Corrosion[1].Monograph.ica 2.01.00

29/10/2001 1

Corrosion of Copper Plumbing Tubes and the Liberationof Copper By-Products to Drinking Water

Gustavo LagosCatholic University of Chile

[Note: document has been partially formatted to ICA Environmental Monographeditorial style]

29/10/2001 2

Table of Contents

EXECUTIVE SUMMARY 4

1- INTRODUCTION 5

2- THEORETICAL ASPECTS OF COPPER CORROSION 8

2.1- Uniform Corrosion 9

2.2- Role of Films 11

3- CORROSION MECHANISMS OF COPPER PIPES IN PRACTICE 15

3.1- Blue Water and Cuprosolvency 16

3.2- Corrosion Mechanisms which have an Effect on Copper By-product Liberation 16a- Type III pitting corrosion 16

Table 3.1-Water Compositions of Type III Mechanism 18a.i- Concluding Remarks on type III Pitting 19

b- MIC Corrosion 20b.i- Preventive Measures 26b.ii - MIC Remedial Measures 27b.iii - Concluding Remarks on MIC 27

c- Jointing Corrosion 28c.i-Welding 28c.ii- Brass, Brazing and Soft Soldering 28c.iii- Mechanical Corrosion at Joints 29c.iv- Valves, Fittings and Meters 30c.v- Fluxes 30

3.3- Corrosion Mechanisms that May Have a Slight Effect on Copper By-Product Liberation 31a- Type I Pitting Corrosion 31b- Type II Pitting Corrosion 32c- Non Classified Pitting Corrosion 32d- Erosion Corrosion 33e- Cavitation 34f- Soil Corrosion 34g- Corrosion in Concrete Slabs 34h- Stray Current Corrosion 35j- Stress Corrosion Cracking 35k- Galvanic Corrosion 35l- Thermogalvanic Corrosion 36

29/10/2001 3

4 - CORROSION AND THE MANUFACTURING OF TUBES 36

4.1 - Carbonaceous Films Formed during Manufacturing 36

5- COPPER BY-PRODUCT LIBERATION TO DRINKING WATER 38

5.1- Introduction 38

5.2- Copper Concentration in Drinking Water 40

5.3- Who is Responsible for Copper in Drinking Water? 44

5.4- Diagnosis of Water Quality and Its Potential to Liberate Copper By-Products 46a- A Solubility Model and its Limitations 47

5.5 - Aspects of Copper By-Product Liberation, Water Composition, Stagnation, and Pipe Aging. 51a-Stagnation, Equilibrium and Aging 51b - pH and Carbon Dioxide 56c - Oxygen 57d - Carbonates and Bicarbonate 57e - Sulfates and Chlorides 58g- Effect of Natural Organic Matter, NOM 59h - Ionic Strength 60k- Sulfide and Chlorine 62

5.6- U.S Lead Copper Rule Approach to Copper By-Product Liberation Prevention. 63a - Exceedence Without the Use of Inhibitors. 63b - Inhibitor Effects 64c- Strategies to Increase pH 66

6- CONCLUSIONS 67

7- REFERENCES 70

29/10/2001 4

Executive summary

This paper discusses both the mechanistic (descriptions of events, i.e.,mechanisms, at a microscopic level at the metal interface with water) and thesolubility approaches (based on solubility models that predict theconcentration of copper in drinking water) in order to understand the corrosionand by-product liberation of copper plumbing tubes. The current corrosiontaxonomy - i.e., pitting corrosion, uniform corrosion, and other types ofcorrosion - is analyzed and classified in two groups: those types of corrosionthat contribute to copper by product liberation to the point of raising thecopper concentration close to or exceeding health based regulatory levels, andthose corrosion types that do not. Plumbing tubes manufacturing processes areanalyzed vis-a-vis pitting corrosion.

Copper by product liberation to drinking water is discussed in the context ofcurrent health based drinking water regulations. The effects of watercomposition, pipe age and stagnation period on copper by-product liberationare analyzed. Finally, the paper discusses regulatory approaches in order toprevent copper by-product liberation.

The main conclusions reached in this analysis are:

• Corrosion processes on copper plumbing tubes, i.e., oxidation/reduction,take place at the copper water interfase and form cuprite. At the cupritewater interfase, cuprite usually reacts electrochemically to form malachiteand tenorite above pH 7.0. In most water compositions, oxidation/reductiondoes not directly give way to liberation of copper by products to water.

• The corrosion products formed give way to dissolution and precipitationprocesses, both of which are the main mechanisms for liberation of copperby products to water and for pipe aging.

• Theory and practice can be used successfully to predict and prevent mostsituations leading to high copper by product liberation to drinking water. Inmany cases remedial action can be taken for the reduction of copper byproduct liberation without replacing plumbing systems, once copper byproduct liberation has started.

• Theoretical models that can predict corrosion products to be formed andaging of plumbing tubes for a wide range of water compositions have notyet been formulated.

29/10/2001 5

1- Introduction

This paper is mostly concerned with the copper by-product liberation todrinking water from plumbing tubes, i.e., with the circumstances which couldaffect the concentration of copper in drinking water, rather than with corrosionper se.

It has been estimated that under given conditions, for instance in newplumbing tubes, only about 3 percent of corroded copper is actually liberatedto water and the remaining 97 percent remains in the film or scale which isformed on the pipe wall (Edwards Schock et.al. 1996). In such case the factorthat controls the liberation of copper by-products to drinking water is thesolubility of the solid present in the scale (Scock, Lytle et.al. 1986). In orderto comply with the Lead-Copper Rule (1991) of the USEPA. whichestablishes the standard for copper in drinking water, the solubility model hasbeen effectively applied to a wide range of water compositions in the U.S.A.resulting in a reduction of copper present in drinking water. This aspect willbe discussed in detail in section 5.

All metal surfaces corrode thus, the essential question should not be whethermetals corrode but rather at what rate they corrode. Corrosion is anelectrochemical process whereby a cathodic and an anodic reaction take placespontaneously and the required electron transfer occurs through an electronicconductor, usually a solid, which connects the anode and the cathode.Therefore, a corrosion reaction requires that there be at least two chemicalcompounds present, one of them which can be oxidized (electron donor) andthe other one reduced (electron acceptor)(Bockris et.al. 1981). The compoundoxidized in a corrosion process has the more negative electrode potential withrespect to the hydrogen electrode potential. Whereas the compound that isreduced is the one that has the more positive potential. The corrosion potentialcan be defined as the product resulting from the combined oxidation andreduction processes and therefore its value is between the two potentials.

The possibility of the existence of a corrosion reaction can be established bythermodynamics, whereby the corrosion rate or corrosion current, is calculatedutilizing electrochemical kinetic theory. The electrochemical corrosionconstant which gives information about the rate of corrosion in a givenreaction is known as the exchange current density and the vast knowledgeabout such constants indicates that corrosion rates can vary by up to 10 ordersof magnitude, depending on the specific compounds involved in the couple

29/10/2001 6

and other conditions existing in the reaction media, such as temperature, watercomposition, presence of catalysts, inhibitors among other variables.

It is therefore possible to have a corrosion reaction taking place due to thenature of the chemical compounds present, but because the rate of corrosion isso slow no appreciable change will occur during relatively long periods oftime (years or tens of years). For instance, platinum corrodes in a 1 molarsulfuric acid solution about 1000 time slower than copper (Bard 1976). Acopper pipe is said to corrode slowly when the rate of corrosion is uniform andequal to 4*10-5 cm/year. Thus, a pipe of 3 mm wall thickness would take over60 years for the wall to be reduced to one half of its initial thickness (Edwards,Ferguson et.al. 1993) provided that the corrosion process is spread uniformlyover the pipe's surface. On the other hand a copper pipe is said to corrodequickly when the rate exceeds 4*10-4 cm/year. If the corrosion process is notuniform but concentrated on a small number of pits, then a 3 mm pipe couldbe perforated, under the appropriate water conditions, in a few months.

Corrosion processes are very dynamic. Once corrosion has started it can eithercontinue up to the point where the original compounds are depleted or it canstop due to the build up of a passivating layer which no longer allows directcontact between the original compounds which formed the electrochemicalcouple, anodic and cathodic.

A second aspect which should be addressed from the start of this review is thatall metal surfaces develop, during or after manufacture, one or more filmswhich can act either as intermediaries or as active elements for the physical,chemical and electrochemical reactions occurring between the metal and theimmersion media. The most common films developed on metal surfaces thatare in contact with the atmosphere or with water of a potable quality, areoxides, hydroxides, and carbonaceous films. But films based on chlorides andsulfates, or even sulfides, can sometimes occur. The thickness of these type offilms does not usually exceed a few hundred microns.

The formation of films on metal surfaces has been extensively studied duringthe last 30 years and it is understood today that a metal surface is never devoidof a film when immersed in water or in air. The use of modern surfaceanalysis techniques require the removal of such films prior to studying themetal itself. The removal process is technically challenging and involvescleaning the metal surface in a vacuum and transferring the metal to the testchamber, also under vacuum, where analysis takes place. The metal must

29/10/2001 7

never be in contact with air during this process. An alternative method to cleanthe surface is in situ ionic bombardment, but this can be very slow and costlywhen films are thick.

Thus, the application of thermodynamic and electrochemical theory to theinteraction between metals and the environment usually predicts the wrongresult if these films are not considered. In some cases these films act only as aphysical barrier, slowing down or impeding ion transport to the metal surface,and in these cases the thermodynamic theory applied to reactions between themetal and the immersion media can be more successful.

The corrosion behavior of metals is determined by the media which come incontact with the metal, during manufacturing, handling, installation and use.Once a known "deleterious" film structure and composition has beenestablished, it is usually difficult to eliminate it and replace it by another filmof "protective" characteristics. It is often much easier and less costly toestablish the preventive condition ex ante rather than the remedial procedureex post. To do this requires detailed knowledge of the processes that cantrigger corrosion.

Even though electrochemical corrosion theory is now highly developed todayand many corrosion mechanisms and reactions are well known, the difficultyto correctly predicting the type and rate of a specific corrosion path is due tothe existence of hundreds of alternatives. The following factors are relevant:initial composition of the copper pipe (depends on manufacture, handling, andinstallation), initial composition of water (with an average significant presenceof many salts, metals, organic compounds, and disinfectant by-products), andconditions of use (stagnation periods, temperature, flow velocity). Thecombination of these factors together with the thermodynamic and kineticcharacteristics of each intermediate set of reactants and products is oftenimpossible to predict. Laboratory experiments directed at reproducing realconditions seldom yield results identical to the original setting. Therefore,prediction of corrosion paths and outcomes is confined mostly to definingbroad sets of characteristics where certain mechanisms are thought to occur. Inthe following sections it will be seen that certain corrosion paths frequentlyintersect the paths of other mechanisms. Thus, corrosion prediction, in termsof a mechanistic approach, is not only an inaccurate practice at present butlikely to remain so in the future despite the progress of science.

29/10/2001 8

The same uncertainty is not true for equilibrium conditions, wherethermodynamics determines the amount of metal that can go into solution.

The following types of corrosion mechanisms will be reviewed in detail sincethey affect the liberation of copper by-products to drinking water: type IIIpitting corrosion and micro biologically induced corrosion, MIC.Additionally, jointing corrosion can contribute to the release of by- productspresent in copper pipe joints, and will therefore also be considered in thisreview.

At the same time, types I and II pitting corrosion, erosion corrosion,cavitation, soil corrosion, corrosion in concrete slabs, stray current corrosion,stress corrosion, thermogalvanic corrosion and galvanic corrosion, can usuallybe regarded as noncontributory in terms of liberation of copper by-products todrinking water. Nevertheless, a brief discussion will be made about thesemechanisms in order to provide the reader with a generalized view of thesubject of corrosion of copper plumbing pipes and also in order to provideinsight of the detailed mechanisms of corrosion, which in the case of type Icorrosion are thought to be known.

2- Theoretical aspects of copper corrosion

The theory of corrosion of metals has been discussed by many authors in greatdetail and it is not the purpose of this review article to revisit these theories.Specifically in the case of copper corrosion, Lucey (1967) proposed in 1967 atheory for type I pitting corrosion, that will be referred to later in section 3.3.This theory is currently accepted at present as the best understanding ofcopper pitting corrosion. However, it is not applicable to all types of pittingcorrosion nor does it explain with the necessary detail the conditions for pitinitiation. Ives and Rawson, (Ives and Rawson 1962 a,b,c,d) published in1967, four papers about the general theory of copper corrosion applicable touniform corrosion.

The following two sections will discuss uniform corrosion and the role of filmformation, both of which are essential to the understanding of copper pipecorrosion.

29/10/2001 9

2.1- Uniform Corrosion

Uniform corrosion in copper plumbing tubes occurs when the total surface ofthe tubes are attacked at an equal rate, and it occurs under specific watercompositions, installation procedures, tube fabrication methods, or design ofthe water installations. Pitting corrosion and uniform corrosion do not usuallyoccur simultaneously (Werner 1995).

Ives and Rawson state that uniform corrosion on copper metal is characterizedby a duplex film, the first of which is a cuprous oxide compact film welladhered to the metal, formed at an early stage and with good electronicconductivity. Due to the compactness of the first film, its growth generatesdisruption and a second, porous film, also cuprous oxide, grows on top of thefirst one. Several simultaneous reactions take place. Copper metal dissolves inorder to form cuprous oxide:

Cu + 1/2 H2O = 1/2 Cu2O + H+

+ e-

(1)

At the same time the reduction of oxygen takes place at the interphasebetween the porous film and the solution:

1/4 O2 + H+

+ e- = 1/2 H2O (2)

Reactions one and two added give the total reaction taking place:

Cu + 1/4 O2 = 1/2 Cu2O (3)

Cupric ion acts as an intermediary in the porous film and its interphase withthe compact film, forming cuprous oxide at the expense of reducing cupricions at one side of the porous film whereas at the other side cuprous oxide isbeing dissolved in order to form cupric ions. According to Ives and Rawsonthis is the reason why oxygen is not depleted, as a true depolarizer should:

Cu2+

+ 1/2 H2O + e-

= 1/2 Cu2O + H+

(4)

1/2 Cu2O + H+

= Cu2+

+ 1/2 H2O + e-

(5)

29/10/2001 10

This model is dependent on pH, as has been pointed out by Cruse et al.(1988).In soft waters of low pH, uniform corrosion is known to increase. When thepH is lower than 5.0, corrosion accelerates rapidly. The relationship betweenpH and copper concentration in water was studied by Shull and Becker (1960).Factors such as water hardness, temperature, the presence of other ions, mayaffect corrosion, and make copper concentration versus pH curves not strictlycomparable to each other. A more detailed discussion of the relationshipbetween copper concentration in water, pH and other variables is carried out insection 5.

Callot et.al.(1978), in perhaps the only study that experimentally confirmedIves and Rawson theory, identified malachite, atacamite, cuprous chloride,tenorite and cuprite in pitted pipes that had been annealed duringmanufacturing. Applying X ray photoelectron spectroscopy, XPS, Copper IIcompounds were found to occur only in the outer layers of the corrosionproducts, and cuprite was identified only in the inner layer. Werneret.al.(1994) suggests the formation of cuprite, with the same mechanismproposed by the theory of Ives and Rawson.

Potential (Eh) versus pH diagrams confirm the formation of cuprite on pipeinner copper surfaces in contact with drinking water. Tenorite, malachite andother copper containing solids can be formed by electrochemical mechanismson top of this cuprite film, depending on the water composition (Cruse et.al.1988). Figure 2.1 shows the Eh – pH diagram for water with 1.37•10-3 M totalcarbon content and 1.4•10-3 M sulfate. No cupric or cuprous species areliberated to solution at this total carbon content.

Figure 2.1: Eh-pH diagram

29/10/2001 11

-0.8

-0.4

0

0.4

0.8

1.2

1.6

2

2.4

-1 1 3 5 7 9 11 13pH

EH (V

)

Cu

Cu2O

CuOMal

Cu2O3hidr

Cu+2

Cu+

Log Cu = -6CuO2

-2

H2CO3 HCO3- HCO3

-CO3

-2

LangiteLog Cu = 0

CuO2-

Uniform corrosion does not usually lead to tube failure but it can produce“blue water” or “green water” and when it does, this is readily detected byconsumers. Green water is also associated with soap residue with which lowcopper ions react, producing green stains.

2.2- Role of Films

Copper corrosion does not occur only in the presence of water because theregion of solid copper stability exceeds the hydrogen potential of pure water.Corrosion requires the presence of electron acceptors, such as oxygen, in orderto occur. In this situation, sparingly soluble solid films are formed and theircomposition may be: cuprous oxide or cuprite (more than one crystallineform), yellow, red or brown in color; cupric oxide, black in color; cuprichydroxide, light blue to blue green in color; malachite, blue-green in color;azurite, blue-green in color; brochantite, light blue in colour; atacamite, greenin color; etc., depending on the species that are initially present in solution.Table 2.2-1 shows some of the compounds that can be formed.

Table 2.2-1- Some copper compounds that can be formed as a result ofpipe corrosion in the presence and absence of phosphate corrosioninhibitors.

Name of Compound FormulaCuprite Cu2O

29/10/2001 12

Cupric oxide CuOCuprous hydroxide Cu(OH)Cupric hydroxide Cu(OH)2Cupric Chloride CuCl2Chalcocyanite CuSO4

Cupric carbonate CuCO3Antlerite Cu(SO4)(OH)4

- CuO.CuSO4Azurite 2CuCO3.Cu(OH)2

Malachite CuCO3.Cu(OH)2Brochantite CuSO4(Cu(OH)2)3Atacamite Cu2(OH)3Cl

Posnjakite Cu4(OH)6SO4.H2O

Langite Cu4(OH)6SO4.2H2O – orthorombic

Wroewolfenite Cu4(OH)6SO4.2H2O – monoclinic

Copper Silicate CuSiO3.H2O

Copper Silicate CuSiO3.2H2O

The thermodynamics under which the cuprous solids are formed is poorlyunderstood, adding uncertainties of up to 100 fold in the prediction of solidphase formation (Schock and Lytle 1985), but it is clear that cuprous oxide isless stable than cupric oxide. Also the aqueous chemistry of cuprous ions,except under complexation (e.g., with chlorides and amines), is negligible incomparison with the aqueous cupric chemistry, due to the weak hydrolysisreactions of the former (Mahapatra et.al. 1967). Hydrodynamic conditions areimportant for the equilibrium between cuprous and cupric oxide. For instance,in crevices it has been shown that cupric oxide forms on the interphase withthe solution, while between the cupric oxide and the metal there is cuprousoxide.

The potentials for the formation and reduction of cuprous and cupric oxideshave been studied by Deutscher and Woods (1986), among others. Their studyshows that the reduction potentials for cuprite and for cupric oxide varies withformation conditions. For example, the major species formed on a copperwire after annealing was cuprite, although present in more than one phase.

29/10/2001 13

This finding refutes earlier research that concluded that both cuprite andcupric oxide were present because different reduction peaks on a voltamogramwere interpreted as showing the presence of copper I and II. However,Wilhelm et al.(1982) found that when the oxides are electrochemically formedit is possible to have both cuprous and cupric oxides co-existing and in contactwith each other. These comments seem to be of relevance because it is notclear from the literature that all studies would have taken these factors intoaccount.

In a recent work, Taylor and Cannington (1993) found that p-type cuprousoxide favored the protection of copper against corrosion whereas n-typecuprous oxide favored the corrosion of copper. Naturally occurring cuprousoxide (cuprite) has been found to be p-type (Shuey 1975) with coppervacancies as the principal acceptor defect. P-type materials act as electronsinks and promote oxidation by electron transfer into the electrode (Gerischer1966). Also p-type cuprous oxide is copper deficient (Cu2-xO) at room

temperature, and has a negligible electronic conductivity. The semiconductorproperties of synthetic cuprite are similar to those of naturally occurringcuprite. When the oxide has oxygen vacancies, i.e., metal excess, then it is ann-type semiconductor.

Synthetic cuprite is seldom reported to be n-type (Bertocci 1978).Nevertheless, when the formation of cuprite is made under potentiostaticconditions (current time experiments) it is observed that the transition from p-to n-type semiconductor occurs between -50 to + 50 mv vs the saturatedcalomel electrode, SCE (Taylor and Cannington 1993). At more positive(anodic) potentials than 50 mv vs SCE, the films formed are always n-type,whereas when the potential is more cathodic than -50 mv vs SCE the filmformed is of the p-type. Films formed with this method have been reported tobe adherent to the metal surface. It was concluded that p-type films corrodedat a slower rate because the corrosion current under potentiostatic conditionsdecayed to lower levels and at a faster rate than the corresponding currents forn-type films.

Taylor and Cannington (1993) propose that the mechanisms producing thechange from p- to n-type semiconductivity of cuprite are not determined andmay be related to water composition, copper surface characteristics, and wateruse factors such as stagnation and surface films. These authors havespeculated about the possible role of the p- or n- type behavior as related toLucey's theory of pitting (1967). This theory proposes that when cuprite is

29/10/2001 14

already present it has a dual role –mediating oxidation reactions (copper metalto Cu+ ion of the cuprite) at the metal-film interface and reduction reactions,usually of oxygen, that occur at the film solution interface. Other coupledreactions happen simultaneously. Since reduction reactions occur more readilyon n-type semiconductors, it is thought that the outer surface of the film is ofthis type, thus promoting the reduction of oxygen. In this case the innersurface of the film would be p-type, thus promoting the dissolution of coppermetal to the cuprous ion present in cuprite. If the outer film is of the p-type,however, there can be no reduction here and the corrosion reaction iseffectively blocked.

Millet et. al.(1995) found that under open circuit conditions and in a NaClsolution a duplex Cu2O layer is created and that this is made of twosemiconducting components of different stoichiometries, namely a p-type andan n-type. The latter, located at the metal oxide interface, was correlated withthe shift of the reduction peak towards more negative values, and wasconsecutive to the formation of the oxide layer via dissolution-precipitationmechanism Cu(I) / Cu(II) species from the electrolyte. In acetate or inhibitor–containing mechanism solution, the time to the formation of the duplex layerat the open circuit potential of the electrode was shorter, suggesting arelationship between the protective properties of the film and its p-n structure.

Light has been observed to retard the growth of cuprite and to promote itsdissolution and oxidation (Taylor and Cannington 1993). The photo-electrochemical behavior of cuprite has been known at least since the 1950'sbut there are still contradicting reports relating the observed photo-currentsand the type of semiconductivity. Although copper plumbing tube corrosionoccurs in the dark it is important to understand the photo electrochemicalproperties of these films because they could possibly be used to measuresome fundamental properties of films formed by corrosion.

In this line of thought it is worthy to mention that Wilhelm et al. (1982) havedemonstrated that the changing conductivity mechanism of cuprous and cupricoxides does not depend on the sign of the photo-current. In other words, photocurrents signals opposite in sign to those expected from conductivity typeexist and this is an indication of the properties of the semiconducting oxides,such as the band gap, and not of the changing conductivity mechanisms.Cupric oxide has a narrow band gap, i.e., a narrow energy gap between thevalency and conduction bands, and therefore a high electron-holerecombination rate and no positive photo-currents should be expected.

29/10/2001 15

Moreover, only certain electronic transitions produce electron-hole pairs incuprous oxide and therefore this type of films should be more resistant tooxidation than an n-type oxide, where recombination takes place at a high rate.One of the consequences of band gap width is that for wide gaps, i.e. cuprousoxide, thickness can grow under anodic conditions and the metal, i.e. copper,on which this oxide grows, would be inert provided that the environment isstable.

In conclusion, Taylor and Cannington (1993) hypotheses regarding theprotective character of p-type cuprous oxide seems to be well basedscientifically but unfortunately this hypothesis has not been shownconclusively. Many aspects of the transition from p- to n- typesemiconductivity in cuprous oxides is not well understood.

The formation of cupric oxide and its equilibrium with other cupric solids willbe discussed in section 5.

3- Corrosion Mechanisms of Copper Pipes In Practice

Pitting corrosion can be defined as the localized electrochemical reaction ofchemical species present in the corroding media, with a specific site on thepipe surface. It constitutes a major concern for the manufacturers of coppertubes, due to the inconveniences that the consumer must face.

Pitting can be localized or uniform, depending on the composition and historyof the system involved. Localized corrosion usually leads the pipe failureunless remedial action is taken, whereas uniform corrosion can lead to bluewater or to the release of copper into the solution in a soluble form. Uniformcorrosion hardly ever leads to the pipe rupture because the time involved forthis effect to occur is relatively long (several years) and detected either via thecoloration of the water or its taste, and remedial action taken.

The corrosion taxonomy already mentioned has been recognized since the1950's but there is agreement at present that it requires improvement or evenrevision since many of the practical cases of corrosion found cannot beclassified in any of the existing categories or they may be often assigned tomore than one category.

29/10/2001 16

The following sections briefly review the corrosion mechanisms that affectcopper plumbing installations and result in by-product release that mayexceed drinking water standards.

3.1- Blue Water and Cuprosolvency

This is an appropriate place to define the meaning of the term "bluewater"(Page 1972), since it appears extensively in recent literature. Blue wateris the result of corrosion and its subsequent release of copper in the form ofinsoluble salts or minerals such as brochantite and posnjakite, to the water. Inthe U.S., 4.4% of the faults of copper pipes were reported to the CDA weredetermined to be blue water corrosion between 1988 and 1993. Blue waterstarts manifesting itself at approximately 5 mg/L copper concentration.

Blue water can be produced by at least three corrosion mechanisms: type IIIpitting corrosion, MIC and by uniform corrosion. These mechanisms can alsolead to the release of copper to a solution without the production of blue water,a situation which will be referred to as cuprosolvency (Hongve et.al. 1995).When the concentration of dissolved copper sulfate in water is larger than 5mg/L the water becomes blue(Beguin-Bruhin 1983), and this situation shouldnot be confused with the “blue water” caused by the presence of solids in thewater.

Another reference to blue water in the literature refers to copper ion reactionwith soap (e.g., sodium oleate) and its precipitation as the calcium salt onplumbing fixtures, staining them (Yamauchi et.al 1986). This is one of thereasons for the W.H.O. classification of copper as an aesthetic parameter.[editor: no longer true. Author:........this is true....the WHO has retained copperas an aesthetic parameter....in the reviews made in 1993 and in 1997]

Finally, it is apparent that the nomenclature "green water" and "blue water"are equivalent.

3.2- Corrosion Mechanisms which have an Effect on Copper By-productLiberation

a- Type III pitting corrosion

29/10/2001 17

Type III pitting corrosion occurs in soft, cold water of low conductivity, lowalkalinity and high pH (Edwards Ferguson et.al.1993). The pits are wide andshallow and have two films, the first one consists of cuprous oxide and thesecond one of brochantite and/or malachite. The latter materials enter thewater and contaminate it, sometimes blocking the pipe.

This form of pitting corrosion has been reported in many places throughoutthe world (Johansson 1989). From 1972 to 1983 there were 170 house ownerswho reported faulty pipes in the district of Floda, Sweden (Linder andLindman 1983). It was found that the faults were produced randomly inhouses built in the 1960´s and 1970´s and the water composition and sourcebecame suspect. In fact, the water source had changed from ground water tosurface raw water from the Oxsjo Works in Lerum. At first both sources weremixed with only a small proportion coming from the Oxsjo Works. After atime the proportion of the surface water increased but did not reach 100%. Thenumber of perforated pipes reported per year reached a constant level. Itshould be noted that houses in Lerum did not report corrosion problems and itwas only after both waters began to be mixed that problems arose in Floda.

The characteristics of the water are consistent with those producing type IIIpitting. After mixing the two water supplies, the alkalinity was lowered from

50-60 mg/L to approximately 40 mg/L. When the HCO3- concentration wasraised to 70 mg/L there was a dramatic drop in the damage frequency(Mattsson 1988). The corrosion products found included langite(Cu4(OH)6.SO4.H2O), brochantite (Cu4(OH)6.SO4) and cuprous oxide.Presence of the elements phosphorus, chlorine, sulphur, silicon and aluminiumwas detected with flame spectroscopy. Damage to the pipes occurred in boththe horizontal and vertical sections of the system.

In order to prevent further events of this type of corrosion, it was proposedthat a film of copper hydroxide carbonate be formed on the copper surface(Mattsson 1988, Linder and Lindman 1983). Concurrently the pH should beadjusted to values greater than 7.4 and the bicarbonate to sulfate ratio shouldbe adjusted to values greater than 1.0. Also, the hardness should be raised withtreatment by limestone and carbon dioxide.

These hypotheses were tested at CSIRO (Moss and Potter 1984) in Australiaand form the basis for the Hunter Water Board's Dundog treatment works,which operates and distributes potable water in the Hunter District. The first

29/10/2001 18

report of corrosion in this District occurred in 1976 at the Newcastle Collegeof Advanced Education and involved the release to the water system ofinsoluble blue-green particles (Moss and Potter 1984). It was subsequentlyfound that water from the tap had copper levels between 10 and 40 mg/L.Between 1976 and 1984, over 100 cases were reported in the area, fromprivate dwellings to large buildings. The most severe cases involved theperforation of the pipe.

During the same period a small number of cases were also reported in largebuildings in Sydney, but there was no perforation observed. According toMoss and Potter (1984), the only comparable events of corrosion had occurredin New Zealand (Page et.al. 1974, Potter 1969) in 1967, especially in theAuckland region. It was later noted in another CSIRO report by Taylor andCannington (1993) that corrosion observed in the aforementioned cases wascomparable to type III pitting cases seen in other countries, such as theoutbreak that occurred to 84 of 200 newly built houses by one developer in ahousing estate in California in the 1990's. An odd aspect of the latter case wasthat another 2000 houses, built by other developers, and that used the samewater did not suffer any problem.

In the Hunter Water District, cuprous chloride, cuprous oxide, copper sulfateand a copper silicate with aluminum, possibly derived from flocculants, werefound as corrosion products. The same products were detected in Auckland,N.Z. post mortem tube analysis, except that the silicates did not containaluminum. The mixture of corrosion products were interpreted to bebrochantite (CuSO4[Cu[OH]2]3) and posnjakite (CuSO4[Cu[OH]2]3H2O) inboth cases. However, in Auckland it was predominantly brochantite whereasin the Hunter Water District it was a similar quantity of both compounds.

The following table presents some data about water composition that has beenrelated with type III pitting. It should be observed, however, that copperconcentration is not quoted. Indeed, this is one of the common flaws found inthe literature.

Table 3.1-Water Compositions of Type III Mechanism

Linder1982(2)

Gilbert1966(4)

Page1973-74

(2)

Moss 1984(4)

Taylor 1993(5)

29/10/2001 19

HCO3- (mg/L)

2-8 29.3-30.5 14-30 22-31.7

pH 6.1-8.2 7.6-9.2 7.9-8.6 6.7-7.2 7

SO4(-2) (mg/L)

14-19 27-35 10.5-15 5-42 0.6-6.9

Cl (mg/L) 10-16 13-20 9-13.5 12-50 5.9-17.3Na (mg/L) 7.5 7-27 5-10.8Ca (mg/L) 20 7-17 1.3-6Mg (mg/L) 8.5 2-25 0.9-2.1TDS (mg/L) 68-90 77.5 30-75Conduct (µS/cm) 35-36 67-108 35-140 97-334 40-92Alkalinity(mg/L as CaCO3)

1.6-6.6 24-39 11.5-24.5 18-26 5.1-12.1

Hardness(mg/L as CaCO3)

8.9 30 23-70 11.9-22.4

K (mg/L) 3 1-3 0.7-1.5CO2 (mg/L) 0.2-6

Note: (2),(4),(5) indicate the number of water compositions included by eachauthor.

a.i- Concluding Remarks on type III Pitting

The following conclusions by two Australian researchers, are valid for typethree pitting, but should be considered in light of other work more recentlypublished, especially in regard to the role of bicarbonate (section 5.3).The conclusions of Moss et al.(Moss and Potter 1984) concerning thecorrosion problems described above can be summarized as follows:-The composition of the potable water has marked and sometimes crucialeffect on the nature, intensity and rate of interaction with copper.-The native rate and extent of corrosion varies depending on the type ofcopper tube, but differences appear less significant than those caused bychanging water composition.-Compositions of water that encourage the formation of passivating layers ofcuprous oxide or of basic copper carbonate have the least probability ofcausing copper contamination problems. Bicarbonate alkalinity offers thisadvantage.

29/10/2001 20

-Dissolved copper sulfate, depending on its relative concentration, candominate the copper/water interaction and, by producing non protective basiccopper sulfate, causes active corrosion of the metal.-Sufficient dissolved chloride likewise causes active corrosion of copper byvirtue of the formation of cuprous chloride, which probably flaws protectivefilms of oxides or basic carbonate.-To passivate copper surfaces in cold potable water the pH value may beraised using bicarbonate alkalinity, which counteracts the combined activatingeffects of dissolved sulfate and chloride. At an electrical conductivity of thewater up to approximately 800 µS cm-1 it is estimated that the pH value needsto exceed 8.3 to be certain of copper passivity. This protective pH valuediminishes with the sulfate to chloride ratio, and falls to 7.3 when no dissolvedsulfate is present.

The conclusions of Taylor et al.(1993) confirmed most of the first reportsfindings and added the following:-bicarbonate dosing decreased the likelihood of formation of basic coppersulfates as a major corrosion product and increased the likelihood of formingmalachite.-bicarbonate dosing was also found to promote the formation of more adhesivecuprous oxide films.-it was determined that p-type cuprous oxide favored the protection of copperwhereas n-type cuprous oxide favored the corrosion of copper. It was alsofound that factory production tubes had a thin p-type film which broke downunder certain water conditions.

The most important cases of type III pitting corrosion have been approachedand diagnostic techniques and preventive solutions have emerged in spite ofthe fact that the mechanisms are still poorly understood and that the taxonomyemployed can be severely criticized. The validity of most of these preventivesolutions is still to be demonstrated. As mentioned before, (in section 5.3) analternative hypotheses for the effect of bicarbonate on copper by-productliberation will be discussed.

b- MIC Corrosion

Micro-biologically induced corrosion (MIC) has been known since the 1940's.Bacteria, fungi, algae, protozoa, diatoms and bryozoa can produce or promotemetal corrosion via the creation of a film, known as biofilm, capable of

29/10/2001 21

adhering to the metal surface and of maintaining concentration gradients ofdissolved inorganic and organic salts (O’Connell 1941, Hadley 1948). It isknown that Pseudomonas and Alcaligenes, which are common species intreated waters, are capable of producing exopolymers, usuallyoligosaccharides and polysaccharides. The build up of these polymers helps toconsolidate the biofilm. Thus, it is not surprising to encounter a biofilm-polymer film adhered to the metal surface.

It was only during the mid-80's that MIC caught the attention of scientists withregards to copper. It was reported then that micro-colonies of exopolymerproducing bacteria, when attached to submerged copper surfaces, can formcopper concentration cells with reactive exopolymers (Geesey et.al. 1986). ApH gradient is established within this film and the film layer close to the metalhas the lower pH. The metal corrodes and since the copper concentration is pHdependent, a copper concentration cell is established within the film and thecopper ions are transported through the film to the film/electrolyte phaseboundary where they react with the electrolyte and form copper oxide. It is notunderstood why a pH gradient is established in the first place.

There is strong evidence that MIC corrosion has been the cause of severalepisodes involving the failure of pipes, especially in institutional buildingssuch as hospitals during the 1980's (Geesey, Kalaiyappan et.al. 1994,Mittelman et.al. 1994). Incidents have been reported about MIC corrosion inhospitals in Germany, Scotland, England, Saudi Arabia and Kuwait. There isknowledge about the symptoms, effects and remedial action of thesephenomena but the mechanisms leading to MIC and the specific action thatmicro-organisms may have on corrosion have only been proposed at ahypothetical level.

Why was MIC not investigated before in relation with copper pipe corrosionwhen it was known to have an important effect on steel corrosion and also onother metals much before?. After the role of MIC became apparent in thecases of the aforementioned hospitals, it is becoming a standard technique tolook for the existence of biofilms when corrosion is reported. But how manyof the cases of corrosion reported before 1990 could have been promoted orproduced by MIC and yet they were assigned to other causes?. These arequestions which can find only partial answer and which indeed blur even morethe interpretation of past analysis of corrosion cases.

29/10/2001 22

Two types of MIC have been detected (Geesey, Bremer et.al. 1994), one thatresembles type 3 pitting and a second which resembles both type 1 and type 2pitting (Angel et.al. 1990) which had been described in the literature as type1.5 pitting. Wagner et al.(1993) have described these two types of MICcorrosion differently, assigning the first classification of Geesey, Kalaippayanet al. (1994), to type I like pitting or hemispherical pitting corrosion while thesecond type they have described as "pepper pot" type pitting. This controversyis a further expression of the inadequacy of the taxonomy of corrosionmechanisms on copper. After analyzing the characteristics of both types ofMIC pitting, the conclusion is that in the first type there are characteristics ofboth types III and type I pitting. In order to avoid further confusion, this paperwill adopt the classification of Wagner et al. (1992)

Arens et al. (1995, 1996, 1999) propose that MIC corrosion occurs in thepresence of soft alkaline surface waters but only if there are long stagnationperiods. MIC should be associated also with the presence of high copperconcentrations, and pitting corrosion. Only the third characteristic isrecognized in many cases. MIC is not associated with hot water except whenparts of the hot water systems have been in contact with cold water for longperiods. MIC occurs in horizontal tubes because the high copperconcentrations generated can settle down on the bottom of the tube, wherecorrosion products build up. Conditions here would also be favorable foraeration cells to build up. Arens also proposes that pitting corrosion can beinfluenced by biofilms, and these would act as selective membranes. Lucey’smodel is valid here, but the membrane plays the role of the oxide andcarbonate layer.

Hemispherical Pitting MIC Corrosion was observed in the hot watersystems of two hospitals in South West Scotland during the 1980's. It involvedwater from surface sources, containing high levels of dissolved organiccompounds, and assimilable organic carbon (AOC), low buffering capacity,humic substances and suspended fine particulates including micro-organisms.Particulates tend to form sediments in “dead” sections of the pipe (INCRA1988) and, along with biofilms, protect bacteria from adverse incidentaltemperatures (usually above 55 to 60 °C). No episodes were reportedinvolving the cold water systems. At the time of the research, one of thehospitals had recently cleaned and chlorinated its system.

Severe corrosion that resulted in pipe perforation was randomly distributedamong the several reported between 1982 and 1988.

29/10/2001 23

Cupric and cuprous oxides were found in the corroded tube. While the formertended to be on the pits, the latter was found some distance away,accompanied by cupric oxides and polysaccharides and oligopeptidesbiofilms. Some general attack, like that obtained in type III corrosion, isfound beneath the black cupric oxide and in some pitting areas. Blue corrosionproducts are present in the latter. The tubercules above pits are basic sulfatesand powdery cupric oxides. Also present are basic carbonate tubercules, likethose of type I pitting. The more severe the corrosion, the more developed thebiofilms found. Cuprous chloride was not found in the corrosion products.

This type of corrosion is accompanied sporadically by the random appearanceof blue water.

Pseudonomas, Alcaligenes, Methylobacterium spp, SRB (sulfate reducingbacteria), and fungi, were found among the corrosion products.

After several hours of stagnation overnight, the oxygen content of the waterwas severely, and sometimes completely, depleted. This depletion was thoughtto be indicative of bacterial colonization and abetted the growth of anaerobicbacteria such as SRB.

The water received by both hospitals was found to be of good quality, but theoperating conditions, especially concerning temperature and the lid of thewater reservoirs was unsatisfactory and thought to account for the conditionsthat lead to the creation of MIC.

The temperature was within the ranges where the bacteria could live and grow,i.e., between 10 and 60 C.

A scheme has been proposed for describing the mechanisms producing thistype of MIC corrosion (INCRA 1988) :

1-Soft upland catchment water, such as received by one of the hospitals,containing natural organic substances derived from the soil e.g. humic acid.2-Aerobes such a Pseudomonas may use these organic substances as nutrientsand produce exopolymers which attach themselves to copper. The inhibitoryeffect of copper on bacteria is rendered ineffective by a film of material whichmodifies the physico-chemical characteristics of the surface and whichpresumably could be there from either fabrication or installation. The

29/10/2001 24

production of polysaccharide also helps protect bacteria against the inhibitoryeffect of copper and promotes the growth of the biofilm.3-Bacteria may be joined by fungi or other species in the formation of acomplex biofilm.4-The aerobic members of the biofilm reduce the oxygen concentration in thebiofilm and conditions are rendered appropriate for the colonization of theanaerobic species, such as SRB, in the region of the biofilm closest to themetal.5-SRB produces hydrogen sulfide which forms copper sulfides and stimulatesa corrosion potential (Hamilton 1985, Jacobs 1997).

These mechanisms for aerobic and anaerobic corrosion may occur sequentiallyand/or simultaneously and are at present under laboratory study in order todemonstrate their viability.

“Pepperpot” like MIC corrosion occurred in one hospital in West Germany,one in South West England and one in Saudi Arabia, and involved cold, warmand hot water systems. It is similar to type I pitting (Geesey, Bremer et.al.1994) in that the pits are hemispherical and deep and are covered withcrystalline cuprous oxide and copper chloride. It resembles type 2 pitting inthat the oxide between the pits is mostly cupric oxide. The mounds above thepits are mainly copper sulfate with a deposit of cupric oxide around theperiphery. This type of pitting has been known for some time as type 1.5 andit has been described as “pepperpot” MIC (Wagner et.al.1992). The authorsdescribe the site of corrosion as: "multiple pits under a common crystallinelayer of cuprous oxide and an outer crust of basic copper sulfate. The areacovered by this oxide membrane is greater than for hemispherical pitting. Theperforations in the oxide membrane correspond to the positions of the pitsbeneath. The pits may coalesce to cause more general undermining of theoxide membrane. Pepperpot pit sites may be adjacent or widely separated".

In all cases there was evidence of polysaccharide biofilm and no carbon filmwas found.

The water, with the exception of one case, was: soft, had a pH between 7.3and 9.8, total hardness of between 25 to 73 mg/L (as CaCO3), alkalinitybetween 10 and 40 mg/L (CaCO3), chloride between 6 and 23 mg/L, sulfatebetween 12 and 24 mg/L, was poorly buffered, and in most cases, had a sulfateto carbonate ratio of two (Geesey, Bremer et.al. 1994, Wagner et.al. 1992).

29/10/2001 25

Pitting in the German hospital appeared only two months after it was openedto the public, in 1986, and it involved pipe failure (Paradies et.al. 1990,Wagner et.al. 1992). A number of failures followed during the next yearresulting ultimately in the replacement of approximately one third of thecopper piping by plastic pipes.

It is crucial to note that the manifestation of the problem was the detection ofcorrosion products in the water. Therefore, whereas in the case of the hospitalsin South West Scotland there was no contamination of the water and thecorrosion problem pertained only to the hot water system, in the case of theGerman Hospital, the corrosion affected the quality of the drinking water.

Citric acid was employed as a rinse treatment which seemingly retarded thecorrosion process but did not stop it.

In 1990 a similar corrosion problem was observed in a court house built at thesame time and in the same county as the Hospital, and which had the samewater supply.

In both cases the corrosion produced to the water a release of copper insoluble and insoluble form. A sample taken after 48 hrs of stagnation at thecourt house (after 5 minutes of rinse) was 2.5 mg/L of dissolved copper.

In the case of the hospital, corrosion was found to be high in horizontal,infrequently utilized sections of the pipe and low on the vertical frequentlyutilized sections. This supports the hypothesis that sediments accumulated inthe horizontal, deadleg sections promote the build up of a biofilm (Wagneret.al. 1993). The pitting was hemispherical, but also uniform in the horizontalpipes whereas it was non-uniform in the vertical pipes. The water containedhumic substances and the conditions of corrosion even after stagnation weredetermined to be aerobic.

The corrosion products over the hemispherical pits consisted of adherentcupric oxide under mounds of cupric sulfate with loose powdery deposits ofcuprous oxide inside, on top and around the perimeter of the pit (Geesey,Bremer et.al. 1994). More than 10 micro-organisms were found, along withcomplexing agents such as metabolic pyruvate and lactate. Three species wereconsistently identified: two strains of Pseudomonas paucimobilis andPseudomonas solanacearum. P. solanacearum was found to be capable ofnitrogen respiration and all exhibited copper tolerance. The high number of

29/10/2001 26

bacteria was not always correlated with pitting. A biofilm was found thatcontained polysaccharides, oligopeptides and n-acetylated derivatives ofglucose, mannose and galactose.

The characteristics of the incidents in the two hospitals in the South West ofEngland were very similar to that of the German and to the Saudi hospitalcases. The quality of water, species, biofilm, type of pitting, sections of thesystem where faults occurred, and correlation between bacterial density andcorrosion, were similar.

It is interesting to note that the water for the two hospitals in England wasobtained from different rivers and also that a third hospital in the same regiondid not have pipe corrosion. Some of the differences between the case of thetwo hospitals in the SW of England and the German case was that no SRBwas found in the former and that the corrosion problems affected only the hotwater system, which indeed was not very hot (30 to 40 ºC).

Another case of MIC corrosion occurred in a hospital in Kuwait, with softwater supply which included sulfate and chloride. The description of the pitscorresponded to type 1.5 but a biofilm was not looked for.

b.i- Preventive Measures

Many preventive measures can be taken in order to avoid MIC:

- pipe system design should avoid dead sections, long periods of stagnation,and, if possible, long horizontal sections where particles can settle.

- use of filters down to 0.2 µm in order to eliminate fine particles; use ofactivated carbon filters in order to eliminate organic matter; hardening of thewater and; ultra violet, (UV) irradiation. These measures have been employedto a limited extent and there has been no evaluation regarding theireffectiveness, except UV which has proven ineffective because it sterilizes thewater but not the pipes.

- maintain water temperature outside the range 25C to 45C because theprobability of corrosion increases here. Above 60C bacterial growth isinhibited.

29/10/2001 27

- water tanks should be properly covered and cleaned periodically.

- the use of periodic chlorine disinfection of the water systems.

b.ii - MIC Remedial Measures

Wagner et al.(1993) summarize these measures as:

- clean the inner surface of the installations with citric acid and/or sulfamicacid and replace the damaged section.

-pretreat the installation with hard water to form a protective layer.

-insulate hot and cold water systems.

-increase the water flow by adding pumps.

-reduce suspended solids and total organic carbon in source water by addingfilters.

-make changes in more than one parameter in order to minimize thereappearance of pitting.

b.iii - Concluding Remarks on MIC

The cases of MIC reviewed here constitute most of the known cases. Morethan half of these cases have not affected potable water systems, but hot watersystems. Nevertheless, it is relevant to consider the totality of the cases ofMIC corrosion because the hot water systems faults could occur in potablewater systems in the future.

An aspect that is still under intense discussion is how to prove that thecorrosion mechanism in each case is microbially induced. The presence ofmicroorganisms in the water, or of biofilms on the tube, is no proof that MICis the causative mechanism of observed corrosion. Mittelman has shown thatbiofilms and bacterial populations exist throughout water distribution systems,and in most cases blue water problems are not associated with their presence

29/10/2001 28

(Mittelman 1994). Even a positive response (i.e. the end of corrosion) with theaddition of chlorine is no proof of the mechanism (Edwards 1990).

Nevertheless, effective prevention, diagnosis and remediation procedureshave been developed in the 1990’s, especially with regard to large potablewater systems, and this knowledge should be made available to institutionsthat deal with drinking water distribution and regulations.

c- Jointing Corrosion

Copper pipes can be jointed by mechanical fittings, by welding, by brazing orby soft soldering (Mattsson 1990). Fluxes are used in brazing and softsoldering.

Corrosion is produced when any of the materials that constitute the joint comeinto contact with water and with copper. The result is usually to accelerate thecopper pipe corrosion at the joint due to chemical attack by the jointingmaterials or by galvanic corrosion. The result of this corrosion is to releaseboth copper and the jointing material to the solution.

Jointing corrosion has been extensively studied and materials and methods areavailable in many countries in order to prevent this type of corrosion.

c.i-Welding

Welding is not commonly used for jointing.

c.ii- Brass, Brazing and Soft Soldering

Brass is often used for valves and fittings. Brass may corrode throughdezincification since zinc is selectively dissolved leaving behind a spongycopper structure (Mattsson 1990, Nielsen 1983). This may lead to the blockingof the pipe. Brass with more than 85% by weight of copper has a goodresistance to dezincification. (Lytle and Schock 1997a) concluded that theamount of composite metal leached from brass copper alloys correspondedwell with alloy composition. Baukloh et.al. (1989), reported that during

29/10/2001 29

brazing films of copper oxides formed on the internal surface of pipes, mayhave a morphology that enhances pitting type I.

Alfa brass with lower contents of copper can be made corrosion resistant bythe addition of inhibitors such as arsenic (0.02 to 0.04% by weight), antimonyor phosphorus. Beta brasses cannot be made resistant to corrosion by theaddition of these inhibitors. Since the early 70's in Sweden, brass for fittingsmust be corrosion resistant and they must pass a test according to standardsISO 6509 and SS 11 71 10. Many types of new corrosion resistant brasseshave been developed since then in order to comply with the standards(Mattsson 1990).

For brazing, silver-phosphor-copper solders are used in Sweden which containat least 2% silver. This solder produces small or acceptable levels of corrosionat the joint with the copper pipe unless the water is corrosive. It is notrecommended though, when nickel containing alloys are present. Cadmiumcontent in solders is banned in most countries. Lead in solders is banned inSweden. Regarding soft soldering, little is done in Sweden. Tin-silver soldersare accepted in this country, with at least 3% by weight of silver.

The Lead-Copper Rule in the U.S. prohibits the use of lead-containing solders.As a result, the content of lead in soft solders has been reduced in the U.S.from 50% in 1986 to 0.2% in 1991. At present the composition of solders isregulated in the standards ASTM B32-Solder Metal (Cohen 1994). Acceptablesolders presently includes 95:5 tin-antimony, several tin-silvers and someproprietary alloys.

c.iii- Mechanical Corrosion at Joints

Joints with irregular, rough edges and surfaces, and exposed solder promotecorrosion and should therefore be avoided.

Streamlined shapes should be used in order to avoid turbulence which leads toerosion corrosion (Mattsson 1990). For instance, lap joints should be used,edges and beads of solder should be avoided. Capillary joints arerecommended. The use of “ASTM Standard Practice B828 - Making CapillaryJoints by Soldering of Copper and Copper Alloy Tube and Fittings” shouldprevent many copper tube corrosion problems, including flux corrosion anderosion corrosion (Cohen 1994).

29/10/2001 30

c.iv- Valves, Fittings and Meters

Valves, fittings and meters can also lead to corrosion due to soldering orgalvanic corrosion, depending on the methods employed for jointing.

In the U.S., companion wrought fittings are produced to ASME B16.22 -Wrought Copper and Copper Alloy Solder Joint Pressure Fittings (Cohen1994). In the case of valves and water meters there are several standardsemployed:-ASTM B62 (Alloy C83600) - Composition Bronze or Ounce Metal Castings.-ASTM B584 (Alloy C84400) - Copper Alloy Sand Castings for GeneralApplications.-ASTM B763 (Alloy C84400) - Copper Alloy Sand Castings for ValveApplication.-AWWA C700 - Cold Water Meters - Displacement Type, Bronze Main Case.-AWWA C800 - Underground Service Line Valves and Fittings.

The correct use of materials and methods can prevent corrosion at the joints.

c.v- Fluxes

Flux residues at the joints usually induce corrosion due to chemical reactionswith copper or other mechanisms. Pitting is produced due to factors such asthe amount of flux applied and the chemical aggressiveness of the flux. Pittingis not restricted to the vicinity of the joint but can be a considerable distanceaway. Generally pitting attack occurs at the periphery of petrolatum basesoldering flux or parallel to the longitudinal axes of the tubes/fittings (Cohen1994).

In the U.S. it is reported that after testing two types of fluxes for 800 days, oneof them produced 34 failures by pitting while the other produced seven(Lyman et.al. 1982). Standard Specification ASTM B813 - Liquid and PasteFluxes for Soldering Applications of Copper and Copper Alloy Tube, definesthe types of fluxes which are required for pitting prevention (Cohen et.al.1995).

Soldering flux induced pitting can be prevented by using industry standardmaterials and by appropriate workmanship during installation.

29/10/2001 31

3.3- Corrosion Mechanisms that May Have a Slight Effect on Copper By-Product Liberation

All of the corrosion mechanisms listed in this category are well known butwith proper design of the piping system and knowledge of the source water itshould be possible to avoid initiation of corrosion by these mechanisms.

a- Type I Pitting Corrosion

Type I pitting, the most common of all, was reported first by Campbell (1950)and a general theory for pit formation was proposed by Lucey (1967). Pittingcorrosion was further elucidated by Smith and Francis (1990), Shalaby, et al.(1989), Holm et.al. (1982), and Riedl et.al. (1989). It occurs in cold (roomtemperature) hard waters, with high conductivity, high alkalinity, high sulfateconcentration, low total organic carbon and micro-organisms. Thesecharacteristics are usually associated with well waters. The pits are spherical(which is an indication of diffusion control of corrosion reactants to the coppersurface). They are covered by several films: the original one is a cuprousoxide film (Campbell 1979), the second to be formed is cuprous chloride andthe third is composed of copper salts, most commonly malachite (Edwards,Schock et.al. 1996). In the U.S.A., copper chloride has rarely been found inthis type of pitting, suggesting that the cuprous chloride film could be absentin some cases.

Lucey (1967) is credited with the theory that pits propagate when oxygen isreduced at the cuprous oxide film which is there from the origin (and acts as acathode on its exterior surface). Simultaneously, the anodic dissolution ofcopper occurs on the inner surface of this film. Thereafter, the dissolvedcopper, usually present as copper chloride, exceeds the solubility product andcuprous chloride precipitates underneath the original oxide film filling the pitcavity. When the local concentration of chloride begins to fall, other coppersalts begin to precipitate. The oxide film prevents copper corrosion productsfrom entering the solution. The pipe eventually ruptures after a few monthsand the problem is detected.

This theory has gained strength with respect to the more classical theory ofcorrosion which states that the oxidation and reduction processes take place onthe copper surface.

29/10/2001 32

Type I pitting was the most common type of pitting corrosion found duringthe last few decades. In the U.S. it accounted for 18% of the total number offaults reported between 1988 and 1993 (CDA 1994). It is the best knownmechanism and has led to the prevention of many of the potential cases wherethis could occur.

b- Type II Pitting Corrosion

Type II pitting was first reported in Sweden (Mattsson et.al. 1968). It occursin soft waters at temperatures above 60°C, when pH is below 7 and there is alow bicarbonate to sulfate ratio. The pits are small in cross section. The maincorrosion products are hard crystalline cuprous oxide capped by mounds ofgreenish black cuprous oxide and copper sulfate(Cruse et.al. 1988). Inbetween the pits, beneath a thin layer of silt, there are both oxides, cuprousand cupric. When the water contains manganese, the pits are reported to belarger and manganese dioxide is added to the corrosion products. The presenceof aluminum has also been reported to induce this type of corrosion (Tunturiet.al. 1968).

Rupture of the pipe typically occurs after some 8 years of use (Mattsson at.al.1968). Hot water pitting accounts for 5% of the copper pipe faults reported inthe USA between 1988 and 1993, and it has also been reported to occur inGermany, the U.K. and Canada.

c- Non Classified Pitting Corrosion

Cohen et.al. (1987) carried out experiments that simulated cold water pitting(11.1 C) occurred in 25 houses in Ohio in 1978. The raw water had pH 7.1,HCO3 298 mg/L, Cl- 28 mg/L, sulfides 63 µg/L (as S), sulfate 362 mg/L, Mg295.7 as CaCO3 total solids 911 mg/L, Ca 362 as CaCO3, free CO2 54 mg/L.Aeration and filtering was applied at the treatment plant, in order to eliminatehydrogen sulfide and iron. Addition of sodium zeolite softened the water tolevels of 250 to 300 mg/L as CaCO3. Pitting was eliminated by raising pH to8.1 or more, by adding soda ash.

Duthill et al.(1996) studied the influence of sulfate and chloride ion containingborate-buffered solutions on pitting corrosion of low alloyed coppercontaining 216 ppm Sn and 103 ppm P, and found that each of these speciesinduce copper pitting. The borate solution was 0.01 M H3BO3 + 0.01 M

29/10/2001 33

Na2B2O7, pH 9. The equations for the dependence of pitting germination orgeneration rate, λο , on the concentrations of sulfate and chloride ions are:

λο = β ([SO42-] – [SO4

2-]c) where β = 1.92 s-1 cm-2 l mol-1 (6)

λο = α ([Cl-] – [Cl-]c) where α = 0.137 s-1 cm-2 l mol-1 (7)

where [SO42-]c and [Cl-]c are the critical concentrations of both ions below

which pitting does not occur, and is equal to 0.004 M and 0.07 M respectively.

With respect to the synergistic effect of chloride and sulfate on pitting, for aconstant sulfate concentration the variation of λο with chloride content isseparated in three regions: region I for low chloride content λο increases withchloride content, region II for medium chloride content λο decreases withchloride content, and in region III, for high chloride contents, λο increaseswith chloride content.

d- Erosion Corrosion

Erosion corrosion is produced primarily by high velocity running waterimpinging on the pipe material. It is easily recognized by the characteristicroughening of the pipe interior (Cruse et.al. 1988). Entrained gases and solidsincrease the rate of corrosion and reduce the velocity threshold at whichcorrosion is initiated.

Water impinging on pipes at high velocity impede the build-up of a protectivefilm. When the protective film is broken, assuming that there is one in the firstplace, a small anodic surface is formed which is surrounded by a largecathodic surface. Corrosion is rapid but the copper ions formed are sweptaway and thus the concentration of copper in the water does not build up. Theconstant water impingement on the active electrodes depolarizes them and thecorrosion rate remains high.

This type of corrosion is usually the consequence of faulty design of the watersystem. It often occurs with hot water lines, where water is extra pressurized.A recommended maximum velocity of 0.5 m/sec should be designed for hotwater systems. In the case of cold water, the velocity is recommended to belower than 2 m/sec.

29/10/2001 34

Erosion corrosion accounted for 23.5% of the faults detected in Japan bySumitomo Light Metal Industries (Sumitomo 1994) in 1993 and for 17.5% ofthe faults accounted for by the CDA in the USA between 1988 and 1993.

e- Cavitation

This type of corrosion is produced in areas located close to pump impellers,where the action of the impeller creates regions of low pressure. Oxygen andnitrogen low pressure bubbles form in aerated waters and when these bubblescollapse on the metal surface pressures as high as 60,000 psi are produced(Cruse et.al.1988) leading to rapid corrosion.

f- Soil Corrosion

When bedding and backfill materials surround a metal pipe, differences in pHor/and soluble salts presence, can produce electrochemical concentration cellsalong the pipe length. Areas in contact with higher soluble salt concentrationshould become cathodic and the anodic area, where concentration of solublesalts is lower, should corrode.

Another case where soil corrosion can occur is when the oxygen concentrationof the soil changes along the length of the pipe. The area along the highestoxygen concentration becomes a cathode.

This type of corrosion is recognizable because some sections of the length ofthe pipe corrode while others, usually with incrustations of calcium carbonate,are free from attack(Cruse et.al.1988).

Avoidance of copper piping soil corrosion is one of good construction practiceand has been reviewed by (Cruse et.al.1988).

g- Corrosion in Concrete Slabs

The origin of this type of corrosion is very similar to that of soil corrosion andis produced by different concentrations or conditions around the pipe bedding.Freshly poured concrete close to a copper pipe which is not well encased canproduce corrosion on the external surface of the pipe with the formation ofcorrosion products such as crysocolla (CuSiO3.2H2O).

29/10/2001 35

Again, the proper design of the piping system should prevent this type ofcorrosion.

h- Stray Current Corrosion

Stray current corrosion occurs when electrical systems are grounded to thepiping system and when the alternating current is not symmetric in its positiveand negative cycles over long periods of time. Specifically, when the phaselag of the AC current is as high as 2 percent it has been considered to be thecause of external corrosion of the pipe (Cruse et.al.1988).

When direct current sources are earthed to the piping system, corrosion is tobe expected at the point where the current leaves the pipe.

j- Stress Corrosion Cracking

Stress corrosion cracking occurs by the simultaneous presence of tensile stressand a specific corrosive medium (Fontana at.al. 1978). In the case of copperpipes this medium can be produced by the presence of ammonia vapors andsolutions, amines and water or water vapor in the vicinity of the pipes. Tensilestress can occur due to a variety of causes including intergranular stress in thepipe material, stress due to the installation of the pipe, etc.

During stress corrosion most of the pipe is free from attack while a section ofthe pipe should exhibit propagating cracks on the exterior surface.

There is much scientific discussion regarding the mechanisms of stresscorrosion, clearly demonstrating the benefits of a well designed piping systemalong with well fabricated pipe in preventing this type of corrosion.

Between 1988 and 1993 the total number of cases of stress corrosion reportedin the USA with respect to the total number of corrosion cases reported, were2.5% (CDA, USA, 1994). In Germany such cases have disappeared almostcompletely (Cruse et.al.1988), while in Japan they are not reported as such(Sumitomo 1994).

k- Galvanic Corrosion

29/10/2001 36

Copper is a noble material and acts as a cathode when in couple with otherless noble metals such as iron, zinc, nickel, lead, etc. Most minerals also haveless noble (more negative) electrochemical potentials than copper and tend tocorrode rather than copper. Copper is less noble than metals such as silver,gold and platinum, rarely present in piping systems.

Galvanic corrosion is very rare and seldom reported. In the USA, only onecase (0.04% of the total of reported cases during this period) was reportedduring the last 6 years.

l- Thermogalvanic Corrosion

This type of corrosion should be expected for hot water pipes which rununderground and where severe temperature differences build up in shortsections of the pipe.

Thermally produced electrochemical potential differences should be relativelysmall and thus induce very slow corrosion processes. Another effect oftemperature is the evaporation of water and the resulting salting out producedin the vicinity of the pipe, which induces concentration cells along the lengthof the pipe, similar to that observed in soil corrosion.

4 - Corrosion and the Manufacturing of Tubes

Copper pipes for water distribution can be broadly classified in two types:first, soft annealed or flexible tubes, used primarily for the connection of tapsand valves where pipes are not visible. And second, hard drawn tube, usedmainly in visible and underground sections of the water distribution systems.In Germany, 85% of the copper plumbing tubes are hard drawn.[editor: what about “half-hard” tube.......author..I have no information aboutuse of hard hard tube in drinking water systems.]

4.1 - Carbonaceous Films Formed during Manufacturing

A controversy has developed during the last 40 years over the effect of acarbonaceous film on the initiation of cold water pitting corrosion (Campbell1950). This film is formed on the surface of the copper pipes duringmanufacturing when residues of the die lubricant are cracked to carbon at hightemperature. This process occurs when annealing flexible tubes, but can also

29/10/2001 37

occur due to high temperature produced by friction between the tube and thedie in the case of hard drawn tubes (Moss and Potter 1984).

Campbell proposed that this carbon film would be noble with respect tocopper, thus creating a galvanic cell where copper is anodic and the film iscathodic. Later, several researchers proposed that the effect of carbonaceousfilms was to raise the potential beyond the critical point, above which coppercorrosion starts. This potential is approximately 100 mV versus the calomelelectrode. Any water composition leading to potentials above 100 mV wouldproduce pitting, and above 170 mV the pitting would be severe (Cornwellet.al. 1973, Lucey 1982).

Callot et.al. (1978) identified an evenly distributed but not continuous carbonfilm across the thickness of a cuprite film which was adhered to the coppersurface. Cornwell et.al.(1976) found that certain types of water compositions,together with low content of natural organic matter, favored pitting type Iprovided that there was a more than critical amount of carbon residue in thebore.

Five major European copper water tube manufacturers employ a proprietaryprocess for producing a carbon free tube bore and its product is marketedunder the registered trademark SANCO, whose first patent was created in1982. The SANCO patents have a validity of 18 years and consist of the insideoxidation of the copper tubes during annealing, in the case of flexible tubes,and in a special cleaning technique in the case of the hard drawn tubes.

Additionally, at least two other major European manufacturers use a blastingcleaning technique to remove the carbonaceous films.

No carbon lubricants are reported to be used in the US in the manufacture ofsoft copper tubes, thus no carbon film should be expected.[editor:delete?......author...perhaps it is worth indicating this...I leave it to theeditor]

European manufacturers of copper tube claim that since the creation of theSANCO process in 1982, most, if not all, type I pitting is caused by thepresence of carbon films (Edwards 1995), and therefore this type of pitting hasbeen eradicated from Europe.

29/10/2001 38

Until recently it was thought that the carbon free tube was the main factor intube manufacturing that contributed to the prevention of type I pitting. Taylorand Cannington (1993) have found, nevertheless, that there is a cuprous oxidefilm which is formed on the pipes surface during the SANCO manufacturingprocess which involves annealing. This film is p-type and thus constitutes acorrosion protective film.

It is important, to examine this debate within the overall context of copperpipe corrosion.

While no one disagrees that carbon films induce corrosion of copper tubes;researchers from Japan, Australia, New Zealand and the U.S. have proven thatcold water pitting (type I) can and does occur in the complete absence ofcarbon films (Edwards 1995).

Moss and Potter (1984) conclude that water composition has a major effect oncopper corrosion whereas the variations in tube manufacturing techniques areof less significance to corrosion. Other authors address only the problem ofwater composition and of design and installation procedures, thus implicitlyacknowledging that these are the principal factors in copper corrosion.

An aggressive water composition can destroy a protective film created duringmanufacturing and can lead to the creation of a non protective film, whereasother types of water contribute to either reinforcing this protective film or tocreate it in the first place. Design and installation procedures can also wipe outwhatever film has been created during manufacturing.

Not only does the action of tube design and installation and of watercomposition, come later in the pipe's life cycle than manufacturing methods,but the great variety of water sources and of people involved in design andinstallation of pipes, make much more difficult to control these variables thanto control the manufacturing method, which is performed by a handful ofcompanies throughout the world.

5- Copper By-product Liberation to Drinking Water

5.1- Introduction

29/10/2001 39

Corrosion prevention has been a main aim of the copper plumbing tubemanufacturing companies throughout the world for many years. Only duringthe 1990’s this emphasis changed to prevention of copper by-productliberation, due to the emerging health based regulations for copper in drinkingwater adopted by the U.S. Environmental Protection Agency, USEPA(USEPA 1985, 1991a, 1994, Wyllie 1957), the World Health Organization,WHO (WHO 1993,1996,1998, Fitzgerald 1995, IPCS 1999), and theEuropean Union (EU 1998).

Present concern about high levels of copper contained in drinking water isrelated to its potential acute gastrointestinal effects. These include nausea,followed sometimes by vomiting, and diarrhoea. Low levels of copper indrinking water have also been under scrutiny from the perspective of thepotential deficiency of an essential element (IPCS 1999).

For many years the world wide practice, based on the absence ofepidemiological data to relate any health effects of copper, was to includecopper in the List of Substances and Parameters in drinking water that maygive rise to complaints from consumers at a value of 1 mg/L (WHO 1993), alevel which was defined as a reference value to avoid staining of sanitary wareand complaints of the bitter taste of the water by consumers but with no healthsignificance.

This changed in 1991, when the U.S. Environmental Protection Agency (EPA)created the Lead/Copper Rule (USEPA 1991b), which established a MaximumContaminant Level Goal, MCLG, of 1.3 mg/L, on a 90th percentile basis, forcopper in drinking water, measured at a tap. The action level, i.e., thetreatment of water, is triggered when 10% or more of the samples measured atthe faucet have a concentration which exceeds the MCLG. Thus, the actionlevel is independent of the degree of exceedence of a particular sample.

This standard was derived from a retrospective case study on a group ofnurses who suffered acute gastrointestinal disturbances after consuming analcoholic beverage with presumed high copper concentration (Wyllie 1957),and has been severely criticized (Fitzgerald 1995).

In 1993, the WHO included copper in its list of Health Significance due tochronic effects (WHO 1993), and in 1997 it changed this to acutegastrointestinal effects (WHO 1998). The guideline level established was 2.0mg/L, measured as an average throughout one day, at a tap.

29/10/2001 40

In 1998 the European Union adopted the WHO guideline (EU 1998) and is inthe process of defining whether the 2 mg/L should be measured as a mean oras a maximum. Countries in the European Union are in the process ofchanging their internal regulations in order to adapt them to the EU.

This section defines maximum, stagnant, mean, minimum, and running watercopper concentrations in drinking water. It analyses data about copperconcentration in drinking water and discusses some implications on drinkingwater regulations. Finally, it reviews theoretical tools to predict theconcentration of copper in drinking water, it discusses the effects of watercomposition on copper concentration and considers methods to reduce copperby product liberation to water.

5.2- Copper Concentration in Drinking Water

The maximum potential copper concentration that can be generated in acopper pipe containing drinking water, depends on, the time that the water hasbeen standing in the pipe, also known as stagnation time, on the watercomposition (i.e., pH, alkalinity, hardness, the concentrations of anions suchas sulfate and chloride, and other parameters), on the water temperature, onthe pipe diameter, and on the age of the pipe.

The stagnant copper concentration may be equal to the maximum potentialcopper concentration provided that, the stagnation time is the same as thechemical equilibration time required to reach the maximum potentialconcentration, and that the diameter of the pipe is not greater than ½ inch. Inpipes of larger diameter the time required for copper to diffuse from the pipewall to the center of the pipe, and thus produce a homogeneous concentrationacross the pipe section, is much greater than 8 hours (Van den Hoven et.al.1995). Therefore, under these conditions, the copper stagnant concentration islower than the maximum potential concentration. Health based copperregulations establish a period of stagnation before sampling, and this can varybetween 6 and 12 hours (USEPA 1991ª, HM Germany 1990, ME Denmark1998).

The minimum concentration of copper or mains copper concentration at awater tap usually occurs when the water has been running for a few minutesand can be considered fresh from the mains. The concentration in waterdistributed through the mains by water utility companies is low and usually

29/10/2001 41

does not constitute a regulatory concern (AWWA 1996, Maggi 1999). Insome cases, nevertheless, the running water copper concentration does notdecrease fully to the mains concentration due to the design and/or length ofthe copper pipes.

Several methods have been developed and applied for measuring stagnant, andrunning water copper concentrations in drinking water (USEPA 1991b, HMGermany 1990, ME 1998). The European Union has developed severalmethods to measure the mean lead concentration in drinking water throughouta day (Kiwa 1998), problem which has several similarities with that of copperin drinking water.

Figure 5.2.1 illustrates copper concentration profiles with time in three housesin Santiago, Chile. The first sample in every house was taken after 8 hour ofwater stagnation. Subsequent samples were taken every one hour. The lastsample was taken after letting water run for a few minutes. The copperconcentration varies according to the use of water in every house. Peakcopper concentrations occurs usually first time in the morning and it does notcoincide always with human exposure to peak copper concentration due todifferent patterns of water ingestion (Lagos et.al.1990). For instance, in houseone of Figure 5.2.1, the peak copper concentration (135 µg/L) occurred at 8AM in the kitchen tap, but this water may have been used for washing,preparation of foods, or may have been flushed before ingesting it. Lagoset.al. (1999) developed a method to estimate peak and average humanexposure to copper in drinking water, and showed that only 4.5% of thepopulation of Santiago, Chile, ingested water at the overnight stagnantconcentration.

Figure 5.2.1: Measurements of copper in drinking water in three housesin Santiago, Chile, 1998.

0

20

40

60

80

100

120

140

160

7:00 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 17:00 18:00

Hour

Cop

per

conc

entr

atio

n (u

g/l)

House 1 House 2 House 3

29/10/2001 42

The data of Figure 5.2.1 allows to estimate the mean copper concentrationthroughout the day, variable that is important for the estimation of compliancewith the WHO guideline for copper in drinking water.

Dodrill and Edwards (1995), and Lagos (1997), examined data originallycompiled during a 1991 American Water Works Association, AWWA, survey(AWWA 1996). The data base analyzed contains the population data of 284utilities, the number of samples taken by each utility, the 90th percentile, thelowest and the highest copper concentrations measured by each utility, the pH,the alkalinity and also the corrosion treatment status of the waters at the timeof monitoring. A fraction of these water utilities (118) had carried out by 1991water treatment (pH adjustment and addition of phosphate inhibitors) in orderto reduce iron corrosion. This reduced the corrosion potential of these waterstowards copper as well.

The 284 utilities supply water to a population of 105.8 million people. A totalof 27407 samples were taken by these utilities in order to comply with theUSEPA regulation about copper and lead in drinking water. Figure 5.2.2shows the copper concentration distributions for the 284 utilities: highest, 90thpercentile, mean and lowest.

Figure 5.2.2: Copper concentration distribution for 284 Water Utilities inthe US, 1991. Data modified from Dodrill and Edwards (1995)

0.001

0.01

0.1

1

10

0.00 20.00 40.00 60.00 80.00 100.00

% of Utilities Below Indicated Concentration

Cop

per

Con

cent

rati

on, m

g/L

lowest

highest

90th%ile

The mean concentration shown in Figure 5.2.2. was estimated by Lagos(1997) and has not been validated for this data. It can be observed that 2.3% of

29/10/2001 43

the population sampled exceeded the MCLG of 1.3 mg/L on a 90th percentilebasis.

Figure 5.2.3 shows the cumulative frequency distribution of first draw watercopper concentration, or stagnant copper concentration, for the US data shownin Figure 5.2.2, of a region in Germany comprising 2577 samples, of SantiagoChile, comprising a population of 5.2 million people, and of the BrumslebyHousing complex, in Copenhagen, Denmark (Force Institute 1999). Thestagnation periods for these distributions were 6 hours for the USA, 8 hoursfor the Brumsleby housing complex, 12 hours for the German data, andvariable stagnation time with a mean of 7.85 hours for Santiago.

The distributions for the USA, Germany and Chile were shown to be log-normal by chi square tests. The Brumsleby distribution is based on only tenpoints and is insufficient to derive detailed statistical information from it. Itsinterest lies on the fact that the variability of water composition in this housingcomplex is small and therefore the copper concentration distribution is flat,with little dispersion (see Table 5.2.1), and is not log normal. The dispersionof data from the USA, Chilean, and German data is equivalent.

Figure 5.2.3: Stagnant copper concentration distributions for the USA(AWWA 1996), Germany, Chile (Lagos et.al. 1999), a Housing Complex

in Copenhagen, Denmark (Force Institute 1999). German data wasmodeled from Meyer (1996).

0.1

1

10

100

1000

10000

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

percentile

Sta

gnan

t Cop

per

Con

cent

ratio

n, u

g/L

Cu max (Copenhagen)

Cu First Draw (USA, AWWA)

Cu Firs Draw (Santiago, Chile)

Cu First Draw (Germany)

It is evident from Figure 5.2.3 and Table 5.2.1 that US waters are moreaggressive with respect to copper dissolution than the German waters shown,and than Santiago’s water.

29/10/2001 44

Table 5.2.1: mean and standard deviation for the four first draw copperconcentration distributions shown in Figure 5.3.1

First DrawCopperConcentration

USA Germany Santiago, Chile Brumsleby,Copenhagen

mean (µµµµg/L) 1100 222.6 123.9 2527.3StandardDeviation

1500 335.9 181.1 369.0

StandardDeviation/Mean

1.36 1.51 1.46 0.15

Figure 5.2.4 shows one utility in the USA that supplies water to a populationof 175 thousand people and its 90th percentile concentration before treatmentwas 2.26 mg/L, with a highest concentration of 4.56 mg/L and a minimumclose to zero. The pH before corrosion control was 7.1 and the alkalinity was268 mg/L as CaCO3 . To meet the action limit the utility raised pH to 7.4 andlowered alkalinity to 98 mg/L. Two subsequent monitoring events conductedafter this water quality change was implemented demonstrated that the 90thpercentile concentration dropped to 0.31 mg/L, and the highest concentrationto 0.6 mg/L. Release of copper by-products was mitigated at all houses.

Figure 5.2.4: Stagnant copper concentration monitoring for one waterutility in Colorado, USA, before and after corrosion treatment. From

Dodrill and Edwards (1995), and Lagos (1997)

0.00.51.01.52.02.53.03.54.04.55.0

0 10 20 30 40 50 60 70 80 90 100

Cumulative frequency distribution

Stag

nant

Cop

per

Con

cent

rati

on (

mg/

l)

before

after

5.3- Who is Responsible for Copper in Drinking Water?

29/10/2001 45

The drinking water quality delivered at the tap of copper distribution systems,can be the responsibility, of the water authority (U.S., Netherlands), of theinstaller - who takes the responsibility for recommending or notrecommending copper as plumbing material - (Germany), or of the consumer.The results of the application of these regulatory philosophies are dramaticallydifferent for the copper tube manufacturers and for the copper industry ingeneral.

When the responsibility lies with the water authority, the result is that allpotable water should be suitable for use with copper plumbing tubeinstallations after water treatment made by water utilities. Copper by-productliberation is reduced at least ten fold in those facilities that apply corrosioninhibitors (Lagos 1997). The concentration of copper in drinking water aftercorrosion treatment is well below the USEPA Lead Copper Rule level. TheLead Copper Rule approach to corrosion prevention will be discussed insection 5.6.

When the responsibility lies with the installer, as in Germany (HM Germany1990), then installation of copper plumbing tubes does not take place when thewater is not suitable. Reportedly, this could be approximately 5% ofGermanys water. Copper pipe compatible waters are considered to be thosewith pH between 6.5 and 9.5, if the alkaline capacity Kb 8.2 is not higher than 1mol/m3 (approximately 44 mg CO2/L). Installations that were put in placebefore this regulation (HM Germany 1990) in 1990 are classified according totheir age. For “new pipes”, copper by-product liberation should be reduced tolevels that are below the German standard (3 mg/L measured as a maximum).This is acknowledged in the German Water Ordinance’s Annex 7 which states“The guide level applies after 12 hrs stagnation. Within two years after theinstallation of copper pipework (old pipes) the guide level shall apply withouttaking stagnation into account”. Also for those private wells with “corrosive”waters that use copper tubes, different approaches may exist, such asrecommending using bottled water for more sensitive populations, or lettingwater run for a few minutes after periods of stagnation.

The German regulation to be adopted after the 1998 European Union WaterDirective will still place the responsibility with the installer, but aspects suchas measurement protocol for compliance (with the 2 mg/L EU regulation), andthe type of waters to be banned for copper pipe installation are not known yet.

29/10/2001 46

When responsibility lies with the consumer, then the Government may or maynot enter into the homes and measure the concentration of copper. Therefore,the most suitable approach to insure that the standard is not surpassed may beto ban the installation of copper plumbing tubes.

5.4- Diagnosis of Water Quality and Its Potential to Liberate Copper By-Products

Several models for predicting copper corrosion and/or liberation of copper todrinking water have elaborated.

Lucey´s empirical nomogram (Lucey 1972a) for English water forecastspitting in the U.K. and Europe but is not valid for U.S. waters which are moremineralized and have more sulfate and chloride than European waters (Cruseet.al. 1988). Other assessments of pitting characteristics of waters differ withLucey's and between themselves (Cohen et.al., 1984). Lucey's nomogram isable to predict the pitting type I propensity of waters when sulfate, chloride,pH, oxygen, sodium and nitrate concentrations are known. This nomogramhas been successful for predicting the "time to failure" via pitting type I inwaters of a limited range of compositions. As was discussed, nevertheless,pitting type I is not one of the corrosion mechanisms that contributes toreleasing great amounts of copper to water.

The U.S. Copper Development Association developed a computer model forpredicting corrosion propensity and "time to failure" (Lyman et.al. 1982),based on a small number of water samples from the U.S. and the data baseemployed by Lucey in the U.K. The results of this empirical model confirmsthat pH is the most relevant variable and that below pH 7.8 waters are morecorrosion. The second most important variable was carbon dioxide, with thedividing line at 25 ppm. Sulfate/chloride ratio, as well as dissolved oxygen,were also found to be of major importance. Lyman et al.(1982) point out thatone of the flaws of this model is that it leaves out too many variables, such asflux, and corrosion inhibitor residues, etc.

The results of these models, even though empirical, have served the purposefor which they were designed but it is evident that a larger data base of watercompositions related to failure is required in order to predict the corrosionpropensity of any type of water via any type of mechanism.

29/10/2001 47

Other corrosion models, such as the Langelier Index and Larson’s Index havebeen used widely for reducing copper by-product liberation to water(Edwards, Schock et.al.1996), however these models are not suitable for thisapplication. Langelier’s Index was derived for waters where precipitation ofcalcite occurs and it is now known that this is not the case for the majority ofpotable waters. On the other hand, Larson’s index, originally developed forthe steel industry, predicts the opposite effect of Shock’s solubility model,which concludes, that adding chloride to a solution, would be more corrosive,whereas it has been shown that high chloride concentrations have beneficialeffects on corrosion. Similarly, Larsson’s index predicts that the morebicarbonate added to a solution, the less corrosive it should be. Edwards,Schock et al.(1996) work show that what occurs is exactly the opposite inmost situations.

a- A Solubility Model and its Limitations

As it was mentioned in the introduction, most corroded copper from plumbingtubes is in solid form and the amount that is liberated into solution depends onthe solubility of these solids in equilibrium with water of a certaincomposition, and also on the amount of solid film that is detached from thefilm and goes in to solution as a solid. In the case of new pipes (Edwards,Schock et.al. 1996) the overwhelming majority of copper present in solution issoluble. In the case of old pipes, copper present in solution is usually mixedbetween soluble and solid, and the soluble fraction is greater than the solidfraction by a factor of 5 to 1, according to Lagos et.al.(1999). The relativefraction of soluble and solid copper in solution may depend on factors such asflow rate used when opening the tap, and also on the composition of the filmin the inner surface of the pipe. Figure 5.4.1 shows [(cstagnant total - cstagnant

dissolved)/cstagnant total] • 100 versus cumulative percentage of the houses surveyed;c is the copper concentration measured in 250 houses in the Santiago watersurvey in 1997 (Lagos et.al.,1999). In about 10% of the houses the dissolvedstagnant copper concentration was equal to the total stagnant copperconcentration. For approximately 20% of the houses the dissolved stagnantcopper concentration is less than 80% of the total stagnant copperconcentration.

Figure 5.4.1: percentage difference between dissolved and total stagnantcopper concentrations measured in the Santiago survey (Lagos et.al. 1999).

29/10/2001 48

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

90.00

0.0 20.0 40.0 60.0 80.0 100.0

percentile

% d

iffer

ence

bet

wee

n di

ssol

ved

and

tota

l sta

gnan

t cop

per c

once

ntra

tions

Shock, Lytle et al.(1995a) carried out extensive thermodynamic calculationsof copper solubility in waters of compositions normally found in drinkingwaters. These authors concluded that the “quantitative prediction of copperlevels in drinking water relies heavily on the solubility and physical propertiesof cupric oxide, hydroxide and basic carbonate solids that comprise mostscales or films in water supplies”.

At pH values normally found in drinking water the formation of the solidphase Cu(OH)2(s) is favored over the formation of cupric oxide or malachite,especially under anodic potentials and in new plumbing systems (Edwards,Schock et.al. 1996). Formation of CuO(s) can occur after growth of thecrystallite of Cu(OH)2(s) has taken place, in an estimated period of one monthor more. The solubility of CuO is much lower than that of cupric hydroxide.While carbon dioxide, chloride and ammonia species can be manipulatedexperimentally when assessing the reliability of solubility constants, nothingcan be done to avoid the presence of hydroxide species such as CuOH2 ,Cu(OH)2

0, Cu(OH)3- , except by operating at low pH.

Copper forms stable chloride and amine cuprous complexes in solution,however the cuprous ion itself is unstable in aqueous solutions.

Shock, Lytle et al. (1995b) also showed that “carbonate complexationdominates copper (II) solubility in most drinking waters over pH 7, and is thekey to developing effective cuprosolvency reduction strategies”. Coppersolubility increases with alkalinity above pH 7, while below this pH the effectis much less severe. Above pH 9.5 the complex CuCO3(OH)2

2- is formed

enhancing the solubility of copper hydroxide even more.

29/10/2001 49

These authors carried out thermodynamic calculations of copper solubility andtheir dependence on ionic strength at pH 8 and above (range of 0.001 to 0.02),and showed that the latter did not alter the former significantly because of “thedominance of uncharged aqueous species over much of the pH/DIC(Dissolved inorganic carbon) range”. The effect of ionic strength on solubilityis greater at lower pH (about 6.5) where it contributes approximately 2 mg/Lto copper solubility.

Edwards, Schock, et al.(1996) concluded that at pH between 7.0 and 8.5 andalkalinities between approximately 10 and 250 mg/L as CaCO3 , the

predominant copper species are Cu(OH)2(aq) , CuCO3(aq) , and CuHCO3+

,and therefore solubility depended on these species. The equation that wasworked out on the basis of solubility constants for these species at 25 °Cyields:

Soluble Cu(mg/L)=10(13.4-2pH)

+0.58+10(5.1-pH)

(alkalinity)+10(11.4-2pH)

(alkalinity) (8)

Thus, for a given pH the relationship between soluble copper and alkalinity islinear. The slope of the line is directly related to the likelihood of formingcopper carbonate complexes, and the lower pH the greater the slope. Thispredicted relationship has been shown to be in excellent agreement withexperimental data.

Taking Schock et al.(1995a)¡Error!Marcador no definido. data, at pH below 6 anddissolved inorganic carbon between 2 and 8 mg C/l the concentration ofcopper depends only on pH and complies with the following equation:

Soluble Cu (mg/L) = exp (12/(e3.9265*pH

)) (9)

Empirical models developed by KIWA (Dutch Institute for the Study ofWaters), (Van den Hoven 1988) agree with these results, but also include theeffect of sulfate according to the following equation:

Cumax = 0.52 TIC - 1.37 pH + 2(SO42-

) + 10.2 (10)

29/10/2001 50

Cumax is in mg/L while total inorganic carbon, TIC, and SO42-

are inmmol/L.

This model is valid, according to KIWA, for old pipes with a stabilizedcorrosion layer and for the following range of water compositions: pH 7-8.45;

Cl- 7-176 mg/L; temperature 8-19.8 C; SO4

2- 0-131 mg/L; TIC 0.75-6.5

mmol/L; KMnO4 0-29 mg/L; O4 4.5-12 mg/L.

Figure 5.4.2: shows the predictions of copper concentration of four equationsdeveloped by: Schock Lytle et. al.(1995a)¡Error!Marcador no definido., Edwards,Schock et al.(1996), and KIWA in the pH range from 7 to 8. The KIWAequation for old pipes includes a sulfate term and 70 mg/L were assumed forthis concentration.

Four Models that Predict the Concentration of Copper Released to Water

0,001,002,003,004,005,00

50 100 150 200 250

Alkalinity

Cop

per

(mg/

l) Kiwa(newpipe),1993

Kiwa(oldpipe),1988

EPA, 1995

Edwards 1996

It is crucial that the corrosion layer be stabilized, and herein lies one of thegreatest uncertainties related to the application of these models to old pipes. Ifthe corrosion layer is not stabilized, i.e., well adhered to the pipe wall, then thesolubility model is not valid because a sizeable part of the corrosion layer goesinto solution as solid particles. Reports of blue and green water correspond tothe presence of solids such as malachite, brochantite, or other solid coppercompounds in the water.

Models for the dissolution of copper compounds other than cupric hydroxide,such as tenorite, malachite, brochantite, etc., can be formulated, but another ofthe important uncertainties present in this approach is that different valuesexist for the thermodynamic equilibrium constants (Paulson 1980, Hidmi andEdwards 1999). Using one value for a constant rather than another one may

29/10/2001 51

signify overestimating dissolution by as much as one order of magnitude insome cases, such as that of tenorite.

One of the advantages of using thermodynamic models such as the onediscussed above is that powerful software, which includes large databases, areavailable such as MINEQL+ (Schecher et.al. 1998), and MINTEQA2 (Allisonet.al. 1991).

These models do not consider the particle or grain size of the scales or filmsbuilt on the pipes. Stumm et.al. (1996) have studied this relationship andconcluded that for particles smaller than approximately one micron, the valueof solubility may increase considerably, and thus, stability should decrease.Therefore, aging of small particles may involve recrytallization into largeones. Precipitation of tenorite is favoured in certain water compositions abovepH 7.5, because its solubility is lower than that of other copper compounds.When the precipitation of tenorite occurs with the formation of fine particles,its solubility increases considerably and can give way to chemical unstabilityand the formation of compounds such as cupric hydroxide which usually hashigher solubility than tenorite (Schindler 1967). Walsh et.al.(1995) found thatduring exposure to aqueous solutions of cupric chloride, the etch rate and thefilm composition is dictated by grain size of the original copper.

5.5 - Aspects of Copper By-Product Liberation, Water Composition,Stagnation, and Pipe Aging.

a-Stagnation, Equilibrium and Aging

Figure 5.5.1 shows the concentration profile as a function of time for oneSantiago house with ½ inch copper pipe diameter, and it is observed that thereis diffusion control (Van den Hoven 1995 and 1998, Lytle and Schock 1997b)i.e., that the diffusion of copper from the pipe wall to the center of a pipe isslower than the copper dissolution kinetics. This has proven to be the case forseveral water compositions, but there may be some water compositions forwhich copper dissolution kinetics is the controlling step.

Figure 5.5.1: Copper loading from a ½” diameter copper pipe. The figureshows standardized copper concentration versus time. Water Composition:

pH 7.4, Ca 84.5 mg/L, Cl- 102 mg/L, SO4 173 mg/L, HCO3 60 mg/L asCaCO3, Temperature 16.1 C.

29/10/2001 52

00.10.20.30.40.50.60.70.80.9

1

0 2 4 6 8 10

stagnation time (hours)

Sta

nd

ard

ized

Un

itsstagnation curve

Diffusion

The dissolution curve of Figure 5.2.2 indicates that equilibrium between thedissolving solid copper containing film on the pipe wall and water may beattained at approximately 8 hours, but it is not conclusive evidence of it.

In some water compositions, and depending on the psotion of the pipe(horizontal, vertical, etc.) there may be convective forces generated bydissolution – precipitation processes. In this case dissolution – precipitationkinetics may take a leading role, rather than molecular diffusion.

The age of the pipe, together with water composition have an effect on thelevel and equilibration time for copper in water. The following figuresillustrate the influence of these parameters as shown by Meyer (1996). Withwaters of alkalinity of 320 (as mg/L of CaCO3) and at pH 7.2, the dissolutionof copper proceeds rapidly up to 4 hours and then becomes slower. After 8hours stagnation, the copper concentration of water was measured to be 90%of the concentration obtained after 16 hours stagnation (see figure 5.5.2). Inwaters of lower pH and the same alkalinity (320 mg/L), the behavior ofsolubility may be quite different, as indicated in Figure 5.5.3. In this case theconcentration reaches a maximum after 4 hours and then the concentrationbegins to decrease, suggesting that precipitation is possibly taking place,presumably due to the change in solution composition. Whether this new solidforms a scale or is in the form of particles is speculation at this stage, butobviously in terms of regulatory impact it is very different. If the solid isliberated to solution in the form of particles, then this will have a regulatoryimpact, whereas if it is not, the only effect would be that of changingsolubility. Another case studied by Meyer is waters of pH 6.9 but of loweralkalinity (about 100 mg/L as CaCO3). In this case it is observed that anapparent equilibrium is reached after 4 hours stagnation. The copper

29/10/2001 53

concentration continues increasing after 8 hours, nevertheless, and after 16hours a new equilibrium is not reached (figure 5.5.4). In figures 5.5.2 and5.5.3 it is evident that the effect of pipe usage age is to decrease the copper by-product release to the water.

Figure 5.5.2- Copper concentration versus stagnation time in a copper pipe22*1 mm, K B8.2(CO2) = 0.82 mmol/L (36.1 mg/L of CO2); DIC = 77 mg/L(Alkalinity = 320 mg/L as CaCO3); pH = 7.2; Modified from E.Meyer(1996)

0

1

2

3

4

5

6

7

0 4 8 12 16

5 month 12 month 24 month 86 month

Copper/(mg/l)

Stagnation Period / h

Figure 5.5.3- Copper concentration versus stagnation time in a copper pipe22*1 mm, K B8.2(CO2) = 1.5 mmol/L (66 mg/L of CO2) ; DIC = 96 mg/L

(Alkalinity = 400 mg/L as CaCO3) ; pH = 6.9; Modified from E.Meyer(1996)

0

2

4

6

8

10

0 4 8 12 16

5 month 8 month 19 month 56 month

Copper/(mg/l)

Stagnation Period / h

29/10/2001 54

Figure 5.5.4- Copper concentration versus stagnation time in a copper pipe22*1 mm, K B8.2(CO2) = 0.6 mmol/L (26.4 mg/L CO2); DIC = 31 mg/L

(alkalinity = 129 mg/L as CaCO3); pH = 6.9 ± 0.2; Modified from E.Meyer(1996)

0

500

1000

1500

2000

2500

0 4 8 12 16

5 month 12 month50 month 76 month

Copper/(µg/l)

Stagnation Period/ h

Note: K Bx.x(CO2) is the basic capacity of water at pH=x.x.

Sauter et.al.(1995) obtained comparable results regarding age, and equilibriumtimes for the corrosion of brass alloys.

Most homes have stagnation periods of less than 8 hours (Lagoset.al. 1999),and therefore for regulatory purposes, a stagnation requirement of 6 to 8 hrsseems adequate, according to Figures 5.5.1 to 5.5.3.

Hidmi and Edwards (1999) have studied the effect of aging on cuprichydroxide solubility in a solution of pH 7.0, and have found that a dissolutionpseudo-equilibrium is established at a copper concentration of 6 mg/L after 7to 8 hours stagnation. During this pseudo-equilibrium the copperconcentration falls slightly to about 5.6 mg/L in a period of about 20 hours,and afterwards the copper concentration begins to fall steadily down to about3 mg/L in a period of approximately 20 hours. This fall in concentration ismarked by a fall in pH and the change in the color of the solid from blue todark brown. A second pseudo-equilibrium is established at this level for aperiod of about 60 hours and also with a slight decrease in copperconcentration. Finally the concentration continues to fall steadily down toabout zero mg/L in a period of approximately 630 hours. Similar results wereobtained at pH 7.5. At pH’s of 8 and higher, a true equilibrium is establishedvery quickly. The authors conclude that during the occurrence of pseudo

29/10/2001 55

equilibrium concentration plateaus, several solids coexist at lower pH’s (7 and7.5), namely gerhardite (Cu2NO3(OH)3), Likasite Cu2NO3(OH)3 and cuprichydroxide (Cu(OH)2). The final solid product formed after 720 hours istenorite. When pH is 8 or greater only quick equilibria due to the formation oftenorite are observed.

One of the few reports about long time aging behavior, i.e., several years, ofcopper pipes was published in Nielsen (1995) for water of pH between 7.4 and7.9, bicarbonate concentration between 4.5 and 5.8 mmol/L, sulfate between0.6 and 1.5 mmol/L, and chloride between 1.4 and 2.8 mmol/L. The datashows that the release of copper after 5 and one half years is still between 2and 3 mg/L of copper, after stagnation. An aging process was not apparenthere, and only during the first 3 weeks were copper concentrations observed tobe between 4 and 5 mg/L. After this period the concentration stabilized at 2 to3 mg/L, save for rare higher fluctuations. Thus, in this case the aging processwas produced extremely rapidly, and also the yield of copper has remainedvery high during more than 5 years. This seems to be an unusual behaviorwhich could be due to the very aggressive composition of the water, especiallyin regard to alkalinity.

Figure 5.5.5 shows the stagnant dissolved copper concentration for theSantiago survey (Maggi 1999, Lagos et.al. 1999) as a function of pipe age.The linear trend line shown indicates that as an average, older pipes dissolveslower than newer pipes. The correlation between old pipes, i.e., more thantwo years old (HM Germany 1990), is not significantly better than that ofFigure 5.5.5. The same applies for new pipes. Also, non dissolved copper andpipe age are poorly correlated. These findings suggest that copper dissolutionmay reach the same level for pipes of different ages, depending on the watercomposition, temperature, and physical factors, such as pipe diameter, circuitdesign, etc.

Figure 5.5.5: Stagnant dissolved copper concentration for 218 houses in theSantiago survey versus pipe age (Maggi 1999, Lagos et.al. 1999). The meanstagnation time was 7.85 hours.

29/10/2001 56

1

10

100

1000

10000

0 10 20 30 40 50

Age (years)

Stag

nant

dis

solv

ed c

oppe

r co

ncen

trat

ion

(ug/

L)

b - pH and Carbon Dioxide

Aspects of the effects of pH and of carbon dioxide on solubility and copperby-product liberation have been discussed in section 5.3. These aspects willnot be reviewed.

The presence of carbon dioxide in (atmospheric) air leads to the dissolution ofcarbon dioxide into water that is in contact with air. In the absence of otherchemical compounds, the pH of water depends on the dissociation of carbondioxide into protons and bicarbonate. Carbon dioxide itself is a non-reactivedissolved gas which seems to play no direct role in corrosion, except throughthe modification of the water's pH that is the most significant variable indetermining the corrosivity of water. Some authors have suggested that CO2plays a direct role in corrosion and pitting, however this has not beenconclusively shown (Sander et.al. 1995). Recently, Edwards, Schock etal.(1996) have indicated that the main role of carbon dioxide could befacilitating copper carbonate complexation. Indeed, soluble copper “tied upwith carbonate species” is a linear function of dissolved carbon dioxide.

In general, acid waters are more corrosive, with respect to copper, than basicwaters, although waters of pH 9 can be corrosive in special circumstances andwaters of pH 6 can be quite innocuous in special cases (see Figure 5.5.6.).Above pH 8, waters are usually non corrosive and below pH 6.5 waters areusually corrosive.

Figure 5.5.6: Stagnant copper concentration versus pH for the Santiagosurvey (Maggi 1999, Lagos et.al. 1999).

29/10/2001 57

1

10

100

1000

10000

6.5 7 7.5 8 8.5 9

pH

Stag

nant

tota

l cop

per c

once

ntra

tion,

ug

/L

c - Oxygen

Oxygen is a cathodic depolarizer and if corrosion is under cathodic control,the corrosion rate should be proportional to the area of the cathode and to theoxygen flux. This occurs in uniform corrosion. In systems which are not undercathodic control, such as type I pitting (Cruse et.al. 1988), oxygenconcentration is not linked to corrosion rate, but oxygen is necessary tosupport the process. Therefore, oxygen is not always a controlling factor and itis rarely a limiting factor because waters are usually aerated. The presence ofoxygen is necessary for corrosion unless a non aerobic process takes place,such as in some types of MIC. Also, due to the work presented in Ives andRawson papers (1962 a,b,c,d) it may be possible that dissolution of copperproceeds after all the oxygen has been depleted, because copper could reactanodically with water to form cuprous oxide and protons (Werner 1995).

d - Carbonates and Bicarbonate

The effects of carbonates and bicarbonates has been extensively studied inrelation to the corrosion of copper (Cruse and Pomeroy 1974, Chen andLyman 1972, Campbell 1971, Shalaby et.al. 1990). The effect of calciumcarbonate, for instance, is thought to be beneficial in spite of the fact that thereason is not well understood. The apparent formation of surface films is oneof the possible reasons for the corrosion inhibition, but the effect is beneficialwell below the saturation of calcium carbonate. The effect of bicarbonate isfound beneficial as a corrosion inhibitor, since it is thought to facilitate theformation of calcium carbonate and malachite protective films. It is alsoapparent that at high concentrations of this ion (>0.3 M) it can complex copper

29/10/2001 58

and prevent the formation of surface protective films. Recently, though, it hasbeen shown that copper corrosion increases dramatically after aging at pHabove 7.0 and in the presence of bicarbonate due to the formation of a filmwhich catalyzes the reduction of oxygen (Edwards, Meyer at.al. 1994). Afteraging and at pH above 8.5 bicarbonate was found to be passivating. Edwards,Schock et al.(1996) confirm the dual effect of bicarbonate: below pH 8.1 it isaggressive and above this value it is passivating, although the explanations forsuch dual behavior remain at the level of hypotheses.

e - Sulfates and Chlorides

The role of sulfate and chloride ions in the corrosion of copper has beenstudied by many authors, but the conclusions of these studies do not lead to agreater understanding of the mechanisms that are behind the greater corrosionobserved when there is sulfate and/or chloride present. Copper corrodes fasterat increasing chloride concentrations (Braun and Nobe 1979, Nishikata et.al.1990, Drogowska et.al.1987), but under some circumstances (Edwards,Ferguson et.al. 1993) high concentrations of chloride can inhibit corrosionwhen sufficient surface protective film was already present when chloridecame in contact with copper. Some authors support the idea that chloridecontributes to corrosion only when present in low concentrations (Lucey1972b, Cohen and Myers 1984). Edwards, Meyer and Rehring (1994) havefound that copper surfaces passivate in the presence of chloride at pH>7.0,whereas copper is activated at pH values between 5 and 10 in the presence ofsulfate, and also of nitrate and perchlorate. At pH 7.0 these authors found thatthe aggressiveness of anions was HCO3->SO42->NO3>ClO4->Cl- and at pH

8.5 the aggressiveness strength changed to SO42->ClO4->NO3>HCO3->Cl-.

Edwards, Ferguson et al.(1993) reviewed the role of sulfate in coppercorrosion and conclude that this is substantiated by numerous correlationsbetween increased pit frequency and high sulfate to chloride ratio, high sulfateto carbonate ratios and high sulfate to hardness ratios. The correlations areconfirmed, according to these authors, by data obtained from corrosion cases.

Most of the discussion found in the literature that relates the roles of sulfateand of chloride ions and corrosion is not linked to the type of mechanism, i.e.I, II or III.

29/10/2001 59

From a point of view of solubility, Schock, Lytle and Clement (1995b)conclude that aqueous sulfate complexes are not likely to be significantinfluences on copper by-product liberation in the pH range of 6 to 8 at low andmoderate dissolved inorganic carbon, DIC, and in the absence of malachite.

Royuela and Otero(1993) studied the effect of chloride and sulfate ions oncorrosion and concluded that the greatest corrosion current or corrosivityoccurs in waters near neutrality, with high salinity [Cl- + SO4

2-] and lowHCO3

-/SO42-, and the lowest corrosivity occurred for solutions with low

salinity. All the corrosion potentials measured at the end of tests are in thestability zone for Cu2O in the Eh-pH diagram. The greatest corrosivitycoincides with the larger conductivities. Copper dissolution was notsignificant in these experiments, since most of it seems to have precipitatedout of solution.

g- Effect of Natural Organic Matter, NOM

Rehring and Edwards (1996) have studied the effects of natural organic matter(NOM) on copper corrosion. This effect has been known to occur for morethan 40 years. NOM is a heterogeneous group of organic molecules that arepresent in natural waters at concentrations between 0.05 and 20 mg/L totalorganic carbon (TOC). The most common components of NOM are humic andfulvic acids, anionic polyelectrolytes with molecular weights ranging fromseveral hundred to several million daltons.

NOM's removal from natural waters is justified because they potentially serveas organic precursors for the formation of carcinogen disinfection by-products(DBP) when water disinfectants such as chlorine or other chemicals are addedto water (Rehring and Edwards 1996). Also NOM's have been recognizedsince the 1950's as having an effect on copper corrosion (Campbell 1971).

Some of the most commonly used methods to remove NOM are adsorptionwith granular activated carbon (GAC), enhanced coagulation using alum orferric chloride.

The GAC method, enhanced coagulation and ozonation were investigatedexperimentally by Rehring and Edwards (1996) and it was concluded that thepresence of NOM increased anodic currents, i.e., corrosion rates, at all pH'salthough this effect is more pronounced when the pH is 6.0 or less. At pHgreater than 7.5 it was found that the effect of NOM on corrosion were minor.

29/10/2001 60

Also, the rate of corrosion increased more with alum ((Al2SO4)3•18H2O)

treatment than with ferric chloride (FeCl3•6 H2O), most probably due to thegreater corrosion effect of sulfate than that of chloride. GAC adsorptionincreased corrosion rates whereas ozonation had little or no impact.

Korshin et al. (1996) found that NOM promotes the formation of pits in acertain narrow range of concentrations (0.1 to 0.2 mg/L of NOM, Chloride 35mg/L, sulfate 96 mg/L, bicarbonate 244 mg/L, temperature 20 C, pH 7.3) andsuppresses this type of corrosion at higher dosages. Leaching, or liberation ofcopper by products, is increased.

h - Ionic Strength

The ionic strength is given by the product of the concentration of each ionpresent in solution multiplied by the square of the valency of each ion. Allsuch products are then added. The physical significance of ionic strength is thereach of the ionic atmosphere in the solution. The larger the number of ionspresent, the larger the ionic strength. On the other hand the activity of eachspecific ion decreases with increasing ionic strength due to greatercompetition for ion exchange.

Alkalinity is the addition of the concentrations of strong bases present inwater, namely carbonate and bicarbonate ions. If only strong bases are presentin solution there is usually proportionality between alkalinity and ionicstrength. When ions such as chloride, sulfate, calcium and magnesium arepresent in appreciable concentrations in solution, then there is noproportionality between ionic strength and alkalinity.

A data base of water compositions published for types I, III and MICcorrosion has been compiled. The following figures show some of therelationships between pH, alkalinity, ionic strength and specific ionconcentrations.

Figure 5.5.7 shows that corrosion type III occurs in water with pH from 6 to9.2 but alkalinity (Figure 5.5.8) is restricted to values between 10 to 38 mgCaCO3/L. Most of the 11 cases compiled for type III pitting occurred at

around pH 7.0 and alkalinity between 10 and 30.

29/10/2001 61

Figure 5.5.7- Alkalinity versus pH for types I and III pitting. Data from thestudies by Moss et al. 1984; Taylor et al. 1993; Linder et al. 1982; Page et al.

1974.; Cohen et al. 1984; Cruse et al. 1974;Cornwell et.al.1976.

0

100

200

300

400

500

6 7 8 9 10

pH

Alka

linity

(mgC

aCO3

/lt)

Type III pitting

Type I pitting

Type I corrosion can occur in waters with pH from 6.5 to 8.5, i.e. a morerestricted range of pH than type III corrosion. For the 28 cases compiled fortype I corrosion, but in only one case (Figure 5.5.8) did the water have analkalinity below 70 mg CaCO3/L. On the other hand, the dispersion ofalkalinity is quite high, reaching up to 450 mg CaCO3/L, but most of the datapoints occur at alkalinity between 200 and 350 mg CaCO3/L and in a pH rangebetween 7 and 7.5.

Figure 5.5.8 shows that for type III pitting the ions present are mainlycarbonate and bicarbonate, whereas in the case of type I pitting there are otherions apart from carbonate and bicarbonate present in solution.

Figure 5.5.8- Ionic strength versus alkalinity for types I and III pitting.Data from the studies by Moss et al. 1984; Taylor et al. 1993; Linder et al.

1982; Page et al. 1974.; Cohen et al. 1984; Cruse et al. 1974; Cornwell et al.1976.

29/10/2001 62

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

0 100 200 300 400 500

Alkalinity (mgCaCO3/lt)

Ioni

c St

reng

ht Type III pitting

Type I pitting

k- Sulfide and Chlorine

The effect of sulfide on liberation of copper by-products to water was studiedby Jacobs et al.(1998). Water with 23 mg/L sodium chloride, 30 mg/L sodiumsulfate, 25 mg/L sodium bicarbonate, pH 6.5 and 9.2, and 5 mg/L sulfideadded as sodium sulfate, was prepared. Water of the same composition butwithout sulfide was also prepared. In test pipes 8 months old with waterstagnated 3 and 6 hours, the release of copper to water increased 5 times (toapproximately 1.5 mg/L) at pH 6.5 and 50 times (between 0.4 and 0.8 mg/L)at pH 9.5, with respect to liberation of copper in water of the samecomposition but without sulfide. Sulfide induced corrosion proved difficult tostop. Removing sulfides from the water, adding chlorine or deaerating mightnot stop corrosion in short periods (up to two months). With chlorine, thecorrosion rate increased markedly, possibly because chlorine is a strongeroxidant than oxygen. This agrees with results reported earlier about the effectof chlorine (Reiber 1989), (Hong et.al. 1998).

Even after one hour of deaeration, purging nitrogen, did not reduce thecorrosion rate. Therefore, it is concluded that even traces of oxygen can allowfast corrosion rates in the presence of a strong catalyst such as sulfide.

Several corrosion cases in the USA, and one in Scotland, analyzed on the MICcorrosion section in this paper, are discussed by Jacobs et al. where sulfidemay have been present but was not associated with the cause of corrosion.

29/10/2001 63

5.6- U.S Lead Copper Rule Approach to Copper By-Product LiberationPrevention.

Lytle et.al. 1995, 1996) investigated the effect of zinc orthophosphate, alkalimetal orthophosphate, and of sodium silicates on the corrosion of lead andcopper in drinking water. Royuela et.al.(1994) investigated the effect ofsilicates and of sodium polyphosphates on copper corrosion.

Two approaches that are used to reduce copper by-product liberation will beexamined in this section: the first, is the use of phosphate inhibitors and thesecond, increasing pH. These methods may be used simultaneously.

The study by Dodrill and Edwards (1995) considered the effects of pH andalkalinity on 90th percentile lead and copper concentrations, with and withoutthe addition of inhibitors. For this purpose, water was classified in severalranges of pH and four ranges of alkalinity, as shown in tables 5.6-1 and 5.6-3.This classification spread the data into equally populated segments

The analysis presented by these authors considers the aggregate data of allutilities, thus the results are valid as a general trend and are not valid for oneutility in particular.

a - Exceedence Without the Use of Inhibitors. When inhibitors are not usedit was found that copper release increased in waters with pH lower than 7.4and within this range, release was greater for alkalinities greater than 7.4 mg-CaCO3/L. Table 5.6-1 shows the percent reduction in copper release due toindicated increase in pH and demonstrates this statement.

The authors maintain that these results agrees with the hypotheses thatCu(OH)2 solids control the solubility of copper (Schock, Lytle et.al. 1995a,Schock and Lytle 1994 and 1996, Schock, Lytle and Clement 1994, Meyerand Edwards 1994) i.e., at high pH, copper solubility decreases, and highcopper release is observed at high alkalinities when pH is < 7.8.

The results shown in table 5.6-1 also mean that when the pH is increased tovalues greater than 8.4 the release of copper is reduced considerably,irrespective of the alkalinity. By increasing the pH from a range of 7.4-7.8 to7.81-8.4 the release of copper is reduced significantly only when alkalinity isabove 74 mg/L.

29/10/2001 64

Table 5.6-1: (reproduced from and Dodrill and Edwards 1995)pH change pH change pH change

Alkalinity(mg/L as CaCO3)

<7.4 to7.4-7.8

7.4-7.8 to7.81-8.40

7.81-8.4 to>8.4

< 30 68% (95 pcc) 1% 75% (95 pcc)30-74 43% (95 pcc) 30% 45% (95 pcc)

75-174 39% 51% (95 pcc) 47% (85 pcc)>174 -1% 53% (90 pcc) 100% (95 pcc)

Pcc = percentage confidence.

No waters with pH > 7.8 exceeded the copper standard and the highestpercentage of exceedances was found in waters of pH 7.4 and alkalinity < 30mg/L.

b - Inhibitor Effects

The presence of phosphate species in solution (e.g., orthophosphate, zincorthophosphate, hexametaphosphate, polyphosphates and various blends)presumably change the solids that are formed when copper plumbing tubes arecorroded. The following copper (II) and orthophosphate minerals werecompiled by Schock, Lytle and Clement 1994):

Table 5.6-2: Reported Minerals Containing Copper (II) andOrthophosphate (from Schock, Lytle and Clement 1994).

Libethenite Cu2PO4OH

Cornetite Cu3PO4(OH)3Reichenbachite Cu5(PO4)2(OH)4

Ludjibaite Cu5(PO4)2(OH)4Pseudomalachite Cu5(PO4)2(OH)4.H2O

Nissonite Cu2Mg2 (PO4)2(OH)4.5H2O

Zapatalite Cu3Al4 (PO4)3(OH)9.4H2O

Turquoise CuAl4 (PO4)4(OH)8.5H2O

Sieleckite Cu3Al4 (PO4)2(OH)12.2H2O

Planerite (Cu,Ca)Al6 (PO4)4(OH)8.H2O

Hentschelite CuFe2 (PO4)2(OH)2

29/10/2001 65

Andrewsite(Cu,Fe

3+)(PO4)4(OH)8.4H2O

ChalcosideriteCu,Fe

3+(PO4)4(OH)8.4H2O

PhosphofibriteKCuFe

3+15(PO4)12(OH)12.12H2O

Early calculations with orthophosphate addition predicted negligible effects onthe solubility of copper solids at neutral pH and low alkalinity. However thesepredictions were based on the assumption that malachite was the controllingsolid phase, rather than solid cupric hydroxide. Unfortunately the solubilitydata for othophosphate species is very poor and at present it is not possible topredict with certainty the solids that should be formed in the presence ofotrhophsophate species in solution. However, if cupric hydroxide is assumedto be the controlling solid phase, then the stability field for at least one of theorthophosphate solids with reported solubility constants, is much larger.

Dodrill and Edwards (1995) found that the addition of phosphate basedcorrosion inhibitors to water was effective only at certain pH and alkalinity.

For instance, the addition of 1 to 5 mg/L of orthophosphate to water waseffective at pH 7.5 and 8.0 (Schock, Lytle and Clement 1995b, Benjaminet.al.1990). Below pH 6.0 the corrosion rate was not affected.

The conclusions of Dodrill and Edwards (1995) regarding the effect of theaddition of inhibitors on the percent of copper release to solution issummarized in table 5.6-3.

The effectiveness of inhibitors on the reduction of copper release are mostlyconfined to the lower pH, although at pH>8.4 and alkalinity 75-174 there isalso a region where the effects are positive. The table also shows that atcertain pH and alkalinities the effects of inhibitors can also be negative, i.e.,their presence would release more copper to solution. This is indicated by thenegative values.

Table 5.6-3-Reproduced from Dodrill and Edwards (1995). Table shows thepercentage release of copper for those utilities that use inhibitor with respectto the utilities that do not use inhibitors.Alkalinity mgCaCO3/L

pH < 7.4 pH 7.4-7.8 pH 7.81-8.4 pH > 8.4

29/10/2001 66

<3030-7475-174>174

56% pcc11%51%23%

13%-2%34%4%

47%-45%-34% (90 pcc)

-26%-5%50% (85 pcc)

pcc=percentage confidence

Dodrill and Edwards (1995) conclude their work with the followingrecommendations to utilities:

Exceedance problems are confined to two water characteristics:

1- pH < 7.0 & alkalinity < 30 mg/L2- pH < 7.8 & alkalinity > 90 mg/L

Phosphate inhibitors can be effective in reducing copper release only belowpH 7.8. Above this value its effects are highly variable.

c- Strategies to Increase pH

As was mentioned, pH is the single variable that contributes most to copperby-product liberation. Increasing pH should have, therefore, the greatest effecton reducing copper by-product liberation to water.

Cohen and Myers(1993) compared the effectiveness of addition of: lime,caustic soda, soda ash, aeration and other methods, for raising the pH of water.Lime is purported as the most effective and cheapest method to elevate pH,and its main limitation is its limited solubility in carbon dioxide free water.Caustic Soda (NaOH) is effective in raising pH and the only cautionrecommended is not to exceed 200 mg/L recommended by the WHO. SodaAsh (Na2CO3) requires higher dosages for the same neutralization benefits aslime, and therefore its use has declined. Aeration removes CO2 and it can alsoreduce sulfides, oxidize Mn and or Fe bicarbonates to their filterable oxides.

Edwards, Schock et al.(1996) consider three alternative methods to raise pH:two methods are based on the addition of caustic (NaOH) or lime (CaOH2). IfpH is raised to a value of 8.5 the added hydroxide contributes to increasingalkalinity, thus, the model discussed in section 5.2-a predicts that copper by-product liberation increases. If the third method is used, which consists ofaeration, dissolved CO2 is stripped from the solution and therefore pH

29/10/2001 67

increases but alkalinity does not (Lytle, Schock et.al., 1998). For instance ifpH is raised from approximately 6.6 to 7.2, the final predicted coppersolubilities are 3.8 mg/L when using lime, 3.4 mg/L using caustic, and 2.8mg/L using aeration. Calcite precipitation is also avoided by adding air tosolutions, rather than lime or caustic.

6- Conclusions

Electrochemical theory predicts copper plumbing tubes oxidize to cuprite oncontact with drinking water. This has been experimentally confirmed for atleast one water composition. The other half cell reaction occurs at the cupritesolution interfase and consists usually of the reduction of oxygen. The cupritefilm remains as the only contact with the metal. Above pH 6.5 cuprite furtheroxidizes, usually to malachite, or tenorite. No cupric ions are liberated tosolution with these electrochemical reactions. Depending on the initial watercomposition, copper compounds bearing sulfates, chlorides, nitrates and othersalts may form on top of cuprite, and ususally at pH values under 7.0. UnderpH values of 6.5 cuprite may oxidize directly to cupric species in rare watercompositions.

Simultaneously with these corrosion processes, dissolution of the corrosionproducts occurs, allowing copper to enter the water as cupric species.Precipitation of copper compounds on the pipes inner surface, already coveredby scales, may start at the same time in certain water compositions, and startsafter dissolution when the solubility product of dissolved copper is reached.Copper may also enter the solution by detachment of particles from the scalesformed during the corrosion processes or after precipitation of coppercompounds.

Aging takes place due to dissolution/precipitation processes, since for a givenwater composition, scales with higher solubilities should dissolve and thecompounds that precipitate are those with lower solubilities. Differences intemperature between summer and winter, water composition variations, andother factors contribute to the formation of more than one copper compoundon the inner surface of copper pipes. After some time, usually a few months,and some times a few years, dissolution slows down because less solublescales have either replaced or covered more soluble scales. It is assumed thatthe rate of the corrosion processes that occur at the copper/cuprite interfasealso slows down when outer scales are dissolving slower.

29/10/2001 68

The proportion between copper solids and dissolved copper species dependson water composition, pipe age, water velocity, and possibly other variables.Available drinking water monitoring data shows that the partitioning betweensolids and dissolved copper species varies between zero and about 90%, andthere is no theory, at present, to predict this proportion. Solubility models havesuccessfully predicted the dissolved copper concentration in certain watercompositions at one given time. These models have not yet been developed topredict aging processes.

Dissolution takes place across the water exposed scales surfaces, and isusually coupled with uniform corrosion. This added to aging, yield long life,possibly 60 years or more, for copper plumbing tubes.

Pitting, on the other hand, can shorten the life of copper plumbing tubes to afew weeks before rupturing. This review classifies pitting corrosion and othertypes of corrosion in two groups, those that may contribute significantly toraise the copper concentration in drinking water, namely uniform corrosion,pitting type III, and jointing and MIC corrosion, and those that do not, i.e.,pitting types I and II, erosion corrosion, cavitation, soil corrosion, corrosion inconcrete slabs, stray current corrosion, stress cracking corrosion, and galvanicand thermogalvanic corrosion.

The conditions for the occurrence of pitting and the other types of corrosionaforementioned, are known and they can be prevented. Sometimes they can berepaired or reverted once the phenomenon has already started. Apart from thethe role of carbonaceous film - formed during annealing of soft tubes in themanufacturing process - on pitting type I, it is accepted that the watercomposition is the main factor contributing to copper liberation from copperplumbing tubes to drinking water.

Regulatory concerns about drinking water fundamentally address at presentthe daily and maximum concentrations of copper in drinking water, and thepotential effects on acute gastrointestinal illnesses. Some drinking waterregulations take into account the aging of copper plumbing tubes and establishdifferent monitoring protocols for new and old pipes. These regulations alsoconsider regulatory differences for aggressive waters towards copperplumbing tubes, i.e., waters that may yield copper concentrations higher thanthe regulation. Some regulations require measurement of the maximum copperconcentration to which people may be exposed, and this occurs due to

29/10/2001 69

dissolution of copper scales during water stagnation, usually overnight. Housetap monitoring is required for stagnation periods between 6 and 12 hours indifferent countries.

When regulatory action levels are exceeded, the USEPA requires waterutilities to treat the water in order to reduce copper by product liberation.Raising pH remains the main variable to control, and since the roles ofbicarbonate and of carbon dioxide on copper by-product liberation are nowbetter understood, it has been shown that raising pH by adding lime or causticsoda raises alkalinity and is less effective than aeration, which strips carbondioxide and does not alter alkalinity. [editor: From my days preparing mediafor toxicology tests, I recall excessive aeration supersaturating with gases suchas CO2. The reverse of this statement…..author: see Edwards, Schock et.al.,1996)] Phosphate inhibitors are also used for corrosion control of iron, copperand lead, and their addition has had a positive impact on the reduction ofcopper by-product liberation to water.

29/10/2001 70

Acknowledgments

This Work was sponsored by the Chilean Government’s Technical AdvisoryCommittee Regarding the World Health Organization Provisional Guidelinefor Copper in Drinking Water and by the International Copper Study Group,ICSG.

Special thanks to Marcelo Andia, who gathered much of the backgroundinformation for this paper. Also many thanks to M. Edwards and G. Geeseywho read one of the versions of the paper and contributed very usefulcomments.

7- References

Allison, J.D., D.S.Brown, and K.J. Novo-Gradac, 1991.MINTEQA2/PRODEFA2, A geological assessment model for environmentalsystems: version 3.0 User´s manual. USEPA/600/3-91/012., Athens, Georgia.

Angel P., H.S. Campbell, and A.H.L. Chamberlain. 1990. MicrobialInvolvement in Corrosion of Copper in Fresh Water", Interim Report,International Copper Association.

Arens P., G.J. Tuschewitzki, M. Wollmann, H. Follner, and H. Jacobi, 1995.Indicators for microbiologically induced corrosion of copper pipes in a cold-water plumbing system. Int. J. of Hygiene and Env. Medicine. 196: 444-454.

Arens, P., G.J. Tuschewitzki, H. Follner, H. Jacobi, S. Leuner., 1996.Experiments for simulating microbiological induced corrosion of copper pipesin a cold-water plumbing system. Materials and Corrosion. 47: 96-102.

Arens P. 1999. Personnal Communication. Deutsches Kupfer-Institut,Dusseldorf, Germany.

AWWA., 1996, American Water Works Association, Database AWWAProject: Initial Monitoring Experiences of Large Utilities USEPA’s LeadCopper Rule. Version 2 including modifications of Dodrill and Edwards.Denver CO., USA.

29/10/2001 71

Bard, A.J. (ed.) 1976. Encyclopedia of Electrochemistry of the Elements.Volume 6: Al, In, Ir, Os, Pd, Pt, Rh, Ru, Sc, Y. Lanthanides New York:Marcel Dekker.

Baukloh A., H. Protzer, and U. Reiter, 1989. Kupferrohre in derHausinstallation – Einflub von Produktqualitat, Verarbeitungs – undIntsallations – bedingungen auf die Bestandigkeit gegen Lochfrab Typ I,Metall, 43, No 1, pgs 26-35.

Beguin-Bruhin, Y., F. Escher, H.R. Roth, 1983, Threshold concentrationdetection of copper in drinking water, Lebensmittel wissenschaft und-technologie. 16(1):22-26.

Benjamin, M.M., S.H. Reiber, J.F. Ferguson, E.A. Vanderwerff and M.W.Mille. 1990. Corrosion of copper pipes in Chemistry of Corrosion Inhibitors inPotable water. pgs. 210. Denver, Colorado: AWWAR.

Bertocci, U. 1978. Photopotentials on copper and copper alloy electrodes. J.Electrochem. Soc. 125: 1598.

Braun, M. and K. Nobe. 1979. Electrodissolution Kinetics of Copper in AcidicChloride Solutions. J. Electrochem. Soc. 126(10): 1666-1671.

Bockris, J.O.M, B. Conway, E. Yeager and R.E. White (eds.) 1981.Comprehensive Treatise of Electrochemistry. Volume 4: ElectrochemicalMaterials Science. New York: Plenum Press.

Callot P., A. Jaegle, A Kalt and G. Nanse, 1978. Pitting Corrosion of CopperTubes and Carbon Deposits: ESCS studies, Werkstoffe und Korrosion 29, pgs519-522.

Campbell, H.S. 1950. Pitting corrosion in copper water pipes caused by filmsof carbonaceous material produced during manufacture, J. Inst. Metals. 77:345.

Campbell, H.S. 1971. Corrosion , Water Composition and Water Treatment.Water Treat. and Exam. 20: 11-34.

Campbell, H.S. 1979. A Review: Pitting Corrosion of Copper and its Alloys",Localized Corrosion. Houston: NACE.

29/10/2001 72

CDA 1994, Annual data:Copper•Brass•Bronze: Copper Supply &Consumption in the USA in the period 1973-1993, Copper DevelopmentAssociation, N.Y., USA, 1994.

Cohen, A., and W.S. Lyman. 1972. Service Experience with Copper PlumbingTube. Material Protection & Performance. 11(2): 48-52.

Cohen, A. and J.R. Myers. 1984. Mitigation of Copper Tube Cold WaterPitting by Water Treatment. Proceedings Corrosion Conference (1984).

Cohen A., and J.R. Myers, 1987. Mitigating Copper Pitting Through WaterTreatment, J.AWWA, vol. 79, number 2, pgs 58-61

Cohen A., and J.R. Meyers, 1993. Water Tretament to Mitigate Corrosion ofCopper Plumbing Systems, Materials Performance, vol. 32, number 8.

Cohen, A. 1994. Occurrence and Control of Corrosion in Copper Water TubeSystems. Proceedings of the AWWA Water Quality Technology Conference,San Francisco, November 1994.

Cohen, A., and J.R. Myers 1995. "Overcoming Corrosion Concerns in CopperTube Systems", Corrosion 95, The NAC International Annual Conference andCorrosion Show, Paper 605.

Cornwell, F.J., G. Wildsmith, and P.T. Gilbert. 1973. Pitting Corrosion inCopper Tubes in Cold Water Service. British Corrosion Journal. 8: 203.

Cornwell F.J., G. Wildsmith, and P.T. Gilbert, 1976. Pitting Corrosion inCopper Tubes in Cold Water Service, Galvanic and Pitting Corrosion – Fieldand Laboratory Studie, ASTM STP 576, American Society for Testing andMaterials, pgs 155-179.

Cruse, H. and R.D. Pomeroy. 1974. Corrosion of Copper Pipes. JAWWA.67(8): 479-483.

Cruse, H., Von Franque O., and Pomeroy R. 1988. Corrosion in Potable WaterSystems. Chapter 5 of Corrosion in Pipes, Published by the American WaterWorks Association, pg. 317- 416.

Deutscher, R.L., and Woods R. 1986. Characterization of oxide layers oncopper by linear potential sweep voltammetry. J. of Applied Electrochem. 16:413-421.

29/10/2001 73

Dodrill, D. and M. Edwards. 1995. Corrosion Control on the basis of utilityexperience, J.AWWA, V.89, No 7, 74-85.

Drogowska, M., L. Brossard, and H. Menard, 1987. Anodic CopperDissolution in the Presence of Cl- Ions at pH 12. Corrosion. 43: 549-552.

Duthill, J.P., G. Mankowski and A. Giusti. 1996. The synergetic effect ofchloride and sulphate on pitting corrosion of copper. Corrosion Science.38(10): 1839-1849.

Edwards, M., J.F Ferguson and S.H. Reiber, 1993. On the pitting corrosion ofcopper. Journal of the AWWA. 86(7): 74.

Edwards, M., T. Meyer, and J. Rehring. 1994. Effect of Selected Anions onCopper Corrosion Rates. J.AWWA, 86:12, pgs. 73-81.

Edwards M., 1995. Worldwide perspective of copper corrosion. Proc. Of the1994 AWWA WQTC in San Francisco, CA.

Edwards, M., M.R. Schock and T.E. Meyer, 1996. Alkalinity, pH, and copper:corrosion by-product release. J.AWWA., 8:3, 81-94.

EU, 1998, Directive 98/93CE from the Council, Official Newspaper of theEuropean Community, 5 December, 1998.

Fitzgerald D.J., 1995, Copper guidelines values for drinking water: reviews inneed of review?, Regulatory Toxicology and Pharmacology, Vol. 21, pgs 001-003.

Force Institute 1999, Pilot project on copper investigation of 10 houses inCopenhagen, Corrosion Department, Force Institute, Copenhagen, Park Alle345, Denmark.

Geesey G.G., M.W. Mittelman, T. Iwaoka, P.R. Griffiths. 1986. Role ofBacterial Exopolymers in the Deterioration of Metallic Copper Surfaces,Materials Performance, 2: 37-40.

Geesey. G., C. Kalaiyappan, 1994. Determination of Biocide Treatment toControl Biofilm Growth and Associated "Blue Water" Phenomenon in CopperPlumbing Systems. Progress Report to ICA, Project number 484-94, NY.,USA.

29/10/2001 74

Geesey, G., P. Bremer, W. Fischer, D. Wagner, C. Keevil, J. Walker, A.Chamberlain, and P. Angel 1994. Unusual Types of Pitting Corrosion ofCopper Tubes Used for Water Service in Institutional Buildings. In Biofoulingand Biocorrosion in Industrial Water Systems. Boca Raton : Lewis Publishers.

Gerischer, H. 1966. Electrochemical behaviour of semiconductors underillumination. J. Electrochem. Soc. 113: 1174-1182.

Hadley, R.F. 1948. Corrosion by Microorganisms in Aqueous and SoilEnvironments. In Corrosion Handbook., H.H. Uhlig.(ed.), Pgs. 466-481. NewYork: John Wiley & Sons.

Hamilton, W.A., 1985. "Sulfate reducing bacteria and anaerobic corrosion",Annual Review of Microbiology. 39: 195-217.

Hidmi, L., and M. Edwards M., 1999. Role of temperature and pH in Cu(OH)2

solubility, ES&T, V.33, No.15, 2607-2610.

HM. Germany 1990. Potable Water Ordinance, Ordinance on Potable Waterfor Food Production Undertakings, Translation (December 1990), HealthMinistry, Germany.

Holm , R., R. Sunberg and E. Mattsson. 1982. Corrosion of Copper Pipes inFresh Waters - Swedish Experience. Conference on Corrosion of Copper andCopper Alloys in Buildings, Japan CDA, Tokyo.

Hong P.K.A, and Y-Y Macauley, 1998. Corrosion of Copper Tubing Exposedto Chlorinated Water, Water, Air and Soil Pollution, 108: pgs 457-471.

Hongve, D., and E. Andruchow. 1995. Cuprosolvency in soft drinking watercontrolled by alkalinity and pH. In Internal Corrosion in Water DistributionSystems, Proceed. International Seminar, Chalmers University of Technology,Goteborg, Sweden, May 22-24, 1995, pg. 197-205.

INCRA. 1988. Interim Report on the International Copper ResearchAssociation (INCRA), Research Report 407.

IPCS, 1999, Environmental Health Criteria for Copper, Vol. 200, Publishedby the WHO.

Ives, D.J., and Rawson A.E., 1962-a. Copper Corrosion I. Thermodynamic

29/10/2001 75

aspects. J. Elec.Soc. 109(6): 447-451.

Ives, D.J., and Rawson A.E., 1962-b. Copper Corrosion II. kinetic studies. J.Elec.Soc. 109(6): 452-457.

Ives, D.J., and Rawson A.E., 1962-c. Copper Corrosion III. Electrochemicaltheory of General Corrosion. J. Elec.Soc. 109(6): 458-462.

Ives, D.J., and Rawson A.E., 1962-d. Copper Corrosion IV. The effects ofsaline additions. J. Elec.Soc. 109(6): 458-462.

Jacobs, S.A., 1997. Sulfide-induced corrosion of copper in potable water.M.Sc. Thesis, Department of Civil Engineering, University of Colorado,Colorado, USA.

Jacobs, S., S. Reiber, and M. Edwards. 1998. Sulfide induced coppercorrosion. J.AWWA. 90(7): 62-73.

Johansson L. 1989. Importance of Water Composition for Prevention ofInternal Copper and Iron Corrosion. Chalmers University of Technology,Goteborg, Sweden.

Kiwa., 1998, Developing a new protocol for the monitoring of lead in drinkingwater, Order number SMT4-CT96-2112, Institute for the study of watre,KIWA, Nieuwegein, The Netherlands.

Korshin, G.V., S.A.L. Perry, J.F.Ferguson, 1996. Influence of NOM on coppercorrosion. J.AWWA, 88: 7, pgs 36-47.

Lagos, G. 1997. Population Exposure to Copper in Drinking Water. NationalEnvironmental Health Forum Monographs, Metal Series No 3, pp 52-59, 1997,South Australian Health Commission.

Lagos, G.E., L.C. Maggi, D. Peters, and F. Reveco, 1999. Model forEstimation of Human Exposure to Copper in Drinking Water. The Sc. Of theTotal Env., 239 (1999), pp 49-70.

Linder M. & E.K. Lindman. 1983. Investigation of Pitting Corrosion Type III ,in Copper Pipes. 9th Scandinavian Corrosion Congress, September, 1983.Swedish Inst. of Corrosion.

Lucey, V.F., 1967. Mechanisms of pitting corrosion of copper in supplywaters, Br. Corrosion J., 2: 175.

29/10/2001 76

Lucey, V.F., 1972a. Pitting Corrosion of Copper in Supply Waters: The Effectof Water Composition. British Non Ferrous Research Association, ResearchReport A1838.

Lucey, V.F. 1972b. Developments Leading to the Present Understanding of theMechanisms of Pitting Corrosion of Copper. Br. Corrosion J. 7: 36-40.

Lucey, V.F. 1982. Pitting Corrosion of Copper: a Review. Proceedings of theInternational Symposium, Corrosion of Copper and Copper Alloys in Building.Tokyo, March 1982.

Lytle D.A., M.R. Schock, and T.S. Sorg, 1995. Investigation on Techniquesand Control of Building Lead and Copper Corrosion by Orthophosphate andSilicate., Proc of The NACE International Annual Conference and CorrosionShow, paper No 609.

Lytle D.A., M.R. Schock, and T.S. Sorg, 1996. Controlling Lead Corrosion inthe Drinking Water of a Building by Orthophosphate and Silicate Treatment, J.NEWWA, vol 110, No3, pgs 202-217.

Lytle, D.A. and M.R. Schock, 1996. Stagnation time, composition, pH andorthophosphate effects on metal leaching from brass. EPA/600/R-96/103.September, 1996, Cincinnati, Ohio: EPA.

Lytle D.A., and M.R. Schock, 1997a. An Investigation of the Impact of AlloyComposition and pH on the Corrosion of Brass in Drinking Water, Advancesin Environmental Research, 1 (2), pgs 213-233.

Lytle D.A., and M.R. Schock, 1997b. Impact of stagnation time on thedissolution of metal from plumbing materials”, Proceedings AWWA AnnualConference, Atlanta, GA, Volume D, pgs. 193-225.

Lytle, D.A., M.R. Schock, J.A. Clement, and C. Spencer. 1998. Using aerationfor corrosion control. J. AWWA. 90:3: 74-88.

Lyman, W.S., A. Cohen, and J.R. Myers. 1982. Causes and Prevention ofPitting in Copper Plumbing Systems in the USA. Proceedings of theInternational Symposium on Corrosion of Copper and Copper Alloys inBuildings, Japan Copper Development Association, Tokyo, 1982.

29/10/2001 77

Maggi. C., 1999, Desarrollo de un modelo para estimar la exposición humanaal cobre contenido en el agua potable, M.Sc. thesis, School of Engineering,Catholic University of Chile, Santiago, Chile.

Mahapatra S., and R.Subrahmanya, 1967. Studies in the Hydrolysis of MetalIons, Part I: Copper, Proc. Indian Acad. Sci. 65:283-290.

Mattsson, E., & A.M. Fredriksson. 1968. 1968. Pitting Corrosion in CopperTubes-Cause of Corrosion and Counter Measures, Br. Corrosion J., 3: 246.

Mattsson E. 1988. Counteraction of pitting in copper water pipes - causes andcountermeasures. Werkstoffe und Korrosion. 39: 499-503.

Mattsson E., 1990. Copper and Brass for Plumbing: a Guide for CorrosionPrevention", Proceedings of Copper 90 Refining, Fabrication, Market, October1990, London: The Institute of Metals.

ME. Denmark 1998. Statutory Order from the Ministry of the EnvironmentNumber 515 of August 29, 1998, on Water Quality Supervision of WaterSupply Plants., Denmark..

Meyer, E., 1996. Determinants of Copper Intake from Water. NIH - Universityof Chile Conference, Genetic and Environmental Determinants of CopperMetabolism., 18-20 March, 1996, Washington D.C.

Meyer, T.E. & M. Edwards. 1994. Role of Bicarbonate in copper corrosion, inCritical Issues in water and Wastewater Treatment, Proceedings of the 1994ASCE National Conference on Environmental Engineering. Braun-BrumfieldPublishers. Greenwich, CT., USA.

Millet, B., C. Fiaud, C. Hinnen and E.M.M. Sutter, 1995. A correlationbetween electrochemical behaviour, composition and semiconductingproperties of naturally grown oxide films on copper. Corr. Sc., 37(12): 1903-1918.

Mittelman M., 1994. Corrosion Workshop Organised by the InternationalCopper Association, ICA, September 1994, London, U.K.

Mittelman M.W., D. Davidson. 1994. Influence of Bacterial BiofilmCommunity Structure and Activity Mobilization of Copper in Domestic WaterSystems, Department of Microbiology, Faculty of Medicine, University ofToronto, Interim Report of Project 484-94, July 1, 1994.

29/10/2001 78

Moss G., and E.C. Potter. 1984. Investigation in the interactions between coldpotable water and copper pipes. CSIRO Restricted Investigation Report 1534R,June 1984, Australia, 72 pgs.

Nielsen K, 1983. Control of Metal Contaminants in Drinking Water inDenmark, Aqua, No 4, pgs 182-182.

Nielsen, K., 1995. Copper release from pipes in high alkalinity water”, inInternal Corrosion in Water Distribution Systems, Proceed. InternationalSeminar, Chalmers University of Technology, Goteborg, Sweden, May 22-24,1995, pgs. 207-217.Nishikata, A., M. Itagaki, T. Tsuru, and S. Haruyama. 1990. Passivation and itsStability on Copper in Alkaline Solutions Containing Carbonate and ChlorideIons. Corrosion Science, 31: 287-292.

O'Connell, W.J. 1941. Characteristics of Microbial Deposits in Water Circuits.Proc. American Petroleum Institute, 11th Mid Year Meeting, 1941, 22(3):66-83.

Page, G.G., 1972. Copper Corrosion: Discussion on Blue Water. MaterialsPerformance. 11(2):53.

Page G.G., P.C.A. Bailey and G.A. Wright. 1974. Mechanisation of NewType of Copper Corrosion in Water. Australasian Corrosion Engineering,(Nov-Dec) pgs. 13-19.

Paradies, H.H., I. Hanbel, W. Fischer,and D.Wagner, 1990. Microbiologicallyinduced corrosion on copper pipes. INCRA Project number 404, July 1990.

Paulson, A. J. & D.R. Kester. 1980. Copper (II) Ion Hydrolysis in AqueousSolution. Journal of Solution Chemistry, 9:4:269.

Potter E.C. 1969. An Investigation of the Green-Water Problem in Auckland,New Zealand and a Discussion of Possible Remedies. CSIRO Report, Sydney,Australia.

Rehring, J.P., and M. Edwards. 1996. Effect of natural organic matter andwater treatment processes on copper corrosion. Corrosion, V.52, No. 4, pgs307-317.

29/10/2001 79

Reiber S., 1989. Copper plumbing surfaces: an electrochemical study.J.AWWA. 81(7):114.

Riedl R., and J Klimbacher, 1989. Pitting corrosion in copper water pipes,Int.J. of Materials and Product Technology, vol. 4, No 2, 159-166.

Royuela, J.J., and E. Otero. 1993. The assessment of short term data of pipecorrosion in drinking water – II. Copper. Corrosion Science. 34(10): 1595-1606.

Royuela J.J., and E. Otero, 1994. Corrosion Inhibition in Drinking Water.Effect of Temperature. Part 2. Copper, Rev. Metal. Madrid, 30(2), pgs 92-99.

Sander, A., A Elfstrom Broo, B. Berghult, T. Hedberg, and E. Lind Johanson.1995. The influence of water quality on corrosion of iron and copper pipematerials. In Internal Corrosion in Water Distribution Systems, Proceed.International Seminar, Chalmers University of Technology, Goteborg, Sweden,May 22-24, 1995, pgs. 105-110.

Sauter W., E. Meyer, and O. von Franque, 1995. Operation of a Test Rig toAssess the Suitability of Alloys in Different Types of Water, Proceed.International Seminar, Chalmers University of Technology, Goteborg, Sweden,May 22-24, 1995, pgs. 207-217.

Schecher, W.D., and D.C., McAvoy. 1998. MINEQL+, A chemicalequilibrium modeling system for PCs, version 4.0 for Windows. The Procterand Gamble Company, Environmental Research Software, Hallowell, Maine,USA.

Schindler, P.W. 1967. Heterogeneous Equilibria Involving Oxides,Hydroxides, Carbonates and Hydroxide Carbonates. In Equilibrium Conceptsin Natural Water Systems, Adv. Chem. Ser., No 67, American ChemicalSociety, Washington DC, p 196.

Schock, M.R., D.A. Lytle, J.A. Clement, 1985. Effect of pH, DIC,Orthophosphate and Sulfate on Drinking Water Cuprosolvency. EPA/600/R-95/085, June 1985, USEPA.

Schock, M.R., & D.A. Lytle. 1994. The importance of stringent control ofDIC and pH in Laboratory Corrosion Studies: Theory and Practice.

29/10/2001 80

Proceedings of the AWWA Water Quality Technology Conference, SanFrancisco, Ca., Nov., 1994.

Schock, M.R., D.A. Lytle & J.A. Clement. 1994. Modeling Issues of CopperSolubility in Drinking water", Proceedings ASCE National Conference onEnvironmental Engineering, Boulder, Colorado, July, 1994.

Schock, M.R., D.A. Lytle & J.A. Clement. 1995a. Effect of pH, DIC,Orthophosphate and sulfate on Drinking water Cuprosolvency". U.S.Environmental Protection Agency, EPA/600/R-95/085, Office of Research andDevelopment, Cincinnati, OH.

Schock, M.R., D.A. Lytle., and J.A. Clement, 1995b. Effects of pH, carbonate,orthophosphate, and redox potential on cuprosolvency. The NACE, Corrosion95, paper N 610, 29 pages.

Shalaby, H.M., Al-Kharafi and Gouda, V.K. 1989. A Morphological Study ofPitting Corrosion of Copper in Soft Tap Water. Corrosion. 45: (7): 536-547.

Shalaby, H.M., F.M. Al-Kharafi and A.J. Said. 1990. Corrosion Morphologyof Copper in Dilute Sulfate, Chloride and Bicarbonate Solutions. Br. CorrosionJ. 25(4): 292-298.

Shuey, R.T. 1975. Semiconducting ore minerals. New York: Elsevier. Pg. 349.Shull, K.E. and Becker R.J. 1960. Cold Water Corrosion of copper tubing. J.AWWA. 52(8):1022.

Smith, S., And Francis, R. 1990. Use of Electrochemical current noise to detectinitiation of pitting conditions on copper tubes. Br. Corrosion J. 25(4): 285-291.

Stumm, W.S; & J.J. Morgan, 1996. Aquatic Chemistry - Chemical Equilibriaand Rates in Natural Waters. John Wiley & Sons, Inc., New York (ThirdEdition).

Sumitomo Light Metal Industries, Ltd., 1994., “Super Tin Coat Copper TubeData Sheet”, Technical Research Laboratories, Japan, September 1994.Fontana, M.G., and N.D. Greene. 1978. Corrosion Engineering. New York:McGraw-Hill.

29/10/2001 81

Taylor, R.J., and Cannington P.H. 1993. Control of Pitting Corrosion ofCopper tubes in Potable Waters", Melbourne: Urban Research Association ofAustralia. August, 1993.

Tunturi, P., and S. Ylasaarl. 1968. "A Special Case of the Pitting Corrosion ofCopper in a Hot Water System", 5th Scandinavian Corrosion Congress,Copenhaguen, Denmark, 1968.

USEPA, 1985, National primary drinking water regulations, EnvironmentalProtection Agency., 40 CFR Part 141. Fed. Reg. 50:46967.

USEPA, 1991a, Maximum contaminant level goals and national primarydrinking water regulations for lead and copper; final rule., 40 CFR Parts 141and 142. Fed. Reg. 56:110.

USEPA, 1991b, Maximum contaminant level goals and national primarydrinking water regulations for lead and copper; final rule., 40 CFR Parts 141and 142. Fed. Reg. 56:110.

USEPA, 1994, CFR, Vol. 59, No 125, Part V, 59 FR 33860, Thursday, June30.1 Wyllie, J., 1957, Copper poisoning at a cocktail party. Am. J. Public Health,number 47, p. 617.

Van den Hoven, Th., H. Brink and N. Slaats. 1995. Copper corrosion and theenvironmental consequences. In Internal Corrosion in Water DistributionSystems, Proceed. International Seminar, Chalmers University of Technology,Goteborg, Sweden, May 22-24, 1995, pgs. 53-57.

Van Den Hoven, Th., M.W.M. Van Eekeren, 1998., Optimal Composition ofDrinking Water. Kiwa Report 100. Groningenhaven 7, PO Box 1072, 3430 BBNieuwegein, The Netherlands.

Wagner, D., W. Fischer, and G. J. Tuschewitzki. 1992. MicrobiologicallyInduced Pitting Corrosion of Copper Pipes. Final Report ICA Project number453, 1992.

Wagner D., Peinemann H., Siedlarek H., Fischer W.R., Arens P., andTuschewitzki G.J., " Microbiological Influenced Pitting Corrosion of CopperPipes", Final Report, ICA Project number 453-A, 1993.

29/10/2001 82

Walsh L.H., and N.B. Feilchenfield, 1995. Characterization of Films FormedDuring Corrosion of Copper in Dilute Solutions of Cupric Chloride,Mat.Res.Soc.Symp. Proceedings Vol 390, Materials Research Sciety, USA,pgs 117-122.

Werner W, H.J Grob, M. Gerlach, D Horvath, and H. Sontheimer, 1994.Untersuchungen zur Flachekorrosion in Trinkwasserleitungen aus Kupfer,GWF, Gas – Wasserfach: Wasser/Abwasser 135, 2, pgs 92-103

Werner, W. 1995. Uniform copper corrosion. In Internal Corrosion in WaterDistribution Systems, Proceed. International Seminar. Chalmers University ofTechnology: Goteborg, Sweden, May 22-24, 1995, pg. 95-103.

WHO, 1993, Guidelines for Drinking Water Quality, Volume 1.

WHO, 1996, Guidelines for Drinking Water Quality, Volume 2.

WHO, 1998, Guidelines for drinking-water quality, second edition, addendumto volume 2, health criteria and other supporting information, WHO/EOS/98-1.

Wilhelm, S.M., Tanizawa Y., Liu C., Hackerman N., 1982. A photo-electrochemical investigation of semiconducting oxide films on copper.Corrosion Science. 22(8): 791-805.

Yamauchi, S., K. Nagata, and S. Sato, 1986. Dissolution from Copper WaterTubes and New Copper Alloys showing Anti-Cuprosolvency. Report presentedto the International Copper Research Association, NY: INCRA.