In 2010 the Swedish Nuclear Fuel and Waste Management Company, SKB, plans to submit an application for a final repository for spent nuclear fuel. The proposed method, called KBS, involves depositing nuclear waste deep into the bedrock in the municipality of Östhammar on the Swedish east coast. It is based on three protective barriers: copper canisters encapsulating the spent nuclear fuel, bentonite clay surrounding the copper canisters, and finally the rock itself.

The geological environment surrounding the copper canisters will be nearly oxygen-free, and one of the premises on which KBS-3 rests is the assumption that copper cannot corrode in such an environment. The scientific findings of a small group of researchers at KTH (the Royal Institute of Technology) in Stockholm, that copper actually can corrode in pure water, free from oxygen as well as from complexing ions [1,2], have therefore been met with widespread interest and debate. Since the release of the KTH findings, the long-term safety of the KBS-3 method has been questioned. SKB has, however, refuted the findings [3].

The scientific controversy

Experimental results published by Gunnar Hultqvist and Peter Szakálos, researchers at KTH (the Royal Institute of Technology) suggest that copper metal can be corroded by pure water without any complexing ions such as sulphides and chlorides [1, 2]. The researchers also challenge the assumption of thermodynamic immunity of copper in water at elevated temperatures. Hultqvist and Szakálos refer to both experimental and theoretical observations and argue that these observations are consistent with their theory that pure oxygen-free water corrodes copper. In the process, hydrogen ions are reduced to hydrogen atoms and a corrosion product is formed whose identity and exact composition are not yet known. According to the research team, the hydrogen atoms will either form hydrogen gas molecules or be absorbed by, and diffuse into, the copper metal. The researchers also concluded that when the partial pressure of hydrogen reaches a level of 1 mbar the copper corrosion ceased. The corrosion process continues only if hydrogen is removed from the system so that the hydrogen pressure is reduced under this critical level.

SKB does not find the scientific evidence for the proposed reaction mechanism convincing; it concludes that no convincing evidence exists that water oxidizes copper [3]. It will nevertheless include the effect of such a corrosion mechanism in the safety assessment, even though its opinion is that the actual corrosion mechanism will not limit the lifetime of the canisters in the repository.

But what is the basis of this controversy? One important element is opposing views with respect to interpretation of thermodynamic data. Hultqvist, Szakálos and others claim that thermodynamic data prove that copper is not immune in pure oxygen-free water. Hence, their result is actually thermodynamically expected, although at a slow reaction rate. One weakness in the work of Hultqvist and Szakálos is, however, that the corrosion product of their proposed reaction has not been identified.

Corrosion reactions

In a KBS-3 type final repository, the copper canisters will be surrounded by a buffer of bentonite clay. The bentonite will gradually be saturated by groundwater. This gradual saturation process may extend over centuries and is a prerequisite for the assumed buffer safety function. A limited amount of oxygenated water is initially present in the bentonite, but the groundwater at that depth (about 500 m) is oxygen-free. There is also oxygen in the air that is present in the pore system of the bentonite before it is saturated. In the KBS-3 repository, copper will react with the oxygen that is present in the bentonite and in the groundwater and form copper oxide. This is scientifically known. When the oxygen is consumed, copper will react with water containing sulphide ions that are naturally present in the bentonite and the groundwater. Copper species including sulphide are the products of these reactions. The corrosion reaction is slow, and SKB has carried out research aimed at minimizing the formation and transport of these ions in the repository, and studied how different concentrations of sulphide affect long-term safety. When oxygen (O2) is present in the repository system, copper will react with oxygen to form copper oxide. When the oxygen is consumed, copper reacts with sulphide ions present in the water to form copper sulphide. A third corrosion reaction involves chloride ions present in the water and forms copper hydroxyl chlorides. The latter reaction is however only expected to be significant in low-pH environments. A consensus prevails regarding these reactions.

However, Hultqvist and Szakálos suggest that another reaction can occur in an oxygen-free environment, namely that copper reacts with the water molecules to form copper hydroxide species and hydrogen gas. In this reaction, hydrogen may also be dissolved in the copper metal:

Cu + water molecules → Cu hydroxides + H2 + H (in Cu) [1]

Reaction [1] is disputed among scientists. However, Hultqvist claims that they have evidence for a reaction product that is solid and porous and that molecular hydrogen can be measured in the gas phase. He also states that hydrogen is absorbed in the metal and that it can be measured. Hultqvist argues that since the pressure of hydrogen in his experiments (10-3 bar at a temperature of 20-80°C) is higher than the natural H2 pressure in air (5·10-7 bar), it can be concluded that copper is corroded by water.

Is copper corrosion in oxygen-free water thermodynamically possible? When a metal corrodes, it disintegrates into its constituent atoms, which are ionized due to chemical reactions (most commonly electrochemical oxidation) with its surroundings (in the presence of an oxidant such as oxygen). Typically oxide(s) and/or salt(s) of the original metal are produced as a result of such a reaction. Thermodynamics deals with the concepts of energy and entropy and can tell us whether a reaction is possible or not. The first law of thermodynamics states that energy cannot be created or destroyed, it can only change forms (that is, the energy of an isolated system is conserved, or in other words, constant in time). Hence, heat supplied to a system must equal the increase in internal energy of the system plus the work done by the system. Entropy is a measure of how organized or disorganized a system is. Thermodynamics states that the entropy of an isolated system which is not in equilibrium will tend to increase over time to a maximum value at equilibrium. This is the second law of thermodynamics. A special case of the second law is the concept of Gibbs free energy, which states that if the pressure is constant, a process will occur spontaneously if the change in Gibbs free energy is less than or equal to zero. In order to make a correct prediction, all the reactants and products in the chemical reaction in question must be determined with respect to their chemical identity. As mentioned earlier, the corrosion product in the reaction proposed by Hultqvist and Szakálos has not been identified, which makes it impossible to predict its spontaneity in terms of Gibbs free energy.

Szakálos claims that it is an undisputed fact among thermodynamics experts that copper is not thermodynamically immune in pure oxygen-free water. Among corrosion scientists, however, this is controversial. Szakálos also states that their experimental results do not conflict with known thermodynamic principles with respect to the corrosion of copper in water. The results can be explained by the formation of an amorphous copper hydroxide. He asserts that several scientific publications suggest the existence of different amorphous hydroxides, including both monovalent and bivalent copper, which can easily be converted to oxides. According to Szakálos, copper corrosion in oxygen-free water is a well-known phenomenon in industrial copper cooling systems and synchrotrons. All cooling systems for power generators and accelerators, such as at CERN in Switzerland, corrode in the region of a micrometer per year. This occurs in water that is deionized and degassed. The industry tries to reduce these corrosion rates and to achieve oxygen-tight metal fittings, such as UHV fittings. Nevertheless, the corrosion is still on the order of a micrometer per year. He illustrates the problem with the presence of partial plugging in the cooling systems by corrosion products such as oxides and hydroxides. The system clogs in a few years’ time. The environment makes the copper hot, around 70 to 90 degrees, which is about the same temperature that the copper canister will attain. In the industrial systems there is, of course, no groundwater, but the water is pure.

SKB and several experts assert that the corrosion process suggested by Hultqvist and Szakálos, and specifically the claim that a new stable phase is formed, challenges some of the basic principles of thermodynamics. However, SKB contends that even if this reaction should occur, it would be of negligible importance and the extent of copper corrosion would still be determined by the amount of sulphide and chloride ions.

Different types of experiments

There are different types of studies that can be used to explore the different aspects of corrosion: laboratory studies, in-situ experiments and analogues (i.e., natural or man-made artefacts). The initial state of a laboratory experiment is well-known, and control of the environment in the experiments is very good. On the negative side, it is less representative of real-world corrosion, due to the simplified system and the short time scale. It is good to get quick results, but it is harder to evaluate the long-term effects. In-situ experiments are investigations in realistic environments, and therefore closer to reality than laboratory results. In this case the initial state is quite well-known and representativeness is rather good. It is possible to perform both short-term and medium-term experiments. In the case of the analogues, the initial state is unknown and the environment cannot be controlled because the reaction has already happened when analogues are examined. However, the analogue entails a long reaction time; some parameters can be measured; and representativeness can range from poor to good, depending on the aspect under consideration. All experiments have weaknesses. In weight-loss experiments it is possible to determine how much has disappeared after a certain time, but it is not possible to distinguish between what happened in the initial phase, which might have occurred quickly, and what occurred over a longer period of time. Another weakness is associated with measuring corrosion depths and corrosion rates in specimens, since it is not possible to differentiate between different mechanisms that may have operated over different lengths of time. This makes it difficult to extrapolate experimental results for use in the safety assessment.

Experimental evidence

In one of the experiments performed by Hultqvist, two glass vessels holding copper foils were kept submerged in pure oxygen free water for 15 years. One glass vessel had a membrane of platinum so that H2 was not evacuated. In the other glass vessel, H2 was removed through a membrane of palladium. The latter vessel shows signs of corrosion: the foils turned black in colour. Hultqvist argues that if H2 is removed, which is always the case in an open system, this corrosion must be expected to happen. Hultqvist and Szakálos argue that the atomic hydrogen that is formed also can be absorbed into the copper metal. Hultqvist et al. have published proofs using two methods. Hydrogen was detected both by secondary ion mass spectrometry and by a quantitative study on outgassing in vacuum. Secondary ion mass spectrometry is sensitive to hydrogen. It is concluded by Hultqvist and Szakálos that the process of copper corrosion in water has been verified by experimental results, such as the formation of hydrogen, increase of weight, hydrogen in the copper metal, chemical analysis of the corrosion product, as well as by visual inspection and metallographic examination.

SKB’s work on copper corrosion includes literature reviews and various kinds of experiments, both short-term and long-term (gas measurements, simpler glass container experiments as well as electrochemistry). Moreover, theoretical calculations are performed that examine equilibrium reactions in water. SKB has found that the corrosion rate decreases with time in the short-term electrochemical and laboratory experiments. Very few, if any, results indicate that the corrosion rate increases with time. In the in-situ experiments, copper(II) corrosion products are often found, indicating that the copper in the experiment has undergone periods of oxidizing conditions. The analogue experiments show that copper in its native form has remained stable for a very long time in both the natural state and engineered artefacts.

One of SKB’s ongoing studies of copper corrosion includes first-principle computer calculations of the thermodynamic properties of Cu-O-H phases [3]. The main objective of the calculations has been to look for a stable product (or phase) between copper, oxygen and hydrogen that could be the final product of the supposed reaction between copper and water that is suggested by Hultqvist and Szakálos. Other objectives are (a) to calculate, from first principles, the thermodynamic properties of known Cu(I) compounds with oxygen and hydrogen, and (b) to analyze the thermodynamic stability of copper and its compounds in oxygen-free water environment. The compounds of copper(I) with oxygen and hydrogen are copper oxide (also known as cuprite, Cu2O) and copper hydride, respectively. Cuprite is a stable and well-known substance with regard to both its chemical and electrical properties. Copper hydride, on the other hand, is a less-studied phase. It does exist, but it is very unstable and loses hydrogen quickly with time. In the study, computer calculations were performed in an attempt to reproduce the experimental information on cuprite and copper hydride. They show that the reaction with copper and oxygen is energetically favourable and that the obtained thermodynamic properties of cuprite are in quite good agreement with existing experimental data. More or less good agreement with experiment was seen also for the hydride, although the reaction is shown to be unfavourable thermodynamically. Hence, it is known that the copper oxide is stable but the hydride is unstable.

The next part of the study was to search for other possible stable Cu-O-H phases. The study showed that copper oxyhydride is not a stable configuration. It also showed that copper hydroxide is a quite unstable species and its formation energy (ΔG) is experimentally known to be positive. If the copper hydroxide is condensed into a solid phase, it was found that the most stable structure was a combination of the structure of cuprite and the structure of ice, ‘cuprice’. The hydroxide has a reasonable electronic spectrum compared with the spectra of cuprite and copper hydride. However, if the stability of copper hydroxide is compared with that of cuprite and water, it is found to be unstable. Thermodynamically speaking, it should spontaneously decompose into cuprite and water. Hence, according to SKB’s study, cuprite is still the most stable of these compounds. Copper hydroxide may exist as a meta-stable phase, but its thermodynamic properties indicate that it is unstable compared with cuprite and water.

The use of archaeological analogues

One argument that is used by Hultqvist et al. to support the idea that water can corrode copper are the copper coins from the warship Vasa. The coins have been exposed to water for over 330 years and have diminished in size. Hultqvist argues that this is due to the fact that copper is corroded by the water itself and not by sulphide, since copper sulphide has extremely low solubility.

Hultqvist’s interpretation has, however, been criticized by SKB and others, who claim it is the presence of sulphide that has caused the corrosion of the copper coins. SKB also use archaeological analogues in their arguments. It claims that the bronze cannons from the warship Kronan, which sank in 1678 and were raised in 1986, are good objects to study, because the environment—the sediment of the Baltic Sea, considered to be oxygen-free and brackish—is similar to what the copper canisters will be exposed to in Swedish repositories. Szakálos disagrees that the cannons are analogous, since the corrosion of bronze differs fundamentally from that of copper. Szakálos argues that because the formation of passivating tin on the bronze surface greatly reduces the corrosion rate in aqueous environments, the corrosion rate on these cannons is around 1,000 times slower than can be expected with pure copper.

What additional information is needed to confirm this specific corrosion process and to assess the importance of the process for the final repository? Szakálos argues that the situation at the planned repository at the Forsmark NPP site is complex and threatening from both corrosion and embrittlement points of view. The copper canisters will initially be exposed to atmospheric corrosion until the oxygen is consumed. Then there is corrosion by water, sulphide, salt, stress corrosion cracking, intergranular corrosion and evaporation-induced corrosion. Szakálos concludes that before the KBS-3 concept can be accepted, it needs to be tested in realistic conditions. He quotes an SKI report from 1996: “Copper of identical composition as the future canisters should be placed in a future site environment, with artificial heating at about 80 degrees, with bentonite, etc. Such an experiment could be monitored for several decades.” He finishes by saying that the problem with copper is that it reacts slowly with everything.

The experts in the panel point to the necessity of further research on this topic to be able to assess whether the results of Hultqvist and Szakálos are realistic or not. Both Ron Latanision and Gaik Chuah argued that hydrogen may be produced by corrosion of copper in oxygen-free water, but that it is essential to know that the corrosion product is thermodynamically stable and that it can be identified and characterized. Latanision refers to the fact that there are a number of sophisticated surface analytical techniques that should be used to demonstrate that the proposed reaction products are indeed formed. Digby Macdonald emphasizes that the kinetics must also be examined, that is, both the corrosion mechanism and how fast the reaction occurs. He also stresses the importance of knowing that the water in the experiment is pure, since even very small amounts of monovalent copper ions and dissolved hydrogen gas in the water are of great importance. Considerable care must be exercised when designing experiments aimed at demonstrating copper corrosion to ensure that corrosion is spontaneous upon initiation of the experiment. Chuah suggests that more experiments should be carried out with the aim of (i) confirming or disproving the formation of hydrogen through direct detection by mass spectrometry; (ii) studying the experimental conditions of Hultqvist and Szakálos under which hydrogen is formed; (iii) examining the reaction products formed on copper using in-situ methods to avoid any phase transformation on exposure to atmospheric conditions; (iv) measuring the strength of the exposed copper; (v) quantifying the thickness of the corrosion layers as a function of time; and other relevant tests. Latanision suggests that definitive well-controlled experiments should be carried out at a third-party laboratory. He also suggests that a third-party institution should test the susceptibility of copper to embrittlement caused by absorbed hydrogen in copper tensile specimens that are electrolytically charged with hydrogen at cathodic current densities that correspond to the corrosion rates associated with the measurements reported by the KTH team.

Furthermore, copper is an obvious material to consider for disposal in chemically reducing environments. If copper corrosion in anoxic environments is observed and confirmed as described in the research, the conditions which have led to corrosion need to be clarified and then either controlled or engineered out of the repository environment. Macdonald argues for more research on the common contaminant bisulphide ions, HS-, in groundwater and the corrosion that it causes. He says that a much more detailed analysis is warranted to fully define the conditions under which immunity of copper might be expected to exist. He also recommends what such an analysis should include. David Shoesmith points to the fact that some of the references cited by Szakálos do not stand up to scrutiny, and that Hultqvist and Szakálos have made an incomplete analysis of the available literature on corrosion of copper cooling systems. He states that a review of the literature provided for the workshop, the presentations made, and a personal search of additional literature indicate that there is no evidence that significant corrosion of Cu can be sustained by water reduction.


It can be concluded that in theory copper may be corroded by pure oxygen-free water with respect to the following reaction:

2 Cu(s) + H2O ↔ Cu2O + H2 [2]

Hydrogen atoms or molecules must be a reaction product, in addition to the possible formation of copper hydroxide or copper oxide species, because only protons can possibly accept electrons from copper atoms. Hence, the key issue is how far to the right reaction [2] can proceed, and the partial pressure of hydrogen that is obtained at equilibrium. According to Hultqvist et al., the hydrogen pressure is 1 mbar, which is much higher than the natural partial pressure in air. Then the corrosion reaction will proceed until the equilibrium pressure is attained, and if the hydrogen is continuously removed the corrosion may be extensive. However, thermodynamic calculations result in an extremely low hydrogen pressure at equilibrium, which is the theoretical justification for the general assumption that copper is resistant to corrosion in pure anoxic water. If it is assumed that the system is so open that H2 can be continuously removed from the reaction area, the reaction will be shifted to the right and corrosion will be favoured.

There are however a number of questions that need to be answered regarding the experiment and results presented by Szakálos and Hultqvist. One such issue is that the proposed corrosion product needs to be identified and characterized. If the corrosion product is chemically characterized, the formation energy can be computed and the overall spontaneity of the corrosion reaction calculated. Then a new substance has to be added to the thermodynamic data tables. Moreover, based on available information presented by Szakálos and Hultqvist, it is not possible to state what the proposed reaction means for a copper canister in the repository environment. It is for instance unclear what the presence of the bentonite buffer will mean for the removal of H2. It is also unclear how complexing ions such as sulphide and chloride that are present in the repository environment will influence the proposed reaction. Hence, there are a number of issues that need to be addressed. The yet unknown corrosion product needs to be identified, the corrosion rate needs to be determined, and corrosion in a realistic repository environment needs to be studied. The panellists also said that further research is required to clarify the experimental results and the analytical methods used by Szakálos and Hultqvist.


This article is an edited extract of “Mechanisms of Copper Corrosion in Aqueous Environments,” a report from a scientific workshop organised by The Swedish National Council for Nuclear Waste (Kärnavfallsrådet) on 16 Nov. 2009;
Workshop panel members include: Ron Latanision, corporate vice president and director of Exponent’s mechanics and materials practice; Gaik Khuan Chuah, teaches and supervises students at the National University of Singapore in heterogeneous catalysis; Digby D. Macdonald, distinguished professor of materials science and engineering at Pennsylvania State University; David Shoesmith, professor in the department of chemistry at the University of Western Ontario, who holds the Ontario
Power Generation industrial research chair in nuclear fuel disposal chemistry; David J. Duquette, the John Tod Horton professor of materials science and engineering at Rensselaer Polytechnic Institute (moderator)
[1] P. Szakálos, G. Hultqvist, and G. Wikmark, Electrochemical and Solid-State Letters 10 C63 (2007)
[2] G. Hultqvist, P. Szakálos, M. J. Graham, et al, Catalysis Letters 2009, Volume 132, Numbers 3-4, 311-316. (Catalysis Letters also published a criticism of the article by
L. O. Werme and P. Korzhavyi (2010, Volume 135, Numbers 3-4, Pages 165-166), and a reply by Hultqvist et al (166-167).
[3] P. Korzhavyi, B Johansson, ‘Thermodynamic properties of copper compounds with oxygen and hydrogen from first principles’ SKB TR-10-30,
[4] O. Karnland, S. Olsson, A. Dueck et al, ‘Long term test of buffer material at the Äspö Hard Rock Laboratory, LOT project’, SKB TR-09-29,
[5] Svensk Kärnbränslehantering AB, ‘Long-term safety for KBS-3 repositories at Forsmark and Laxemar – a first evaluation. Main report of the SR-Can project’, SKB TR-06-09,
See also: “Nuclear Waste State of the Art Report 2010—Challenges for the Final Repository Programme” (