On the corrosion of copper inpure water

T E Eriksen1, P Ndalamba1 and I Grenthe2

1 The Royal Institute of TechnologyDepartment of nuclear chemistry

2 The Royal Institute of TechnologyDepartment of inorganic chemistry

Mars 1988

ON THE CORROSION OF COPPER IN PURE WATER

T E Eriksen1, P Ndalamba1 and I Grenthe2

1 The Royal Institute of TechnologyDepartment of nuclear chemistry

2 The Royal Institute of TechnologyDepartment of inorganic chemistry

Mars 1988

This report concerns a study which was conductedfor SKB. The conclusions and viewpoints presentedin the report are those of the author(s) and do notnecessarily coincide with those of the client.

Information on KBS technical reports from1977-1978 (TR 121), 1979 (TR 79-28), 1980 (TR 80-26),1981 (TR 81-17), 1982 (TR 82-28), 1983 (TR 83-77),1984 (TR 85-01), 1985 (TR 35-20), 1986 (TR 86-31) and1987 (TR87-33) is available through SKB.

ON THE CORROSION OF COPPER IN PURE WATER.

T-E. Eriksen1, P. Ndalamba1 and I. Grenthe2

Departments of nuclear1 and inorganic2 chemistryThe Royal Institute of TechnologyS-100 44 Stockholm

11

ABSTRACT.

Due to a remarkable publication by Hultquist questioning wellknown thermodynamic data the corrosion of copper in distilledwater has been studied. No hydrogen evolution was observedduring an exposure period of 61 days using a gaschromatographictechnique. CU2O was the only corrosion product detected bymeans of ESCA and cathodic reduction. The corrosion ratesobtained for two different qualities are much lower thancorrosion rate measured in the study by Hultquist and isascribed to the reaction between the copper foils and restoxygen initially present in the water. In conclusion thepresent investigation confirmed well establishedthermodynamics, which means that oxidation of copper by puredeoxygenated water under the formation of hydrogen as proposedby Hultqvist is not thermodynamically feasible.

Ill

CONTENTS

Page

ABSTRACT ii

SUMMARY iv

INTRODUCTION 1

EXPERIMENTAL 1

RESULTS 2

DISCUSSION 2

CONCLUSIONS 4

REFERENCES 5

TABLES 6

FIGURE LEGENDS 10

FIGURES

SUMMARY.

Hydrogen evolution by corrosion of thin copper foils indeareated distilled water was studied in laboratory experimentsusing a gaschromatographic technique.The corroded copper surface was analyzed by ESCA and cathodicreduction.No hydrogen evolution was observed during an exposure time of61 days.The formation of the corrosion product Cu^O is ascribedto a reaction between the copper surfaces and small amounts ofresidual oxygen in the reaction vessels.

INTRODUCTION

Copper has been proposed as a reasonably inert- material for theproduction of canisters for radioactive waste. The mainadvantage of copper is its stability in ground water (Cu is anoble metal). A thorough discussion of the stability of copperand possible corrosion reactions has been given in technicalreports published by the Swedish Nuclear Fuel and WasteManagement Co (The Swedish Corrosion Institute and itsreference group 1978,1983). The work has also been summarizedby Mattsson (1984). However, the corrosion of copper in pureoxygen free water has been disputed in a recent publication(Hultguist 1986). In his study the corrosion of copper ininitially aerated distilled water was followed by monitoringthe hydrogen (H,) concentration in the reaction vessel with asolid electrolyte probe described by Lyon and Fray (1984),working on the basis of a concentration cell. From the voltageversus time of exposure plot the Ho production by a 99.7 wt%copper foil with surface area 1460"cm3 was found to be 2*2'10~6

g'h"1' corresponding to a corrosion rate of 0.1 jig'cm"2"h"1. Infact the highest partial pressure of H? is close to 0.1 atm.This observation is at variance with tne very well establishedthermodynamic data for the Cu-H2O-O2 system as shown e.g. inpotential-pH diagrams.

Hultquist indicates that his finding is a result of a non-reversible cathodic reaction for which "we cannot predict anupper hydrogen pressure (in an oxygen-free corroding system)".This statement is of course a violation of the second law ofthermodynamics, a fact which seems to have escaped the author,the reviewers and the editor of Corrosion Science. The papermight have been left to obscurity were it not for the nuclearwaste aspects. In fact the results of Hultquist were firstreported in an interview in a Swedish weekly technical journal"Dagens Industri" (Hultquist 1984). References to Hultquistsarticle have also been made by various organizations (includingsome universities) when reviewing the Swedish nuclear wastemanagement research program.

Ignorance of thermodynamics seems to be more wide-spread thanwe thought possible. Due to the important application of coppercanisters for final storage of nuclear waste we have made areinvestigation of the copper corrosion by pure deoxygenatedwater using an alternative technique to monitor hydrogen. Inthis way we will also demonstrate what has gone astray withHultqui6t's experimental work.

EXPERIMENTAL.

The copper foil qualities used in this investigation are Merck>99.7 wt% (0.1 mm thick, 1600 cm2 surface area) and Alfa99.9995 wt% (0.025 mm thick, 600 cm2 surface area).

The water used was deionized, double distilled in a quartzapparatus and deaerated by purging with Ar (AGA SR-quality)containing less than 1 ppm oxygen. This procedure does not 6eemto have removed all oxygen as 6een from our discussion below.The foil6 were rinsed with methanol followed by ethanol and

quickly dried in a stream of Ar before use. The foils wereplaced in glass vessels (1365 cm3 volume) in such a way thatall parts of the surface areas were accessible for contact withwater and the vessels flushed with Ar (AGA SR-quality). Waterwas thereafter transferred to the copper containing vessels byapplying an overpressure of Ar to the water reservoir. Thewater and gas volumes of each vessel were 900 and 4 65 cm3

respectively. Assuming the solubility of O, and H, to be 1.4and 0.85 mmol'dm"3 respectively at 25°C (CRC 1965) more than98% of these gases should be in the gas phase of the reactionvessels. To monitor the H2 the vessels were connected to thesampling loop of an Argograf (AGA; allowing small volumes ofthe gas phase to be transferred to the carrier-gas stream (Fig.1). The gaschromatograph was calibrated by applying the samesampling procedure to a corresponding vessel containing Ar andknown concentrations of H2.

At the end of the experiment the exposed copper foils werequickly dried in a stream of Ar and small areas of the surfacesanalyzed by ESCA and cathodic reduction. Larger samples (50 cm2

surface area) were immersed for 120 minutes in 50 cm3 of athoroughly deaerated 1 mol'dm"3 HCIO^ solution. The solutionswere thereafter analyzed for copper by atomic absorptionspectrophotometry. Unexposed copper foils were used to obtainreference solutions.

RESULTS.

The exposure lasted for 61 days and no H2 was detected duringthe time of the experiment. The detection limit wasapproximately 10 ppm (Figs.2,3). To ensure that the absence ofH2 was not due to any leakage small volumes of K2, yielding aconcentration in the range 100-300 ppmv in the gas phase of thereaction vessels were added and the H2 concentration monitoredfor 12 days. No change in concentration i.e. no leakage wasobserved (Fig.4).

The two species present on the surface of the exposed copperfoils were, according to the ESCA data Cu and Cu2O. The resultof the ESCA measurements was corroborated by the potential ofthe cathodic reduction, showirg that Cu2O was the onlycorrosion product. The thickness of the corrosion product layeron small areas of the most corroded parts of the copper foilswas calculated from the integrated current during the time ofreduction (Zakipour 1987). The results are given in table 1.The amount of Cu2O going into solution in 1 mol'dm"

3 HC1O4 fromexposed samples are summarized in table 2.

Discussion.

The surface of the exposed copper foils were unevenly corroded,large areas were seemingly unaffected whereas smaller areaswere clearly discoloured. The cathodic reduction was carriedout on 0.5 cm2 samples of the discoloured parts and the resultsobtained from dissolving the corrosion products in 1M HC1O4

probably better characterize the average thickness of thecorrosion product layer. The ratio between the corrosion depthsmeasured by the two methods is approximately 10. A maximum

pitting factor of 25 was assumed by the Swedish CorrosionInstitute (1978) in the previous report on copper corrosion.

Based on the detection limit of 10 ppm H2 and the reactionH2O + 2Cu — > Cu2O + H2 the upper limits for the coppercorrosion were calculated and are given in table 3 togetherwith the average corrosion rates calculated from thedissolution data in table 2.

It is quite clear, based on the upper limit for hydrogenproduction, that the amount of copper corrosion caused byoxidation of water is at most 4*10~5 i^g'cm"2^"1. By using analternative method Simpson and SchenK 11987) have decreased thedetection limit of hydrogen to 3.7*10~8 q'h~l'm2 as compared to0.1 Atg'cnT^h"1 reported by Hultquist.

Corrosion may be caused by oxygen present in the reactionvessels i.e. according to the thermodynamically feasiblereaction 2Cu + 1/2 02 —> Cu2O, the total amount of oxygenrequired by the mass balance of the corrosion products aregiven in table 4. As can be seen if the deoxygenation by Ar-purging was less efficient than approximately 97%, which isquite probable as no extreme measures were taken, the oxygencontent of the reaction vessels would be sufficient to explainthe corrosion. The higher rate of corrosion of the Alfa99.9995% copper foil with the smaller surface area also givessupport to the assumption of oxygen corrosion. The corrosionrates calculated from dissolution of the corrosion product Cu2Ofrom the exposed copper surfaces in our study are at least 5times smaller than the corrosion rate given by Hultquist. Itshould be noted that more oxygen was initially available perunit surface area in the work of Hultquist and this probablyaccounts for the higher rate of corrosion observed in hisexperiments.

Our results clearly put doubts on Hultquist's method ofmeasuring hydrogen. It is to be noted that the long timestability of the electrolyte probe has not been satisfactorilydemonstrated as the longest time of exposure in the work ofLyon and Fray is 28 hours. A likely explanation of the timedependence observed by Hultquist is that his emf systemgradually deteriorated.

By using standard free energies of formation of Cu2O and H-,0we can calculate the highest possible partial pressure of &2for the reaction

2 Cu + H2O --> Cu2O + H2(aq,P).

We obtain P(H2) * 3*10"18 atm

The concentration of molecular hydrogen in the near surfaceatmosphere has been determined to be approximately 0.65 ppmvand the concentration in water 0.44 nmol'dm"3 (Herr and Barger1978) which makes oxidation of copper by waterthermodynamically impossible.

CONCLUSIONS.

No hydrogen evolution by corrosion of copper in distilled waterwas observed during an exposure period of 61 days.

The corrosion product, determined by ESCA and cathodicreduction to be Cu2O, is most probably formed in a reactionwith small amounts of residual oxygen in the reaction vessels.

The following claims by Hultquist are erroneous

1. Hydrogen is evolved during corrosion of copper inwater i.e. copper is oxidized by water

2. The escape rate of hydrogen from the corroding systemis of decisive importance for the corrosion kineticsin water are erroneous.

The equilibrium pressure of H2 according to the equation2 Cu + H20 — > Cu,0 + H2 is 3'10"

18 atm, which is certainlymuch lower than the pressure of H2 in any aqueous system. Hencecorrosion by this reaction is not thermodynamically feasible.

ACKNOWLEDGE! ENTS.

This work was supported by the Swedish Nuclear Fuel and WasteManagement Co .The experimental work by Mr. S.O.Engman isgratefully acknowledged.

REFERENCES.

Handbook of Chemistry and Physics46th EditionThe Chemical Rubber Co 1965

F.L. Herr and W.R BargerJ. Geophys. Res. 8_3_, 6199 (1978).

G HultquiPtDagens Industri, 19 June (1984).

G. HultquistCorrosion Science ££, 173 (1986).

S.B. Lyon and D.J. FrayBr. Corros. J. 1£, 23 (1984).

E. Mattsson inScientific Basis for Nuclear Waste Management,Ed: G.J. McCarthyVol 1, 271 (1978).

J.P. Simpson and R. SchenkCorrosion Science 21i 1365 (1987).

The Swedish Corrosion Institute and its reference group,Copper as canister material for unreprocessed nuclearwaste, evaluation with respect to corrosion.KBS TR-90 (1978).

The Swedish Corrosion Institute and its reference group,Corrosion resistance of a copper canister for spentnuclear fuel.KBS TR 83-24 (1983).

S. ZakipourKorrosionsinstitutet 1987-04-02 (in Swedish).

Table 1

Thickness of corrosion product layer by cathodic reduction.

Sample

Merck>99.7

Alfa99.9995%

(1)(4)

(2)(3)

Thickness

1347484

46133229

(A)

Table

Corrosicji products dissolved in 1M HC104

(50 cm2 surface area, *0 cm solution)(50 cm2 surface area, 4

cm solution)

Copperquality

Merck> 99.7%

Alfa99.9995%

DissolvedCu mg'dm"2

21.519.621.5

11.415.09.0

304025

Average corrosionlayer thickness (A)

235

132

354

Comments

as received

discolouredsurface

slightlydiscolouredsurface

discolouredsurface

Table 3

Average Cu corrosion rate calculated from dissolution ofcorrosion products and H2 evolution at the detection limit10 ppm.

Copper quality Corrosion rate ( jig* cm"2 "h"1)

corrosion prod H2(detection limit)

Merck > 99.7% 0.014 l.l*10"5

0.0082

Alfa 99.9995% 0.021 3.6'10"5

Table 4

Oxygen consumption from average corrosion rates.

Cu-quality CU consumed Req initial O2 Purgingcone in gasphase efficiency

mg % %

Merck> 99.7% 4.3 0.65 96.4

2.4 0.36 98.0

Alfa99.9995% 1.9 0.30 98.3

rota meter

to gaschromatograph

2/f\1

gas tightsyringe,.

0)

o-rinseal

carrier gas ir(Ar)

.mixing chamber

on-off valve

torr-seal

I/flat flange reactionvessel

copper foilinwater

Fig 1: Schematic of reaction vessel and gas sampling system.

0:

10ppm

sampleN202

to

T-»s—(Hl)

Fig 2: Chromatograms for gas phase samples from reaction vessel 720 h after start ofexperiment and reference gas mixtures containing 10 and 40 ppmv respectively.

I I

oCM

LO

E6-o

LO

OCO

OLO

Oro oCM

(Aiudd)?H '

Fig 3: Calibration curve for gas sampling system.

>sO

U

O-

I >=:

oin

Fig 4: Height of H2 peak in gas chromatogram versus time plot.H2 was added to the reaction vessel at three different.

List of SKB reportsAnnual Reports1977-78TR121KBS Technical Reports 1-120.Summaries. Stockholm, May 1979.

1979TR 79-28The KBS Annual Report 1979.KBS Technical Reports 79-01 - 79-27.Summaries. Stockholm, March 1980.

1980TR 80-26The KBS Annual Report 1980.KBS Technical Reports 80-01 - 80-25.Summarios. Stockholm, March 1981.

1981TR 81-17The KBS Annual Report 1981.KBSTechnical Reports81-01 -81-16.Summaries. Stockholm, April 1982.

1982TR 82-28The KBS Annual Report 1982.KBSTechnical Reports82-01 -82-27.Summaries. Stockholm, July 1983.

1983TR 83-77The KBS Annual Report 1983.KBSTechnical Reports83-01 -83-75Summaries. Stockholm, June 1984.

1984TR85-01Annual Research and Development Report1984Including Summaries of Technical Reports Issuedduring 1984. (Technical Reports 84-01-84-19)Stockholm June 1985.

1985TR 85-20Annual Research and Development Report1985Including Summaries of Technical Reports Issuedduring 1985. (Technical Reports 85-01-85-19)Stockholm May 1986.

1986

TR 86-318KB Annual Report 1986Including Summaries of Technical Reports Issuedduring 1986Stockholm, May 1987

1987TR 87-33SKB Annual Report 1987Including Summaries of Technical Reports Issuedduring 1987Stockholm, May 1988

Technical Reports

7988TR 88-01Preliminary investigations of deep groundwater microbiology in Swedish granitic rocksKarsten PedersenUniversity of GöteborgDecember 1987

TR 88-02Migration of the fission products strontium,technetium, iodine, cesium and the actinidesneptunium, plutonium, americium in graniticrockThomas Ittner1, Börje Torstenfelt1, Bert Allard?1 Chalmers University of Technology2University of LinköpingJanuary 1988

TR 88-03Flow and solute transport in a single fracture.A two-dimensional statistical modelLuis Moreno1, Yvonne Tsang?, Chin Fu Tsang2,Ivars Neretnieks1

1 Royal Institute of Technology, Stockholm, Sweden2Lawrence Berkeley Laboratory, Berkeley, CA, USAJanuary 1988

TR 88-04Ion binding by humic and fulvic acids:A computational procedur» based on func-tional site heterogeneity and the physicalchemistry of polyelectrolyte solutionsJ A Marinsky, M M Ready, J Ephraim, A MathutiiuUS Geological Survey, Lakewood, CA, USALinköping University, LinköpingState University of New York at Buffalo, Buffalo, NY, USAApril 1987

TR 88-05Description of geophysical data on the SKBdatabase GEOTABStefan SehlstedtSwedish Geological Co, LuleåFebruary 1988

TR 88-06Description of geological data in SKBs data-base GEOTABTomas StarkSwedish Geological Co, LuleåApril 1988

TR 88-07Tectonic studies in the Lansjärv regionHerbert HenkelSwedish Geological Survey, UppsalaOctober 1987

TR 88-08Diffusion in the matrix of granitic rock.Field test in the Stripa mine. Final reportLars Birgersson, Ivars NeretnieksRoyal Institute of Technology, StockholmApril 1988

TR 88-09The kinetics of pitting corrosion of carbonsteel. Progress report to June 1987G P Marsh, K J Taylor, Z SooiMaterials Development DivisionHarwell LaboratoryFebruary 1988

TR 88-10GWHRT - A flow model for coupled ground-water and heat flowVersion 1.0Roger Thunvik1, Carol Braester*1 Royal Institute of Technology, Stockholm2 Israel Institute of Technology, HaifaApril 1988

TR 88-13Validation of the rock mechanics HNFEMPcode against Colorado school of .ninesblock test dataOve Stephansson, Tomas SavilahtiUniversity of Luleå, LuleåMay 1988

TR 88-14Validation of MUDEC against Coloradoschool of mines block test dataNick Barton, Panayiotis Chryssanthakis,Karstein MonsenNorges Geotekniske Institutt, Oslo, NorgeApril 1988

TR 88-15Hydrothermal effects on montmorillonite.A preliminary studyRoland PuschOla KärnlandJuni 1988

TR 88-16Swedish Hard Rock LaboratoryFirst evaluation of preinvestigations 1986-87and target area characterizationGunnar GustafsonRoy StanforsPeter WikbergJune 1988

TR 88-11Groundwater numerical modelling of theFjällveden study site - Evaluation ofparameter variationsA hydrocoin study - Level 3, case 5ANils-Åke Larsson1, Anders Markström2

1 Swedish Geological Company, Uppsala2 Kemakta Consultants Co, StockholmOctober 1987

TR 88-12Near-distance seismological monitoringof the Lansjärv neotectonic fault regionRutger Wahlström, Sven-Olof Linder,Conny HolmqvistSeismological Department, Uppsala University,UppsalaMay 1988

ISSN 0284-3757

CM-Tryck AB, Bromma 1986