Anodic oxidation and dissolution of CANDU fuel(UO2) in slightly alkaline sodium perchlorate

solutions

S. Sunder, L. K. Strandlund and D. W. Shoesmith*

Atomic Energy of Canada Limited, Whiteshell Laboratories, Pinawa, Manitoba, Canada R0E 1L0

(Received 9 July 1997; in revised form 12 November 1997)

AbstractÐThe anodic dissolution of UO2 in aqueous sodium perchlorate solutions at pH09.5 shows twodistinct regions of oxidation/dissolution behavior. In the potential (E) range 0.10E < 0.35 V (vs SCE) cur-rents decay continuously with time owing to the formation of oxidized surface ®lms which gradually blockfurther oxidation of the electrode. An analysis of anodic and cathodic charges (QA, QC, respectively)obtained by integration of the anodic current±time plots (QA) and cathodic potential scans to reduce accu-mulated oxidized surface ®lms (QC) in this range, shows that 90% of the anodic oxidation current goes toproduce these ®lms. The ®lms blocking anodic dissolution appear to be either UO2.67 or, more probably,UO3�xH2O. These may be located primarily at grain boundaries, although this has not been con®rmed.For E>0.35 V, steady-state currents are obtained and measurements of QA and QC show that the majorityof the current goes to produce soluble species. It is proposed that, at the higher potentials, rapid oxidationand dissolution followed by the hydrolysis of dissolved uranyl species leads to the development of acidicconditions in the grain boundaries. At these lower pH values UO3�2H2O is soluble and does not accumu-late. Analyses of Tafel slopes suggest that the observed behavior is consistent with dissolution underacidic conditions, although other possibilities have not been excluded. Crown copyright # 1998 PublishedbyElsevier Science Ltd. All rights reserved

Key words: UO2, oxidation, anodic, CANDU fuel, dissolution.

INTRODUCTION

Methods for the disposal of used nuclear fuel are

being investigated in many countries such as

Canada, U.S.A., Sweden, Germany and Spain,

etc. [1±20]. The assessment of the geological dispo-

sal of used nuclear fuel requires a prediction of its

dissolution rate as a function of disposal time. This

dissolution rate is expected to be a function of

redox conditions, since used nuclear fuel is mainly

UO2 (>95%), a solid for which the solubility

increases by many orders of magnitude under oxi-

dizing conditions [20±28]. Although the ground-

waters at the planned depth of a Canadian disposal

vault are reducing [2, 7], the possibility remains that

the fuel will encounter oxidizing conditions on ex-

posure to groundwater, due to the potential for

production of oxidants by the radiolysis of water

by gamma, beta and alpha radiation [1, 2, 6±9, 19].

It has been shown that the dissolution (corrosion)

of UO2 is an electrochemical process [6±8, 20, 29±

37]. Recently, we have published a simple electro-

chemical model to predict the dissolution rates of

UO2 as a function of the redox conditions prevail-

ing in the solution environment to which the fuel is

exposed [6, 7], and applied this model to assess the

behavior of used fuel in a waste disposal vault [38].

In this model the dissolution currents (iD) measured

on UO2 electrodes as a function of applied electro-

chemical potential (E) are plotted as a Tafel plot

(log iD vs E) and extrapolated to the steady-state

corrosion potential [(ECORR)SS] measured under

natural corrosion conditions. The value of the cor-

rosion current (iCORR) obtained is the dissolution

rate of UO2 for the redox conditions prevailing in

that particular solution. By measuring a series of

values of (ECORR)SS in solutions of di�erent redox

Electrochimica Acta, Vol. 43, Nos 16±17, pp. 2359±2372, 1998Crown copyright # 1998 Published by Elsevier Science Ltd. All rights reserved

Printed in Great Britain0013±4686/98 $19.00+0.00PII: S0013-4686(97)10174-8

*Author to whom correspondence should be addressed.

2359

conditions, the dissolution rate can be predicted asa function of the concentration of various oxidizingagents. If reliable dissolution rates are to be pre-

dicted, the Tafel relationship must be known withcon®dence. Our previous predictions relied on asingle set of dissolution current measurements (6

data points recorded on a single UO2 specimen in0.1 mol lÿ1 NaClO4 (pH = 9.5) [34]. Here, wedescribe the results of a more extensive series of

measurements on a number of UO2 specimens alsoperformed in 0.1 mol lÿ1 NaClO4 (pH = 9.5).

EXPERIMENTAL

General

Two UO2 electrodes were constructed fromdi�erent pellets of sintered polycrystalline UO2

taken from unused CANDU fuel elements [37]. TheUO2 discs used in preparing the electrodes were

about 3 mm thick and 1.4 cm in diameter with ageometric surface area of 01.6 cm2. The resistivityof the electrodes ranged between 5 and 10 kO�cm.

The electrodes were polished using 600-grit SiCpaper initially and before each series of potentialstep experiments. After polishing, the electrode was

washed with pure water and reduced at ÿ2.0 V vsSCE (saturated calomel electrode) for 5 min in theelectrochemical cell. All potentials were measured,

and are quoted, against this reference electrode.The cell was of the standard three-electrode±three-compartment design and was purged with nitrogen.The electrolyte used was 0.1 mol lÿ1 NaClO4. The

solution pH was adjusted using HClO4. TheNaClO4, HClO4, and NaOH were obtained fromAldrich, Mallinckrodt, and Fisher Scienti®c

Company, respectively. All experiments were con-ducted at room temperature (0228C).Electrochemical data were obtained using a

potentiostat/galvanostat (EG and G PrincetonApplied Research model 273) controlled by a PCcomputer using a program PARC [38]. Data were

recorded using an Electronic Controls Design Inc.model 50 data-logger. The angular velocity of therotating-disc electrode was controlled by a PineInstruments Model AFASR analytical rotator.

Unless otherwise stated, experiments were carriedout at an angular velocity (o) of 16.7 Hz so as toprovide constant stirring conditions with consistent

removal of dissolved species during the polarization.Potentials were monitored using a Keithley 614electrometer.

Procedure to record anodic oxidation/dissolution

curves

Figure 1 shows the potential-time pro®les used torecord the anodic oxidation/dissolution currents onrotating disc electrodes. In all cases the electrode

was ground and cathodically reduced at ÿ2.0 V for~5 min to remove any ``oxidized'' ®lms.Subsequently, two di�erent procedures were used to

record currents as a function of applied potential:(1) The potential was pulsed directly from ÿ2.0 V

to a potential in the range +0.100 V to +0.500 Vand the current recorded as a function of time [(2),

Fig. 1. Schematic showing the potential-time pro®les used to obtain anodic dissolution currents for rotating disc electro-

des (RDE) of UO2: (1) potential at which the electrode was preoxidized [+0.1 V (vs SCE) for 24 h]; (2) anodic dissol-

ution current recorded at individual potentials with repreparation of the electrode in between; (3) anodic dissolution

current recorded at a sequence of consecutive potentials; (4) voltammetric scan to ÿ2.0 V (vs SCE) at 20 mV sÿ1 after

recording anodic currents by either (2) or (3), above.

S. Sunder et al.2360

Fig. 1]. The potential was then swept back toÿ2.0 V and the current±potential relationship

recorded [(4), Fig. 1]. The electrode was then repre-pared and cathodically reduced again at ÿ2.0 V for5 min before repeating the procedure at a di�erent

potential.(2) The electrode was preoxidized at +0.100 V

for various lengths of time between 1.5 and 26 h

[(1), Fig. 1]. Then the potential was pulsed to a con-secutive series of more positive potentials, in therange +0.100 to +0.500 V and the current

recorded for up to 1.5 h at each potential beforemoving to the next [(3), Fig. 1]. In a small numberof cases the currents were measured for >1.5 h atcertain potentials. Following the recording of cur-

rent at the most positive potential, the potentialwas swept at a rate of 20 mV sÿ1 from this anodiclimit to ÿ2.0 V and the current±potential relation-

ship recorded [(4), Fig. 1].

RESULTS

Figure 2 shows logarithmic current (i)±time (t)

plots recorded individually at a series of potentialswithout preoxidation at +0.100 V. At short times[log t< 0.1 (in minutes)], the current shows a simi-lar dependence on time for all potentials although

the absolute current increases with increasing poten-tial. For longer times (log t>0.1), the currentdecays with time over the duration of the exper-

iment providing E < 0.35 V. This suggests oxi-dation of UO2 to produce ®lms which eventually

block its further oxidation. The ability of these

®lms to block further oxidation of the surface can

be appreciated by the potentiostatic behavior

observed at +0.10 V over 26 h (Fig. 3). After this

time, the current has fallen to 015 nA (03 nA cmÿ2

assuming a uniform distribution of current and an

electrode surface area of 04.8 cm2 due to a rough-

ness factor of03 [37]) and shows no apparent trend

towards a steady-state value. This current represents

a lower limit below which a meaningful value is

unobtainable. If a steady state does exist it is only

at an extremely low current value. For E>0.35 V,

the current achieves a steady-state value, indicating

that oxidation of the surface does not lead to its

blockage (Fig. 2). The potential range over which

this transition from decaying to steady-state cur-

rents occurs is not particularly well-de®ned, and

variability in behavior is often observed in the

range 0.30 < E < 0.40 V.

Figure 4 shows a similar series of logarithmic i±t

plots recorded consecutively at a sequence of poten-

tials after preoxidation at +0.100 V for 024 h.

Again, for E < 0.35 V, the current decays continu-

ously with time over the 1.5 h period for which the

potential is held at each value in the sequence. The

currents at short times are almost an order of mag-

nitude lower than those in the absence of preoxida-

tion due to the partial blockage of the oxidation

process by ®lm growth at the preoxidation potential

of +0.100 V. At higher potentials (+0.450 V in

Fig. 4, but generally at any potential >0.35 V) the

current increases with time ®nally achieving a

Fig. 2. Logarithmic current±time plots recorded individually (by Procedure 2, Fig. 1) at a series of potentials by pulsing

directly from ÿ2.0 V without preoxidation at 0.1 V (vs SCE) in 0.1 mol lÿ1 NaClO4 (pH= 9.5); o= 16.7 Hz.

Anodic oxidation and dissolution of CANDU fuel 2361

steady-state value (iss) after 01.5 h. This increase

suggests that the oxidized surface ®lms grow and

block oxidation at lower potentials, but are

removed at higher potentials. The oxidized layer

remaining on the surface at high potentials, is essen-

tially ine�ective in blocking the oxidation of UO2

[Figs 3 and 4 (vide-supra)].

The presence of oxidized surface ®lms is con-

®rmed by the i±E relationships recorded when the

potential is scanned from the anodic limit to ÿ2.0 V(Fig. 5). These scans were recorded after completion

of the anodic oxidations for which the log iÿ log t

plots are shown in Fig. 2. Three reduction peaks [1

(ÿ0.67 to ÿ0.75 V), 2 (ÿ0.83 to ÿ0.87 V) and 3

(ÿ0.1 to ÿ0.3 V)] are observed and, as demonstrated

previously [30, 37], can be attributed to the re-

duction of ®lms formed on the UO2 surface during

anodic oxidation. Peak 1 is formed even at very low

Fig. 3. Logarithmic current±time plot recorded at +0.100 V (vs SCE) in 0.1 mol lÿ1 NaClO4 pH= 9.5; o= 16.7 Hz.

Fig. 4. Logarithmic current±time plots recorded consecutively (by Procedure 3, Fig. 1) at a sequence of potentials after

preoxidation at 0.100 V (vs SCE) for024 h in 0.1 mol lÿ1 NaClO4 (pH = 9.5), o= 16.7 Hz.

S. Sunder et al.2362

potentials, i.e., <ÿ0.3 V [30]. Peak 2 is present after

oxidation at 0.175 V and increases in size as the

anodic potential is increased up to 00.35 V.Although not shown here, peak 2 can be observed

after extensive oxidation (for 24 h) at potentials as

low as +0.100 V. It appears that the growth of thisoxide is the cause of the current decay (Figs 3 and

4). After oxidation at anodic limits >0.35 V peak 2

is either much smaller or absent altogether. Thebehavior of the minor peak 3 is di�cult to deter-

mine but appears to be similar to that of peak 2,

although it is present to a more anodic potential of

ÿ0.375 V compared to peak 2.

The charge consumed in reducing these oxidizedsurface ®lms, QC, can be determined by integrating

the area under the reduction peaks in Fig. 5 as indi-

cated by the shaded area for the i±E pro®lerecorded after 1.5 h of anodic oxidation at

+0.175 V. Figure 6 shows a plot of the charge ratio

QC/QA as a function of the applied anodic potential(EA) for many such experiments. All the data in

this plot are from experiments in which the oxi-

dation was performed potentiostatically at a singlepotential. The total charge, QA, is that consumed

by anodic oxidation at this single potential (EA).

For oxidation at potentials <0.30 V the QC/QA

charge ratio is 00.9, an indication that the majority

of the charge consumed during anodic oxidation isrecovered during the cathodic reduction of the sur-

face oxidized ®lms. (The single value of QC/QA>1

(at 0.260 V) suggests a value of QC in¯uenced bythe reduction of residual traces of oxygen in the sol-

ution.) Thus, in this potential range, in which cur-

rents decay logarithmically with time (Fig. 2),

oxidation leads predominantly to the formation ofsurface ®lms. The small fraction, 0.1, of the anodic

charge unaccounted for by the subsequent ®lm re-

duction process could be lost by dissolution (asUVI). Alternatively, its loss could be attributable to

the <100% e�ciency of the reduction process.

Analysis of the reduced surface by X-ray photo-electron spectroscopy shows that its composition is

0UO2.1, and photoelectrochemical studies con®rm

that the oxidation is not totally reversible [30].

For E>0.30 V, the ratio QC/QA decreases mark-

edly, mainly due to an increase in the value of QA.Values of 00.2 are achieved for E>0.40 V and, for

E = 0.50 V, the ratio is e�ectively zero. Since little

of the anodic charge is recovered during the catho-dic reduction sweep it can be considered consumed

to produce dissolved UVI species which are lost by

transport to the bulk of the solution. This occurs inthe potential region where steady-state currents are

observed (Figs 2 and 4); therefore, these currents

can be attributed to oxidative dissolution.

The ®nal values of the currents recorded over the

whole potential range (+0.100 to +0.500 V) areplotted logarithmically as a function of applied po-

tential in Fig. 7. On the basis of the evidence pre-

sented above, the plot can be divided into twodistinct regions: a region, for E < 0.35 V when oxi-

dized ®lms build up and block the electrode surface,

and a region, for E>0.35 V, when steady-state dis-solution predominates. The two major sets of data

Fig. 5. Cathodic stripping voltammograms (CSV) recorded at each potential after completing the anodic oxidations

described in Fig. 2. The curves are vertically o�set by 50 mA.

Anodic oxidation and dissolution of CANDU fuel 2363

Fig. 6. The charge ratio QC/QA, as a function of applied anodic potential (EA). QA values were obtained from exper-

iments in which the anodic oxidation was performed potentiostatically at a single potential. QC values were obtained

from CSVs.

Fig. 7. Logarithmic current (i)±potential (E) plots recorded on rotating disc electrodes cut from di�erent UO2 pellets in

0.1 mol lÿ1 NaClO4 (pH = 9.5), o= 16.7 Hz: (r) currents recorded on electrode L; (w) currents recorded on electrode

R; (R) currents recorded on electrode L after 14 to 20 h of oxidation at each potential; (.) current recorded on electrode

R after 26 h of oxidation. (Ð) linear ®t to the points for E>0.30 V (vs SCE); (ÐÐÐ) linear ®t from previously pub-

lished data [29].

S. Sunder et al.2364

points (r, w) show currents obtained on two UO2

electrodes cut from di�erent UO2 fuel pellets. Thecolumns of data points at +0.100, +0.200 and+0.300 V show the e�ects of prolonged potentio-

static oxidation at these potentials. The larger cur-rents were obtained after oxidation for 1.5 h

whereas the lower currents were obtained after pro-gressively longer periods of oxidation. The lowest

current plotted for +0.10 V was recorded after 26 hof oxidation (Fig. 3) and the lowest points at+0.20 and +0.30 V were recorded after 024 h.

Even for these durations, the currents had shownno tendency to achieve steady-state, con®rming that

a steadily increasing blockage of the oxidation pro-cess by ®lm formation was occurring.

Although the results are not included in Fig. 7,some experiments were also performed as a function

of electrode rotation rate. The e�ect of electrode ro-tation rate (0 to 25 Hz) on the oxidation currentwas minor demonstrating that the formation of

oxide ®lms was not signi®cantly a�ected by thetransport of dissolved uranyl species away from the

electrode surface. Clearly, a simple dissolution/pre-cipitation process cannot be invoked to explain ®lm

formation. If anything, there is some evidence thatthe currents in the transition region between non-steady-state and steady-state behavior (0.30 to

0.40 V) are slightly higher at stationary electrodes.For E>0.35 V, the currents increase with poten-

tial, but are quite scattered, e.g., the currents for0.40 V range over nearly two orders of magnitude.

A linear ®t to the data for E>0.30 V yields a Tafelslope of 5825 mVÿ1 (thick line in Fig. 7). This isclose to the 62.5 mVÿ1 obtained with our previously

published data [15] (thin line in Fig. 7). If we ®t the

data points for the two di�erent electrodes indepen-dently then Tafel slopes of 6228 mVÿ1 [r, Fig. 8]and 5424 mVÿ1 [w, Fig. 9] are obtained for elec-

trodes L and R, respectively. Closer inspection ofthe data points obtained with electrode R suggests

a ®t to two distinct linear regions, 0.35RER0.40 Vand 0.40RER0.55 V, may be more appropriate.

When the data for electrode R are ®tted in thismanner, Tafel slopes of 3124 and 109222 mVÿ1

are obtained [Fig. 10].

A similar scatter is also apparent in the values ofQA and QC when plotted as a function of EA, the

potential at which the oxidation was performed(Fig. 11). Around 00.35 V, QA begins to increase

markedly, re¯ecting the di�erences in i±t behaviorobserved above and below this potential (Figs 2

and 4). The ranges of QA and QC values are wide,especially in the potential range 0.35E < 45 V.Within this potential range, low values of QC are

inevitably recorded in experiments for which highvalues of QA are obtained and vice versa. Thus, a

signi®cant amount of scatter in the dissolution cur-rents in the steady-state region of Fig. 7 can be

attributed to the irreproducibility of the balancebetween oxidative ®lm formation on, and oxidativedissolution of, the UO2 electrode surface.

Figure 12 shows plots of iss and QA as functionsof the ®nal pH of the solution. In this ®gure QA is

the total charge consumed either by anodic oxi-dation at a single potential, EA>0.35 V, or at a

sequence of potentials culminating in a ®nal oxi-dation at EA, and iss is the ®nal steady-state dissol-ution current achieved at EA and plotted in Fig. 7.

Fig. 8. Tafel plots for the data recorded at E>0.300 V for electrode L (from Fig. 7).

Anodic oxidation and dissolution of CANDU fuel 2365

Clearly, the higher the anodic dissolution current(iss) and the larger the total extent of oxidation(QA), the lower is the ®nal pH of the solution due

to the hydrolysis of dissolved uranyl species.

DISCUSSION

The surface chemistry of UO2 is complex in thepotential region investigated here (+0.10 to

+0.50 V) [30, 36, 37], and the results presenteddemonstrate it has a major in¯uence on the oxi-dative dissolution process. Using a combination of

electrochemical and XPS techniques, we have pre-viously assigned peak 1 (ÿ0.67 to ÿ0.75 V) in Fig. 5to the reduction of UO2.33 to UO2+x, peak 2 (ÿ0.83to ÿ0.87 V) to the reduction of UO2.67 to UO2+x

and peak 3 (ÿ0.1 to ÿ0.3 V) to the reduction ofUO3�xH2O to UO2+x [30, 37].The occurrence of peak 3 at a stripping potential

as positive as 0ÿ 0.2 V indicates that this species iseasily reduced, and our original assignment of thispeak to the reduction of a secondary phase formed

by a solution-mediated recrystallization(UO3�xH2O) now seems less certain [37]. Toaccount for the fact that only a very small charge

(QC) was associated with the reduction of this ®lm,irrespective of whether it was formed electrochemi-cally or by natural corrosion, we postulated that it

was only present in small quantities in grain bound-aries. This suggests that both electrochemical andcorrosion reactions are concentrated at grainboundary sites.

Despite the small amount of charge involved, this

layer, when present as a consequence of natural

corrosion, supports higher currents for the electro-

chemical reduction of oxygen than those measured

when it is absent. This is not consistent with its

being an insulating layer. The reduction of oxygen

on UO2 electrodes appears to involve the utilization

of donor±acceptor sites on the electrode

surface [39, 40], i.e., adjacent uranium cations must

be in di�erent oxidation states (i.e., UO2.33,

UO2.67). The optimum con®guration would be a

surface monolayer containing UVI cations with

adjacent U cations in a lower oxidation state (i.e.,

UV or UIV). The presence of such a monolayer

would be consistent with our claim that a layer con-

taining UVI species exists on the electrode surface

and can act as a precursor to either the formation

of oxide ®lms or dissolved uranyl species [37]. In

our earlier papers we designated this as an adsorbed

UO22+ layer. While its actual chemical identity may

remain in doubt, the results presented here are con-

sistent with its existence as a precursor to ®lm for-

mation/dissolution. Whether or not it exists only at

grain boundaries has not been determined.

The assignment of peak 1 (Fig. 5) to the re-

duction of a UO2.33 layer to UO2>+x is consistent

with all our previous observations under both elec-

trochemical and natural corrosion conditions. A

layer close in stoichiometry to UO2.33 has been

shown to form by oxidation of the UO2 at poten-

tials as low as ÿ0.100 V ( [41]).

UO2 � 0:33H2Oÿÿÿ4UO2:33 � 0:66H � � 0:66eÿ �1�

Fig. 9. Tafel plot for the data recorded at E>0.300 V for electrode R (from Fig. 7).

S. Sunder et al.2366

The present results show that such a layer remains

on the electrode surface even after steady-state

anodic dissolution for extended periods of time at

positive potentials >0.35 V (E in Fig. 5).

It is the ®lm reduced at peak 2 (Fig. 5) that

appears to account for the current decay observed

for E < 0.35 V. The size of this peak increases with

increasing potential, or with the duration of oxi-

dation at a single potential, in the low potential

region, but decreases and eventually disappears

under the steady-state dissolution conditions which

prevail at higher potentials (Fig. 7). Even when a

substantial layer of this oxide is grown at

E < 0.35 V, subsequent potentiostatic oxidation at

higher potentials leads to its removal and an

increase in the current until steady-state dissolution

conditions are eventually achieved (Fig. 4).

We have shown that anodically grown oxide ®lms

suppress the cathodic reduction of oxygen (at

E <ÿ0.30 V) [39, 40]. While a direct correlation

between this suppression and the presence of the

®lm reduced at peak 2 (Fig. 5) remains to be estab-

lished, it is clear that this oxide possesses insulating

properties. This puts in doubt our previous assign-

ment of this peak to the reduction of a layer of

UO2.67. If our previous claim were correct then the

removal of UO2.67 at more positive potentials

(+0.40 V in Fig. 4) could be attributed to the oxi-

dative dissolution of this phase to produce dissolved

UO22+. Although the UO2.67 phase is expected to be

unstable with respect to UO3�xH2O at a potential of

0ÿ 0.20 V, it may not be totally converted to

UO3�xH2O on account of kinetic factors. Since

UO2.67 is not an insulator [8] it would not be

expected to block the oxidation of UO2 by requir-

ing an overpotential of >0.55 V for its oxidative

dissolution.

An alternative possibility is that peak 2 is due to

the reduction of a layer of UO3�xH2O that does

possess insulating properties. From the charges as-

sociated with the reductions at peaks 1 and 2

(Fig. 5) after oxidation in the potential range 0.35

to 0.40 V, we estimate that, if evenly distributed

across the electrode surface, these oxide ®lms would

be 020 to 30 nm thick. Since peak 2 requires ap-

proximately one half of this charge, the UO3�xH2O

would be 10±15 nm thick. This is extremely thin for

an insulating layer formed by a solution-mediated

recrystallization. Such layers would be expected to

be somewhat porous and to require growth to

much greater thicknesses in order to block the elec-

trode surface. Also, if formed uniformly across the

electrode surface via dissolution and precipitation,

we would expect the thickness of this oxide to be

sensitive to the angular rotation rate of the disc

electrode. This is not observed, the dependence of

current on electrode rotation rate being very small.

All this evidence suggests that the UO3�xH2O

could be located at occluded sites on the electrode

surface, most likely grain boundaries and other

reactive surface asperities. Then the insensitivity to

hydrodynamic conditions would not be surprising.

The formation and removal of this oxide ®lm

would then depend on the local chemical conditions

Fig. 10. Tafel plot for the data recorded at E>0.300 V for electrode R (from Fig. 7) assuming two distinct linear regions

(0.35RER0.40 V and 0.40RER0.55 V) (vs. SCE).

Anodic oxidation and dissolution of CANDU fuel 2367

at these sites. Scanning electron microscopic exam-

ination of electrodes subjected to extensive electro-

chemical oxidation showed extensive attack at grain

boundaries [42].

In an electrochemical, as opposed to natural cor-

rosion experiment, the anode and cathode are sep-

arated on di�erent electrodes in separate

compartments of the cell. Consequently, there is the

possibility that local excursions in chemical con-

ditions, especially in pH, could exist at the working

electrode, in this case, the UO2 electrode. The incor-

poration of O2ÿ into surface ®lms during ®lm

growth (UO24UO2+x) and the production of

hydrolyzable UO22+ species due to dissolution will

both lead to the production of protons. Under

natural corrosion conditions this will be, at least

partially, neutralized by the production of hydroxyl

ions due to O2 reduction

O2 � 2H2O� 4eÿÿÿ44OH ÿ �2�at adjacent sites on the same surface. Under electro-

chemical conditions the buildup of local acidity will

not be neutralized, and if occurring predominantly

in occluded sites, will not necessarily be prevented

by rotation of the electrode. The data plotted in

Fig. 12 clearly show that extensive anodic dissol-

ution does lead to a decrease in pH.

We can summarize the electrochemical oxidation/

dissolution of UO2 in the following mechanism.

Oxidation of the UO2.33 layer leads to the for-

mation of an intermediate UVI species,

UO2:33 � 0:33H2Oÿÿÿ4�UO2�2 �ads � 0:66OH ÿ � 1:34eÿ

�3�This intermediate may subsequently dissolve orbecome incorporated into oxidized surface phases,

3UO2:33 � 3�UO2�2 �ads � 3H2Oÿÿÿ46UO2:67 � 6H �

�4�

�UO2�2 �ads � �1� x�H2Oÿÿÿ4UO3 � xH2O� 2H �

�5�Of these two options (i.e., equation (4) or

equation (5)) the formation of UO3�2H2O by thehydrolysis of (UO2

2+)ads appears the kineticallymore feasible reaction.

In the low potential region (E< 0.35 V) in which®lm growth predominates, the development ofacidity by the occurrence of equation (4) and/or

equation (5) at surface asperities does not appear tobe su�cient to increase the solubility of UVI speciesand prevent their incorporation into oxidized sur-

face phases which eventually block equation (3). Anattempt to illustrate this process is shown schemati-cally in Fig. 13(A) and (B). While these schematics

show blockage occurring predominantly in grainboundaries it is acknowledged that the e�ect couldbe occurring at a wide range of surface occluded

sites.At higher potentials (E>0.35 V) the production

of uranyl species (UO22+) is rapid and dissolution

more extensive. The rapid and extensive hydrolysis

Fig. 11. QA (*), QC (w) values as a function of the single applied potential at which the anodic oxidation was per-

formed (i.e., Procedure 2, Fig. 1). Values of QA obtained from integrated i±t plots: values of QC from CSVs. The lines in

this ®gure are only to show the trends for the QA and QC values.

S. Sunder et al.2368

of these uranyl species

2UO2�2 � 2H2Oÿÿÿ4�UO2�2�OH �2�2 � 2H � �6�

3UO2�2 � 5H2Oÿÿÿ4�UO2�3�OH ��5 � 5H � �7�

could lead to the development of a low pH at the

oxidation sites in occluded areas. The solubility of

UO3�xH2O increases markedly below pH05.5 [15±

19] and its formation could thus be prevented. This

would leave the oxidation sites unblocked by the

presence of oxidized phases and free to dissolve at a

current dictated by the applied potential, as indi-

cated in Fig. 13(C). The large variations in dissol-

ution current values (Fig. 7) and in QC values

(Fig. 11) for the same anodic treatment of the elec-

trode indicates that the conditions in occluded sites

are widely variable, and not necessarily reproduci-

ble, both on the same electrode and from one speci-

men of UO2 to the next. It is clear from the steady-

state dissolution currents plotted in Fig. 7 that,

although the dissolution mechanism may remain

Fig. 13. Schematic illustrating how oxidation and dissolution of UO2 could be localized at grain boundaries. (A)

Oxidation/dissolution at short times at ER0.350 V; (B) oxidation/dissolution at long times at ER0.350 V; (C) steady-

state dissolution at E>0.350 V.

Fig. 12. Plot of the steady±state currents (iss), Q; and anodic charges (QA), q; as a function of the ®nal pH of the bulk

solution for anodic oxidation at Er0.35 V (vs SCE).

Anodic oxidation and dissolution of CANDU fuel 2369

unchanged, no single simple equation precisely

describes the relationship between potential and dis-

solution current for all UO2 specimens.

Previously, Nicol and Needes proposed the fol-

lowing mechanism for the anodic dissolution of

UO2 [32],

UO2 �H2Oÿÿÿ4UO2OH�H � � eÿ �8�

UO2OHÿÿÿ4UO3 �H � � eÿ �9�

UO3 � 2H �ÿÿÿ4UO2�2 �H2O �10�

While di�erent in format, this mechanism has close

similarities to the one presented in this report.

While we have not speci®cally proposed a UV inter-

mediate, the thin UO2.33 layer (~5 to 8 nm [37]) can

be thought of as a surface layer in which the

uranium species are in the oxidation states IV and

V with two UV species for each UIV. The UO3 layer

proposed by Nicol and Needes as an intermediate

in the dissolution process is similar in concept to

the UVI intermediate we claim as reduced at peak 3

(Fig. 5).

Given these similarities, the kinetic scheme of

Nicol and Needes can be adapted in an attempt to

describe our results for dissolution under steady-

state conditions (E>0.35 V). In acidic solutions,

equivalent to those proposed to exist in occluded

sites, the dissolution of UO3 will be rapid compared

to its rate of production and coverage of the surface

by the UO3 intermediate will be negligible. The dis-

solution current is then given by

iD � 2Fk2y exp��1ÿ b2�

EF

RT

��11�

where y is de®ned by Nicol and Needes [43] as the

fractional surface coverage by UO2OH, k2 is the

rate constant and b the symmetry factor for

equation (9). According to our reaction scheme, it

is more appropriate to view y as a variation in stoi-

chiometry of the UO2.33 with applied potential.

Some experimental evidence exists to support this

view [42].

Under steady-state conditions the rate of for-

mation of Uv (or the UO2.33 layer) is given by

v1 � k1�1ÿ y�exp��1ÿ b1�

EF

RT

��12�

and its rate of disappearance by

v2 �k2y exp��1ÿ b2�

EF

RT

�

� kÿ1�H ��y exp

�ÿ b1

EF

RT

��13�

where k1, kÿ1 are the rate constants for equation (8)

which is taken to be reversible and b1 is the sym-

metry factor for this reaction.

Eliminating y by substitution in equation (11)

yields

iD �2Fk1k2 exp��1ÿ b1�

EF

RT

�

exp

��1ÿ b2�

EF

RT

�� Aÿ1 �14�

where A is given by

A �k1 exp��1ÿ b1�

EF

RT

�� k2 exp

��1ÿ b2�

EF

RT

�

� kÿ1�H ��exp�ÿ b1

EF

RT

��15�

Two possible approximations exist.

(a) For low anodic potentials

kÿ1 � k1, k2

and only the third term in A will be signi®cant. The

dissolution current is then given by

iD � 2Fk1k2�kÿ1�ÿ1�H ��ÿ1exp��2ÿ b2�

EF

RT

��16�

which predicts a Tafel slope of 40 mVÿ1 if it is

assumed that b2=0.5. It also predicts an inverse

acid dependence with a cathodic shift of 40 mVÿ1

for each unit increase in pH. This equation does

describe both the Tafel and pH behavior observed

by Nicol and Needes for pH < 2, but the Tafel

slope is inconsistent with our observation

(5825 mVÿ1).(b) For high anodic potentials

kÿ1 � k1, k2

and the last term in A can be neglected yielding for

iD the expression

iD � 2Fk1k2�k1 � k2�ÿ1exp��1ÿ b� EF

RT

��17�

providing it is assumed b1=b2=b. This expression

predicts no dependence on pH and a Tafel slope of

120 mVÿ1, a value which is again inconsistent with

our observations.

For pH values around 9, Nicol and Needes

obtained Tafel slopes of 70 and 80 mVÿ1 in sodium

perchlorate solutions. Our values for the two di�er-

ent UO2 electrodes 5424 and 6228 mVÿ1 are clo-

ser to the values they obtained in sodium sulphate

solutions. They noted that slopes close to 60 mVÿ1

suggested a single two electron transfer was

involved but placed little faith in such a conclusion.

For the L electrode a ®t to two Tafel slopes,

3124 mVÿ1 at low potentials and 109222 mVÿ1

at higher potentials, could be obtained, consistent

with the two limiting kinetic cases for acidic sol-

utions (pH < 2) expressed by equations (16) and

(17). However, given the scatter in the data, such a

®t must be considered tentative at best, and we can-

S. Sunder et al.2370

not unequivocally conclude that the dissolutionreaction occurs in two kinetically distinguishable

one electron transfer steps as expected in acidic sol-utions. An equally feasible explanation for thedecreased dependency of dissolution current on po-

tential observed for electrode L at high potentials[Fig. 10] is the reaccumulation of precipitatedUO3�xH2O. For the high anodic dissolution cur-

rents achieved, the rate of local production ofUO2

2+ species in occluded sites may be su�cient toexceed the solubility product despite the accompa-

nying local acidi®cation.

SUMMARY AND CONCLUSIONS

The anodic oxidation and dissolution of UO2 inneutral and slightly basic solutions can be divided

into two distinct regions of behavior.(1) For 0.10RE < 0.35 V the electrode becomes

covered with oxidized surface ®lms which block the

further oxidation of the electrode. The electrode iscovered by a thin (5 to 8 nm) layer of UO2.33 and adeposited ®lm of UO2.67 or, more probably,UO3�xH2O which appears to be located at reactive

occluded surface sites.(2) For E>0.35 V the electrode undergoes

steady-state dissolution and the dissolution currents

obey a Tafel relationship, although they vary fromone electrode to another. Dissolution appears to beconcentrated at occluded sites. Hydrolysis of the

uranyl dissolution product leads to local acidi®ca-tion at these sites. This drop in pH either preventsthe formation of the oxidized surface layer which

blocks oxidation and dissolution at lower potentialsor causes its redissolution if it has been preformedat lower potential.The Tafel slope of 5825 mVÿ1 suggests a single

two electron transfer is involved but this seemshighly improbable. There is some evidence with oneof the two electrodes used that dissolution under

steady-state conditions may occur in two kineticallydistinguishable one electron transfer steps asexpected for acidic conditions. However, given the

scatter in the data, this evidence cannot be con-sidered conclusive.

ACKNOWLEDGEMENTS

We are grateful to W. H. Hocking and S. Stroes-

Gascoyne for helpful comments on the manuscript.This work is part of the Canadian Nuclear FuelWaste Management Program funded by Ontario

Hydro.

REFERENCES

1. Johnson, L. H., Leneveu, D. M., King, F., Shoesmith,D. W., Kolar, M., Oscarson, D. W., Sunder, S.,Onofrei, C. and Crosthwaite, J. L., The disposal ofCanada's nuclear fuel waste: a study of post-closure

safety of in-room emplacement of used CANDU fuelin copper containers in permeable plutonic rock. Vol.2, Atomic Energy of Canada Limited Report, AECL-111494-2, COG-95-552-2, 1996.

2. L. H. Johnson, D. M. Leneveu, D. W. Shoesmith, D.W., Oscarson, M. N. Gray, R. J. Lemire and N. C.Garisto, The disposal of Canada's nuclear fuel waste:the vault model for postclosure assessment, AtomicEnergy of Canada Limited Report, AECL-10714,COG-93-4, 1994.

3. B. Grambow, A. Loida, P. Dressler, H. Geckeis, P.Diaz, J. Gago, I. Casas, J. de Pablo, J. Gimenes andM. E. Torrero, Long-term safety of radioactive wastedisposal: reaction of high burnup spent fuel and UO2

in saline brines at room temperature,Kernforschungszentrum, Karlsruhe, KfK 5377, 1994.

4. R. S. Forsyth and L. O. Werme, Spent fuel corrosionand dissolution, J. Nucl. Mater. 190, 3 (1992).

5. W. J. Gray, H. R. Leider and S. A. Steward,Parametric study of LWR spent fuel dissolution kin-etics, J. Nucl. Mater. 190, 46 (1992).

6. D. W. Shoesmith and S. Sunder, The prediction ofnuclear fuel (UO2) dissolution rates under waste dispo-sal conditions, J. Nucl. Mater. 190, 20 (1992).

7. D. W. Shoesmith and S. Sunder, An electrochemistry-based model for the dissolution of UO2, AtomicEnergy of Canada Limited Report, AECL-10488,1991.

8. S. Sunder and D. W. Shoesmith, Chemistry of UO2

fuel dissolution in relation to the disposal of usednuclear fuel, Atomic Energy of Canada LimitedReport, AECL-10395, 1991.

9. D. W. Shoesmith, J. C. Tait, S. Sunder, W. J. Gray,S. A. Steward, R. E. Russo and J. D. Rudnicki,Factors a�ecting the di�erences in reactivity and dis-solution rates between UO2 and spent nuclear fuels,Atomic Energy of Canada Limited Report, AECL-11515, COG-95-581, 1996.

10. I. Casas, J. Gimenez, J. de Pablo and M. E. Torrero,in Materials Research Society Symposia Proceedings,ed. C. G. Interrante and R. T. Pabalan, Vol. 294,Scienti®c Basis for Nuclear Waste Management XVI,Pittsburgh, PA, 1992, p. 67.

11. J. Gimenez, E. Baraj, M. E. Torrero, I. Casas and J.J. de Pablo, J. Nucl. Mater. 238, 64 (1996).

12. A. Loida, B. Grambow and H. Geckeis, J. Nucl.Mater. 238, 11 (1996).

13. P. D. W. Bottomley, D. H. Wegen and M. Coquerelle,J. Nucl. Mater. 238, 23 (1996).

14. V. V. Rondinella and Hj. Matzke, J. Nucl. Mater.238, 44 (1996).

15. D. J. Wronkiewicz, J. K. Bates, S. F. Wolf and E. C.Buck, J. Nucl. Mater. 238, 78 (1996).

16. J. Engelhardt, F. Feldmayer, M. Laske and G. Marx,J. Nucl. Mater. 238, 64 (1996).

17. J. Bruno, E. Cera, L. Duro, T. E. Eriksen and L. O.Werme, J. Nucl. Mater. 238, 110 (1996).

18. L. H. Johnson and D. W. Shoesmith, Spent fuel, inRadioactive Waste Forms for the Future, ed. W. Lutzeand R. C. Ewing. Elsevier Publishers SciencePublishers B.V., 1988, p. 635.

19. H. Christensen, S. Sunder and D. W. Shoesmith,Development of a kinetic model to predict the rate ofoxidation and dissolution of nuclear fuel (UO2) by theradiolysis of water, Atomic Energy of Canada LimitedReport, AECL-11102, COG-93-488, 1994.

20. B. Grambow, Used fuel dissolution and oxidation. Anevaluation of literature data, SKB Technical Report,SKB 89-13, 1989.

21. I. Grenthe, J. Fuger, R. J. M. Konings, R. J. Lemire,A. B. Muller, C. Nguyen-Trung and H. Wanner, inChemical Thermodynamics of Uranium, Volume 1,

Anodic oxidation and dissolution of CANDU fuel 2371

Chemical Thermodynamics, ed. H. Wanner and I.Forest. North Holland, Amsterdam, 1992.

22. J. Paquette and R. J. Lemire, Nucl. Sci. Eng. 79,26 (1981).

23. R. J. Lemire and P. R. Tremaine, J. Chem. Eng. Data25, 361 (1980).

24. F. Garisto and N. C. Garisto, Nucl. Sci. Eng. 90,103 (1985).

25. R. J. Lemire and F. Garisto, The solubility of U, Nb,Pu, Th and Tc in a geological disposal vault for usednuclear fuel, Atomic Energy of Canada LimitedReport, AECL-10009, 1989.

26. J. Bruno, I. Casas and I. PuigdomeÁ nech, Radiochim.Acta 44±45, 11 (1988).

27. J. Bruno, I. Casas and I. PuigdomeÁ nech, Geochim.Cosmochim. Acta 55, 647 (1991).

28. G. A. Parks and D. C. Pohl, Geochim. Cosmochim.Acta 52, 863 (1988).

29. S. Sunder, D. W. Shoesmith, R. J. Lemire, M. G.Bailey and G. J. Wallace, Corros. Sci. 32, 373 (1991).

30. D. W. Shoesmith, S. Sunder and W. H. Hocking,Electrochemistry of UO2 nuclear fuel, inElectrochemistry of Novel Materials, ed. J. Lipkowskiand P. N. Ross. VCH Publishers Inc., New York,1994, p. 297.

31. C. R. S. Needes, M. J. Nicol and N. P. Finkelstein,Electrochemical model for the leaching of uraniumdioxide: 2 Ð Alkaline carbonate media, in Leachingand Reduction in Hydrometallurgy, ed. A. R. Burkin.Inst. Min. Metall., London, U.K., 1975, p. 12.

32. M. J. Nicol and C. R. S. Needes, Electrochim. Acta20, 585 (1975).

33. M. J. Nicol, C. R. S. Needes and N. P. Finkelstein,Electrochemical model for the leaching of uranium

dioxide: 1 Ð Acid media, in Leaching and Reductionin Hydrometallurgy, ed. A. R. Burkin. Inst. Min.Metall., London, U.K., 1975, p. 1.

34. D. W. Shoesmith, S. Sunder, M. G. Bailey and G. J.Wallace, Corros. Sci. 29, 1115 (1989).

35. J. D. Rudnicki, R. E. Russo and D. W. Shoesmith, J.Electroanal. Chem. 372, 63 (1994).

36. D. W. Shoesmith, W. H. Hocking, S. Sunder, J. S.Betteridge and N. H. Miller, J. Alloys and Comps.213±214, 551 (1994).

37. S. Sunder, D. W. Shoesmith, M. G. Bailey, F. W.Stanchell and N. S. McIntyre, J. Electroanal. Chem.130, 163 (1981).

38. EG and G PARC, in Softcore Corrosion Measurement,Software operating manual, 1989.

39. W. H. Hocking, J. S. Betteridge and D. W.Shoesmith, The cathodic reduction of dioxygen onuranium oxide in dilute alkaline aqueous solution,Atomic Energy of Canada Limited Report, AECL-10402, 1991.

40. W. H. Hocking, J. S. Betteridge and D. W.Shoesmith, J. Electroanal. Chem. 379, 339 (1994).

41. S. Sunder, D. W. Shoesmith, N. H. Miller and G. J.Wallace, in Materials Research Society SymposiaProceedings, Vol. 257, ed. C. G. Sombret. Scienti®cBasis for Nuclear Waste Management XVI,Pittsburgh, PA, 1992, p. 345.

42. L. H. Johnson, D. W. Shoesmith, G. E. Lunansky, M.G. Bailey and P. R. Tremaine, Nucl. Technol. 26,238 (1982).

43. M. J. Nicol and C. R. S. Needes, The mechanism ofthe anodic dissolution of uranium dioxide, Nat. Inst.Met. Repub. S. Africa, Report No. 7079, 1973.

S. Sunder et al.2372