W.R. Alexander

Nagra, Switzerland (russell@nagra.ch)

J.A.T. Smellie

Conterra AB, Sweden (john.smellie@conterra.se)

Abstract The Maqarin Natural Analogue Project site (in NW Jordan) appears to be unique in that the hyperalkaline groundwaters in the area are the product of low temperature leaching of an assemblage of natural cement minerals produced as a result of high temperature/low pressure metamorphism of marls (i.e. clay biomicrites) and limestones. Investigations provide a consistent picture explaining the origin of the hyperalkaline waters, the persistence of some of the hyperalkaline springs/seepages, and the sequence of alteration occurring when such waters react with various rock-types. Additionally, the studies have produced good quality measurements of the concentration (and general speciation) of some repository-relevant elements in high pH groundwaters, along with microbial and colloidal populations.

The Maqarin natural analogue site therefore provides a rare oportunity to examine the mechanisms of processes associated with cementitious repositories for radioactive and chemotoxic wastes, particularly when cement pore fluids will be dominated by the dissolution of portlandite and calcium silicate hydrate gel phases.

1: early, active, high pH Na/KOH leachates (Western Springs, Maqarin; see Fig. 2 below)

2: intermediate, active, lower pH Ca(OH)₂ buffered leachates (Eastern Springs, Maqarin; Adit A-6, Railway Cutting and Waterfall Road Cutting in Fig 2 below)

3: late, inactive (fossil), lower pH silica dominated leachates (Daba region in central Jordan).

Whilst Phase I and Phase II were very much site-specific and process oriented (e.g. studies of the source term and its interaction with the host rock; testing the applicability of available thermodynamic data to hyperalkaline conditions; predicting the extent of high pH water/rock interaction using coupled models etc.), Phase III provided a more regional perspective to the geological and hydrogeochemical evolution of the entire cementitious system.

The Maqarin area (Fig. 2 below) comprises Cretaceous-Tertiary carbonate rocks overlain by Quaternary basalts, soils and alluvium. The E-W trending Yarmouk River Valley is a relatively recent geomorphological feature, having eroded through some 400 m of the Irbid Plateau exposing the Bituminous Marl Formation which is overlain by the Chalky Limestone Formation, in turn capped by basaltic lava flows. There is no direct evidence that the Yarmouk Valley is structurally controlled by tectonic events associated with the Jordan Rift Valley system.

Regionally, the exposed strata show very little deformation and are almost horizontal to gently dipping; flexures with very small curvatures are also evident. Locally, atAdot A-6, the Bituminous Marl Formation has been slightly uplifted by a N-S trending anticline which plunges to the NE. Detailed structural mapping on a local scale underlines the heterogeneity of the system, in part explained by gravity tectonics (i.e. slumping) facilitated by the steep hillslope angles (~ 45^o). Localised slump features bordering the Yarmouk River Valley are widely observed in the Eastern Spring area.

Less is known from the Western Springs locality where talus and colluvium cover most of the hillslopes. There is, however, some indication of a N-S trending fault system and the possibility of localised shear and rotation between such faults. Gravity slumping features on the scale present at the Eastern Springs locality are not readily observed, but evidence exists of valley bulging on the floor of Wadi Sijin and the presence of a landslide on the western valley side of the wadi.

The mechanisms which have activated spontaneous combustion in the Bituminous Marl Formation to produce the metamorphosed "cement zones", and subsequently the formation of the hyperalkaline groundwaters in the area, are still uncertain. Whilst earthquake shock is not seen as a likely mechanism for creating fissures and joints directly in the rock, it may be a significant factor in triggering landslides and causing abrupt rupturing of superficial strata followed by erosional downloading. If this has contributed to the process of combustion in the Eastern Springs area, then, from geomorphological considerations, a probable date of ignition is around 600 ka, and certainly not later than 150 ka. More hypothetically, a second, later phase of triggered landslides, might also be considered to explain the geomorphological features observed in the Western Springs area. Speculatively, such an event might have occurred in the last 100 ka (Late Quaternary), or even as recently as the last 10 ka (Holocene). These two phases of landslip activity indicate that the Eastern Spring system is geologically older than the Western Springs system.

In addition to the possibility of these two major landslip events, reactivation on a smaller scale may be expected to be recurrent, triggered rapidly by heavy rainfall and more gradually over longer timescales by continued lateral erosion of the Yarmouk River. This may account for the periodic sealing and reactivation of hyperalkaline groundwater conducting fracture pathways indicated from mineralogical studies and field observations.

In northern Jordan, regional hydrostratigraphic units dominate with flow generally from the central plateau westwards to the Jordan River Valley and northwest to the Yarmouk River Valley, with a component from southern Syria flowing southwest towards the Yarmouk Valley. On the local scale (see Fig 3 below), three main groundwater bodies can be recognised in the Eastern Springs area flowing northwards to the Yarmouk River Valley:

1: Deep confined aquifer water from the Amman Formation (B₂); some of these waters may move upwards along discontinuities in the overlying Bituminous Marls where there is an artesian head.

2: Infiltration of meteoric water through the Chalky Limestone Formation in the plateau lying east of Maqarin, penetrating the Bituminous Marl Formation (B₄/B₅ units) down to the contact (B₃) between the metamorphosed "cement zones" and unmetamorphosed clay biomicrites. The groundwaters follow this interface eventually discharging as seeps and springs along the Yarmouk Valley sides. This is the main groundwater source and their passage through the cement zones gives rise to the hyperalkaline varieties.

3: Water infiltrates directly into the Chalky Limestone Formation above Adit A-6, down into the Bituminous Marl and the cement zone, possibly mixing with type (2), or locally discharging along the roof and walls of the adit. If the main hyperalkaline source is in the vicinity of Adit A-6, the maximum extent of the hyperalkaline plume downflow to the river 400-500 m.

In the Western Springs area the groundwater flow is also northwards towards the Yarmouk River Valley. The waters totally originate from rapid vertical recharge into the Chalky Marl Formation at the plateau south of the Western Spring location, facilitated by widespread open karst features. These groundwaters are eventually channelled through the Bituminous Marl Formation and the cement zone, to discharge through a thick colluvium sequence along the Yarmouk Valley at an elevation of 1-4 m above the river bed. Reaction with the cement zone has also produced the hyperalkaline groundwaters at this locality, and the extent of the hyperalkaline plume downflow to the river is somewhat less than at the Eastern Springs.

Hydrogeochemical and isotope studies largely support the hydrogeological conceptualisation of the major groundwater flow pathways. The recharge groundwaters are meteoric and local in origin, of normal pH and dilute bicarbonate in type, contain measureable tritium and are the probable precursors to the high pH groundwaters. No significant mixing of groundwaters from the underlying confined Amman Formation aquifer is indicated. Interaction of the normal pH groundwaters with the cement zones in the Bituminous Marl Formation has produced two chemically distinct high pH groundwaters; Eastern Spring Ca-OH type and the Western Spring Ca-K-Na-OH-SO₄ type. Both types are believed to be older than at least 40 years (tritium-free), although minor amounts of tritium (~4 TU) in some samples may indicate slightly younger ages if mixing with recent recharge waters is excluded. The absence of thermonuclear ³⁶Cl is also consistent with recharge prior to 1950.

Hydrogeochemical modelling suggests that the observed differences in groundwater composition between the Eastern and Western Springs localities can be attributed to different stages of water/rock interaction evolution. There is no convincing evidence that the high Na, K and SO₄ concentrations at the Western Springs locality are due to initial differences in the composition of the cement zone, variable infiltration conditions, or anthropogenic contamination. A possible scenario is that the Eastern Springs area groundwater circulation has occurred over a much longer timespan and readily soluble mineral phases are already completely dissolved. This supports the analogue that the Western Springs groundwaters represent the first pore volume discharging from the cement zone, and that the Eastern Springs groundwaters represent late, more evolved, discharge. The scenario is consistent with the field observations which suggest that the Eastern Springs system is geologically older than the Western Springs system (Smellie, 1998).

To model the evolution of a hyperalkaline plume, in a cementitious repository setting, it is assumed that that the initial leachates are pushed from the cement in a piston effect, due to the flow of groundwater into the cement further upstream in the repository. At the cement/host rock interface (the proximal part of the plume; see Fig. 4 below), the hyperalkaline leachates have not yet reacted with the host rock and so have a high pH and high concentrations of Na, K and Ca, reflecting the cement porewater chemistry. As the plume reacts with the host (aluminosilicate-bearing) rock, the pH falls, as do the Na, K and Ca concentrations in the groundwater, while the concentrations of Al and Si rise fractionally. Beyond the distal edge of the plume, in the, as yet, undisturbed host rock, the groundwater pH is near neutral, the Na, K and Ca concentrations are low, while the concentrations of both Al and Si are high. This pattern has consequences for the secondary mineralogy: C-S-H phases will be found in the fractures (through which the plume has migrated) in the proximal part of the plume, reflecting the fact that the leachate has not yet reacted with the host rock and is equilibrated with the C-S-H phases which make up the cement. As the leachate moves downstream and interacts with the aluminosilicates in the host rock (and the host rock groundwater and porewater), the Al concentration increases, precipitating C-A-S-H phases. At the distal edge of the plume, the leachate has reacted with an even larger volume of host rock (and the host rock groundwater and porewater) and eventually precipitates zeolites as the Al and Si concentration in the groundwater becomes high enough and the pH low enough.

If it is assumed that the plume continues to migrate, then the zeolite zone produced by the distal part of the plume will later be over-run by the middle of the plume, inducing some re-dissolution of the zeolite phases and replacement by C-A-S-H phases. This mixed zeolite/C-A-S-H zone will be then over-run by the proximal part of the plume, inducing further re-dissolution of the zeolite/C-A-S-H phases and replacement by C-S-H phases. These complex mixtures have been observed repeatedly at Maqarin (Alexander, 1992, Linklater, 1998).

The fracture mineralisation and alteration of the clay biomicrite/limestone due to interaction with hyperalkaline groundwaters has been examined in detail at the Eastern Springs and Western Springs localities. Several successive stages of fracture mineralisation have been observed, reflecting both the evolution of the cement leachates and changes in rock-water interactions. The system is highly complex; in reality, the sequence of secondary mineralisation may be reversed, repeated or parts may be missing entirely in areas, depending on the specific evolution of the cement zone and the reactivation of previously sealed fractures. In addition, individual fractures may seal at any point in the sequence and, unless reactivated tectonically, remain sealed.

5.1. Western Springs location

The data currently available are limited to the examination of large boulders and clasts of basalt and limestone/clay biomicrite (including chert), embedded in Quaternary colluvium deposits which characterise the Yarmouk River Valley sides above the river bed. The hyperalkaline groundwaters preferentially flow through, and along the base of the colluvium deposits, before discharging as seepages and small flows into the Yarmouk River. As a general conclusion, quartz, chert, K-feldspar, glass and plagioclase are the most reactive phases with the hyperalkaline groundwaters, dolomite is moderately reactive, augite, hypersthene (orthopyroxene) and olivine are weakly reactive, and apatite and Ti-Fe oxides appear to be largely unreactive. Similar reactivity is observed in the sand-silt matrix which characterises the colluvium.

Although the reported results refer to a porous media (namely the colluvium), nevertheless it is still possible to produce a paragenetic mineral sequence. Mineral paragenesis of the secondary alteration phases indicates the following sequence: 1) a carbonate stage, 2) a very hydrous Ca-K-Na-Al ‘zeolite-type’ stage, 3) a CSH gel stage with a variable Ca:Si ratio covering the range CSH(I) to CSH(II) (i.e. suolinite-afwillite), 4) a more Si-rich CSH gel (i.e. okenite-type), and 5) a low Ca-Si ratio CSH phase (i.e. truscottite-type). The latest stage of alteration is the filling of remaining pore spaces and blanketing of the CSH and CASH phases by calcite or aragonite. Stages 3, 4 and 5 repeat in places, indicating pulses in the hyperalkaline plume flow.

5.2. Eastern Springs location

As the groundwater flow at the Eastern Springs location is fracture/joint controlled, most alteration processes and mineral paragenesis are confined to these features. Most wallrock alteration takes place within 0.5 to 4 mm of the fracture. Within this zone, the fine-grained matrix calcite, kaolinite, silica, traces of illite, albite and organic matter are dissolved and pyrite, trace sulphides and glaucony are oxidised. The wallrock porosity is significantly enhanced for up to 1 mm from the fracture and the original wallrock mineralogy tends to have been totally replaced by fine-grained secondary phases.

The fracture mineralisation is highly complex and the hyperalkaline groundwaters appear to have exploited existing calcite veins in the clay biomicrite/limestone, utilising partings which opened up between the calcite vein-fill and the wallrock. The exposed surfaces have been mineralised initially by needles of aragonite. The remaining fissure has been infilled by a complex mixture of gypsum, ettringite-thaumasite, amorphous C-S-H gel (or C-A-S-H gel) and very hydrous, fibrous to gel-like zeolites and gypsum. Textural evidence, such as the existence of complex zonations within the mineralised fractures that contain different minerals in different textural settings, and the sharp boundaries between the zones, indicate that the fractures were repeatedly re-opened (presumably by tectonic movements) and re-sealed by fracture minerals. Based on this field evidence, it is suggested that each phase of mineralisation (and possibly also of cement leaching) is preceded by tectonic reactivation of the fractures, which in turn are efficiently sealed by secondary phases precipitating from groundwaters percolating through the open fractures. It is worth noting that each reactivation event only affected a fraction of all fractures present in the rocks, such that the exact sequence of mineralisation phases differs from fracture to fracture.

The alteration pattern, closest to the ‘source’ and best observed deep in Adit A-6, reveals the following mineral paragenesis: 1) aragonite stage, 2) ettringite -thaumasite stage, 3) C-S-H stage, and 4) zeolite stage. Further downflow along the hyperalkaline plume (e.g. nearer the mouth of Adit A-6 and at the Railway Cutting), at least eleven paragenetic episodes of mineralisation have been identified. The temporal alteration can be summarised as: 1) Initiation Stage (carbonate - portlandite - brucite/hydrotalcite), 2) Ettringite-Thaumasite Stage, and 3) C-S-H Stage.

5.3. Summary and conclusions

The spatial and temporal evolution of the secondary phases observed at Maqarin have been placed in the framework of the conceptual model for a hyperalkaline plume discussed above. The data from the Western Springs colluvium present a snapshot of the likely temporal evolution of a plume in any flow system where the central part of the plume may over-run the distal edge which may then be over-run by the proximal portion, with each new section of the plume replacing wholly, or in part, the previous mineral assemblage. Thus, the observations from the Western Springs have proved immensely useful in suppporting the validity of the conceptual model but, as the data are only of relevance to a highly advective system (e.g. in a highly porous rock or in a fractured rock in a tectonically highly active region), care is required if the observations are to be extrapolated to a repository site.

The Eastern Springs may be seen to represent a snapshot of the likely spatial evolution of a plume in any flow system with the observed mineralogical changes with distance from the cement source in close agreement with the conceptual model. Although the Eastern Springs data initially proved extremely difficult to interpret, the overall pattern at the Eastern Springs is now seen to differ little from that at the Western Springs; but the fact that the system seals and re-opens in a complex sequence makes interpretation difficult. However, what is clear at Maqarin is that the fractures seal; in a tectonically quiet area the fractures would presumably remain sealed, but at Maqarin they are constantly re-activated, so producing the complex paragenetic patterns observed.

A literature compilation of thermodynamic data of zeolites has been carried out because of the potential of zeolites to sorb and retard certain radionuclides of importance to the long-term performance assessment of a cementitious repository. The zeolites observed at Maqarin are not in thermodynamic equilibrium with the current hyperalkaline groundwaters; the concentrations of major aqueous species are probably controlled by equilibrium with portlandite and an ettringite-thaumasite solid solution. The formation of zeolites at the Eastern and Western Springs localities is intimately linked with pore fluid composition (pH, activities of aluminium and silica) and probably reflects alteration in the very early stages of sealing (see Fig 3 below).

A detailed study of the extent of rock matrix diffusion was carried out on four profiles taken perpendicular to water-conducting fractures from Adit A-6 (Linklater, 1998). Each profile was analysed for a suite of elements and natural decay series radionuclides along with porosity variations and the results are highly ambiguous. Significant variations in the unaltered clay biomicrite signature has made it impossible to detect any potential perturbations due to hyperalkaline water/rock matrix interaction (i.e. the background noise is too great). In only one case is there some suggestion of a clear signal: the ²²⁶Ra/²³⁸U ratios in two of the profiles suggest relatively young rock/water interaction at up to 4-7cm depth. However, these depths should be treated cautiously considering that all four samples are heavily influenced by microfracture networks extending several centimetres into the rock, and by pre-existing lithological variations (bedding etc.).

It is intended to collect additional samples from a less tectonically disturbed site (Waterfall Cutting, to the east of the Eastern Springs) for further matrix diffusion studies in the hope of obtaining an unambiguous picture of the effects of the hyperalkaline leachates on the host rock.

Some vein minerals (and associated fracture wallrock) from sealed fractures in Adit A-6 have been analysed for natural decay series radionuclides and dates can be very tentatively assigned to the hyperalkaline groundwater alteration. A diagenetic calcite vein was dated by the ²³⁰Th ingrowth method at between 500 ka and 2 Ma, providing a maximum age for the metamorphism which produced the cement zone mineral assemblage. This can be compared with the geomorphological assessment of uplift and erosion in the Yarmouk Valley area which suggests a probable date of ignition at around 600 ka, and certainly not younger than 150 ka. Analyses of samples of tobermorite vein filling and jennite-ettringite vein filling produced ages of 90 ka and 80 ka respectively. A reported ¹⁴C age of less than 650 a for recarbonated cement zone material from Adit A-6 does not necessarily contradict the ²³⁰Th ingrowth ages, rather it confirms the scenario of repeated re-activation of fractures (due to gravitational tectonics) with consequent mutli-phase fracture infill.

Indeed, the persistence of secondary fracture materials of differing ages suggests that, once sealed, the secondary fracture-filling mineralogy in the hyperalkaline disturbed zone can remain stable for safety assessment relevant timescales - with the implication that radionuclides associated with these secondary phases also can be isolated from the evolving groundwaters. No data exist for the porous media flow Western Springs system although the petrological data cited above suggests that the system has remained open throughout its lifetime, implying that any age calculated must be a reflection of a complex multi-phase reaction.

8.1. Microbes

Microbiological studies have clearly shown the presence of populations (~10⁵ microbes mL^-1) of heterotrophs and sulphate reducing bacteria (SRB) in the groundwaters and, at least an order of magnitude more bacteria on the fracture faces. Gene sequencing work on microbial material from Maqarin carried out during Phase III have shown that the total number of bacteria are within the range of other subterranean sites and that there is a diverse microbial population present. Some in situ evidence was found of attached biofilms on fracture faces in contact with hyperalkaline water, and also signs of growth in some cultures (albeit at a maximum pH of ~11).

Unfortunately, many of the bacteria present in the groundwaters still remain unidentified because of the difficulty of maintaining appropriate conditions (especially pH) in the laboratory in which to cultivate enough cells to enumerate and classify.

8.2. Organics

Studies have shown that the high DOC content in the groundwater from the Western Springs locality can hardly be explained by dissolution of organic matter from the Bituminous Marl by percolating high pH groundwaters. Solution experiments show that only a minor part of the organics could dissolve in the water; in addition, aromatics are found in the DOC whereas organic material is predominantly non-aromatic in the marls. Organic matter does not appear therefore to play any major role in the geochemistry of the trace elements studied. Furthermore, in situ groundwater speciation analyses suggest that organic complexation is insignificant in these hyperalkaline groundwaters.

8.3. Colloids

In the case of a cementitious repository, degradation of the cement may provide a significant source of colloids at both the near-field/far-field interface and at the margins of the hyperalkaline plume (which may extend downstream of the repository). Evidence from the Maqarin groundwaters suggests that the numbers of colloids generated in the cement zone will be low. This is supported by the very low amounts of colloids measured from the Adit A-6 (10⁷ colloids mL^-1). It has been also demonstrated that no uranium (or other trace elements) is associated with the colloidal material characterised. However, as a note of caution, sampling for colloids was conducted under oxidising conditions, not representative of the reducing conditions expected around a repository.

Studies of smectite stability under hyperalkaline conditions have been hindered by the small amounts of clay (~5 wt/%) present in the host Bituminous Marl. Nevertheless, studies of altered clay biomicrite samples from Maqarin (e.g. near the entrance of Adit A-6) showed the presence of smectite alteration products, namely volkonskite (chrome smectite). However, the smectite seems to be an exogenic precipitate within hyperalkaline-altered chalks and hydrated marbles, rather than an in situ reaction product of primary clay-rich lithologies. In general, more field studies are necessary to better assess the availability of clay-rich zones.

10.1. Thermodynamic Databases (TDB)

Two tests of a suite of geochemical databases have been carried out within the project to date. All the TDBs were limited in the representation of the controlling solid phases. Most of the predicted solubility controlling solids were non-representative of the natural system, being pure end-members whereas the actual mineralogical data indicated the involvement of trace elements in solid solutions or as trace incorporations with C-S-H gels, for example. Nevertheless, the TDBs generally produced conservative estimates of trace element solubility, as is required in a repository performance assessment.

10.2. Geochemical codes

A limited test of geochemical codes was conducted during the early stages of the Maqarin study with the aim of comparing code predictions in the relatively high ionic strength (up to 0.1M) hyperalkaline waters. Two codes, MINEQL/PSI and PHREEQE were examined (using similar thermodynamic databases) and the results indicated that, for the Maqarin groundwaters and model cement waters (ionic strength 0.2M), the differences in activity coefficient calculations carried out by the codes were usually one order of magnitude smaller than the uncertainties in the equilibrium constants contained in the databases (and so the resulting speciation calculations and predicted solubilities were almost identical).

10.3. Coupled (geochemistry-transport) code testing

To date, only two tests have been carried out; in both instances, a comparison was carried out between an equilibrium approach (CHEQMATE) and a kinetic approach (MARQUISS). Note, both tests were carried out prior to all site and mineralogical data being available; this would probably warrant an alternative approach in hindsight.

Two modelling scenarios were identified, for the Eastern and Western Springs localities respectively:

Western Springs:

The most significant result of the tests show that the codes' predictions are relatively insensitive to the lithological, hydrogeological and hydrochemical differencies between the sites, with the predicted sequence of mineral reactions being very similar in all cases and, in gross terms, in agreement with the observed mineralogy.

Initially, CHEQMATE predicted fracture sealing with ettringite within five years while MARQUISS predicted fracture sealing around 0.6m from the cement source, also with ettringite. CHEQMATE also predicted the observed calcite 'front' within the clay biomicrite matrix but it has not been possible to check the predictions for other secondary phases in the matrix due to the complexity of the fracture re-activations. Later calculations showed reasonable agreement between the predicted reaction rims on the clasts in the colluvium and observations, although the thickness of the rims was significantly over-predicted.

10.4. Microbiology code testing

For modelling purposes, the Maqarin site was divided into three zones, A (upstream of the cement zone), B (cement zone) and C (downstream of the cement zone). The code predicted that, in zones A and B, the limit on maximum microbial growth could be defined by the phosphorous availability whereas in zone C, growth is limited by the amount of energy available (i.e. carbon). The results for zones A and C are surprising in that the groundwaters from all three zones are saturated with respect to apatite, a common accessory mineral in the rock. It may be that the phosphorous in apatite is not easily accessible or that the microbes can only use phosphorous from an organic source.

The MGSE code properly predicted that zones A and C contained the higher populations (it was presumed that the lower population in B was due to loss of energy and nutrient sources during the combustion event) but, for all three zones, the predicted populations were three to five orders of magnitude greater than were observed, suggesting that further code development may be necessary and that more appropriate data are still required from the site (e.g. better definition of the in situ carbon source, better hydrogeological data etc.).

The Maqarin site now provides a consistent picture explaining the origin of the hyperalkaline waters, the persistence of some of the hyperalkaline springs/seepages, and the sequence of alteration occurring when such waters react with various rock-types. Additionally, the studies have produced good quality measurements of the concentration (and general speciation) of some relevant elements in high pH groundwaters, along with microbial and colloidal populations).

The Maqarin natural analogue site is therefore a unique location to be able to examine the mechanisms of processes associated with cementitious repositories, particularly when cement pore fluids will be dominated by the dissolution of portlandite and calcium silicate hydrate gel phases. Evidence from Maqarin shows that:

• the conceptual model for the evolution of a hyperalkaline plume in a host rock is largely consistent with observations at the site,
• hyperakaline pore fluid conditions generated by minerals analogous to those envisaged for cements are long-lived (in excess of tens of thousands of years),
• predictive models of the solubility of elements of interest to radioactive waste disposal provide conservative estimates of solubility (i.e. solubilities are overestimated),
• the amounts of colloidal material generated at the cement/host-rock interface zone probably will be low,
• sequences of minerals predicted by thermodynamic and coupled modelling are similar to those observed in hyperalkaline alteration zones at Maqarin,
• due to ambigious data, it is not yet possible to say if the rock matrix will be accessible to diffusion of aqueous species even during the phase of on-going wallrock alteration,
• small aperture fractures will be self-healing, larger aperature possibly so,
• tectonic effects upon fracture sealing and the site hydrology need to be considered on a repository site-specific basis.

Alexander, W.R. (Ed.)., 1992. A natural analogue study of cement-buffered hyperalkaline groundwaters and their interaction with a sedimentary host rock - I: Source-term description and geochemical code database validation. NAGRA Technical Report Series (NTB 91-10), Nagra, Wettingen, Switzerland.

Alexander, W.R. and Mazurek, M. (1996). The Maqarin natural analogue: possible implications for the performance of a cementitious repository at Wellenberg. Nagra Unpublished Internal Report, Nagra, Wettingen, Switzerland.

Khoury, H.N., Salameh, E., Mazurek, M. and Alexander, W.R. (1998). Geology and hydrogeology of the Maqarin area. Ch. 2 in J.A.T.Smellie (editor). Maqarin Natural Analogue Study: Phase III. SKB Technical Report. (TR 98-04, Vols I and II). SKB, Stockholm, Sweden.

Linklater, C.M. (Ed.), 1998. A natural analogue study of cement buffered, hyperalkaline groundwaters and their interaction with a repository host rock II. Nirex Report no. S/98/003, UKNirex, Harwell, Oxon., UK.

Savage, D. (1998). Zeolite occurrence, stability and behaviour. Ch.8 in J.A.T.Smellie (editor). Maqarin Natural Analogue Study: Phase III. SKB Technical Report. (TR 98-04, Vols I and II). SKB, Stockholm, Sweden.

Smellie, J.A.T. (Ed.), 1998. Maqarin natural analogue project: Phase III. SKB Technical Report (TR 98-04, Vols I and II), SKB, Stockholm, Sweden.

The material cited above, while open, has a somewhat restricted circulation. Interested readers may also find further information on the project in:

Alexander, W.R. (1995) Natural cements: How can they help us safely dispose of radioactive waste? Radwaste Magazine 2, vol 5, Sept. 1995, pp 61-69.

Alexander, W.R., Dayal, R., Eagleson, K., Eikenberg, J., Hamilton, E., Linklater, C.M, McKinley, I.G. and Tweed, C.J. (1992) A natural analogue of high pH cement pore waters from the Maqarin area of northern Jordan II: results of predictive geochemical calculations. J. Geochem. Explor. 46, pp l33-146.

Barnes, I., Presser, T.S, Saines, M., Dickson, P. and Koster van Groos, A.F. (1982) Geochemistry of highly basic calcium hydroxide groundwater in Jordan. Chem. Geol. 35, pp147-154.

Chambers, A.V. (1994) Use of the quasi-stationary state approximation to determine the migration of mineral alteration zones at a natural analogue for the disturbed zone of a cementitious radioactive waste repository. Sci. Basis Nucl. Waste Manag. XVIII, pp 639-644

Clark, I.D., Dayal, R. and Khoury, H.N. (1994) The Maqarin (Jordan) natural analogue for 14-C attenuation in cementitious barriers. Waste Manag. 14, pp 467-477.

Kattan, Z. (1995) Chemical and environmental isotope study of the fissured basaltic aquifer systems of Yarmouk Basin (Syria).IAEA-SM/336/28, IAEA Int. Symp. Isot. Wat. Res. Manag. (Vienna, March 20-24, 1995).Khoury, H.N. and Nassir, S. (1982) High temperature mineralisation in the Maqarin area, north Jordan. Neues Jahrb. Miner. Abh. 144, pp 197-213.

Khoury, H.N., Salameh, E. and Abdul-Jaber, Q. (1985) Characteristics of an unusual highly alkaline water from the Maqarin area, Northern Jordan. J. Hydrol. 81, pp 79-91.

Khoury, H.N., E.Salameh, I.D.Clark, P.Fritz, W.Bajjali, A.E.Milodowski, M.R.Cave and W.R.Alexander (1992) A natural analogue of high pH coment pore waters from the Maqarin area of northern Jordan I: introduction to the site. J. Geochem. Explor. 46, pp ll7-132.

Linklater, C.M., Y.Albinsson, W.R.Alexander, I.Casas, I.G.McKinley and P.Sellin (1996) A natural analogue of high pH cement pore waters from the Maqarin area of northern Jordan: comparison of predicted and observed trace element chemistry of uranium and selenium. J.Contam. Hydrol. 21, pp 59-69.

McKinley, I.G. and Alexander, W.R., (1992) A review of the use of natural analogues to test performance assessment models of a cementitious near field. Waste Manag. 12, pp 253-259.

Saines, M., Dickson, P and Lambert, P. (1980) An occurrence of calcium hydroxide groundwater in Jordan. Groundwater 18, p 503.

Smellie, J.A.T., Karlsson, F. and Alexander, W.R., (1997) Natural analogue studies: present status and performance assessment implications. J.Contam. Hydrol. 26, pp 3-18.

Tweed, C.J and Milodowski, A..E (1994) An overview of the Maqarin natural analogue project - a natural analogue study of a hyperalkaline cement groundwater system. in H.von Maravic and J.A.T.Smellie (editors), 5th NAWG Workshop, Toledo, Spain, 5-9 October, 1992, CEC EUR 15176 EN, Brussels, Belgium.

West, J.M., Coombs, P., Gardner, S.J. and Rochelle, C.A. (1995) The microbiology of the Maqarin site, Jordan. A natural analogue for cementitious radioactive waste repositories. Sci. Basis Nucl. Waste Manag. XVIII, pp 181-189.