Groundwater dating applied for geological disposal of radioactive waste

Transcription

1 J. Jap. Assoc. Hydrol. Sci., Vol. 44, No. 1, p.39 64, 2014 Groundwater dating applied for geological disposal of radioactive waste A review of methods employed worldwide Katsuhiro HAMA* 1 and Richard METCALFE* 2 Abstract Groundwater dating methods employed in projects to develop deep geological repositories for radioactive wastes, or to research technologies and methods that may be used when developing a repository, have been reviewed. The reviewed projects are being, or have been, undertaken in Japan, Finland, Sweden, Belgium, the UK, Germany, France, Switzerland, Canada and the U.S.A.. A wide range of actual and potential repository host rock types and hydrogeological settings have been investigated. The most commonly used dating methods are based on physical hydrogeology and measurements of δ 18 O/ 16 O, 3 H, 14 C, 36 Cl and 4 He in groundwater or pore water. Only two of the reviewed projects have used noble gas data to estimate recharge temperatures, and hence deduce the timing of recharge. A single project used 129 I. Analyses of U-series isotopes have generally not been used successfully, owing to the complex radiological and redox processes that influence U migration, which complicates the interpretation of such U-series data. None of the reviewed projects have used methods based on Tritium 3 H Helium-3 3 He Chlorofluorocarbons CFCs or 85 Kr. These methods have not been needed because they indicate the presence of water recharged within the last ca. 50 ca. 60 years, which could be identified readily at all the sites by the presence of 3 H. In the case of 81 Kr, sampling and analytical difficulties have probably prevented its use. Recently, improved analytical techniques enable smaller samples to be analyzed, but 81 Kr gives similar information to 36 Cl, which can be investigated more easily. Whatever the combinations of methods that have been used in a particular program, their results have invariably been interpreted in combination with one another and in the context of other information that is available for the investigated site. That is, a particular groundwater dating method is never applied in isolation. Key words: groundwater dating, radioactive waste disposal, site characterization I. Introduction To assess the suitability of any site that might in future be considered to host a deep geological repository for radioactive wastes, it is important to determine the rate at which the groundwater system is evolving. Generally, at a suitable site, groundwater fluxes will be sufficiently low that the functioning of engineered barrier systems will not be compromised and any radionuclides that are released from these barriers will be transported through the geosphere sufficiently slowly as to not pose a threat to safety. In order to meet this goal, it is 1 JAEA, Tono Geoscientific Research Unit 2 Quintessa Limited 39

2 HAMA and METCALFE important to determine the ages of the groundwater and pore water that are present at the site. This paper reviews methods by which this goal may be achieved and the ways in which these methods have been employed worldwide in projects aiming at the deep geological disposal of radioactive wastes. The reviewed projects include fundamental research that was not site-specific, projects aiming to characterize a rock formation that may possibly host a repository, though not at the investigated location, and characterization of actual or potential repository sites. Primarily the review concerns developments in groundwater dating since 2000, particularly in deep disposal programs for radioactive wastes that have been active since this date. Hence, detailed descriptions of all the various dating methods are not given, but instead they are summarized in Section III. For more details of these methods, readers are referred to Bethke and Johnson 2008 and Metcalfe et al In this paper, groundwater usually refers to water within relatively permeable rock, that flows naturally, or that can be induced to flow. In contrast pore water refers to water that is located predominantly within poorly connected pores and that can migrate only by diffusion. However, where used without qualification the terms groundwater dating and groundwater ages should be understood to apply equally to groundwater and pore water. Also, dating refers to obtaining absolute or relative age information from groundwaters, for which numerous techniques have been developed. In reality, any natural groundwater will be a mixture of waters originating in different sources and consequently the date or age of a groundwater must always be explained in context; generally it is necessary to interpret the past evolution of a sampled groundwater using a range of age and origin indicators. Any given groundwater sample may have several recognizable ages, including: the times since the different components of the water recharged the groundwater system; the times for which these different components have been resident in the different rock units through which the water has flowed; and the residence times of the various solutes within the water components. II. Groundwater dating in national programs 1. Descriptions of national disposal programs considered There are radioactive waste disposal programs in many countries. The main focus of the present paper is on programs concerned with the deep 200 m geological disposal of High Level Waste HLW and Spent Fuel SF The paper also covers relevant experience during other kinds of radioactive waste management program, notably deep geological disposal of Intermediate Level Waste ILW and some Low Level Waste LLW Also, covered are disposal programs for TRU waste, which is a particular classification of LLW used in some countries for waste, that has relatively high activities of transuranic elements that have long half-lives. A feature of all these projects is that the long half-lives of many of the isotopes within the considered wastes means that the safety of disposal must be assessed over a very long period, typically in the order of 1 Ma. As a basis for doing this, it is necessary to establish how the groundwater/pore water present in the rocks at actual or possible disposal sites have 40

3 Groundwater dating applied for geological disposal of radioactive waste behaved over similarly long periods of time in the past. Hence, methods for groundwater dating have been applied in all programs. The different programs are at varying stages of development, which means that they have used groundwater dating methods for different purposes. As a basis for determining the aims of groundwater dating exercises, this section therefore reviews briefly the stage of development of the various deep geological disposal programs. Table 1 summarizes the kinds of investigations undertaken during the reviewed deep geological disposal programs. Included in the table are programs that are on-going and programs that, though suspended or terminated, have been active since The table excludes past projects that have been completed or abandoned prior to 2000, except where information was published within the post-2000 timeframe of concern to this paper. For this latter reason, information about the Nirex investigations near Sellafield, UK, which effectively finished in 1997, is also included. Excluded are projects aimed at shallow disposal of radioactive wastes. Table 1 covers national programs in their entirety and presents the degree of development of each one, relative to its eventual goal of completing a repository. The table does not, therefore, distinguish different aspects of a national program that may be running in parallel. For example, undistinguished in this table are Swedish investigations in the Hard Rock Laboratory HRL an underground laboratory at Äspö, and at the proposed repository site at Forsmark. Whereas the former investigations are undertaken for research purposes, the latter are undertaken to develop a repository. The overall stage to which the program has developed, however, is one of surface-based investigation of an actual site. 2. Purposes of groundwater dating in national disposal programs The various geological disposal programs outlined in Section II. 1 have used information from groundwater dating studies in varied ways, reflecting the differing stages of the programs. Those programs that are presently generic, that is without a specific site having been identified Japan, UK, France, Belgium, Germany, Switzerland, Canada SF program are either developing and testing groundwater dating methodologies Japan carrying out such development and testing while at the same time characterizing a rock formation that may eventually host a repository France, Belgium, Germany and Switzerland or else are not actively employing groundwater dating methods UK, Canada SF Program Among those countries with generic programs, but which are also selecting a site, France, Belgium and Germany have employed groundwater dating methods within regions and/or at sites, within which a repository may eventually be sited. For example, in France, it is proposed that a repository will be sited somewhere in the region around the Bure underground research laboratory URL Therefore, groundwater dating methods that have been employed in this facility may well be directly relevant to the site that is eventually selected for a repository. Similarly in Belgium, a repository may be sited in an area near to the HADES URL at Mol and hence groundwater dating at this site may be directly relevant to the safety assessment of the eventual repository site. In Germany, it is conceivable that eventually the Gorleben site will be chosen to host a repository, in which case 41

4 HAMA and METCALFE Table 1. Summary of the status of deep 200 m geological disposal programs for radioactive wastes considered in the review. groundwater dating undertaken here will also be relevant. At research facilities, Äspö in Sweden, Mol in Belgium, Bure in France, Mont Terri in Switzerland and Horonobe and Mizunami in Japan groundwater dating methods have been employed to: help to understand past and present groundwater flow at the site of the facility, and thereby help to develop a context for 42

5 Groundwater dating applied for geological disposal of radioactive waste the overall investigation program; test methods for groundwater dating under conditions relevant to a repository; develop groundwater dating methods. At proposed repository sites that have been abandoned or suspended ILW repository at Sellafield UK, HLW/SF repositories at Gorleben, Germany and Yucca Mountain, U.S.A. groundwater dating was used to varying degrees to: build confidence that the environment of the emplaced wastes will be sufficiently stable over the time frame of a performance assessment PA ; help calibrate and/or check the validity of models for groundwater flow. Similarly at proposed repository sites with on-going investigations SF repository at Forsmark in Sweden, SF repository at Olkiluoto in Finland, ILW/LLW repository at Bruce in Canada groundwater dating methods have been used either directly or indirectly to assess whether a repository will meet safety and other performance criteria a process called performance assessment, PA However, the precise ways in which the dating information has been used is variable. Mostly, groundwater dating has been used to build confidence that the underground environment around the emplaced wastes will remain sufficiently stable during the assessment time frame typically 1 million years The approach used has been to demonstrate that over a comparable past time scale known external influences, typically those due to climate change, have not perturbed this underground environment significantly. The only operating deep repository covered by the review is the Waste Isolation Pilot Plant WIPP in New Mexico. This facility first received waste in March 1999 prior to the period covered by this review. All the groundwater dating studies that were undertaken at this site occurred during its characterization phase, which was well before the period covered by this paper. It appears that these earlier groundwater dating studies were used to support the conclusion that the repository host rock would remain stable throughout the PA period, and that there are no fast pathways for groundwater flow from the repository host rock, via the overburden, to the biosphere. Following the start of operations in 1999, groundwater at the site has been monitored to provide hydrogeological data and groundwater samples for analysis. However, the purpose of the monitoring is to ensure environmental protection and there appears to have been no further groundwater dating studies. III. Groundwater dating methods and applications 1. Descriptions of groundwater dating methods A brief overview of the main methods that have been used to date groundwaters in general, is given in Table 2. All the commonly used methods are included, together with several less widely applied methods notably Tritium 3 H Helium-3 3 He method, SF 6, 21 Ne, 85 Kr and 81 Kr, 14 C in DOC and 129 I In addition to the listed chemical tracers, others have occasionally been used to indicate groundwater ages and solute residence times. In particular, other anthropogenic chemicals or isotopes are occasionally used. For example, at Yucca Mountain, analyses for 99 Tc were undertaken since this isotope is a product of atomic bomb testing and its presence in deep rock samples would indicate the passage of water since the 43

6 HAMA and METCALFE Table 2. Methods that have been employed to date groundwater. 44

7 Groundwater dating applied for geological disposal of radioactive waste onset of the atomic weapons program in the U.S.A.. The number of isotopes and anthropogenic chemicals that could be used in this way is very large, but most will be suitable for application at only a small number of sites e.g. sites near to nuclear facilities in the case of 99 Tc For this reason, it is impractical to cover these tracers individually in this review. A large number of natural isotopes and chemical tracers have also been used to indicate the origins of groundwater and its solute load, and then this information has been used together with hydrogeological models to constrain groundwater residence times. For example, Br/Cl, 11 B/ 10 B, 34 S/ 32 S and 37 Cl/Cl have all been used to indicate the source of groundwater and/or its solute load. If the groundwater flux can be determined independently, then the time for a solute to be transported from the indicated solute source can be calculated and the residence time of the solutes within the rock can be estimated. Fluxes can be determined using hydrogeological data if transport is by advection, or concentration gradients, and experimentally determined diffusion coefficients if transport is by diffusion. In reality, such approaches are hybrid hydrogeological hydrochemical evolution techniques in the first case and hydrochemical evolution techniques in the second. Consequently, these techniques are not considered separately in this review. It is also important to recognize that ideally groundwater dating at a particular site should not rely upon a single method, since each method has advantages and disadvantages. Additionally, it is always important to interpret a particular age indicator in the context of all available information about the characteristics and history of a site. For methods based on analyzing groundwater, general sampling requirements are to ensure that: contamination is sufficiently low to not cause the age-sensitive parameter to be perturbed significantly; significant mixing between waters of different age does not occur as a result of sampling. Invariably, there will be some degree of contamination. It is not possible to state precisely what degree of contamination is acceptable, since this will depend upon many site- and sample- specific factors. It may be possible to correct for the effects of contamination, for example, if a tracer is added to drilling fluid, allowing the proportion of drilling fluid contamination within a sample to be calculated. The degree to which mixing between water bodies of different ages may be caused by sampling, will depend upon the physical characteristics of the sampled rock body and the volume of water that is sampled. Generally, mixing is likely to be a greater issue the larger is the sample size, which means that ages calculated by certain noble gas isotopes 39 Ar, 81 Kr may be particularly affected. 2. Groundwater dating methods applied in different lithologies The techniques used to date groundwaters by the various deep geological disposal programs outlined in Section II. 1, are summarized in Table 3, which also indicates the lithologies considered by each program. It should be noted that at any particular site, groundwater dating techniques tend to be applied to all or a sub-set of the lithologies that are present. That is, programs do not usually investigate only the lithology that could potentially host a repository, or that is being researched as a potential kind 45

8 HAMA and METCALFE Table 3. Summaries of the lithologies investigated and the groundwater dating techniques employed in the different reviewed programs. of repository host rock. Usually, the understanding of groundwater age/history at a site is derived from an integrated interpretation of age-relevant information obtained from all the investigated lithologies. Hydrogeological techniques can be applied only in rocks where advection of groundwater occurs. In lower-permeability rocks, notably most argillaceous rocks, unfractured, nonkarstic limestone, many evaporite rocks and unfractured igneous rocks, groundwater and solute movement will occur by diffusion rather than advection. In these cases, observed solute gradients can be interpreted in terms of groundwater and solute residence times Mazurek et al., 2011 and references therein In this paper, this approach is covered by the group of techniques based on hydrochemical 46

9 Groundwater dating applied for geological disposal of radioactive waste evolution Table 2 The ability to obtain samples of groundwater for analysis influences strongly the applicability of the different geochemical techniques. In the cases of permeable rock types, through which groundwater may flow, it may be relatively straightforward to obtain large samples for analysis. In contrast, in low-permeability indurated or plastic argillaceous rocks, it will typically be necessary to obtain samples of water by squeezing or long-term in-situ collection. Alternatively, solutes can be leached. These activities mean that large samples cannot be obtained, which normally prevents analysis of noble gases and/or noble gas isotopes 81 Kr, 85 Kr, 222 Rn IV. Results of groundwater dating studies 1. Rationale A large number of groundwater dating studies have been covered by this review. The considered deep disposal projects for radioactive wastes have proposed to use several different kinds of host rocks, which are located in different kinds of rock sequence in different kinds of hydrogeological setting. The groundwater dating studies that have been undertaken in each program usually cover not only the host rocks, but also the lithologies that surround them. The age of groundwater within the host rock is then interpreted from integrated interpretations of all the age data that are obtained. The following sub-sections consider the main lithologies and hydrogeological settings that have been targeted as possible repository host rocks by one or more radioactive waste disposal programs. Important results of groundwater dating studies that have been applied to these lithologies and settings are presented. By subdividing the review in this way, it is aimed to highlight the factors that influence the applicability of the different sampling and dating methods. That is, it is aimed to illustrate those lithology-dependent factors and hydrogeological setting-dependent factors that influence the choice of dating methods. 2. Host rocks below the water table a. Fractured crystalline host rocks Groundwater in fractured crystalline host rocks, or rocks that behave like fractured crystalline rocks, have been the subject of numerous groundwater dating studies Louvat et al., 1999; Mahara et al., 2001; Puigdomeneck, 2001; Gascoyne, 2004; Pitkänen et al., 2004; Almén and Stenberg, 2005; Bath, 2005; Degnan et al., 2005; Bath et al., 2006; Metcalfe et al., 2007; Pitkänen and Partamies, 2007; Laaksoharju et al., 2008; Mahara et al., 2008; Grolander, 2009; Laaksoharju et al., 2009; Delos et al., 2010 These various studies show that the available groundwater dating methods can distinguish between groundwater bodies with differing residence times. It is apparent that variations in residence times are related to the heterogeneous permeability distribution of these rocks and that compartmentalization is a common feature. For example, in the Swedish sites of Äspö, Forsmark and Laxemar, and the Finnish site of Olkiluoto, at least 5 different groundwater components can be recognized: old shield brine; glacial melt water; meteoric water; Baltic seawater and chemically altered marine water. Each of these water types has distinctive chemical and isotopic characteristics and their spatial distributions appear to be related to the distribution of relatively conductive features in the rock mass. 47

10 HAMA and METCALFE At these sites, stable oxygen and hydrogen isotopic data allow groundwaters and pore waters with different origins to be distinguished. The waters that are most depleted in heavy isotopes are thought to have been recharged under cold-climate, sub-glacial conditions. At Äspö, δ 18 O of water with a glacial recharge component was as low as 15.8, significantly lower than present meteoric water, which has δ 18 O around 11. Data for 14 C shed some light on the age or ages of this glacial recharge component, but the fact that water samples are mixtures of water from different sources, including water that is dead to 14 C was found to complicate the interpretation of 14 C data. At Äspö, Mahara et al found that these mixing relationships could be distinguished using correlations between 1/Cl, 36 Cl/ Cl, and 3 H concentrations. They found that the majority of the 36 Cl and excess dissolved 4 He of interstitial groundwater in fractures at this site could be explained by sub-surface production of these isotopes. The 36 Cl/Cl ratios that can be inferred for the brine end-member based on these results are consistent with a long residence time of the chloride in the rock, most likely in excess of 1.5 Ma. At Laxemar, near to Äspö, the glacial component of the waters was found to be very low in bicarbonate and 14 C in inorganic carbon Laaksoharju et al., 2009 The very low or undetectable 14 C was considered to be consistent with the inferred water origin in glacial recharge, even though the latest glaciation in this area occurred within the last 35,000 years, which is within the range amenable to dating by 14 C. The low 14 C content was explained by the long residence time of glacial recharge in the ice before recharging the bedrock or, alternatively, the extremely low bicarbonate content of the water. Laaksoharju et al noted that some of the residence time estimates at Laxemar based on 4 He data were unrealistically long hundreds of thousands to several millions of years considering that there is clearly a component of post-glacial water present. They noted that uncertainties in the 4 He ages are due to: the possibility that He was transported from greater depths e.g. mantle origin possible He production from uranium and thorium precipitated on fracture coatings and altered host wall rock; and gas circulation facilitated by variable porosity and diffusivity, for example caused by alteration of the fracture wall rock. They noted that in addition to 4 He data, 3 He data would have helped to resolve these issues since it would allow the origins of the He to be interpreted more confidently. In spite of the uncertainty about some of the long groundwater ages based on He data, not all the He ages are unreasonable and 36 Cl data for Laxemar do help to corroborate some values. Laaksoharju et al reported the following general results of 36 Cl studies: High and variable 36 Cl/Cl ratios occur in recently recharged dilute low-cl waters. Water samples dominated by Baltic Seawater have relatively low 36 Cl/Cl ratios. Water samples with a large component of Littorina Seawater a palaeo-seawater show the lowest 36 Cl/Cl ratios. The highest 36 Cl/Cl occur at deeper depths in brackish non-marine and saline groundwaters. In these latter waters, the 36 Cl/Cl approaches equilibrium with in-situ produced 36 Cl, indicating a long residence 48

11 Groundwater dating applied for geological disposal of radioactive waste time in excess of 1.5 Ma. At Forsmark, the chemical characteristics of the groundwaters are broadly similar to the chemical characteristics of groundwaters at Laxemar and Äspö Laaksoharju et al., 2008 At Forsmark, 3 H and 14 C were valuable for indicating the depth to which recently recharged groundwater has penetrated into the bedrock. There are marked decreases in 3 H and 14 C contents at a depth of about 150 m depth, indicating that these shallower waters have residence times in the order of a few decades to a few hundred years. There are marked chemical contrasts between groundwater from transmissive fracture zones and less transmissive parts of the rock mass at Forsmark. The compositions of water samples from depths greater than about 200 m indicate a brackish marine water Littorina Sea, a palaeo-seawater having concentrations of Cl between 2,000 mg l 1 and 6,000 mg l -1. This kind of water occurs to depths of 600 to 700 m depth in transmissive gently dipping fracture zones in the south-eastern part of the candidate area. In contrast, in the rock remote from this fracture zone, where there is a low frequency of water-conducting fractures, the brackish marine water occurs only at depths between about 150 m and about 300 m. At depths greater than these, the water is dominantly brackish to saline and non-marine and is interpreted to have a longer residence time. The chemistry of matrix pore water at Forsmark is also consistent with a low rate of groundwater turnover in parts of the rock mass remote from transmissive fracture zones. Compared with the fracture groundwaters, pore water from such locations in the rock is generally lower in chloride content and enriched in δ 18 O. These observations were interpreted to indicate that the pore waters evolved from older dilute groundwaters that circulated over a very long period through only a few of the fractures cutting the rock mass. As at Laxemar and Äspö, 36 Cl and 4 He data were used to estimate the residence times of groundwaters at Forsmark. Once again, it was found that waters with a high proportion of Littorina Seawater contained unrealistically large concentrations of 4 He, taking into account 14 C data that indicated ages of these waters of a few thousand years post- last glaciation This result was again interpreted to indicate migration of 4 He from below, possibly facilitated by the greater fracture frequency and transmissivity at shallower levels. In contrast, the 4 He data for the brackish to saline non-marine groundwaters was consistent with other age indicators. It was calculated that a time greater than 1.5 Ma would be required to reach the measured 4 He concentration by considering in-situ production of 4 He. This result is consistent with the 36 Cl data, which indicates that 36 Cl in the deeper waters is in secular equilibrium with respect to in-situ production of 36 Cl in the rock mass. The hydrogeochemitry of Olkiluoto in Finland is broadly similar to that of the Swedish sites described previously Pitkänen et al., 2004 However, at Olkiluoto during approximately the last 2500 years, flushing of palaeowaters by modern meteoric waters has occurred only to relatively shallow depths compared to Forsmark and Äspö. At Olkiluoto, fresh groundwater with low total dissolved solids TDS 1 g l -1 is found only in the uppermost tens of meters and modern meteoric water has flushed Littorina Seawater down to depths of only about m. Prior to this, over about 5,000 years, the Littorina water replaced 49

12 HAMA and METCALFE most of the previously recharged cold-climate glacial water down to m. It has been interpreted that at Olkiluoto, glacial meltwater penetrated to only around m, possibly due to the presence of highly saline, and therefore dense, shield water at fairly shallow depth. Alternatively, following the last glaciation this latter water may have moved upwards to the present, relatively shallow depths at which it occurs. The spatial distribution of glacial meltwater and the depths to which it penetrated may have been controlled by the distribution of transmissive structures in the rock mass. Investigations at Olkiluoto used the same groundwater age indicators as those employed at these Swedish localities. Helium data were obtained, but have been used only to deduce qualitatively that groundwater residence time increases downwards, based on increasing He concentration with respect to depth and salinity. The 36 Cl/Cl ratios in the shallow dilute groundwaters are similar to those in surface water and rain water. In contrast, the 36 Cl/Cl ratios in the deep groundwaters approach a secular equilibrium value with respect to the 36 Cl produced in-situ in the rock mass. This finding implies a residence time of more than about 1.5 Ma. All of the Swedish and Finnish sites have low relief and hence only subdued driving hydraulic gradients for groundwater flow. Accordingly, apart from circulation of recently recharged meteoric water through shallower less than 100 m 150 m relatively fractured rock, groundwater is flowing very slowly if at all. The spatial distributions and mixing proportions of chemically distinct groundwater components reflect largely palaeo-processes, such as glaciation and shore line migration. In contrast, at Sellafield in the UK, which was investigated by Nirex during the 1990s, groundwater is subjected to a hydraulic gradient that reflects the location of the site between the coast to the west and mountains in the east up to ca.1,000 m high Bath et al., 2006; Metcalfe et al., 2007 A further difference between Sellafield and these other sites is that the fractured Borrowdale Volcanic Group BVG which was investigated as a possible repository host rock 1, is overlain by sedimentary rocks, which include sandstones, shales, evaporates and limestones. These cover rocks thicken from east to west across the investigated site, towards a sedimentary basin, the East Irish Sea basin, which is centered offshore. Groundwater flow patterns reflect not only the hydraulic gradient across the site, but also the differing properties of the BVG basement and the sedimentary cover rocks. At the reviewed sites in Sweden and Finland, fractured crystalline rocks extend from the surface downwards, except where covered by glacial drift deposits. At Sellafield, groundwater dating was carried out not only to understand the past evolution of the groundwater system, but also to help build confidence in models for present groundwater flow Milodowski et al., 1998; Bath et al., 2006; Metcalfe et al., 2007 Here, the following data were used to estimate groundwater residence times either qualitatively or quantitatively: 18 O/ 16 O, 2 H/ 1 H, 3 H, 14 C, 13 C/ 12 C, 36 Cl/Cl, 4 He and other inert gases Ne, Ar, Kr and Xe and mineralogical data. All 1 The project to develop a repository at this site was cancelled by the UK Government in

13 Groundwater dating applied for geological disposal of radioactive waste these kinds of data were obtained from both the fractured BVG and the overlying sedimentary rocks, using boreholes sited so that variations in these parameters approximately parallel to the general groundwater flow path and perpendicular to this flow path could be interpreted. The data were used to estimate absolute and relative residence times of groundwaters in both the BVG and the overlying sedimentary rocks. Spatial variations both lateral and vertical in these times were found to be consistent with the conceptual model for groundwater flow based on hydrogeological data groundwater head data and rock hydraulic conductivities This consistency built confidence in the estimated age of groundwater in the BVG, as well as the overlying rocks. As in the Swedish and Finnish sites, stable oxygen and hydrogen isotopic data from Sellafield reflect partly the origins of the water and enabled a component of cold climate recharge to be discerned in the deeper groundwater system near to the proposed repository. Among the reviewed projects, the Sellafield investigations were the only ones in crystalline rock the BVG to use noble gas data to calculate recharge temperatures and hence support the interpretation of these oxygen and hydrogen isotopic measurements. The only other identified study to calculate noble gas recharge temperatures was in the eastern Paris Basin. Using the measured concentrations of dissolved Ne, Ar, Kr and Xe, estimates of recharge temperatures were made based on the fact that the solubilities of these gases increase as temperature decrease. The calculated noble gas recharge temperatures averaged 4.8 C for deeper saline waters near to the proposed repository, in contrast to an average of 9.0 C for shallower fresh waters. This result supports the conclusion that the relatively light oxygen and hydrogen isotopic signatures of this deeper water reflect cold climate recharge. An attempt was made to use 14 C for groundwater dating at Sellafield. However, in all the boreholes from which samples were obtained for analysis, contamination of the groundwater by drilling fluid was a problem. A major issue was the use of an organic polymer in the drilling fluid, which caused perturbation of the carbon system in the groundwater. Accordingly, all model ages that were calculated for the groundwaters were minima. Minimum ages are in the ranges 13 to 18 ka for a brackish groundwater at the top of the basement rocks and 0 to 21 ka in fresh groundwaters towards the base of the sedimentary formation. The Sellafield investigations used both 4 He and 36 Cl data to estimate groundwater residence times. In saline waters and brines at the site, there was no correlation between 4 He and 36 Cl, suggesting that simple two-component mixing does not explain the 4 He contents. However, while there are uncertainties concerning 4 He accumulation, the estimated residence times using 4 He data are consistent with the most saline waters to the west of the proposed repository brines having residence times of several million years. This finding is consistent with the origin of these brines in a sedimentary basin of the East Irish Sea. The 4 He concentrations in saline groundwaters within the BVG near the proposed repository are slightly higher than those in the basinal brines to the west of this proposed repository. However, the 4 He concentrations in the BVG are consistent with Pleistocene recharge ages, if allowance is made for the lower porosity of basement rocks relative to basinal sedimentary rocks since the lower porosity results in a 51

14 HAMA and METCALFE smaller volume in which the 4 He may accumulate, leading to higher 4 He concentrations However, uncertainties in the data mean that the quantitative significance of these 4 He measurements is unclear. The 36 Cl/Cl data were consistent with this conclusion since they indicated that the brines are at equilibrium with in-situ production of 36 Cl in the sedimentary rocks of the basin. This finding indicates a residence time in excess of 1.5 Ma. Similar long residence times were also deduced for deeper saline water in the BVG to the east, near to the proposed repository, where 36 Cl/Cl ratios are again consistent with secular equilibrium. However, the 36 Cl/Cl ratios of groundwater in the deep BVG in the east of the area are about twice the ratios in the groundwater in the overlying sedimentary rocks of the west of the area, reflecting higher U and Th contents of the BVG compared to the sedimentary cover rocks. In contrast, in the central part of the investigated area, 36 Cl/Cl ratios across a transition between saline water and brine are not in secular equilibrium with the host sedimentary rocks. This observation was interpreted to indicate that Quaternary water is mixing with pre-existing Tertiary brines in this zone, indicating that the brine and saline water have moved between the different formations within the last 1.5 Ma. Significantly, this result is consistent with patterns of groundwater mixing that are deduced from groundwater flow modeling based on hydrogeological data. b. Argillaceous host rocks Argillaceous host rocks are being actively investigated in Belgium, France and Switzerland. Since these rocks have very low permeabilities, such that solute and pore water movement is by diffusion, methods typically employed to obtain water for analysis are different to those used to sample water from more permeable rocks through which groundwater can flow. Pumping from boreholes is typically used to sample groundwater from these more permeable rocks, including fractured crystalline rock. In contrast, usually pumping cannot be employed to obtain samples from low-permeability argillaceous rocks. Instead samples of water and solutes are usually obtained from argillaceous rocks by extraction from rock cores, using one or more specialist techniques: squeezing; vacuum distillation; diffusion; or advective displacement under a pressure gradient. Alternatively, pore water may be sampled from packed-off sections of boreholes drilled within argillaceous rocks, if the boreholes are left for a sufficiently long time period to allow pore water to accumulate. Gases may be sampled by outgassing from preserved core of argillaceous rock. Solutes may also be extracted by leaching of rock core. In recent years, several investigations have considered how to use the information obtained in order to determine the origins and residence times of the pore waters Rübel et al., 2002; Marty et al., 2003; Pearson et al., 2003; De Craen et al., 2004; De Craen et al., 2006; Fourré et al., 2010; Cornaton et al., 2010; Jean- Baptiste et al., 2010; Smith et al., 2010; Gimmi and Waber, 2004; Mazurek et al., 2011 At the reviewed sites, profiles of conservative relatively unreactive solutes such as Cl or stable oxygen and hydrogen isotopes across argillaceous formations, can be explained by diffusion over periods of several thousands to several millions of years. One approach to estimating solute and pore water residence times has been to use these profiles. The initial concentrations of these constituents within the 52

15 Groundwater dating applied for geological disposal of radioactive waste argillaceous formation have been estimated, using chemical and mineralogical evidence for the origins of the original pore fluid and the solutes contained there-in. The concentrations of the same constituents in the groundwaters from the aquifers bordering the argillaceous formation have also been measured. Hence, by making assumptions about the variations in the chemistry of water in these aquifers over time, it is possible to model the observed diffusion gradient across the argillaceous formation. The modeled time for the observed diffusion gradient to form can then be compared with the ages of the geological strata and the timing of geological processes e.g. folding, uplift etc. that have affected the rocks, as estimated from geological data. Hence, potentially the approach can indicate whether groundwater residence times have been affected by these geological processes. For example, substantially shorter times to develop the diffusion profiles than the known age of the argillaceous rock would indicate shorter groundwater and solute residence times than the age of the rock. A problem with this approach is that it relies on knowledge of the initial chemistry of the waters within the argillaceous formation and bounding aquifers, and the chemical evolution of waters in the bounding aquifers. Since this knowledge is typically imperfect, estimates of pore water residence times will usually be uncertain. In the Opalinus Clay at Mont Terri in Switzerland, heavy stable oxygen and hydrogen isotopes are relatively enriched within the formation, compared to values in groundwaters from shallower and deeper aquifers Pearson et al., 2003 This variation mirrors a variation in water salinity, Cl contents being higher within the Opalinus Clay up to ca. 15,000 mg l 1 than in the aquifers around 5,000 mg l 1 The isotopic data show that the waters in the aquifers are similar to recent recharge waters. These differences in isotopic and solute concentrations between the Opalinus Clay and the adjacent aquifers is consistent with the pore water of the very low-permeability Opalinus Clay having a longer mean residence time i.e. being older than the groundwater in the relatively permeable aquifers. This interpretation is supported by He concentration data, since the pore waters of the Opalinus Clay have higher He concentrations compared to the groundwaters of the aquifers. However, the quantity of He present is only 3% 6% of the levels that would have occurred if all He produced in the formation had remained there. That is, substantial He loss has occurred during the history of the Opalinus Clay. Fourré et al present measurements of dissolved He in groundwater sampled from Oxfordian and Dogger limestone aquifers in the Bure area of France, which respectively lie above and below a 150 m thick sequence of very low-permeability Callovo-Oxfordian clays. These clays comprise the potential host formation for a deep geological repository. Fourré et al calculated groundwater fluxes and residence times for the Dogger aquifer the deeper of the two aquifers by using a typical crustal helium flux of mol m -2 a -1 and assuming that the Callovo-Oxfordian formation provides an effective cap. In this way, they obtained a mean velocity of Dogger groundwaters in the range m a -1 and a He age of the groundwaters in the range 50 ka and 200 ka. Fourré et al considered these ages to be minimum values since the crustal He flux used to calculate them is probably greater than the actual flux; formations underlying the Dogger aquifer may actu- 53

16 HAMA and METCALFE ally act as a barrier to upwardly-diffusing He. In contrast, they were unable to calculate a residence time for the water in the shallower Oxfordian aquifer, owing to uncertainty about the quantity of He that diffuses upwards from the Callovo-Oxfordian clays. Smith et al report He concentrations obtained by outgassing core taken from the Callovo-Oxfordian clays. These authors found that the He concentrations within the pore waters of the clays are greater than those in the overlying and underlying aquifers, with maximum concentrations occurring near the middle of the clay formation. They showed that the He concentration profiles could be modeled by diffusion of He from the Callovo-Oxfordian towards the aquifers. Using measured He concentrations for the aquifers and estimating the He production rate in the Callovo-Oxfordian from U and Th concentrations, they calculated an effective He diffusion coefficient of m 2 s -1. However, they did not report estimates of He pore water residence time within the clay. An integrated study of groundwater residence times within the Oxfordian and Dogger aquifers is described in Lavastre et al These authors used a combination of 3 H, 14 C, 36 Cl and noble gas He, Ne, Ar, Kr, Xe data to determine the residence times of the waters in these two aquifers. An important motivation was to establish that the Callovo-Oxfordian clay formation, the potential repository host rock, forms an effective barrier to water and solute transport between the aquifers. They concluded that this clay formation in fact behaves as a seal and that there is no mixing between waters from the Oxfordian and Dogger aquifers. Lavastre et al found that noble gas data from the shallower Oxfordian aquifer indicate recharge temperatures between 3 and 8.6 C, 2 7 C less than the present average surface temperature, indicating recharge before the end of the last glacial period, 10 ka or longer ago. Additionally, in the Oxfordian aquifer, concentrations of radiogenic 4 He are cm 3 STP g -1 and cm 3 STP g -1, 2 3 orders of magnitude greater the concentration of 4 He in air-equilibrated water, consistent with residence times of several hundred thousand years. However, there is doubt as to whether the aquifer behaves as a closed system with respect to 4 He and therefore such a long residence time could not be confirmed. Compared to the groundwater in the shallower Oxfordian aquifer, the 4 He content of groundwater from the deeper Dogger aquifer was found to be up to cm 3 STP g -1, more than an order of magnitude higher and consistent with a much longer residence time of at least several hundred thousand years. Even though once again there was some doubt as to whether the Dogger aquifer has behaved a closed system with respect to 4 He, raising some doubt as to the reliability of the calculated residence time, there is considerable confidence that the residence time of the water in the Dogger aquifer is much greater than the residence time of the water in the Oxfordian aquifer. Importantly, there was no evidence for transport of 4 He across the Oxfordo-Callovian clay formation. c. Salt host rocks Two of the reviewed programs, at Gorleben in Germany and at the WIPP in New Mexico, U.S.A., have investigated halite host rocks. Activities at the former site are presently suspended, although at the time of writing they are on the point of restarting and in principal, it is possible that eventually a repository could 54

17 Groundwater dating applied for geological disposal of radioactive waste be sited in salt deposits at Gorleben. The WIPP, which has been developed in a halite formation, is significant in so far as it is the only presently operating deep geological repository for radioactive wastes anywhere in the world. Groundwater dating studies for these two programs are reported in several sources Chapman, 1986; Lambert, 1987; Lambert and Carter, 1987; Davies, 1989; Kröhn and Schelkes, 1996; Buckau et al., 2000; DOE/WIPP, 2005; Klinge et al., 2007 Direct groundwater/pore water dating in the salt host rocks themselves has not been undertaken. Apart from pockets of brine within the salt, they are essentially water-free. Furthermore, their very low permeabilities and selfsealing characteristics justify the assumption that water will move very slowly indeed through these lithologies if at all. Instead, groundwater dating in the reviewed projects has focussed upon the overburden and has aimed to build confidence in the overall understanding of the investigated sites. d. Limestone host rocks At only one of the reviewed sites, Bruce in Ontario, Canada, it is proposed that limestone should be the host rock of a deep repository for ILW and some LLW Intera Engineering, 2011 Groundwater dating studies at the Bruce site and adjacent region are described in Hobbs et al and Clark and Herod 2011 At this site, the proposed repository host rock, an Ordovician limestone-dominated formation, the Cobourg Formation, is overlain by a rock sequence of more than 650 m including substantial thicknesses of shales and limestones. The host rock is unfractured and of very low permeability, which means that transport of water and solutes through it is thought to occur by diffusion. Its pore water is hypersaline and believed to have been resident in the formation for many millions of years. Owing to the low permeability, it has been difficult to obtain hydrogeochemical information from the host rock formation. The Nuclear Waste Management Organization of Canada NWMO has commissioned several studies to investigate different methods for acquiring such information, including data from which pore water ages may be determined. One such study is reported in Clark and Herod 2011 and investigated the use of stable oxygen and hydrogen isotopes, 3 H, 14 C, 4 He, 36 Cl, 129 I. The stable isotope data, 3 H and 14 C data were obtained only for flowing groundwater samples taken from depths of ca. 300 m in boreholes. In contrast, 36 Cl and 129 I data were obtained both from flowing groundwater samples in shallower rocks in the sequence above the host formation, and from solute leachates obtained from the lower-permeability rock units, including the proposed limestone host rock. Data for 4 He were obtained by microdrilling of rock core samples followed by their encapsulation in a sealed, evacuated tube. The He was then outgassed and measured by magnetic-sector mass spectrometry. Both He concentrations and 3 He/ 4 He ratios were measured. However, the concentrations were normalized to concentration in the rock cc He STP/g and not pore water concentrations. Therefore, the concentrations do not necessarily reflect the pore water concentration profile that would drive diffusion through connected porosity. Furthermore, these measured He concentrations do not include all the He retained within minerals, since the rock samples were not heated during extraction. Consequently, the measured He concentrations provide minimum es- 55

18 HAMA and METCALFE timates of the He concentration in the rock, which is useful as a constraint on pore water residence time. Of significance is the fact that the He concentrations correlate closely with the concentrations of U in the limestone host rock of the Cobourg Formation and in the overlying Ordovician shales, consistent with little He moving into or out from these rocks noting that α-decay of U and its daughter isotopes is the main source of 4 He However, the 3 He/ 4 He ratios in the shales and underlying Cobourg Formation were found to be similar and close to theoretical values calculated for in-situ production in the shales, but lower than values calculated for in-situ production in the Cobourg Formation. These He concentration and He isotope data in combination are consistent with low concentrations of He produced in-situ within the Cobourg Formation, having mixed with a component of He that originated in the overlying shales and then diffused downwards. The much higher 3 He/ 4 He of the He diffusing from the shales resulted in the He within the Cobourg Formation also acquiring similarly high 3 He/ 4 He ratios. Based on estimates of in-situ production rates in the overlying shales, it was concluded that these shale pore waters had been resident within the formation for million years. Consequently, it was proposed that the pore waters in the Cobourg can have a similar residence time. The 129 I/I ratios for leachates from the Cobourg Formation were found to be consistent with secular equilibrium, indicating consistency with the He age estimates. 3. Host rocks above the water table The only radioactive waste disposal program to have considered host rocks located above the water table to site a deep geological repository is the Yucca Mountain Project in Nevada, U.S.A.. Tuffaceous rocks of Tertiary age were considered. These rocks are located at depths of around 300 m below the ground surface and at about 300 m above the water table. Safety arguments placed considerable emphasis on the lack of flowing water to transport radionuclides. Therefore, at this site, there was no groundwater dating as such, but considerable efforts were expended to identify so-called fast pathways for the possible migration of water in future Paces et al., 2001; Lu et al., 2003; Cizdziel, 2006 The investigations focused on tracers for groundwater movement preserved within the rocks. Paces et al described deposits of calcite and opal, which coat open fractures and lithophysal cavities in the unsaturated tuffs. They interpreted these deposits in terms of past water movement. The outer layers of the calcite deposits yielded radiocarbon ages of 16,000 to 44,000 years and thorium-230/uranium ages of 28,000 to more than 500,000 years. These ages are much younger than the age of the rock, which is around 13 Ma. The age data, together with micrometer-scale layering and distinctive crystal morphologies, were interpreted to indicate that the minerals formed very slowly from water films that migrated through open pore space within the rock. Crystal growth was shown to be very slow, probably 5 mm per million years. A major issue during the Yucca Mountain investigations was reports of elevated 36 Cl/Cl in rock core from repository depths at the site. These elevated ratios were interpreted as being possibly due to rapid penetration of water labelled with 36 Cl of atomic bomb origin. This finding, if verified, would imply water penetration to these depths within the last 50 years. 56