ATOMIC ENERGY »"5Ä L'energie ATOMIQUE OFCANADA LIMITED Tibv DU CANADA LIMITÉE

Home , Zirconium alloy

AECL-8702

ATOMIC ENERGY »"5Ä L'ENERGIE ATOMIQUE OFCANADA LIMITED TiBV DU CANADA LIMITÉE

MECHANISMS OF HYDROGEN ABSORPTION BY ZIRCONIUM ALLOYS

Mécanismes d'adsorption de l'hydrogène par les alliages de zirconium

B. COX

Presented at 1984 Fall Meeting of the Materials Research Society, Boston, November 26-30. 1984

Chalk River Nuclear Laboratories Laboratoires nucléaires de Chalk River

Chalk River, Ontario

January 1985 Janvier ATOMIC ENERGY OF CANADA LIMITED

MECHANISMS OF HYDROGEN ABSORPTION BY ZIRCONIUM ALLOYS

B. Cox

(Presented at 1984 Fall Meeting of the Materials Research Society, Boston, November 26-30, 1984)

AECL-8702

Materials Science Branch Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0 Canada 1985 January L'ENERGIE ATOMIQUE DU CANADA, LIMITEE

Mécanismes d'adsorption de l'hydrogène par les alliages de zirconium

par

B. Cox (Rapport présenté à la réunion d'automne 1984 de Materials Research Society tenue à Boston du 26 au 30 novembre 1984)

Résumé Ce rapport résume la façon dont nous comprenons actuellement les trois mécanismes primaires par lesquels les isotopes d'hydrogène peuvent pénétrer dans les alliages de zirconium en service. Voici ces mécanismes: (i) Réaction avec l'hydrogène et adsorption de cet élément gazeux provenant d'atmosphères n'ayant pas suffisamment d'oxydant pour que soit maintenue la pellicule d'oxyde protectrice. (ii) Diffusion dans le métal d'une fraction de l'hydrogène dégagé dans la cellule partielle cathodique en cours d'oxydation dans l'eau ou la vapeur. (iii) Diffusion des isotopes d'hydrogène au travers d'une liaison métallurgique comprenant des métaux dissemblables formant un orifice traversant la pellicule d'oxyde normalement protectrice. Bien que ces trois mécanismes puissent être spécifiés qualitative- ment avec une certaine certitude et bien que des preuves circonstanciées soient souvent disponibles, on indique dans le rapport que les constantes de vitesse des diverses étapes des réactions sont essentiellement incon- nues.

Département de science des matériaux Laboratoires nucléaires de Chalk River Chalk River, Ontario, Canada KOJ 1J0 Janvier 1985 AECL-8702 ATOMIC ENERGY OF CANADA LIMITED

MECHANISMS OF HYDROGEN ABSORPTION BY ZIRCONIUM ALLOYS

B. Cox

(Presented at 1984 Fall Meeting of the Materials Research Society, Boston, November 26-30, 1984)

ABSTRACT

This paper summarizes our present understanding of the three primary mechanisms by which hydrogen isotopes can enter zirconium alloys in service. These are:

(i) Reaction with and adsorption of hydrogen gas from atmospheres containing insufficient oxidant to maintain the protective oxide film.

(ii) Diffusion into the metal of a fraction o.l: the hydrogen released in the cathodic partial cell during oxidation in water or steam.

(iii) Diffusion of hydrogen isotopes through a metallurgical bond with a dissi- milar metal which provides a window through the normally protective oxide film.

It is shown that while all three mechanisms can be specified qualita- tively with some certainty, and while supporting circumstantial evidence is often available, the basic rate constants for the various steps in the reactions are largely unknown.

AECL-8702

Materials Science Branch Chalk River Nuclear Laboratories Chalk River, Ontario KOJ 1J0 Canada 1985 January -1-

1 . INTRODUCTION

Delayed hydride cracking (DHC) of Zr-2.5 wt% Nb alloy pressure tubes was first observed in CANDU reactors in 1974 (Pickering-3) and 1975 (Pickering-4), and again in 1982 (Bruce-2) [1]. In the first two of these incidents, hydrogen absorption - as opposed to hydrogen in the tubes as installed - played no part in the process. Nevertheless, some unexpected hydrogen ingress into the tube ends was observed. The sudden failure of a Zircaloy-2 pressure tube in Pickering-2 in August 1983 [2], although crack propagation again was essentially by delayed hydride cracking until the critical crack length was obtained, was significantly influenced by hydrogen absorption after the installation of the tube. Thus, our earlier interests in hydrogen absorption mechanisms, which had been dormant for many years, were revived, not least by the large differences in hydrogen uptake behaviour shown by different zirconium alloys [3].

Zirconium alloys are normally protected against hydrogen ingress by the surface oxide film, which presents a good barrier both to ingress, and to the egress of hydrogen already in the metal. However, laboratory work has shown that, under conditions of straining at a notch, this oxide film offers little protection, and zirconium alloys are susceptible to rapid cracking (up to 10"^ m/s) in hydrogen gas [4,5]. These experiments revealed only small differences in the crack velocities of the same alloys which show surprisingly large differences in crack velocity under DHC conditions with only internal hydrogen [6].

As a result of this revived interest in hydrogen absorption, as a critical first step in any hydride cracking processes at elevated temperatures, earlier work on hydrogen absorption (some previously unpublished) has been reviewed, and new experimental programmes have been started. There are three fundamentally different processes by which hydrogen isotopes can enter a fabricated zirconium alloy component. The first, and most trivial mechanistically, is by diffusion into the zirconium alloy via a direct metallurgical contact with another metal having a high diffusivlty for hydrogen, and a higher fugacity of hydrogen than the zirconium. The other two ingress routes operate through the surface oxide film. The first of these occurs when zirconium is exposed in a hydrogen atmosphere containing insufficient oxidizing species to maintain the protective nature of the oxide, while the second occurs as part of the normal oxidation process in aqueous media. These three processes will be dealt with individually.

2. ABSORPTION OF HYDROGEN GAS

All zirconium components, unless held at high temperature in a good vacuum, carry a ZrÛ2 film on their surfaces. In air at room temperature, the thickness of this oxide is limited by electron tunneling to 3-5 nm. Thickening of this oxide can proceed by a variety of processes. By increasing the electric field across the oxide, further oxygen ion transport through the air-formed film can be initiated either at room temperature or above. This same process will proceed by thermal activation if the temperature is raised in the presence of an oxidant. Such an oxide film is normally a very good barrier against reaction with hydrogen gas and, provided sufficient oxidant is present in the environment, -2-

may remain so indefinitely. "Sufficient" in this context was established by Shannon [7] to be that required to maintain the normal oxidation rate at the temperature concerned (Figure 1); the "normal" oxidation rate being that obtained at environmental pressures sufficiently high for the pressure dependence of the oxidation rate to approach zero [8].

These observations led to the postulate that, for any conditions of temperature, pressure and oxidant, there should be a critical hydrogen/oxidant ratio above which the oxide film would remain protective, and gross reaction with hydrogen would be prevented [9]. There could still be a slow absorption of hydrogen by reaction with the oxidant if this were a hydrogen containing molecule. Several investigators [9,10] have established these critical ratios, for a limited range of environments, and they are generally observed to be in the range ÎO^-IO^ (Table 1). Thus, very small concentrations of oxidizing species in hydrogen gas are sufficient to prevent direct surface reaction with hydrogen at reactor operating temperatures. This effect was borne out in experiments on cracking in hydrogen gas where small oxygen additions were sufficient to inhibit crack growth [5]. However, it is notable (Table 1) that more oxygen appeared to be necessary to ensure passivation of strained metal at the crack tip, than appears to be the case for reaction on a smooth surface.

If insufficient oxidant is present to maintain the oxide film in good repair, then, after a very variable incubation time required for "breakdown" of the surface oxide film [10-13], a rapid direct reaction with hydrogen gas ensues (Figure 2). The progress of the oxide "breakdown" process can be monitored, during the incubation period, by measuring the electrical resistivity (Figure 3) of the oxide film [7]. The resistivity of the oxide declines rapidly with time, as a result of the increasing hypostoichiometry of the oxide, and hydriding commences once the resistivity declines below a critical level. The hypostoichiometry changes because the rate of dissolution of oxygen atoms from the oxide into the metal now exceeds the rate at which replacement oxygen atoms can be acquired from the environment. In addition to the increasing electronic conductivity associated with this process, an increase in the number of anion vacancies will result.

As a result of these observations, investigators studying the rate of reaction uf hydrogen with oxidized zirconium surfaces have been tempted to postulate that hydrogen is migrating via these anion vacancies as interstitial H2 [14,15]. From their experiments they have then calculated dif fusivitief; for hydrogen in the Zr(>2 lattice. However, there is no unambiguous evidence from these experiments to support an argument that hydrogen is migrating via the ZrÛ2 lattice at all. The diffusivities which they calculate are very high for such a lattice diffusion process, and lie in a range more typical of surface dtffusivities. Experiments of this type give permeation rates 10? or more times those measured by effusion experiments using tritium [16-18]. From these, a diffusion coefficient for tritium in the surface layer (arbitrarily defined as the first 5 Mm) of 10~6 cm2/s at 300°C was estimated [17]. This is in good agree- ment with a value of 3 x 10"^ which can be calculated from the autoradiographs in ret.19 for 400-500°C.

There is some circumstantial evidence that hydrogen entering the metal under these conditions does not pass through the oxide lattice. Tritium _ r> _

autoradiography [19] has shown that, when a specimen preoxidized in O9 or air exposed to T2 gas, tritium appeared in the metal core but the oxide film remained essentially free of tritium (Figure 4). Other experiments showed th tritium, built into the oxide fiitn during reaction with T2O, remained fixed o the specimen was cooled and did not migrate into the metal, or exchange with hydrogen in the environment, over periods of years [16] . This again suggests that the diffusivity in the oxide is orders of magnitude less than that calcu ted from the reaction rate experiments [14,15]. Conversely, tritium absorbed porous oxide films, by exposure to T7O, exchanged rapidly with hydrogen in th environment and vanished in periods of, at most, a few days [16]. Thus, at t very most, any tritium which was present in the oxide on the specimens heated T'i gas had exchanged with the laboratory environment, and vanished, by the time the autoradiograph was exposed. This indicates strongly that the T2, wh entered the metal in these experiments, did so via pores or cracks in the oxi and not by diffusion through the oxide lattice. Autoradiographs of similar specimens preoxidized in oxygen and thon exposed to T7O showed tritium both i the metal and in the oxide (Figr'e 5). This shows the fundamental difference between reaction with T2 and T2O.

The experiments of Shannon and others gave no direct clues to the micr copie nature of the "breakdown process" in the oxide film, other than for the accompanying changes in conductivity and stoichiometry. The implicit assumpt that these would be homogeneous changes is carried by the deductions subseque made- Our electron microscope studies [20,21] of the oxide films formed on zirconium, of similar purity to that used in the hydrcgen diffusivity studies showed that the oxide films formed in a far from uniform and parallel-sided manner. Ridges of thick oxide form preferentially along many grain boundarie and big variations in the oxide thickness occur from grain to grain (Figure 6 Many of these features have also been recognized from optical microscopy [22] However, the lines of cracks (Figure 7) which form in the oxide along these ridges, and along grain boundaries separating grains which grow thick and thi oxide films [20-23], are less easily seen in the optical microscope. These cracks are encouraged by the curvature in the oxide/environment interface int duced by the local variation in oxide thickness and the high Pilling-Bedworth ratio of ~1.5. These factors cause the outer surface of the oxide to go int tension, whereas oxide films formed as a planar layer on a smooth surface rem in compression at the same point in time.

However, even in regions where cracks form in the outer part of the oxide, the inner layers of the oxide will remain in compression because the n layers of oxide form at the oxide-metal interface by inward diffusion of oxyg Thus, there will probably always be a thin layer of protective oxide at the bottom of these cracks. This argument is supported by the observation that cracks are visible in the oxide surface formed during oxidation, and yet thes oxide films continue to present excellent barriers to the ingress of hydrogen Hence, such cracks are not the routes by which hydrogen enters the metal, although they may help to initiate, or become part of such a route during the "breakdown" process. That these ciarks do not penetrate to the oxide/metal interface is shown in SEM studies of stripped oxide films. Examination of th oxide/metal interfaces [23,24] clearly showed the ridges of thick oxide along grain boundaries (Figure 8), and the variations in oxide thickness from grain grain (Figure 9), but did not reveal any yigns of cracks penetrating through the interface at these locations. In instances when the stripping process induced cracking of the oxide, these cracks generally cut across oxide ridges at ,;rain boundaries (Figure 10), suggesting that these sites were not even regions of wuakness in the oxide. Only occasionally have cracks running parallel to, and within, the oxide ridges been seen at the oxide-metal interface. Some other route, which allows direct access of hydrogen to the oxide/metal interface, must develop during the incubation period, therefore.

When a specimen, with an oxide film such as those above, is heated in an environment (vacuum or H2) containing insufficient oxidant to maintain the normal growth of the oxide, the rate of oxygen dissolution in the metal will exceed the rate of formation of new oxide. Under these conditions, the oxide 1 ifm will slowly dissolve in the metal. This process has been observed (.Figure IL) by following the thinning ot interference-coloured oxide films in the optical microscope [25,26]. On the scale of resolution of this instrument the dissolution process appears to be uniform, provided that the metal core has not become saturated in oxygen [26,27]. However, autoradiography ot the oxygen distribution using the ^0(p,n)^F reaction [27] shows that preferential diffusion 01 oxygen along zirconium grain boundaries occurs.

When the dissolution of oxide films is studied by electron microscopy, examination of the oxide/raetal interfaces shows that the thinning of the oxide is not uniform. Because of the preferential diffusion of oxygen along the grain boundaries in zirconium, the dissolution of the oxide ridges at these sites occurs preferentially and inhomogeneously (Figure 12). This irregular dissolution of the grain boundary ridges leads to the formation of large pores in the oxide. These pores are thought to pass right through the oxide film because similar rows ÖL pores have been seen when oxide films on specimens exposed to low oxygen partial pressure environments were viewed from the outside by a replica technique (Figure 13). These pores also occur in the oxide away from the prior metal grain boundaries, perhaps because of enhanced local oxygen diffusion at incoherent twins, or dislocation pile-ups.

The uniform dissolution of the bulk of the oxide, seen in the interference colour studies [25,26], is probably the process which leads to the observed drop in oxide resistivity [7,10]. However, the preferential dissolution of the oxide at grain boundaries, and other sites, which leads to the formation of arrays of pores at these locations, is probably the process which terminates the incubation period, and leads to the rapid direct reaction of hydrogen with the metal. Since these pores appear to pass right through the oxide, there will be no oxide diifusion barrier to prevent such a reaction at these sites. Thus, none of the investigations which purported to measure the diffusivity of hydrogen in ZrO2 by exposing preuxidized specimens in hydrogen gas will have measured such a quantity, and alternative techniques for measuring the hydrogen diffusion coel1 ifient in ZrO2 must be sought.

3. HYDROGEN UPTAKE DURING OXIDATION

Direct reaction of gaseous hydrogen with the zirconium at the bottoms of pores is not a possible route for hydrogen ingress during normal oxidation. Even when oxide films become porous, in the post-transition region of oxidation kinetics, the pores which develop will not penetrate up to the oxide/metal -5-

interlace, .'. 1 thuu^ii L lit? y may approach quiLe close to it, because there will always bo c-nougli oxi.dant available to maintain at least a thin barrier layer at the bottom of the pore.

Oxid.ition studies using T2/H2O mixtures [16] have shown that, during normal oxidation, no 'Ij enters the metal (Table 2) until the thermally-induced exchange reaction has progressed to the point where a measurable fraction of T2O has been formed. Thus, the hydrogen isotopes which enter the metal do so as an integral part of the reaction of the zirconium with water molecules, and not by reaction with any dissolved hydrogen in the water- Studies have shown that this situation persists (Figure 14) until hydrogen overpressures in the system o(" tens of MPa are present [28].

During the reaction cf zirconium alloys with water the hydrogen is liberated, initially, as an adsorbed hydrogen atom on the oxide surface, when a proton is discharged by an e lee. L run ei;u--rt; i. :,;, through the oxide film. The site at which this occurs depends on the morphology of the oxide film (Figure 15). Thus, for alloys of the Zircaloy type, which contain additions of Fe, Cr and Ni (to a total of~0.3 wt%) in the form of a distribution of second-phase particles, the elecuron current flows primarily at sites where intermetallics partially, or completely, short-circuit the oxide [29]. In alloys containing no, or few, such intermetallic particles of the transition metals (e.g. Zr-2.5 wt% Nb) the electron current flows homogeneously through the bulk of the oxide [29].

The hydrogen atoms released by the discharge of the protons (either locally at intermetallics, or uniformly over the surface) then either recombine, and are evolved as hydrogen gas, or diffuse into the metal. It is the competi- tion between the rates of these two processes which determines the proportion of the corrosion-hydrogen which enters the metal. Any local situation which affects the relative probabilities of recombination or ingress into the metal will, therefore, affect the hydrogen uptake percentage. Despite the importance of this step in the reaction no measurements of hydrogen recombination rates on either pure or doped zirconia surfaces are known.

The presence of alloying additions, such as nickel, which are thought to decrease the rate of the recombination reaction, or even, under some conditions, to directly dissociate molecular hydrogen dissolved in the water [28], could then lead to high percentage uptakes for the corrosion hydrogen by allowing a greater opportunity for the hydrogen atoms to diffuse into the metal. It was for this reason that nickel was eliminated from the alloy in the development of Zircaloy-4 from Zircaloy-2 [30]. It is also notable that the hydrogen uptake percentage shown by Zircaloy-4 in water is virtually independent of the hydrogen overpressure (Figure 14), unlike that of Zircaloy-2 [28], thus adding further weight to the argument that the presence of nickel can lead to the direct disso- ciation and ingress of hydrogen dissolved in the water, when this is present to excess.

The presence of dissolved oxygen, or other oxidizing species (e.g. nitric acid) is able to speed up the removal of the discharged protons from the Zr02 surface and hence reduce the percentage hydrogen uptake (Figure 16). The precise mechanism by which this is achieved, whether by direct reaction of oxygen dissolved in the water with hydrogen atoms on the surface, or by the substitution -6-

of the reduction of oxygen by the electrons for the reduction of protons (as the cathodic reaction), has not been identified. The presence of dissolved hydrogen, as we have seen above, has virtually no effect on alloys without nickel containing precipitates, and only a small effect on Zircaloy-2, until very large concentrations are present. Once the oxide film becomes porous, in the post- transition oxidation region, the hydrogen apparently finds it more difficult to escape from the oxide, for there is a general tendency for percentage uptakes to increase [31]. However, these effects are smaller at temperatures typical of reactor operation than at higher temperatures (Figure 17).

The strong correlation (Figure 18) of hydrogen uptake percentages with the nature of the elements in the intermetallic precipitates [32], coupled with the observations of localized electron conduction at these sites, leads to the inference that the hydrogen entering the metal does so via some specific property of the oxide formed over these intermetallics. We are only now elucidating the nature of these oxide films [33], so the understanding of this aspect of the uptake mechanism may change rapidly in the near future. However, preliminary results suggest that elements like Fe and Ni may remain in the zero oxidation state in the oxide on an intermetallic. The state of agglomeration of these atoms has not yet been established, but the possibility of metallic stringers within these oxides cannot yet be ruled out. Such stringers could account for both the high local electrical conductivity and the easy ingress of hydrogen at these sites.

Although it is thought that, in alloys containing large numbe's of intermetallic paticles large enough to partially, or completely short-c i_rcuit the oxide, the hydrogen from the corrosion reaction enters the metal at these sites, it is known from tritium autoradiography that the uniform oxide between the intermetallics contains a significant (Figure 19), but unknown, concentration of hydrogen [19]. That the quantity of tritium in such oxides appears to be roughly proportional to oxide thickness (Figure 20), for interference-coloured oxides, suggests a relatively constant concentration of tritium in these oxides, with little concentration gradient thrcigh them. Attempts .-, measure this concentration using early infra-red spectroscopy techniques [34] were unsuccessful, but indicated that the concentration of OH~ in Zr02 films was probably ~ 5%. No application of modern infra-red techniques to this problem has yet been made.

Despite the known presence of hydrogen in the uniform oxide film there is no knowledge of its mobility. The long-term stability of autoradiographic specimens against exchange of the tritium with laboratory water vapour suggests that the mobility is low, but does not permit the calculation of other than an upper bound for the mobility. If, as observed, the tritium content of an 0.5 fim oxide film remains apparently unchanged after one year at room temperature, the permeability of tritium through these films must have been less than ~10~^ mole cm~2.s~l at room temperature. This is very much the value extrapo- lated from the elevated temperature tests [14,15], which we have already argued represents surface diffusion down pores (Figure 21).

Work presently in progress by Fourier Transform NMR on ZrO2 films formed in 500°C steam, suggests that the hydrogen in the ZrO2 is tightly bonded and relatively immobile. However, we have not yet been able to extract a value for the diffusivity from these experiments. -7-

Thus, we arrive at a picture whereby hydrogen uptake in alloys containing intermetalLics of Fe, Cr and Ni is controlled by reLease of the hydrogen atoms at the sites of these particles, and migration of them via the same intermetal- llcs, perhaps by way of stringers of metallic Fe or Ni in the oxide, into V. he •netal. [n alloys containing few sucli particles (e.g. Zr-2.5 wt% Mb) the hydrogen must migrate nore uniformly through the oxide, which has a low diffusivity for hydrogen. This difference in mechanism can account for the large differences in the observed deuterium uptake rates in Zircaloy-2 and Zr-2.5% wt Mb alloy pressure tubes [3]. If we assume that the very low rates of uptake by Zr-2.5 wt% Nb are entirely controlled by diffusion through the oxide film, then we can use these rates to calculate a lower bound for the diffusivity in ZrO2• This gives us a value of ~ 10~^ moles.cm~^.s~^ for the permeability at 290°C. This is in the same range as rhe results from effusion into steam [16], especially if the actual barrier layer oxide thickness is quite thin (i.e. If the oxide is post- transition)- Without a knowledge of the hydrogen concentrations in these oxides, it is not possible to estimate a true dif tus ion coefficient, but the permeabilities estimated above are not inconsistent with the very low diffusion coefficients in the surface layer measured by the workers at North Carolina State University [17,18].

4. HYDROGEN ABSORPTION VIA METALLIC CONTACTS

Although hydrogen absorption through a metallic contact may be the most trivial of the three mechanisms mechanistically, it nevertheless has important practical aspects as it may be one route for ingress of excess deuterium into the ends of pressure tubes (Figure 22) from the 403 SS end fitting [1]. The zirconium alloy pressure tube is in good metallic contact with the end-fitting steel in the three grooves of the rolled joint. The factors which control ingress by such a route should, therefore, be considered.

When a galvanic couple between a zirconium alloy and a nickel or iron alloy is exposed to high temperature water the zirconium alloy becomes anodic and the transition metal alloy cathodic. The cathodic part of the oxidation process then occurs on the transition metal and protons are discharged there and may be absorbed by the transition metal. Thj effective surface areas of the two parts ot the couple will be determined by the conductivity of the water, and in high purity water only small areas adjacent to the point of contact should be effective, although under irradiation this limitation may be relaxed by the effect of irradiation on the conductivity of the electrolyte [36]. The distance from the points of closest approacli of the two metals in the water and the point of metallurgical contact will be important for the hydrogen migration, but not lor the oxidation. This is evident in the instance of localized oxidation opposite a platinum implant [36], where the electrical contact was distant from the point of closest approach of the platinum implant and the facing zirconium surface. In this instance, accelerated oxidation but no excess hydriding occurred, whereas adjacent to the platinum implant both accelerated oxidation and enhanced hydrogen contents were observed.

In the case of the Inconel to Zircaloy bonds [35,37], where the interface between the Inconel and the Zircaloy-2 was exposed to the water, and diffusion distances for both hydrogen and electrons were short, both excess oxidation and -8-

excess hydriding were observed adjacent to the bond. No enhanced hydrogen uptake was observed in Zircaloy/Zircaloy crevices [36,37], but enhanced oxidation, perhaps from enhanced radiolysis, was seen. Although it might be expected that the diffusivity of hydrogen in the transition metal would affect the rate of migration into the zirconium alloy via the metallurgical bond there is no evidence for comparison (say) between Inconel and stainless steel. However, where high temperature water is in contact with zirconium alloy/transition metal bonds close to or at the interface between them the rate of the cathodic reaction on the transition metal may be more important than the hydrogen diffusivity in it. In this circumstance, the higher nickel content: of Inconel than of stainless steel and the specific effects of nickel with regard to the hydrogen recombination reaction may mean higher excess hydrogen absorption for zirconium alloy/lnconel bonds than for zirconium alloy/stainless steel bonds. This seems to be borne out by the results of Urbanic's [38] experiments on Zr-2.5% Nb tubes with plugs of various metals inserted in them. If these results are compared with some properties of the various metals (Table 3) it would appear that the exchange current density for the hydrogen evolution reaction is more important than the hydrogen permeability in determining the uptake by zirconium alloys. However, there is insufficient evidence at present to reach a firm conclusion about the important factors determining the rate of excess hydrogen uptake in zirconlum alloys metallurgically bonded to transition metals; tue indications are that we should look at the efficiency of the coupled metal as a cathode, rather than its diffusion properties.

5. CONCLUSIONS

1 have summarized, above, our present understanding of the three princi- pal mechanisms by which hydrogen isotopes may enter zirconium alloy components in service. It will be seen that many of the basic physical rate constants which would be useful in confirming these uiechanisms have not been measured. The diffusion coefficients for hydrogen isotopes are not known to within many orders of magnitude, and other rate constants such as the recombination rate of hydrogen atoms on zirconia surfaces (doped with realistic alloying additions) are not known at all. The radiation chemistry within porous ZrO2 films is, likewise, an unknown quantity with regard to any differences between it and that in bulk water. There is much scope, therefore, for further work in t'.ls area.

REFERENCES

E.C.W. Ferryman, "Pickering Pressure Tube Cracking Experience", Nucl. En., 1978, 1_7_, 95.

G.J. Field, J.T. Dunn and B.A. Cheadle, "Analysis of the Pressure Tube Failure in Pickering NGS 'A1 Unit 2", Presented at Can. Nucl. Ass. Ann. Meeting, Saskatoon, Jun.- 1984; Atomic Energy of Canada Ltd., Report AECL- 8335 (June 1984). -9-

[3] V.F. Urbanic and B. Cox, "Long-Term Corrosion and Deuteriding Behaviour of Zircaloy-2 Under Irradiation", Proc. 1984 C.I.M. Conf., Quebec City, Aug. Can. Met. Quart, (to be published).

[4] H.G. Nelson and H.F. Wachob, "Stress Corrosion Cracking of Zircaloy" in Electric Power Research Institute Report EPRI-NP-717, Palo Alto, CA, March 1978, Section 6.

[5] C.E. Coleman and B. Cox, "Cracking of Zirconium Alloys in Hydrogen", Proc. of 6th Int. Conf. on Zirconium in the Nuclear Industry, Vancouver, 1982, ASTM-STP-824, p.675.

[6] CE. Coleman and J.F.R. Ambler, "Delayed Hydrogen Cracking in Zr-2.5 wt% Nb Alloy", Revs, of Coatings & Corrosion, 1979, 3^ 105, Freund, Tel Aviv.

[7] D.W. Shannon, "Role of Oxidation Rate on the Hydriding of Zirconium Alloys in Gas Atmospheres Containing Hydrogen", Corrosion, 1963, 19, 414t, and U.S. Report HW-76562 REV. (Feb.1963).

[8] B. Cox, "Some Effects of Pressure on the Oxidation of Zircaloy-2 in Steam and Oxygen", J. Less Common Met., 1963, _5_, 325.

[9] R.F. Boyle and T.J. Kisiel, "Hydrogen Permeation of Zircaloy-2 Corrosion Films", U.S. Report, Bettis Tech. REv., WAPD-BT-10 (1958) p.31.

[10] Katsumi Une, "Kinetics of Reaction of Zirconium Alloy with Hydrogen", J. Less. Common Met., 1978, _57_, 93.

[11] S. Aronson, "Some Experiments on the Permeation of Hydrogen Through Oxide Films on Zirconium", U.S. Report, Bettis Technical Rev., WAPD-BT-19 (1960) p.75.

[12] E.A. Gulbransen and K.F. Andrew, "Mechanism of the Reaction of Hydrogen with Zircaloy I. Role of Oxide Films, Pretreatments and Occluded Gases", J. Electrochem. Soc, 1954, 101, 348.

[13] Idem, "Reaction of Hydrogen with Preoxidized Zircaloy-2 at 300° to 400°C", ibid., 1957, 104, 709.

[14] T. Smith, "Mechanism of Hydrogen Permeation of Oxide Films on Zirconium", J. Electrochem. Soc., 1965, 112, 560.

[15] Idem, "Kinetics and Mechanisms of Hydrogen Permeation of Oxide Films on Zirconium", J. Nucl. Mat., 1966, 1£, 323; and U.S. Report, NAA-SR-6267.

[16] B. Cox and C. Roy, "The Use of Tritium as a Tracer in Studies of Hydrogen Uptake by Zirconium Alloys", Atomic Energy of Canada Ltd., Report AECL- 2519 (Dec.1965).

[17J T.S. Elleman and K. Verghese, "Surface Effects on Tritium Diffusion in Niobium, Zirconium and Stainless Steel", J. Nucl. Mat., 1974, 53, 299. -10-

[18J J.H. Austin, T.S. Ellenan and K. Verghese, "Tritium Diffusion in Zircaloy-2 in the Temperature Range -78 to 204°C", J. Nucl. Mat., 1974, 2l_, 321 .

[19] C. Roy, "Hydrogen Distribution in Oxidized Zirconium Alloys by Autoradio- graphy", Atomic Energy of Canada Ltd., Report AECL-2085 (Sept.1964).

[20] B. Cox, "Oxide Breakdown on Arc-Melted Sponge Zirconium", Corrosion, 1960, lb_, 306; and U.K. Report AERE-2874 (1959).

[21] B. Cox and A.B. Mclntosh, "The Oxide Topography on Crystal-Bar and Reactor Grade Sponge Zirconium", Atomic Energy of Canada Ltd., Report AECL-3223 (Nov.1968).

[22] J.K. Wanklyn, CF. Britton, D.R. Silvester and N.J.M. Wilkins, "The Corrosion of Zirconium and its Alloys. Part III. The Influence of the Environment on Oxidation", J. Electrochem. Soc., 1963, 110, 856; and U.K. Report AERE-R4130 (August 1962).

[23] B. Cox, "Examination of Oxidized Zirconium Alloys with the Photo-Emission and Scanning Electron Microscopes", Atomic Energy of Canada Ltd., Prog. Rep. PR-CMa-9 (1969) p.76.

[24] B. Cox, "The Zirconium-Zirconia Interf?r" ", J. Aust. Inst. Met., 1969,

[25] J.P. Pemsler, "Diffusion of Oxygen in Zirconium and its Relation to Oxidation and Corrosion", J. Electrochem. Soc, 1958, 105, 315.

[26] J.P. Pemsler, "Studies of Oxygen Gradients in Corroding Zirconium Alloys", J. Nucl. Mat., 1962, ]_, 16.

[27] W.W. Doertfler, "A Contribution to the Mechanism of Dissolution and Diffusion of Oxygen in Zirconium", Swiss Report, EIR-82 (Sept.1965).

[28] E. Hillner, "Hydrogen Absorption in Zircaloy During Aqueous Corrosion. Effect of Environment", U.S. Report, WAPD-TM-411 (1964).

[29] N. Ramasubramanian, "Localized Electron Transport in Corroding Zirconium Alloys", Proc. 1975 NACE Annual Meeting, Toronto, Preprint 160; and J. Nucl. Mat., 1975, 55_, 134.

[30] S. Kass and W.W. Kirk, "Corrosion and Hydrogen Absorption Properties of Nickel-Free Zircaloy-2 and Zircaloy-4" , ASM Trans. Quart., 1962, _56_, 77.

[31] B. Cox, "Some Factors which Affect the Rate of Oxidation and Hydrogen Absorption of Zircaloy-2 in Steam", U.K. Report AERE-R4348 (July 1963).

[32] B. Cox, "Oxidation of Zirconium and its Alloys", Adv. in Corr. Sei. and Tech., Eds. Fontana and Staehle, Plenum, N.Y., Vol.5 (1976) p.173. -11-

[33] R.A. Ploc and R.D. Davidson, "Auger Electron Analysis of Oxides Grown on Dilute Zirconium Alloys", Proc. of Int. Metall. Soc. Ann. Conf., Philadelphia, July 16-19, 1984.

[34] T. Maekawa and M. Terada, "Infra-Red Spectra Study on the Oxidation Film of Zirconium", Trans. Jap. Inst. Met., 1963, 4L, 47.

[35] A.B. Johnson, Jr., "Aqueous Corrosion and Hydriding of Zirconium Alloys in Nuclear Reactor Environments", Proc. 4th Int. Cong, on Met. Corr., Amsterdam, Sept. 1969, NACE, p.168.

[36] A.B. Johnson Jr., J.E. LeSurf and R.A. Proebstle, "Study of Zirconium Alloy Corrosion Parameters in the Advanced Test Reactor", Proc. of Symp. on Zirc. in Nucl. Appl., Portland, Aug. 1973, ASTM-STP-551, p.495.

[37] A.B. Johnson, Jr., "A Review of Corrosion Phenomena on Zirconium Alloys, Niobium, Titanium, Inconel, Stainless Steel and Nickel Plate under Irra- diation", Revs, on Coatings and Corro., Freund, Tel Aviv, 1975, 4_, 299.

[38] V.F. Urbanic, "Observations of Accelerated Hydriding in Zirconium Alloys", Proc. 6th Int. Symp. on "Zirconium in the Nuclear Industry", Vancouver, 1982, ASTM-STP-824, p.554. -12-

TABLE 1

Critical H2/0xidant Ratios for Direct Hydriding

Temperature H. Pressure Critical Material (*C) (atm) Ratio Author Ref.

Zircaloy-2 343 6.5 x 10"2-1.0 106-108 Boyle & Kisiel 9

400 1.3 x 10 102 Shannon 7

320 1.0 105 Gibby BNWL-150 (1965)

300 .-v .„ , „ .. Une 10 400 1.0 10Z

25 0.1 102-103 .„ . _ . / Coleman & Cox 5 Zr-2.5% Nb 25 0.1 10-10 -13-

TABLE 2

Oxidation in T2/H2O Mixtures

Hydrogen Absorption Temp. Time Wt. Gain Sample (°C) (h) (mg/dm2) ppm % Theor. Observations

GRAO Zr-2.5 wt% Nb (1) 400 66.0 15.5 14 21.2 No significant tritium uptake.

GR41 Zr-2.5 wt% Nb (2) 400 113.0 14.5 13 23.4 Small amount of tritium in oxide and metal.

GR42 Zircaloy-2 (3) 500 24.5 24.5 39 23.0 No tritium detected in oxide or metal.

(1) Heated at 1000°C in vac. for 1 h and quenched.

(2) Heated at 88CCC in vac. for 72 h and quenched, reheated at 500°C for 6 h. The large amount of isothermal a-phase present in this sample probably results from oxygen absorption during the long anneal at 880°C.

2 (3) Preoxidized in 02 at 500°C for 24 h; Aw = 28.2 mg/dm . TABLE 3

Comparison of H Uptake by Zirconium Alloy/Metal Couples with Physical Properties of the Coupled Metal

Exchange Current Diffusion Coeff. Enthalpy of Permeability Hydrogen Uptake Density for H, (1) of B2 at 400*C Solution of to B, at 400*C 2 J Ref.37 Ref.38 dog i > H 1 -1 -2 -I Metal o 2 (« (STP)aawb .cm at» '} (3) Cu -6.70 9.0 x 10"6 (4) +0.47 (A) 3.6 x 10"4 (5) - 13 Fe -5.93 1.5 x 10"A (4) +0.28 (A) 1.5 x lO"1 (6) - - Inconel -5.46 8.0 x 10"7 (7) - 5.4 x 10"3 (7) 152 - 30A S/S -5.37 3.0 x 10"7 (8) - 3.5 x 10~3 (6) - 380 A03 S/S - 5.8 x 10"5 (9) - 4.6 x 10~2 (9) - 1380 Ni -5.17 5.0 x 10~6 (4) +0.16 (A) 3.0 x 10"2 (6) - 3730 Pt -3.73 5.1 x 10"3 (4) +0.75 (A) 1.1 x 10~3 (6) 354 >10000(9)

(1) Comprehensive Treatise of Electrochemistry, Vol.2, Eds. Bockris, Conway, Yeager & White, Plenum, NY, 1981. (2) mg/kg H2 in Zircaloy-2 after 60 d coupled to metal in 300°C water under irradiation. (3) Peak H2 concentration at centre of Zr-2.5 wtZ Nb tube after 10 days in pHIO LiOH with 1.1 MPa H2 overpressure. (4) J. Vblkl and G. Alefeld, Chapter 5, Diffusion in Solids, Recent Developments, Eds. Nowick and Burton, Ac. Press, New York, 1975. (5) G.R. Caskey et al., Corrosion, 1976, 32. (9). (6) R.W. Webb, Permeation of H through Metals, NAA-SR-10462 (1965). (7) W.M. Robertson, Met. Trans. A, 1977, 8A. (8) M.R. Louthan and R.G. Derrick, Corr. Sei., 1975, JJ>, (9) V.F. Urbanic and A.F. Coles, unpublished results, CRNL. -15-

100

CNi

v - Samples that Picked Up More than 100% I of Corrosion Product H2

o - Samples that Picked Up Less than 100% of Corrosion Product H2

• - Weight Gain in Pure ^O Vapor

<7 1 1 I I I 1 l 1 1 10 50

Time (h)

FIGURE 1. Plot showing influence of oxidation rate during exposure to H2/H2O mixtures at 400°C on hydrogen uptake (Ref.7) 320

2*0

Û CURVE I , UNOXlOlZED PLATES Q CURVES 2, OXIOiZEO PLATES O CURVES 3, OXIDIZED PLATES • CURVE 4, OXIDIZED PLATES, TREATED IN VACUO • CURVE S, OXIDIZED PLATES, TREATED IN VACUO }} INDICATE!. RUN INTERRUPTED AT END OF WORKING DAY (SEE TABLE I)

6 7 10 12 TIME IN HOURS

FIGURE 2. Incubation times for absorption of hydrogen gas by zirconium at 400°C (Ref.11). -17-

500 K —

Time (days)

FIGURE 3. Plot of electrical resistance for Zircaloy-2 exposed to gaseous environments at 45O°C (Ref.7). Oxide

Oxygen Di ffus ion Zone

03 I

aZr+Zriu

X500 Autoradiograph X500 Autoradiograph in-situ

FIGURE 4. Autoradiograph showing lack of tritium absorption by a thick oxide film formed in O2 when subsequently exposed to T2• (Oxidized in O2 at 6OOCC for 66.5 h to weight gain of 6.84 mg/dm2; Exposed to T2 gas at 800°C for 16.5 h; Réf.19). ox ido

diffusion zone

aZr +ZrHx

^%V^

X500 Autoradiograph X500 Micrograph

FIGURE 5. Autoradiograph showing tritium incorporated in a thick oxide film formed in O2 when subsequently exposed to T2O. (Oxidized in O2 at 600°C for 66.5 h to weight gain of 727 mg/dm2; Exposed in T2O vapour at 400°C for 72 h, additional weight gain 5 mg/dm2; Ref.19). ,*^ Î^V y*» *

*«•**•"

FIGURE 6. Scanning electron micrograph showing regions of thick and thin oxide formed on different zirconium grains together with ridges of thick oxide along grain boundaries and micro-twins, X2000 (Ref.21). -21-

FIGURE 7• Higher magnification micrograph of junction between thick and thin oxide areas and oxide ridge, X6000 (Ref.21). FIGURE 8. View of oxide-metal interface showing appearance of oxide ridges at grain boundaries, X3000 (Ref.24) -23-

.'. " ~ "'''^.. '.:....'••- :.i-^..- -' •• ------J

FIGURE 9. View of oxide-metal interface showing areas of thick and thin oxide comparable to those seen in Figure 6, X1000. Note oxide has not fractured along grain boundaries.

crack crack

crack

(a) (b)

FIGURE 10. Oxide-metal interface of stripped oxide films showing cracks induced by the stripping, X3000. Note these cracks do not follow oxide ridges . -24-

490 1 1 1 1 1 1 1 1 1 1 1 I

420 - 9 '•

e / & 350 4

i 280 - u r. •- 6 2'° ç 1 \ S 140 • i D ....I_J 3 a 2tfcotoy î* Siol.c Vacuum 70 3A Or«*» **•

1 -/l 1 1 1 1 1 1 1 1 i i i t i t | 1.0 2JO 4.0 Tim«, (days)

FIGURE 11. Decrease in thickness of interference-colour oxide films as a function of annealing time in vacuo at 400°C (Ref.25). - 25 -

(a) (b)

FIGURE 12. Oxide-metal interface of specimen after vacuum annealing at 500°C. Note break-up of oxide ridges and formation of pores. (a) X1000. (b) X3000.

(a) (b)

FIGURE 13 . Replica of zirconium oxide surface of specimen heated in a low oxygen environment (He), showing a similar array of pores to those seen in Figure 12. Diameter at surface is smaller than at oxide-metal Interface. (a) X4000, (b) X10 000. A * ZlRCALOY-2,AS ROLLED D * 2IRCALOY-2 ,/9-TREATED •~ I00-WIL THICK SPECIMENS 60 _O ' ZIRCALOY-4 , AS ROLLED O • ZIRCAL0Y-4, ^-TREATED

ZIRCALOY-2,/3-TREATED 50

ZIRCALOY-2 AS-ROLLED

ZIRCALOY-4 AS-ROLLED OR 0-TREATED

500 1000 1500 2000 2500 3000 HYDROGEN OVERPRESSURE, PS!

FIGURE 14. Hydrogen pickup by Zircaloy-2 and Zircaloy-4 as a function of hydrogen overpressure after 14-day exposure in 343°C water (Ref.28). ETCH PIT PRODUCED OXYGEN , AIR DURING INITIAL 020R H20 ADSORBED PICKLING OR STEAM PROTONS OR REACTION :- 2 ELECTRONS TRAVEL —2O ~ BY SURFACE CONDUCTION

THICK OXIDE NJ ELECTRONS ^DEVELOPS COLUMNAR \ TED THR0UC4 0 2" DIFFUSES EM1T Fe2O3OR MONOCLINIC CRYSTALLITE STRUCTURE \TH,N |rO2 LAYER, DOWN Fe3O4LAYER CRYSTALLITE BOUNDARIES CRYSTALLITES NUCLEATE AMORPHOUS OXIDE

...„,„, THIN AMORPHOUS LAYER OF ZrO2 ZIRCONIUM REACTION 2 2O ~+Zr-»ZrO2+4e INTERMETALLIC I MICRON PARTICLE STILL IN METALLIC CONTACT WITH THE MATRIX APPROXIMATE SCALE

FIGURE 15. Schematic diagram of oxide film on Zircaloy-2 and the processes occurring in it during oxidation (Ref.32). ZIRCALOY - 2 680*F DEGASSED H2O NICKEL-FREE ZIRCALOY-2 680'F DEGASSEO H,0

>^680*F HzO + IOOO PSI ALL MATERIALS THEORETICAL (INSTANTANEOUS) I co I

ù, • ZIRCALOY-2 • AÏ «... I S SURVEY EXP REfEATJ O« Ni "FREE ZIRCALOY-2 • »J D • ZIRCALOY-4

SO 73 100 125 ISO 175 200 OXYGEN WEIGHT GAIN, MG/DM2

FIGURE 16. Effect of oxygen content of water on hydrogen uptake by Zircaloys in 360°C water (Ref.28) -29-

O6 O / STEAM ' latm V 3OO°C

O-4 % a. /

O2 ZIRCALOY-2 O-OOS'thcct 0 2 4 • g.

o o. e o z

IO SO 6O 7O 8O Oxygen uptake

FIGURE 17. Change in hydrogen uptake rate with oxide film thickness for Zircaloy-2 (Ref.31). CD CAL 1HVDROGEh »ORBED. 100 80 7O 90 50 60 30 20 10 FIGURE 18 . Effect o f alloying element s in intermetallic for m on hydrogen uptake percentage (Ref.32) BASED ONALLOYSCONTAINING 2. WEBERRY 3. JNWANKLY(UNPUBLISHEDDATA) 1. THISWORK Ti VC r MnF e CoNL Cu 1-5 7©ADDITIONS Nl I Zn I Go I Ge T o u> i i X1000 Autoradiograph X1000 Micrograph

FIGURE 19. Autoradiograph showing uniform tritium content in thin zirconia film formed in T2O for long enough to saturate oxide (oxidized in O2 at 400°C for 66.5 h to weight gain of 12.7 mg/dm2, oxidation conti- nued in T2O at 400°C for 48 h to total weight gain of 114.4 mg/dm2). Compare with non-uniform distribution in Figure 5 (Ref.19). X200 Autoradiograph X200 Micrograph

FIGURE 20. Surface autoradiograph shows tritium content proportional to oxide thickness. Note that the darkness of the interference colour oxides in the micrograph is not proportional to thickness because of colour response of black and white film. (Oxidized in T2O at 400°C to weight gain of 1.9 mg/dm2, Ref .19). FIGURE 21. Comparison of permeability data of T. Smith with other estimates of hydrogen in ZrC>2 films. FIGURE 22. Increase in deuterium concentration at the inlet ends of Zr-2.5 wt% t?b alloy pressure tubes in the Pickering reactors (Ref.l). ISSN 0067 0367 ISSN 0067-0367

To identify individual documents in the series Pour identifier les rapports individuels fai: 3 we have assigned an AECL number to each. partie de cette série nous avons assigné un numéro AECL- àchacun.

Please refer to the AECL numberwhen re- Veuillez faire mention du numéro AECL- si questing additional copies of this document vous demandez d'autres exemplaires de ce rapport

from

Scientific Document Distribution Office Servicede Distribution des Documents Office Atomic Energy of Canada Limited L'Energie Atomique du Canada Limitée Chalk River, Ontario, Canada Chalk River, Ontario, Canada KOJ 1J0 K0J1J0

Price $3 00 per copy Prix $3.00 par exemplaire

ATOMIC ENERGY OF CANADA LIMITED, 1985

0285-85