ÐÓ°ÉÔ­´´

GIVEN THE CHOICE between living next to a pig farm or a deep repository for nuclear waste, most people would choose the pigs. ÐÓ°ÉÔ­´´s can show that farming is considerably more of a risk to health and the environment than an ideal repository, but most people are unconvinced. Perhaps this is because the researchers show what they mean by safety only through complex mathematical models that predict releases of radioactivity in the far distant future, on timescales which few people can imagine. One solid way to explain safety to both the sceptical specialist and the concerned neighbour is to show what happens to naturally radioactive rocks through geological time.

James Hutton, one of the founding fathers of geology in the 18th century, rejected the idea of experiments to simulate geological processes on account of ‘the immensity of the natural agents’, which, he believed, worked in ways far beyond our reach. Geologists have resorted to analogues ever since. Natural analogues provide evidence that an underground repository could cope with nuclear waste. In addition, many aspects of the models that demonstrate their safety can be tested on scales of time and space that are inconceivable in a laboratory experiment.

The main components of a deep disposal site are the waste, the canisters that enclose it, the filling and liners of the tunnel, which protect it, and the rocks in which the whole system sits. Such a repository should remain intact until slow, gradual processes such as erosion, or very unlikely catastrophes such as a volcanic eruption or a meteorite landing on top, eventually release its contents. Safety assessments consider what might happen to a site in a future that could include these catastrophes, ice ages and other changes in Britain’s climate (see ‘A million years in the life of a waste site’, New ÐÓ°ÉÔ­´´, 15 October 1988). But the main threat comes from water, which will almost always flow through the repository and slowly breach its barriers.

These physical barriers are the first line of defence. Although at first they isolate the waste, they will inevitably fail through chemical corrosion, physical breakage or even the action of microbes. Once water has penetrated, the waste will begin to dissolve and migrate out into the rock. The rate at which radionuclides escape depends on how quickly water reaches them, how fast they dissolve and how much they interact with their surroundings, a process known as sorption. The dissolved isotopes move through cracks, fissures and pores in the surrounding rock, which is chosen for its low permeability. Because there is a large volume of rock compared with the small quantities of waste, sorption can be an important factor in slowing down the escape. Once the barriers fail, the chemistry of the radionuclides, and their interactions with the surrounding rocks, control when and at what levels isotopes from the waste reach the biosphere.

Some of the earliest analogues for movement of radioactive elements over millions of years are the natural nuclear reactors at Oklo, in Gabon, which hit the scientific headlines in the 1970s. This deposit of uranium ore, formed more than 2000 million years ago, turned out to have an extremely unusual composition. In some of the richest ore bodies in the open cast mine, researchers found far less of the fissile isotope uranium-235 than they expected, relative to the amount of the more common uranium-238. In other natural uranium, this ratio is constant; such depletions could be explained only by nuclear fission in a natural chain reaction, accelerating the disappearance of the lighter isotope.

Oklo’s chain reaction happened in an unusual geological situation. The ore was very rich in uranium, and, originally, contained enough of the lighter isotope to start the chain reaction. Uranium-235 has a half-life of 713 million years, and so decays much more quickly than uranium-238, with a half life of 4150 million years. About 2000 million years ago, the natural concentration of ²³⁵U relative to ²³⁸U was over 3 per cent – about the same as in today’s artificially-enriched reactor fuels. The combination of a high total content of uranium (between 50 and 70 per cent uranium oxide), enrichment in ²³⁵U and a suitable geochemical environment allowed the chain reaction to take place, consuming **²³⁵U and leading to the depletion now seen at Oklo.

The chain reaction happened intermittently for more than 500 000 years. Despite the temperatures of more than 600 degrees, and the radiation damage which the rocks and minerals sustained, the Oklo deposits have behaved like a natural store for spent reactor fuel in the 2000 million years since. Many of the ‘waste’ radionuclides produced in the ‘fuel’, such as plutonium, neptunium and thorium, either stayed where they formed or moved only short distances in the surrounding rock before they decayed to stable isotopes. But some of the radionuclides produced by fission were lost from immediately around the deposit. These included noble gases, and isotopes of elements which are highly soluble and poorly sorbed, such as halogens.

The way in which these radionuclides appear to have escaped matches predictions made in safety assessments of model repositories. Research on how these tricky elements behave will enable designers to plan how best to contain them. Additional engineered barriers should limit the release of even the more mobile elements from a repository.

Although containing a fair amount of naturally produced ‘waste’, the geology at Oklo, with radioactive rocks at great depths, and free circulation of water, bears little resemblance to the environments being sought for disposal of spent fuel today. To find a more appropriate example, geologists have to go to northern Canada. Near Cigar Lake, in Saskatchewan, prospectors discovered a 1300-million year old uranium deposit, 450 metres below ground. They found the site in the course of an electromagnetic survey which picked up graphite, a conducting mineral often found with ores. The uranium ore is extremely high-grade, containing 14 per cent uranium oxide by weight on average, and more than 40 per cent in places. This is so rich that the uranium may have to be extracted by remote control, because of the risk to miners from radiation. By chance, the geology of the Cigar Lake ore body and its surrounding rocks looks similar to repositories proposed for spent nuclear fuel in Sweden, Finland and Canada.

Remarkably, there are no clues at the surface at Cigar Lake, for example in streams or groundwater, to indicate that this uranium is there. Despite the fact that the rocks have been saturated with water for around a thousand million years, the uranium has not moved significantly. This provides considerable reassurance to those modelling the very slow fluxes expected to arise around nuclear waste at depth.

The simplest analogues for the movement of radionuclides through rocks and groundwater systems are those where there is a clear source of radionuclides, and where the time during which they have been moving can be established clearly. This is the case at the bottom of Loch Lomond in Scotland. Modellers are interested in the rate at which radionuclides will diffuse through clays, which are often used to surround containers of waste. The muddy sediments at the bottom of Loch Lomond have settled out from the water of the lake over the past 10 000 years. They include a layer of silty clay deposited between 5400 and 6900 years ago, when the loch was connected to the sea. This marine layer is different from the freshwater deposits above and below it, containing, in particular, higher concentrations of uranium, radium, iodine and bromine. In the thousands of years since the sediments settled, the system has remained undisturbed; researchers can compare the distributions of different elements in the clay layers and their pore-waters with models of how they may have been redistributed by the flow of water or diffusion.

Iodine is one of the key elements in assessing the safety of a method of disposal, because its isotope iodine-129 is common in radioactive waste, and appears to move relatively freely in groundwater, reacting little with the rock. Consequently, iodine is often predicted to be one of the first radionuclides to return to our environment. In the Loch Lomond marine bed, however, iodine and bromine appear to have stayed where they were originally deposited, most probably through interaction with carbon in organic matter in the mud. They have not been free to migrate significantly for thousands of years.

The concentrations of elements through the layers of sediments can be analysed to calculate their diffusion coefficients. These numbers tell researchers how quickly an element such as iodine spreads through a fluid. The coefficients turn out to be comparable with, or lower than, those measured over weeks or months in the laboratory.

There are other important factors that also cannot be addressed solely through experiments. The fabric of the repository, the canisters and the tunnel fillings, will degrade as time passes. As an example, in most repositories the waste will be sealed in metal containers. Laboratory experiments that take only months or years have to provide data for models of corrosion lasting thousands of years. But researchers can test these models by comparing the corrosion predicted by experiments with that observed in artefacts left by earlier civilisations.

Provided that the archaeological record is very detailed, and the chemistry of the environment resembles that expected in a repository, tests can give quite precise values for parameters such as rates of corrosion . There are analogues for the degradation of the many forms of waste, types of container and their surrounding materials. Metal artefacts such as cannon or armour can act as models for the way in which metal canisters corrode. Volcanic glasses mimic the glass made from some types of waste, thick deposits of clay match the clay that lines tunnels and shafts, and bitumen, native copper and iron meteorites can be models for other parts of a repository.

The variety of types of waste and barriers means that the chemistry within a repository can be complicated; even so, there are usually suitable analogues. For example, waste encapsulated in bitumen and surrounded by concrete in a repository will be highly alkaline. This matches the chemistry of some spring waters in Jordan, which bubble up through muddy limestones that contain bitumen. The spring’s pH of more than 12.5 results from the cement mineral, portlandite, which has formed naturally within the rocks. Research on these springs could show how elements move through water and rock in a repository made from bitumen and concrete.

Containers of spent fuel from nuclear reactors will themselves alter the chemistry of water held in pores in the surrounding rock. Initially, this water acts chemically as a reducing agent. Once the metal container has corroded through, radiation and groundwater will interact in a process known as radiolysis which could make the water a strong oxidiser. The rock surrounding the containers will act as a buffer, consuming the oxidants in the water, and being oxidised itself. As the oxidising water diffuses outward, a ‘redox front’, the boundary between oxidising and reducing conditions, will move slowly through the rock. It carries behind it dissolved uranium and other elements from the fuel that are also more soluble in oxidising conditions. In the rocks, when some elements encounter reducing conditions at the front, they precipitate as minerals. This chemical situation arises in yet another natural analogue.

The area around the small spa town of Pocos de Caldas in the Brazilian highlands of Minas Gerais is well known for its hot springs and mineral deposits, including what was until very recently Brazil’s principal uranium mine. The most obvious feature of this ore deposit is an abrupt change in the colour of the rock. This represents a redox front; water rich in oxygen moved downwards from the surface, oxidising iron in the volcanic rocks from Fe2+ to Fe3+. This changed its colour from pale green, where minerals contain reduced iron, to red, representing hydrated iron oxides. As the redox front moved, the uranium in the rock dissolved under the oxidising conditions and precipitated again on the reduced side of the front, concentrating the element into an ore body. This movement could also happen around canisters of waste. The high concentrations of radionuclides in the rocks and groundwaters also make Pocos de Caldas an ideal place to test the geochemical and thermodynamic models being used to predict the solubility of isotopes in underground water.

Not far from the mine is one of the most naturally radioactive places on the surface of the Earth, a small hill called Morro do Ferro (the hill of iron). At the head of a small valley drained by a stream, there sits an ore-body containing about 30 000 tonnes of thorium and its daughter products, such as radium and rare-earth elements. The plants growing on top of the ore-body have absorbed so much radium-228 that they produce an image when placed on photographic film. The gamma radiation levels at the surface are so high that they would give someone on the top of Morro do Ferro an annual dose of 250 millisieverts. The average dose each year in Britain is about 0.4 millisieverts; living on top of this hill for about six weeks would give the equivalent of a lifetime’s British dose.

This site is being studied because it provides an opportunity to model the release and migration of thorium and radium in groundwater. Merril Eisenbud of New York University Medical Center, and his team, have used thorium as a chemical analogue for plutonium when dissolved in water. Thorium and plutonium both come from the same section of the periodic table, the actinide series, and their ions behave in similar ways. By monitoring water in streams around the Morro do Ferro, Eisenbud showed that the rate of mobilisation of thorium from the deposit has been only one part in 1000 million per year. If thorium is a realistic analogue for plutonium, then the concentration of plutonium in the streams draining this radioactive hot-spot would be a factor of 20 less than the permissible levels for drinking water set by the US government – even though the water flows through the highly weathered ore-body.

So, with all this information, how can researchers evaluate the risks of storing radioactive waste? One way is to apply the same sort of limits that control the day-to-day risks from radiation. Regulations are often based on limiting the dose of radiation likely for particular individuals or groups. Some people may receive a higher dose because of how they live or what they eat. Unlikely events, such as a repository suffering a direct hit from a large meteorite, would add to the risks from radiation, although the overall effects of an impact big enough to damage a deep repository would probably dwarf the risks arising from the release of radioactive isotopes.

Dose limits are usually set in terms of a numerical target value, always considerably less than the variation in dose that comes from the background radiation to which we are all exposed. Last year, Roger Clarke and Richard Southwood of the National Radiological Protection Board showed that, on average, someone in Cornwall receives an annual radiation dose three times the average in Britain; some people receive more than eight times. The difference is almost entirely due to the higher levels of radon produced by uranium decaying in Cornish granite.

The problem with using a radiation dose to understand the consequences of waste disposal is that it condenses predictions about processes and events far in the future into a single number. These predictions often have to assume that any people around to receive a radiation dose will live in circumstances similar to present day societies. To assess what will happen in the distant future, thousands of years hence, there has to be a different kind of target: ensuring that an ancient and degraded waste repository does not cause a significant increase in the natural radioactivity.

The National Swedish Institute of Radiation Protection has suggested that estimated rates of mobilisation of radioactivity from a repository could be compared with the rates at which erosion releases them from nearby rocks. This compares the concentrations of radionuclides in water at and near the surface, that erodes the neighbouring rocks, with levels in water deep underground, that has passed through the waste. The rates calculated for Morro do Ferro, together with what we know about the solubility of the elements in question, suggest that the two are likely to be broadly comparable.

Using this method, researchers can calculate how some of the longest-lived radionuclides in Britain’s proposed repository for low and intermediate level wastes will behave . Despite the large amount of uranium in the waste, it will contribute perhaps 1 per cent or less to the total amount of uranium released to the environment by erosion of the surrounding rocks over the next few million years. A few million years ago, humans had not evolved. It is hard to imagine, let alone quantify, the changes that people and society on Earth could undergo in that same period in the future.

People will have to live with radioactive waste for the foreseeable future. Without belittling the extra hazards of the radioactive isotopes that we have mined and concentrated for power, medicine, research, industry and arms, no one should forget that we have lived with radioactivity in the past. Everyone, especially those who make decisions about the future, must understand the results and limitations of safety assessments. Simple comparisons with natural examples, combined with rigorous assessment of how a repository performs, are going to be increasingly important as we peer uncertainly into the far distant future.

* * *

Box 1: Roman nails and Swedish cannon

AS THE Romans consolidated their hold on Britain in the 1st century AD, they abandoned their farthest-flung military encampments as the frontier of civilisation withdrew southwards. Although they took most valuables with them, they could not carry everything. Anything of military value, such as metal, they buried to hide them from the warlike native tribes.

In typical Roman fashion, many of these hoards were meticulously documented. For example, in the year 86, about 12 tonnes of iron nails were buried at the Roman fort at Inchtuthil in Scotland. Archaeologists have recovered more than 875 000 nails from this site. Although nails from around the outer few centimetres of the hoard are heavily corroded, many of those inside are almost intact, after nearly two millennia. So it is reasonable to expect the lifetime of massive steel canisters, perhaps 25 centimetres thick, in deep repositories, to exceed 1000 years.

On 1 June 1676, the Swedish warship Kronan sank in the Baltic Sea about 3 miles from the eastern coast of Oland. Archaeologists have recovered a bronze cannon from this ship, and analysed both it and the sediment, to see what happened during its 300-year-long burial.

The Baltic muds that surrounded the cannon excluded oxygen, and the cannon was made of more than 95 per cent copper. They calculated that the average loss of copper in this environment, without oxygen, was about 40 grams per square metre in 300 years, equivalent to a rate of corrosion of about a hundred-millionth of a metre per year. Even allowing for pitting, which would increase the maximum depth of penetration by a factor of five, a layer 5 centimetres thick would take a million years to corrode, much the same timescale considered in assessments of the lifetime of copper canisters that isolate the waste. These canisters would be at least 5, and possibly 10 centimetres thick. This study also showed that the copper leached from the cannon had migrated only about 4 centimetres in the 300 hundred years of its burial. This confirms the very low mobility of metals in environments rich in clay – the major reason why clay is a common material for packing around canisters in waste production.

* * *

Box 2: How to hide nuclear waste

WILL THE radioactivity from a repository thousands or millions of years after it is built be detectable against the natural background? The aim of the designers of such sites is that it will not. One way they try to achieve this is demonstrated by looking at the likely behaviour of the repository currently proposed for low- and intermediate-level radioactive waste by UK Nirex Ltd, the company responsible for dealing with Britain’s nuclear debris.

The design allows for disposal of about 19 000 tonnes of uranium in various isotopic forms, predominantly uranium-238. After hundreds of thousands of years, much of the radioactivity remaining in the waste will arise from the original uranium; other isotopes, with shorter half-lives, will already have decayed.

Current plans for disposal in hard fractured rocks in Britain place the repository at a depth of about 800 metres, with its caverns occupying an area of 85 hectares (about 700 by 1200 metres). A model for the flow of groundwater around the site, including dissolution of uranium from the waste and transport to the surface, might look at a block as big again as the repository on each side.

Hard basement rock in Britain, such as the Borrowdale volcanic series of the Lake District, contains about 1 part per million uranium. This block of rock would contain about the same amount of uranium as the proposed repository. At least for uranium and its daughter radionuclides, there seems to be some logic in comparing the amounts introduced by the wastes with those already present.

We have made some rough calculations of how much uranium escapes. One metre in 100 000 years is the average rate at which rock erodes. This releases natural uranium from the rock into our environment at a rate equivalent to 1 per cent of the uranium in the repository escaping every million years.

Researchers from the Harwell Laboratory, who have studied how uranium in the waste might dissolve and be removed from the repository, predict that it will release a steady 0.026 per cent of its total uranium every million years. The overall uranium content of the repository and the rock immediately surrounding it are equivalent, so the repository will contribute insignificantly to the background re lease of uranium in the distant future.

Neil Chapman is a geologist working for INTERA-ECL in Melton Mowbray, and used to live on a pig farm. Ian McKinley is with the Swiss National Co-operative for Radioactive Waste Disposal (NAGRA). They are the authors of The Geological Disposal of Radioactive Wastes, published in 1987.