of 1-5 wt.%). The NOM pool is distributed between the liquid and the solid phase. The organic matter found in Boom Clay is typical for its depositional environment during the Rupelian: clearly marine but with near-shore terrestrial influences (LAENEN, 1997; VANDENBERGHE, 1978; WOUTERS and VANDENBERGHE, 1994). The organic matter from Boom Clay was already subject of detailed petrographic (VANDENBERGHE, 1978) and geochemical (LAENEN, 1997; LAENEN, 1998) studies, focussing on samples from the outcrop region. More recent studies (e.g., Deniau et al. (2001) and Blanchart (2011)) started out from (so-called) "undisturbed" samples typical for the Mol- Dessel region. Overviews of Boom Clay organic matter were already given in Wouters and Vandenberghe (1994) and in Van Geet et al. (2003).

In this report we will mainly focus on the most recent studies performed on (undisturbed samples from) Boom Clay from the Mol-Dessel region. Two types of studies can be distinguished. The first type was performed by research groups (Institut Français du Pétrole, IFP and Laboratoire G2R: Géologie et Gestion des Ressources minerals et énergétiques UMR 7566, Université de Nancy) specialised in organic geochemistry, and focused on questions related to the bulk geochemical and molecular organic geochemistry of the organic matter. The second type was performed by SCK•CEN and focused predominantly on the transport characteristics in the framework of rapid colloidal transport of radioactive isotopes.

The report is subdivided following the operational definitions proposed in section 1.1. This means that distinguishment is made between studies on the solid organic matter (or kerogen) fraction (chapter 2) and studies on the dissolved organic matter fraction (chapter 3). In both chapter 2 and 3 the focus will lie on the undisturbed organic matter. In the final chapter, the influence of possible perturbations on the organic matter from Boom Clay is discussed.

2. Solid Boom Clay organic matter

2.1. Geochemistry of kerogen ‐ general

Kerogen is defined as the insoluble sedimentary organic matter capable of generating petroleum and natural gas (DURAND, 1980; VANDENBROUCKE, 2003; VANDENBROUCKE and LARGEAU, 2007). Its insolubility in normal organic solvents is derived from the huge molecular weight (upwards of 1000 Daltons) of its component compounds. Thus, in opposition with usual chemistry nomenclature, the name kerogen does not represent a substance with a given chemical composition. Indeed kerogen is a generic name, in the same sense as lipids or proteins. Several organic precursors and their mixtures may lead to kerogen incorporated in sediments. Moreover, with geological burial of sediments and temperatures ranging from a few tens to less than 200 °C, during time periods from tens to thousands of million years, the chemical composition of kerogen dispersed in sedimentary layers (called source rocks) progressively changes, a part of it is being transformed into petroleum (oil window ca. 60-160 °C) and gas (gas window ca. 150-200 °C). The operational definition of kerogen on the basis of insolubility does not account for these compositional variations due to source and evolution. Another major drawback of the pragmatic definition of kerogen is that the composition and chemical features of kerogen will be closely dependent on the organic solvent and extraction procedure used for its separation. Besides solvent extraction needed for separation of soluble and insoluble organic matter in rocks, analytical techniques used for kerogen characterisation often require its isolation from minerals, or at least its concentration.

Apart from kerogen, several other commonly used terms related to solid organic matter and kerogen geochemistry include: • Bitumen1: a mixture of organic liquids that are highly viscous, black, sticky, entirely soluble in carbon disulphide, and composed primarily of highly condensed polycyclic aromatic hydrocarbons. In contrast with kerogen, bitumen is operationally defined as that part of the organic matter which is soluble in organic solvents. Bitumen typically results from kerogen transformation in source rocks along the pathway of crude oil generation. • Asphalthenes: this fraction is operationally defined as the n-heptane (C7H16)-insoluble, toluene (C6H5CH3)-soluble component of bitumen or coal. Asphalthenes have been shown to have a distribution of molecular masses in the range of 400 Da to 1500 Da with an average around 750 Da. Asphalthenes are today widely recognised as soluble, chemically altered fragments of kerogen which migrated out of the source rock during oil catagenesis. • Resins: similar structure and chemistry as asphalthenes, but smaller.

Most of the organic matter in sediments consists of dispersed insoluble kerogen. The bitumen form only a small part of the organic fraction, and can be divided in three classes: asphalthenes, heterocomponents or polar nitrogen-sulpur-oxygen compounds (NSO) and hydrocarbons (LAENEN, 1997). The first two groups mainly consist of polymers. During progressive diagenesis, these high molecular weight polymers are broken down and defunctionalised to give rise to hydrocarbons. The third group contains the aliphatic and

1 Bitumen is chemically equivalent to asphalt, although the latter term is more commonly used as the manufactured end product of a crude oil distillation process, while bitumen refers to natural deposits.

aromatic structures, whose skeleton is made up of C and H-atoms. In immature sediments, the molecular composition of these hydrocarbons is strongly related to the source of the organic matter; hydrocarbon concentrations are low because most of the organic matter is still functionalised and incorporated in the kerogen matrix. When temperature rises hydrocarbons are broken from the kerogen matrix. This process is paralleled by a decrease in the amount of asphalthenes and resins due to thermal cracking and defunctionalisation.

2.1.1. Isolation and analysis of kerogen

The isolation of the kerogen from Boom Clay is achieved according to the IFP (Institut Français du Pétrole) chemical procedure, designed by Durand and Nicaise (1980). Its flowchart is presented on figure 2.1. It is based on the destruction of major minerals (carbonates, sulphides, sulphates, hydroxides, clay minerals, quartz, silicates) from the finely ground sedimentary rock by non-oxidant acid attacks at temperatures between 60 and 70 °C under an inert atmosphere, in order to prevent any oxidation of kerogen during preparation. Once isolated by this procedure, the kerogen concentrate still contains residual minerals. The main remaining mineral is pyrite, frequently enclosed inside the organic network of kerogen, as a normal by-product of bacterial kerogen formation and incorporation in sediments (DURAND and NICAISE, 1980). The isolated kerogen is usually subjected to two types of analysis: elemental analysis and pyrolysis (VANDENBROUCKE, 2003).

A first major technique for geochemical studies on kerogen is the elemental analysis of kerogen concentrates, which has shown that the major elements are the following: C, H, N, O, S, and possibly Fe from pyrite, with generally C and N amounting respectively for the highest and the lowest weight % (VANDENBROUCKE, 2003). These analyses are used for classifying the kerogen in a “van Krevelen”-like diagram (van Krevelen originally designed this diagram for coals (VAN KREVELEN, 1961)). This diagram plots the main elements, C, H and O as atomic ratios of H/C vs. O/C. Starting from the wt.% analyses, these atomic ratios are obtained by multiplying the weight ratio of H to C by 12, and the weight ratio of O/C by 0.75 (i.e. 12/16). In a “van Krevelen diagram”, kerogens from sedimentary source rocks are classified into three (or four)2 main organic matter types according to their specific depositional environment and their evolution during maturation (with increasing burial) in geological conditions following “carbonisation paths” (figure 2.2).

A second major technique for geochemical studies on kerogen is pyrolysis, which is used both for characterisation of its molecular building blocks and for kinetic studies. Pyrolysis is a widely used degradation technique that allows breaking a complex substance into fragments, by heating it under an inert gas atmosphere. The small compounds thus obtained are building blocks of the complex substance, but they can often be analysed more easily, eventually up to a molecular level, and quantified (VANDENBROUCKE, 2003). Applying this technique to hydrocarbon generation from kerogen thermal cracking also means that geological conditions with long time intervals at low temperature can be replaced by laboratory conditions with short experimental duration at high temperature.

2 The three main kerogen types are labelled sapropelic (type I), planktonic (type II) and humic (type III) – see also section 2.1.2. Type IV kerogen is the residual part in the form of polycyclic aromatic hydrocarbons and, as opposed to the other three types, has no potential to produce hydrocarbons.

Figure 2.1: IFP procedure for kerogen isolation Reproduced from Vandenbroucke (2003)

Of the different types of pyrolysis, the Rock-Eval technique is probably the most frequently used. Rock-Eval pyrolysis was designed for evaluating the oil and natural gas potential and the maturity of source rocks in relation with petroleum and gas exploration (ESPITALIE et al., 1977). The Rock-Eval apparatus consists of an open pyrolysis system, with an inert gas flow extracting the effluents generated during programmed heating of the weighed whole rock (VANDENBROUCKE, 2003). Heating to 300° C results in a thermovaporisation of hydrocarbons already present as such in the rock (S1 peak), whereas hydrocarbons produced in the 300-550 °C temperature range result from kerogen cracking (pyrolysate, S2 peak). The temperature at which the expulsion of S2 hydrocarbons reaches a maximum is called Tmax (°C) (temperature of maximum pyrolysis yield). With certain restrictions, this temperature is a measure for the maturity of the kerogen. Normalisation of S2 to TOC gives the hydrogen index (HI, mg HC/g TOC). Up to a temperature of 390 °C, the oxygenated components are measured in a parallel flow. From this measure the weight fraction of CO2 expelled from the sample is derived (organic derived carbon dioxide, S3), which after normalisation to TOC results in the oxygen index (OI) (mg CO2/g TOC). Rock-Eval pyrolysis determines kerogen properties inside source rocks, without need to isolate it from minerals. The obtained indexes reflect the

potential of the kerogen to generate hydrocarbonaceous compounds (HI) and CO2 (OI) under open conditions.

Figure 2.2: General scheme of kerogen evolution from diagenesis to metagenesis in the Van Krevelen diagram (taken from Vandenbroucke and Largeau (2007), modified after Tissot and Welte (1978), p. 186)

2.1.2. The classification of kerogens

Below follows a brief description of the chemical characteristics of kerogen reference types I to III and their associated depositional environment (VANDENBROUCKE, 2003) (figure 2.2).

 The immature organic matter from type I kerogen is very aliphatic, with atomic H/C frequently higher than 1.5. The oxygen content is often low (0.03 < O/C < 0.1). The analysis of oxygen functional groups in immature kerogen shows a major participation of unreactive oxygen assumed to be located in aliphatic ether bonds. The extracts, oils or pyrolysis products are very rich in long chain n-alkanes. This type is not really associated with specific biological precursors, sedimentary environments or preservation conditions.  Immature kerogens from type II are characterised by an atomic H/C ratio of about 1.3 and a corresponding O/C ratio around 0.15. This organic matter contains much more aliphatic cyclic moieties than in type I, and a large amount of aromatics. The n-alkane distribution in extracts or pyrolysates is generally restricted to the carbon range under 25 atoms. Sulphur is always associated with this type of organic matter, either as pyrite and free sulphur, or in organic structures. The depositional environment is moderately deep marine environment with planktonic input as primary source.

 Type III organic matter is frequently found in deltaic settings and derives from higher plant debris (terrestrial environment), often highly reworked due to the oxidant transportation conditions associated with detritic sedimentation in shallow marine environment. The organic fraction is for a large part ligneous debris with a predominantly aromatic structure, together with moieties derived from the protective constituents of higher plants with a predominantly aliphatic structure rich in long chain alkanes. The atomic H/C ratio is lower than 0.8, while the extracts and oils are highly paraffinic in comparison with products generated from type II.

2.1.3. Kerogen transformation on burial and its evolution path

Diagenesis3 of kerogen stricto sensu is a step where kerogen loses large amounts of oxygen, mainly as CO2 and H2O. This step is all the more evident if the initial kerogen is rich in oxygen and is thus mainly observed for type III organic matter, but it is also significant for type II. Biological processes have a major influence on the amount and composition of the organic matter during early diagenesis. During this step, the main loss of nitrogen occurs due to consumption of proteins, and sulphur may become incorporated in the organic matter, if sulphur bacterial colonies are present. Preservation of organic matter during early diagenesis is strongly dependent on the oxygen content of water and on the water circulation inside deposited sediments. A very low oxygen level (anoxia) reduces the efficiency of bacterial populations able to feed on organic debris, while low porosity reservoirs preclude the free circulation of bacteria into sediment pores. Clay sediments are therefore more likely to contain kerogen, not only because they may protect the organic matter by physico-chemical adsorption, but also because their structure settles as "house of cards", increasing water pressure in the porosity and preventing water input during further compaction on burial (VANDENBROUCKE, 2003).

Catagenesis is the main step of hydrogen (and carbon) loss from the kerogen, and is consequently the stage of petroleum formation. Considering a general formula of generated hydrocarbons as (CH2)n, two H atoms are lost for one C atom, thus the atomic H/C of the residual kerogen decreases in the van Krevelen diagram, reflecting an increasing aromatization. It appears immediately that the petroleum potential deduced from the H/C variation range decreases strongly in the order from type I to type III. During this stage, compounds with lighter and lighter molecular weights are formed during this stage, leading finally to gases. An insoluble aromatic residue similar to coke is generated in late steps of catagenesis as the result of secondary cracking of petroleum products issued from kerogen. This residue, which has almost no petroleum potential, can be observed alone as “pyrobitumen” in high temperature reservoirs, but is also formed in source rocks and thus mixed with the residual kerogen.

The transition between catagenesis and metagenesis zones is located for all kerogen types at an atomic H/C ratio around 0.6. Below this transition zone, all kerogen types merge into a single group. Metagenesis is the last stage of kerogen reorganisation into an almost pure carbon structure. During this step, the aromatic lattice rearranges by expelling its "defects" in

3 Diagenesis = all physical , chemical and biological changes within a sediment or rock, starting from its deposition and until metamorphosis.

a manner similar to crystal formation. The first eliminated product is late methane, resulting from release of methyl groups on aromatics. Non-hydrocarbon gases are also formed during metagenesis by elimination of heteroelements from resistant bonds (O from ethers) or from aromatic structures (S, N, O).

The basic skeletons of biomarkers incorporated in the kerogen are well preserved during diagenesis and early catagenesis. Therefore, in many cases pyrolysis products can still be unambiguously linked to precursor molecules. Bitumen are more sensitive to bacterial degradation, defunctionalisation and thermal alteration, but in immature sediments they generally are only slightly transformed and bear useful information about precursors and preservation conditions. The biomarker composition of bitumen also gives information about diagenetic conditions and maturity. The diagenetic pathways of various biomolecules strongly depend on pH and redox conditions of pore waters, bacterial degradation and the presence of specific catalysts (LAENEN, 1997).

2.1.4. Kerogen conceptual models

Because kerogen is a mixture of various non-polymeric macromolecules, comprising both resistant biomacromolecules and recombined biodegradation products, it will never be possible to represent a true kerogen structure. However, a hypothetical average structure of kerogen, representing a large amount of information from various analyses, can provide a synthetic view of the main resemblances and differences among sedimentary organic matters (VANDENBROUCKE, 2003).

Although it is very difficult to describe such complex macromolecular mixtures by detailed chemical structures, many structural or chemical models of sedimentary organic matter were published in the past. Their aim was to visualise a number of its properties or analyses in a synthetic form to interpret or understand others. Thus a model aimed at representing physical or spectral properties can be a structural model whereas a model for describing chemical interactions or cracking reactions will rather be a molecular model.

Both atomic (providing quantitative information on the number of atoms, their chemical environment or their spatial distribution) and molecular (providing qualitative and quantitative information on kerogen building blocks by studying degradation products of macromolecular kerogen constituents) analyses are used for kerogen modelling and their classification. Once quantitative atomic and molecular analyses are obtained on a given kerogen, the quantitative combination of molecular building blocks should ideally fit atomic data. Eventually, by trial and error, structural models representing the three kerogen types at the main stages of their geological evolution were constructed (figure 2.3).

In more recent years, computer-aided modelling was used to provide a more realistic view of structures by building 3-D kerogen models (figure 2.4).

The notions of kerogen types and evolution paths, even though they may appear as an oversimplification of unique environmental situations, enabled a great deal of geochemical data to be condensed and explained in a simple manner.

The processes of kerogen formation in present and past geological times, in other words the mechanisms of transformation and preservation of the organic matter finally incorporated into sediments, are still a matter of debate and active research (VANDENBROUCKE, 2003;

VANDENBROUCKE and LARGEAU, 2007). Despite numerous advances in knowledge of kerogen precursors, many important points, such as the atomic structural arrangement in kerogen moieties, are still far from being elucidated.

Figure 2.3: Structural models of kerogens: comparison of the chemical composition of kerogen types I, II and II at the beginning of the diagenesis stage (VANDENBROUCKE, 2003).

Figure 2.4: 3-D molecular structure of type III kerogen types at the evolution stage c: end of catagenesis (VANDENBROUCKE, 2003)

2.2 Studies on Boom Clay organic matter and kerogen

Source, degradation and maturity are the most important processes in control of the composition of the organic matter in a geological deposit. In sediments that have only experienced weak thermal maturation the first two parameters will be determining.

The Boom Clay is located in the Campine Basin of northern Belgium. It is a Lower Oligocene siliciclastic sediment deposited in an open marine environment at the southern border of the epicontinental North Sea Basin (VAN KEER and DE CRAEN, 2001; VANDENBERGHE, 1978; VANDENBERGHE and VAN ECHELPOEL, 1987). A marine transgression, which started 32.7 Ma ago, started with the deposition of silty clay in the sedimentary basin (Belsele-Waas Member). The increasing water depth resulted in the deposition of a gradually finer clay sediment (Terhagen Member). A small regression then occurred, resulting in the deposition of coarser sediments, known as the "double band" (base of the Putte Member). Later, finer sediments were again deposited due to a second transgressive pulse. The maximum water depth of this transgression corresponds to S60 (Putte Member). The transgression is again interrupted by a short regression, but quickly followed by a third transgressive pulse. This transgression is followed by an important regression, which finishes the deposition of the clays, 28.5 Ma ago (sedimentation of the Eigenbilzen Sands) (VAN KEER and DE CRAEN, 2001).

In the following sections, the main important investigations into the characterisation of the (solid) organic matter from Boom Clay are resumed.

2.2.1 PhD thesis of Noël Vandenberghe

Vandenberghe (1974; 1978) was the first to make a stratigraphical map of the Rupelian deposits, based on samples gathered from outcrop areas. Both granulometric, mineralogic and geochemical analysis was performed.

Vandenberghe (1978) identified three lithostratigraphic members in the type area of the Boom Clay (the vicinity of Antwerp): the silt-rich Belsele-Waas Member, the Terhagen Member composed of grey, partially carbonate-rich clays, and the organic-rich Putte member. He also showed that the sharp colour boundary between the Terhagen and the Putte Member (observed in the outcrops of the Boom Clay) is due to a difference in the amount of organic matter. In the upper part of the section the rhythmicity in organic matter content largely obscures the silt/clay alternation, which is responsible for the banded appearance of the lower Boom Clay members.

Especially in the Putte Member, TOC and the coarseness of the sediment reveal a similar, high frequency cyclicity, although the maxima are slightly out of phase. At the top of each silt bed, TOC sharply increases from a background level between 1 and 1.5 wt.% up to values of 3 wt.%. In the overlaying clay bed, the amount of organic matter gradually diminishes towards the background level. These metrescale cycles are superimposed on a gradual increase in TOC from approximately 0.5 wt.% in the lowermost silt beds of the Belsele-Waas Member to 1.5 - 2 wt.% in the Putte Member.

By means of microscopy and C/N analyses Vandenberghe (1978) discriminated between three groups of organic matter, i.e., two autochthonous groups that could be related to marine

organisms and bacteria, and a group of allochtonous, continental organic matter. This last group largely consists of higher plant remains, but also includes a few coal fragments. The Putte Member appears to contain more phytoclasts than the underlying grey clays.

Based on the analysis of two wood fragments from the Boom Clay and a compilation of literature data, Vandenberghe (1978) concluded that the C/N ratios of the bulk sediment are the result of mixing a pool of nitrogen poor organic matter derived from higher plants with a nitrogen rich, marine pool. The dark clays from the Putte Member have a higher average C/N ratio than the grey clays what suggests that the Putte samples contain the largest proportion of higher plant material.

A petrographical study on the phytoclast fragments present in the Boom Clay enabled to distinguish five groups of fragments: poorly evolved material (algae remnants, badly conserved liptinite), liptinite (vegetal resins and waxes), inertinite (residue of wood fire), vitrinite (cellular tissue of plants), and coal fragments. Based on these petrographical data, the phytoclast fragments are divided in primary allochthonous grains, secondary allochthonous grains and coal fragments.

The vitrinite reflectivity of the primary allochthonous particles (R0 = 0.2-0.83 %) points towards an initial diagenetic stage with methane development corresponding to a lignitic rank. The reflectivity of the secondary allochthonous particles ranges between 1.5 and 1.6 %. The reflectivity of the coal fragments (R0 = 0.5-1.5 %) point towards an origin of a Palaeozoic rock, most likely deriving from the British Coals.

Overall, the vitrinite reflectivity of Boom Clay organic matter shows a low level of maturity, as reflected by low vitrinite reflectance in the 0.25-0.4 % range for most wood fragments (VANDENBERGHE, 1978). Indeed, the Boom Clay was never deeply buried during its geological history so that the organic matter did not experience significant thermal stress.

2.2.2 PhD thesis of Ben Laenen

Laenen (1997; 1998) studied the general organic geochemical trends in the Boom Clay and compared the results with the known relative sea-level history. Relative sea-level changes can be expected to have an equally strong influence on the composition of organic matter as it has on the geometric pattern of lithofacies. To this purpose, Laenen (1997) discussed both the bulk organic composition, which allowed a rough determination and quantification of the pools of autochtonous and allochtonous organic matter, and the molecular composition of extracted bitumen from 36 selected samples by medium pressure liquid chromatography, gas chromatography (GC) and mass spectrometry (MS).

2.2.2.1 Bulk organic matter composition

The bulk organic matter composition was examined through samples spanning the entire stratigraphical interval exposed in the outcrops of the type area. Samples were taken in freshly excavated parts of the section at a depth of 20 to 30 cm. Nevertheless, (at least) superficial oxidation cannot be excluded by this procedure.

The TOC profile can be divided in two distinct stratigraphic zones (figure 2.5). Throughout the Belsele-Waas and Terhagen Members, the TOC values range from 0.3 wt.% to 1.4 wt.%

and variance is low. At the Terhagen/Putte boundary the TOC suddenly increases to reach a maximum of 2.9 wt.%. In the Putte Member the TOC distribution is dominated by two high frequency cyclicities, related to the 41 ka and 100 ka Milankovitch cycles. The 100 ka cycles reach maxima near each silt-to-clay transition. It can therefore be concluded that most of the variance of TOC is a function of grain size and the thickness of the silt/clay couples. This latter covariation indicates that a significant part of the variance in TOC can be explained by the dilution of organic matter by detrital materials. The highest TOC concentrations are found in parts of the section with low sedimentation rates that were deposited during quiet periods. Maximum flooding surfaces are also marked by high fluxes of organic carbon.

Rock-Eval analysis showed that the Boom Clay samples are all characterised by low amounts of bitumen (S1) and pyrolysable kerogen compounds (S2). S2 slightly increases from the base of the section up to the Terhagen/Putte boundary. Above this level the average value and the variance increase.

Figure 2.5: Distribution of TC (wt.%), TOC (wt.%) and TIC (wt.%) throughout the section. The boundaries between the members are indicated with short-dotted lines (LAENEN, 1997)

Figure 2.6: Variations in Rock-Eval parameters through the type section of the Boom Clay (LAENEN, 1997)

The average Boom Clay value of the HI/OI ratio is low at 1.08±0.56 (figure 2.6). The stratigraphic distribution of this ratio closely parallels the HI curve (not shown). The Tmax values are typical for immature kerogens (416°C ± 11 °C) (figure 2.6). The Boom Clay samples plot close to the OI-axis of the Tissot diagram (figure 2.7a). This could either point to a mixture of a type III and IV kerogens. The coal fragments found in the Boom Clay may be the source of this inert (type IV) organic matter. However, the trend can also be explained by oxidation of a type III and/or type II source. This is supported by a plot of HI versus Tmax in which the Boom Clay samples lay within the type III area close to the type II-III boundary (figure 2.7b). Thus, the observations suggest that the kerogen of the Boom Clay consists of a mixture of degraded type II and type III organic matter, and hence, is mainly derived from algae and higher plants. The low HI and high OI are indicative of strong oxic degradation of the organic matter. The small amounts of coal fragments, which correspond to a type IV kerogen, might have further lowered the HI (LAENEN, 1997). The septaria layers contain more type II kerogen than the clays immediately above and below.

The OI, HI and Tmax values plead for a mixture of oxic degraded type II and III organic matter, probably with lower amounts of a coaly type IV kerogen. The increase in the amount of pyrolysis products and the Rock-Eval ratios across the Putte/Terhagen boundary is indicative of an increase in hydrogen-rich kerogen in the Putte Member. This most probably is due to a higher amount of type II organic matter in the black clays and a better preservation of reactive organic matter.

Figure 2.7: a. Tissot classification of the kerogen of the Boom Clay. The bold lines mark maturation pathways for the different types of kerogen, the dashed lines indicate the effects of the loss of H2O, CO2 and CH4. b. Plot of the HI versus Tmax. The bold line marks the boundary between the fields of type III and type II kerogens (LAENEN, 1997)

Nitrogen and phosphorus and sulphur closely covary with TOC. Elemental analysis shows that the bulk rock ratios of C : N : P of the organic matter of the Boom Clay are 100 : 4.1±0.2 : 1.0±0.2. Although the organic matter is rather poor in N (compared to other published values), it is in fact the primary host of nitrogen. The C/N ratio of 24.4±2.7 derived from linear regression, either pleads for a large contribution of continental organic matter, or may reflect intense bacterial reworking of nitrogen in the water-column.

There is also a positive correlation between S and TOC, which asks for a relationship between the amount of organic matter buried in the sediment and the formation of pyrite. It is therefore suggested that higher amounts of organic matter resulted in the formation of more pyrite, probably due to the fact that more reactive organic matter was available.

2.2.2.2. Molecular organic geochemistry

Because of the low maturity, the molecular composition of bitumen extracted from the Boom Clay is mainly a function of source and preservation of the organic matter. Laenen (1997) evaluated the origin and diagenetic fate of the biomarkers found in the aliphatic and aromatic hydrocarbon fraction of the Boom Clay, and interpreted the variations in molecular composition observed throughout the type section in terms of changes in marine primary production, continental organic matter and early diagenetic conditions. To this purpose, 36 samples were taken from layers that take key positions in the sequence stratigraphic and sedimentologic framework of the Boom Clay.

Generally, the Boom Clay samples yielded low amounts of organic extract. The solvent- extractable fraction mainly consists of nitrogen-sulphur-oxygen (NSO) heterocomponents4. This may be due to the low maturity of the Boom Clay, leaving the majority of the hydrocarbons still bound to the kerogen matrix. The abundance of S-compounds in Boom Clay is low and fits with the abundance of pyrite. Most of the sulphide reacted with iron, or was lost to the overlaying water column. Molecules with a chroman skeleton are present in the aromatic fraction of the Boom Clay, which may point to an early diagenetic formation by the condensation of phytol and methylphytols.

The aliphatic and aromatic fractions were also investigated. Among the aromatic molecules are the polyaromatic hydrocarbons (PAH)5. Beside PAH derived from higher plants, small concentrations of PAH can be found in the Boom Clay and are related to the coal fragments known from a maceral study (VANDENBERGHE, 1978). The aliphatic molecules could be classified in six groups, i.e., acyclic alkenoids6 (mainly alkane-type), steroids7 (including

4 Most of the nitrogen-sulphur-oxygen components (NSO) found in bitumen have a high molecular weight and belong to the resin and asphalthene fraction. In addition, a significant part of organic S and minor amounts of oxygen are incorporated in the aromatic hydrocarbon fraction under the form of thiophenes and chromans. Both are aromatic structures with respectively S and O incorporated in one of the rings.

5 Polyaromatic hydrocarbons are fully aromatised, strongly condensed organic molecules, formed at high temperatures. Consequently, PAH are abundant in mature rocks and coals as late diagenetic and catagenetic reaction products. Nevertheless, fully aromatised structures are also formed in immature sediments.

6 Alkenoids are aliphatic, acyclic organic molecules. The most wide-spread members of this group are the straight-chained, saturated n-alkanes and acyclic, methylated alkanes. Both molecular classes are abundant in cell membranes and protective waxes of various organisms, where they mainly occur as fatty acids, alcohols and n-alkanes. In immature sediments the carbon atom distribution and the number and position of the double bonds of normal and methylated alkenoids are source specific. For example, n-alkanes derived from higher plant waxes have a strong odd over even predominance (OEP) and a chain length in the range of 23 to 33 carbon atoms. In contrast, alkenoids biosynthesised by marine phytoplankton such as algae show a dominance of C17 n-alkane.

7 Steroids are tetracyclic triterpenoidal structures formed by the enzymatic oxidation and cyclisation of squalane. Steroids are abundant in plants and animals and act as rigidifiers of cell membranes. The position of double bonds and methyl groups in the skeleton of the steroids are characteristic for certain groups of organisms. The number of carbon atoms can give information about the source of the molecules. Higher plants generally are rich in C29 steroids, but C28 skeletons can also be abundant. Cholesterol, a C27 structure, is abundant in animals, but high amounts of C27 steroids are also found in green and blue-green algae. Zooplankton is rich in C27 cholesterol, while C28 predominates in phytoplankton.

steranes, diasterenes and sterenes), hopanoids8, higher plant pentacyclic triterpenoids9 (HPPT), and sequi- and diterpenoids10.

The regular alkanes are the most abundant molecules in the aliphatic fraction of the Boom Clay. The n-alkanes extend from C11 to C37 and have a bimodal distribution. The long chain n- alkanes, which range from C21 to C33 and maximises at C31, are the most abundant group. This points to a significant contribution of higher plant waxes. The short chain population is only a minor constituent and shows no distinct OEP, which may point to a bacterial or algal origin.

The steroids found in the aliphatic fraction of the Boom Clay can be divided in three subclasses: the regular steranes, the diaster-Δ13(17)-enes and the regular sterenes. The range and distribution of regular steranes are indicative of a strong contribution of continental organic matter to the organic fraction of the Boom Clay, and of a lesser contribution of marine organic matter to the bitumen.

The high relative abundance of hopanoids in the aliphatic and aromatic fractions are indicative of a large contribution of bacteria to the organic fraction of the Boom Clay. The aliphatic hopane series ranges from C27 to C33. The hopane distribution maximises at C31 and the concentrations of extended hopanes decreases rapidly with increasing carbon number, which may be explained by oxic degradation during the initial stages of burial. The second important group are the hopenes and their distribution, which maximises at C30, suggests different precursors compared to the hopanes.

Most of the HPPT biomarkers encountered in the aliphatic fraction of the Boom Clay have an oleanoidal skeleton, pointing to angiospermal plants as the main source of these structures. Beside the regular structures, a number of degraded molecules could be identified and are indicative of oxidative biodegradation. It is suggested that this degradation did not occur during or after deposition of the Boom Clay, but could have taken place in the source area or during transportation. The concentrations of resinous sesquiterpenoids, diterpenoids and sesterterpenoids in the aliphatic fraction of the Boom Clay are low.

Summarising, the strongest source parameters for aliphatic and aromatic compounds are related to continental organic matter (higher plant precursors). The higher plant derived structures can be divided in two groups, i.e., molecules derived from angiospermal plants and resinous compounds with a gymnospermal origin. For a number of biomarkers a marine, probably algal origin seems more plausible. The most abundant members of this group are short chained n-alkanes. The high concentration of autochthonous hopanoids indicates a strong bacterial activity in the depositional environment and the sediment.

The diversity in sources of the various molecular groups is reflected in slightly different distributions throughout the section. The concentrations of all organic compounds correlate

8 Hopanoids are pentacyclic triterpenoids of bacterial origin with a five membered E-ring and a side chain of variable length at C-21. The length of the side chain depends on environmental and/or early diagenetic conditions.

9 Although a wide range of higher plant derived pentacyclic triterpenoids (HPPT) have been isolated from organisms, only a limited number of skeletal structures are encountered in sediments.

10 Diterpenoids are especially abundant in the resins of conifers, although some marine organisms, mainly algae, also contain minor amounts of diterpenoids. The sequiterpenoids form a large group of molecules and originate mainly from resins of higher plants, mainly gymnosperms.

positively with TOC. This is due to the fact that the organic-rich layers yielded more extract. After normalisation to TOC, the concentrations of the aliphatic molecules showed only small variations to throughout the section, which indicates a very homogeneous composition of the aliphatic fraction. All classes reach the highest concentrations in the lower part of the Putte Member.

The relationship between the different compound classes and sedimentology was clarified by factor analysis. The first factor groups the hopanoids and the molecules derived from higher plants, mainly angiosperms. The higher plant derived compounds have a clear terrestrial origin. The molecular data and the fact that the Terhagen samples have higher scores for this factor than those from the Kruibeke quarry, advocate for a terrestrial origin of this family, and point to a southern source area, most probably the Brabant Massif. The hopanoids most probably derived from autochthonous bacteria. Their distribution is rather constant and shows no link with sedimentology.

The second factor reveals high loadings for compounds the main source of which is believed to be gymnospermal resins. This group of organic matter positively correlates with TOC and its proportional contribution to the aliphatic fraction is the highest in the black layers. This suggests that the high TOC values mainly resulted from a large supply of this gymnospermal organic matter.

The third factor group is represented by compounds for which an autochthonous source has been proposed repeatedly. The distribution of these molecules shows a link with grain size and the autochthonous signal is the strongest in the clay beds.

The weak thermal maturation of the Boom Clay is reflected in all molecular maturity parameters. No change in maturity could be detected throughout the section and only a number of parameters that depend on the presence of specific precursors in immature sediments, show significant inter-sample differences. For the PAH, present in high amounts in the Boom Clay and generally believed to be the products of strong thermal maturation, alternative pathways have been proposed (e.g., reworking of mature organic compounds from the British coalfields).

The gymnospermal derived group of organic compounds correlates with a number of diagenetic ratios that are indicative of acidic pore-water conditions. This suggests that the deposition of this pool of allochthonous organic matter was accompanied by changes in the early diagenetic conditions. Additionally, the correlation of these variables with TOC may be indicative of an influence of the amount of organic matter on early diagenetic pore-water chemistry. The organic-rich layers appear to have experienced more acidic early diagenetic conditions and are characterised by high amounts of diasterenes and thiophenes. The lowering of the pH may be explained by higher amounts of organic acids generated by the aerobic bacterial degradation of the large quantities of organic matter buried in these parts of the section. In these layers, sulphate reduction was high and probably even exceeded the formation of pyrite. Intense bacterial sulphate reduction might also have been responsible for the high amounts of hopanoids encounted in the black layers.

In the Boom Clay, only a limited number of biodegraded molecules were recognised. Also, the predominance of n-alkanes and the high abundances of short chain, acyclic structures plead against strong biodegradation in the depositional environment.

Two groups of allochthonous organic molecules are present in the Boom Clay (figure 2.8). The first one, which is dominated by HPPT, fernenes and aromatised structures derived from C3-oxygenated precursors, has the highest concentrations in the lowest part of the section. The molecular composition is characteristic for angiospermal, allochthonous organic matter. The angiospermal group positively correlates with sediment supply and reaches high relative abundances in intervals where strong continental conditions prevailed, i.e., near sequence boundaries and in periods of glacio-eustatic lowstand. This pool might therefore be part of the detrital sediment population supplied from nearby, southern landmasses.

The second group is dominated by resinous di- and sesquiterpenoids, derived from gymnosperm resins. This pool of allochthonous organic matter also includes the mature PAH. The composition of organic matter in the black layers is also characterised by a larger contribution of the gymnospermal population. This suggests that the cyclic occurrence of organic-rich layers is due to periodic fluctuations in the supply of continental organic matter. The correlation with kaolinite, and the high abundances in the Kruibeke samples compared to those from Rumst, suggest that the gymnospermal population derived from a distant, northern source. High relative contributions of this population mark strongly marine periods.

The proportional abundance of both groups of continental organic matter changes with relative sea-level. The contribution of the angiospermal group reaches maxima near sequence boundaries. The gymnospermal population is concentrated near flooding surfaces. The fact that this supply reaches maxima near maximum flooding surfaces may be explained by the flooding and the contemporaneous destruction of coastal soils, peat bogs and vegetation.

Figure 2.8: Palaeogeographical scheme of the Lower Oligocene together with the sources of allochthonous organic matter present within the Boom Clay (taken from Van Geet et al. (2003) after Wouters and Vandenberghe (1994) and Laenen (1997))

The sharp increase in TOC across the Putte/Terhagen boundary resulted from a reduction in sediment supply and an increased flux of continental organic matter. At the same time, the

high TOC values in the lower part of Putte seem to have triggered a change towards more acidic, and probably also suboxic, early diagenetic conditions. This led to a better preservation of all organic matter, and may have further increased TOC.

The same mechanism might be responsible for the variations in TOC on the scale of the silt/clay alternations (figure 2.9). In a sequence stratigraphic sense, each silt-to-clay transition can be interpreted as a transgressive system tract. The centre of the clay layer corresponds to the maximum flooding surface. The rise in relative sea-level might have increased the flux in continental organic matter near the silt to clay transition. Higher up in the parasequence, the higher sedimentation rates of the clay would have diluted the high fluxes of continental organic matter. In the highstand system tract, the input of continental organic matter decreased again to reach a minimum in the upper part of the clay or the lower part of the overlying silt bed.

Figure 2.9: Sequence stratigraphic interpretation of a silt/clay couple, and the distribution of organic matter through a parasequence (LAENEN, 1997)

Besides the two allochthonous populations, a group of acyclic aliphatic biomarkers is present, which most probably is of algal origin. The short chain n-alkanes (C<21) and branched acyclic isoprenoids are believed to represent this autochthonous pool of organic matter. Also in the aromatic fraction, no exclusively marine biomarkers could be identified. The concentrations of the short chain n-alkanes and the branched acyclic isoprenoids closely vary with glacio- eustasy. The maximum flooding surfaces are marked by a strong marine signal, while the contribution of marine organic matter is low near sequence boundaries and high in clay beds. This trend results in a negative correlation with the > 32 µm grain size fraction. Both the third-order and Milankovich trend may either be due to variations in primary production or to

changes in preservation conditions. In the first case, the increase in marine primary production in the transgressive system tracts can be explained by a larger supply of nutrients due to upwelling or flooding and erosion of coastal soils. In the second case, the fine grain size and low turbulence of the clay beds may have caused less oxidative degradation of labile organic matter than in the silts.

The amount of the marine molecular population positively correlates with the amount of nitrogen. Consequently, part of the variability in N may be assigned to changes in the type of organic matter. Slightly lower values of C/N near maximum flooding surfaces correspond to a higher contribution of autochthonous nitrogen-rich organic matter, and the high values in the Putte Member coincide with a large contribution of higher plant material (figure 2.10).

Figure 2.10: Variations in the abundance of autochthonous, probably algal organic molecules (BA and C<21 n-alkanes) throughout the glacio-eustatic sequences recognised in the type section of the Boom Clay. R(N%-TOC) shows the variations in nitrogen after subtraction of the variance related to TOC. The right-hand column shows the variations in continental (TOCcont.) and marine (TOCauth.) organic carbon (LAENEN, 1997)

The fourth important group of organic constituents are the hopanoids. This group is of bacterial origin. The high amounts of hopanoids point to intense bacterial activity at the depositional site of the Boom Clay. The stratigraphic distribution of this group of organic compounds correlates positively with TOC, suggesting that bacterial activity was the highest during the deposition of organic-rich layers. As the relative abundance of the hopanoids is

almost constant, bacterial activity was more a function of the supply of organic matter, than of its composition. Nevertheless, the positive correlation between the relative abundances of the hopanoids, the steroids and the HPPT might indicate that the bacteria preferentially degraded the pool of fresh angiospermal organic matter. The relative abundances of aromatised hopanoids in the Putte Member might result from strong acidic early diagenetic conditions.

The composition of the solvent-extractable hydrocarbon fraction points to differences in the eogenetic conditions between the black and grey clays. The high relative abundances of diasterenes, thiophenes, retene as well as other tri- and tetracyclic, partially hydrogenated terpenoids, aromatised hopanoids and aromatised higher plant derived molecules are indicative of acidic and probably suboxic eogenetic conditions during deposition of the organic-rich parts of the Putte Member, whereas the lower relative abundances of these molecules in the Terhagen Member reflect oxic and less acidic conditions in the uppermost parts of the sediment.

2.2.3 PhD thesis and Post-doc of Isabelle Deniau

A study on Boom Clay kerogen was performed by Deniau and co-workers (DENIAU et al., 2001). They isolated kerogen from a Boom Clay sample collected at 230 m depth in the HADES underground research facility of SCK•CEN at Mol. The sample showed a TOC value of 1.5 wt.%. As usual for immature sedimentary rocks, the organic matter is dominated by kerogen. The kerogen fraction, isolated in the classical way after destruction of the mineral matrix by HF/HCl treatment and extensive solvent extraction (Durand and Nicaise, 1980; Vandenbroucke, 2003) accounts for above 80 wt.% of the total organic matter (DENIAU et al., 2001; DENIAU et al., 2004).

The isolated kerogen and the whole rock samples were then subjected to a classical combination of spectroscopic, microscopic and pyrolytic methods, including "off-line" pyrolysis (LARGEAU et al., 1986). Rock-Eval pyrolysis was performed both on the whole rock and on the isolated kerogen. Kerogen and its pyrolysis residues were analysed by elemental analysis and Fourier transform infrared (FTIR) spectroscopy and 13C nuclear magnetic resonance (NMR). "Off-line" pyrolysis was also performed, and each residue was extracted and analysed by gas chromatography – mass spectrometry (GC-MS). Scanning (SEM) and transmission (TEM) electron microscopy were also performed.

Rock-Eval pyrolysis of the whole rock sample and the isolated kerogen showed that the organic matter in the sample belonged to type II. The hydrogen to carbon atomic ratio (H/C = 1.3) indicated a rather high aliphatic nature. The kerogen is relatively sulphur-poor but contains a substantial level of nitrogen. The FTIR and 13C NMR spectra confirmed the aliphatic character of the kerogen (with long polymethylenic chains) and revealed contributions of aromatic and olefinic carbons, and non-conjugated carboxyl groups. Pyrolysis showed that its chemical structure is chiefly based on a macromolecular network of long, normal alkyl chains probably cross-linked by ether bridges located at various positions on the chains. However, a substantial amount of aromatic structures also occurs in the kerogen along with minor branched hydrocarbon chains. Some alcohol and acid moieties, slightly degraded when compared to biological compounds, were also detected. This chemical structure points to a major contribution of phytoplanktonic material with a low input of terrestrial and bacterial components.

SEM showed that Boom Clay kerogen appeared to be mainly composed of aggregates of various sizes, ranging from 30 to 200 µm, without well-defined shapes. Some well-preserved structures derived from terrestrial material (pollen, plant debris) were also observed but they occur only in minor amounts, in agreement with the small contribution of terrestrial organic matter deduced from pyrolysis data. Natural sulphuration was at most a minor preservation mechanism: H2S produced by sulphate-reducing bacteria was probably trapped by available iron and transformed into pyrite framboids. TEM revealed that the kerogen is mainly amorphous even at high magnification. Two types could be distinguished: granular and homogeneous, along with a few cell walls (lamellar structures of ca. 200 nm thick, inherited from biological precursors). The predominance of amorphous organic matter indicates that the selective preservation pathway did not play a major role for Boom Clay kerogen.

The occurrence in the pyrolysate of nitrogen-containing products related to proteins along with a substantial N-content point to a role for the degradation-recondensation or mineral protection mechanisms in the formation of Boom Clay kerogen. Moreover, the distribution of fatty acids in the pyrolysate indicates that the initial degradation step was rather limited, i.e. lipids may have been quickly incorporated into the macromolecular network or protected before incorporation. Examination of polished sections using backscattered scanning electron microscopy (BSEM) indicated that the organic matter is mainly concentrated in small (3-10 µm) particles with regular outlines embedded in the clay matrix. Some other small particles, systematically associated with pyrite framboids, were also observed, just as a few structureless particles with larger size (50-150 µm) and less regular outlines interfingering with the mineral matrix. TEM observations confirmed that no regular organo-mineral organisation occurred: organic matter neither appears as a thin coating of mineral grains nor as nanolayers alternating with mineral grains. Based on these observations it appears that no tight association between organic matter and clay minerals occurs in Boom Clay. Accordingly, the Boom Clay sample did not show any indication of a major role for physical protection by minerals as a mechanism of kerogen formation.

Taken together, the organic matter in Boom Clay was mainly preserved through the degradation-recondensation pathway. However, the occurrence of a large number of mid- chain ketones in the pyrolysate of the Boom Clay kerogen points to the additional involvement of the so-called oxygen cross-linking pathway in its formation.

During the initial heating step (300°C for 20 min) of the "off-line" pyrolysis, carried out prior to the actual cracking step (400°C for 1 h) of the macromolecular structure of the low maturity kerogen, Deniau et al. (2001) observed a release of a large amount of compounds. These compounds, essentially thermolabile products which are thought not to be representative of the bulk kerogen structure, accounted for ca. 25 wt.% of the total organic matter of the kerogen. As such, they might be of importance in relation to the influence of heat emission on the kerogen during long term storage (see chapter 4). The nature and distribution of the various components of this thermolabile fraction were determined by combined gas chromatography/mass spectrometry (GC/MS) (DENIAU et al., 2004).

In principle, these components could have three modes of occurrence (DENIAU et al., 2004): 1. "Free" compounds, potentially soluble but non-extractable due to tight physical trapping within macromolecular structures. Two distinct origins can be considered for these "free" compounds – either they occurred as trapped components in resistant biomacromolecules that contributed to the kerogen through selective preservation, or they were secondarily trapped from the bitumen fraction during kerogen formation. Such compounds should exhibit a low extent of degradation.

2. "Labile" moieties linked to the macromolecular structure of the kerogen by covalent bonds with low thermal stability. 3. "Bound" moieties linked to the macromolecular structure of the kerogen by relatively stable covalent bonds undergoing partial cleavage at 300°C.

Deniau et al. (2004) used a Boom Clay sample collected at 223 m depth in the HADES underground facility of SCK•CEN at Mol, identical to the one used by Deniau et al. (2001). Comparison of the FTIR spectrum of unheated kerogen with that after heating at 300°C showed a preferential release of carboxyl and C─O functions. In agreement with these spectroscopic features, the chromatogram of the thermolabile fraction is dominated by a series of carboxylic acids (as methyl esters). Several other types of compounds, including hydrocarbons, ketones, phenols, furans and pyrroles were identified by GC/MS either directly or through selective detection of diagnostic ions (DENIAU et al., 2004).

The thermolabile fraction of the Boom Clay kerogen (ca. 25 wt.% of total organic matter), released upon short heating at 300°C, contains a large variety of components including hydrocarbons (n-alkanes, pristenes, sesquiterpenes and hopanoids), ketones (isoprenoid and normal), fatty acids, phenolic, furanoic and pyrrolic compounds11. This fraction corresponds

11 A detailed description of the thermolabile fraction is given below: 1. Hydrocarbons a. n-Alkanes in the thermolabile fraction ranging from C12 to C30 with a maximum at C14. These n-alkanes appear to correspond to "free" components that were tightly trapped in the macromolecular structure of the kerogen. This distribution points to contributions from several sources: terrestrial higher plants and marine bacteria. In contrast, n-alkanes occur in the bitumen ranging from C15-C31, with a maximum at C25. Such variations can be accounted for by differences in the extent of biodegradation, with a greater preservation of the trapped n- alkanes. Note: acid treatment of crude rock also results in the release of saturated hydrocarbons, but these are only "weakly" trapped. b. Two regular C19 isoprenoid alkenes (prist-1-ene prist-2-ene), thought to originate from the cracking of phytyl (C20 regular isoprenoid) moieties covalently linked to the macromolecular structure of kerogen and derived from the phytyl side chain of chlorophyll and/or tocopherol. c. Three sesquiterpenoids, which are ubiquitous components of higher plant resins and essential oils and are thus indicators of a terrestrial input. These compounds were probably "free" components secondarily trapped in the macromolecular structure of the kerogen. d. Hopanes (C27, C29 and C30), along with predominant C27 and C29 hopenes. Hopanoid hydrocarbons are characteristic biomarkers of bacteria. The distribution differences with hopanoids detected in bitumen reflect either (i) the occurrence as "free", secondarily-trapped compounds, or (ii) a difference in extent of degradation. 2. Ketones a. The isoprenoid C18 ketone 6,10,14-trimethylpentadecanone occurred in relatively large amount and could originate from the cracking of ether-bound "labile" moieties. These are thought to correspond to phytyl chains covalently linked to the macromolecular structure of the kerogen by oxidative incorporation. b. Normal, straight chain alkan-2-ones ranging from C10 to C25 with a bimodal distribution and a marked odd-over-even carbon number predominance. These long chain ketones may correspond to "free" components trapped in biomacromolecules that contributed to the kerogen. 3. Carboxylic acids. Saturated normal C8 to C28 fatty acid methyl esters dominate the chromatogram, although some unsaturated compounds are also detected. The saturated series exhibits a maximum at C16 and even-carbon-numbered components predominate. Such a distribution points to multiple sources: degradation products of longer acids, markers of microalgal inputs, bacterial contribution and markers of higher plant waxes. Long-chain acids probably originate from the oxidation of n-alkan-2-ones. They likely correspond to "free" trapped compounds trapped in an algal structure. 4. Numerous phenolic components occur in the thermolabile fraction, mostly corresponding to guaiacyl- type compounds (methoxyphenols). They are typically formed upon cracking of the macromolecular structure of lignin and lignin-derived products and are thus largely used as higher plant markers. They

to "free" components trapped in the macromolecular structure of the kerogen and to "labile" moieties linked by covalent bonds with relatively low thermal stability. A part of the "free" compounds was probably secondarily trapped from the bitumen during kerogen formation. In contrast, some long chain n-alkanones and fatty acids may have been present as trapped compounds in resistant biomacromolecules selectively preserved in the kerogen. These trapped compounds exhibit a low mobility and cannot be extracted from the unheated kerogen, but they can become mobile and soluble upon thermal treatment.

It is remarked that the open pyrolysis and heating conditions maintained in Deniau et al. (2004) do not afford specific information on the behaviour of the kerogen under the thermal stress that would be associated with the disposal of high-activity nuclear waste. The aim of these experiments was rather to obtain information on the chemical composition of the Boom Clay kerogen.

2.2.4 TRANCOM-II studies

Van Geet (2002) describes the results of Rock-Eval data on selected samples of core HADES 2001/4 (borehole for the Morpheus piezometer). The sampling interval comprises part of the Terhagen and Putte Members of the Boom Clay. Figure 2.11 illustrates the TOC measurements of these samples to their position of sampling and compares with the data found by Laenen (1997). TOC values vary between 0.85 and 4.13 wt.%. From this plot it is clear that the TOC value is sharply increasing above the Putte-Terhagen boundary. A histogram of the TOC data shows that there is a bimodal distribution of TOC values with two maxima (figure 2.12). A cut-off value of 2 wt.% was chosen to separate the populations in order to perform a statistical analysis on the data. A similar observation can be made for HI (which is linearly correlated with TOC for the Boom Clay), and a cut-off of 130 mg HC/g TOC can be chosen (VAN GEET, 2002).

correspond to "labile" moieties released through the cleavage of covalent bonds with relatively low thermal stability. Because demethoxylation takes place during the first steps in the diagenetic degradation of lignin, their large contribution indicates a low level of lignin degradation in Boom Clay kerogen. The lignin in Boom Clay probably originated from resinous plants, i.e. gymnosperms. 5. Various furans and related cyclic ketones were found. They are typical pyrolysis products of polysaccharides and thus reflect the presence in Boom Clay kerogen of "labile" polysaccharide moieties that are cleaved under moderate thermal stress. 6. Several nitrogen- and oxygen-containing compounds, based on a pyrrole or tetrahydropyrrole ring, were identified, the ethyl methyl pyrroledione being dominant. They likely correspond to "free" compounds secondarily trapped in the kerogen and are derived from porphyrin degradation.

Figure 2.11: Stratigraphical distribution of kerogen types based on a cut-off value of the HI of 130 mg HC/g TOC, including data from Laenen (1997) and Van Geet (2002)

Figure 2.12: Histogram of measured TOC values within core HADES 2001/4 (VAN GEET, 2002)

A crossplot of S2 values vs. TOC allows to define the principal kerogen types. Figure 2.13 illustrates that the first population (TOC < 2 wt.%) is composed of type III kerogen, while the second population (TOC > 2 wt.%) seems to consist of both type II and type III kerogen. Moreover, the regression lines through the data clouds have a positive intercept on the TOC axis. These intercepts can be interpreted as the amounts of "dead organic carbon", i.e., carbon not associated with hydrogen and therefore incapable of producing hydrocarbons upon pyrolysis (also denoted as type IV kerogen). It can thus be derived that the first population has a dead carbon value of 0.7 wt.% while the second population has a dead carbon value of 1.3 wt.%. The crossplot of S2 vs. TOC can be corrected for this type IV kerogen, allowing for a better appraisal of the different kerogen types within both the populations.

Figure 2.13: a) Crossplot of S2 versus TOC defining kerogen type: b) crossplot of S2 versus corrected TOC ("dead carbon-free") defining kerogen type (VAN GEET, 2002)

The Tmax data of the Rock-Eval analysis range between 412 °C and 428 °C for all samples, indicating a low maturity over the entire stratigraphic range.

2.2.5 PhD thesis of Pascale Blanchart

Within her PhD, Blanchart (2011) studied different biomarkers contained within the organic extractable fraction of the Boom Clay organic matter: • Hydrocarbons o n-alkanes are constituted of a linear carbon chain. Their distribution is indicative of the source (marine or continental) of the organic matter o iso-alkanes are alkanes containing one or more methyl groups. Examples include pristane (C19) and phytane (C20), two acyclic isoprenoids derived from the degradation of the phytol chain (linked to chlorophyll). o terpenoids are derived from the diagenesis of bioterpenoids, i.e., lipids with a large diversity in structure (linear or cyclic), chemical functionalities (alcohols, aldehydes, ketones, etc.) and biological functionalities. These molecules originate from the assemblage of isoprene units and are classified according to the number of units they contain. During diagenesis, the functional groups are progressively eliminated and ring structures are either aromatised or hydrogenated. . Monoterpenoids (two isoprene units) . Sesquiterpenoids (three isoprene units) . Diterpenoids (four isoprene units) . Sesterpenoids (five isoprene units) . Triterpenoids (six isoprene units) are derived from squalene • Hopanoids contain five hydrocarbon rings and are differentiated following their carbon content, configuration and presence (or not) of unsaturated bonds. • Oleanoids, usanoids and lupanoids • Steroids originate from the enzymatic oxidation of squalene followed by ring formation. They contain 26 to 30 C atoms and display a large diversity in structure reflecting source and modification pathways of the organic matter (figure 2.14). • Aromatic biomarkers originate from the reorganisation of aromatic biomolecules. • Acyclic lipids o Fatty acids are amphiphilic molecules containing a carboxylic acid group. The length of their carbon chain and the degree of unsaturated bonds are indicative of the source of organic matter o Alkanoles are amphiphilic molecules containing an alchol group. • Lignin and its derivates are biopolymers synthesised by plants with a rigid cell wall (wood structures)

Blanchart (2011) used (undisturbed) Boom Clay samples collected in the HADES underground research facility in Mol (Belgium). One sample originated from the excavation of the PRACLAY gallery (October 2007), the other sample was taken from the window located at URL R16W. The kerogen from both samples falls within the envelope of type III organic matter. The molecular organic composition is consistent with a higher plants contribution to the Boom Clay organic matter but also suggest additional contributions of

marine organic matter and bacteria. All geochemical data indicate that the organic matter is immature (BLANCHART, 2011).

Figure 2.14: Relationship between carbon numbers in sterols and their environmental source (BLANCHART, 2011)

However, despite the fact that both samples were collected at locations not far apart in terms of vertical and lateral distances, some slight geochemical variations could be noticed. For instance, the samples differed in TOC (1.0 vs 1.7 wt.%), HI (87 vs 150 mg HC/g TOC) and in relative contributions of some specific biomarkers. This indicates organic facies variability within the Boom Clay at the scale of the galleries (BLANCHART, 2011).

2.3.Comparison of kerogens from Boom Clay, Callovo‐Oxfordian argillite and Toarcian shales

Besides Boom Clay, other argillaceous formations are being researched by other national waste management organisations for their suitability to host a radioactive waste disposal facility. Callovo-Oxfordian argillites (Bure Meuse/Haute-Marne, France)12 are investigated by the French Agency for Nuclear Waste Management (ANDRA), while, in parallel, the Institute for Radioprotection and Nuclear Safety (IRSN) conducts experiments on Toarcian argillites at the Tournemire experimental site (Aveyron, France)13 in order to improve the knowledge of argillaceous media. All these argillaceous geological formations contain a substantial level of sedimentary organic matter which is dominated by the insoluble macromolecular fraction called kerogen.

12 The Callovo-Oxfordian argillite occurs in the eastern part of the Paris Basin. It originates from the Callovo- Oxfordian formation, which is investigated via the underground research laboratory of Andra at Bure (Meuse/Haute-Marne, France). In this area, the argillite occurs as a ca. 130 m thick marine formation. It was deposited on a sea bed, far from the coast 150 Ma ago.

13 The Tournemire experimental site, investigated by IRSN, is located in southern France (Aveyron). An old railway tunnel, about 2 km length, excavated through the Tournemire massif, is being used as a test laboratory. This sedimentary rock corresponds to ca. 250 m of thick indurated Toarcian shales.

Some studies were previously performed on the organic matter of the Callovo-Oxfordian Clay (ELIE et al., 2000; ELIE et al., 2004; LANDAIS and ELIE, 1999). In particular, Landais and Elie (1999) showed that the Callovo-Oxfordian kerogen at Bure is immature and generates low quantities of liquid effluents under a thermal stress. In the case of the Toarcian Clay at Tournemire, if previous analyses showed a high degree of maturity for the organic matter of the black shales (lower Toarcian), they did not give clear-cut indications on the maturity of the organic matter of the argillite (upper Toarcian) (DE WINDT et al., 1999).

Deniau et al. (2008) performed a global geochemical characterisation of the kerogen isolated from samples originating from the Bure (core from Andra EST 209 borehole) and Tournemire (core from EX1 borehole) sites and derived information on their thermal reactivity. Elemental analysis showed a pyrite content for the two kerogens as high as 65 wt.%. The TOC content in the Bure sample amounted to 0.60 wt.% and in the Tournemire sample to 0.90 wt.%. The results for the kerogens were compared with those obtained on Boom Clay kerogen.

It was found that Boom Clay kerogen is the most aliphatic one, with an H/C atomic ratio as high as 1.27, while weaker ratios (in the 0.9-1 range) are observed for the Bure and Tournemire kerogens. The Boom Clay kerogen is also characterised by an especially high O/C atomic ratio (0.27) whereas the abundance of oxygenated groups is markedly lower for the Tournemire one and, to a lesser extent, for the Bure kerogen. The H/C atomic ratios versus the O/C atomic ratios (van Krevelen) diagram are reported in figure 2.15. For the three kerogens, a mixture of type II (marine) and type III (terrestrial) organic matter is observed, but with different levels of maturity. The diagram shows that the Boom Clay sample is located at the onset of the diagenetic evolution stage while, on the contrary, the Tournemire kerogen is towards the end of this stage (i.e., the kerogen has lost a large part of its oxygenated functions). The Boom Clay, Callovo-Oxfordian and Toarcian kerogens therefore appear to contain markedly different quantities of oxygenated functions.

Boom Clay

Toarcian

Callovo-Oxfordian

Figure 2.15: Elemental analyses of the kerogens from the Boom Clay, Callovo-Oxfordian and Toarcian argillites plotted as H/C versus O/C atomic ratios in a Van Krevelen diagram. Type I: lacustrine; Type II: marine; Type III: terrestrial (DENIAU et al., 2008)

The observed Tmax values, ranging from 404 to 430 °C, increase from Boom Clay to Callovo- Oxfordian, and from Callovo-Oxfordian to Toarcian kerogens. The hydrogen indexes are

relatively low with HI values around 330 mg HC/g TOC for the Boom Clay and Tournemire kerogens and a value of ca. 220 mg HC/g TOC for the Bure one. Comparison of the data also shows a pronounced decrease in OI values from Boom Clay to Bure, and from Bure to Tournemire kerogens.

The FTIR spectra of the three kerogens exhibit the same absorption bands, but with differences in relative intensities. The major difference between the three spectra deals with the relative intensity of the band corresponding to C=O groups, which is relatively intense for the Boom Clay kerogen whereas it only corresponds to a weak shoulder rather than a true band in the case of Tournemire, while for Bure an intermediate situation is observed. Similarly, the absorptions corresponding to C-O groups are relatively more intense for the Boom Clay kerogen.

The predominant compounds generated from the three kerogens by pyrolysis are aliphatic (n- alkanes and acyclic isoprenoids) and aromatic (alkylbenzenes, polycyclic aromatic compounds) hydrocarbons. A large variety of oxygen-containing compounds is identified in the pyrolysate of the Boom Clay kerogen whereas a much lower number of components are found in the case of the Tournemire sample and, to a lesser extent, of the Bure one. Common O-containing compounds include furanic and phenolic derivatives while series of normal saturated carboxylic acids and ketones are detected only in the pyrolysates of Boom Clay and Callovo-Oxfordian kerogens. The furanic compounds are indicative of the presence of polysaccharidic components in the kerogens. The phenolic compounds are typical pyrolysis products of lignin and altered lignin although a part may also originate from other components such as cellulose and chitins. Most of the above oxygen-containing compounds are characterised by a relatively high aqueous solubility, like phenols.

Concluding, the Boom Clay sample appears to be the most immature kerogen. On the contrary, the Tournemire sample is more mature and is located towards the end of diagenesis. It was submitted to some thermal stress during its geological history and thus has lost a large part of its oxygenated functions. The Bure kerogen exhibits an intermediate position corresponding to mid-diagenesis.

2.4. Conclusion on the solid Boom Clay organic matter

Organic matter in Boom Clay originates from different sources: allochthonous (terrestrial; angiospermal and gymnospermal + coal fragments) and autochthonous (marine + bacterial). Throughout the stratigraphy, all sources are always found, but in slightly different ratios. The organic-rich layers, for example, reflect a clearly higher continental input (VANDENBERGHE, 1978), but are the result of an increase of the material from all sources. Moreover, a better preservation and a lower sediment supply enhanced the organic carbon content within these layers (LAENEN, 1997).

The organic matter of the Boom Clay shows little effects of alteration, except for the reworked organic matter from older sediments. The organic matter preserved within early diagenetic concretions (e.g., woodfragment in S41 concretion, photograph in De Craen (1998), p. 57) is not very different from the one found in the surrounding clay matrix, but only better preserved (VAN GEET et al., 2003).

From the TOC and Rock-Eval analyses, two populations of organic matter can be discriminated. The first population has TOC values below 2 wt.% and is omnipresent throughout the stratigraphically sampled parts (Putte, Terhagen and Belsele-Waas Members). The second population has TOC values above 2 wt.% and is mainly constrained to the Putte Member. The Putte Member above the HADES level and the Boeretang Member are respectively low- and non-sampled. The obtained data are comparable between Van Geet (2002) and Laenen (1997), which implies that the overall trends in the stratigraphy can be extrapolated from samples collected in Mol to regional scale. However, variations are also present due to the influence (mixing) of the different source areas of the organic matter (LAENEN, 1997).

From the analysis of the Rock-Eval data it is concluded that the second population of data points consists of a mixture of type II, III and IV kerogen, while population 1 only contains type III and IV kerogen. Figure 2.11 gives the stratigraphical distribution of the two populations based on a cut-off value of the HI of 130 mg HC/g TOC for the data of Van Geet (2002) and Laenen (1997). Consequently, a mixture of organic matter derived from higher plants and reworked organic matter is found throughout the section of Boom Clay. However, in the Putte Member an additional input of autochthonous marine organic matter is found which may vary between 30 and 50 wt.% of the non-dead carbon fraction (VAN GEET, 2002). The results of Van Geet (2002) are comparable to those of Laenen (1997), who argued that type II kerogen was mainly found in sections with a high TOC content. However, as pointed out by both Laenen (1997) and Vandenberghe (1978), these sections also displayed a higher input of terrestrial organic matter.

The stratigraphical distribution shows that S2 slightly increases from the Belsele-Waas Member up to the Terhagen/Putte boundary. Above it, the mean value and variance is increased (Putte Member: S2 = 1.76±1.42 and Terhagen Member: S2 = 0.74±0.49). The HI and OI values mirror each other in the stratigraphic distribution, but higher values are reported by Van Geet (2002) compared to Laenen (1997). This might be related to the different sampling (clay pit versus fresh borehole) or to a lateral variation in kerogen composition over the sedimentary basin. However, the similar and low Tmax values measured in both studies show that effects of oxidation are negligible.

Due to a correlation of TOC and HI, it is suggested that organic-rich layers have the highest relative abundance of type II kerogen. Moreover, the pyrolysable oxygen compounds (S3) increase as a function of TOC. Consequently, the black layers have the highest proportional contributions of type II organic matter, but at the same time the highest absolute concentrations of both pools of organic matter.

The Tmax values show only a limited variation between 412 and 428 °C, indicating a low and homogeneous maturity throughout the stratigraphical section.

3. Dissolved organic matter

3.1. Dissolved organic matter – definitions and background

Dissolved organic matter (DOM) is defined as the fraction of the total water-soluble organic matter which passes through a 0.45 µm filter. It is composed of non-humic substances and of humic substances. Non-humic substances are composed of easily degradable molecules originate from (the degradation of) cellular compounds or from microbial synthesis. Examples include carbohydrates, proteins, amino acids, lipids, tannins, lignin and fatty acids.

On the other hand, humic substances are defined as those organic compounds found in the environment that cannot be classified as any other chemical class of compounds (GAFFNEY et al., 1996; STEVENSON, 1982). They arise from the decomposition of plant and animal tissues yet are more stable than their precursors. Their size, molecular weight, elemental composition, structure, and the number and position of functional groups vary, depending on the origin and age of the material (GAFFNEY et al., 1996). Examples of different functional entities are aromatics, aliphatics, phenolics and quinones (GAFFNEY et al., 1996). However, the similarities between different humic substances are more pronounced than their differences. The range of elemental composition of humic materials is relatively narrow, being approximately 40-60 % carbon, 30-50 % oxygen, 4-5 % hydrogen, 1-4 % nitrogen, 1-2 % sulphur and 0-0.3 % phosphorus. Because the natural organic matter in Boom Clay has already undergone substantial transformation after deposition, Boom Clay DOM are considered to consist solely of humic substances (the non-humic fraction being negligible).

Humic substances in natural groundwater are present in the form of humic colloids, consisting of the organic entities and associated mineral structures and complexes metal ions. These humic colloids can play a major role in radionuclide migration in natural systems (CHOPPIN, 1992; GECKEIS et al., 2002; MAES et al., 2010; MAES et al., 2006; TRANCOM-II, 2004). The potential impact of humic-colloid-mediated radionuclide transport depends on (i) the sources/concentrations of humic substances, (ii) their stability and mobility and (iii) the interaction with radionuclides. In addition to the complexation strength, the complex kinetic behaviour needs to be considered as well (GECKEIS et al., 2002; MAES et al., 2010; MAES et al., 2006).

The importance of the complexation of radionuclides of higher oxidation states (Z ≥ 2) with natural organics in natural systems is widely recognised. In groundwater, soluble humic substances, including humic and fulvic acids, are already complexed with metal ions of water constituents, e.g. Fe3+/Fe2+, Ca2+, REE-ions etc. and are present as humic colloids. These colloids act as soluble exchangers and hence, readily create pseudocolloids of these ions. Such humic colloids or pseudocolloids are easily separable from solution by ultrafiltration. Humic substances may thus cause an enhancement of migration, or retention, of some radionuclides depending on the filtration capacity of a given system. For the quantification of each of these processes, a basic knowledge of the nature of humic acids and their complexation behaviour under groundwater conditions is indispensable (KIM et al., 1991).

Humic substances are operationally classified into humic acids (HA) and fulvic acids (FA), based on differences in solubility in acidic and alkaline media. Humic and fulvic acids are soluble in alkaline media, whereas humic acid flocculates under acidic (pH < 2) conditions.

Table 3.1 represents a statistical evaluation of the elemental composition of humic and fulvic acids isolated from samples taken all over the world (RICE and MACCARTHY, 1991; STEVENSON, 1982). On a weight basis, C and O are the most important elements. Humic acids have a higher contribution of C than fulvic acids. In contrast, humic acids have a lower O content while the H content is about the same. This difference in composition also reflects differences in size and structure (GAFFNEY et al., 1996). The structure of fulvic acids is somewhat more aliphatic and less aromatic than humic acids; and fulvic acids are richer in carboxylic acid, phenolic and ketonic groups. This is responsible for their higher solubility in water at all pH values. Humic acids, being more highly aromatic, become insoluble when the carboxylate groups are protonated at low pH values. This structure allows the humic materials to function as surfactants, with the ability to bind both hydrophobic and hydrophilic materials.

Humin is soil science term for the fraction of natural organic materials that is insoluble in water at all pH values and was discussed in the previous section.

Table 3.1: Mean elemental composition of humic and fulvic acids from different sources expressed as weight percent on a dry and ash-free basis

Element Humic acidsa Fulvic acidsa Humic acidsb Fulvic acidsb (%) (%) (%) (%) C 55.0±2.6 48±4 50-60 40-50 H 5.1±0.5 5.1±0.7 4-6 4-6 N 2.7±0.9 2.8±0.9 2-6 <1-3 S 1.6±1.2 1.4±0.5 0-2 0-2 O 36±4 44±4 30-35 44-50 a Data from Rice and MacCarthy (1991) b Data from Stevenson (1982)

Over the years, several chemical structures for natural organic matter were proposed (see figures 3.1 and 3.2). The properties of humic acid included in the more realistic pictures include: a three dimensional structure of high aromaticity bridged by aliphatic chains, the presence of nitrogen atoms and a large number of metal complexing sites, the possibility of intermolecular and intramolecular aggregation, etc. These properties are responsible for the polyelectrolyte nature of humic acid (SCHULTEN and SCHNITZER, 1997).

Figure 3.1: 2D drawing of a humic acid structure proposed by Schulten and Schnitzer (1997)

Figure 3.2: Geometrically optimised 3D structure of humic acid (C308H335O90N5, 738 atoms) using molecular mechanics calculations (SCHULTEN and SCHNITZER, 1997)

In general, the molecular weight of humic substances may range from several 100 Da for 6 aquatic materials to more than 1×10 Da for soil-derived materials (MACCARTHY, 2001; STEVENSON, 1982). Therefore, humic substances are said to be polydispersive. The lowest molecular weight fraction possesses the highest total acidity on a weight basis. Humic acids have a larger content of aliphatic chains, a lower amount of functional groups and usually a higher molecular weight as do fulvic acids. The aliphatic chains of the fulvic acids are more oxidised, resulting in a higher carboxylic group content, charge, solubility, and lower molecular weight. The molecular weight can be an extremely important parameter to predict the transport of colloids in fractured and porous media. The carrier function of the organic colloids can be limited to the smallest particles. When studying the transport of natural organic matter, it can sometimes be important to distinguish between the different fractions.

Fractionation of the mixture can easily be performed by centrifugation, ultrafiltration, or gel permeation chromatography.

Despite the aforementioned differences between humic and fulvic acids, they should not be considered as two totally different types of products but rather as two 'extremes' in a continuum of compounds. Humic and fulvic acids are mostly classified as colloidal material, with a size varying between 1 nm and 1 µm (STEVENSON, 1982).

3.2. Functional group content of dissolved organic matter in Boom Clay

3.2.1 Introduction

Many of the chemical properties of dissolved organic matter (DOM) in relation to cation binding and charge development depend on the type and quantity of acid organic functional groups associated with the humic molecules (KIM et al., 1991; LUMSDON and FRASER, 2005; STEVENSON, 1982). Therefore, numerous attempts have been made to elucidate the structural composition of humic materials through the acquisition of physical evidence from analytical methods.

Numerous methods are used to study the chemical functional groups in a qualitative and quantitative way: • Visible light, ultraviolet and fluorescence spectroscopy, • 13C-, 1H- and 31P-nuclear magnetic resonance, • Fourier transform infrared spectroscopy, • Mass spectroscopy, • Electron spin resonance, • Pyrolysis gas chromatography and/or pyrolysis mass spectrometry, • Electron microscopy, • Scanning transmission X-ray microscopy (STXM), • pH titration (or potentiometric titration)

Frequently used spectroscopic methods are UltraViolet / VIsible Spectroscopy (UV/VIS), Fourier-transform infrared (FTIR) and fluorescence spectroscopy. UV/VIS spectrograms typically show a monotonous increase with decreasing wavelength due to aromatic and other organic chromophores. On the other hand, fluorescence (excitation and emission) signals give information related to the structure, functional groups, conformation and heterogeneity. However, due to significant overlapping and peak broadening linked to the multicomponent nature, straightforward identification and interpretation of fluorescence signatures is difficult.

A variety of functional groups and structural entities has been identified with the above mentioned methods and these groups are linked to the natural organic matter matrix which is represented as R in figure 3.3 (R may be both aliphatic and aromatic) (STEVENSON, 1982). Because of the large variety of functional groups, humic and fulvic acids are often referred to as polyfunctional structures.

Figure 3.3: Important structural groups of organic molecules (STEVENSON, 1982)

The functional groups present in natural organic matter are of great importance as centers of metal ion binding. The functional group content of soil humic substances ranges from 6.4 to 14.2 eq/kg for fulvic acids and from 5.6 to 7.7 eq/kg for humic acids (STEVENSON, 1982). Average values for the COOH, the acidic OH and the weakly acidic and alcoholic OH content in humic acids are respectively 3.6, 3.9 and 2.6 eq/kg. For the fulvic acids the values for these groups are respectively 8.2, 3.0 and 6.1 eq/kg (STEVENSON, 1982). These groups are considered as the most important in metal ion complexation reactions of natural organic matter.

3.2.2 Elemental and spectroscopic characterisation of Boom Clay dissolved organic matter

Within the EC project MIRAGE-II, member laboratories of the COCO-Club14 launched a joint action to characterise selected humic and fulvic acids for their elemental compositions, inorganic impurities, spectroscopic properties, size distributions and proton exchange capacities (KIM et al., 1991). Among these were also humic and fulvic acids extracted and purified from Boom Clay pore water. Humic acids were isolated by acid precipitation and centrifugation, while fulvic acids were obtained by solid phase extraction on a XAD-8 column. Both fractions were further purified by acid washing and NaF treatment. It is remarked here that a first attempt to use humic acid isolated from the Boom Clay solid phase was not successful, because of its insolubility after isolation by acid treatment (KIM et al., 1991)15. In a second, successful attempt, humic substances are extracted from pore water of the Boom Clay. The humic acid was used in protonated form (Boom-Clay-HA (H+)) while the fulvic acid was kept in its Na-form (Boom-Clay-Fa (Na+)) due its small quantity.

14 COCO is the abbreviation of Colloids and Complexes. For Belgium, the participating laboratory was the Laboratory for Colloid Chemistry, Katholieke Universiteit Leuven (prof. A. Cremers and prof. A. Maes)

15 This is probably due to the fact that larger humic molecules are liberated when suspending Boom Clay solid phase in water, which are irreversibly flocculated upon acid treatment (see later).

Elemental analysis showed that the elemental composition of Boom-Clay-HA (H+) was very different from Aldrich16 and Gorleben17 humic acids, having a much higher carbon content and lower oxygen content (table 3.2). On the other hand, Boom-Clay-FA (Na+) resembles very much the Aldrich and Gorleben humic substances. The O/C and H/C ratios can be useful to indicate the presence of various structures. The H/C value approaching unity implies that the chemical structure of humic acid predominantly consists of aromatic bodies containing carboxyl, ester, ether and quinone groups. Aliphatic structures including primary amino groups increase the H/C ratio to greater than unity. The O/C ratio reflects the carbohydrate content: the larger the ratio, the higher the amount of carbohydrate that might be involved. The atom ratios of H/C and O/C for the investigated humic and fulvic acids together with data for aquatic humic and fulvic acids in general are shown in table 3.3. The humic and fulvic acids from Boom Clay show somewhat higher H/C ratios than the literature values for a "mean" humic/fulvic acid and a considerably lower value of O/C for the humic acid.

Table 3.2: elemental composition of humic and fulvic acids (in percent) normalised to 100% or organic components (KIM et al., 1991), compared to literature values (RICE and MACCARTHY, 1991; STEVENSON, 1982)

Element Boom-Clay-HA Boom-Clay-FA Lit.values HA Lit.values FA (H+) (Na+) C 62.4 56.8 50-60 40-50 H 6.1 5.5 4-6 4-6 N 2.9 1.8 2-6 <1-3 O 27.0 34.7 30-35 44-50 S 1.7 1.2 0-2 0-2

Table 3.3: the H/C and O/C ratios of humic and fulvic acids (KIM et al., 1991), compared to literature values (RICE and MACCARTHY, 1991; STEVENSON, 1982)

Boom-Clay-HA Boom-Clay-FA Lit.values HA Lit.values FA (H+) (Na+) H/C 1.16 1.16 0.94 – 1.10 1.02 – 1.28 O/C 0.32 0.46 0.47 – 0.50 0.51 – 0.76

The major inorganic impurities in the purified humic substances solutions were found to be Al, Ca, Cr, Fe, Ti, Mg, Na and Si. The relatively high Fe content in purified substances is indicative of the strong complexation of Fe with humic acid, which cannot be quantitatively decomposed with 0.1 M HCl. However, the purification procedure is satisfactory with the total impurities being about 0.30 meq/g for Boom-Clay-HA, which represents ca. 7.1 % of its proton exchange capacity (see section 3.2.3). Fulvic acids generally show lower concentrations of inorganic impurities. Trace metal ions (transition metals, REE, metalloids) were also analysed and were observed not to be substantially decreased upon purification. This reflects the fact that multivalent metal ions such as U, Th, REE, etc. are difficult to

16 A commercial humic acid from Aldrich Co. It is available as a Na-salt and is of natural mined origin.

17 Gorleben is located in the northern German Plan (Lower Saxony). The site consists of a Permian salt dome overlain by an aquifer and was researched as a possible nuclear waste repository in the 80 s and 90s.

remove completely by acid precipitation of humic acid, because of their strong complexation properties. The fulvic acids are purified finally by ion exchange and show lower concentrations of trace metal ions than humic acids. In both purified fulvic and humic acids the total trace metal concentrations are found to be in general negligibly small.

The UV spectrum of humic acids shows an absorption continuum without recognisable peaks, increasing with decreasing wavelength. The E4/E6 value, used widely to characterise humic acids is the ratio of absorption intensities at 465 and 665 nm. However, its interpretation is not straightforward: it might be dependent on the extent of humification, condensation of aromatic constituents, or particle size (molecular weight). The E4/E6 ratio is therefore of limited use for the characterisation of humic substances. For Boom Clay humic and fulvic acids, the E4/E6 ratios (pH 8.5, I = 0.1 M NaClO4) are respectively 6.3±0.4 and 5.9±0.5.

IR spectroscopy of humic acid provides information on particular bond vibrations (functional groups) at specific frequencies. However, IR spectra reveal little about the chemical structure of humic acids. Thus, IR spectroscopy is useful only for a gross characterisation of humic acids of diverse origins, and for the evaluation of the effects of different extraction processes and chemical modifications such as methylation, acetylation, saponification and the formation of derivatives. In addition to purified Boom Clay humic and fulvic acids, also humic acids extracted from the solid Boom Clay have been analysed (Boom-Clay-HAS(H+)) (figure 3.4). The absorption peaks found on the spectra were attributed to OH stretching (3400 cm-1), stretching vibrations of aliphatic C-H bands (2920 and 2860 cm-1), C=O stretching (1720 cm- 1), C=C double bonds conjugated with C=O and/or COO- (1600-1650 cm-1), bending of aliphatic C-H groups (1450 cm-1), C-O stretching and OH bending (carboxylic groups) (1220 cm-1).

Figure 3.4: IR spectra of Boom Clay humic acid from the clay matrix as well as humic and fulvic acids from the pore water (KIM et al., 1991)

3.2.3 Potentiometric characterisation of Boom Clay dissolved organic matter

Various functional groups are associated with humic acids. However, proton exchange capacity measurements usually take only two groups into consideration: carboxyl and phenol groups. Methods are available to distinguish between these groups: the Ba(OH)2 procedure can be used for the total exchange capacity and the Ca-acetate procedure for the carboxyl group capacity. The difference between the total and carboxyl group capacities is taken to be the phenol group capacity. The latter represents, in fact, the capacity of all weak acid groups, of which the phenol group is known to be predominant. Additionally, the proton exchange capacity can be measured by direct pH titration (KIM et al., 1991).

Titration hysterics may occur during direct pH titration. The causes for these hysterics are not well understood. Determination of the proton exchange capacity by alkaline treatment may also be falsified by the contact of even trace amounts of O2(g) as well as CO2(g). At high pH,

alkali is consumed through auto-oxidation of organic substances in the presence of even trace amounts of O2(g), with a corresponding drop in pH on standing. In addition, trace amounts of CO2(g) can result in a serious carbonate effect in alkaline solution.

3.2.3.1. Direct pH titration in the MIRAGE II project (KIM et al., 1991)

Within the EC-project MIRAGE II, direct pH titration by dissolving humic acid in excess base and titrating with acid (backtitration) followed by base (forwardtitration) was performed + on purified Boom Clay humic acid (Boom-Clay-HA(H ) 0.2 g/L; I = 0.1 M NaClO4; T = 20°C) (KIM et al., 1991). The first derivative of the titration curve is plotted against the amount of excess base and the amount of excess base at the maximum of the first derivative represents the exchange capacity of the humic acid. This maximum appears in the neutral pH range and is equal to 4.22±0.02 meq/g for Boom Clay humic acid (six separate determinations). Direct pH titration of humic acids therefore gives the amount of proton exchanging groups in the acidic to neutral range (mainly carboxylic groups and acidic OH groups) but do not include groups that dissociate at high pH. Boom-Clay-HA(H+) has a lower exchange capacity than the Aldrich and Gorleben humic acids as expected from their elemental composition as well as spectroscopic characteristics (KIM et al., 1991).

3.2.3.2. Potentiometric titrations in the TRANCOM-Clay project (1995-2000)

Within the EC project TRANCOM-Clay, a new titration methodology was developed to allow determining the functional group content of organic matter in Boom Clay extracts and Boom Clay pore water without a preliminary purification step (TRANCOM-CLAY, 2000). This new titration methodology was used to determine the functional group capacity under in situ pH conditions (pH 8.2 - 8.3).

Different types of DOM were investigated: • Real Boom Clay Water 'RBCW', sampled from a piezometer installed in the 'Extension Gallery, Bottom Shaft' or EG/BS). This water extract was considered as a good representation of the aqueous phase in equilibrium with Boom Clay solid phase. • Boom Clay extract, obtained by mixing Boom Clay solid phase18 with Synthetic Boom Clay water 'SBCW' (also termed "leachings"). This Boom Clay extract contains "easily available humic substances" (mainly humic and fulvic acids). The reason for investigating this extract is because its composition is relevant for batch experiments with radionuclides (e.g., Eu, Am) from which complexation data are calculated. • The humic acid fraction of TROM (TROM is concentrated from EG/BS water by means of solid phase extraction on DEAE cellulose). Humic acid fractions were isolated from TROM samples by means of acid precipitation, were purified and transformed to the proton form. This fraction is called Boom Clay pore water humic acid (BCPHA).

18 The Boom Clay sample used originated from the ANDRA gallery in the HADES underground research facility (URF), and was taken between ring 4 and 5. The coring was vertically downwards. The sample was taken at 5.0-5.2 m (core 1,4) on 28/5/1996. The sample was packed in PE/Al/PET foil (UCB) in order to avoid oxidation, and was further kept in a stainless steel container at room temperature in the glove box. The water content of the clay (determined at 60°C) is 19.3 wt.%.

Apart from these solutions, Dierckx et al. (1997) describe the results of potentiometric titrations on the humic acid fractions of five different Boom Clay extracts, prepared by various leaching procedures. In table 3.4 a brief overview of the procedures and obtained extraction yields is presented. The maximum extraction yield is 37.9% and is likely to correspond to the maximum water extractable organic matter in Boom Clay (DIERCKX et al., 1997). All extracts were then subjected to a purification procedure in which the humic acid fraction was precipitated by addition of HCl and dialysed to salt-free conditions.

Table 3.4: Comparison of the organic matter extraction yields for different leaching conditions (DIERCKX et al., 1997)

Leach test Extraction step Extraction yield Code -2 1 Extraction with 1.35×10 M NaHCO3 4.7% BC(1) 2a Prior freeze drying 9.2% BC(2) 2b Alkaline extraction of previous residue 2.2% BC(3) 2a+2b 11.4% 3a Prior acidification (pH = 3) and 17.3% BC(4) redispersion at pH 8.6 3b Alkaline extraction of previous residue 20.6% BC(5) 3a+3b 37.9%

The purpose of investigating these different DOM solutions was to construct a suitable database representative for Boom Clay humic material in both the surface and solution phase.

Methods used to determine functional group content 1. Functional group content from acid-base titration Potentiometric titration of DOM is a well-known method to determine the acid functional group content. The charge (functional group capacity) of the organic matter due to deprotonation of ionic groups can be calculated from the charge balance:

[]= ()− + [ + ]− [ − ] []  functional group capacity CB CA H OH HA equation (3.1) where CB denotes the concentration of added base (NaOH; NaHCO3) in M; CA the + concentration of added acid (HClO4) in M; [H ] the free concentration of proton calculated from pH measurement in M; [OH -] the free concentration of hydroxyl calculated from pH measurement in M; [HA] the amount of organic matter in solution (in g/l) measured by UV. Acid-base titrations are performed at a predefined and fixed ionic strength (e.g., at 0.1 M NaClO4).

2. Functional group content from index cation adsorption This method essentially consists in adding an excess of index cation (in this case, cobaltihexamine) to induce the precipitation of the DOM and allows to determine the cobaltihexamine cation exchange capacity (CEC) by difference (MAES et al., 1992). The cobaltihexamine CEC was proven to correspond to the maximum charge measured by acid- base titration. The method was used for the case of the humic acid fraction of easily extractable organic matter from Boom Clay (BCEHA). The method does not apply for fulvic

acids and also fails in samples such as EG/BS waters, which contain a mixture of low molecular weight acids and humic and fulvic acids.

3. New titration procedure for in situ samples Because of the inorganic carbon content (~ 0.014 M NaHCO3) in EG/BS water or Boom Clay extracts, the organic matter content will only contribute for a few percent to the acid neutralising capacity. To allow for the measurement of the organic matter acid-base properties, the in-house developed procedure eliminates first the carbonates by slowly titrating with acid (0.5 M HCl) to pH 2 to 3 before the titration is made from low to high pH. The acid solution is left to stand overnight. Next morning the solution is stirred during 3 hours and is subsequently slowly titrated with 0.5 NaOH to pH values between 10 to 11 in steps of about 0.5 pH units with 10 minutes waiting in between two additions (Dosimat 665, Metrohm). All manipulations are done in a 95% N2 / 5% H2 glove box.

Results 1. Functional group content of EG/BS water from titration Three different EG/BS water samples, whose chemical composition is given in TRANCOM- Clay (2000), were titrated according to the procedure described before. The charge in meq/g versus pH is shown in figure 3.5. The first end point is observed at a pH value of about 7-8.

Figure 3.5: Charge versus pH curve for three EG/BS waters (TRANCOM-CLAY, 2000)

The value of the functional group capacity at the first end point (inflection point) is calculated relative to the minimum around pH 3. Corrections (up to about 1 eq/kg) for the uptake of OH- relative to the minimum around pH 3 were made on the basis of the analytical contents and the theoretical OH- uptake in the considered pH domain according to the acid dissociation constants. Rather high values (about 5 to 6 eq/kg) are obtained for the functional group capacity. They reflect the presence of fulvic acids, known to have a larger functional group capacity than humic acids.

The titration curves show a strong increase beyond pH 8 which assigned to the fact that Fe2+ is present as a free ion and as partly associated with DOM. The interpretation of the titration curves beyond pH 8 is therefore difficult in this region. Consequently these titration curves are not modelled in detail.

It is also mentioned that the cobaltihexamine method (MAES et al., 1992) was tried out for these EG/BS waters. However, only partial flocculation was observed when EG/BS water was contacted with cobaltihexamine. The reason for this behaviour is that apart from humic acids, EG/BS waters also contain fulvic acids and eventually other small molecules. The latter two do not entirely flocculate with cobaltihexamine and therefore this methodology failed.

2. Functional group content of Boom Clay extract The acid-base behaviour of Boom Clay extract was investigated by means of potentiometric titrations and compared to EG/BS water and the (purified) humic acid fractions. Figure 3.6 shows the titration curve. The charge at the first inflection point and relative to the value at pH 3 is 2.37 eq/kg. After correction for the co-titration of cations and anions in the extract a value of 1.8 eq/kg is obtained.

Figure 3.6: Charge versus pH curve for a Boom Clay extract (TRANCOM-CLAY, 2000)

3. Functional group content of humic acid fractions of Boom Clay organic matter

a) humic acid fraction of TROM

TROM organic matter is concentrated from EG/BS water. A continuous titration was performed of a 12.5 g/l humic acid suspension in a 0.25 mol/l NaClO4 background electrolyte under N2 atmosphere. The results are shown in figure 3.7. The end point between pH 7 and 8 corresponds to about 2.85 eq/kg.

Figure 3.7: Titration of Boom Clay TROM Humic acid. The solid line represents a FITEQL fitting (TRANCOM-CLAY, 2000)

b) humic acid fractions of "extractable humic substances"

The humic acid fractions of different humic extracts from Boom Clay solid phase were also subjected to acid-base titrations (DIERCKX et al., 1997). An analysis of the iron content shows that freeze drying and acidification result in an enhanced iron content of the extracts. This Fe pool was assumed to be present in colloidal form. On the other hand, the Fe content of the alkaline extraction steps are low and of the same magnitude as in the NaHCO3 extract and in Boom Clay pore water (EG/BS). This indicates that the humic material in the extracts and in interstitial waters have similar complexing properties for iron (DIERCKX et al., 1997).

Figure 3.8: The charge evolution (meq/g) of Boom Clay humic acids as a function of pH. For the different codes BC(i) see table (DIERCKX et al., 1997)

Figure 3.8 compares the charge versus pH curves obtained at 0.1 M ionic strength for the five different extracts. Except for extract BC(3), the dissociation behaviour of the different humic acid extracts is very similar. At the inflection point around pH 8, the charge of the humic acids varies from 2 to 2.2 eq/kg. This value is close to the value of 1.8 eq/kg obtained for Boom Clay extract and indicates that the extract (almost) solely contains humic acids.

+ Figure 3.9: Co(NH3)6 CEC versus pH on the humic acid fraction of Boom Clay extract (taken with 0.015 mol/l NaHCO3 and treated with NaF in the purification step). CEC values on the sample not treated with NaF are also indicated (TRANCOM-CLAY, 2000)

The cobaltihexammine capacity (MAES et al., 1992) of the humic acid fraction of a NaHCO3 extract (BC(1)) was also determined as a function of pH and is shown in figure 3.9. The end point around pH 8 shows a functional group capacity of about 2.1 eq/kg, in agreement with the acid-base titrations.

Conclusions Different organic matter samples were characterised by titration methods. Essentially two types of organic matter samples were studied: a) samples used as such (EG/BS water and a Boom Clay extract) without further purification treatment, and b) samples containing only the humic acid fraction (from different Boom Clay extracts and from TROM, a concentrate of EG/BS). The functional group contents corresponding to the inflection point around pH 7 or 8 were taken as representative for a comparison of the different organic matters related to Boom Clay. This inflection point is assumed in literature to correspond to the end point of the titration of the carboxylic functional groups. The comparison of the different results is given in table 3.5.

Table 3.5: Comparison of the functional group capacity at the inflection point in the pH range 7-8 for different organic matter samples

Capacity (eq/kg) Figure Notes EG/BS 5.95 Natural sample – titration EG/BS 6.19 Natural sample – titration EG/BS 5.10 Natural sample – titration BCE 1.80 Single extract with SBCW – titration BCP-HA 2.85 TROM humic acid – titration BCE-HA 2.10 Successive extractions humic acid fractions – 3+ Co(NH3)6 method BCE-HA 1.80 – 2.30 Different extracted humic acid fractions – titration

All Boom Clay extracts investigated show a constant functional group capacity of about 1.8 to 2.3 eq/kg. These humic substances are mainly present whenever batch experiments are performed with Boom Clay. The capacity of easily extractable Boom Clay organic matter is

almost identical to the capacity of the humic acid fraction of this extract, indicating that easily extractable Boom Clay organic matter is essentially composed of humic acid. It is therefore concluded that at least the humic acid fraction from Boom Clay is fairly homogeneous. Compared to published values for soil organic matter, the functional group content of Boom Clay humic acids is rather low (DIERCKX et al., 1997). This difference can be ascribed to the different nature (soil versus marine biomass) or to diagenetic processes.

It is also clear that EG/BS water contains a considerable amount of highly charged fulvic acids in view of the large difference in capacity between the original EG/BS sample and the humic acid fraction of EG/BS (BCP-HA). The capacity of the humic acid fraction of EG/BS water (2.85 eq/kg) also exceeds the capacity of the Boom Clay extracted humic acids (2 eq/kg). The pore water DOM is more relevant for laboratory column and in situ experiments.

3.2.4 Modelling of potentiometric titrations

As far as the thermodynamic behaviour of organic matter is concerned, four major types of modelling approaches can be distinguished (TRANCOM-CLAY, 2000): • the free ligand approach, • the charge neutralisation approach, • the continuous site distribution approach, • the surface complexation approach. These approaches are based on thermodynamic complexation reactions at specific functional groups of the organic molecule.

The free ligand approach considers functional groups as if they were simple ligands19, freely distributed in space. Electrostatic effects due to the impact surface potentials are neglected, which makes the free ligand approach applicable for small molecular weight fractions, and for 1:1 stoechiometric reactions only.

The charge neutralisation approach has been brought forward by Kim and co-workers (CZERWINSKI and KIM, 1996; CZERWINSKI et al., 1994; KIM and CZERWINSKI, 1996; KIM et al., 1997). It is based on the assumption of a metal Mz+ neutralising z functional negative charges of the molecule. Furthermore, the approach uses an amended mass-action law in order to anticipate the interdependency of the functional groups. And finally, the model introduces the notion of a loading-capacity, in order to account for competing cations, such as protons.

None of the previous approaches includes some mechanism for explaining the smeared appearance of acid-base titration of organic matter. The continuous site distribution models introduce a wide range of site-affinities for notably deprotonation reactions, leading to the observed 'smeared' behaviour. Other types of models are derived from this approach, such as Tipping's model n.

Another approach, sometimes used for organic molecules, is the surface complexation model. Normally, this approach should be reserved for inorganic surfaces with a high surface site density. Indeed, the electrostatic correction factor which generally forms part of the surface complexation model, is based on the assumption of a homogeneously distributed surface charge. Organic molecules, which are 2 to 3 dimensional hollow structures with an irregular

19 - 2- Simple ligands are ligands with fully characterised nature, structure and properties (e.g. Cl , CO3 , amino acids, …).

functional group distribution, do not respect these assumptions. Nevertheless, the electrostatic correction also leads to a smeared appearance of the titration curves.

Three of the four possible approaches have been applied to model acid-base titration curves in the TRANCOM-Clay project: the free ligand approach, surface complexation with electrostatic correction and a generalised form of the charge neutralisation model (TRANCOM- CLAY, 2000). The approach for the modelling of titration data for both extracted Boom clay humic acid, as well as pore water-derived humic acid, was based on a data fitting procedure in which three different functional groups are distinguished. The following deprotonation reactions were considered:

- + ≡HA(1)-OH ⇔ ≡HA(1)-O + H (K1) - + ≡HA(2)-OH ⇔ ≡HA(2)-O + H (K2) - + ≡HA(3)-OH ⇔ ≡HA(3)-O + H (K3)

Both the different constants corresponding to reactions with groups ≡HA(1)-OH, ≡HA(2)-OH and ≡HA(3)-OH, as well as the functional group concentrations, were fitted. Both the surface complexation approach (with electrostatic correction) and the free ligand approach required the three types of functional groups in order to obtain a reasonable fit of the smeared titration curves. The results are listed in table 3.6 en visualised in figure 3.10.

The data could be reasonably well fitted with both approaches, and the following conclusions were drawn from the parameter tables: • three types of functional groups are needed to describe the titration behaviour over the pH range studied (3.5-10); • the extracted humic acid fraction has a two times lower total functional group concentration compared to the pore water fraction. This is consistent with experimental results showing that humic acids from the pore water are smaller, hence exhibiting more accessible functional groups; • the thermodynamic behaviour of the extracted humic acid fraction resembles the behaviour of the pore water humic acid fraction (in terms of log(K)'s).

Table 3.6: Parameter configuration after data fitting of Boom Clay extracted and pore water humic acid (HA) acid-base titration. K1, K2 and K3 are the deprotonation constants, all determined within ±0.05 log unit. Q1, Q2 and Q3 are the proton exchange capacities corresponding to group 1, 2 and 3 respectively, expressed in meq per gram HA. QT is the total proton exchange capacity. The FITEQL fit was obtained with the fitting programme FITEQL, based on a free ligand approach without correction for ionic strength (TRANCOM-CLAY, 2000)

Boom Clay extracted HA Boom Clay pore water HA Surface Compl. Free ligand Surface compl. Free ligand FITEQL fit

Log(K1) -3.3 -4.1 -3.3 -4.45 -4.45 Log(K2) -5.1 -6.2 -6.4 -6.2 -6.23 Log(K3) -7.7 -9.0 -7.8 -9.8 -9.09 Q1 1.1 1.1 2.8 1.9 2.0 Q2 0.5 0.5 0.5 0.95 0.75 Q3 0.4 0.4 0.7 1.15 0.7 QT 2.0 2.0 4.0 4.0 3.45

Figure 3.10: Result after fitting the experimental Boom Clay extracted humic acid data for a) the surface complexation model with electrostatic correction and b) the free ligand model (TRANCOM-CLAY, 2000)

3.3 Composition and source of Boom Clay dissolved organic matter

3.3.1 Introduction

As shown in section 2, the Boom Clay contains relatively large levels of immature organic matter (up to 4 wt.%). However, measurements on piezometer waters indicate that only a small fraction (0.1 %) of this total NOM is present in the interstitial solution (VAN GEET, 2004). Due to this discrepancy, the rules governing the partitioning of Boom Clay NOM into a (hydro)soluble and non-soluble fraction have to be known in order to understand and predict DOM concentrations over the long term.

To this end, several research studies were undertaken in order to: • Better characterise Boom Clay dissolved organic matter • Obtain information on the origin and diagenetic status of Boom Clay DOM • Understand the link between the soluble and non-soluble fractions of NOM.

3.3.2 Work performed during TRANCOM-II (MORPHEUS piezometer)

In order to study and characterise the soluble and non-soluble20 organic matter, a cored borehole (HADES 2001/4) has been drilled starting from the HADES underground research facility nearly 40 m downward and a multifilter piezometer was installed (figure 3.11). The piezometer obtained the codename MORPHEUS21 and was designed to have filters at different levels of the Boom Clay exhibiting specific different characteristics (grain size, carbonate content, organic matter content, …) (VAN GEET, 2004).

Table 3.7: Characteristics of the sediment next to the different filters of the MORPHEUS piezometer (VAN GEET, 2004)

20 In the reports related to this research, the terms "mobile" and "immobile" organic matter are used, rather than "(hydro)soluble/dissolved" and "non-soluble". However, both sets of terms relate to the same problem statement, namely the link between the DOM observed in piezometer water samples and the total NOM pool 21 MORPHEUS stands for Mobile Organic matter and Pore water extraction at the Hades Experimental Underground Site.

Morpheus

N2TD N2CG

Figure 3.11: Localisation and details of the piezometers frequently sampled for use and analysis of dissolved Boom Clay organic matter in the HADES underground research facility (Mol, Belgium)

Table 3.7 gives an overview of the characteristics of the host rock surrounding each filter of the MORPHEUS piezometer. A variation of grain size, carbonate and organic matter content is observed in the sampled layers. Figure 3.12 shows the calculated hydraulic conductivities for the different installed filters. At the position of the double band (filter F8) a noticeably higher permeability is observed.

Figure 3.12: Calculated hydraulic conductivity for each filter of the MORPHEUS piezometer (VAN GEET, 2004)

The pore water composition analysed at each filter level is described in detail in De Craen et al. (2004b). Briefly, after an initial period influenced by an oxidative perturbation induced by the drilling, quite similar concentrations for the major elements are measured in all filters, except for the filter in the double band (F8). Here, a higher concentration of nearly all elements is measured.

The analysis of the solid phase organic matter was described before, in section 2.2.4. Figure 3.13 shows the TOC content of the sediment next to the filters. The two lowermost (within the Terhagen Member) and the upper (within the Putte Member) filter positions have the lowest contents of TOC (below 2 wt.%) and consequently have probably no or only limited input of kerogen type II. The Tmax values for the sediments is limited between 419 and 428 °C.

Figure 3.13: TOC (%) values measured in cores from the HADES 2001-4 borehole at the positions the twelve different filters present within the MORPHEUS piezometer plotted with regard to their depths (VAN GEET, 2002)

The sampling of dissolved organic matter from the Boom Clay is performed in three ways: • Squeezing of clay cores • Leaching of clay sediment (with SBCW, resulting in Boom Clay extract) • Direct sampling of piezometer Each technique results in different concentrations of measured DOC (table 3.8).

Table 3.8: TOC measurements of the solid and the pore water at 12 different levels within the Boom Clay at the corresponding depths of the MORPHEUS piezometer. TOC of the pore water is analysed by extracting water in 3 different ways (squeezing, leaching and in situ sampling) (VAN GEET, 2004)

Squeezing results in quite low concentrations of DOC. Increasing the pressure on the clay sample lowers the pore space and could consequently prevent the movement of (larger) organic molecules.

On the other hand, the leaching technique results in very high concentrations of DOC (MAES et al., 2003b). The amount of DOC in the first leachate (or extract) with SBCW is dependent on the solid-to-liquid ratio and seems to correspond to a simple dilution of a rather constant amount of soluble organic matter pool (figure 3.14). This feature is already known for a long time and has been observed by different authors on different Boom Clay samples (e.g., Henrion et al. (1985), TRANCOM-Clay (2000), Maes et al. (2003a)) and is an important process which should be taken into account when studying the uptake behaviour of radionuclides in Boom Clay batch suspensions. The solubilisation of organic matter is probably caused by the destruction of pores, allowing a large fraction of water-soluble organic molecules to get into solution. However, it should be noted that even for this leached organic matter no apparent correlation exists with the solid phase organic matter. Probably, not all compounds of the solid phase organic matter can be leached at the same rate (VAN GEET, 2004).

Figure 3.14: Dissolved organic matter concentration (measured as absorbance at 280 nm) found in the first extraction step with SBCW per amount of Boom Clay in solution (g·L-1). The UV absorbance is recalculated to g DOC/l following the relationship: DOC = 23.2×10-3×UV (MAES et al., 2003a; MAES et al., 2003b)

When increasing the number of leaching steps (every time replacing the supernatant solution in the Boom Clay suspensions with fresh SBCW), the amount of organic matter in solution is observed to decrease (figure 3.15) (MAES et al., 2003b; TRANCOM-CLAY, 2000). Although there is some replenishment from the solid phase, the pool of hydrosoluble organic matter appears to become exhausted.

Figure 3.15: Dissolved organic matter concentration (measured as absorbance at 280 nm) per extraction step with SBCW, identified per solid-liquid (S/L) ratio (expressed in kg·L-1). The lines serve merely an illustrative purpose and were not modelled. The apparent minimum in in the fourth extraction step therefore has to be disregarded. Rather, a plateau was observed (MAES et al., 2003a; MAES et al., 2003b)

The piezometer pore water has an intermediate concentration of DOC. It is remarked that the DOC concentration sampled from the different filters in the MORPHEUS piezometer showed strong variations (figure 3.16). Generally, a strong decrease of the DOC concentration is noted in all filters, until a more or less constant value is reached after 2 years. This feature was attributed to the initial oxidative perturbation of the clay surrounding the piezometer and will not be further discussed here (see section 4.4). All following data refer to pore water sampled after the initial perturbed state was assumed to be finished.

Figure 3.16: Evolution of the DOC in the different filters of the MORPHEUS piezometers in function of time (VAN GEET, 2004)

After about 3 years the DOC content in the different filters ranges from (roughly) 50 to 150 mgC/L. This range is in agreement with the DOC concentrations sampled from other piezometers (EG/BS, Spring 11622, N2TD, N2CG23) in the HADES underground research facility (between 60 and 129 mgC/L (BLANCHART, 2011)). Filter 8, which samples from the 'double band', shows a higher DOC concentration, at about 200-250 mgC/L. The DOC content of the pore water appears not to be correlated with the TOC content of the neighbouring sediment. Also, the measured concentrations are far lower than expected based on an extrapolation of the leached organic carbon concentrations (figure 3.14) to in situ solid- liquid ratio (1700 g/L). This indicates that a large fraction of the hydrosoluble organic matter is immobile in the confined pore space of the Boom Clay. Currently, it is believed that piezometer pore water has the most representative composition of pore water, including DOC concentration, as the in situ conditions are as much preserved as possible (DE CRAEN et al., 2004b; VAN GEET, 2004). In the following paragraphs, the results of some spectrometric analysis on the DOM sampled from MORPHEUS are presented.

The UV-Vis absorption spectra of the piezometer and squeezed organic matter were recorded. All spectra exhibit a high shoulder/peak in the early UV region (200-230 nm) and a subsequent strong decrease towards the Vis region. This indicates a large variety of functionalised organic groups on the organic matter molecules. Due to the smeared nature of the spectra and limits in signal quality a further analysis in terms of individual contributing chromophores was not possible. The UV-Vis absorption characteristics were however further analysed by the ratio of UV465/UV665 (or: E4/E6). This ratio is roughly characteristic of the nature of the DOM: a value < 5 points to humic acids while a value between 6 and ~8.5 to

22 Spring 116 stands for Source Piezonest at RING 116 of the Test Drift Gallery. It is a horizontal piezometer constructed in 1999 (presence of O2) and equipped with four large filters permitting to sample large volumes of water 23 The N2TD and N2CG piezometers were drilled in anoxic conditions (N2) in December 2004 in order to sample water that would not be disturbed by (oxidation during) the excavation process. The piezometers are located in the Test Drift (N2TD) and Connecting Gallery (N2CG), respectively

fulvic acids. Figure 3.17 shows that for all filters the ratio falls within the range for humic acid. Also the organic matter found in Boom Clay extracts (leachings) fall within this range.

Figure 3.17: The ratio of UV465/UV665 (E4/E6) absorbance plotted versus the UV280 absorbance for the different filters in the MORPHEUS piezometer (VAN GEET, 2004)

The molecular characterisation of piezometer waters was performed using gas chromatography – mass spectrometry (GC-MS) following a solid phase extraction (SPE) or liquid-liquid extraction procedure and (in some cases) a methylation step. EG/BS water was also characterised following the same procedure (VAN GEET, 2004). It was found that the EG/BS waters contain many fatty (i.e., carboxylic) acids, which are unsaturated indicating the immature nature of the organic matter. Many compounds that could not be readily identified were observed as well. The GC-MS profiles of EG/BS waters also showed contributions from plasticizers, probably due to a contamination in the pore water sampling strategy.

The GC profiles of the different MORPHEUS samples all showed the presence of a single compound, obscuring the evaluation of other peaks in the profile. This compound was identified as NBBS, a plasticizer probably originating from the non-metallic piezometer components24. Besides this peak, the GC profiles of the pore waters of all filters of the MORPHEUS piezometer are very similar indicating hardly any recognisable difference in organic compounds at the different stratigraphic levels.

24 It is noted that the presence of this plasticizer did not markedly influence DOC measurements (Van Geet, 2004)

Figure 3.18: FTIR spectra of freeze-dried material from three representative filters of the MORPHEUS piezometer (VAN GEET, 2004)

Freeze-dried material from the MORPHEUS piezometer was also analysed using Fourier Transform Infrared Spectroscopy (FTIR) and pyrolysis-gas chromatography-mass spectrometry (Py-GC-MS). Py-GC-MS analysis showed profiles that look very similar for each filter (VAN GEET, 2004). FTIR spectra of the material from the different filters of the MORPHEUS piezometer look very similar (figure 3.18). Generally, the signals of aromatic and aliphatic C-H bonds, hydroxyl groups, carboxylic C=O and aromatic C=C bonds can be clearly observed. The difference in the range of 80 to 1500 cm-1 do not differ significantly. However, three groups of filters could be discriminated (figure 3.18) based on differences in intensity and some specific bands (e.g., aromatics): • Group 1: F15, F13 and F8 • Group 2: F12, F10 and F6 • Group 3: F9, F4 and F2 Up to now, this grouping could not be linked with any other observation of those stratigraphic horizons (VAN GEET, 2004).

3.3.3 PhD of Pascale Blanchart

During her PhD, Blanchart (2011) sampled different piezometers in order to study the variability of the DOM in the interstitial water surrounding the HADES underground research facility. Four different piezometers were studied: EG/BS, Spring 116, N2TD and N2CG (figure 3.11). The DOC of the different piezometers varied from 60 (N2TD) to 129 (EG/BS) mgC/L. DOM was analysed by a multi-technique approach, including 3D-Fluorescence, FTIR, size exclusion chromatography (HPLC-SEC) and pyrolysis-GC/MS (see also section 4.4).

Figure 3.19: Emission-excitation matrices from water samples taken from EG/BS (including separate humic and fulvic acid fractions), N2TD and Spring piezometers (BLANCHART, 2011)

The humic and fulvic acid fractions of EG/BS were determined by acid (pH < 2) precipitation. It appeared that about 58 % of the total DOC belonged to the fulvic acid fraction. This is higher than earlier reported values (70 % HA, 30 % FA) (TRANCOM-CLAY, 2000). Moreover, it appears that piezometer water contain considerably more (smaller) fulvic acids compared to Boom Clay extracts (which are assumed to consist almost solely of humic acids) (see also section 3.2.3).

Blanchart (2011) also developed a quick screening technique based on 3D-Fluorescence allowing to verify whether the sampled DOM was affected by oxidation. 3D-Fluorescence is a non-destructive analysis technique in which a characteristic "map" (or emission-excitation matrix) of the fluorophores is obtained (figure 3.19). As one of the main conclusions of her PhD thesis, it was put forward that the piezometer water samples exhibited a signature that is

typical of non-perturbed DOM (BLANCHART, 2011). Analysis of the HPLC-SEC and 3D- Fluorescence signals (figure 3.19) has shown that the DOM signature of the four piezometers is very similar.

Fourier-transform infrared spectroscopy (FTIR) was also frequently applied to scope samples containing Boom Clay DOM. However, due to the nature of the Boom Clay pore water, the infrared spectra are typically influenced by several bands corresponding to inorganic salts (sulphates and carbonates). Nonetheless, it is possible to observe typical bands corresponding to aliphatic, aromatic and oxygenated organic compounds (figure 3.20).

Figure 3.20: Infrared spectra from water samples taken from piezometers EG/BS, N2TD, N2CG and Spring 116 (BLANCHART, 2011)

Figure 3.21: THM-GC-MS analysis results of EG/BS water. Relative abundances of A) linear and aromatic compounds, B) different aliphatic families and C) different aromatic families (BLANCHART, 2011)

Figure 3.22: Pyrogram of EG/BS water sample and legend of identified components (BLANCHART, 2011)

Molecular analysis of EG/BS piezometer water was performed by flash pyrolysis – gas chromatography – mass spectrometry (Py-GC-MS). This technique allows to analyse the molecules generated at 620°C (vaporisation through cracking). Before analysis, the lyophilised EG/BS piezometer water sample was methylated (THM-GC-MS) to protect carboxylic and hydroxyl functional groups. The pyrogram (figure 3.22) shows a high abundance of linear components from the carboxylic acid and n-alkane families as well as molecules derived from lignine (p-hydroxyphenyl and guaiacyl) (figure 3.22). The aromatic compounds are monocyclic and their functional groups are mostly oxygenated and of the acidic and/or methoxy type. The observed functional groups are in agreement with the bands identified through FTIR. The lignine derivates may indicate a terrestrial source for the DOM.

3.3.4 Conclusion with respect to the source and composition of dissolved organic matter in Boom Clay

The different investigations over the course of 20 years have allowed to shed some light on the characteristics of the dissolved organic matter in the Boom Clay and the way this DOM is related to the total (or solid) organic matter pool. The most important findings may be summarised as follows: • The organic matter collected from piezometer water contains both fulvic and humic acids (in approximately equal amounts). The origin may be related to the terrestrial pool of NOM in the Boom Clay. The DOM exhibits a wide range of functional groups which are capable of complexing transition metals and radionuclides in the pore water. The total DOM concentration seems dependent on the pore size distribution (reflected in the hydraulic conductivity parameter) and on the nature of the NOM pool. The latter statements need further verification. • Suspending Boom Clay solid phase results in the release of a more hydrophobic, humic acid-type organic matter in the solution phase. Although the functional groups of this DOM appear to be similar to the ones found for DOM in piezometer water, the amount of functional groups is 2-3 times smaller.

Despite these important results, research until now has not resolved in the complete elucidation of the mechanisms and processes behind the distribution of the Boom Clay NOM pool between kerogen, bitumen and dissolved organic matter. On the other hand, such research is quite exceptional within the scientific world since respectively kerogen/bitumen and DOM research normally belong to two distinct scientific communities (respectively petroleum/organic geochemistry and soil chemistry). Bringing the concepts of these two communities together is therefore not an easy task. However, in order to evaluate the current Boom Clay conditions and to extrapolate to disposal time scales (Ma), such research efforts need to be continued.

3.4 Redox reactivity of dissolved organic matter

The redox properties of natural organic matter have already long time been acknowledged (ALBERTS et al., 1974; BRUGGEMAN et al., 2012; GOODMAN and CHESHIRE, 1982; SKOGERBOE and WILSON, 1981; SZILAGYI, 1973; VISSER, 1964). Several studies have focussed on the ability of NOM to act as electron-transfer agents in biogeochemical redox processes (figure 3.23) (BENZ et al., 1998; DUNNIVANT et al., 1992; LOVLEY et al., 1996; LOVLEY et al., 1998). The reduction potential of natural organic matter has been measured most frequently using a titration with either the Fe(III)/Fe(II), or the I2(0)/I(-I) redox couple (MATTHIESSEN, 1994; PALMER et al., 2006; STRUYK and SPOSITO, 2001). The use of the latter can however be criticised, because of the possible irreversible bonding of iodine to natural organic matter, termed "iodination" (REILLER et al., 2006; SCHLEGEL et al., 2006). The ability to accept and donate electrons can be characterised by the electron-carrying capacity (ECC), which represents the number of mole equivalents of electrons transferred from a donor to an acceptor per gram NOM (RATASUK and NANNY, 2007). Helburn and Maccarthy (1994) determined redox properties of humic acid by alkaline ferricyanide titration and reported a value of 1.7 meq/g at pH 9. Matthiessen (1994) used ferricyanide consumption to determine the redox capacity of humic materials as a function of pH, and reported a value of ~1 meq/g at pH 7, which could increase up to ~7 meq/g at pH 10. Natural NOM of various origins exhibited different pH dependencies. Struyk and Sposito (2001) titrated three standard humic acids with I2 and reported a reducing capacity that ranged from 3.3 to 11.5 meq/g at pH 7. Palmer et al. (2006) reported reduction capacities obtained from I2 titrations of 3.3 and 9.5 meq/g for two types of humic acid. Bauer et al. (2007) found for peat dissolved NOM an electron acceptor (EAC) and an electron donor capacity (EDC) of respectively 6.2 meq/g (pH 6.5) and 1.5 meq/g (pH 4.5), and that the EAC and EDC were controlled by the redox potential of the acceptor or donor. Among the other methods used to characterise organic matter redox state, Wilson and Webber (1977) used electron spin resonance to detect semiquinone radicals in humic substances. More recently, Klapper et al. (2002) and Cory and McKnight (2005) used fluorescence spectroscopy to detect fulvic and humic acid oxidation states, while Aeschbacher et al. (2010; 2009) used electrochemical methods to assess the redox properties of humic substances.

Figure 3.23: Natural organic matter acting as electron carrier from a bulk donor to a pollutant (BRUGGEMAN et al., 2012)

Figure 3.24: Reduction of benzoquinone to the intermediate semiquinone radical and further reduction to hydroquinone (BRUGGEMAN et al., 2012)

Considerable evidence has accumulated that quinones (figure 3.24) are the dominant redox- active moieties associated with NOM (DUNNIVANT et al., 1992; KLAPPER et al., 2002; LOVLEY et al., 1996; LOVLEY et al., 1998; NURMI and TRATNYEK, 2002; SCOTT et al., 1998; STRUYK and SPOSITO, 2001). Quinones are a versatile class of biomolecules found in living cells, in extracellular material, and in detrital organic material. Quinones can cycle between three redox states: oxidised, semiquinone radical, and the reduced, or hydroquinone state (Figure). Using cyclic voltammetry, NOM has been shown to undergo sequences of two one- electron, quasi-reversible, diffusion controlled, electron transfers at a Pt electrode surface (NURMI and TRATNYEK, 2002). These observations are in line with electron transfer reactions to quinones, which leads to an increase in semiquinone radical intermediates as well as hydroquinones. Cory and McKnight (2005) observed the ubiquitous presence of oxidised and reduced quinones in dissolved organic matter, and found that the fractions of reduced, respectively oxidised, quinones changed concurrently with environmental and/or laboratory redox gradients. Ratasuk and Nanny (2007) identified the redox sites in NOM through characterisation and quantification of the ECC of reversible redox sites, and found that they consisted of quinone as well as non-quinone structures. The contribution of quinone sites to the total ECC was estimated to be 44-79 %. Because the electron transfer in reduction and + oxidation processes of quinones is coupled to H -transfer, the redox potentials (Eh) of such reactions are pH-dependent.

Pirlet (2003) determined the redox capacity of Boom Clay pore water by means of a redox titration with potassium ferricyanide, K3Fe(CN)6. Buffer solutions (TRIS) with a constant concentration of 0.005 M were used to keep the pH at a constant value (8.3) over the whole reduction time and the ionic strength was fixed to 0.02 M by adding NaCl. The reduction of ferricyanide to ferrocyanide by the Boom Clay pore water was followed by spectrophotometry. The total reaction time was 24 hours. The reaction kinetics of the reduction indicated the existence of at least two different reduction mechanisms. The total redox capacity of the Boom Clay pore water, attributed essentially to the oxidation of humic functional groups, amounts to 9.3 ± 0.1 meq/g OM at pH8.3 (PIRLET, 2003). This redox capacity is high compared to soil-derived humic substances (MATTHIESSEN, 1994).

The determination of the redox capacity of Boom Clay organic matter is also part of the ongoing research plan on EUROBITUMEN. More results are expected as this research plan proceeds.

The reducing effect of Boom Clay dissolved organic matter has been shown to play in the case of U(VI) (BRUGGEMAN and MAES, 2010), Np(V) (PIRLET, 2003; PIRLET and VAN ISEGHEM, 2003) and Se(IV) (BRUGGEMAN et al., 2007). In other studies, dissolved organic matter has been shown to play a role in the reduction of both radionuclides, like Np(V) (SHCHERBINA et al., 2007b; ZEH et al., 1999) and Pu(IV,V) (ANDRE and CHOPPIN, 2000;

BLINOVA et al., 2007; MARQUARDT et al., 2004; SHCHERBINA et al., 2007a), and heavy metals, like Fe(III) (LOVLEY et al., 1996; SZILAGYI, 1973), Mn(IV) (SUNDA and KIEBER, 1994), V(V) (WILSON and WEBER, 1979), As(V) (PALMER et al., 2006), Cr(VI) (WITTBRODT and PALMER, 1997), Mo(VI) (GOODMAN and CHESHIRE, 1982) and Hg(II) (ALBERTS et al., 1974; MATTHIESSEN, 1998; SKOGERBOE and WILSON, 1981).

3.5 Size distribution of dissolved organic matter in Boom Clay

3.5.1 Introduction

From the previous section it appeared that the amount of hydrosoluble organic material that is potentially mobile in Boom Clay pore water may be strongly affected by the lay-out of the pore space of the confined Boom Clay. The size restriction imposed by the physical structure of the pore space and the surrounding fabric (possibly together with charge repulsion phenomena) may indeed hinder the free movement of larger organic molecules and may be responsible for the difference in DOC concentrations observed in leaching and squeezing tests, and in piezometer samples.

Indeed, while dissolved organic matter (DOM) is commonly defined as the hydrosoluble organic matter fraction that passes through a 0.45 µm filter, the Boom Clay may impose an effective cut-off which is smaller than this size. Since humic acids are generally larger than fulvic acids, this reduced effective cut-off might also explain the differences found between humic/fulvic ratio in piezometer water and in Boom Clay extracts.

3.5.2 Characterisation of Boom Clay dissolved organic matter size distribution

Several physical fractionation methods (gel permeation chromatography (GPC), ultrafiltration, ultracentrifugation, etc.) can be used to determine the size distribution of the different DOM fractions. Since humic acid is a polyelectrolyte, its aquatic molecular size varies with pH and ionic strength. Whatever methods known hitherto to determine the size distribution of humic acid, no procedure can provide information on true sizes because of the complex structural conformations of humic acid. They are sensitively changing with surrounding conditions, such as proton activity, ionic strength, concentration of complexing ions, etc. Every method involves molecular coagulation of humic acid, so that experimental results are at best phenomenological (KIM et al., 1991).

3.5.2.1 Gel permeation chromatography

In gel permeation chromatography, molecules larger than the pores of a given gel will exit the column with an elution volume equal to the dead volume of the column. Smaller molecules which are able to diffuse unhindered into the pores of the gel, will be eluted with an elution volume equal to the total volume of the column. Molecules of intermediate size will be eluted with an elution volume logarithmically proportional to their molecular size. The size fractionation of a column can be calibrated with standard molecules of different sizes (e.g., globular proteins of known molecular weight) (figures 3.25). The concentration of organics is measured by UV absorption (e.g., at 280 nm). In figure 3.26 a linear correlation is shown between the elution volume and the logarithm of molecular weights on the left ordinate as well as the logarithm of hydrodynamic diameter on the right ordinate. From this calibration curve both equivalent molecular weight (Dalton) and hydrodynamic diameter of unknown substances can be determined. However, the size fractionation of humic acids by GPC involves also some difficulties which are attributed to the complex nature of these molecules.

These difficulties involve the influence of both hydrodynamic size and shape on the diffusion through the gel, the repulsion between gel and molecules, the sorption interaction between gel and molecules, and the agglomeration of molecules.

Figure 3.25: GPC elution profiles of calibration standard molecules with Sephadex 100-120, -3 I = 0.1 (NaClO4) and pH = 9.2 (10 M borate buffer): Dextrane blue, Albumin, Carbonic anhydrase, Cytochrome C, Aprotinin and Benzylic alcohol (starting from the left) (KIM et al., 1991)

Figure 3.26: Molecular size calibration in Dalton (left) and nm (right) as a function of the GPC-elution volume (KIM et al., 1991)

Figure 3.27: GPC analysis of ultrafiltered (Amicon filters) piezometer waters from filters F8 and F12 of the MORPHEUS piezometer (VAN GEET, 2004)

Boom Clay piezometer water 1.0 Boom Clay extract (leachate)

0.8

0.6

0.4

0.2 normalised absorption UV

0.0

0306090 Elution volume (mL)

Figure 3.28: Gel permeation chromatograms of Boom Clay extract and Boom Clay piezometer water (KULeuven, unpublished data).

Van Geet (2004; 2003) describes GPC measurements on piezometer water samples from MORPHEUS and on ultrafiltered samples (Amicon filters) thereof. Tests were carried out on water from two filters (F8 and F12, see section 3.3.2). The results show rather a decrease in absorption (total peak surface) than a shift in peak position (figure 3.27). Only a first peak corresponding to a size fraction > 100 kDa could be distinguished. Probably, the used set-up was not very efficient in discriminating between different size fractions < 100 kDa.

In another test performed at K.U.Leuven (laboratory of prof. A. Maes, unpublished data), gel permeation chromatograms taken from piezometer water on the one hand, and a Boom Clay extract (with SBCW) on the other hand, show a clearly observable difference in size distribution (figure 3.28). The Boom Clay extract apparently contains a lot more humic material of higher molecular weight compared to piezometer water samples. This feature might explain the difference in DOC concentration measured in extracts versus piezometer samples (see section 3.3).

Blanchart (2011) compared the distributions of the DOM from four piezometers (EG/BS, Spring 116, N2TD, N2CG) and of their humic and fulvic fractions by means of HPLC-SEC (PLaquagel-OH Agilent, 300×7.5 mm id, 8 µm pore size) equipped with a diode array (DAD) and fluorescence (FLD) detector (eluent composition 0.01 M Na2HPO4 + 0.01 M KH2PO4 + 0.1 M Nacl, pH 6.8). The distributions of the DOM, humic and fulvic acids from the four piezometers were very similar (figure 3.29). The fulvic acid fractions eluted at (slightly) higher times compared to the humic acid and total DOM samples, indicating the presence of molecules of lower molecular weight.

Figure 3.29: Normalised DAD signal at 254 nm as a function of time for the samples EG/BS, N2CG, N2TD and Spring 116 (total DOM, fulvic and humic fractions) (BLANCHART, 2011)

3.5.2.2 Ultrafiltration

Kim et al. (1991) describe results from ultrafiltration performed by means of a 10 mL stirring cell connected to a flow-through cell of a UV spectrometer to quantify the humic acid concentration in the filtrate. Filters of various pore sizes (Amicon YM and XM membranes from 500 MWCO up to 300000 MWCO) were used. The filtration results of Boom Clay humic and fulvic acid are shown in table 3.9 and figure 3.30. Boom Clay humic acids display a wider size range compared to fulvic acids. Most of the fulvic acids are constrained within the range < 30 kDa (~2.1 nm). On the other hand, a significant portion of humic acids is situated in the size range 10 kDa – 50 kDA, while another fraction is higher than 100 kDa.

Table 3.9: Molecular size distribution determined by ultrafiltration at pH 8.5 and I = 0.1 (NaClO4): permeation through the filters (in percent) (KIM et al., 1991)

Filter Poresize Poresize Boom-Clay- Boom-Clay- (Da) (nm) HA(H+) FA(Na+) Amicon XM300 300000 ~15 87.5±1.2 96.8±0.7 Amicon YM100 100000 ~5 71.7±0.9 95.7±0.7 Amicon XM50 50000 ~3 57.0±2.4 93.4±1.9 Amicon YM30 30000 ~2.1 29.9±2.6 87.3±3.1 Amicon YM10 10000 ~1.5 16.5±0.3 66.3±1.8 Amicon YM5 5000 ~1.3 9.1±0.7 44.4±3.7 Amicon YM2 1000 ~1 1.6±0.1 11.5±3.2 Amicon YC05 500 <1 - -

Figure 3.30: Molecular size fractionation of humic acids by ultrafiltration with various pore sizes; 0.1 M NaClO4 at pH 8.5, HA concentration: 30 mg/L (KIM et al., 1991)

At SCK•CEN, microfiltration and ultrafiltration were used to distinguish between different sizes of the dissolved organic matter. Microfiltration is typically carried out using 0.22 µm or

0.45 µm PVDF membranes (VWR). For ultrafiltration either Millipore Amicon Diaflo regenerated cellulose or Pall Microsep Omega polyether sulphone (PES) membranes have been used, with different cut-offs. In table 3.10 the relationship between membrane cut-off value and nominal size for some selected ultrafilters, based on information gained directly from the producers, is given.

Table 3.10: relationship between cut-off value in Da and Stokes radius (Millipore) or nominal pore size (Pall) for Millipore Amicon Diaflo (regenerated cellulose) and Pall Microsep * Omega (polyether sulfone – PES) ultrafilters (TRANCOM-II, 2004)

Cut-off Millipore Amicon Pall Microsep Omega Linear Globular 30 kDa 2.5 nm 4.5 nm 50 kDa 5 nm 100 kDa 3.5 nm 7 nm 10 nm 300 kDa 6 nm 12 nm 35 nm * info obtained directly from Millipore and Pall Life Sciences

Figure 3.31: Molecular weight distribution of the pore water organic matter found in the different filters in 9 stratigraphical levels in the MORPHEUS piezometer, sampled 352 days after installation, by means of Amicon ultrafilters (TRANCOM-II, 2004)

Van Geet (2004; 2003) measured the molecular size distribution of piezometer water found in 9 different filters from MORPHEUS using Amicon filters (consisting of regenerated cellulose) at 1, 10, 50 and 100 kDa cut-off (figure 3.31). The results illustrate a bimodal size distribution, with about 45 % of the DOM smaller than 1 kDa and 45 % larger than 50 kDa. A similar result was obtained using centrifugal ultrafiltration units with PES25 membranes (Pall Life Sciences, Omega membranes).

In the framework of the EC project FUNMIG, Bruggeman et al. (2010b) studied the size distribution of EG/BS piezometer water (RBCW) and Boom Clay extracts made with both RBCW (RBCW+BC) and SBCW (SBCW+BC) by means of centrifugal ultrafiltration units

25 PES stands for Poly Ether Sulfones

with PES membranes (Pall Life Sciences, Omega membranes) From figure 3.32 it can be observed that the natural organic matter pool in the interstitial piezometer water (RBCW) has a bimodal distribution consisting of very small (< 10 kDa) dissolved organic matter molecules and larger (30-300 kDa) colloids. This bimodal distribution is also apparent in the RBCW+BC and even the SBCW+BC organic matter pools, but the most striking feature for both these NOM solutions is the additional presence of very large (> 300 kDa) organic matter colloids. Therefore it is stated that by suspending Boom Clay solid phase in SBCW or RBCW, large colloids are liberated or solubilised that were originally trapped within the confined pore structure of the clay. These larger colloids are assumed to be immobile under in situ conditions.

40% RBCW 35% RBCW+BC 30% SBCW +BC 25%

20%

15%

10% 5%

m -5% Da Da Da Da m 1kDa < 0)k 0)k )k 5)um 45u 0)k .4 -0 .22 u >0. (1-10)kDa (10-3 (30-5 0-100 a (5 0-100 (100-300)kDa(30 (0.22-0 1000kD

Figure 3.32: Size distribution of the three DOM-containing solutions (RBCW, RBCW+BC, SBCW+BC) used in the experiments, as determined by ultrafiltration and microfiltration at various cut-offs (BRUGGEMAN et al., 2010b)

In an earlier study26, Righetto et al. (1991) investigated the size fractionation of Boom Clay -2 DOM colloids (extracted (1.4×10 mol/L NaHCO3; pH 8.6) Boom Clay humic acids) by means of ultracentrifugation (55000 rpm, 1h, estimated colloid cut-off 10-15 nm). The results indicate a pH-dependent polydispersity of HA (figure 3.33). Additional ultrafiltration measurements indicate that molecular units smaller than 1.5 nm (~ 15 %) are present at all pHs, while aggregation occurs with decreasing pH. The electrostatic repulsion between the negatively-charged functional groups hinders the aggregation of the HA particles at high pH. However, with decreasing pH and degree of ionisation, inter- and intramolecular hydrogen bonding among protonated sites becomes possible, and clusters of HA molecules begin forming. Accordingly, figure 3.33 shows that an enhancement of HA aggregation occurs with increasing HA concentration.

26 Study performed in the context of the EC-coordinated project MIRAGE, working group on Colloids and Complexes (COCO)

Figure 3.33: pH dependence of "soluble HA", i.e., not removed by ultracentrifugation, as obtained by absorbance measurements (at 240 nm) (RIGHETTO et al., 1991)

3.5.2.3 Other techniques

Various other techniques are available to determine the size distribution of humic colloids and have been tested on Boom Clay DOM.

Thang et al. (2001) optimized and applied the flow field flow fractionation (FFFF) for studying the size distribution of (among others) humic colloids from the EG/BS piezometer. The system was calibrated by using sulfonated polystyrene (PSS) reference colloids. The carrier solution consisted of a 5×10-3 mol/L Tris-buffer at pH 9.1. Boom Clay humic colloids were observed to have a peak molecular weight, Mp at 1.0 (±0.02) kDa, a number averaged molecular weight, Mn at 1.1 (±0.03) kDa, and a weight-averaged molecular weight, Mw at 1.8 (±0.04) kDa. This was found to correspond to a hydrodynamic diameter of 1.8 nm (THANG et al., 2001).

M.Bouby (2007) investigated different Boom Clay DOM samples (EG/BS, TROM36, Boom Clay extracts with SBCW and EG/BS) using asymmetric flow field flow fractionation (AsFFFF) coupled to a UV-Vis spectrophotometer. The fractograms of the different DOM samples showed bi- or tri-modal distributions (figure 3.34), and were qualitatively in agreement with the results from ultrafiltration (see previous section). About half of the DOM molecules in TROM36 and EG/BS water was located in the smallest size range, i.e. from 0.5 – 7 kDa. The remainder was located in the size-range from 7 – 100 kDa. The Boom Clay extract with EG/BS showed the same qualitative distribution as (pure) EG/BS water, but the relative proportion of the higher size range was larger. The Boom Clay extract with SBCW displayed the lowest relative proportion of small-sized molecules.

The microstructure and pore space distribution of Boom Clay is subject to recent investigations using cryo-SEM (scanning electron microscopy at cryogenic temperature), with ion beam cross-sectioning to prepare smooth, damage free surfaces (DESBOIS et al., 2010; HEMES et al., 2011; HEMES et al., 2012). The resolution of the SEM is about 10 nm.

Three main types of pore morphology could be distinguished: (1) Type I – elongated pores between similarly oriented clay sheets (< 100 nm), (2) Type II – crescent-shaped pores in saddle reefs of folded sheet of clay (100-1000 nm) and (3) Type III – large jagged pores surrounding clast grains (> 1 µm). About 87 % of the pores have an equivalent radius less than 100 nm. Porosity fabrics seem to be highly anisotropic with pores preferentially orientated parallel to the bedding. Moreover, interconnectivity of pores is not obvious and corroborates complex connectivity. These latter two factors might be very important for colloidal transport because the mobility of dissolved organic matter is not determined by the pore size, but by the pore throat size (Reszat and Hendry, 2009).

Organic matter from Real Boom Clay Water 0 100 200 300 400 500 600 700 800 900 1000 0.04 [C]= 100 µg.L-1 each

0.03 15.2 kD 15.2 81.8 kD 81.8 1.29 kD void peak void 0.02 4.4 kD

29 kD

0.01 UV. 225 nm / a.u. nm 225 UV.

0.00 0 100 200 300 400 500 600 700 800 900 1000 0.025 0.06

DOC / mg.L-1 0.020 0.049 0.098 0.19 0.04 0.015 0.49

0.010 0.98 0.02 UV. 225 nm /UV. nm 225 a.u. 0.005

0.000 0.00 0 100 200 300 400 500 600 700 800 900 1000 Elution time / s Eluent: MQ + NaOH pH 8.5 100 µL injected

Figure 3.34: Fractogram of Boom Clay pore water (EG/BS) by means of (asymmetric) flow field flow fractionation. Left figure from Thang et al. (2001). Right figure from Bouby (2007)

The size distribution of piezometer DOM was also investigated by Blanchart (2011) by means of dialysis at 3500 Da (Spectra/Por Dialysis, dialysate was milliQ H2O and was changed every 24 h, total dialysis time was 5 days) and of Atmospheric Pressure Photo-Ionisation Quadrupole-time of flight spectrometry (APPI-Qtof, Bruker Daltonics). APPI-Qtof allows to separate ionized ions present in the nebulized sample to be separated according to their m/z ratio.

Dialysis of EG/BS piezometer showed that about 85 % of the DOC was able to pass through a dialysis membrane of 3500 Da. The fulvic acid fraction was found to consist almost completely (98 %) of molecules smaller than 3500 Da while the humic acid fraction contained about 24 % molecules larger than 3500 Da (BLANCHART, 2011).

On the other hand, APPI-Qtof of samples taken from the N2CG and N2TD piezometers showed that the DOM of both waters had a similar unimodal distribution, comprised within a m/z range from 100 to 600, with a maximum at 160 m/z (figure 3.35).

Figure 3.35: APPI-Qtof mass spectra of water samples taken from the N2CG and N2TD piezometers (BLANCHART, 2011)

3.5.3 Conclusion: mobile organic matter concentration in Boom Clay

As a conclusion, we state that the cut-off equivalent to ultrafiltration at 3×105 Da (Pall Microsep Omega PES membranes), or about 35 nm (according to the specifications of the producer27), can be used as a realistic value for the maximum size of mobile dissolved organic matter molecules in the Boom Clay (undisturbed conditions, Mol-Dessel region). Dissolved organic matter molecules retrieved from piezometer water are indeed smaller than this cut-off (as observed in the case of the EG/BS and MORPHEUS piezometers). On the other hand, organic matter molecules that are leached from the Boom Clay (Boom Clay extract) were found to be larger than this cut-off. This feature may explain the discrepancy found between the DOC concentration in piezometer water and the high DOC concentrations found when leaching Boom Clay at high solid-to-liquid ratio. Furthermore, we consider that all mobile dissolved organic matter display sufficiently similar characteristics and can thus be regarded as one homogeneous pool (further evidence hereof is given in the section on organic matter transport, section 3.6).

However, there is still some large uncertainty related to this value. On the one hand, the size distribution analysis technique used may impose some errors on the measured value. Indeed,

27 http://www.pall.com/pdfs/Laboratory/08.2414_CentrifugalDevice_SS.pdf

differences in the experimental molecular weight distribution exist in the literature depending on the applied method and the fractionation conditions. This fact is partly ascribed to the real change of hydrodynamic size due to ionic strength and pH, but also to possible fractionation artifacts (THANG et al., 2001). Therefore it is recommended that the results from ultrafiltration are to be reconfirmed by other analysis techniques. On the other hand, it remains uncertain if the results gathered through such size analysis techniques are applicable and fully transferable to the transport of dissolved organic matter through the pores. This uncertainty is related to the (possible) aggregate structure of humic material. For example, Randall et al. (1996) used radiolabelled humic material to investigate the size distribution of humics. When a fraction containing smaller sized molecules was separated from a sample of humic material, the smaller molecules aggregated to form larger molecules in an attempt to re-establish the original distribution of sizes. This may indicate that a larger humic aggregate is capable of breaking down upon trying to pass a narrow pore throat. Dedicated migration experiments with different size-fractionated DOM pools may offer more insight into the existence and possible influence of this mechanism.

Because of the assumptions made above, the mobile dissolved organic matter concentration range in Boom Clay can be gathered from DOC measurements on piezometer water from the Boom Clay. The MORPHEUS piezometer samples pore water from different levels in the Boom Clay and was specifically intended to scope the variation in DOC (both in concentration and characteristics) over the formation. The evolution of DOC concentration versus time in the different filters from the MORPHEUS piezometer is displayed in figure 3.16.

Therefore, the dissolved organic matter concentration in Boom Clay pore water (undisturbed) is situated in the range between (roughly) 50 – 150 mgC/L with the exception of the concentration in the so-called double band (Filter 8) which is higher and amounts to 200 mgC/L. The DOC concentrations in the other piezometers which are used as the source of RBCW (EG/BS, Spring 116), are lying within the same range. We therefore postulate a DOC expert range of 50-150 mgC/L and an average content of 100 mgC/L.

It is remarked that the concentration range postulated here refers to an experimental range sampled mainly at the level of, and below the HADES underground research facility. The upper part of the Putte Member and the Boeretang Member are largely unexplored.

3.6 Migration of dissolved Boom Clay organic matter

The transport of dissolved organic matter in the Boom Clay under undisturbed conditions in the Mol-Dessel region is mainly determined by 1) the diffusion coefficient in the pore space, as advection is negligible, and 2) the retention of organic matter on the solid surfaces surrounding the pore space, either through (pure) sorption (equilibrium approach) or through colloid filtration (kinetic approach) mechanisms.

3.6.1 Sorption of dissolved organic matter onto Boom Clay

The distribution of organic matter between the liquid and the solid phase is of significant relevance with respect to migration in Boom Clay: how does organic matter migrate through the formation and how does the 'mobile' organic matter fraction interact with the 'non-mobile'. Distribution coefficients for 14C-labelled dissolved OM contacted with Boom Clay suspensions at different solid-to-liquid ratios were determined experimentally during the TRANCOM-Clay and the TRANCOM-II projects (TRANCOM-CLAY, 2000; TRANCOM-II, 2004).

When dispersing Boom Clay at different liquid/solid ratios a constant amount of organic matter is released into solution (figure 3.36). Because of this release, 14C radiolabelled DOM has been used to study sorption (TRANCOM-CLAY, 2000) in two experimental setups: 1) batch sorption experiments at different SBCW/Boom Clay ratios, and 2) a sequential batch equilibrium sorption experiment.

Figure 3.36: Amount of organic matter in solution upon dispersal of increasing amount of Boom Clay and centrifugation (2 hours at 27000g) (TRANCOM-CLAY, 2000)

In the first type of experiments, 2 mL aliquots of a stock solution of 14C radiolabelled DOM (isolated and concentrated from piezometer (EG/BS) water during the TRANCOM project; coded TROM (TRANCOM-CLAY, 2000)) in SBCW were contacted with Boom Clay suspensions at different liquid-to-solid ratios (~3.9, ~6.7, ~17.4, ~50.8 L/kg) and for different time intervals (6, 13, 20 days). After phase separation by centrifugation (2 hours at 27000 g, cut-off ~ 25 nm), the supernatant solution was monitored for UV absorbance, 14C and pH. The experimental results are shown in figure 3.37. The decimal logarithm for the distribution coefficient of the 14C-DOM decreased with increasing liquid/solid ratio and was independent on the equilibration time.

Figure 3.37: Distribution of 14C organic matter versus absorbance in Boom Clay (TRANCOM-CLAY, 2000)

14 14 When comparing the distribution coefficient for the C labelled DOM, Rd( C), with the distribution of the organic matter between the Boom Clay (the total organic pool, assuming ~2.2 wt.% organic carbon) and the weight in solution (calculated from UV absorbance 14 measurements), Rd(DOM), it appears that Rd( C) is about 200 times smaller. This result indicates that part of the organic matter pool is not involved in the 14C distribution. Otherwise stated the 14C labelled organic matter does not distribute over the entire organic matter pool. This is also consistent with the observation that upon dispersion of Boom Clay in SBCW or RBCW, larger OM molecules are solubilised that do not feature in piezometer water.

Similar results were obtained during the TRANCOM-II project (see figure 3.38).

Figure 3.38: 14C-organic matter distribution ratio versus the Boom Clay batch solid-liquid ratio (in g/l) (TRANCOM-II, 2004)

Further evidence for the uptake of dissolved OM was gathered by studying the sorption of three DOM solutions (RBCW, RBCW+BC and SBCW+BC; see section 3.5.2.2) by illite (BRUGGEMAN et al., 2010b). The DOC was measured both after centrifugation and after ultrafiltration (here, 30 kDa (Pall Microsep Omega membrane) was used as it represents the boundary between the very small dissolved organic matter molecules, and larger organic matter colloids in RBCW). Figure 3.39 shows the calculated DOM solid-liquid distribution

DOM DOM coefficient, Kd (L/kg), derived from these measurements. When calculating Kd it is assumed that all DOM which is not in the supernatant (after centrifugation) or in the percolate (after ultrafiltration), is adsorbed by the solid phase. This is evidently not always the case: especially for the DOM retained on the filter during ultrafiltration, the mere size distribution DOM of the DOM pool (and not sorption) will dictate the value of Kd . This can be most readily observed for the SBCW+BC samples after ultrafiltration: almost no DOC is left (in the order DOM of <5 mg/l), and, thus, the calculated log Kd is very high (between 3.4 and 3.7).

RBCW centr 4 RBCW ultrafiltr RBCW+BC centr RBCW+BC ultrafiltr SBCW+BC centr SBCW+BC ultrafiltr

2 NOM d logK

0 1020304050607080 TOC after centrifugation/ultrafiltration (mg·l-1)

DOM Figure 3.39: Calculated solid-liquid distribution coefficient, log Kd , of three different DOM pools (RBCW, RBCW+BC, SBCW+BC) between illite and SBCW (BRUGGEMAN et al., 2010b)

After centrifugation, only a small part of the initial dissolved OM concentration is removed from solution, possibly by adsorption onto/interaction with suspended illite particles. With respect to this initial concentration, only ~ 5 % is removed in RBCW-containing samples, ~ 5- 10 % in RBCW+BC samples, and ~ 10-15 % in SBCW+BC samples. Thus, an increase in NOM removal is observed for the pools containing larger colloids, but the overall interaction between Boom Clay-derived NOM and illite is considered small.

3.6.2 Overview of migration experiments with Boom Clay dissolved organic matter

Different experiments have been performed over the course of more than 20 years to study transport of dissolved organic matter (DOM) in Boom Clay. This section provides an overview of these migration experiments, together with a discussion of their most prominent outcome. Table 3.10 gives an overview of all experiments.

3.7.1.1. Early (pre-TRANCOM) experiments

The earliest experiments (HENRION et al., 1985; HENRION et al., 1991; PUT et al., 1992) on natural organic matter (or dissolved organic matter) transport in Boom Clay used either non- labelled humic material from the formation itself, or small 14C-labelled compounds which served to mimic dissolved humic molecules.

Henrion et al. (1991) investigated the effect of compaction on the diffusion of non- or weakly sorbing tracers. The main focus was on small labelled organic compounds (sucrose, lactose, triiodothyronine (TIT) and phenylalanine) and HTO and I- were used as reference compounds. Through-diffusion experiments (SCK•CEN code A3) were performed on reconsolidated "clay pastes" having no preferred orientation. The diffusion parameters that were investigated were the product (ηR) of the diffusion accessible porosity (η) and the retardation factor (R), the apparent diffusion coefficient (Dapp) and the effective diffusion coefficient (Deff = ηRDapp). The diffusion parameters that were obtained from these experiments were shown to adhere to an equation similar to Archie's law. However, this outcome is nowadays questioned because 1) Archie's law is only valid on an element-by-element basis; 2) the investigated species combined both neutral, anionic, sorbing as well as non-sorbing tracers.

Instead of working with simulators for the dissolved organic matter, Put et al. (1992) used DOM from the Boom Clay itself. They performed two types of experiments: (1) percolation of an intact clay core with consecutively synthetic Boom Clay water (SBCW) and RBCW (so- called "down flooding" and "up flooding" experiments). Each percolation cycle is continued for at least 20 renewals of the pore volume of the clay sample. (2) an impulse of RBCW is injected at the end of the last cycle with SBCW, and percolation with SBCW is continued (pulse injection experiment). In both methods, the concentration of organic material in the percolated liquid at the outlet of the clay core is monitored as a function of time by UV280 absorbance measurement. The diffusion-advection equation is solved for the respective given boundary and initial conditions and fitted to the measured concentration evolution to obtain i the model parameters ηR and the apparent dispersion coefficient (D app). A total of eleven tests have been done (table 3.10 and figures 3.40 and 3.41).

Table 3.10: Overview of all migration experiments performed at SCK•CEN on organic matter transport in Boom Clay. Displayed parameters are: relevant literature reference, experiment type (see section 0.3), percolating solution and/or organic matter spike used, stratification, -1 reconsolidation pressure, darcy velocity (vd [m·s ]), product of diffusion accessible porosity and retardation factor (ηR), apparent dispersion i -1 -1 -1 coefficient (D app [m²·s ]), apparent diffusion coefficient (Dapp [m²·s ]), effective diffusion coefficient (Deff, [m²·s ]) and pore diffusion coefficient -1 (Dpore, [m²·s ])

Figure 3.40: Example of the total quantity of percolated dissolved organic material (DOM) through a clay core for percolation with real Boom Clay water (RBCW) as a function of time (test 2). The square markers on the curves are the measured percolated quantity of DOM as a function of time and the solid line is the fitted model calculation (PUT et al., 1992)

Figure 3.41: Example of the total quantity of percolated dissolved organic material (DOM) through a clay core for impulse injection with real Boom Clay water (RBCW) as a function of time (test 4). The square markers on the curves are the measured percolated quantity of DOM as a function of time and the solid line is the fitted model calculation (PUT et al., 1992)

Samples were cored perpendicular (V) or parallel (H) to the stratification of the formation. Most samples were reconsolidated at a pressure of 2.28 MPa and then percolated with SBCW leaching out the DOM present in the liquid of the fresh clay core, then percolated with RBCW up to saturation, followed by a percolation with SBCW at the end of which an impulse of 3 cm³ of RBCW is injected at the inlet and the percolation with SBCW is continued. Other samples were confined (no reconsolidation pressure applied) and percolated with either SBCW or they were first percolated with SBCW and then with RBCW.

The results show a lower apparent dispersion coefficient for the DOM in the fresh clay core after reconsolidation than for the DOM in RBCW collected in the underground laboratory. A i -11 -1 mean value for D app for the consolidated systems is equal to 5.9×10 m²·s . The ηR values obtained by pulse injection (SCK•CEN code A3) are about one third of those obtained by

percolation (SCK•CEN code C4). The results show a minor influence of the stratification on the apparent dispersion coefficient (a maximum factor of 2). The influence of the reconsolidation of the clay cores on the mobility of the DOM seems to be trivial.

To test the hypothesis of the formation of new small organic molecules by disintegration of larger molecules, the percolation of SBCW was halted for 55 and 30 days during tests 9 and 10, respectively, and resumed afterwards. No change in concentration is measured after resumption of the percolation, disproving the hypothesis of the formation of new, small organic molecules.

Because this experimental set-up did not allow to make distinction between injected DOM and DOM that was leached from the solid phase of Boom Clay, subsequent migration experiments used labelled organic material.

3.7.1.2. Experiments performed during TRANCOM-Clay

In the framework of TRANCOM-Clay, single- and double-labelled migration experiments with organic material were performed. The mobility of the DOM was measured by percolation experiments with pulse injection of radiolabelled (125I and 14C) DOM through clay cores (PUT et al., 1998; TRANCOM-CLAY, 2000). For labelling clay water was collected under anaerobic conditions from a filter in the URF and the mobile organic matter was concentrated with DEAE-cellulose. Details on the labelling procedure can be found in Warwick et al. (1993). The stability of the 125I label was proved to be unstable in contact with Boom Clay. 14C (applied as methylamine) as radio-label was proved to be stable and suitable for pulse injections (TRANCOM-CLAY, 2000).

Table 3.10 gives details on the experimental conditions used for the impulse injection experiments with 14C-labelled NOM through intact Boom Clay cores. The DOM concentration evolution at the outlet as a function of time was monitored and this allowed to calculate the migration parameters for a simple diffusion-advection equation. The following experiments were performed: • Single labelled pulse injection experiments (14C-DOM) o In confined clay core holders o In consolidated clay core holders, at different consolidation pressures: 2.7, 1.7 and 0.8 MPa o In a background of synthetic Boom Clay water (SBCW) • Single labelled in-diffusion experiments (14C-DOM) • Double labelled pulse injection experiments (14C-DOM in contact with 241Am) • In situ large-scale single labelled diffusion experiment (14C-DOM).

To study the influence of the molecular size on the mobility of the organic matter, the following three fractions were used for pulse injection: - Low size range (Da < 1000); “L” in table 3.10. - High size range (Da > 100000); “H” in table 3.10. - Full size range (all mobile NOM present); “F” in table 3.10. The fractions were separated by ultrafiltration (Amicon Diaflo ultrafilters; YM1 with cut-off = 1000 Da; YM100 with cut-off = 100 kDa). Filtration was performed in an anaerobic glove box. Gel Permeation Chromatography (GPC) of different batches and the total distribution showed that different size fractions were indeed injected. The experiments were performed in

a temperature controlled room at 25°C, and RBCW was forced through the clay core at a pressure difference of 1 MPa.

A total of eight experiments were performed. The results of the eight tests are given in table 3.10. Figure 3.42 shows examples of the 14C concentration evolution at the outlet of the tests, for the low, high and full size range. The figures show a tailing of the concentration of DOM at the outlet. Possibly, storage and filtration, followed by a gradual release of the stored molecules, are the most acceptable colloid retardation mechanisms for the observed behaviour.

Figure 3.42: Concentration evolution profiles at the outlet of a clay core for impulse injection 14 experiments with C labelled mobile organic matter (TRANCOM-CLAY, 2000)

This idea is supported by the evolution of the recovery curves given in figure 3.43. For the large size fraction a recovery of less than 20 % is observed. But even for the small size fraction no complete recovery is obtained under the experimental conditions. Thus, the behaviour of colloids in Boom Clay shows a kinetic hindrance of a part of the mobile organic matter: the larger the molecules, the larger the storage. For the experiments with the large size range, the size distribution in the percolate at the outlet has been examined by GPC-runs performed for different percolated volumes. From these results it was concluded that only small molecules are percolated, indicating that larger molecules are filtered by the clay.

Figure 3.43: Cumulative recovery curves from the impulse injection experiments, showing the influence of the molecular size fractions of NOM and the influence of the effective stress in the clay pres (tests number 3 to 8) (PUT et al., 1998)

Figure 3.43 also shows the difference of behaviour for the tests in consolidation cells and in confinement cells. The recovery for the consolidation cells (oedometers) is much higher than for the confinement cells. This is explained by the higher porosity, resulting from a lower effective stress, for the clay cores in the consolidation cells. Figure 3.43 also supports the retardation due to filtering and storage of the colloids, because the higher the porosity, the higher the recovery.

The following observations can be summarised from the results of table 3.10: (1) the values of i the apparent dispersion coefficient D app show no systematic trend for the different fractions, and the mean value is 3.2(±1.0)×10-11 m²·s-1; (2) the product ηR shows a higher value of 2.1 for the high size range, the mean value for the low and full range fraction is 0.51. The high size range molecules are more retarded than the low and full range. But the values of the i apparent dispersion coefficient D app are about the same. This cannot be explained by a higher sorption of the larger molecules but by an irreversible hold-up (filtration or straining) in the smaller pores. A part of the stored molecules can escape and continue through larger pores, but the bulk is immobilised.

Pulse injection experiments in a background of synthetic Boom Clay water The breakthrough curves (BTCs) from two different clay cores, one in a background of RBCW and the other in a background of SBCW, are compared in figure 3.44. Surprisingly, the BTC and especially the tailing do not differ much from each other. Although in the case of SBCW the organic molecules in the pores and those readily extractable from the surface were removed, a pronounced sorption effect of the pulse of organics could be expected. Apparently, there is no competition or saturation effect.

Figure 3.44: Comparison of normalised elution profiles after injection of labelled organic matter (< 1000 Da) in a background of RBCW and SBCW (TRANCOM-CLAY, 2000)

In-diffusion experiments Two in-diffusion tests with labelled organic matter have been set up. In this setup clay cores of about 30 mm long were contacted with two reservoirs. The source reservoirs contained the full size fraction of labelled organic matter in one set up, and the fraction of organics < 1000 Da in the other set up. The outlet compartments contained real unlabelled clay water. The source compartments were sampled at regular time intervals. After 28 days (full size fraction) and 27 days (< 1000 Da), the clay is cut in slices in the glove box and the 14C activity was extracted. The 14C profile for the small molecules is presented in figure 3.45. To the author's knowledge, this profile has never been fitted.

Figure 3.45: Relative activity profile in the clay core after diffusion of the 14C-labelled low size fraction of organic matter during 28 days (TRANCOM-CLAY, 2000)

3.7.1.3. Experiments performed during TRANCOM-II

In the framework of the EC TRANCOM-II project (TRANCOM-II, 2004), laboratory migration experiments were performed to investigate transport process characteristics of Boom Clay, taking into account dissolved (natural) organic matter. For this purpose, radionuclide sources (241Am) were prepared with concentrations as close as possible to their expected equilibrium

concentration under in-situ Boom Clay conditions and in contact with 14C-labelled Boom 14 Clay organic matter ( C-DOM), so-called “double-labelled” migration experiments (MAES et al., 2006; TRANCOM-II, 2004). The labelling procedure was different from the one used in TRANCOM-Clay and was intended mostly to provide a more stable label. Laboratory migration experiments were of the percolation type (SCK•CEN code C4). Different experiments were performed with a fixed total plug length but with varying end-plug lengths (i.e., distance of the source to the outlet of the clay core), and varying Darcy velocities (table 3.10). The 14C-DOM solution is characterised by large molecular weight molecules (>30 kDa), and the DOM particle sizes range between 2.1 and 5 nm.

Observed breakthrough curves for the 14C-labelled DOM reveal that peak concentrations and time of the peak are influenced by 1) the position of the source within the clay core, and 2) by the flow rate (see figure 3.46). For similar water fluxes, 14C recovery is independent of travel length and on average about 60 %. For the higher flow rates, recovery is higher too (~ 72-73 %).

Figure 3.46: Experimental 14C-labelled natural organic matter breakthrough curves (IONESCU et al., 2008)

Several modelling approaches were tried to simulate the transport of DOM under in situ Boom Clay conditions. These modelling approaches encompassed a 3-D random walk model with a lognormal distribution of velocities (HYTEC) (TRANCOM-II, 2004), a straightforward model using the classical advection-diffusion equation but considering different retardation factors for different "classes" of DOM (PORFLOW) (TRANCOM-II, 2004), and two colloid transport codes with kinetic attachment-detachment (HYDRUS-1D (IONESCU et al., 2008) and POPCORN (HICKS, 2008; TRANCOM-II, 2004). The latter two are discussed more in detail hereafter.

The HYDRUS code In the HYDRUS transport code, colloidal transport is described by a combination of the convection-dispersion-retardation equation and colloid attachment theory (ŠIMŮNEK et al., 2007; ŠIMŮNEK et al., 2005):

∂()η ∂()s ∂() ∂() ∂ ∂ ∂() c + ρ e + ρ s1 + ρ s2 = η i c  − qDc − μ η − μ ρ ()+ + b b b  Dpore  w HTO c s b se s1 s2 ∂t ∂t ∂t ∂t ∂x  ∂x  ∂x

equation (3.2)

The parameters used in this model are summarised in table 3.11.

Sorption to equilibrium sites can generally be described by:

κ ⋅ c β s = equation (3.3) 1+ χ ⋅ c where κ [m³·kg-1], β [-] and χ [m³·kg-1] are empirical coefficients. When β = 1, equation (3.3) becomes the Langmuir equation, when χ = 0, equation (3.3) becomes the Freundlich equation and when both β = 1 and χ = 0, equation (3.3) leads to a linear adsorption isotherm.

Table 3.11: Parameters of the HYDRUS colloidal transport model.

Parameter Units Definition -1 c Nc·L colloid concentration in the aqueous phase. Nc - Number of colloids - ρb kg·dm Bulk dry density 3 s Nc·kg-1 Colloid solid phase concentration, subscripts e, 1 and 2 represent equilibrium and two kinetic sorption sites, respectively -1 μw s Represents inactivation and degradation processes (e.g., decay rates) in the liquid phase -1 μs s Represents inactivation and degradation processes (e.g., decay rates) in the solid phase i -1 i D pore m²·s Pore dispersion coefficient; D pore = Dpore + αv -1 Dpore m²·s Pore diffusion coefficient α m dispersivity -1 v m·s Pore water velocity; v = qD/η -1 qD m·s Darcy flux η - Diffusion accessible porosity (or "effective porosity") for colloids ηHTO - Diffusion accessible porosity (or "effective porosity") for HTO

The relationship between the colloid concentrations in the aqueous and solid phase may be given by the Freundlich sorption isotherm equation (non-linear sorption):

= κ ⋅ β s F c equation (3.4)

-1 where β is the Freundlich exponent [-] and κ F is the Freundlich sorption coefficient [m³·kg ].

Net kinetic accumulation of colloids on the solid phase from the aqueous phase can be described as the result of attachment/straining and detachment/liberation (the indexes 1, respectively 2, for the kinetic sorption sites have been omitted from the following equation):

∂s ρ =η k Ψ c − k ρ s equation (3.5) b ∂t att det b

-1 where katt is the first-order deposition (attachment) coefficient [s ], kdet is the first-order entrainment (detachment) coefficient [s-1], and Ψ is the colloid retention function [-]. Attachment is the removal of colloids from solution via collision with and fixation to the solid phase, and it is dependent on colloid-colloid, colloid-solvent, and colloid-porous media interactions (BRADFORD et al., 2004; BRADFORD et al., 2005). To simulate reductions in the attachment coefficient katt as a result of filling the favourable attachment sites, a Langmuirian dynamics equation may be used to describe the decrease of Ψ with increasing colloid mass retention (~blocking mechanism) (ŠIMŮNEK et al., 2005):

Ψ = − 1 S Smax equation (3.6)

-1 where Smax is the maximum solid phase concentration [Nc·kg ]. Alternatively, a depth- depending blocking coefficient may be invoked to characterize the so-called straining process, where straining means the entrapment of colloids in down gradient pores and at grain junctions that are too small to allow particle passage:

−β  + −  Ψ = dc x x0   equation (3.7)  dc  where dc is colloid diameter [m], β is a fitting parameter [-] that controls the shape of the colloid spatial distribution, x is depth [m] and x0 is depth of the column inlet or textural interface [m].

Observed breakthrough curves for the 14C-labelled OM reveal that peak concentrations and time of the peak (figure 3.46) are influenced by 1) the position of the source within the clay core, and 2) by the flow rate. Ionescu et al. (2008) developed a modelling strategy with the HYDRUS-1D (ŠIMŮNEK et al., 2005) numerical code in order to estimate meaningful values for migration parameters capable of describing colloid transport. Typical colloid transport submodels tested included kinetically controlled attachment/detachment and kinetically controlled straining and liberation. Size and/or charge exclusion has been taken into account by using a diffusion accessible porosity, η, of 0.13, instead of the porosity of 0.37 obtained for the conservative tracer HTO. For the dispersivity, α, a value of 1×10-4 m has been used, based on initial sensitivity analysis.

The delay of the peak arrival time Tp, compared to the advective water travel time through the clay core Tw, in the experimental breakthrough curves suggests the occurrence of a sorption type mechanism for DOM transport. The calculated retardation coefficient R varies between 2.4 and 3.1, which suggests mildly retarded DOM transport.

Colloid transport parameter starting values have been taken from calculations based on simulations with the POPCORN code, except for the values for straining and liberation rates, since these processes are not modelled within the POPCORN code (HICKS, 2008; TRANCOM-

100

II, 2004). Modelling of the migration of DOM was done by considering three different sorption sites: site (1) equilibrium with non-linear Freundlich sorption (submodel 1), site (2) kinetically controlled attachment/detachment (submodel 2, with katt1 and kdet1 as attachment- detachment rates) and site (3) kinetically controlled straining and liberation (with katt2 = kstr and kdet2 = klib as straining-liberation rates). Initial sensitivity analysis without considering the detachment (kdet = 0, for the attachment-detachment model) and liberation terms (klib = 0, for the straining model) failed in predicting the observed breakthrough curves. Therefore, further calculations always included mechanisms of detachment and liberation.

Part of the 14C-DOM colloids are assumed to be sorbed on the solid surface (the equilibrium site – submodel 1). Presence of the ψ-parameter in the attachment/straining component of the mass-transfer equation allows one to account for different straining/blocking mechanisms. The simulation of breakthrough curves for 14C-DOM colloids taking into account advection, dispersion and “clean-bed” attachment/detachment model (ψ = 1 and kdet = 0) for both kinetic sites failed in reproducing both the peak (including peak arrival time) and the tailings of the BTC curves. Including both attachment-detachment and depth-dependent straining in the modelling did however improve the modelling and simulated BTCs were in good agreement with experimental BTCs.

Table 3.12: Optimised parameter values for the fit of colloidal transport using the HYDRUS code (IONESCU et al., 2008)

Parameter Layer M1 & M3 Layer M2 η 0.14 0.3 (f) Dpore 6.43 3.94 κF 92.01 0 (f) β 0.43 1 (f) katt1 0.015 (0.016) kdet1 0.0012 (0.0015) katt2 (kstr) 0.090 (0.015) kdet2 (klib) 0.0095 (0.0015) 25 Smax2 (β) 0.45 10 (f) dc 0.00095 0.00091

Figure 3.47: Fitted and observed breakthrough 14C curve for a double-labelled 14C-241Am migration experiment (IONESCU et al., 2008)

101

Three sections (materials) were considered for modelling purposes: the inlet domain M1, a very thin source layer M2, and the outlet domain M3. The source layer M2 is modelled as a medium with a higher effective porosity (η = 0.30). The fitted parameter values using inverse calculations are shown in table 3.12, while observed and fitted BTC are shown in figure 3.47.

Using the optimised parameters for the inverse simulation of a first experiment, the other experimental BTCs were directly calculated. Figure 3.48 shows that the forward simulations are in good agreement with the experimental data when the source is located in the middle of the clay column (top two figures), but for experiments with a different position of the source (bottom two figures), the simulations fail in representing the data.

Figure 3.48: fitted and observed 14C breakthrough curves for a double-labelled 14C-241Am migration experiment. Fits are obtained by forward simulations. Upper figures are for experiments where source is located in the middle of the clay column while lower figures are for experiments with a different position of the source (IONESCU et al., 2008)

Owing to the difference in travel time between these latter two experiments and the experiments where the source is applied in the middle of the column, several processes (for example, dispersion, kinetically controlled attachment/detachment, etc.) that are distance and/or time dependent may not be well described on the basis of parameter values derived from tests with different space and time scales. Additional sensitivity analyses have shown that the most influential parameters for the experiments are: the attachment rate katt, Freundlich distribution κF, and pore diffusion coefficients Dpore. Straining did not seem to influence the peak travel time and peak height, but it has an important role in explaining the tailing of the BTC. These results therefore suggest that colloid migration parameters are sensitive to the experimental setup, more specific to the spatial scale used in deriving parameter values. Therefore, it was concluded that extrapolation of parameter values obtained from small-scale core samples to large scale in-situ conditions must be done with care.

102

The POPCORN code. The POPCORN radionuclide transport model includes representation of the effects of dissolved organic matter on radionuclide migration in a porous medium, but focuses mainly on the radionuclide itself while DOM features only as an "accessory" transport vector. Dissolved radionuclides may be sorbed on the rock matrix or may be complexed with OM. Mass transport occurs by advection and dispersion of the dissolved radionuclide and radionuclide-OM complexes. The movement of radionuclide-OM complexes may be affected by exchange between mobile and immobile phases by, for example, filtration and remobilisation in the rock matrix. The one-dimensional advection and dispersion equations for transport of dissolved radionuclides and radionuclides complexed to mobile OM are:

∂c ∂ 2c ∂c R s = D s − u s − Rλc − κ ()K c − c − κ ()K c − c ∂t s ∂x 2 s ∂x s m m s m f f s f equation (3.8)

∂c ∂ 2c ∂c m = D m − u m − λc + κ ()K c − c − (s c − s c ) ∂t m ∂x 2 m ∂x m m m s m m m f f equation (3.9)

and the behaviour of immobile radionuclide-OM complexes is described by the following:

∂c f = −λc + κ ()K c − c + (s c − s c ) ∂t f f f s f m m f f equation (3.10)

3 where cs, cm and cf (Bq/m ) are the concentrations of radionuclides in solution, complexed to mobile OM, and complexed to immobile OM. The parameters in equations (3.8) to (3.10) are defined in table 3.12 and the modelled processes are illustrated in figure 3.49.

Clay

Immobile C R κ f k f f cs κ c s c f f Cs us,Ds m m s c Dissolved f f Mobile Cm um,Dm k κ C m m s κ Pore mcm

Figure 3.49: Illustration of radionuclide transport processes represented in POPCORN (HICKS, 2008)

103

Table 3.12: Parameters of the POPCORN radionuclide transport model (HICKS, 2008)

Parameter Units Definition -1 -1 u m·s Mean fluid velocity, which is Us/φs, where Us (m·s ) is the fluid Darcy s velocity and φs is the total porosity. -1 -1 u m·s Mean velocity of fluid carrying OM, which is Um/φm, where Um (m·s ) is m the Darcy velocity for fluid carrying OM and φm is the diffusion accessible porosity for OM. m²·s-1 α Ds Ls us Solute pore dispersion coefficient given by RDsapp + , where Dsapp α (m²·s-1) is the apparent solute diffusion coefficient and Ls (m) is the intrinsic longitudinal dispersivity. m²·s-1 α Dm Lm um Hydrodynamic pore coefficient for mobile OM given by Dmapp + α -1 Lm where Dmapp (m²·s ) is the apparent OM diffusion coefficient and (m) is the intrinsic OM longitudinal dispersivity. -1 λ s Radionuclide decay constant. Ingrowth is not modelled. R - Retardation coefficient for radionuclides sorbing to the clay surface, = + ()−φ ρ φ R 1 1 s Kd s -1 which is given by , where Kd (L·kg ) is the distribution coefficient for radionuclides sorbing to the clay and ρ (kg/dm3) is the density of the solid rock. κ /s Kinetic constant for complexation of radionuclides with mobile OM. m κ /s Kinetic constant for complexation of radionuclides with immobile OM. f K - Partition coefficient for complexation of radionuclides with mobile OM, m α Kdm m 3 which may be defined as , where Kdm (m /kg) is equivalent to a distribution coefficient for complexation of radionuclides with mobile 3 OM, and αm (kg/m ) is the mass of mobile OM per unit fluid volume. K - Partition coefficient for complexation of radionuclides with immobile f α Kdf f 3 OM, which may be defined as , where Kdf (m /kg) is equivalent to a distribution coefficient for complexation of radionuclides with immobile 3 OM, and αf (kg/m ) is the mass of immobile OM per unit fluid volume. sm /s The rate of exchange of radionuclides between mobile and immobile OM. sf /s The rate of exchange of radionuclides between immobile and mobile OM.

The model assumes linear, reversible, instantaneous adsorption of dissolved radionuclides on the clay surface according to the retardation coefficient R. Radionuclide-OM complexation is described by a linear, kinetic model that is analogous to a non-instantaneous adsorption- -1 desorption model. The term κm (s ) is a kinetic constant for the complexation reaction for radionuclides and mobile OM, and the dimensionless term Km is effectively a partition coefficient for the reaction, which may be expressed in the form Km = Kdmαm where Kdm 3 3 (m /kg) is equivalent to a distribution coefficient and αm (kg/m ) is the mass of mobile OM per unit fluid volume. The term κm(Kmcs – cm) in equations (3.8) and (3.9) thus represents mass transfer between radionuclides in the dissolved phase and radionuclides complexed with mobile OM. Similarly, the term κf(Kfcs – cf) in equations (3.8) and (3.10) represents mass transfer between radionuclides in the dissolved phase and radionuclides complexed with immobile OM.

104

Exchange of radionuclides between mobile and immobile OM complexes is characterised by the term smcm - sfcf, which may represent an OM filtration and detachment process in the clay or an additional dissociation and complexation process. It is assumed that none of these processes affects porosity.

Simulation of the migration experiments shown in figure 3.46 using the POPCORN model were performed using the following constraint for OM: negatively-charged OM will be electrostatically repelled from clay surfaces and, therefore, will have a lower diffusion- accessible porosity, φm. In TRANCOM-II (2004) a diffusion-accessible porosity of 0.13 for OM in the Boom Clay was reported. The hydrodynamic dispersion coefficient for OM, Dm, is -11 -10 2 considered to be in the range 10 to 10 m /s (TRANCOM-CLAY, 2000; TRANCOM-II, 2004), with any retardation characterised by the kinetic terms described above. Mobile OM represents less than 0.1 % of the total OM content of the clay (TRANCOM-II, 2004) and therefore a value of the coefficient Kf for immobile OM has been assumed to be a factor of 3 10 greater than the value Km for mobile OM.

The above-mentioned values were considered as a starting point for the POPCORN analyses of the migration experiments. It has been possible to simulate the migration of 14C-DOM in the migration experiments with some success by introducing the terms sm and sf, which describe the rates of interaction between mobile and immobile OM. For the 11 experiments 14 involving the TROM34 solution, the C-DOM percolation curves were matched with sm in a -8 -8 -1 -8 -1 fairly narrow range of 5×10 to 12×10 s with sf fixed at 1.4×10 s in each case. As the rates sm and sf are increased, a linear, reversible instantaneous adsorption model is approached with the retardation coefficient equal to 1 + sm/sf . Based on the derived values of sm and sf the retardation coefficient for DOM would lie in the range of about 4 to 10.

There appears to be a correlation between the derived values of sm and the mean DOM velocity, um, which is consistent with observations of colloid filtration behaviour in porous media (Hicks, 2008). The relationship between these terms can be seen in figure 3.50, which presents the derived values of 1/sm against the estimated peak advection-driven OM breakthrough time. Not surprisingly, the two experiments involving the TROM6 solutions are not consistent with the correlation for the TROM34 experiments, as indicated in figure 3.50. Most of the 14C-OM in the TROM34 solution is of much greater molecular weight than that in the TROM6 solution and, therefore, smaller filtration rates sm would be expected for the TROM6 solution.

There is also a tendency for a higher hydrodynamic dispersion coefficient to be required to fit the percolation curves at higher mean fluid velocities. This behaviour may reflect velocity- dependent longitudinal dispersivity.

Because the pressure gradients under which the migration experiments were carried out may be higher than would be expected under repository conditions, it was concluded that a model in which there is no OM filtration may be more suitable and conservative for any applications in performance assessments.

105

350 y = 0.3369x + 92.15 300 R2 = 0.85 250 200 (days)

m 150

1/S 100 50 0 0 100 200 300 400 500

Core length/Um (days)

TROM34 TROM6 Linear (TROM34)

Figure 3.50: The relationship between the core length divided by the mean fluid velocity and 14 the term 1/sm based on the POPCORN analyses of the C-OM percolation 2 experiments. A least squares fit trend line and R value are shown (HICKS, 2008)

3.6.3.4. Long-term in situ diffusion experiment

As a last type of experiment, a large-scale in situ migration experiment with 14C-labelled DOM was performed to study the DOM migration behaviour on a large scale (m), on the long-term (> 10 years) and in directions parallel and perpendicular to the bedding plane (transport anisotropy). The experiment was initiated during the TRANCOM-Clay project, and was followed up during TRANCOM-II and thereafter.

The outline of the in-situ experiment is as follows. Diluted 14C-NOM batches, transferred into teflon-coated containers and stored in a refrigerator to prevent bacterial growth, were pumped continuously over a filter placed in the Boom Clay to allow the labelled organic matter to diffuse out into the formation (figure 3.51). One horizontal piezometer (TDR41H, started June 9, 1997) and one vertical piezometer (TDR41V, started June 25, 1997) were used. At regular time intervals, the source filter and neighbouring filters are sampled for 14C analysis. The injection in the horizontal filter is sampled and the first activity is detected in the neighbouring filter, just above the detection limit (TRANCOM-CLAY, 2000).

106

Figure 3.51: Schematic representation of the emplacement of the piezometers in the HADES URF (SCK•CEN, Mol, Belgium) for the migration experiment with 14C-labelled Organic Matter (MARTENS et al., 2010)

1.0x1011

1.0x1010

1.0x109 m³) / 1.0x108

1.0x107

1.0x106

C activity(Bq injection filter - V

14 5 1.0x10 injection filter - H first neighbouring filter - V 1.0x104 first neighbouring filter - H

1.0x103 0 500 1000 1500 2000 2500 3000 3500 time (days)

Figure 3.52: Measured 14C-activity in the sampling filters nearest to the injection filters (filters 7) and in the injection filters (filters 8). Data for the horizontal and vertical piezometer are indicated with a "H" and "V" respectively (MARTENS et al., 2010)

The experimental data from the in-situ migration test are shown in figure 3.52. The in-situ experiments demonstrate on a larger scale the mobility of natural organic matter. Some first modelling attempts were done during the TRANCOM-II project but these attempts were unable to consistently reproduce the experimental breakthrough curves. This was attributed to the colloidal behaviour of NOM. Martens et al. (2010) aimed to model the experimental data using the HYDRUS 2D/3D code (ŠIMŮNEK et al., 2007), which includes a colloid transport module. For the vertical piezometer, an axisymmetric model was gradually developed, by increasing the complexity of the model step by step by introducing observed processes. First, only diffusion and linear sorption were included. In a subsequent step advection was added

107

and non-linear sorption (Freundlich) was examined. Finally, the effect of adding a colloid transport model was evaluated. The initial parameter values were taken from small-scale lab experiments. The final model was then validated on the dataset from the horizontal piezometer. The anisotropy of the clay was accounted for through the hydraulic conductivity and the pore diffusion coefficient.

Step-by-step model building for the vertical piezometer dataset a) Diffusion and linear equilibrium sorption only

Figure 3.53a shows the results for a simulation including diffusion as the only transport process. Linear equilibrium sorption was also included in the model by applying a small retardation coefficient R = 3.1, based on the average ηR value obtained from small-scale lab experiments with similar labelled DOM. The diffusion coefficient is then the fitting parameter starting with the Dpore value derived from lab experiments. A reasonably good fit was -11 -1 -11 obtained for a pore diffusion coefficient of 8.5×10 m²·s (equals a Dapp=Dpore/R= 2.7×10 m²·s-1). However, if we zoom in on this curve (figure 3.53b) it is also clear that variation of the diffusion coefficient does not allow to obtain exactly the same shape as the experimental curve, which indicates that other processes need to be included. b) Diffusion, equilibrium sorption (linear vs. non-linear sorption) and advection

In a next simulation, advection was added to the model. The gallery induces a hydraulic gradient in the clay. The pressure head between filter 7 and 8 was measured and the average of these measurements (7.37 m/m) was used in the simulation. For the vertical hydraulic conductivity of the Boom Clay a value of 1.7×10-12 m·s-1 was used. The pore diffusion coefficient was set at 8.5×10-11 m²·s-1, i.e. the best fit obtained in part a. Predictions for the simulation with and without advection are compared in figure 3.53c. The magnitude of the difference between the two curves indicates that advection is not a dominant transport process for this experiment, but neither can it be neglected. It is also clear that introduction of an advective term does not allow to obtain the same shape as the experimental curve, which indicates that still other processes need to be included.

β Therefore we first investigated non-linear sorption (Freundlich formalism, sk=k*C ) to account for the specific shape of the diffusion curve. An excellent fit to the experimental values was obtained with following (pseudo-)Freundlich parameters: β=1.325 and k=4.5 10-7 m³·kg-1 (figure 3.53d). However, for a conventional Freundlich sorption mechanism, β should be < 1 (hence the name "pseudo-Freundlich"). A surface precipitation process can sometimes be described with a β>1, but such a process has not been observed for DOM in Boom Clay. Therefore we only consider linear equilibrium sorption in the following steps.

c) Diffusion, linear equilibrium sorption, advection and kinetic colloid attachment/detachment

The size distribution of the 14C-DOM from the sampling filters was also analysed. It can clearly be seen that the size distribution is shifting towards smaller molecules as the fraction < 1 kDalton increases while the fraction between 1-10 kDalton decreases compared to the distribution of the source solution. This indicates that a filtration process is influencing the NOM transport.

108

Therefore a colloidal mechanism (besides linear equilibrium sorption) was introduced in the model. Kinetically controlled colloid attachment/detachment was included in its most simple form. The colloid retention function Ψ was set to 1 and the starting values for the attachment and detachment coefficients were taken from Ionescu et al. (2008) in which similar colloid transport modelling was done on comparable migration experiments involving complexed 241Am-14C-labelled DOM in Boom Clay. The parameters were then optimized. The detachment rate has been considered to be one order of magnitude smaller than the attachment rate, referring to a situation where the process of a colloid entering a pore and interacting with the surface (attachment) occurs faster than the reverse process (detachment).

1.2x108 2.5x107 -10 -11 Dp= 1.5×10 m²/s Dp= 8.5×10 m²/s 8 1.0x10 7 -10 2.0x10 experimental Dp= 1.0×10 m²/s 7 8.0x10 -11 Dp= 8.5×10 m²/s 1.5x107 6.0x107 experimental 1.0x107 4.0x107

6 C activity (Bq/m³)C activity 7 (Bq/m³)C activity 5.0x10

14 2.0x10 14

0.0x100 0.0x100 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500 time (days) time (days)

2.5x107 2.5x107 diffusion & lin. sorption only diffusion,pseudo-Freundlich sorption + advection diffusion & lin. sorption + advection diffusion, lin. sorption, advection + colloid transport 2.0x107 2.0x107 experimental idem but R=0 idem, R=0, optimised kinetics 1.5x107 1.5x107 experimental

1.0x107 1.0x107

6 6 C activity activity (Bq/m³) C C activity(Bq/m³) 5.0x10 5.0x10 14 14

0.0x100 0.0x100 0 500 1000 1500 2000 2500 3000 3500 0 500 1000 1500 2000 2500 3000 3500 time (days) time (days)

Figure 3.53: 14C-activity in filter 7 (Bq m-³) as a function of time (days). Experimental values and modelled curves for the vertical piezometer. (a) Model considering only diffusion & linear sorption: sensitivity analysis on the pore diffusion coefficient. (b) Model considering only diffusion & linear sorption: best fit. (c) Model considering only diffusion & linear sorption versus a model including advection -11 -1 (both for Dpore = 8.5×10 m² s ). (d) Comparison between simulation results for a model including diffusion, linear sorption vs. non-linear sorption (pseudo-Freundlich because fitted β>1), advection and a model also including kinetic colloid attachment/detachment kinetics with or without linear sorption taken -11 into account (both with a Dpore = 9.5×10 m²/s). (MARTENS et al., 2010)

-9 -1 -10 -1 The final parameter values are katt = 8.1×10 s and kdet = 8.1×10 s and also the Dpore was -11 -1 slightly adapted to Dpore = 9.5×10 m² s . The model results obtained with these parameters are shown in figure 3.53d. Including colloid transport clearly improved the shape of the

109

modelled curve but still the fit is not perfect. Probably, this can be improved by optimising the colloid model, e.g. by the introduction of blocking/straining mechanisms. However, this increases the number of fitting parameters for which no independent data are available. As mentioned in the beginning, a robust, reasonable description with the lowest complexity and the least number of fitting parameters was strived for.

To verify the additive approach of the different processes, it was tested if the linear sorption could be neglected and only a kinetic interaction with the solid phase considered. In fact, sorption is then incorporated in the kinetics part of the colloid model. The results of this -8 -9 -1 exercise (figure 3.53d), with optimised katt and kdet (respectively: 3.2 10 and 3.2 10 s ) show that by neglecting the linear sorption, no reasonable reproduction of the experimental curve could be obtained.

Testing of the model against the horizontal piezometer dataset

Due to the anisotropy of the Boom Clay, the hydraulic conductivity and the pore diffusion coefficient are different for the horizontal and vertical direction. In this modeling exercise, a horizontal hydraulic conductivity of 4.1×10-12 m·s-1 was used (within the range of typical measured values for the Boom Clay (YU et al., 2011). This is 2.4 times higher than the vertical hydraulic conductivity. There should be no anisotropy for sorption and colloid attachment/detachment, so the same parameters as for the vertical dataset were used for these processes. The pore diffusion coefficient is thus the only fitting parameter which means that a validation of the sorption and colloid processes is performed. Furthermore, the average measured horizontal hydraulic gradient (6.50 m/m) is somewhat lower than the vertical gradient.

The model including linear sorption, diffusion, advection and kinetically controlled colloid attachment/detachment (model c) was then tested for the horizontal dataset. Results are shown in figure 3.54. From this figure it is clear that the model, which was deduced based on the vertical dataset, is also applicable for the horizontal dataset. The best fit (figure 4.18) was -10 -1 obtained for Dpore = 1.33×10 m²·s . This is 1.4 times higher than the best fit Dpore value for the vertical pore diffusion coefficient, but somewhat lower than the anisotropy value of 2 used for a blind prediction of iodide migration on the same piezometer set-up based on lab derived transport parameters.

5.0x107 diffusion, advection and colloid transport experimental 4.0x107

3.0x107

2.0x107

7 C activity(Bq/m³) 1.0x10 14

0.0x100 0 500 1000 1500 2000 2500 3000 3500 time (days)

Figure 3.54: Activity in filter 7 (Bq/m³) as a function of time (days). Experimental values and modeled curve for the horizontal piezometer (MARTENS et al., 2010)

110

It was concluded that more elaborate transport models (including a colloidal transport part) can indeed better describe the migration of the colloidal NOM in the Boom Clay, but this remains a fitting exercise because there are no independent determinations of the attachment/detachment kinetics. Nevertheless, the modeling exercise revealed that the "lumped" parameters obtained from lab experiments can be transferred to larger scale experiments and that even a classic diffusion-advection description gives already quite good results.

3.6.3 Conclusion on the migration of dissolved organic matter in Boom Clay

From the description of the different DOM migration experiments and of the different colloidal transport models that were used to simulate them, it is clear that there exists currently no unique set of colloidal transport equations that are able to grasp the features observed in these experiments. Moreover, most colloidal transport equations take into account a whole range of processes and mechanisms (linear and non-linear equilibrium sorption- desorption, kinetic attachment-detachment, straining, etc.). All these processes are in their turn described by a number of parameters whose values need to be assessed. This would require a huge number of experiments and data, which are currently not available.

From laboratory-scale transport experiments, values of the product of the diffusion accessible porosity (η) and the retardation factor (R) can be deduced, as well as values for the apparent i dispersion (D app) and/or diffusion (Dapp) coefficient. According to Martens et al. (2010), the modelling exercise of the in-situ 14C-DOM diffusion experiment revealed that these "lumped" parameters can be transferred to larger scale experiments and that a classic diffusion- advection description gives already quite good results for describing colloidal transport behaviour of DOM in Boom Clay. It is therefore assumed (MAES et al., 2011) that the following transport equation describes colloidal DOM transport in Boom Clay:

∂C D ∂ 2C Tc−DOM = pore,OM Tc−DOM equation (3.11) ∂ ∂ 2 t ROM x

This equation entails that no irreversible or non-equilibrium processes take place. Moreover, only linear, reversible sorption is considered as a factor in the retardation of organic matter- related colloids. This means that, contradictory to several experiments, no straining or filtration of the colloids is considered. All colloids that are at a certain point mobile in the confined Boom Clay structure, will remain so in the future and will not be filtered out due to pore size or pore throat constrictions. This mobile DOM colloid concentration is assumed to correspond to the DOC concentration found in piezometers (more precisely, the Morpheus piezometer, which samples organic matter from different horizons within the Boom Clay at Mol). The maximum size of the DOM colloids found in these piezometers corresponds to the maximum size of colloids that are assumed to be mobile in the Boom Clay.

These assumptions follow the recommendations made in the POPCORN report (Hicks, 2008) which state that "a model in which there is no OM filtration may be more suitable and conservative for any applications in performance assessments" and correspond with the approach adopted in section 3.5 to estimate the ranges for the mobile DOM colloid concentration.

111

3.6.3.1. Diffusion accessible porosity and retardation factor

We have shown in previous sections that the retardation factor for dissolved organic matter is difficult to determine because several processes are involved: (straightforward) sorption- desorption on (e.g. oxide) solid phases, association with other organic matter molecules, colloid filtration processes. In all experiments we have witnessed non-conservative behaviour for dissolved Boom Clay organic matter molecules, pointing to both sorption-desorption (resulting in retardation of the breakthrough curves) and colloid filtration (resulting in incomplete recovery). It is evidenced that larger organic matter molecules are more subject to these retardation processes than smaller molecules. Moreover, from the existence of different DOC concentrations in sampled piezometer water, it is hypothesised that the DOC concentration in each stratigraphic horizon is determined both by the pore structure (which would be different for silty and clayey horizons) and the kerogen composition (BLANCHART, 2011). Therefore colloid filtration theory might play a crucial role to describe dissolved organic matter movement throughout the formation. Because this theory is currently not fully developed for Boom Clay, we take into account only (reversible) sorption-desorption processes in the current parameter estimation. It is noted that this approach might be over- conservative.

The lower and upper limits for ηR based on data of Put et al. (1998; 1992) are 0.38 to 2.7. The value ranges for the diffusion accessible porosity, η, are based on the parameter ranges selected for iodide (BRUGGEMAN et al., 2010a), because of the negative charge associated with dissolved organic matter (colloids). The best estimate (BE) and source range are taken identical to iodide, while the expert range combines the lowest value for anions (value taken 2- from data collection form of selenate, SeO4 ) with the upper limit of the expert range for iodide.

Source - Expert range Best Estimate Expert range Source range Diffusion accessible porosity 0.16 0.05 – 0.20 0.05 – 0.40 for DOM colloids [ηDOM, -]

If we combine the range for ηR with the best estimate value for η (= 0.16, BE anion accessible porosity for iodide), this gives a value range for R equal to 2.4 – 16.9. The mean value for the retardation of the "low" and "full" size ranges of dissolved Boom Clay organic matter, fitted from diffusion experiments (PUT et al., 1998) with ηR = 0.51 and BE for η = 0.16 gives R = 3.2. These lower values for R are presumed to be representative for the smallest fraction of DOM. We take R = 3 as the lower limit for the expert range. The upper limit for R is obtained from fitting breakthrough profiles of sequential migration experiments (MAES et al., 2011). These profiles correspond to radionuclides that are assumed to be 100 % associated with Boom Clay DOM. From earlier experiments it was already clear that 1) larger size fractions of DOM have the highest influence on RN transport and 2) DOM-associated RN breakthrough curves lag with respect to the early breakthrough of the smallest DOM fractions.

112

The best estimate value for R is obtained by expert judgement, combining the higher retardation values from sequential migration experiments with the knowledge that in percolation experiments, DOM-associated RN breakthrough profiles lag slightly behind the breakthrough of DOM itself.

Source range values for R were derived as follows. Because the uptake of DOM by solid phases was independently confirmed by batch experiments with (14C-labelled) Boom Clay organic matter (BRUGGEMAN et al., 2010b; TRANCOM-II, 2004) (Kd range from 0.5 to 10 L/kg on illite and Boom Clay), we do not take into account migration as a conservative tracer (R = 1) as a realistic possibility. The lower limit is therefore taken equal to the lowest value for R based on a BE value for η (see above). The upper limit is obtained by combining the highest value for ηR with the lowest value for η (= 0.05) and rounding off.

Source - Expert range Best Estimate Expert range Source range Retardation factor for DOM 10 3-30 2-50 colloids [RDOM, -]

3.6.3.2. Diffusion coefficients

From results of DOM migration experiments in the Boom Clay at small scale (PUT et al., 1998; PUT et al., 1992; TRANCOM-CLAY, 2000; TRANCOM-II, 2004) it was observed that different measurement techniques and size fractions did not have much influence on the i i -11 apparent dispersion coefficient (D app) and an average D app = 4.5(±2.1)×10 m²/s was i -11 -10 -1 obtained. The total range of D app values was from 1.9×10 to 1.8×10 m²·s . As a i simplifying assumption, we will neglect the contribution of advection in these D app values and therefore the apparent dispersion coefficients are judged to be equal to apparent diffusion coefficients.

On the other hand, the larger-scale in situ migration experiment with 14C-labelled DOM could -11 be fitted with Dapp value of 2.7-3.1×10 m²/s (MARTENS et al., 2010). This value is taken as best estimate because of the larger scale (~m) used in these experiments compared to ordinary -11 laboratory (percolation) tests. The corresponding Dpore value is about 8.5-9.5×10 m²/s.

The minimum and maximum values for the expert range are taken from the experiments of Put et al. (Put et al., 1998), combining different percolation conditions and different sizes of mobile organic matter. The source range is derived by rescaling the best estimate for Dapp taking into account an R source range of 2 to 50 and inverse dependency between Dapp and R: -12 LL = BE×BE/UL = 4.9×10 m²/s and UL = -10 BE×BE/LL = 1.5×10 m²/s.

Source - Expert range Best Estimate Expert range Source range Apparent diffusion coefficient 3.0×10-11 1.9×10-11 – 6.0×10-12 – for DOM colloids 8.9×10-11 1.5×10-10 -1 [Dapp, DOM, m²·s ]

113

From the modelling of the in situ test by Martens et al. (2010), an anisotropy factor of about 1.4 was obtained. This is somewhat lower than the anisotropy value of 2 used for a blind prediction of iodide migration on the same piezometer set-up based on lab derived transport parameters (BRUGGEMAN et al., 2010a). We therefore select the range of 1.4 – 2.0 as appropriate estimates for the anisotropy of the DOM diffusion coefficient in Boom Clay.

114

4. Perturbations and evolving conditions related to Boom Clay organic matter

4.1 Thermal perturbation

Due to heat liberation by high activity waste, the surrounding clay will be submitted to some thermal stress during long term disposal. The intensity of the thermal stress would be controlled by the technical conditions of the long-term storage and by how long the pre-burial storage aimed at decreasing this heat liberation would last. According to a thermal model developed by Sillen and Marivoet (2007), temperature nearby the disposal galleries would reach a maximum of 80°C within the first 10 years, and then it would take a few thousand years for the whole clay formation to cool down and return to normal conditions.

As a result, the immature kerogen in the clay may release various compounds (DENIAU et al., 2004; DENIAU et al., 2005b). It is therefore important to derive detailed information on the Boom Clay kerogen and its behaviour under thermal stress.

4.1.1. Release of thermolabile components

Hydrocarbonaceous compounds are mainly generated under pronounced thermal stress during catagenesis, as the result of extensive cracking of carbon-carbon bonds in kerogen. Contrary to the hydrocarbonaceous compounds, the oxygen-containing components should start to be released for much weaker thermal stresses. Furthermore, due to their polarity and physico- chemical features, such oxygen-containing compounds may change the chemical confining properties of the argillaceous formations.

Deniau et al. (2004) investigated the thermolabile components of Boom Clay kerogen released in the first stage of pyrolysis (see also section 2.2.3). Such components can correspond to (i) molecules tightly trapped within the macromolecular structure, hence not extracted by solvents or (ii) constitutive units of the whole macromolecular structure of the kerogen exhibiting a low thermal stability. The "thermal extracts" thus obtained represent only a low fraction of the kerogens. However, taking into account their great thermal reactivity, they are the first compounds that would be released upon a thermal stress.

Deniau et al. (2005b) elaborated on the study of Deniau et al. (2004) by researching the low molecular weight polar products, which could affect the effectiveness of the geological barrier due to, e.g., complexation of radionuclides or changes in the physico-chemical features of the clay like pH. To better simulate natural conditions (as compared to the methodology from Deniau et al. (2004)), closed pyrolysis in sealed tubes was carried out under various temperature/time conditions. The conditions were selected in order to simulate the thermal stress related to the CERBERUS (NOYNAERT et al., 1998; NOYNAERT, 2000) experiment (80 °C for 5 a) and to long-term disposal (80 °C for 1 ka and 120 °C for 3 ka). The temperature/time conditions used for simulating the different thermal stresses to be examined, with a higher temperature compensating for a shorter time, were determined using the classical approximation, i.e., a twofold increase in the reaction rate for a temperature increase of 10 °C. Bulk quantitative features, i.e., the total amount of gaseous and soluble products

115

generated, were determined for the above set of conditions; detailed molecular analysis of the C12+ soluble products was performed by GC/MS and the main series of O-containing polar products quantified. In the analysis procedure, the most labile liquid components were totally or partially lost (up to C12 for the n-alkanes) so that the soluble fraction obtained in fact corresponds to the "C12+" compounds.

Deniau et al. (2005b) used a clay sample collected from a core drilled at 230 m depth in the HADES underground facility at Mol. The isolated kerogen showed a large contribution of O functions as reflected by a high O/C atomic ration of 0.35. Analysis of the bulk features points out that as early as for the weakest thermal stress applied (80°C for 5a), low but significant amounts of gaseous and soluble compounds are generated. The production of these two fractions markedly increases with the extent of stress and, together, they account for about 20 % of the total weight of the initial kerogen for the 120 °C/3 ka experiment. These bulk observations confirm that a large fraction of the Boom Clay kerogen exhibits a rather low thermal stability.

In the simulation of the CERBERUS (NOYNAERT et al., 1998; NOYNAERT, 2000) experiment (80 °C/5 a), significant amounts of soluble components are generated comprising a wide variety of polar compounds dominated by mono- and dicarboxylic acids. Normal saturated carboxylic acids are generated ranging from C11 to C28 with a maximum at C16 and a marked predominance of even-C-numbered compounds. This series accounts for ca. 0.3 wt.% of the total mass of the unheated kerogen. A series of C4 to C22 dicarboxylic acids, with a maximum at C9 and a similar abundance corresponding to about 0.3 wt.% of the total mass of unheated kerogen, is also observed. Several other series and individual components also occurred, but in lower relative abundance. Most of these minor compounds correspond to O-containing (and, to a lesser extent, also N-containing) products (a.o., isoprenoid carboxylic acids, branched carboxylic acids, benzoic acids, normal alkan-1-ols, etc.). The bulk of the acids probably corresponds to trapped compounds and was made soluble through the weak thermal stress applied in the experiment. Some compounds, however, are more likely formed through the cracking of thermally labile units of the kerogen. These observations are in line with those made during the in situ CERBERUS experiment: a drop in pH was monitored which revealed a substantial production of CO2 (BEAUCAIRE et al., 2001; NOYNAERT et al., 1998; NOYNAERT, 2000). This CO2 is a major pyrolysis product of low maturity kerogens and starts to be released already under mild conditions (DENIAU et al., 2005a).

The simulation of stronger thermal stresses, related to the range expected for long-term disposal of high-activity nuclear waste, resulted in substantial cracking of the kerogen along with secondary degradation of some of the primary pyrolysis products. In both simulations, n- alkanes appear now as the predominant series of the C12+ fraction. For the 80 °C/1 ka stress, a number of compounds detected for the weak stress are still observed in the C12+ fraction. However, (i) the amount of di- and, to a lesser extent, monocarboxylic acids decreases due to secondary cracking, (ii) these secondary reactions are also reflected by the presence of a series of alkylated aromatic hydrocarbons formed by cyclisation-aromatisation and (iii) ketones formed by the cracking of ether bridges are then obtained. When the 120 °C/3 ka stress is simulated, (i) the carboxylic acids are no longer observed at the end of the experiment, as well as other O- and/or N-containing products, due to complete elimination through secondary degradation, (ii) the relative abundance of alkyl-substituted aromatic hydrocarbons increases along with their ring number and (iii) several phenolic compounds, likely formed via cracking of the lignin fraction of the kerogen, are obtained. These phenolic compounds account for about 0.2 wt.% of the total mass of the unheated kerogen (DENIAU et al., 2005b).

116

The 3 simulation experiments therefore yielded a soluble fraction comprising a wide variety of polar O- and/or N-containing compounds. Moreover, the nature and/or the relative abundance of these compounds exhibit strong variations, with the extent of the thermal stress, reflecting the primary cracking of different types of structures with different thermal stability and the occurrence of secondary degradation reactions. The amount of an individual series of polar compounds generated under a given stress is rather low. Furthermore, some of these polar compounds are only transiently formed and, due to extensive secondary degradation, they are no longer present at the end of the stress. Thus, the mono- and dicarboxylic acids obtained when simulating the 80 °C/5 a stress account for 0.6 wt.% of the initial kerogen, but such products are absent at the end of the 120 °C/3 ka simulation. However, it should be noted that (i) due to the large variety of the O- and/or N-containing compounds in the C12+ fraction the total amount of these polar products should be substantial, (ii) the lower homologues of the polar series, corresponding to the C12- soluble products, were not taken into account, (iii) the primary products transiently observed in closed laboratory experiments may migrate away from the hot zone under natural conditions and thus escape secondary degradation, and (iv) if such a migration would not take place, the degradation of these polar products would generate low molecular weight compounds in the C12- fraction and, eventually, CO2 in the gaseous fraction.

Deniau et al. (2005b) determined, through closed pyrolyses under a wide range of temperature/time conditions, the kinetic parameters related to the formation of soluble hydrocarbonaceous compounds from the Boom Clay kerogen. These kinetic parameters were used to simulate the production of such compounds for a maximal thermal stress presently estimated for the hot phase of a high-level waste repository (80 °C during 1000 years). Under these conditions, a very weak production of only ~0.53 mg "HC"/g TOC (corresponding to only ~16 mg "HC"/kg rock) was calculated. This low value indicates that only a very small fraction (0.53 mg/332 mg, i.e. less than 0.2 %) of the total potential for hydrocarbonaceous compound production would be released under such a type of thermal stress. Due to this low production, added to their apolar (or weakly polar) nature, the production of hydrocarbonaceous compounds should not, therefore, significantly disturb the confining properties of the host rock for the Mol site.

4.1.2. Production of CO2 from the Boom Clay kerogen under thermal stress

Deniau et al. (2005a) and Lorant et al. (2008) simulated the production of CO2 from the Boom Clay kerogen under low thermal stress and determined kinetic parameters allowing to simulate CO2 production upon disposal of high activity nuclear waste.

It is well-documented that type II kerogen from immature sedimentary rocks (i) is prone to generate relatively large amounts of liquid and gaseous compounds under thermal stress and (ii) heteroatomic compounds, including CO2, are first generated due to the relatively low thermal stability of O-containing bonds. Thermal cracking of kerogen is the sum of a large number of elementary reactions and, thus, only apparent kinetic parameters can be calculated from experimental data (DENIAU et al., 2005a). The reactions of this kinetic scheme are assumed to obey first-order kinetics and the Arrhenius law relating the rate constant k to temperature:

117

()− k = Ae E RT equation (4.1) where A is the frequency factor, E is the activation energy, R is the gas constant and T the absolute temperature.

The production of gaseous compounds under moderate thermal stress, with focus on CO2, may significantly influence physico-chemical parameters (pH, Eh, ionic strength) of the geological barrier.

The main purpose of the study by Deniau et al. (2005a) is to (i) determine the nature of the gaseous compounds generated from the Boom Clay kerogen under moderate thermal stress; (ii) assess the kinetic parameters (activation energy distribution and frequency factor) related 28 to the early production of CO2; (iii) simulate this production of "labile" CO2 (i.e., of CO2 generated from thermolabile moieties of the kerogen) for the range of thermal stress considered for long-term disposal of high activity waste. To this end, the kerogen was submitted to closed system isothermal pyrolysis in Au tubes for various temperature/time conditions. The kerogen and the residues recovered after heating in closed tubes were examined by open, non-isothermal Rock-Eval pyrolysis and elemental analysis. The gas fractions generated during the closed experiments were quantified and analysed by GC. The studied core sample (20 cm in length, 10 cm in diameter), collected in May 2001, was drilled dry from the HADES underground research facility in Mol in the Putte Member of the formation at a depth of 220.9-221.1 m.

Rock-Eval analysis of the crude rock sample showed a TOC content of 2.95 wt.%. The isolated kerogen concentrate showed relatively large ash content (~35 %) composed mainly of pyrite and minor amounts of titanium oxide (rutile). Elemental analysis data for the kerogen showed an H to C atomic ratio of ca. 1.35 (in the range typically observed for low maturity type II kerogens) and an O to C ratio of 0.27. The Tmax value was rather low (404 °C), reflecting the immaturity of the kerogen. HI and OI values of 332 mg HC/g TOC and 92 mg CO2/g TOC, respectively, were obtained. The OI is relatively high compared to usual values for type II kerogens (≤ 50 mg CO2/g TOC), reflecting a high content of O-containing functional groups. In addition, CO2 production was monitored starting from a pyrolysis temperature of 150°C up to 700°C (figure 4.1). The obtained curve is non-symmetrical and extends over a wide temperature range. It exhibits a steep initial slope from 150°C and shows a substantial shoulder in the low temperature part of the curve with a maximum around 165°C. The large width and irregular shape reflect the presence of a variety of O-containing functional groups exhibiting a wide range of thermal stability. Closed pyrolyses were performed to derive the kinetic parameters related to the formation of "early" CO2 from thermolabile (T < 200°C in figure 4.1) groups in the Boom Clay kerogen.

28 The CO2 thus generated is termed "early CO2", because it is formed far before the cracking of the macromolecular structure of the kerogen and, hence, before the onset of the genesis of hydrocarbon compounds.

118

Figure 4.1: CO2 production versus temperature during Rock-Eval pyrolysis of the Boom Clay kerogen (DENIAU et al., 2005a)

Isothermal pyrolyses in closed tubes were carried out for various temperature/time conditions, with temperatures ranging from 150 to 375 °C and time intervals from 1 to 144 h. Rock-Eval data obtained for residual kerogen after heating show that a sharp OI decrease takes place with increasing severity of thermal stress applied. In contrast, the HI index remains fairly constant. Only when the OI decreases to values < 50 mg CO2/g TOC, HI values start to go down. For the highest thermal stresses applied, the OI becomes negligible (< 5 mg CO2/g TOC), while a rather large HI value (~100 mg HC/g TOC) is retained. Similar features were observed when considering the evolution of the H/C and O/C atomic ratios obtained by elemental analysis.

Analysis of the nature and abundance of the gaseous compounds generated shows that the total amount of gas systematically increases with the severity of the thermal stress applied. The gaseous fraction corresponds to ca. 1 up to ca. 15 wt.% of the mass of the initial kerogen. H2S is the only S-containing gas generated and is observed already for the mildest thermal stress applied, in agreement with the low thermal stability of S-containing groups. At its maximum, H2S production corresponds to ca. 1.7 wt.% of the mass of kerogen, which is a low figure reflecting the low content of organic S in the Boom Clay kerogen. Generated hydrocarbons (dominated by CH4) are always extremely low (≤ 0.1 % of the mass of total C in unheated kerogen) and is even absent in the mildest test conditions. This indicates that no significant cleavage of C-C bonds takes place and that the global macromolecular structure of the kerogen is not significantly affected. CO2 is always, by far, the main gaseous component formed and corresponds to up to ca. 10 wt.% of the total mass of the initial kerogen. Furthermore, even for the mildest stress applied, a significant amount of CO2 is released.

119

Figure 4.2: Experimental data and calculated cumulative generation curves for CO2 (mg/g kerogen) during closed pyrolysis of the Boom Clay kerogen at 150, 175 and 200°C (DENIAU et al., 2005a)

The cumulative production of "labile" CO2 from the Boom Clay kerogen at 150, 175 and 200 °C is reported in figure 4.2. The calculated curves, obtained through optimisation to first- order kinetics, are also presented. Based on these curves, it appears that the maximum amount of "labile" CO2 that can be generated by the Boom Clay kerogen is about 55 mg/g of kerogen. Taking into account the O content of the kerogen, it appears that the O contained in the "labile" CO2 represents around 20 % of the total O of the kerogen. Moreover, it appears that a part corresponds to a very early "flash" production, representing ca. 8 mg/g C. The obtained kinetic parameters suggest that a single type of reaction (or similar reactions) is implicated in the bulk of the early CO2 production from the Boom Clay kerogen.

The kinetic parameters obtained for "labile" CO2 production through closed pyrolysis were used for calculating CO2 genesis associated with highly radioactive waste disposal. According to these calculations, a large part of the "labile" CO2 potential of the Boom Clay kerogen would be rapidly released even at 80 °C. At this temperature, within 200 a, the released CO2 (ca. 27 mg/g of kerogen) would account for about half of the total potential and at 100 °C almost all of the "labile" CO2 would be freed in the same period. This generation of CO2 might be able to induce a chemical perturbation on Boom Clay pore water, resulting from the combination of 3 processes: CO2 production itself, the dissolution of carbonate minerals by H2CO3 and the diffusion out of the heated zone of all species whose concentration is increased. For a given disposal configuration, the actual impact depends on the relative kinetics of these 3 processes.

However, the kinetic model established by Deniau et al. (2005a) suffers from a lack of robustness for two reasons (LORANT et al., 2008). Firstly, the kinetic model was established assuming a set of 21 potential competitive reactions, all having the same frequency factor. The relatively low amount of experimental data made the optimisation process, from a mathematical point of view, therefore under determined. Secondly, since the "flash" production of CO2 seemed to be completed even at the lowest experimental thermal conditions, the data did not allow the kinetics of this "flash" reaction to be constrained at lower temperatures. As a result, the model predicted that some CO2 generation would occur at the natural temperature of the Boom Clay (ca. 16 °C). In fact, the kerogen is in a steady state

120

in the formation and, consequently, the modelled CO2 generation at present formation temperature was unrealistic. Therefore, Lorant et al. (2008) developed a new kinetic model and validated this model by additional "low temperature" (80 °C for 1, 3 and 6 months) closed pyrolysis experiments. This model is based on a kinetic scheme involving only 3 competitive (i.e. parallel) reactions with first-order rate laws. The rate constants were assumed to be temperature dependent following the Arrhenius law (equation (4.1)). The kerogen used in the experiments originated from the same batch as used by Deniau et al. (2005a). The experimental data and calculated results with the adjusted new kinetic model are shown in figure 4.3. The total potential determined for early CO2 production amounts to ca. 49 mg/g of organic carbon and agrees with the observations from Deniau et al. (2005a).

Figure 4.3: Result of the calibration of the new kinetic model: computed (solid lines) versus measured (dots) CO2-amounts (LORANT et al., 2008)

The model also correctly predicted the CO2 yields recovered during the 80 °C pyrolysis experiments. It is worth remarking that the amount of gas recovered at 80 °C/6 months is similar to the yield of the "flash" of CO2 observed at higher temperature (i.e., close to 8 mg/g C). By comparing reported activation parameters for decarboxylation reactions with the kinetic parameters found during model calibration, Lorant et al. (2008) concluded that the initial "flash" of CO2 resulted from the decarboxylation of "activated" carboxyl functions in the kerogen. This "activation" points to the presence of a neighbouring group, such as a second carboxylic acid group in oxalic acid or an α amide group in oxamic and oxanilic acids. The "flash" CO2 reaction kinetics would dominate CO2 formation following nuclear waste disposal within the Boom Clay formation.

The kinetic model was extrapolated to thermal conditions in the Boom Clay formation as predicted by Sillen and Marivoet (2007). The results showed that CO2 release from the kerogen would start about one year after disposal, would continue for ~100 years in the neighbourhood of the waste galleries, and last at least 1000 years towards the limits of the clay layer. The cumulated mass of CO2 formed within the Boom Clay 1000 years after nuclear waste disposal might reach ca. 3 tons per meter of gallery (figure 4.4).

121

Figure 4.4: Predicted CO2 yields in the Boom Clay formation, for different distances from the waste gallery (chart A), and corresponding absolute mass of CO2 formed within the clay layer for 1 m of waste gallery (chart B) (LORANT et al., 2008)

The kinetic model proposed by Lorant et al. (2008) was derived from anhydrous closed system experiments using purified kerogen, free of clay mineral matrix. These laboratory conditions are therefore not representative of the in situ characteristics of the geological barrier. Especially the role of possible interactions between kerogen, water and clay minerals on the formation of CO2 proves difficult to predict. Indeed, clay minerals may exhibit catalytic activity while water may significantly promote the oxidation of organic compounds and the formation of CO2. The model developed by Lorant et al. (2008) therefore provides minimum rates and amounts of CO2 that can be released by the organic matter in the Boom Clay.

In order to get a more comprehensive picture of the total CO2 production under more representative conditions, Michels et al. (2010) performed a dedicated study on both isolated kerogen29 and (whole) Boom Clay30. The samples were loaded in gold capsules under Ar atmosphere. Pyrolysis was performed at temperatures of 150, 180 and 200 °C for durations of 3 to 72 h under 220 b pressure (water as liquid phase). The CO2 production curves obtained for the pyrolysis of the isolated kerogen (figure 4.5) are consistent to those obtained by Lorant et al. (2008) although the absolute quantities are inferior. This is most likely related to the differences in initial oxygen content of the samples used in each study31 and highlights the influence of kerogen type and composition on the CO2 yield (MICHELS et al., 2010).

In contrast, the amounts of CO2 produced as well as the shape of the production curves obtained with Boom Clay are very different from that of isolated kerogen (figure 4.6). CO2 production curves show first a strong increase, before reaching a maximum. The, CO2 amounts decrease as a function of pyrolysis time. It is currently assumed that this decrease is related to the buffering effect of carbonate inside the sample. The maximum quantity of CO2

29 The kerogen was isolated from a clay sample taken during the excavation of the PRACLAY gallery

30 The Boom Clay sample originated from R13U-07, 7m60-7m775 (HADES borehole 2003/9). The TOC content is 1.11 wt.%. The water content was "as received" (i.e., about 18 wt.%).

31 The kerogen from the PRACLAY sample contained 55.8 wt.% pyrite, 31.9 wT.% C, 2.5 wt.% H and 9.8 wt.% O, while the sample from Lorant et al. (2008) contained 35.8 wt.% pyrite, 43.8 wt.% C, 4.6 wt.% H and 15.8 wt.% O (Michels et al., 2010)

122

obtained is ~119 mg/g TOC for Boom Clay, compared to ~48 mg/g TOC for Lorant et al. (2008) and ~18 mg/g TOC for the PRACLAY sample (MICHELS et al., 2010).

Figure 4.5: CO2 quantities (in mg of CO2 formed per g of TOC) measured from the pyrolysis of isolated kerogen from Boom Clay (PRACLAY sample) (MICHELS et al., 2010)

Figure 4.6: CO2 quantities (in mg of CO2 formed per g of TOC) measured from the pyrolysis of a whole Boom Clay sample (MICHELS et al., 2010)

Another investigation on the potential of CO2 formation by Boom Clay is currently ongoing at SCK•CEN. The experimental set-up consists of Boom Clay batch suspensions (containing higher liquid-to-solid ratios compared to the in situ conditions) heated up to 80°C for longer time periods (years). The results of these experiments will serve to confirm the pyrolysis experiments described in this section.

4.1.3. Comparison to other geological formations.

Landais and Elie (1999) showed that the Callovo-Oxfordian kerogen at Bure is immature and generates low quantities of liquid effluents under a thermal stress. In the case of the Toarcian

123

Clay at Tournemire, if previous analyses showed a high degree of maturity for the organic matter of the black shales (lower Toarcian), they did not give clear-cut indications on the maturity of the organic matter of the argillite (upper Toarcian) (DE WINDT et al., 1999).

The production of hydrocarbonaceous compounds from the Boom Clay kerogen under mild thermal stress is quite low (DENIAU et al., 2005b). An even lower production would occur for the other two sites, especially Bure, due to weaker values for HI and/or TOC (DENIAU et al., 2008). This feature is in agreement with the study reported by Landais and Elie (1999), which shows that the organic matter of the Callovo-Oxfordian clay in Bure submitted to a thermal stress, produces a very limited quantity of soluble compounds.

The potential for the production of oxygen-containing compounds from the three kerogens is reflected by the production of CO2 upon Rock-Eval pyrolysis and the derived oxygen indexes. The curves of CO2 production obtained by Rock-Eval pyrolyses are shown in figure 4.7 (DENIAU et al., 2008). The large width and irregular shape of these curves reflect the presence of a variety of O-containing functional groups exhibiting different thermal stabilities in the three kerogens. The Boom Clay kerogen releases a higher quantity of CO2 when compared to the other two kerogens and especially to the Toarcian clay, whatever the considered temperature. These results are in agreement with the O/C atomic ratios and the differences in the relative intensity of the band corresponding to C=O functions in the FTIR spectra.

Although the potential production of "early CO2" is weaker (which is in coherence with their higher maturity), the Callovo-Oxfordian and Toarcian kerogens also contain some thermolabile CO2-generating groups. These thermolabile groups should also release "early CO2", although in lower amounts compared to Boom Clay kerogen, under the kind of thermal stress presently conservatively estimated in Belgium for HLW (80 °C for 1000 years) (DENIAU et al., 2008).

Figure 4.7: CO2 production versus temperature during Rock-Eval pyrolysis of the kerogens from the Boom Clay, Bure and Tournemire URL. A flash release of CO2 is observed, corresponding to a shoulder on the curve at ~220 °C (DENIAU et al., 2008)

124

4.2. Impact of alkaline perturbation

Cement will be present in the foreseen nuclear waste repository, mainly as part of engineered structures. In case of water intrusion, cement dissolution will lead to high pH values. Such high pH values may, among others, result in the mobilisation of dissolved organic matter by means of alkaline attack on the kerogen fraction. It is therefore expected that an alkaline plume with a high pH should wash NOM off the clay. Percolation experiments with young 32 cement water (YCW) have indeed shown such a phenomenon (WANG et al., 2004). On the other hand, similar percolation experiments with evolved cement water (ECW)33 showed either no, or a much slower breakthrough (WANG et al., 2004) (figure 4.8).

Figure 4.8: The effect of alkaline plumes on the percolated concentration of NOM quantified by UV absorbance or DOC. Top figures are results of the module which has been percolated with EG/BS water for 1 year before cement waters. Bottom figures are for modules for which no EG/BS water was percolated before cement waters (WANG et al., 2004)

Similar investigations on the impact of alkaline solutions (pH 13.2) on 4 selected clay samples from the Callovo-Oxforidan formation showed substantial generation of hydrophilic dissolved organic carbon (243-355 mg DOC/L) from hydrophobic clay organic matter (CLARET et al., 2002; CLARET et al., 2003). Claret et al. (2003) characterised the generated dissolved organic carbon by means of spectroscopic methods (UV/Vis absorption, IR and

32 Young cement water composition: K+ 5500 mg/L, Na+ 1490 mg/L, pH 13.2, TIC 176 mg/L, TOC 113 mg/L

33 Evolved cement water composition: Na+ 550 mg/L, Ca2+ 409 mg/L, pH 12.5, TIC 3 mg/L, TOC 53 mg/L

125

fluorescence properties). It was shown that the DOM has the typical properties of humic and fulvic acids. Consequently, alkaline perturbation is an additional potential source for mobile DOM.

126

4.3. Impact of ionic strength perturbation

Variations in ionic strength represent the most common chemical disturbance resulting in the release or in situ mobilisation of colloidal particles. However, the detailed mechanisms and appropriate kinetic processes of the release phenomena within natural systems are still poorly understood (GROLIMUND et al., 2007).

Among others, a change in ionic strength would affect the dissolved organic carbon concentration and size. Indeed, humic substances and organic material behave like weak-acid polyelectrolytes. Negative charges arise from the ionisation of acidic functional groups like carboxylic COOH, phenolic OH, enolic OH, imide (=NH), and possibly other groups. When studying the influence of neutral salt on the titration properties of humic acid it was found that ionisation constants for acidic functional groups increased with decreasing ionic strength (STEVENSON, 1982). This is expected because neutral salt will partly "mask" dissociated functional groups. The remaining protons can therefore be more easily removed from the macromolecule, resulting in a decrease in the dissociation constant (pKa) compared to titrations in which the ionic strength is lower.

On the other hand, slopes of the curves obtained by application of the so-called extended Henderson-Hasselbalch equation:

pH = pKa – n log[(1-α)/α] equation (4.2) were independent of ionic strength. Accordingly, it was concluded that the slope of the titration curve was not caused by interactions of similar groups on the same molecule, but rather to a distribution of pKa values within and between molecules. Because of this observed ionic strength effect, the titration properties of DOM deviate from the theory that they behave like typical "polyelectrolytes". The adopted view is as follows (STEVENSON, 1982): in the presence of electrolyte, the humus macromolecule changes from a hydrophilic to a hydrophobic-type colloid. This was believed to occur in the following way. In aqueous solution acidic functional groups of the humic colloid are more or less dissociated and the polymer assumes a stretched configuration because of mutual repulsion of negatively charged functional groups (e.g., COO-). When electrolyte is added, the cation is attracted to negative groups, thus causing a reduction in intermolecular coulombic repulsion in the polymer chain. This in turn favours coiling of the chain and causes a reduction in the amount of hydration water held by the colloid.

The extent to which flocculation occurs depends upon such factors as pH and the nature of the electrolyte (GROLIMUND and BORKOVEC, 2006; GROLIMUND et al., 2001). Colloid concentration is also of importance, particularly for monovalent cations. Under proper pH conditions, trivalent cations, and to some extent divalent cations, are effective in precipitating DOM from very dilute solutions (in agreement with the Schultz-Hardy rule34); monovalent

34 The Shultz-Hardy rule states that the destabilization of a colloid by an indifferent electrolyte is brought about by ions of opposite charge to that of the colloid, and that coagulation effectiveness increases with charge. Thus, monovalent, divalent, and trivalent species should be effective approximately in the ratio of 1 : 100 : 1000. In most practical systems, the Shultz-Hardy rule is "violated" because the electrolytes are not different. Inorganic compounds added to water react with solution electrolytes and form complexes and precipitates. Hence the coagulation power for monovalent, divalent, and trivalent species is taken as 1:60:700.

127

cations are generally effective only at relatively high particle concentrations. For both monovalent and divalent cations, coagulating power increases with decreasing radius of the hydrated ion, the order of increasing effectiveness being K+ > Na+ > Li+ and Ca2+ > Mg2+. It is remarked that complete precipitation with salts of monovalent cations is seldom achieved. According to one theory of coagulation of colloids, the critical concentration of ions required to initiate flocculation is inversely proportional to the sixth power of their valency. Thus: Monovalent:divalent:trivalent = (1/1)6:(1/2)6:(1/3)6 = 1 : 0.016 : 0.0014

Because of their high molecular weights, humic acids are more easily coagulated than fulvic acids.

In Moors (2005) Boom Clay cores were percolated with electrolytes of different and decreasing ionic strength. It is known from literature that such a decrease in ionic strength is usually accompanied by a release or remobilisation of colloidal material (GROLIMUND et al., 2007; GROLIMUND and BORKOVEC, 2001). The effect on the dissolved organic carbon concentration was followed by monitoring the UV-absorbance at 280 nm of the outlet solution (figure 4.9). The increase of UV-signal, each time the ionic strength is lowered, might therefore be explained by a re-suspension or release process of organic material. However, during some periods noisy signals are measured (figure 4.9). This noise appears to be correlated with not only the ionic strength changes but also with variations in advection rate, temperature and resident time of the feed water inside the Boom Clay. The latter is clearly demonstrated by the strong UV signal peak observed at the same moment, in all three Boom Clay cores, after the period of no percolation. It is not clear what process has caused these peaks, as some experimental details from this period are lacking. It has to be noted that the UV-signal, and consequently the DOC concentration, was almost never equal to the UV measurement of the percolation feed water.

Figure 4.9: Evolution of UV-signal at 280 nm as function of volume of percolated solution (MOORS, 2005)

As a small side effect (compared to the changes in size, flocculation or release of colloids), a change in ionic strength would also affect the diffusion accessible porosity. The effect of ionic strength on the diffusion accessible porosity of anions is studied in De Cannière et al.

128

(1996), Moors (2005) and Aertsens et al. (2009). It was shown that the value of ηR for iodide (I-) increased from an average of 0.17 using RBCW as background electrolyte to an average of 0.28 when percolating Boom Clay cores with 1 mol/L NaCl. This effect can be explained by the decrease in anion exclusion due to a decrease of the double layer thickness with increasing ionic strength. A similar effect would be expected also for dissolved organic carbon.

129

4.4 Impact of oxidation

4.4.1 Introduction

The reducing nature of Boom Clay natural organic matter (see section 3.4) makes it particularly susceptible to an oxidative perturbation. Such a perturbation is acknowledged as a side effect of gallery excavation and operation (ventilation) (DE CRAEN et al., 2011; DE CRAEN et al., 2008). Since oxidation may affect the clay barrier geochemistry and the engineered barrier systems, it has been subject to numerous investigations in the past, mainly focussing on pore water chemistry and mineralogy effects (DE CRAEN et al., 2008; DE CRAEN et al., 2004a). The main effect of oxidation was evidenced in the pore water composition, with high sulphate concentrations observed through either pore water sampling from piezometers or leaching of clay material in the first meter surrounding the gallery. The effect of oxidation on the mineralogy is less pronounced and less extended. Conceptually, oxidation is assumed to be related to fracturing. After sealing of the fractures, oxidation products are redistributed in the clay matrix by a combined diffusion–advection transport. In addition, a continuous oxygen diffusion from the gallery into the clay is maintained. However, the latter process seems to have only a minor effect. In the case of the Boom Clay, the extent of the oxidised zone equals the extent of the fractured zone and is about 1 m, even after 20 years of ventilation (DE CRAEN et al., 2011; DE CRAEN et al., 2008).

In addition to the (mainly inorganic) processes described in aforementioned publications, oxidation might also inadvertently affect the natural organic matter of the Boom Clay. Both the elemental and molecular composition of the organic matter and the relative proportions of the different organic matter fractions (kerogen, bitumen, dissolved organic matter) may change as a function of oxygen exposure. These changes were investigated by Blanchart during her PhD (BLANCHART, 2011). The results from these studies are briefly described in the following sections.

4.4.2 Impact of oxidation on the compositions of kerogen and bitumen

In order to study the long-term impact of air-clay contact onto the organic matter contained in the Boom Clay, two types of samples were studied: 1) a reference "natural" series of clay samples (taken from an observation window in the Test Drift) having been in contact with the atmosphere of the HADES gallery for increasing times, and 2) clay samples (taken during the excavation of the Praclay gallery) submitted to "artificial" oxidation in a ventilated oven at 80°C up to 9 months. The evolution of geochemical data of the two series was compared using Rock-Eval pyrolysis, bitumen extraction by soxhlet, GC-MS and GPC-HPLC analysis. These two series were chosen 1) to obtain a comprehensive geochemical investigation on kerogen as well as organic extracts allowing to identify major oxidation stages relevant to the underground laboratory conditions, and 2) to compare the results from both sets of samples in order to confirm the validity of the experimental simulation of oxidation.

The organic matter of the unaltered samples in both series is mainly of type III with some contribution of type II. TOC from natural as well as the artificial series is low, ranging from 1.78 to 0.85 wt.% and decreasing with oxidation. The oxygen index (OI) increases with

130

alteration duration while the hydrogen index (HI) decreases. Moreover, both series seem to follow the same trend (figure 4.10), indicating hydrogen loss and oxygen uptake of the kerogen. The results from other analyses equally show that the evolution of geochemical markers for the two series are very similar. A new, aliphatic enriched bitumen is released with an average molecular weight lower than the initial. The distribution of molecular markers, such as n-alkanes, 2-alkanols and 2-alkanones, is shifted to lower molecular weights.

Figure 4.10: Evolution of normalised values for OI and HI during oxidation for natural and artificial series of Boom Clay (BLANCHART, 2011)

In order to further decipher the role of kerogen, bitumen and clay in the oxidation of the organic matter of Boom Clay, Blanchart (2011) also compared the artificial oxidation of kerogen isolated from Boom Clay to that of the initial as well as the bitumened Boom Clay. The molecular composition of the initial kerogen shows a significant decrease of compounds chain length and molecular weight with increasing oxidation. As a consequence, the oxidation induces first the release of mainly aliphatic compounds as well as alkanoic acids (present as genuine constituents of the initial kerogen), then the enrichment of the residual kerogen in aromatic constituents, which are partly oxidised to benzoic acids. The comparison of the oxidation experiments of isolated kerogen and pre-extracted clay evidences the role of clay minerals: the n-alkanes released from the kerogen into the rock matrix undergo oxidation catalysed by clay minerals. Reaction products of the n-alkanes lead to the formation of new oxygen bearing compounds, following several reaction steps.

All these evolutions are characteristic of the air oxidation process whose impact on the organic matter can be described as a combination of three major phenomena: kerogen decomposition and oxidation, bitumen release and subsequent oxidation catalysed by clay minerals.

131

4.4.3 Impact of oxidation on the fraction of dissolved organic matter

The perturbing effect of oxidation on the dissolved organic matter concentrations measured in piezometer samples was suspected after excavation and instalment of the MORPHEUS piezometer in the HADES underground research facility. As the MORPHEUS piezometer was not drilled under anoxic conditions, oxidation cannot be ruled out and might have an effect on the observed pore water chemistry. Indeed, Van Geet (VAN GEET, 2004; VAN GEET et al., 2003) observed a strong decrease in DOC in the different filters during the first years. Generally, the range of DOC concentrations lowered from 175 – 425 mgC/l to 50 – 125 mgC/l after ~3 years' time (figure 4.11). The decrease seems to be highest in the first year after instalment and then tends towards a more or less constant value. Superposed on this general trend, some erratic variation is noticed as well (possibly correlated to on- and off- switching of valves to extract pore water). The general decreasing trend is not immediately noticeable for filter F8 (located in the "double band"). However, the flow rate in this filter is much higher and the first waters of this filter were not collected. Therefore, the decrease was probably missed. Van Geet (2004) also observed that this decrease in DOC coincided with a general decrease in sulphate concentration in the collected water samples. This observation feeds the assumption that the observed trend is due to an artefact caused by oxidation.

Figure 4.11: Evolution of the DOC in the different filters of the MORPHEUS piezometer as a function of time (VAN GEET, 2004)

In order to minimise the oxidation effect during instalment of a piezometer, a new drilling procedure was tested and applied on two piezometers in the framework of the EC-project NF- PRO35. The N2TD and N2CG piezometers, located respectively in the Test Drift and the Connecting Gallery, were drilled and installed under continuous flushing with nitrogen gas (DE CRAEN et al., 2011). Each piezometer contained 6 filter screens, located at various distances from the gallery wall. From sampling of the pore water, it was observed that the

35 NF-PRO was a four-year Integrated Project (2004-2007) under the Sixth Framework Programme of the European Commission. In particular, NF-PRO investigated key processes affecting the long-term barrier performance of the near-field system, which is an essential component of the geological disposal system ensuring the long-term safety of disposal.

132

water collected from the first filters, located closest to the gallery wall (at ~0.40 and 0.70 m), were still affected by oxidation (evidenced from the higher sulphate and thiosulphate concentration), but that the other water samples collected from filters located further away (> 1.20 m) were not. Therefore it was concluded that the zone disturbed by oxidation was limited to about one meter outside the gallery wall (DE CRAEN et al., 2008). The concentrations of DOC measured in the piezometer water samples also showed high values collected initially in the first filters closest to the gallery wall, i.e. in the oxidised zone (figure 4.12). After one year (and in all other filters), the concentrations dropped towards 'normal' concentrations, between 100 and 200 mgC/l (DE CRAEN et al., 2011).

.Figure 4.12: Dissolved organic carbon concentrations in the pore water as a function of distance from the concrete/clay interface for Test Drift and Connecting Gallery (dotted line: initially after piezometer installation; full line: one year after piezometer installation) (DE CRAEN et al., 2011)

In order to test the hypothesis that oxidation would lead to an increase in dissolved organic carbon concentration, Blanchart (2011) studied the effect of controlled air oxidation on the amount and composition of dissolved organic molecules, and tested several analytical techniques allowing a fast screening of piezometer water samples on oxidation effects. It was observed that with increasing oxidation time, the amount of organic carbon (with respect to the TOC) increased from -7 to ~21 %, and then decreased back to ~17 %. The "natural" oxidation series exhibited lower amounts of water-extractable organic carbon, possibly due to reworking by aerobic bacteria. APPI-Qtof and HPLC-SEC shows that the size distribution of DOM decreases (slightly) towards lower masses upon oxidation. Also, the amount and proportion of oxygenated functional groups seems to increase with oxidation time. Notably, a strong increase in polyacidic (dicarboxylic) functional groups is noticed with Py-GCMS.

The most interesting tool to study the effect of oxidation on the dissolved organic matter characteristics was found to be 3D-Fluorescence spectroscopy. Indeed, during oxidation, a new fluorescence peak was observed with wavelengths close to peak characteristic for the fulvic acid fraction. The 3D-Fluorescence signals were further analysed by defining several new indices, I1 and I2, which represented the different peaks in the spectra (figure 4.13).

133

Figure 4.13: Formulas for the calculation of the indices I1 and I2 (BLANCHART, 2011)

st Index I1 describes the relative proportion (in intensity) of the (so-called) 1 peak (Massif n°1, the peak corresponding to the non-oxidised background) with respect to the sum of the two peaks taken together (Massif n°1 and Massif n°2). Index I2 describes the relative proportion of the 2nd peak (appearing in oxidised samples) with respect to the sum of both peaks. With oxidation, the I2 index increases. The nominal values of these indices allow to assess the degree of oxidation of the investigated DOM samples (figure 4.14).

In a complementary experiment, Blanchart (2011) also investigated the effect of oxidation on the isolated kerogen fraction. The dissolved organic matter generated by kerogen during oxidation is mainly composed of oxygenated aromatic compounds. The chemical properties (amount of functional groups) change with increasing oxidation: the DOM becomes more soluble over the entire pH range, resembling a "fulvic acid"-like behaviour. The molecular weight distribution is not significantly affected. In the I1 versus I2 diagram, the DOM from natural and artificial oxidation clay samples follows the trend towards the DOM extracted from oxidised kerogen isolate (figure 4.14).

Finally, by comparing the I1 and I2 indices from actual piezometer water samples with the oxidised samples (figure 4.14), Blanchart (2011) concluded that the water samples (at the present conditions) are not affected by oxidation.

134

Figure 4.14: Position of different samples of natural and artificial oxidation series in a diagram of 3D-Fluoresence indices I2 versus I1 (BLANCHART, 2011)

135

4.5 Impact of compaction

The degree of compaction has a profound influence on the diffusion accessible porosity and the pore diffusion coefficient because it will determine the constrictivity, the pore size distribution and pore throat diameters. It is furthermore expected that a decrease in HTO- accessible porosity due to increasing compaction will result in a relatively bigger decrease in anion-accessible porosity. This is due to the fact that anions, and especially negatively- charged colloidal species such as humic acids, will especially enter larger pores containing free water and will be excluded from smaller pores where double layers of soil minerals may overlap. An increase in compaction likely results in a relatively larger decrease of the number of larger pores, while the smaller pores will be almost unaffected.

Henrion et al. (1991) studied the influence of compaction on the diffusion of non-sorbed species and especially humic-like molecules in Boom Clay (consolidation pressure up to 6.86 MPa). HTO and I- were used as reference molecules to estimate parameters such as tortuosity. Among the molecules used to simulate small humic molecules, labelled sucrose, lactose, triiodothyronine and phenylalanine were used. Triiodothyronine (TIT), having aromatic rings and a carboxylic group, is structurally closer to humic acids. The same can be said of the smaller phenylalanine. Separate batch tests showed that the latter is slightly sorbed on the clay (mean Kd for 3 measurements: 1.2 L/kg), an effect which is reflected in ηR values which are higher than expected.

As a general observation, the volume fraction accessible for diffusion (table 4.1 and figure 4.15) decreases as the consolidation pressure increases. Compared to HTO however, the porosity decrease for anions is much more significant and values for η as low as 0.03-0.04 were reached. Larger molecules (such as lactose and TIT) show also a consistently lower value for ηR compared to iodide. In Henrion et al. (1991) all data could be fitted to one unique equation, a modified version of Archie’s law. However, this generalisation (data for different advective velocities, molecules, etc. brought together in a single plot) is probably not correct.

136

Figure 4.15: ηR as a function of consolidation pressure for different species on clay pastes consolidated to different pressures (HENRION et al., 1991)

Table 4.1: Data relative to type through-diffusion tests for different species on clay pastes consolidated to different pressures (HENRION et al., 1991)

137

-1 Figure 4.16: Dapp (in cm²·s ) as a function of consolidation pressure for different species on clay pastes consolidated to different pressures (HENRION et al., 1991)

Compared to the changes in diffusion accessible porosity, the decrease in apparent diffusion coefficient with increasing consolidation pressure is much less pronounced (table 4.1 and figure 4.16). Moreover, the same trend is noted for small organic molecules as for HTO. A decrease of factor 2 to maximum 3 is noted when increasing the consolidation pressure from 0.98 to 6.86 MPa.

The influence of destressing (TRANCOM-CLAY, 2000) To study the influence of destressing on the mobility of organic matter, the consolidation pressure in the oedometers is lowered from 2.7 to 1.7 MPa, after the first pulse died out. Later, the consolidation pressure has been lowered a second time to 0.8 MPa and the behaviour of the clay core was monitored by means of the displacement. Destressing the consolidation pressure from 2.7 MPa to 1.7 MPa did not influence the migration behaviour of organic matter in clay, as the elution profiles were exact copies. In other words, the porosity appears unchanged. This set-up with decreasing consolidation pressures may be compared to a situation where the geological layers above the disposal galleries are eroded due to glacial cycles. That no influence of destressing is observed may be explained by the hysteresis behaviour of the clay or the oedometer. Further destressing to 0.8 MPa allowed the clay core to swell and resulted in an increase of the hydraulic conductivity (figure 4.17). This increase of hydraulic conductivity is caused by an increasing porosity, which also explains the increase in UV-signal because the clay core then acts as a less performing filter.

138

Figure 4.17: The influence of destressing the consolidation pressure on the UV-signal in the percolate and the hydraulic conductivity (< 1000 Da) (TRANCOM-CLAY, 2000)

139

5. General conclusions and recommendations

The Boom Clay contains a significant amount of low-maturity organic matter. This organic matter substantially influences, among others, the pore water chemistry and radionuclide geochemistry. It is well-known and sufficiently described how the (total) organic matter varies over the stratigraphical column in the Putte, Terhagen and Belsele-Waas Members. However, the organic matter from the Boeretang Member as well as the impact of location in the sedimentary basin (transferability) remains largely understudied.

Because of its low immaturity and relatively high O/C content, the total organic matter pool is able to generate a large amount of dissolved organic matter. This DOM impacts radionuclide solubility and sorption. However, the mobility of DOM is restricted due to the confined pore structure of the compacted Boom Clay at the Mol-Dessel reference site. Up to date, it is unclear how the solid and dissolved organic matter pools are related. A relationship with the origin and composition of the solid organic matter may be put forward, but the impacts of burial history, diagenesis and pore water composition remain to be investigated. These open questions also make it difficult to predict DOM mobility on the formation scale (Mol-Dessel reference site), and to extrapolate in case of transferability and long-term evolution (undisturbed conditions) issues. However, considering current knowledge, it is not expected that DOM concentrations in the pore water will increase with respect to the present-day in situ conditions in the Mol-Dessel reference site.

The highest impact on organic matter composition and fractionation is assumed to arise from near-field perturbations, mainly related to thermal gradient, alkaline plume and oxidation. The thermal gradient associated with the disposal of vitrified high level waste and spent fuel is currently considered to result mainly in the dissociation of carboxylic functional groups, leading to a release of CO2(g). The foreseen gradient is not assumed to result in a substantial cracking of the solid organic matter pool. The alkaline plume and oxidation processes occurring in the near field are considered to have an impact mainly on the composition of the solid organic matter pool and would likely result in a local increase of the dissolved organic matter concentration. Ionic strength increases stemming from seawater intrusion are likely to neutralise acid organic functional groups, resulting in the partial flocculation of the DOM pool. However, replacement of high ionic strength water with low ionic strength electrolyte would result in a resuspension of DOM.

140

6. References

Aertsens, M., De Canniere, P., Moors, H., and Van Gompel, M., 2009. Effect of ionic strength 14 - on the transport parameters of tritiated water, iodide and H CO3 in Boom Clay. In: Burakov, B. E. and Aloy, A. S. Eds.)Scientific Basis for Nuclear Waste Management XXXIII. MRS Materials Research Society, Warrendale, Pennsylvania, USA, St. Petersburg, Russia. Aeschbacher, M., Sander, M., and Schwarzenbach, R. P., 2010. Novel Electrochemical Approach to Assess the Redox Properties of Humic Substances. Environmental Science & Technology 44, 87-93. Aeschbacher, M., Schwarzenbach, R. P., and Sander, M., 2009. Electrochemical characterization of humic substance redox properties. Geochimica Et Cosmochimica Acta 73, A14-A14. Alberts, J. J., Schindle.Je, Miller, R. W., and Nutter, D. E., 1974. Elemental mercury evolution mediated by humic acid. Science 184, 895-896. Andre, C. and Choppin, G. R., 2000. Reduction of Pu(V) by humic acid. Radiochimica Acta 88, 613-616. Bauer, M., Heitmann, T., Macalady, D. L., and Blodau, C., 2007. Electron transfer capacities and reaction kinetics of peat dissolved organic matter. Environmental Science & Technology 41, 139-145. Beaucaire, C., Pitsch, H., De Canniere, P., and Noynaert, L., 2001. Effects of irradiation and heating on Boom clay: geochemical implications. Benz, M., Schink, B., and Brune, A., 1998. Humic acid reduction by Propionibacterium freudenreichii and other fermenting bacteria. Applied and Environmental Microbiology 64, 4507-4512. Blanchart, P., 2011. Influences de l'oxydation et de la biodégradation anaerobie sur la matière organique de l'argile oligocène de Boom (Mol, Belgique): Conséquences sur la formation d'espèces organiques hydrosolubles, Institut National Polytechnique de Lorraine, Université de nancy. Blinova, O., Novikov, A., Perminova, I., Goryachenkova, T., and Haire, R., 2007. Redox interactions of Pu(V) in solutions containing different humic substances. Journal of Alloys and Compounds 444, 486-490. Bouby, M., 2007. NOM samples characterization by asymmetric flow field flow fractionation (AsFlFFF) coupled to UV-Vis spectrophotometer. INE/KIT. Bradford, S. A., Bettahar, M., Simunek, J., and van Genuchten, M. T., 2004. Straining and attachment of colloids in physically heterogeneous porous media. Vadose Zone Journal 3, 384-394. Bradford, S. A., Simunek, J., Bettahar, M., Tadassa, Y. F., van Genuchten, M. T., and Yates, S. R., 2005. Straining of colloids at textural interfaces. Water Resources Research 41. Bruggeman, C., Aertsens, M., Maes, N., and Salah, S., 2010a. Iodine retention and migration behaviour in Boom Clay - Topical Report, First full draft, SCK•CEN external report SCK•CEN-ER-119, Mol, Belgium. Bruggeman, C., Liu, D. J., and Maes, N., 2010b. Influence of Boom Clay organic matter on the adsorption of Eu(3+) by illite - geochemical modelling using the component additivity approach. Radiochimica Acta 98, 597-605. Bruggeman, C., Maes, A., and Vancluysen, J., 2007. The interaction of dissolved Boom Clay and Gorleben humic substances with selenium oxyanions (selenite and selenate). Applied Geochemistry 22, 1371-1379.

141

Bruggeman, C. and Maes, N., 2010. Uptake of uranium(VI) by pyrite under Boom Clay conditions – influence of dissolved organic carbon. Environmental Science & Technology 44, 4210-4216. Bruggeman, C., Maes, N., Christiansen, B. C., Stipp, S. L. S., Breynaert, E., Maes, A., Regenspurg, S., Malstrom, M. E., Liu, X., Grambow, B., and Schafer, T., 2012. Redox-active phases and radionuclide equilibrium valence state in subsurface environments - New insights from 6th EC FP IP FUNMIG. Applied Geochemistry 27, 404-413. Buffle, J., 1988. Complexation reactions in aquatic systems - an analytical approach. Ellis Horwood, Chichester, West Sussex, England. Choppin, G. R., 1992. The role of natural organics in radionuclide migration in natural aquifer systems. Radiochimica Acta 58-9, 113-120. Claret, F., Bauer, A., Schafer, T., Griffault, L., and Lanson, B., 2002. Experimental investigation of the interaction of clays with high-pH solutions: A case study from the Callovo-Oxfordian formation, Meuse-Haute Marne underground laboratory (France). Clays and Clay Minerals 50, 633-646. Claret, F., Schafer, T., Bauer, A., and Buckau, G., 2003. Generation of humic and fulvic acid from Callovo- Oxfordian clay under high alkaline conditions. Science of the Total Environment 317, 189-200. Cory, R. M. and McKnight, D. M., 2005. Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Environmental Science & Technology 39, 8142-8149. Czerwinski, K. and Kim, J. L., 1996. Complexation of transuranic ions by humic substances: Application of laboratory results to the natural system. In: Gray, W. J. and Triay, I. R. Eds.)Symposium on Scientific Basis for Nuclear Waste Management XX, Boston, Ma. Czerwinski, K. R., Kim, J. I., Org Econ, C., and Dev, 1994. Effects of humic substances on the migration of radionuclides: Complexation of actinides with humic substances. A commission of the European communities programNEA Workshop on Binding Models Concerning Natural Organic Substances in Performance Assessment, Bad Zurzach, Switzerland. De Cannière, P., Moors, H., Lolivier, P., De Preter, P., and Put, M., 1996. Laboratory and in situ migration experiments in the Boom clay. European Commission Nuclear Science and Technology report EUR 16927 EN, European Commission, Luxembourg, Grand Duchy of Luxembourg De Craen, M., 1998. The formation of septarian carbonate concretions in organic-rich argillaceous sediments. PhD-dissertation, K.U.Leuven. De Craen, M., Honty, M., Van Geet, M., Weetjens, E., Sillen, X., Wang, L., Jacques, D., and Martens, E., 2011. Overview of the oxidation around galleries in Boom Clay (Mol, Belgium) - status 2008. NF-PRO Final activity report for SCK•CEN. SCK•CEN, Mol, Belgium. De Craen, M., Van Geet, M., Honty, M., Weetjens, E., and Sillen, X., 2008. Extent of oxidation in Boom Clay as a result of excavation and ventilation of the HADES URF: Experimental and modelling assessments. Physics and Chemistry of the Earth 33, S350-S362. De Craen, M., Van Geet, M., Wang, L., and Put, M., 2004a. High sulphate concentrations in squeezed Boom Clay pore water: evidence of oxidation of clay cores. Physics and Chemistry of the Earth 29, 91-103. De Craen, M., Wang, L., Van Geet, M., and Moors, H., 2004b. Geochemistry of Boom Clay pore water at the Mol site. Status 2004. SCK•CEN, Mol, Belgium.

142

De Windt, L., Cabrera, J., and Boisson, J. Y., 1999. Etude géochimique des propriétés de confinement de la formation argileuse de Tournemire. Deniau, I., Behar, F., Largeau, C., De Canniere, P., Beaucaire, C., and Pitsch, H., 2005a. Determination of kinetic parameters and simulation of early CO2 production from the Boom Clay kerogen under low thermal stress. Applied Geochemistry 20, 2097-2107. Deniau, I., Derenne, S., Beaucaire, C., Pitsch, H., and Largeau, C., 2001. Morphological and chemical features of a kerogen from the underground Mol laboratory (Boom Clay Formation, Oligocene, Belgium): structure, source organisms and formation pathways. Organic Geochemistry 32, 1343-1356. Deniau, I., Derenne, S., Beaucaire, C., Pitsch, H., and Largeau, C., 2004. Occurrence and nature of thermolabile compounds in the Boom Clay kerogen (Oligocene, underground Mol Laboratory, Belgium). Organic Geochemistry 35, 91-107. Deniau, I., Derenne, S., Beaucaire, C., Pitsch, H., and Largeau, C., 2005b. Simulation of thermal stress influence on the Boom Clay kerogen (Oligocene, Belgium) in relation to long-term storage of high activity nuclear waste - I. Study of generated soluble compounds. Applied Geochemistry 20, 587-597. Deniau, I., Devol-Brown, I., Derenne, S., Behar, F., and Largeau, C., 2008. Comparison of, the bulk geochemical features and thermal reactivity of kerogens from Mol (Boom Clay), Bure (Callovo-Oxfordian argillite) and Tournemire (Toarcian shales) underground research laboratories. Science of the Total Environment 389, 475-485. Desbois, G., Urai, J. L., and De Craen, M., 2010. In-situ and direct characterization of porosity in Boom Clay (Mol site, Belgium) by using novel combination of ion beam cross-sectioning, SEM and cryogenic methods - Motivations, first results and perspectives. External Report, SCK•CEN-ER-124, SCK•CEN, Mol, Belgium. Dierckx, A., Maes, A., Henrion, P., and De Canniere, P., 1997. Potentiometric titration of humic substances extracted from Boom Clay. In: Drozd, J., Gonet, S. S., Senesi, N., and Weber, J. Eds.), The role of humic substances in the ecosystems and in environmental protection. IHSS - Polish Society of Humic Substances, Wroclaw, Poland. Dunnivant, F. M., Schwarzenbach, R. P., and Macalady, D. L., 1992. Reduction of substituted nitrobenzenes in aqueous-solutions containing natural organic-matter. Environmental Science & Technology 26, 2133-2141. Durand, B., 1980. Sedimentary organic matter and kerogen. Definition and quantitative importance of kerogen. In: Durand, B. (Ed.), Kerogen: insoluble organic matter from sedimentary rock. Editions TECHNIP. Durand, B. and Nicaise, G., 1980. Procedures for kerogen isolation. In: Durand, B. (Ed.), Kerogen: insoluble organic matter from sedimentary rock. Editions TECHNIP. Elie, M., Faure, P., Michels, R., Landais, P., and Griffault, L., 2000. Natural and laboratory oxidation of low-organic-carbon-content sediments: Comparison of chemical changes in hydrocarbons. Energy & Fuels 14, 854-861. Elie, M., Faure, P., Michels, R., Landais, P., Griffault, L., Mansuy, L., and Martinez, L., 2004. Effects of water-cement solutions on the composition of organic compounds leached from oxidized Callovo-Oxfordian argillaceous sediment. Applied Clay Science 26, 309-323. Espitalie, J., Laporte, J. L., Madec, M., Marquis, F., Leplat, P., Paulet, J., and Boutefeu, A., 1977. Méthode rapide de caractérisation des roches mères de leur potentiel pétrolier et de leur degré d'évolution. Revue de l'Institut Français due Pétrole 32, 23-42. Gaffney, J. S., Marley, N. A., and Clark, S. B., 1996. Humic and Fulvic Acids and Organic Colloidal Materials in the Environment, Humic and Fulvic Acids. American Chemical Society.

143

Geckeis, H., Rabung, T., Manh, T. N., Kim, J. I., and Beck, H. P., 2002. Humic colloid-borne natural polyvalent metal ions: Dissociation experiment. Environmental Science & Technology 36, 2946-2952. Goodman, B. A. and Cheshire, M. V., 1982. Reduction of molybdate by soil organic-matter - electron-paramagnetic-res evidence for formation of both Mo(V) and Mo(III). Nature 299, 618-620. Grolimund, D., Barmettler, K., and Borkovec, M., 2007. 1 colloid facilitated transport in natural porous media: Fundamental phenomena and modelling. Grolimund, D. and Borkovec, M., 2001. Release and transport of colloidal particles in natural porous media 1. Modeling. Water Resources Research 37, 559-570. Grolimund, D. and Borkovec, M., 2006. Release of colloidal particles in natural porous media by monovalent and divalent cations. Journal of Contaminant Hydrology 87, 155-175. Grolimund, D., Elimelech, M., and Borkovec, M., 2001. Aggregation and deposition kinetics of mobile colloidal particles in natural porous media. Colloids and Surfaces a- Physicochemical and Engineering Aspects 191, 179-188. Helburn, R. S. and Maccarthy, P., 1994. Determination of some redox properties of humic- acid by alkaline ferricyanide titration. Analytica Chimica Acta 295, 263-272. Hemes, S., Desbois, G., Urai, J., De Craen, M., and Honty, M., 2011. Comparative study on porosity in fine- and coarse-grained Boom Clay samples (Mol-Dessel reference site, Belgium). Belgian Nuclear Research Centre, Mol, Belgium. Hemes, S., Desbois, G., Urai, J., De Craen, M., and Honty, M., 2012. Variability of the morphology of the pore space in Boom Clay from BIB-SEM, FIB and MIP investigations on representative samples. Belgian Nuclear Research Centre, Mol, Belgium. Henrion, P. N., Monsecour, M., Fonteyne, A., Put, M., and Deregge, P., 1985. Migration of Radionuclides in Boom Clay, Radioactive Waste Management and Environmental Restoration. Harwood Acad Publ Gmbh. Henrion, P. N., Put, M. J., and Vangompel, M., 1991. The Influence of Compaction on the Diffusion of Non-Sorbed Species in Boom Clay. Radioactive Waste Management and Environmental Restoration 16, 1-14. Hicks, T., 2008. POPCORN Modelling of Migration Experiments to Investigate the Effects of Organic Matter on Radionuclide Transport in the Boom Clay, 0438-2 Version 1, Draft 1, Galson Sciences ltd, Oakham, Rutland, UK. Ionescu, A., Maes, N., and Mallants, D., 2008. Modelling transport of C-14-labelled Natural Organic Matter (NOM) in Boom Clay. In: Lee, W. E., Roberts, J. W., Hyatt, N. C., and Grimes, R. W. Eds.), Scientific Basis for Nuclear Waste Management Xxxi. Kim, J. I., Buckau, G., Klenze, R., Rhee, D. S., and Wimmer, H., 1991. Characterization and complexation of humic acids. Commission of the European Communities, Nuclear Science and Technology, Luxembourg. Kim, J. I. and Czerwinski, K. R., 1996. Complexation of metal ions with humic acid: Metal ion charge neutralization model. Radiochimica Acta 73, 5-10. Kim, J. I., Rhee, D. S., Buckau, G., and Morgenstern, A., 1997. Americium(III)-humate interaction in natural groundwater: Influence of purification on complexation properties. Radiochimica Acta 79, 173-181. Klapper, L., McKnight, D. M., Fulton, J. R., Blunt-Harris, E. L., Nevin, K. P., Lovley, D. R., and Hatcher, P. G., 2002. Fulvic acid oxidation state detection using fluorescence spectroscopy. Environmental Science & Technology 36, 3170-3175. Laenen, B., 1997. The geochemical signature of relative sea-level cycles recognized in the Boom Clay. PhD-dissertation, K.U.Leuven.

144

Laenen, B., 1998. The geochemical signature of relative sea-level cycles recognised in the Boom Clay. Aardkundige Mededelingen, University Press Leuven 9, 61-82. Landais, P. and Elie, M., 1999. Utilisation de la géochimie organique pour la détermination du paléoenvironnement et de la paléothermicité dans le Callovo-Oxfordien du site de l'Est de la France. Etude de l'Est due Bassin Parisien.Actes des Journées Scientifiques CNRS/Andra, Bar-le-Duc, 20 et 21 octobre 1997. EDP Sciences. Largeau, C., Derenne, S., Casadevall, E., Kadouri, A., and Sellier, N., 1986. Pyrolysis of immature torbanite and of the resistant bio-polymer (PRB A) isolated from extant alga botryococcus-braunii - mechanism of formation and structure of torbanite. Organic Geochemistry 10, 1023-1032. Lorant, F., Largeau, C., Behar, F., and De Canniere, P., 2008. Improved kinetic modeling of the early generation of CO2 from the Boom Clay kerogen. Implications for simulation of CO2 production upon disposal of high activity nuclear waste. Organic Geochemistry 39, 1294-1301. Lovley, D. R., Coates, J. D., BluntHarris, E. L., Phillips, E. J. P., and Woodward, J. C., 1996. Humic substances as electron acceptors for microbial respiration. Nature 382, 445- 448. Lovley, D. R., Fraga, J. L., Blunt-Harris, E. L., Hayes, L. A., Phillips, E. J. P., and Coates, J. D., 1998. Humic substances as a mediator for microbially catalyzed metal reduction. Acta Hydrochimica Et Hydrobiologica 26, 152-157. Lumsdon, D. G. and Fraser, A. R., 2005. Infrared spectroscopic evidence supporting heterogeneous site binding models for humic substances. Environmental Science & Technology 39, 6624-6631. MacCarthy, P., 2001. The principles of humic substances. Soil Science 166, 738-751. Maes, A., Bruggeman, C., Geraedts, K., and Vancluysen, J., 2003a. Quantification of the interaction of Tc with dissolved boom clay humic substances. Environmental Science & Technology 37, 747-753. Maes, A., Bruggeman, C., Van Geet, M., Deniau, I., and Largeau, C., 2003b. Topical report on: Methodologies developed to identify the source of mobile Boom Clay organic matter - TRANCOM II Migration Case Study: Transport of Radionuclides in a Reducing Clay Sediment. SCK•CEN, Mol, Belgium. Maes, A., Tits, J., Mermans, G., and Dierckx, A., 1992. MEASUREMENT OF THE POTENTIALLY AVAILABLE CHARGE AND THE DISSOCIATION BEHAVIOR OF HUMIC-ACID FROM COBALTIHEXAMMINE ADSORPTION. Journal of Soil Science 43, 669-677. Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S., and van Gompel, M., 2010. A consistent phenomenological model for natural organic matter linked migration of Tc(IV), Cm(III), Np(IV), Pu(III,IV) and Pa(V) in Boom Clay. Submitted for publication. Maes, N., Bruggeman, C., Govaerts, J., Martens, E., Salah, S., and Van Gompel, M., 2011. A consistent phenomenological model for natural organic matter linked migration of Tc(IV), Cm(III), Np(IV), Pu(III/IV) and Pa(V) in the Boom Clay. Physics and Chemistry of the Earth 36, 1590-1599. Maes, N., Wang, L., Hicks, T., Bennett, D., Warwick, P., Hall, T., Walker, G., and Dierckx, A., 2006. The role of natural organic matter in the migration behaviour of americium in the Boom Clay - Part I: Migration experiments. Physics and Chemistry of the Earth 31, 541-547. Marquardt, C. M., Seibert, A., Artinger, R., Denecke, M. A., Kuczewski, B., Schild, D., and Fanghanel, T., 2004. The redox behaviour of plutonium in humic rich groundwater. Radiochimica Acta 92, 617-623.

145

Martens, E., Maes, N., Weetjens, E., Van Gompel, M., and Van Ravestyn, L., 2010. Modelling of a large-scale in-situ migration experiment with (14)C-labelled natural organic matter in Boom Clay. Radiochimica Acta 98, 695-701. Matthiessen, A., 1994. Evaluating the redox capacity and the redox potential of humic acids by redox titrations. Humic Substances in the Global Environment and Implications on Human Health, 187-192. Matthiessen, A., 1998. Reduction of divalent mercury by humic substances - kinetic and quantitative aspects. Science of the Total Environment 213, 177-183. Michels, R., Faure, P., and Schumacher, A., 2010. Influence de l'eau et des minéraux sur la formation de CO2 pendant la pyrolyse de l'argile de Boom: Conséquences sur le bilan de masse et la cinétique de formation. Comparaison de la pyrolyse de l'argile de Boom et de son kérogène isolé. CREGU (Centre de Recherches sur la Géologie des Matières Premières Minérals et Energétiques), Vandoeuvre les Nancy, France. Moors, H., 2005. Topical report on the effect of the ionic strength on the diffusion accessible porosity of Boom Clay. External Report SCK•CEN-ER-02, Waste & Disposal Department, SCK•CEN, Mol, Belgium. Noynaert, L., Volckaert, G., De Canniere, P., Meynendonckx, P., Labat, S., Beaufays, R., Put, M., Aertsens, M., Fonteyne, A., and Vandervoort, F., 1998. The Cerberus project. Demonstration test to study the near-field effects of an HLW canister in argillaceous formation. European Commission. Nuclear science and technology series., Luxembourg. Noynaert, L. E., 2000. Heat and radiation effects on the near field of a HLW or spent fuel repository in a clay formation (CERBERUS project). European Commission. Nuclear science and technology series., Luxembourg. Nurmi, J. T. and Tratnyek, P. G., 2002. Electrochemical properties of natural organic matter (NOM), fractions of NOM, and model biogeochemical electron shuttles. Environmental Science & Technology 36, 617-624. Palmer, N. E., Freudenthal, J. H., and von Wandruszka, R., 2006. Reduction of arsenates by humic materials. Environmental Chemistry 3, 131-136. Pirlet, V., 2003. The investigation of the neptunium complexes formed upon interaction of high-level waste glass and Boom Clay medium. PhD, Université de Liège. Pirlet, V. and Van Iseghem, P., 2003. Neptunium speciation in humic acid-rich clay water upon interaction with radioactive waste glass samples. In: Finch, R. J. and Bullen, D. B. Eds.), Scientific Basis for Nuclear Waste Management Xxvi. Put, M. J., Dierckx, A., Aertsens, M., and De Canniere, P., 1998. Mobility of the dissolved organic matter through intact boom clay cores. Radiochimica Acta 82, 375-378. Put, M. J., Monsecour, M., and Fonteyne, A., 1992. MOBILITY OF THE DISSOLVED ORGANIC MATERIAL IN THE INTERSTITIAL BOOM CLAY WATER. Radiochimica Acta 58-9, 315-317. Randall, A., Warwick, P., Carlsen, L., and Lassen, P., 1996. Fundamental studies on the interactions of humic materials. Final report. European Commission, Nuclear Science and Technology, Luxembourg. Ratasuk, N. and Nanny, M. A., 2007. Characterization and quantification of reversible redox sites in humic substances. Environmental Science & Technology 41, 7844-7850. Reiller, P., Mercier-Bion, F., Gimenez, N., Barre, N., and Miserque, F., 2006. Iodination of humic acid samples from different origins. Radiochimica Acta 94, 739-745. Reszat, T. N. and Hendry, M. J., 2009. Migration of Colloids through Nonfractured Clay-Rich Aquitards. Environ Sci Technol 43, 5640-5646. Rice, J. A. and Maccarthy, P., 1991. Statistical evaluation of the elemental composition of humic substances. Organic Geochemistry 17, 635-648.

146

Righetto, L., Bidoglio, G., Azimonti, G., and Bellobono, I. R., 1991. Competitive actinide interactions in colloidal humic-acid mineral oxide systems. Environmental Science & Technology 25, 1913-1919. Schlegel, M. L., Reiller, P., Mercier-Bion, F., Barre, N., and Moulin, V., 2006. Molecular environment of iodine in naturally iodinated humic substances: Insight from X-ray absorption spectroscopy. Geochimica Et Cosmochimica Acta 70, 5536-5551. Schulten, H. R. and Schnitzer, M., 1997. Chemical model structures for soil organic matter and soils. Soil Science 162, 115-130. Scott, D. T., McKnight, D. M., Blunt-Harris, E. L., Kolesar, S. E., and Lovley, D. R., 1998. Quinone moieties act as electron acceptors in the reduction of humic substances by humics-reducing microorganisms. Environmental Science & Technology 32, 2984- 2989. Shcherbina, N. S., Kalmykov, S. N., Perminova, I. V., and Kovalenko, A. N., 2007a. Reduction of actinides in higher oxidation states by hydroquinone-enriched humic derivatives. Journal of Alloys and Compounds 444, 518-521. Shcherbina, N. S., Perminova, I. V., Kalmykov, S. N., Kovalenko, A. N., Haire, R. G., and Novikov, A. P., 2007b. Redox and complexation interactions of neptunium(V) with quinonoid-enriched humic derivatives. Environmental Science & Technology 41, 7010-7015. Sillen, X. and Marivoet, J., 2007. Thermal impact of a HLW repository in clay. Deep disposal of vitrified high-level waste and spent fuel. SCK•CEN, Mol, Belgium. Šimůnek, J., Šejna, M., and Van Genuchten, M. T., 2007. The hydrus software package for simulating the two- and three-dimensional movement of water, heat and multiple solutes in variably-saturated media. User manual, version 1.0. PC-Progress, Prague, Czech Republic. Šimůnek, J., Van Genuchten, M. T., and Šejna, M., 2005. The HYDRUS-1D Software package for simulating the One-dimensional movement of water, heat and multiple solutes in variably-saturated media, version 3. Skogerboe, R. K. and Wilson, S. A., 1981. Reduction of ionic species by fulvic-acid. Analytical Chemistry 53, 228-232. Sparks, D. L., 2002. Environmental soil chemistry, second edition. Academic Press. Sposito, G., 1989. The chemistry of soils. Oxford University Press, New York. Stevenson, F. J., 1982. Humus Chemistry - Genesis, Composition, Reactions. John Wiley & Sons, Inc. Struyk, Z. and Sposito, G., 2001. Redox properties of standard humic acids. Geoderma 102, 329-346. Sunda, W. G. and Kieber, D. J., 1994. Oxidation of humic substances by manganese oxides yields low-molecular-weight organic substrates. Nature 367, 62-64. Szilagyi, M., 1973. Redox properties and determination of normal potential of peat-water system. Soil Science 115, 434-437. Thang, N. M., Geckeis, H., Kim, J. I., and Beck, H. P., 2001. Application of the flow field flow fractionation (FFFF) to the characterization of aquatic humic colloids: evaluation and optimization of the method. Colloids and Surfaces a-Physicochemical and Engineering Aspects 181, 289-301. Tissot, B. P. and Welte, D. H., 1978. Petroleum Formation and Occurrence, first ed. Springer Verlag, Berlin. Trancom-Clay, 2000. Transport of radionuclides due to complexation with organic matter in clay formations (Trancom-clay). European Commission - Euratom, Brussels, Belgium. Trancom-II, 2004. Migration Case Study: Transport of radionuclides in a reducing Clay Sediment (TRANCOM-II). In: Maes, N., Wang, L., Delecaut, G., Beauwens, T., Van

147

Geet, M., Put, M., Weetjens, E., Marivoet, J., van der Lee, J., Warwick, P., Hall, A., Walker, G., Maes, A., Bruggeman, C., Bennett, D., Hicks, T., Higgo, J., and Galson, D. Eds.). Euratom - European Commission nuclear science and technology, Brussels, Belgium. Van Geet, M., 2002. Interpretation of Rock-Eval data on samples of core HADES 2001/4. Topical report in the frame of D.S. 2.82 Characterisation of Boom Clay organic matter: mobile and immobile fraction. SCK•CEN, Mol, Belgium. Van Geet, M., 2004. Characterisation of Boom Clay organic matter: mobile and immobile fraction. Interim report for the period 2001-2003. SCK•CEN, Mol, Belgium. Van Geet, M., Maes, N., and Dierckx, A., 2003. Characteristics of the Boom Clay organic matter, a review. Geological Survey of Belgium, Professional paper 2003/1, N. 298, 1- 23. Van Keer, I. and De Craen, M., 2001. Sedimentology and diagnetic evolution of the Boom Clay: State of the art. SCK•CEN, Mol, Belgium. Van Krevelen, D. W., 1961. Coal: typology, chemistry, physics, constitution. Elsevier. Vandenberghe, N., 1974. Een sedimentologische studie van de Boomse klei, Doctorale verhandeling, K.U.Leuven. Vandenberghe, N., 1978. Sedimentology of the Boom Clay (Rupelian) in Belgium. . Verhandeling Koninklijke Academie voor Wetenschappen, Letteren en Schone Kunsten van België, Klasse Wetenschappen XL147, 1-137. Vandenberghe, N. and Van Echelpoel, E., 1987. Field guide to the Rupelian stratotype. Bulletin de la Société Belge de Géologie 94, 325-337. Vandenbroucke, M., 2003. Kerogen: from types to models of chemical structure. Oil & Gas Science and Technology-Revue De L Institut Francais Du Petrole 58, 243-269. Vandenbroucke, M. and Largeau, C., 2007. Kerogen origin, evolution and structure. Organic Geochemistry 38, 719-833. Visser, S. A., 1964. Oxidation-reduction Potentials and Capillary Activities of Humic Acids. Nature 204, 581. Wang, L., De Canniere, P., Jacques, D., Moors, H., Van Gompel, M., and Aertsens, M., 2004. Experimental and modelling studies of Boom Clay exposed to an alkaline perturbation. Final Report to NIRAS/ONDRAF, DS 2.13 (2002-2003). SCK•CEN, Mol, Belgium. Warwick, P., Zhao, R., Higgo, J. J. W., Smith, B., and Williams, G. M., 1993. THE MOBILITY AND STABILITY OF IODINE-HUMIC AND IODINE-FULVIC COMPLEXES THROUGH SAND. Sci Total Environ 130, 459-465. Wilson, S. A. and Weber, J. H., 1977. Electron-spin resonance analysis of semiquinone free- radicals of aquatic and soil fulvic and humic acids. Analytical Letters 10, 75-84. Wilson, S. A. and Weber, J. H., 1979. EPR study of the production of vanadium(V) to vanadium(IV) by fulvic-acid. Chemical Geology 26, 345-354. Wittbrodt, P. R. and Palmer, C. D., 1997. Reduction of Cr(VI) by soil humic acids. European Journal of Soil Science 48, 151-162. Wouters, L. and Vandenberghe, N., 1994. Geologie van de Kempen - Een synthese. NIRAS/ONDRAF-94-11. Yu, L., Gedeon, M., Wemaere, I., Marivoet, J., and De Craen, M., 2011. Boom Clay Hydraulic Conductivity. A synthesis of 30 years of research. SCK•CEN, Mol, Belgium. Zeh, P., Kim, J. I., Marquardt, C. M., and Artinger, R., 1999. The reduction of Np(V) in groundwater rich in humic substances. Radiochimica Acta 87, 23-28.

148