Nuclear waste is one of the biggest downsides to nuclear power, and can remain dangerous for hundreds of thousands of years.  Geological disposal is often stated as the most preferable way of dealing with it, but what does it entail?  What are the problems that need to be overcome, and how are governments going about overcoming them?  Fortunately, most governments are trying to be transparent about the process, with thousands of reports available on the web. 

For some coursework earlier this year, I looked at the subject from a geological point of view.

Green protesters with their "nuclear waste"
Green protesters with their "nuclear waste".  Photo CC-BY-SA Picasa user John.

Nuclear waste is produced in fission reactions. The majority comes from civilian power and military weapon production, with medical and research isotope manufacture also contributing. It includes spent fuel, along with any other material that has become radioactive, such as reactor parts.  As well as being radioactive, it also produces large amounts of heat. For example, one tonne of spent fuel has a thermal output of 1,300 W at the time of geological disposal. While some isotopes decay quickly, the amount of time some high-level waste remains dangerous is measured in thousands to hundreds of thousands of years.

Nuclear waste is usually split into low, intermediate and high level waste (LLW, ILW and HLW respectively). While HLW is the most dangerous and the main candidate for geological disposal, some ILW can also be long-lived. Short-lived intermediate and low level waste is often stored in surface facilities; the low level of the radiation and the short halflives mean that it is less dangerous. This blog will focus on the disposal of HLW and long-lived ILW; these are the most dangerous, and the hardest to dispose of safely. According to the IAEA, in 2008 there was almost 360,000 m3 of HLW and over 3,000,000 m3 of long-lived low and intermediate level waste (LILW) in storage waiting to be disposed of around the world.

Time taken for the radioactivity of HLW and spent fuel to decay below the radioactivity of naturally occurring uranium oreThe main issue with the disposal of nuclear waste is its longevity.  The diagram to the left shows the amount of time taken for HLW and spent fuel to decay to below the radioactivity of naturally occurring uranium ore (data for HLW here, for spent fuel here). Its radioactivity presents a danger to humans and other living beings, and therefore must be isolated from the biosphere until it decays to a safe activity level. Because this takes so long, conventional storage facilities are inadequate. Uncertainties about the future of society means that the maintenance these facilities require cannot be guaranteed, and dangerous radionuclides are likely to escape. Geological disposal, as outlined below, is designed to be passively safe; it will keep the waste contained even if left alone for thousands to millions of years.

Most concepts involve a multi-barrier approach, where engineered barriers are combined with the surrounding geology to minimise the chance of the escape of any dangerous materials. Waste is first processed into a form that restricts the mobility of dangerous compounds. This often involves vitrifying it. The processed waste is then packed into containers and placed in the repository. The final layer of engineered barriers is the backfill used to fill some or all of the repository. Bentonite clay is often suggested, because of its very low permeability.

There are many problems that need to be overcome in the siting and construction of a geological repository. Some of these are political – such as local opposition to such schemes – others engineering in nature. I will look mainly, however, with the geological constraints that must be placed on such facilities.   While this I will concentrate on deep burial in a stable geological formation, it must be pointed out that other geological options have been considered. The most discussed of these options is disposal in a subduction zone. However, these areas are prone to earthquakes, which could easily disrupt nuclear waste. Subduction rates are so slow that the waste will have decayed to safe levels before it is subducted to any significant depth. The London Convention also makes disposal at sea illegal.

So what geological problems are there, and how have they been approached?

Water Circulation

While depositories are engineered to prevent release of the nuclear waste contained within, engineering barriers will eventually fail. Water circulation, as well as providing a path for soluble waste to escape to the surface, will increase the speed at which engineered barriers such as metal casing will degrade. It is therefore imperative that the amount of groundwater reaching the repository is minimised.

Groundwater circulates in two distinct places within the bedrock, within the pores and within fractures. Depending on the rock type, one or the other of these will be more significant. Crystalline rocks, such as those investigated at the Swedish underground Hard Rock Laboratory in Äspö and the proposed Yucca Mountain repository (now defunct, but much research was done there) in the US, have very low permeabilities but are often highly fractured. In clays, such as those investigated in French and Belgian underground laboratories, the permeability is higher (although still low) but fractures are much rarer.

Groundwater, and any radionuclides dissolved in it, can move in two separate ways: through advection and through diffusion. Advection occurs in high permeability rocks, quickly transporting dissolved radionuclides and hence making them unsuitable for a nuclear waste disposal. Advection also can occur in fractures in crystalline rock. Diffusion, on the other hand, is a slow process. While ideally there would be no transport of radionuclides away from the repository, in practice this is unlikely to be possible for the large time-scales required. It is therefore favourable to site depositories in rocks where the primary groundwater transport is by diffusion.  How to ensure this varies on the type of rock.

Crystalline Rock

Lokation of Äspö and Forsmark
Sweden has one of the most advanced research programmes for the disposal of nuclear waste. Much of the research was done in situ in the underground Äspö Hard Rock Laboratory, however the final disposal of nuclear waste will occur in Forsmark (see map right, geology from OneGeology). They are both part of the Baltic Shield, which is dominated by mainly high-grade metamorphic and intrusive rocks. This was metamorphosed during Svecokarelian orogeny (mountain building event) in the Proterozoic (almost 2 billion years ago), although later orogens also had an affect.

In Forsmark, there are four generations of fractures. These have been dated using cross-cutting relationships and Ar-Ar methods on the minerals that have precipitated along these fractures. The first generation formed close to 1.8–1.7 Ga, towards the end of the Svecokarelian orogony. The second were dated at about 1 Ga, during the early stages of the Sveconorwegian orogeny, while the third generation were dated at between 456–277 Ma. The final generation were much more recent, and sub-horizontal instead of the steeply dipping fractures of other generations. The origin of the final generation of fractures is thought to be glacial loading and unloading.

Mapping and characterising fractures present in the proposed area for a repository is an important first step in assessing the potential groundwater flow. Dating movement on these fractures can be especially important if there is a chance they could move again. In Forsmark, the final generation of fractures is likely due to glacial loading and unloading. While the evidence suggests there is little current movement on these fractures, Scandinavia is likely to undergo future glaciations before waste has decayed to a safe activity level despite current warming trends. This is likely to reactivate these fractures, changing the groundwater flow.

The geochemical properties of the groundwater are useful. Groundwater interacts with the engineered barriers, degrading them. Knowledge of the pH and oxidation state is necessary to model the effect of the groundwater on the engineered barriers, and the geochemical profile of the groundwater can be compared to likely sources to constrain the amount of mixing. This has been used in Forsmark, for example, to show that there has been little mixing between surface water and water in the bedrock since the end of the last glaciation.

All this information can be used to create a model of predicted water flow, and from that the chance and likely impact of radionuclides reaching the biosphere. As Forsmark is a coastal location, climate change also must be taken into account. Sea-level rise can lead to changes in the hydrogeology of the area; a rise can lead to increased salinity of the groundwater surrounding the repository, while a drop in sea-level could lead to a drop in the water table causing oxidation processes. Both of these scenarios will have an effect on the speed of corrosion of engineered barriers and transport of radionuclides away from the repository.


Volcanic tuffs are also candidates for nuclear waste depositories, such as the Yucca Mountain site in the U.S.A.. Welded tuffs can have low permeabilities, but the most attractive property of tuffs is their ability to trap some radionuclides through sorption and anionic exclusion. Sorption occurs when positively charged cations are fixed at the surface of some minerals. Anionic exclusion occurs when negatively charged anions are repelled by these surfaces, lowering the effective porosity of the rock. Both of these effects slow diffusion, and the release of radionuclides.

Yucca Mountain also differs from other suggested site in that rather than just relying on the low porosity and permeability of the surrounding geology, the site sits above the saturated zone. While this is possible in arid areas such as Nevada, in more temperate regions it is harder to bury the waste at large enough depths without reaching water-saturated bedrock. It is worth pointing out that future uncertainties over the climate may mean that Yucca Mountain receives more rain, leading to water-saturated rocks at the level of the repository.

An interesting method has been used in Yucca Mountain to calculate the infiltration of surface water to the depths at which the repository will be constructed. Atomic bomb tests at the nearby Nevada Test Site have lead to increased production of 36Cl since the 1950s. This ‘bomb-pulse’ can be used as a tracer to determine the rate of infiltration. However, there are disagreements over whether 36Cl found in the excavations has come from infiltration or is a contaminant on the tunnel surface.


Location of the Belgian HADES and French Meuse/Haute-Marne underground labsSubterranean laboratories have been constructed in clays in both Belgium and France. The French facility at Meuse/Haute-Marne is constructed at a depth of 490 m in the Callovo-Oxfordian argillite formation, and the Belgian HADES (what an awesome name for a lab!) facility near Mol is located at a depth of 225 m in the Boom Clay.

Although permeability is higher than in unfractured crystalline rocks such as granite, permeabilities in the argillite at Meuse/Haute-Marne can be as low as 10-12–10-14 m/s. Diffusive transport dominates in the Callovo-Oxfordian formation. ANDRA carried out extensive seismic reflection surveys of the area, and no faults with a throw greater than a few meters were found. This means there are no high-permeability pathways for the transport of water and dissolved radionuclides, and most transport will be by diffusion. Diffusion is slowed down, as in tuffs, by sorption and anionic exclusion. Borehole research has shown that the Callovo-Oxfordian formation is relatively homogeneous and at least 130 m thick over a large area. ANDRA defined a 200 km2 ‘transposition zone’, within which the properties of the argillite are suitable for a repository.

Another factor to be considered is the mechanical strength of the rock. Excavating tunnels for the repository leads to stain on the surrounding rock, and creates a damaged zone of increased microfractures (Balland et al., 2009). The Callvo-Oxfordian contains a large proportion of carbonates, which increase its mechanical strength. It also contains a large proportion of smectite. The swelling properties of smectite allows a certain degree of deformablity, allowing some ‘healing’ of the damaged zone.

Earthquakes and Volcanic Eruptions

Earthquakes and volcanic eruptions, although rare need to be considered due to the long operating time of depositories. In geologically active regions, such as Japan and the western U.S.A. this is obviously more important, however there are still hazards which need to be taken into account in more stable regions.


Earthquakes present a hazard to depositories because their movement could damage the engineered barriers and allow the escape of radionuclides. Most of the largest, magnitude nine earthquakes are restricted to subduction zones. As already stated, subduction zones are not ideal for nuclear waste disposal, partly due to this earthquake hazard. However, even on land earthquakes up to magnitude eight are found. A magnitude six earthquake produces a slip of about 30 cm, with a magnitude eight slipping about 8 m. Anything smaller than the slip from a magnitude 6 will likely be taken up by slip in the backfill, especially if this backfill is something like bentonite clay, however earthquakes of magnitude 6–8 could potentially release radionuclides from containment. It is therefore imperative that the location of active faults is known.

Very slow strain rates can still lead to damaging earthquakes over the lifetime of a nuclear waste repository. There are additional uncertainties as the repeat rate of such faults would be much greater than the historical record, while still being inside planned repository lifetimes. This means that there may be many faults capable of causing damage situated within the location of the repository that are unknown at the time of construction. Careful record of any fractures encountered while excavating are therefore crucial in identifying possible problems. Current strain rates in actively deforming regions can also be measured by GPS. Although they primarily measure the elastic intersesimic strain, this can be modelled to try and extract the location of future earthquakes. Although not exact, any area of high strain rates can be ruled out.

It is worth pointing out, however, that if the fault does not pass directly through the repository, damage will be limited. Underground facilities move with the rock, therefore the loads caused by inertia are far smaller than would be expected at a surface facility. In places where little recent tectonic activity has occurred, such as the Paris basin in which the French Meuse/Haute-Marne site is located, depositories can easily be built to withstand any seismic shock. ANDRA’s worst case scenario of a magnitude 6.1 earthquake locate 6 km from the site would not lead to any significant damage, indeed it is calculated the repository could withstand much larger earthquakes.

More important is the earthquake hazard in zones of diffuse deformation. Yucca mountain, for example, is part of the Basin and Range area, and is surrounded by many active faults. Similarly in Japan, most of the country is being deformed. The offshore subduction zones cause crustal shortening. While their siting program calls for a location away from any active fault, it is only in the first stages and no detailed site evaluation has been made.

In the stable Fennoscandian craton, where the Swedish Forsmark and Finnish Olkiluoto sites are located, glacial cycles create an added hazard. There are many examples of faults that are thought to have occurred during the retreat of the ices sheets, as the continent was unloaded and isostatic equilibration took place. The longest of these is the Pärvie fault, at 160 km in length. These are inferred to have ruptured in one event, suggesting a magnitude 7–8 earthquake. This hazard also applies to any other sufficiently northern region.

While models have been constructed to demonstrate the effect of loading and unloading the bedrock with ice, these models are currently not sophisticated enough to be able to predict the likely location of any glacially induced faulting. As these movements are likely to reactivate old features, the Swedish waste disposal program does not allow waste emplacement within a certain ‘respect distance’ of major fracture zones.

Volcanic Eruptions

Volcanic eruptions, like earthquakes are concentrated in active zones. Most continental volcanoes are situated above subduction zones or areas of extension. These areas are also very seismicly active, potentially ruling them out as candidate locations. A third series of volcanoes, however, can occur in the centre of stable continental crust. These occur over ‘hot spots’ mantle plumes, and examples include Yellowstone in the U.S.A and the Eiffel volcanic province in Germany. They are thought to be the surface expression of mantle plumes, bringing higher temperature mantle from deeper within the earth to the base of the lithosphere.

While, like earthquakes, most damage to a repository is caused by a direct hit, a nearby eruption or intrusion would have important hydrogeological consequence. The presence of a heat source would set up hydrothermal circulation, which could affect the repository. Hydrothermal circulation would heat the groundwater and change its chemistry, both of which could increase the speed of corrosion of the engineered barriers and speed the release of radionuclides. Hydrothermal circulation often leads to the alteration of the rocks it passes through, potentially changing mechanical and chemical properties of the rock surrounding the repository.

The initiation of volcanic activity in a new region, through the impact of a new mantle plume or the commencement of a new tectonic regime, have occurred in the past on the timescales of 10s – 100s of millions of years globally. The probability of a new volcano erupting in a region with no geologically recent volcanic activity is so small it can be neglected. Regions where recent volcanism has taken place, however, need to be carefully studied to assess the extent of volcanic risk.

Yucca Mountain is an example of a repository site in a region of recent volcanism. Silicic volcanism occurred between 15 and 8 million years ago at the Timber Mountain caldera complex, to the north of Yucca Mountain. Since 8 million years ago basaltic volcanism has continued, and although the region is in a fairly quiescent period, there have been three eruptions in the last 80,000 year. Reoccurrence times of 3.7–12 events per million years are commonly used in probabilistic prediction studies. These suggest that there is only a small of a volcanic event disrupting the repository during its working lifetime.

Japan again presents a problem for nuclear waste disposal. Because of its location over subduction zones, much of area is volcanic. NUMO’s exclusion criteria prohibit sites within 15 km of a Quaternary volcanic centre. Closer examination of Japanese volcanism, however, suggest that there are patterns. Subduction zone processes responsible for the generation of magma are thought to be stable over periods of millions of years. A volcanic front, a region dense in volcanoes, forms parallel to the trench. Although the reasons for this front are not well understood, dating of volcanic products shows that they are stable for millions of years. Trenchward of these fronts volcanism is greatly reduced. In the back-arc region of Japan, volcanoes are clustered along east-west trending zones. These correlate with gravity troughs and regions of low S-wave velocity, along with mountain chains. This suggests that they are regions of warm, upwelling mantle and that the spacing of these zones is related to underlying mantle convection patterns. In between these zones volcanism is again reduced. Were Japan to construct a nuclear waste repository, it would need to be in one of these regions of little volcanism.

Models have been constructed to show the impact of a direct intrusion into a repository. These are mainly concerned with Yucca Mountain. While the dynamics of the magma itself are well developed, it is unclear how much of the nuclear waste would be transported to the surface. The use of dead-end tunnels may stop circulation of the magma and limit transport of waste to the surface. However, with the capability to disperse any waste that reaches the surface over a wide area, volcanism should still be avoided.

Uplift and Erosion

Erosion has the potential to expose the repository at the surface, allowing nuclear waste to escape. If this happens before the activity of the waste has decayed to a safe level, this is dangerous.

Erosion takes place primarily on mountainous areas, and regions undergoing rapid uplift. Parts of Japan are ruled out, as the uplift rate would uncover any repository before the waste had decayed to safe levels.

Erosion also is a concern at Yucca Mountain. Erosion has been calculated to reach the level of the repository in between 500,000 years to a 5 million years. While this is the time until stream incision reaches the repository level, negative effects could occur much earlier. By removing the overburden the hydrology of the repository rock will potentially be substantially altered.

Switzerland has a major issue with erosion. Studies on the impact of erosion on possible depositories have been carried out, some in the fore-Alps near Oberbauenstock and Wellenberg on mountainous terrain, and others on deep disposal in crystalline basement or clays in the north. In the Alpine studies, uplift and erosion was estimated at between 0.2–2 mm/yr, with a ‘best guess’ of 1 mm/yr. This would uncover a repository at 500 m depth within 500 000 years. In order to model the worse-case impacts of exposure, it was assumed that the repository would be uncovered by 100 000 years. Although this did lead to did lead to exposure of anyone living at the surface, it was only slightly above the 0.1 mSv/yr specified in the regulatory guidelines and below background radiation levels. It was also assumed that the population in this area would be sparse, with only about 80 people currently living in the region.

The other scenario studied by the Swiss was the change in hydrology due to erosion of the Rhine to the north. This could affect the hydrology of the suggested depositories, especially if the Rhine changed its course to the south to directly overlie the crystalline basement containing the suggested repository. Again, the calculated exposure was low, 0.002 mSv/yr. ANDRA also predicted the erosion of the Paris Basin around the Meuse/Haute-Marne laboratory 1 million years into the future. Although there is little chance of the repository itself becoming exposed, they show that erosion will cause a change in direction of the flow in the over- and underlying aquifers, due to new contacts with the surface being formed. This could potentially increase the speed at which radionuclides are transported to the surface after escaping from the Callovo-Oxfordian layer the repository is situated in.

Although erosion is one of the smallest forces acting on a repository, it is often one of the most pressing in the mind of the public. The large timescales repositories are expected to maintain containment for means that erosion cannot be ignored.


Repository sites have been suggested in many varied geological settings around the world. Sweden and Finland have both selected sites for the construction of their final repositories, in Forsmark and Olkiluoto respectively. They are both situated in crystalline basement, on the stable Fennoscandian Shield. France have extensive research on a clay formation in the Paris Basin at Meuse/Haute-Marne and. In the U.S.A. tuffs of Yucca Mountain had been adopted as the site for nuclear waste disposal, although the Obama administration has recently cast doubt on this.

An important consideration in the siting of repositories is the tectonic stability of the surrounding region. Cratons, which have often been relatively undisturbed for up to a billion years, provide ideal candidates for repositories. There are, however, issues with these sites. There are often well developed fractures, and in Fennoscandia in particular there is the prospect of glacially induced faulting. The Canadian shield would also be affected by this. Large basins, such as the Paris Basin, can also be relatively stable. The Paris Basin is far enough away from active tectonic regions like the Rhine Graben and the Alps to have low seismicity.

Volcanic activity should also be avoided. Regions where volcanic activity are common do not make good candidates for repositories, however political motivations can sometimes dominate. This is true of Japan, where the entire country is volcanically active but exporting nuclear waste is politically difficult. Political reasoning may also have lead to the selection of the Yucca Mountain site in the U.S.A., which was already a site of nuclear testing, despite the recent volcanism there.

Erosion also has to be considered. Studies on mountainous regions such as Yucca Mountain and the Swiss Alps have shown that erosion can expose repositories in 100 000s of years, within the timescale for the radioactivity of the waste to decay to safe levels. Erosion also plays a part in changing the hydrogeology of the repository rock.

Most of the research on nuclear waste repositories, however, has concerned the water flow around the repository. Water is the chief agent of the corrosion
of the engineered barriers, and likely to be the main from of transport to the surface for the radionuclides. The chemical properties and circulation of groundwater is therefore one of the foremost challenges for any suggested repository site. Low permeability and few fractures are positive qualities in a host rock, which limits the selection to crystalline or clay-rich rocks. An ability to restrict the movement of radionuclides is also favourable, and is shown by the tuffs at Yucca Mountain and the argillite at Meuse/Haute-Marne.

Overall, there are many suitable geological formations for nuclear waste disposal. A unique blueprint, however, is not available; each site must be considered on its own merits. Current storage facilities can keep nuclear waste safe for decades to centuries, so although disposal facilities will be required that last longer, the situation is not disparate. This time will allow research to be carried out that ensures that any waste entered into a repository stays there until safe.

Further Reading

During my coursework I came across a couple of really good resources.  Free on the web is a peer-reviewed report commissioned by ANDRA, Dossier 2005.  It covers the research conducted at Meuse/Haute-Marne in great detail, and also has covers quite a bit of research from other parts of the world.

Also very useful was the following book:
Connor, C.B., Chapman, N.A.&Connor, L.J. eds., 2009. Volcanic and tectonic hazard assessment for nuclear facilities, Cambridge: Cambridge University Press.