16593 17 tủ tài liệu bách khoa

Plumbing material Corrosion resistance Primary contaminants from pipeCopper Good overall corrosion resistance; subject to corrosive Copper and possibly iron, zinc, tin, antimony, arsenic

CHAPTER 17 INTERNAL CORROSION AND

Michael R Schock, Chemist

U.S Environmental Protection Agency Water Supply and Water Resources Division

Cincinnati, Ohio

Corrosion is one of the most important problems in the drinking water industry Itcan affect public health, public acceptance of a water supply, and the cost of provid-ing safe water Deterioration of materials resulting from corrosion can necessitatehuge yearly expenditures of resources for repairs, replacement, and system Manytimes the problem is not given the attention it needs until expensive changes orrepairs are required

Corrosion tends to increase the concentrations of certain metals in tap water.Two potentially toxic metals (lead and cadmium) are attributable almost entirely toleaching caused by corrosion Three other metals—copper, iron, and zinc—causestaining of fixtures, or metallic taste, or both Low levels of tin and antimony can becaused by the corrosion of lead-free solders (Herrera, Ferguson, and Benjamin,1982; Subramanian, Connor, and Meranger, 1991; Subramanian, Sastri, and Connor,1994) Nickel has sometimes been mentioned as a potential contaminant from theplating of decorative plumbing fixtures The promulgation of the Lead and CopperRule by the U.S Environmental Protection Agency (USEPA) in 1991 has created anemphasis on corrosion control in distribution systems, as well as domestic, public,

and institutional plumbing systems (Federal Register, 1991a,b, 1994a).

Corrosion products attached to pipe surfaces or accumulated as sediments in thedistribution system can shield microorganisms from disinfectants (see Chapter 18).These organisms can reproduce and cause problems such as bad tastes, odors,slimes, and additional corrosion Several researchers have recently promoted cor-rosion control within the distribution system as an effective way to maintain waterquality and adequate disinfection (Rompré et al., 1996; Schreppel, Frederickson,and Geiss, 1997; Camper, 1997; Kiéné, Lu, and Lévy, 1996; Norton et al., 1995;Olson, 1996)

17.1

1 The views expressed in this paper are those of the author and do not necessarily reflect the views or policies of the U.S Environmental Protection Agency.

The term corrosion is also commonly applied to the dissolution of cement-based

materials, and the leaching of their free lime component The most common festation of this problem is the increase in pH, which can be detrimental to disinfec-tion and the aesthetic quality of the water, as well as reducing the effectiveness ofphosphate corrosion inhibitor chemicals intended to control the corrosion of metals.The release of asbestos fibers is of regulatory concern, and in extreme cases, thechemical attack on the pipe by the water may cause a reduction of structuralintegrity and, ultimately, failure

mani-Even when a water system passes all regulatory requirements, the release of rosion by-products by miles of distribution system and domestic piping, and theapplication of corrosion inhibitor chemicals containing metals such as zinc, can besignificant sources of metal loading of wastewater treatment plants This contami-nation source can affect their ability to meet discharge or sludge disposal limits.Phosphate-based corrosion inhibitors can provide unwanted nutrients to waste-water plants and can cause violations of wastewater or other discharge regulations,

cor-or water quality problems in ecosystems receiving the water

Corrosion-caused problems that add to the cost of water include the following:

1 Increased pumping costs caused by tuberculation and hydraulic friction

2 Loss of water and water pressure caused by leaks

3 Water damage to the dwelling, requiring that pipes and fittings be replaced

4 Replacing hot water heaters

5 Customer complaints of “colored water,” “stains,” or “bad taste,” for which the

response may be expensive both in terms of money and public relations

6 Increased wastewater and sludge treatment and disposal costs

7 Increased dosage of chlorine to maintain a distribution system residual

Corrosion is the deterioration of a substance or its properties because of a reactionwith its environment In the waterworks industry, the “substance” that deterioratesmay be a metal pipe or fixture, the cement mortar in a pipe lining, or an asbestos-cement (A-C) pipe For internal corrosion, the “environment” of concern is water.All waters are corrosive to some degree A water’s corrosive tendency willdepend on its physical and chemical characteristics Also, the nature of the materialwith which the water comes in contact is important For example, water corrosive togalvanized iron pipe may be relatively noncorrosive to copper pipe in the same sys-tem Corrosion inhibitors added to the water may protect a particular material, butmay either have no effect or may be detrimental to other materials

Physical and chemical interactions between pipe materials and water may causecorrosion An example of a physical interaction is the erosion or wearing away of apipe elbow from high flow velocity in the pipe An example of a chemical interaction

is the oxidation or rusting of an iron pipe Biological growths in a distribution system(Chapter 18) can also cause corrosion by providing an environment in which physi-cal and chemical interactions can occur The actual mechanisms of corrosion in awater distribution system are usually a complex and interrelated combination ofthese physical, chemical, and biological processes They depend greatly on the mate-rials themselves, and the chemical properties of the water The purpose of this chap-ter is to provide an introduction to the concepts involved in corrosion anddeposition phenomena in potable waters

Each material that can corrode has a body of literature devoted to it Detail onthe form of corrosion of each metal or piping material and specific corrosion inhibi-

tion practices that might be employed can be found in a comprehensive text(AWWARF-TZW, 1996; Trussell, 1985; Snoeyink and Kuch, 1985) and water treat-ment journal articles Table 17.1, modified slightly from the original source (Singley,Beaudet, and Markey, 1984; AWWA, 1986), briefly relates various types of materials

to corrosion resistance and the potential contaminants added to the water In eral, plastic plumbing materials are more corrosion-resistant, but they are not with-out their own potential problems

gen-CORROSION, PASSIVATION, AND IMMUNITY

Electrochemical Reactions

Metal species can be released into water either from the simple dissolution of ing scale materials, or actual electrochemical corrosion followed by dissolution Insome cases, scale materials formed from corrosion by-products may be eroded fromthe pipe surfaces Almost all mineral salts dissolve in water to some extent, frominsignificant traces to high concentrations in seawater This section will provide ageneral overview of some aspects of the electrochemistry of metallic corrosion as itapplies in the context of drinking water treatment However, many specialized texts

exist-on electrochemistry and electrochemical corrosiexist-on are available (Pirexist-on, 1991; Ailor,1970; Bockris and Reddy, 1973; Butler and Ison, 1966; NACE, 1984; Pourbaix, 1966,1973; Thompson, 1970) and should be consulted by readers who are interested in acomprehensive examination of the subject

For corrosion of any type to occur, all of the components of an electrochemical

cell must be present These include an anode, a cathode, a connection between the anode and cathode for electron transport, and an electrolyte solution that will con-

duct ions between the anode and cathode The anode and cathode are sites on themetal that have different electrical potential Differences in potential may arisebecause metals are not completely homogeneous If any one of these components isabsent, a corrosion cell does not exist and corrosion will not occur

Oxidation and dissolution of the metal takes place at the anode The electronsgenerated by the anodic reaction migrate to the cathode, where they are discharged

to a suitable electron acceptor, such as oxygen The positive ions generated at theanode will tend to migrate to the cathode, and the negative ions generated at thecathode will tend to migrate to the anode Migration occurs as a response to the con-centration gradients and to maintain an electrically neutral solution

At the phase boundary of a metal in an electrolyte solution an electrical tial difference exists between the solution and the metal surface This potential is theresult of the tendency of the metal to reach chemical equilibrium with the elec-trolyte solution This oxidation reaction, representing a loss of electrons by themetal, can be written as

Equation 17.1 indicates that the metal corrodes, or dissolves, as the reaction goes tothe right This reaction will proceed until the metal is in equilibrium with the elec-trolyte containing ions of this metal

The current that results from the oxidation of the metal is called the anodic

cur-rent In the reverse reaction, the metal ions are chemically reduced by combining

with electrons.The current resulting from the reduction (the reaction, Eq 17.1, going

Plumbing material Corrosion resistance Primary contaminants from pipe

Copper Good overall corrosion resistance; subject to corrosive Copper and possibly iron, zinc, tin, antimony, arsenic,

attack from high flow velocities, soft water, chlorine, cadmium, and lead from associated pipes and dissolved oxygen, low pH, and high inorganic carbon solder

levels (alkalinities) May be prone to “pitting” failures

Lead Corrodes in soft water with pH <8, and in hard waters with Lead

high inorganic carbon levels (alkalinities) and pH below

∼7.5 or above ∼8.5

Mild steel Subject to uniform corrosion; affected primarily by high Iron, resulting in turbidity and red-water complaints

high dissolved oxygen and chlorine levels, and poorly buffered waters

Cast or ductile iron Can be subject to surface erosion by aggressive waters Iron, resulting in turbidity and red-water complaints.(unlined) and tuberculation in poorly buffered waters

Galvanized iron or steel Subject to galvanic corrosion of zinc by aggressive waters, Zinc and iron; cadmium and lead (impurities in

especially of low hardness; corrosion is accelerated by galvanizing process)

contact with copper materials; corrosion is accelerated at higher temperatures as in hot-water systems; corrosion is affected by the workmanship of the pipe and galvanized coating

Asbestos-cement, concrete, Good corrosion resistance; immune to electrolysis; Asbestos fibers; increase in pH, aluminum, and calcium.cement linings aggressive (soft) waters can leach calcium from cement;

polyphosphate sequestering agents can deplete the calcium and substantially soften the pipe

Brass Good overall resistance; different types of brass respond Lead, copper, zinc

differently to water chemistry; subject to dezincification bywaters of pH >8.3 with high ratio of chloride to carbonate hardness Conditions causing mechanical failure may not directly correspond to those promoting contaminant leaching

* Source: Adapted from Singley, J E., B A Beaudet, and P H Markey, “Corrosion manual for internal corrosion of

water distribution systems,” U.S Environmental Protection Agency, EPA/570/9-84-001 Prepared for Office of Drinking

Water by Environmental Science and Engineering, Inc., Gainesville, FL, 1984.

to the left) is called the cathodic current At equilibrium, the forward reaction

pro-ceeds at the same rate as the reverse reaction, and the anodic current is equal to thecathodic current Thus, no net corrosion is occurring at equilibrium The velocity of

an electrochemical reaction, unlike that of a normal chemical reaction, is stronglyinfluenced by the potential itself

Corrosion results from the flow of electric current between electrodes (anodicand cathodic areas) on the metal surface These areas may be microscopic and invery close proximity, causing general uniform corrosion Alternatively, they may belarge and somewhat remote from one another, causing pitting, with or withouttuberculation Electrode areas may be induced by various conditions, some because

of the characteristics of the metal and some because of the character of the water atthe boundary surface Especially significant are variations in the composition of themetal or the water from point to point on the contact surface Impurities in the metal, sediment accumulations, adherent bacterial slimes, and accumulations ofthe products of corrosion are all related either directly or indirectly to the develop-ment of electrode areas for corrosion circuits

Figure 17.1 shows an example of corrosion reactions taking place on a fresh pipesurface with proximate anodic and cathodic areas (Snoeyink and Jenkins, 1980) Inalmost all forms of pipe corrosion, the metal goes into solution at the anodic areas

As the metal dissolves, a movement of electrons occurs and the metal develops anelectric potential Electrons liberated from the anodic areas flow through the metal

to the cathodic areas where they become involved in another chemical reaction, andthe metal develops another electric potential The focus of corrosion control bywater treatment methods is usually attempting to retard either or both of the pri-mary electrode reactions

The Nernst Equation

The Nernst equation is a relationship that allows the driving force of the reaction to

be computed from the difference in free energy levels of corrosion cell components.The free energy difference under such conditions depends on the electrochemicalpotential, which, in turn, is a function of the type of metal and the solid- and aqueous-phase reaction products Electrons (electricity) will then flow from certain areas of ametal surface to other areas through the metal A metal may go into solution as anion, or may react in water with another element or molecule to form a complex, anion pair, or insoluble compound

FIGURE 17.1 Adjoining anodes and cathodes during the corrosion of iron in

acidic solution (Source: Water Chemistry, V L Snoeyink and D Jenkins

Copy-right © 1980, John Wiley & Sons, Inc Reprinted by permission of John Wiley &

Sons, Inc.)

The equilibrium potential of a single electrode can be calculated by using theNernst equation for the general reaction (Eq 17.1):

{ } denote activity of the ion Me z+

R=the ideal gas constant (about 0.001987 Kcal/deg⋅mol− 1)

T=the absolute temperature (°K)

F=the Faraday constant (23.060 Kcal/V)

z=the number of electrons transferred in the reaction

The Me/Me z+subscript indicates the reaction written as a reduction The Nernst tion, written as Eq 17.3, is for a single electrode, assuming that the electrode is cou-pled with the normal hydrogen electrode, at which the reaction

equa-2H++2e−→H2(g) E0=0.00 (17.3)takes place, and reactants and products are assumed by thermodynamic convention

to equal 1 (Stumm and Morgan, 1981; Garrels and Christ, 1965) The driving forcecomputed by the Nernst equation is directly related to the Gibbs free energy for theoverall reaction, through the relationship ∆Gr0 = −zFE, where ∆Gr0 is the free-energy change for the complete reaction (Snoeyink and Jenkins, 1980; Stumm andMorgan, 1981) In this convention, standard “half-cell” potentials are tabulated withthe reactions written as reductions

It is usually more useful to use the Nernst equation in a general form for the anced net reaction of two half-cells, each being of the form

where Q is the reaction quotient ({red}/{ox}) At equilibrium, no electrochemical

current is generated, and the oxidants and reductants are at their equilibrium

activ-ities Thus, the reaction quotient Q becomes equal to the equilibrium constant K for

the overall reaction (Snoeyink and Jenkins, 1980)

In drinking water systems, the oxidation half-cell reaction of a metal, such as iron,zinc, copper, or lead, is coupled with the reduction of some oxidizing agent, such asdissolved oxygen or chlorine species Example half-cell reactions are the following:

0.0591

{red}

{ox}

O2+2H2O +4e−⇔4OH− (17.7)HOCl +H++e−⇔(1/2) Cl2(aq) +H2O (17.8)

1⁄2 Cl2(aq) +e−⇔Cl− (17.9)Equations 17.8 and 17.9 can be combined to yield the net reaction:

HOCl +H++2e−⇔Cl−+H2O (17.10)which represents a significant oxidizing half-cell reaction for metals in drinkingwater

If Eq 17.7 is written in the Nernst form (Eq 17.5), and the ionic strength of thewater is low enough that it can be assumed to have unit activity ({H2O} =1), then

EO2/OH−=E0

The oxidation potential for this reaction clearly depends on pH, because the [OH−]

is raised to the fourth power in the numerator Similarly, the oxidation potential ofthe hypochlorous acid reaction is directly related to pH:

EHOCl/Cl−=E0

because of the [H+] term in the denominator

By thermodynamic definition, the corrosion (and, hence, dissolution of metalsfrom plumbing materials) can only occur if the overall cell potential exceeds theequilibrium cell potential (Pourbaix, 1973; Snoeyink and Jenkins, 1980; Stumm andMorgan, 1981)

Reactions such as Eq 17.7 and Eq 17.10 can be combined with metal oxidationhalf-cells (Eq 17.1) to show overall corrosion reactions likely to occur in drinkingwater Examples are

2Pb (metal) +O2+2H2O ⇔2Pb2 ++4OH− (17.13)2Fe (metal) +O2+2H2O ⇔2Fe2 ++4OH− (17.14)

0.0591

4

the materials, where “fresh” metal is available to oxidize Anodic and cathodic trode areas may be induced by various conditions, some because of the characteris-tics of the metal, and some because of the character of the water at the boundarysurface Impurities in the metal, sediment accumulations, adherent bacterial slimes,and accumulations of the products of corrosion are all related in some way to thedevelopment of electrode areas that can enable the operation of corrosion circuits.

elec-Corrosion Products on Pipe Surfaces

Metal surfaces may be protected either by their being “immune” or by rendering

them “passive.” If a metal is protected by immunity, the metal is thermodynamically

stable, and is therefore incorrodible (Pourbaix, 1973) For some metals, such as per, this can occur in groundwaters that are somewhat anoxic (Lytle et al., 1998).Sometimes, this region of electrochemical behavior is only possible when water itself

cop-is not chemically stable, so it cop-is only encountered in potable water systems when the

consumption of externally supplied energy (cathodic protection) occurs.

Passivation occurs when the metal is not stable, but becomes protected by a

sta-ble film The protection can be perfect or (more usually) imperfect, depending uponwhether the film effectively shields the metal from contact with the solution (Pour-baix, 1973) True passivation films must satisfy several requirements to effectivelylimit corrosion Particularly, they must be electrically conductive, mechanically sta-ble (neither flaking nor cracking), and continuous

Analysis of corrosion problems is complicated by the variety of chemical tions that take place across the surface For example, consider the reactions at aniron or steel surface, in water where oxygen is the only oxidant, and aqueous ironcomplexation is negligible The primary reaction occurring at the anodic sites is:

2FeCO3(s) +1/2 O2+H2O ⇔2FeOOH(s) +2CO2 (17.22)3FeCO3(s) +1/2 O2⇔Fe3O4(s) (magnetite) +CO2 (17.23)Reactions such as Eqs 17.21 to 17.23 can reduce the rate of oxygen diffusing to theanode, thus the formation of oxygen concentration cells Other reactions that affectcorrosion may take place, depending on the composition of the water and the type

of metal

At the same time as the anodic reactions are taking place, a variety of cathodicreactions may be occurring Perhaps the most common in drinking water distribu-tion systems is the acceptance of electrons by O2

e−+1/4 O +1/2 HO ⇔OH− (17.24)

This reaction causes an increase in pH near the cathode and triggers the ing additional reactions:

follow-OH−+HCO3 −⇔CO3 −+H2O (17.25)

Ca2 ++CO3 −⇔CaCO3(s) (17.26)These reactions can cause CaCO3(s) to precipitate from some waters in which thebulk solutions are undersaturated with this solid, because the pH increase in thevicinity of the cathode forms enough CO3 −to cause supersaturation with respect toCaCO3(s)

Several studies have shown that the pH at the surface of pipe can be significantlydifferent from that in the bulk solution (Snoeyink and Wagner, 1996), althoughmany studies reporting extremely high pHs at the surface have neglected to ade-quately include consideration of buffering by the carbonate system in the water, andthe possible role of solids such as CaCO3(s) and Mg(OH)2(s) (Dexter and Lin, 1992;Lewandowski, Dickinson, and Lee, 1992; Watkins and Davies, 1987) The depositsthat form on pipe surfaces may be (1) a mixture of corrosion products that dependboth on the type of metal that is corroding and the composition of the water solution[e.g., FeCO3(s), Fe3O4(s), FeOOH(s), Pb3(CO3)2(OH)2(s), Zn5(CO3)2(OH)6(s)];(2) precipitates that form because of pH changes that accompany corrosion [e.g.,CaCO3(s)]; (3) precipitates that form because the water entering the system is super-saturated [e.g., CaCO3(s), SiO2(s), Al(OH)3(s), MnO2(s)]; and (4) precipitates orcoatings that form by reaction of components of inhibitors, such as silicates or phos-phates with the pipe materials (e.g., lead or iron)

The nature of the scales or deposits that form on metals is very importantbecause of the effect that these scales have on the corrosion rate The formation ofscales, such as CaCO3(s) and iron carbonates on corroding iron or steel, are normallythicker and have higher porosity than the passivating films Deposits and scales donot decrease the corrosion rate as much as true oxide films do, and the same corro-sion current-potential relationship for passivating films does not occur for suchscales The complex interactions can be illustrated by the case with steel corrosion.Scale formation on steel by minerals such as calcium carbonate reduces the corro-sion rate by decreasing the rate of oxygen transport to the metal surface, therebydecreasing the rate of the cathodic reaction Passivating iron oxide films on steelcause an anodic-controlled corrosion reaction

A very long time, from many months to years, may be required for the corrosionrate of iron and steel to stabilize because of the complex nature of the scales.A muchshorter time may be sufficient for other metals If a scale reduces the rate of corro-

sion, it is said to be a protecting scale; if it does not, it is called nonprotecting The importance of scale is also demonstrated by the phenomenon of erosion cor-

rosion, observed at points in the distribution system or in domestic plumbing

sys-tems where a high-flow velocity or an abrupt change in direction of flow exists Themore intense corrosion that often is observed at such locations can be attributed tothe abrasive action of the fluid (caused by turbulence, suspended solids, and soforth) that scours away or damages the scale, and to the velocity of flow that carriesaway corrosion products before they precipitate and that facilitates transport of cor-rosion reactants more efficiently (Snoeyink and Wagner, 1996)

Changes in water treatment or source water chemistry over time can producesuccessive layers of new solid phases, remove or change the nature of previouslyexisting deposits, or both Figure 17.2 shows an example of the complex nature ofscale on a cast-iron distribution pipe (Singley et al., 1985; Benjamin, Sontheimer, andLeroy, 1996) Scales of similar chemical composition can have a significantly differ-ent impact on corrosion and metal protection because properties such as uniformity,

adherability, and permeability to oxidants can vary depending upon such factors astrace impurities, presence of certain organics, temperature of deposition, length oftime of formation, and so forth.

Scales that form on pipes may have deleterious effects in addition to the cial effect of protecting the metal from rapid corrosion or limiting the levels of toxicmetals (such as lead) in solution Water quality should be controlled so that the scale

benefi-is protective but as thin as possible, because as bulk of scale increases, the capacity

of the main to carry water is reduced The formation of uneven deposits such astubercles increases the roughness of the pipe surface, reducing the ability of themains to carry water, and may provide shelters for the growth of microorganisms

To properly interpret field and laboratory data from corrosion control studies, it

is important to understand that there may be significantly different reactions ring between the water constituents and the surfaces of “new” pipes compared with

occur-“old” pipes Conceptually, this is illustrated in Figure 17.3 for lead pipe On the new

surface [Figure 17.3(a)], the full corrosion reaction can occur, with oxidation of lead

followed by the development of a passivating film Once the film is sufficiently

developed [Figure 17.3(b)], the oxidants in the water no longer can directly contact

the metal of the pipe material itself Therefore, the oxidation step will not occur, andmetal release will become a function of the physical adherence or the solubility ofthe surface deposit, unless water conditions become anoxic and the metal(s) in thesurface deposit become electrochemically reduced Thus, corrosion inhibitor chemi-cals that stifle reactions occurring at cathodic surface sites may appear much better

in tests using new metal surfaces than they may operate when applied to distributionsystem pipes covered by thick scales or corrosion deposits With well-developed sur-face deposits present, the solubility and surface sorption chemistry of the existingscales is much more important in developing water treatment targets than predic-tions based on the pure corrosion chemistry of the metal

FIGURE 17.2 Schematic of scale on a cast-iron distribution

pipe, showing complex layered structure (Source: Internal

Cor-rosion of Water Distribution Systems, 2nd ed., American Water

Works Association Research Foundation, Denver, CO, 1996.)

Corrosion Kinetics

A three-step process is involved in governing the rate of corrosion of pipe: (1) port of dissolved reactants to the metal surface, (2) electron transfer at the surface,and (3) transport of dissolved products from the reaction site (Trussell, 1985) When

trans-either or both of the transport steps are the slowest, rate-limiting step, the corrosion reaction is said to be under transport control When the transfer of electrons at the metal surface is rate-limiting, the reaction is said to be under activation control The

formation of solid natural protective scales that inhibit transport are often an tant factor in transport control This section will only present an overview of the con-cepts involved in the rates of corrosion reactions in potable water systems Numerousreference articles and texts exist that present a detailed development of the theoriesthat are the basis for many direct electrochemical rate-measuring techniques.Corrosion is often described in terms of numerous tiny galvanic cells on the sur-face of the corroding metal Such localized anodes and cathodes as those describedare not fixed on the surface, but are statistically distributed on the exposed metalover space and time The electrochemical potential of the surface is determined bythe mixed contributions to potential of both the cathodic and anodic reactions, aver-aged over time and over the surface area Both the individual anodic and cathodichalf-reactions are reversible and occur in both directions at the same time When the

impor-“electrode” is at its equilibrium, the rates of reaction in both the cathodic and anodichalf-cells are equal

Given the thermodynamic basis for corrosion described above, and the body ofknowledge about kinetic factors that affect the rate of corrosion of metals, severalproperties of the water passing through a pipe or device that influence the rate ofcorrosion can be identified Some of the water-related properties are: (1) concentra-tion of dissolved oxygen, (2) pH, (3) temperature, (4) water velocity, (5) concentra-tion and type of chlorine residual, (6) chloride and sulfate ion concentration, and (7)concentration of dissolved inorganic carbon (DIC) and calcium

These properties interrelate, and their effect depends on the plumbing material

as well as the overall water quality References specific to the type of plumbing ation (pipe, soldered joint, galvanic connection, faucet, or flow-control device) andthe material of interest should be consulted for the most appropriate information oncorrosion rate control Some generalizations will be considered in a later section

situ-New Lead Surface

Both anodic and cathodic sites

Example Anodic site reactions

Old Lead Surface

No anodic and cathodic sites

Pb3(CO3)2(OH)2 + 2PO43– Pb3(PO4)2 + 2CO32– + 2OH –

5[Pb3(CO3)2(OH)2] + 9PO43– 3[Pb5(PO4)3OH] + 10CO32– + 7OH –

FIGURE 17.3 Schematic representation of different surface reactions and their relation to

passi-vation between (a) new and (b) old lead pipe surfaces.

Solubility Diagrams

The solubility of passivating films on the pipe surface is the most important factor indetermining whether a given water quality can meet many drinking water regula-tions that are based on health effects of ingested metals Asbestos fibers are fre-quently released into the water as a result of the dissolution of the cement matrixoriginally holding them Solubility places a lower limit on the level to which metalscan be controlled by modifying water chemistry at the treatment plant Solubilityreflects an ideal equilibrium condition, does not correspond directly to tap waterlevels, and should not be expected to (Schock, 1990a; Britton and Richards, 1981).Other factors are important, such as the physical location of the plumbing materialsrelative to the sample collected, the release of particulates from the deposits on thepipes, the rate of the chemical reactions that affect the mobilization of the metals rel-ative to the standing time of the water before collection, and many other variables.However, it is widely believed that trends in the response of tap water metal levels

to changes in key water chemistry variables usually follow the predictions of bility models, and that has been recently verified in principle by some recent surveys

solu-of over 2500 U.S utilities solu-of all sizes (Edwards, Jacobs, and Dodrill, 1999)

To display solubility relationships in a relatively simple two-dimensional manner,the total solubility of a constituent is plotted as a function of a master variable(Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981, 1996; Bard, 1966) Othersolution parameters can affect solubility such as the ionic strength, or the concen-tration of dissolved species (ligands) that can form coordination compounds or com-plexes with the metal The solubility diagram does not necessarily give theinformation needed to minimize the rate of corrosion, because rate is affected bykinetic parameters that depend on the relative rates of the oxidation, dissolution,diffusion, and precipitation reactions It also cannot predict the ability of a pipe coat-ing to adhere to the pipe surface, or the permeability of a coating to oxidants fromthe water solution or pitting agents It does, however, give important information forestimating how water quality affects attainment of maximum contaminant levels(MCLs) or action levels (ALs) for drinking water, the potential for precipitatingpassivation films on pipe surfaces, or the deposition of other solids important inwater treatment, such as calcium carbonate, octacalcium phosphate, and aluminum

or ferric hydroxide

To construct this type of diagram, an aqueous mass balance equation must be

written for the metal (or other constituent of interest), with the total solubility (ST)

as the unknown The mass balance expression should include the concentration ofthe uncomplexed species (free-metal ion), along with all complexes to be included inthe model For lead (II), the simple relationship for the free ion, plus only thehydroxide and carbonate complexes (the simplest system) is:

ST,Pb(II),CO3=[Pb2 +] +[Pb(OH)2] +[Pb(OH)3 −] +[Pb(OH)4 −] +2[Pb2(OH)3 +]

+ 3[Pb3(OH)4 +] +4[Pb4(OH)4 +] +6[Pb6(OH)8 +] +[PbHCO3 +]

+ [PbCO3] +[Pb(CO3)2 −] (17.27)

If the concentration of a complex is going to be relatively negligible compared withthe total solubility across the pH range of interest, it can be excluded from the modelfor simplicity Frequently, only four or five species are significant in a system Anexample of two such complete mass balance expressions are the following for lead(Schock, 1980, 1981b) and copper (Schock, Lytle, and Clement, 1995a,b) systems

only containing carbonate species in addition to hydrolysis species, which is the imum system composition for potable waters.

In these equations,β’m,h,crepresents the formation constant for the complex having a

stoichiometry of m metal ions, h hydrogen ions, and c carbonate ions, corrected for

ionic strength and temperature (Stumm and Morgan, 1981, 1996) To obtain the dicted solubility, the solubility constant expression for each solid of interest is rear-ranged and solved to isolate the free species concentration, and is substituted into eachterm of the mass balance equation (e.g., Eq 17.28 or 17.29) Sometimes, an iterativetechnique is used to simultaneously solve several related mass balances, dependingupon exactly the type of diagrams desired and assumptions about the availability ofcomponents involved in the precipitation and dissolution reactions in the system

pre-A diagram is then constructed for each solid, and the curves are superimposed.Thepoints of minimum solubility are then connected, giving the final diagram This proce-dure is discussed in more detail by Schock (1980, 1981b) for lead (II); Schock, Lytle,and Clement for copper (1995a,b); and Snoeyink and Jenkins (1980) for iron (II).When additional aqueous species are present, such as orthophosphate or sulfate,they are just added to the basic mass balance expressions, such as those shownabove For example, the contribution to lead(II) solubility from aqueous orthophos-phate species is represented by:

ST, Pb(II), PO4=[PbHPO4] +[PbH2PO4 +] (17.30)

So, the total lead solubility ST, Pb(II)=ST, Pb(II),CO3+ST, Pb(II), PO4, or

ST, Pb(II)=[Pb2 +] +[Pb(OH)2] +[Pb(OH)3 −] +[Pb(OH)4 −] +2[Pb2(OH)3 +]

+ 3[Pb3(OH)4 +] +4[Pb4(OH)4 +] +6[Pb6(OH)8 +] +[PbHCO3 +]

+ [PbCO3] +[Pb(CO3)2 −] +[PbH2PO4 +] +[PbHPO4] (17.31)Similarly, other ligands (such as chloride and sulfate) can simply be added to theexpressions

Because these diagrams are inherently two-dimensional (solubility on the y-axis,

pH on the x-axis), there are often additional variables that have an important impact

on solubility (such as temperature, ionic strength, carbonate concentration or linity, and orthophosphate concentration), these diagrams must display solubilityand species concentration lines for fixed concentrations or values of the other vari-ables For instance, Figure 17.4 displays a solubility diagram for lead in the system

alka-FIGURE 17.4 A solubility diagram showing dissolved lead (II) species in equilibrium with lead (II) solids in a pure system containing 3 mg C/L DIC (2.5 × 10−4 mol/L) at 25 °C and I= 0.005 mol/L.

(Source: data from Schock and Wagner, 1985.)

containing only carbonate species and water, showing the important aqueousspecies and the stability domains of two lead solids In this figure, the dissolved inor-ganic carbon (DIC) concentration was fixed at 0.00025 mol/L (3 mg C/L), with atemperature of 25°C and an assumed ionic strength of 0.005 Figure 17.5 shows thesame basic system, but with a total DIC concentration of 0.0025 mol/L (30 mg C/L).

In both diagrams, the implicit assumption was also made that the solution redoxpotential was not high enough to cause the formation of Pb(IV) aqueous or solid

FIGURE 17.5 A solubility diagram showing dissolved lead (II) species in equilibrium with lead (II) solids in a pure system containing 30 mg C/L DIC (2.5 × 10−3 mol/L) at 25 °C and I= 0.005 mol/L.

(Source: data from Schock and Wagner, 1985.)

species, or a different diagram would apply for the higher oxidation state species oflead Additionally, when metal solubility becomes high, care must be taken to ensurethat computational constraints on the system, such as the fixed ionic strength andthe total ligand concentration (mass balance), are not violated by the presence ofhigh concentrations of dissolved metal species.

Note that if the same complex formation constants were used, but a less solubleconstant was used for PbCO3(s) (cerussite) solubility, the simple carbonate solidwould be predicted to be stable over a wider pH range Likewise, a less soluble con-stant for Pb3(CO3)2(OH)2(hydrocerussite) would expand its stability field relative

to that of cerussite Analogous information can be obtained through the careful struction and study of solubility diagrams for other metals, such as copper and zinc.When constructing solubility diagrams for any metal, the selection of solid andaqueous species must truly represent the system to be modeled, or very erroneousconclusions can result For example, the aragonite form of calcium carbonate is fre-quently found in deposits formed in systems having galvanized pipe, rather than thecalcite form Also, the ferric iron deposits formed in mains are frequently a relativelysoluble hydroxide or oxyhydroxide form [Fe(OH)3 or FeOOH] rather than anordered form such as hematite (Fe2O3) To go along with this concept, the equilib-rium constants must also be accurate to give realistic concentration estimates, andknowledge of changes in the equilibrium constants with temperature is essential,especially when projections of depositional tendency have to be made into hot orvery cold water-piping systems Critical evaluation of data appearing in handbooksand published papers is necessary to avoid using incorrect values, and occasionallyreview articles or major works by rigorous researchers can be consulted for reliablevalues

con-An important assumption behind the diagrams is that the system must reach modynamic equilibrium for the calculations to be truly valid, unless kinetic factorsare incorporated into the model Sometimes improvements can be made in predic-tions by using metastable species in the calculations, although it is not thermodynam-ically rigorous to do so Metastable solids have been found to govern copper (Schock,Lytle, and Clement, 1994, 1995a,b; Edwards, Meyer, and Schock, 1996) and sometimeslead (Schock, Wagner, and Oliphant, 1996) levels in drinking water

ther-FIGURE 17.6 Solubility and saturation A schematic

solu-bility diagram showing concentration ranges versus pH for

supersaturated, metastable, saturated, and undersaturated

solutions (Source: Aquatic Chemistry, W Stumm and J J.

Morgan Copyright © 1980, John Wiley & Sons, Inc Reprinted

by permission of John Wiley & Sons, Inc.)

Stumm and Morgan have discussed the formation of precipitates and how this isrelated in a conceptual way to solubility diagrams, as illustrated by Figure 17.6(Stumm and Morgan, 1981, 1996).They define an “active” form of a compound as onethat is a very fine crystalline precipitate with a disordered lattice It is generally thetype of precipitate formed incipiently from strongly oversaturated solutions Such anactive precipitate may persist in metastable equilibrium with the solution and mayconvert (“age”) slowly into a more stable, “inactive” form Measurements of the sol-ubility of active forms give solubility products that are higher than those of the inac-tive forms The formation of some of the iron hydroxide or oxyhydroxide solids inpipe deposits mentioned previously provides an example of this phenomenon.Hydroxides and sulfides often occur in amorphous and several crystalline modi-fications Amorphous solids may be either active or inactive Initially formed amor-phous precipitates or active forms of unstable crystalline modifications may undergotwo kinds of changes during aging Either the active form of the unstable modifica-tion becomes inactive or a more stable modification is formed With amorphouscompounds, deactivation may be accompanied by condensation or dehydration.When several of the processes take place together, nonhomogeneous solids can beformed upon aging Similar phenomena can occur with basic carbonates, such as atransition from one form to another with changes in pH or DIC over time.

Rather than construct a different detailed diagram for each level of a secondaryvariable, such as DIC, frequently multiple lines representing the different levels of thisvariable are added to a single diagram, and the aqueous species are omitted For met-als (such as zinc, copper, and lead) with solubilities that tend to be influenced by com-plexation, the expansion of the diagram to include a “third dimension” is often useful.For a qualitative, conceptual understanding, a three-dimensional (3-D) surfacecan be constructed that can show multiple trends in a complex system at a glance.Figure 17.7 shows the two “troughs” representing PbCO3(s) and Pb3(CO3)(OH)2(s),the different trend of solubility with DIC concentration for each solid, and the dis-

FIGURE 17.7 Three-dimensional representation of the effect of DIC tration and pH on lead (II) solubility, assuming ionic strength = 0.005, and tem- perature = 25 °C (Source: Schock and Wagner, 1985.)

concen-tinct solubility minimum in the system (Schock, 1980, 1981b; Schock, Wagner, andOliphant, 1996; Schock and Gardels, 1983).

To obtain a better quantitative estimate of trends of solubility resulting from relationships between two major variables, diagrams such as Figure 17.8 can be con-

inter-structed Operating on the same principles as topographic maps, such contour

diagrams present a “map view” of surfaces such as Figure 17.7 The diagrams are

derived by interpolating levels of constant concentration within a three-dimensionalarray of computed solubilities at different combinations of the other two master vari-ables (e.g., pH and DIC) Several different mathematical algorithms are widely used incommercially available computer software, and considerable care must be exercised inselecting algorithms and data point spacing to prevent the creation of erroneous con-touring artifacts Rapid changes in solubility with respect to a master variable (here,

pH or DIC) are shown by closely spaced contour lines A series of these diagrams atlevels of a third master variable (such as orthophosphate concentration, temperature,and so forth) can be useful to help display multiple interactions with a minimum of dia-grams They also enable a direct reading of estimated solubilities, without having toguess from an indirect perspective, such as with Figure 17.7 One problem with thistype of diagram, as well as with the 3-D surface plots, is that most metals can undergo

a change in solubility of three orders of magnitude or even more, over the range of ditions that might be reasonable for potable waters Thus, logarithmic scales are oftennecessary for the metal concentrations, which can be somewhat confusing to read

con-FIGURE 17.8 Contour diagram for lead (II) solubility assuming the

forma-tion of PbCO 3 and Pb 3 (CO 3 ) 2 (OH) 2, computed for I= 0.02, and temperature =

25 °C (Source: equilibrium constant data from Schock and Wagner, 1985.)

Pourbaix or Potential/pH Diagrams

Using the Nernst equations for appropriate electrochemical half-reactions, it is sible to construct potential-pH diagrams, which are also called Eh-pH, or Pourbaix,diagrams These diagrams have been popularized by Pourbaix and his coworkers inthe corrosion field (Pourbaix, 1966, 1973; Obrecht and Pourbaix, 1967), by Garrelsand Christ, and by Stumm and Morgan in geochemistry (Garrels and Christ, 1965;Stumm and Morgan, 1981) A similar type of diagram uses the concept of electronactivity, pE, which is analogous to the concept of pH (Snoeyink and Jenkins, 1980;Stumm and Morgan, 1981)

pos-The Pourbaix diagrams include the occurrence of different insoluble corrosionproducts of the dissolved metal that limit the concentration of the free metal ion.These diagrams mainly give information about thermodynamically stable productsunder different conditions of electrochemical potential The position of the bound-aries of each region is also a function of the aqueous concentrations (activities) ofions that participate in the half-cell reactions

Potential-pH diagrams are particularly useful to study speciation in systems thatcould contain species of several possible valence states within the range of redoxpotential normally encompassed by drinking water, such as manganese, iron, arsenic,and copper Obtaining an accurate estimate of the redox potential of the drinkingwater is usually an important limitation in using potential-pH diagrams The dia-grams are also useful to gauge the possible reliability of electrochemical corrosion-rate measurement techniques Measurement methods that rely on the imposition of

a potential to the pipe surface may shift the pipe surface into the stability domain of

a solid that would not normally form when freely corroding The imposed potentialmight also serve to alter the nature of the surface phase, leading to an erroneousidentification of the dominating corrosion or passivation reactions as the result of asurface compound analysis

A potential-pH diagram is related tothe solubility versus pH plots discussedearlier This relationship is shownschematically in Figure 17.9 If a con-ventional two-dimensional solubility dia-gram is considered as a vertical plane, apotential-pH diagram may be thought

of as a “slice” that is perpendicular tothe solubility plane, which cuts throughthe plane at a single concentration ofthe metal In actuality, potential-pH dia-grams are usually computed in terms ofactivities rather than concentrations,but the difference is usually not impor-tant for practical purposes The activi-ties of all aqueous and solid speciesmust be fixed, while the electricalpotential of the solution and the pHbecome the master variables If a givenwater constituent (such as calcium)does not oxidize or reduce under mean-ingful physical conditions, no additionaluseful information is gained (beyond

FIGURE 17.9 Schematic relationship between

potential pH and conventional solubility versus

pH diagrams.

that directly available with a solubility diagram) by constructing a potential-pHdiagram.

Figure 17.10 is an example of a potential-pH diagram for iron in water DVGW, 1985) This particular diagram assumes Fe2O3 and Fe3O4to be the solidphases that can control iron solubility The diagram in Figure 17.10 shows that ironand water are never thermodynamically stable simultaneously, because the ironmetal field (Fe) falls below the line where water is reduced to H2gas At such a lowelectrode potential, the iron will not corrode (i.e., it is immune) The iron potential isreduced to the immune region, for example, by cathodic protection To accomplishthis, the iron must be coupled with another, more easily corrodible material, such asmagnesium At low pH (<5) and intermediate to high potential (approximately −0.5

(AWWARF-to 1.3 V), the diagram shows that the stable iron species is Fe2 +or Fe3 + Corrosion willoccur at a high rate under these conditions (i.e., the metal is active) In the high-potential and high-pH regions, solid products such as Fe2O3(s) or Fe3O4(s) may formand deposit on the surface of the iron Figure 17.10 also shows the pH-potentialregions for H2−H2O−O2stability At very low potential, water is reduced to H2, and

at high potential, water is oxidized to O2 The stability of water thus limits the rangeover which the potential of a metal can be varied if it is in contact with water.The same restrictions that apply to solubility diagrams apply to Pourbaix dia-grams Realistic aqueous and solid species must be used, including consideringmetastable solids that are viable controls on metal solubility within the time frame

of interest for scale formation Also, the selected activities (or concentrations) of the

FIGURE 17.10 Potential pH diagram for the iron-water system at 25 ° C (considering Fe, Fe 3 O 4 , and Fe 2 O 3 as solid substances and [Fe 2 + ] or [Fe 3 + ] = 10−6M The cross-hatched region is of the most interest for drinking water (Source: After Internal Corrosion of Water Distribution Sys- tems American Water Works Association Research Foundation, Denver, CO, 1985, Fig 1-3.)

dissolved species should be close to the real situations under study Because the totalactivity is fixed across the entire pH range, the phase relationships predicted by thepotential-pH diagram may deviate somewhat from those predicted in the previouslydescribed solubility diagrams.

Figure 17.11 illustrates the relative oxidizing ability (and instability) of chlorinespecies introduced for water disinfection (Snoeyink and Jenkins, 1980) The pres-ence of species such as Cl2, HOCl0, or OCl−will drive the oxidizing state of the waterupward toward its stability limit (line A), which will affect the corrosion and specia-tion of many of the metals and metal surfaces in contact with the water These aremore powerful oxidizing agents than oxygen itself (line A)

Figure 17.12 shows that the location of the Fe(OH)3(s) stability field provides areasonable explanation for the frequent observation that an amorphous ferrichydroxide forms an outer deposit on cast-iron pipes, and colloidal iron forms in chlo-rinated drinking water or aerated groundwater Figure 17.12 was constructed with amixture of solids and aqueous species of iron that is more typical of a drinking waterenvironment than Figure 17.10 (Schock, 1981a) In waters with high alkalinity (highcarbonate concentrations), siderite [FeCO3(s)] can be found in corrosion deposits

on iron pipe (Singley et al., 1985; Benjamin, Sontheimer, and Leroy, 1996; heimer, Kolle, and Snoeyink, 1981)

Sont-Figures 17.13 and 17.14 are revised potential-pH diagrams based on the samespecies as were used to create Figures 17.4, 17.5, 17.7, and 17.8 Areas of immunityand passivation are unstippled Note how the area of passivation virtually vanishes

as the carbonate level increases from 2.4 mg/L (Figure 17.13) to 24 mg/L (Figure17.14), for activities (concentrations) of aqueous lead species set at the old (1989)0.05 mg/L MCL

Typical protective scales may contain a wide variety of solid phases, and apotential-pH diagram that would incorporate all of these would be hopelessly

FIGURE 17.11 The potential pH diagram for aqueous chlorine; 25 °C, CT,Cl = 1 × 10−4M (Source: Water Chemistry, V L Snoeyink and D Jenkins Copyright © 1980, John Wiley &

Sons, Inc Reprinted by permission of John Wiley & Sons, Inc.)

complicated even if the necessary thermodynamic data were available The diagram

is most appropriate for the conditions of the potential and the pH near the pipe face, when those conditions may differ significantly from those of the bulk waterphase (Bockris and Reddy, 1973) Thus, the type of scale that will form cannot nec-essarily be predicted if the composition of only the bulk water phase is known Fur-ther, thermodynamic considerations alone will never give information about thevelocity of the corrosion process itself Nevertheless, potential-pH diagrams can still

sur-FIGURE 17.12 Potential pH diagram of iron in carbonate-containing water at 25 °C at I= 0 Stability fields are shown for dissolved iron species activities of 0.1 mg/L (—) and 1.0 mg/L (– – –) Dissolved carbonate species concentrations are 4.8 mg C/L (4 × 10−4M).

FIGURE 17.13 Potential pH diagram for the Pb-H 2 O-CO 2

system at 25 ° C.Areas of passivation and immunity are

unstip-pled Dissolved lead species activities = 0.05 mg/L Dissolved

carbonate species activities = 2.4 mg C/L (2 × 10−4 mol/L).

(Source: Water Quality and Treatment, 4th ed., 1990.)

predict or describe corrosion and passivation processes for many water qualities andsystems, especially for metals like copper, iron, and manganese with multiple valencestates possible in drinking water.

In conclusion, Bockris and Reddy (1973) provide several good suggestions andcautions for the use of potential-pH diagrams

● Potential-pH diagrams can be used to give yes or no answers on whether a ular corrosion process is thermodynamically possible or not

partic-● The diagrams provide a compact pictorial summary of the electron-transfer, transfer, and electron-and-proton-transfer reactions that are thermodynamicallyfavored when a metal is immersed in a particular solution

proton-● When a potential-pH diagram indicates that a particular metal is immune to rosion, it is so provided the pH in the close vicinity of the surface is what it isassumed to be

cor-● Even if a potential-pH diagram indicates that a particular corrosion process canspontaneously take place, it does not mean that significant corrosion will beobserved The corrosion reaction rate cannot be predicted thermodynamically

TYPES OF CORROSION

Many different types of corrosion exist (Pourbaix, 1973; Snoeyink and Wagner,1996), and only a few of the most significant to the corrosion of drinking water sys-tems will be covered here Specialized corrosion texts should be consulted for morecomprehensive discussions

The kind of attack depends on the material, the construction of the system, thescale and oxide film formation, and the hydraulic conditions Corrosion forms range

FIGURE 17.14 Potential pH diagram for the Pb-H 2 O-CO 2

system at 25 ° C Areas of passivation and immunity are

unstippled Dissolved lead species activities = 0.05 mg/L

Dis-solved carbonate species activities = 24 mg C/L (2 × 10−3

mol/L) (Source:Water Quality and Treatment, 4th ed., 1990.)

from uniform to intense, localized attack The different forms of corrosion are marily influenced by the distribution of anodic and cathodic areas over the corrod-ing material If the areas are microscopic, and very close to each other, corrosionmay be relatively uniform over the entire surface If the areas are scattered, how-ever, and especially if the potential difference is large, pits may form, sometimes cov-

pri-ered with the irregular and sometimes voluminous deposits called tubercles, which

can greatly impede water flow

Uniform Corrosion

According to one model, when uniform corrosion of a single metal occurs, any onesite on the metal surface may be anodic at one instant and cathodic the next Anodicsites shift or creep about the surface, so rate of loss of metal becomes relatively uni-form over the metal surface

An alternative model for uniform corrosion is that oxidation at a metal surface isaccompanied by electron transport through an adherent film (see Snoeyink andWagner, 1996, Figure 1.7) Reduction of oxygen occurs at the film surface, and trans-port of ions to and away from the oxide film takes place Electrons are probably nottransported through an external portion of the metal The overall rate of corrosion

is controlled by the presence and properties of the film or by transport of reactionproducts, especially hydroxide ion, away from the film-solution interface This modelhas been used for several kinds of metallic corrosion, notably by Ives and Rawson(1962a) for copper Its applicability to drinking water systems having chlorine, orchlorine plus oxygen as the oxidizing agents, is somewhat uncertain

The reasons for development of corrosion cells on metals are varied The singlemetals themselves may be heterogeneous, with possible differences in potentialexisting between different areas because of differences in crystal structure, or imper-fections in the metal Also, the concentrations of oxidants and reductants in solu-tions may be different, causing momentary differences in potential

Frequently, the terms plumbosolvency and cuprosolvency are applied to the

phe-nomenon of uniform corrosion when soluble lead and copper, respectively, arereleased from the piping

Galvanic Corrosion

Galvanic corrosion occurs when two different types of metals or alloys contact eachother, and the elements of the corrosion cell are present One of the metals serves as

the anode, and thus deteriorates, whereas the other serves as the cathode Available

metals and alloys can be arranged in order of their tendency to be anodic, and the

resulting series is called the galvanic series This order, in a potable water

environ-ment, depends on the temperature and solution chemistry (which affect the dynamic activities of the cell components), as well as the simple relative ordering ofthe standard electrode potentials of the oxidation or reduction half-cells or simplemetal/ion redox reaction couplings

thermo-Depending upon generalizations about the range of chemical characteristicsexpected for the water in contact with the metals, different “galvanic series” can becomputed A particularly relevant and useful one is that presented by Larson,54

which is excerpted in Table 17.2

If any two of the metals from different groups in this table are connected in anaqueous environment, the metal appearing first will tend to be the anode and thesecond will be the cathode In general, the farther apart the metals in Table 17.2, the

greater will be the potential for corrosion, because the potential difference betweenthem will be greater.

The rate of galvanic corrosion is increased by greater differences in potentialbetween the two metals It is increased by large areas of cathode relative to the area

of the anode, and it is generally increased by closeness of the two metals and byincreased mineralization or conductivity of the water The relative size of the cath-ode relative to the anode may be of particular concern in the corrosion of lead/tin-soldered joints in copper pipe, or when piping of two different types isinterconnected, such as replacement of lengths of lead pipe with copper pipe (Brit-ton and Richards, 1981; Schock, Wagner, and Oliphant, 1996)

Galvanic corrosion is often a great source of difficulty where brass, bronze, orcopper is in direct contact with aluminum, galvanized iron, or iron Proper selection

of materials and the order of their use in domestic hot- and cold-water plumbing tems is critical to the control of corrosion To prevent galvanic corrosion, for exam-ple, only copper tubing should be used with copper-lined water heaters Brass valves

sys-in contact with steel and galvanized plumbsys-ing sys-in waters with high total dissolvedsolids cause corrosion of the steel and galvanized pipes Dissolved copper can attackspots on galvanized pipe, thereby causing copper-zinc galvanic cells (Kenworthy,1943; Fox et al., 1986)

* Metals within each of the groupings have tively similar corrosion potentials.

rela-† Anodic, or least “noble.”

‡ Cathodic, or most “noble.”

Waters Bulletin 59, Illinois State Water Survey, 1975.

protection It actually takes little metal loss to cause a hole in a pipe wall, and failurecan be rapid Pitting failures can occur in water supplies that meet the regulatoryaction level for copper, and unfortunately, in many cases copper release into thewater is not diagnostic for potential pitting failure problems Pitting can begin orconcentrate at a point of surface imperfections, scratches, or surface deposits Oneproposed mechanism for pitting is that a cause may be ions of a metal higher in thegalvanic series plating out on the pipe surface For example, steel and galvanizedsteel are subject to corrosion by small quantities (about 0.01 mg/L) of soluble met-als, such as copper, that plate out and cause a galvanic type of corrosion.

Recent research has shed some light on pitting causes, including the implication

of a role of sulfide for initiation and microorganisms for exacerbating it Though able prediction from water chemistry data is still generally impossible (Jacobs andEdwards, 1997; Shalaby, Al-Kharafi, and Gouda, 1989; Fischer and Füßinger, 1992;Edwards, Rehring, and Meyer, 1994; Edwards, Ferguson, and Reider, 1994; Duthil,Mankowski, and Giusti, 1996; Ferguson, von Franqué, and Schock, 1996), Taxén(1996) has recently attempted to quantitatively predict the propagation and evolu-tion of pits in relation to equilibrium water chemistry and model the mass transport

reli-Concentration Cell Corrosion

Concentration cell corrosion is usually deduced by inference It occurs when ences in the total or the type of mineralization of the environment exist Corrosionpotential is a function of the concentration of aqueous solution species that areinvolved in the reaction, as well as of the characteristics of the metal Differences inacidity (pH), metal-ion concentration, anion concentration, or dissolved oxygencause differences in the solution potential of the same metal Differences in temper-ature can also induce differences in the solution potential of the same metal.When concentration cell corrosion is caused by dissolved oxygen, it is often

differ-referred to as differential oxygenation corrosion, as discussed by Snoeyink and

Wag-ner (1996) Common areas for differential oxygenation corrosion are between twometal surfaces; for example, under rivets, under washers, under debris, or in crevices.Oxygen concentration cells develop at metal-water interfaces exposed to air,such as in a full water tower, accelerating corrosion a short distance below the sur-face The dissolved oxygen (DO) concentration is replaced by diffusion from air andremains high at and near the surface, but does not replenish as rapidly at lowerdepths because of the distance Therefore, the corrosion takes place at a level slightlybelow the surface rather than at the surface

Tuberculation

Tuberculation occurs when pitting corrosion products build up at the anode next to

the pit (Snoeyink and Wagner, 1996) In iron or steel pipes, the tubercles are made

up of various iron oxides and oxyhydroxides These tubercles are usually rust ored and soft on the outside and are both harder and darker toward the inside.Sometimes, they are considerably layered with iron minerals indicative of anoxicenvironments, such as sulfide minerals (Singley et al., 1985; Benjamin, Sontheimer,and Leroy, 1996) When copper pipe becomes pitted, the tubercle buildup is smallerand is a green to blue-green color, reflecting the deposition of basic cupric carbon-ate, sulfate, or chloride salts, or a mix of them

col-Tuberculation is often associated with poorly buffered waters, where pH can getextremely high under localized surface conditions Hence, many consultants andresearchers have recommended a “balance” of minimum hardness and alkalinity toprovide a water that is resistant to iron corrosion and tuberculation (Weber andStumm, 1963; Stumm, 1956, 1960; Merrill, Sanks, and Spring, 1978; Légrand andLeRoy, 1995; Larson and Skold, 1957).

Crevice Corrosion

Crevice corrosion is a form of localized corrosion usually caused by changes in acidity,

oxygen depletion, dissolved ions, and the absence of an inhibitor.As the name implies,this corrosion occurs in crevices at gaskets, lap joints, rivets, and surface deposits

Erosion Corrosion

Erosion corrosion mechanically removes protective films, such as metal oxides,

hydroxycarbonates, and carbonates, that serve as protective barriers against sive attack It can also remove the metal of the pipe itself Erosion corrosion can beidentified by grooves, waves, rounded holes, and valleys it causes on the pipe walls

corro-Dealloying or Selective Leaching

Dealloying, or selective leaching, is the preferential removal of one or more metals

from an alloy in a corrosive medium, such as the removal of zinc from brass ification), or the removal of disseminated lead from brass (Schock and Neff, 1988;Oliphant and Schock, 1996; Lytle and Schock, 1996, 1997) This type of corrosionweakens the metals and can lead to pipe failure in severe cases The stability ofbrasses and bronzes in natural waters depends in a complex manner on the dissolvedsalts, the hardness, the dissolved gases, and the formation of protective films Dezinc-ification is common in brasses containing 20 percent or more zinc and is rare inbrasses containing less than 15 percent zinc The occurrence of plug-type dezincifi-cation and dezincification at threaded joints suggests that debris and crevices mayinitiate oxygen concentration cells and result in dezincification

(dezinc-Lead occurs in lead/tin solder as a disseminated phase, acting principally as a ent for the tin that actually does the binding with the copper (Parent, Chung, andBernstein, 1988) Factors governing the removal of lead from the soldered joint arepresumably similar to those affecting the leaching of lead from brass Soldered joints

dilu-in service for a long time frequently show a depletion of lead from the exposed solder

In the United States, the leaching of lead, cadmium, zinc, and other metals frombrass and solders is regulated in an indirect way The Safe Drinking Water Act(SDWA) Amendments of 1996 limit the content of Pb in solder and flux to be used

in contact with drinking water to 0.2 percent, given in Section 1417(d) (Safe

Drink-ing Water Act Amendments, 1996) Because the definition of lead-free for pipes, well

pumps, plumbing fixtures, and fittings under the original statute was a content of 8percent or less, it was ineffective in controlling contamination from this source, asvirtually all fixtures and fittings implicated in high levels of metal leaching containmuch less than 8 percent Pb (Oliphant and Schock, 1996; Lytle and Schock, 1996,1997) Thus, the SDWA Amendments of 1996 incorporated a performance standard

in Section 1417(e), requiring fixtures and fittings intended to dispense water forhuman consumption to be certified under a voluntary standard established by theAct (NSF International, 1998) Additionally, U.S states having primacy for enforce-ment of the drinking water regulations (at the time of this writing, 49 of the 50 states)may promulgate more stringent standards if they desire.

Selective leaching also applies to the dissolution of asbestos-cement pipe, or thedeterioration of cement mortar linings of iron water mains Highly soluble compo-nents, such as free lime, calcium carbonates, and a variety of silicates and alumi-nosilicates, can be dissolved by aggressive waters (Schock, 1981; Légrand andLeRoy, 1995; LeRoy et al., 1996; Holtschulte and Schock, 1985; Douglas and Merrill,1991) In some cases, the attack can be so severe as to cause weakening in the walls

of the pipe, and the dislodging of mats of fibers

Graphitization

Graphitization is a form of corrosion of cast iron in highly mineralized water or

waters with a low pH that results in the removal of the iron silicon metal alloy ing up one of the phases of the cast-iron microstructure A black, spongy-appearing,but hard mass of graphite remains

mak-Microbiologically Influenced Corrosion

Microbiologically influenced corrosion (MIC) pertains to the general class of

corro-sion resulting from a reaction between the pipe material and organisms such as teria, their metabolic by-products, or both (Little and Wagner, 1997) Algae andfungi may also influence corrosion by producing changes in the pH, dissolved oxy-gen level, or other chemistry changes to the microenvironment at the metal surface

bac-or under cbac-orrosion deposits Microbiologically influenced cbac-orrosion may be animportant factor in the taste and odor problems that develop in a water system, aswell as in the degradation of the piping materials Microbial activity can also stronglyinfluence the mineralogy of iron corrosion deposits (Camper, 1996; Lewandowski,Dickinson, and Lee, 1997; Brown, Sherriff, and Sawicki, 1997; Lazaroff, Sigal, andWasserman, 1982; Postma and Jakobsen, 1996; Tuhela, Carlson, and Tuovinen, 1992;Allen, Taylor, and Geldreich, 1980; LeChavallier et al., 1993)

Biofilms in pipes are often characterized by stratification, with different families

of organisms with different functions existing in the different zones and conditions(Little and Wagner, 1997) Controlling such growths is complicated because they cantake refuge in many protected areas, such as in mechanical crevices or in accumula-tions of corrosion products Bacteria can exist under tubercles, where neither chlo-rine nor oxygen can destroy them Mechanical cleaning may be necessary in somesystems before control can be accomplished by residual disinfectants Preventivemethods include good corrosion control to avoid tuberculation and accumulatingcorrosion by-products, avoiding dead ends, and preventing stagnant water in the sys-tem by flushing or bleeding

Ainsworth (1980) noted that organic carbon appears to be a major factor in trolling the numbers of microorganisms in a distribution system (see Chapter 18).They noted that the numbers of microorganisms increased along with increasing loss

con-in total organic carbon (TOC) and oxygen through the distribution system Themicrobial activity was found to be concentrated in the surface deposits and pipe sed-

iments It is not clear if one of the mechanisms for reduction of biofilms and bial growth by corrosion control and the application of inhibitors is the prevention

micro-or saturation of smicro-orption sites at the pipe surfaces, allowing less natural micro-organic ter (NOM) sorption and attachment (Camper, 1996; LeChavallier et al., 1993) Thus,there would be a decreased attached nutrient base for the colonizing organisms ifpotential organic material attachment sites were already occupied by phosphate orother corrosion inhibitor material

mat-The ways in which bacteria can increase corrosion rates are numerous Slimegrowths of nitrifying (and other) organisms may produce acidity and consume oxy-gen in accordance with Eqs 17.32 and 17.33

NH4 ++3/2 O2⇔NO2 −+2H++H2O (17.32)

NH4 ++2O2⇔NO3 −+2H++H2O (17.33)These reactions can cause oxygen concentration cells that produce lowered pH andlocalized corrosion and pitting Lee, O’Connor, and Banerji (1980), for example,showed increased localized corrosion when cast iron was exposed to a water withextensive biological activity, compared with the same water under sterile conditions

A very recent study found an association between nitrification activity and coppercorrosion problems (Murphy, O’Connor, and O’Connor, 1997a–c) Microbes mayalso enhance corrosion or solubility because extracellular material and metabolites

of microorganisms may include polymers that are good complexing agents (Geesey

et al., 1986, 1988; Geesey and Bremer, 1991), or they may influence the redox istry of the scales at the surface (Lewandowski, Dickinson, and Lee, 1997; Allen, Tay-lor, and Geldreich, 1980; LeChavallier et al., 1993; Lee, O’Connor, and Banerji, 1980).Iron bacteria derive energy from oxidation of ferrous iron to ferric iron Nuisanceconditions often result because the ferric iron precipitates in the gelatinous sheaths

chem-of the microbial deposits and these can be sloughed chem-off and be the cause chem-of water complaints Sontheimer, Kolle, and Snoeyink (1981) noted that they can alsointerfere with the development of passivating scales and, thus, that the rate of corro-sion is higher in their presence than in their absence

red-The sulfate-reducing bacteria may be the most important organism involved withMIC in many water systems (Little and Wagner, 1997) Sulfate can act at the cathode

in the place of oxygen,

SO4 −+8H++8e−⇔S2 −+4H2O (17.34)and the sulfate-reducing organisms apparently catalyze this reaction The localizedhigh pH could promote corrosion, and reduce the effectiveness of disinfectantsagainst the organisms mediating the reactions Sulfate reducers have been found inthe interior of tubercles, and thus may be responsible for maintaining corrosion atthese locations (Ainsworth, 1980) Sulfate reducers are often present regardless ofwhether the supply is aerated (Little and Wagner, 1997), and then activity may beregulated by TOC levels Reductions in TOC should lessen their activity Someobservations that higher levels of sulfate cause more corrosion (Larson, 1975) maypossibly be related to microbial activity (Snoeyink and Wagner, 1996) Recentresearch has found sulfide to be a contributor to copper pitting that is particularlydestructive and difficult to control (Jacobs and Edwards, 1997), and once started, themicrobial mediation may be very difficult or impossible to stop The high-chlorideand extreme pH environments of pits will lead to continued propagation, even if themicrobes are killed by biocides (Little and Wagner, 1997)

Stray Current Corrosion

Stray current corrosion is a type of localized corrosion usually caused by the

ground-ing of home appliances or electrical circuits to the water pipes It occurs more often

on the outside of pipes, but does show up in house faucets or other valves This ject remains somewhat controversial, although strong evidence has been presented

sub-by a number of researchers that both AC and DC current can affect corrosion ratesand metal levels in the water (Williams, 1986; Horton, 1991; Horton and Behnke,1989; Bell et al., 1995; Bell, Schiff, and Duranceau, 1995) In municipal distributionsystems, increased corrosion of steel-reinforced concrete pressure pipe has occa-sionally been noted, and has been believed to be related to stray currents

PHYSICAL FACTORS AFFECTING CORROSION

AND METALS RELEASE

The characteristics of drinking water that affect the occurrence and rate of corrosioncan be classified as (1) physical, (2) chemical, and (3) biological In most cases, cor-rosion is caused or increased by a complex interaction among several factors Some

of the more common characteristics in each group are discussed in this section tofamiliarize the reader with their potential effects Controlling corrosion may requirechanging more than one of these because of their interrelationships

Essentially no statement regarding corrosion or the general use of a material can

be made that does not have an exception (NACE, 1984).The corrosion of metals andalloys in potable water systems depends both upon the environmental factors (solu-tion composition) and the composition of plumbing or fitting material Concernabout the consequences of corrosion varies with the material Sometimes, it is toxic-ity from trace metal dissolution and contamination of the drinking water, but formaterials such as iron, steel, and mortar linings, the concern is more aesthetic con-cerns and material degradation Therefore, this section will address some of themajor factors that contribute to corrosion in potable waters, but the reader mustrefer to literature that comprehensively describes the relationship for the materials

tify “high,” but a common guideline for 1⁄2-in ID copper is about 4 ft/s (AWWARF,1990), and plumbing handbooks should be consulted for the materials of interest forinitial guidelines.

A water that behaves satisfactorily at medium to high velocities may still causeincipient or slow corrosion of iron or steel with accompanying red-water problems

at low velocities, because the slow movement does not aid the effective diffusionrate of the protective ingredients to the metal surface

Temperature. The influence of temperature is also often misunderstood sion can often be avoided by considering basic equilibria in water chemistry, and byremembering that temperature effects are complex and depend on both the waterchemistry and the type of plumbing material present in the system Multiple phe-nomena often operate simultaneously, such as changes in solubility of solids;changes in the formation of complex ions; changes in diffusion rates of dissolvedgases and solution species; changes in composition or physical properties of the met-als, alloys, or solids; and changes in water properties Thus, generalizations are oftenmade, but they are rarely accurate

Confu-The electrode potential (the driving force for any corrosion cell) is proportional

to the absolute temperature, and therefore, theory predicts that the corrosion ratewill increase with temperature This relationship is observed in some controlled lab-oratory experiments, but in practice is less obvious unless wide differences occur(i.e., hot- versus cold-water systems) In distribution systems, temperature fluctua-tions are somewhat limited over short time frames, so whatever effect they have isobscured by other factors “Seasonal” changes in temperature are often accompa-nied by significant changes in one or more major chemistry parameters Also, whenconsidering household plumbing systems, the relatively high thermal conductivity ofthe metallic piping materials normally used causes the water standing inside pipes,

or passing through long stretches of pipe, to rapidly equilibrate to the temperature

of the air surrounding the pipe

Sometimes, hot water is observed to be more corrosive than cold water Watershowing few corrosive characteristics in the distribution system can cause severedamage to copper or galvanized iron hot-water heaters at elevated temperatures Insome hot-water systems, however, the high temperature can turn a corrosive or non-scaling water into a scaling water with reduced corrosivity, through a combination ofraising the pH, decreasing the solubility of calcium carbonate, driving off dissolvedoxygen or carbon dioxide, and speeding up the reduction of any chlorine speciespresent

Temperature significantly affects the dissolving of CaCO3(calcite) Less CaCO3

dissolves at higher temperatures, which means that CaCO3tends to come out ofsolution (precipitate) and form a protective scale more readily at higher tempera-tures Excessive deposition of CaCO3can clog hot-water lines Some other mineralsbehave similarly, such as the aragonite form of CaCO3, calcium sulfate (anhydrite),and many silicates

Larson has pointed out that the effect of temperature on pH is seldom nized (AWWA, 1971) For pure water, as temperature increases the water dissociatesmore (the pKwgoes down), causing increases in both the H+concentration and the

recog-OH−concentration Drinking waters all contain carbonate species at some tration, which strongly influence the water pH So, the degree of influence of tem-perature on pH is also a function of the alkalinity (inorganic carbon content) of thewater, and the dissociation constants for carbonic acid also change with tempera-ture Increasing concentrations of bicarbonate increasingly buffer or reduce this

concen-effect of temperature on pH The net pH decrease from heating tends to be less thanthe net decrease in overall CaCO3solubility, causing scaling For waters of low alka-linity (less than approximately 50 mg CaCO3/L), the higher temperatures decreasethe pH at a rate that is greater than the rate of decrease in solubility of CaCO3 Theeffect on the saturation state of calcium carbonate is particularly significant for

10 mg CaCO3/L alkalinity, and even for 25 mg CaCO3/L alkalinity with a ture change to 40°C (130°F) and more so if the change is to 55°C (157°F)

tempera-Consider a water at 15°C (59°F) having the following characteristics: pH =8.71;

Ca =17.3 mg/L; Na =17 mg/L; Cl =35 mg/L; and total alkalinity =30.6 mg CaCO3/L.The water is at saturation equilibrium with calcite Therefore, it has a saturationindex (or Langelier Saturation Index) (Trussell, 1985; Snoeyink and Kuch, 1985;Snoeyink and Jenkins, 1980; Snoeyink and Wagner, 1996; Merrill, Sanks, and Spring,1978; Rossum and Merrill, 1983; Joint Task Group, 1990) of 0.00.At this temperature,the solubility constant for the simple dissolution of calcite

Ca2 ++CO3 −⇔CaCO3(s) (17.35)

is equal to 10−8.43 If that water is warmed to 55°C, the saturation index drops to

−0.02, even though the solubility constant decreases to 10− 8.71 This phenomenonoccurs because the pH also decreases to 8.16

Figures 17.15 and 17.16 illustrate how the pH of waters of different alkalinitieschanges when warmed to a temperature of 55°C from 15°C or 25°C, respectively.These figures are based on calculations that assume the following:

FIGURE 17.15 Change in pH for waters of different alkalinities caused by warming a water originally at 15 ° C to 55 ° C under closed-system conditions, assuming an ionic strength of 0.001 The pH predicted for 55 °C is obtained by adding the pH correction to the original (observed)

pH at 15 ° C (e.g., pH = pH + pH ).

● An ionic strength of 0.001 mol/L

● No change in inorganic carbon content

● No redox or hydrolysis reactions that significantly affect the proton balance of thesystem

These temperature effects illustrated in Figures 17.15 and 17.16 explain, in part,the problems of some water supplies, whose alkalinities are too low to buffer theeffect of temperature The low flow velocities in hot-water tanks often aggravatecorrosivity, by limiting the ability of some chemical inhibitors to be effective(AWWA, 1971) More details on a general computational approach and some addi-tional figures for computing the pH change resulting from intrinsic changes to waterand carbonic acid dissociation have been published (AWWARF, 1990)

An additional effect of temperature is that an increase can change the entirenature of the corrosion For example, a water that exhibits pitting at cold tempera-tures may cause uniform corrosion when hot (Singley, Beaudet, and Markey, 1984;AWWA, 1986) Although the total quantity of metal dissolved may increase, theattack is less acute, and the pipe will have a longer life

Another example in which the nature of the corrosion is changed as a result ofchanges in temperature involves a zinc-iron couple Normally, the anodic zinc is sac-rificed, or corroded, to prevent iron corrosion In some waters, the normal potential

of the zinc-iron couple may be reversed at temperatures above 46°C (140°F) Thezinc becomes cathodic to the iron, and the corrosion rate of galvanized iron is muchhigher than would normally be anticipated, a factor that causes problems in manyhot-water heaters or piping systems

FIGURE 17.16 Change in pH for waters of different alkalinities caused by warming a water originally at 25 ° C to 55 ° C under closed-system conditions, assuming an ionic strength of 0.001 The pH predicted for 55 °C is obtained by adding the pH correction to the original (observed)

pH at 25 ° C (e.g., pH 55 = pH 25 + pH corr ).

By changing solubility (as well as pH), temperature changes can influence theprecipitation of different solid phases or transform the identities of corrosion prod-ucts These changes may result in either more or less protection for the pipe surface,depending on the materials and water qualities involved.

Manufacturing-Induced Characteristics

For several types of common plumbing materials, the manufacturing processes usedmay play an important role in determining the types of corrosion that will occur, andthe durability of the pipe or fixtures Pitting of galvanized pipe has been associatedwith several characteristics of its manufacture, such as thin, improper, or uneven gal-vanizing coating, poor seam welds, and rough interior finish (Trussell, 1985; JointTask Group, 1990; Fox, Tate, and Bowers, 1983) Impurities in the galvanizing dipsolution, such as lead and cadmium, could cause concern because of their potentialfor leaching into the water (Trussell, 1985)

In England, rapid pitting failures of copper pipe were found to be associated with

a carbon film on their interior surfaces (Ferguson, von Franqué, and Schock, 1996;Campbell, 1954, 1963, 1964; Cruse and Pomeroy, 1974) The film was a residue fromdrawing oil that had become carbonized during annealing Soft-annealed tubes havebeen more susceptible to pitting than half-hard tubes, which are more susceptiblethan hard-temper tubes (Ferguson, von Franqué, and Schock, 1996; Cruse andPomeroy, 1974) Though it has not been proven by systematic study in many loca-tions, many copper corrosion researchers feel that the early history of the pipe andinitial installation in residential and building plumbing has a profound impact on itssusceptibility to localized corrosion later (Lane, 1993) Debris from manufacturing,transport and storage, and patchy or incomplete oxidation coatings from air andmoisture exposure may set up conditions favorable to differential oxygenation cor-rosion, depending on the initial exposure of the installed piping If plumbing is notflushed thoroughly and housing is not occupied soon after construction, anoxic con-ditions favorable to microbiological growth and, later, to differential oxygenationcorrosion, will likely result, and the corrosion cells may be hard to stifle once normalwater usage is started

CHEMICAL FACTORS AFFECTING CORROSION

Dissolved substances in water have an important effect on both corrosion and rosion control This section provides a brief overview to point out some of the mostimportant factors

cor-General

Table 17.3 lists some of the chemical factors that have been shown to have an tant effect on corrosion or corrosion control Several of these factors are closelyrelated, and a change in one changes another The most important example of this isthe relationship among pH, carbon dioxide (CO2), DIC concentration, and alkalinity.Although CO2is frequently considered to be a factor in corrosion, there is no clearevidence that direct corrosion reactions include CO as a reactant (Singley et al.,

impor-TABLE 17.3 Chemical Factors Influencing Corrosion and Corrosion Control

pH Low pH may increase corrosion rate and the strength of oxidizing agents; high

pH may protect pipes by favoring effective passivation films and decrease sion rates; possibly causes or enhances dezincification of brasses

corro-Alkalinity/DIC May help form protective carbonate or hydroxycarbonate films; helps control pH

changes by adding buffering Low to moderate alkalinity reduces corrosion ofmost materials High alkalinities increase corrosion of copper at all pHs, andlead at high pH

DO Increases rate of many corrosion reactions when metal is passive or immune

under anoxic conditions May help form better passivating oxide films on somematerials, such as iron

Chlorine residual Increases metallic corrosion, particularly for copper, iron, and steel

TDS TDS is a surrogate for ionic strength, which increases conductivity and corrosion

rate unless offset by the formation of passivating films

Hardness Ca may precipitate as CaCO3and thus provide protection and reduce corrosion(Ca and Mg) rates May enhance buffering effect in conjunction with alkalinity and pH.Chloride, sulfate High levels increase corrosion of iron, copper, galvanized steel, possibly lead.Hydrogen sulfide Increases corrosion rates; may cause severe pitting in copper

Ammonia May increase solubility of some metals, such as copper and lead

Polyphosphates May reduce tuberculation of iron and steel, and provide smooth pipe interior

May enhance uniform iron and steel corrosion at low dosages Attacks and ens cement linings and A-C pipe and cement linings Increases the solubility oflead and copper Prevents CaCO3formation and deposition Sequesters ferrousiron and reduced manganese, especially at pH below 7

soft-Silicate Forms protective films on many materials, especially iron and cement lining and

A-C pipe, and galvanized pipe in hot water Often needs preexisting scale to bemost effective Sequesters ferric iron Forms most effective films at high pH Sili-cate chemicals good for raising pH in low-alkalinity waters

Orthophosphate Forms protective films on iron, galvanized pipe, and lead Slows oxidation of

copper at neutral pH Tends to form colloidal lead and maybe other metalspecies at pHs above 8 Interferes with calcium carbonate nucleation andgrowth

Natural color, May decrease corrosion by coating pipe surfaces over long term Some organics organic matter can complex metals and accelerate corrosion or metal uptake, especially when

surfaces are new

Iron, zinc, or May react with compounds on interior of A-C pipe to form protective coating.manganese

Copper May cause pitting in galvanized pipe

Magnesium May inhibit the precipitation of calcite from CaCO3on pipe surfaces, and favor

the deposition of the more soluble aragonite form of CaCO3.Aluminum May form diffusion barrier films on iron, lead, and other pipe materials, such as

aluminum hydroxide or aluminosilicate precipitate Reduces effectiveness oforthophosphate if present at high concentration

1984; AWWA, 1986) In some cases, the rate of CO2hydration might influence thebicarbonate concentration, and, hence, the buffering ability of the water, but theimportant corrosion effect usually results from pH and complexation by bicarbonate

or carbonate ions The dissolved CO2concentration is interrelated with pH and DICconcentration Knowing all of the complex equations for these calculations is not nec-essary, but knowing that each of these factors plays some role in corrosion is useful.The material that follows describes some of the corrosion-related effects of thefactors listed in Table 17.3 A better understanding of how they are related to oneanother will aid in understanding corrosion and, thus, in choosing corrosion controlmethods

pH

pH is a measure of the activity of hydrogen ions, H+, present in water In mostpotable waters, the activity of the hydrogen ion is nearly equal to its concentration.Because H+is one of the major substances that accepts the electrons given up by ametal when it corrodes, pH is an important factor to measure At pH values belowabout 5, both iron and copper corrode rapidly and uniformly At values higher than

9, both iron and copper are usually protected Under certain conditions, however,corrosion may be greater at high pH values Between pH 5 and 9, pitting is likely tooccur if no protective film is present The pH also greatly affects the formation orsolubility of protective films for both metallic and cementitious materials

Alkalinity/Dissolved Inorganic Carbon (DIC)

Alkalinity is a measure of the ability of a water to neutralize acids and bases(Trussell, 1985; Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981; Weber andStumm, 1963; Pankow, 1991; Butler, 1982; Faust and Aly, 1981; Morel, 1983; Loewen-thal and Marais, 1976) In most potable waters, total alkalinity is mainly described bythe relationship:

TALK =2[CO3 −] +[HCO3 −] +[OH−] −[H+] (17.36)where [ ] indicates concentration in mol/L and total alkalinity is in equivalents/L(eq/L) Operationally, it is defined by alkalametric titration to the carbonic acidequivalence point (Stumm and Morgan, 1981; Butler, 1982; Loewenthal and Marais,1976; Schock and George, 1991) The concentration of bicarbonate and carbonateions is directly related to the pH of the water and the DIC concentration through thedissociation of carbonic acid Many distribution diagrams have been presentedshowing the fraction of the DIC that is present in each form, and they can be readilycalculated (Trussell, 1985; Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981;AWWARF, 1990; Pankow, 1991; Butler, 1982; Faust and Aly, 1981; Morel, 1983;Loewenthal and Marais, 1976)

Dissolved inorganic carbon is defined as the sum of all dissolved containing species When ion pairs and complexes, such as CaHCO3 +, MgCO3, arenegligible in concentration, DIC is simply:

carbonate-DIC =[H2CO3* ] +[HCO3 −] +[CO3 −] (17.37)where [H2CO3*] represents the sum of dissolved aqueous CO2gas molecules and car-bonic acid molecules (Stumm and Morgan, 1981; Pankow, 1991; Loewenthal and

Marais, 1976) The concentration of DIC may be either directly analyzed (Schockand George, 1991) or computed from a total alkalinity titration and pH, with propercorrections for ionic strength, proton-consuming species (e.g., HPO4, H3SiO4 −, NH3,and so on) (Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981; AWWARF, 1990;Butler, 1982; Loewenthal and Marais, 1976; Schock and George, 1991) When alka-linity is defined adequately by Eq 17.33, DIC (in units of mg C/L) may be derivedfrom total alkalinity (in units of eq/L) and pH using the relationship:

Total alkalinity in conventional units of mg CaCO3/L can be converted to units ofeq/L by dividing by 50044.5 (a factor derived from the equivalent weight of calciumcarbonate and conversion of grams to milligrams) When Eqs 17.36 through 17.38are applicable, total alkalinity is related to DIC in the manner shown graphically byFigure 17.17, which is similar to figures developed by Deffeyes (Stumm and Morgan,1981; AWWARF, 1990; Schock and George, 1991; Deffeyes, 1965) For these calcula-tions, a temperature of 25°C and an ionic strength of 0.005 were assumed The totalalkalinity/DIC relationship is affected by temperature and ionic strength, so usingother assumptions would change the slopes of the lines in Figure 17.17

When bases other than the carbonates (HCO3 −and CO3 −) and OH−are present

in significant quantities, they will consume protons in the alkalinity titration to thecarbonic acid equivalence point Then, the alkalinity definition must be expanded to

FIGURE 17.17 Relationship between total alkalinity and the total

dissolved inorganic carbon concentration for a water at 25 ° C and an

ionic strength of 0.005.

accommodate them For example, the alkalinity of a water containing phate, hypochlorite, ammonia, silica, and some singly charged organic acid species[OA−] would then be (Snoeyink and Jenkins, 1980):

orthophos-Total alkalinity =[HCO3 −] +2[CO3 −] +[OCl−] +[HPO4 −]

+[OA−] +[NH3] +[H3SiO4 −] +[OH−] −[H+] (17.39)This equation assumes that the concentrations of PO4 −and H2SiO4 −are negligi-ble, because their dissociation constants are so small the pH would need to be inor-dinately high for them to exist in significant concentration in potable waters Alsonote that at the carbonic acid equivalence point, HPO4 −is not fully converted to

H2PO4 −, so the stoichiometric multiplier is presumed to be 1 as an approximation,rather than the factor of 2 that might be commonly assumed based on the charge ofthe species

Hydrogen ion–consuming complexes of metals, such as CaHCO3 +, Fe(OH)2,Al(OH)3, MgCO3, and Pb(CO3)2 −, also contribute to alkalinity, but their concen-trations are usually small enough that their contribution can be neglected If a com-plex reacts slowly with the acid in an alkalinity titration, it should not be included inequations used to derive DIC from pH and the titration alkalinity

The bicarbonate and carbonate species affect many important reactions in sion chemistry, including the ability of a water to form a protective metallic carbon-ate scale or passivating film, such as CaCO3, FeCO3, Cu2CO3(OH)2, Zn5(CO3)2(OH)6,

corro-or Pb3(CO3)2(OH)2 They also affect the concentration of calcium ions that can bepresent, which, in turn, affects the dissolution of calcium from cement-lined or fromasbestos-cement (A-C) pipe

The formation of strong soluble complexes with metals such as lead, copper, andzinc (Schock, 1980, 1981; Schock, Lytle, and Clement, 1994, 1995a,b; Edwards, Meyer,and Schock, 1996; Schock and Gardels, 1983; Schock, 1989) can accelerate corrosion orcause high levels of metal pickup, given the right pH/alkalinity or pH/DIC conditions

Buffer Intensity (β), Buffer Capacity, Buffer Index

The ability of a water to provide buffering against a pH increase or decrease caused

by a corrosion process or water treatment chemical addition is closely related to thealkalinity, DIC concentration, and pH of the water (Trussell, 1985; Pourbaix, 1973;Snoeyink and Jenkins, 1980; Stumm and Morgan, 1981; Weber and Stumm, 1963;Pankow, 1991; Butler, 1982; Faust and Aly, 1981; Loewenthal and Marais, 1976) The

buffer intensity of a water is defined as βC=(∂A/∂pH)DIC, which is essentially theinverse of the slope of the alkalinity titration curve (Stumm and Morgan, 1981;Weber and Stumm, 1963; Butler, 1982) The basic equation for buffer intensity,βtot, interms of total alkalinity (TALK) is the following (Snoeyink and Jenkins, 1980;Stumm and Morgan, 1981; Weber and Stumm, 1963; Butler, 1982; Loewenthal andMarais, 1976) for a system containing only the carbonic weak acid system in addition

to water

(17.40)TALK is expressed in eq/L units, [H+] is the hydrogen ion concentration in mol/L,

K'Wis the dissociation constant for water, and K'1and K'2are the first and second sociation constants for carbonic acid, respectively, all corrected for temperature and

dis-K 'W

[H+]

ionic strength Similarly, buffer intensity can be written in terms of DIC (mol/L)rather than alkalinity The resulting equation is then:

(17.41)The units of βare normally mol/L per pH unit When other weak acids or bases such

as orthophosphate or silicate are present, or if the ammonia concentration is high,additional terms must be added to Eqs 17.40 or 17.41 These expressions relate to

what is often termed a homogeneous buffer system, in which all buffering

compo-nents are aqueous species

Figure 17.18 shows the effect of temperature for a DIC concentration of 48 mgC/L Figure 17.19 shows the buffer intensity for several different concentrations ofDIC Note the minimum point near pH 8, which corresponds to the point where pH

is equal to 1⁄2(−log K'1+ −log K'2) for carbonic acid The buffering at the pH extremes

is from either hydrogen ion (low pH) or hydroxide ion (high pH) Figure 17.20 showsthe effect on buffer intensity of the presence of 20 mg SiO2/L silicate, and 5 mg PO4/Lorthophosphate, in a water having a DIC concentration of only 4.8 mg C/L Clearly,the carbonate system provides almost all of the buffering in the system, except whenthe pH is over about 9 and the silicate anion (present at a high total SiO2concentra-tion) contributes sufficiently This is important, because the figure demonstrates thatcorrosion inhibitor chemicals of all types at normal drinking water dosages are a neg-ligible component of pH buffering for all but waters of extremely low DIC When awater body is open to the atmosphere, the exchange of carbon dioxide gas will affectthe DIC concentration, and both the buffer intensity and capacity

The difference between alkalinity and buffering intensity is often misunderstood,and Table 17.4 shows clearly that they are not the same The four waters shown have

K 'W

[H+]

K '1K ' 2+[H+]2



(K '1K ' 2+K '1[H+]+[H+]2)2

FIGURE 17.18 Effect of temperature on buffer intensity of

water for a system with DIC =48 mg C/L, assuming I= 0 Ionic

strength has little effect on the position of the curves.

identical alkalinity, but different pH values These waters have different DICs aswell, but the DIC concentrations do show the same trend as buffer intensity Table17.4 shows that the buffer intensity trend corresponds to that in Figures 17.19 and17.20, where the lowest value is near pH 8 Thus, utilities distributing water at a pHaround 8 to 8.5 will have much poorer pH stability than if the water with the same

FIGURE 17.19 Effect of DIC on buffer intensity, for

25 °C, I= 0.

FIGURE 17.20 Combined components of buffer intensity, for

DIC, silicate, and orthophosphate at high dosages and low DIC.

Computed for 25 °C, I= 0.005.