Content uploaded by Michael R Schock
Author content
All content in this area was uploaded by Michael R Schock on Dec 09, 2014
Content may be subject to copyright.
United States
Environmental Protection
Agency
Effect of pH, DIC,
Orthophosphate and
Sulfate on Drinking Water
Cuprosolvency
Office of Research and
Development
Washington DC 20460
EPA/600/R-95/085
June 1995
Contact
Michael R. Schock and Darren A. Lytle are the EPA contacts for this report. They are
presently with the newly organized National Risk Management Research Laboratory,
new Water Supply and Water Resources Division in Cincinnati, OH (formerly the
Risk Reduction Engineering Laboratory). The National Risk Management Research
Laboratory is headquartered in Cincinnati, OH, and is now responsible for research
conducted by the Water Supply and Water Resources Division.
Effect of pH, DIC, Orthophosphate and Sulfate on
Drinking Water Cuprosolvency
by
Michael R. Schock and Darren A. Lytle
Water Supply and Water Resources Division
National Risk Management Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
and
Jonathan A. Clement
Black and Veatch
Cambridge, Massachusettes 02140
Project Officer
Michael R. Schock
Water Supply and Water Resources Division
National Risk Managment Research Laboratory
Cincinnati, OH 45268
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
EPA/600/R-95/085
June 1995
ii
Disclaimer
The information in this document has been funded wholly or in part by the U. S. Environmen-
tal Protection Agency. It has been subjected to the Agency’s peer and administrative review, and it
has been approved for publication as an EPA document. Mention of trade names or commercial
products is for explanatory purpose only, and does not constitute endorsement or recommendation
for use.
iii
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the Agency strives
to formulate and implement actions leading to a compatible balance between human activities and
the ability of natural systems to support and nurture life. To meet these mandates, EPA's research
program is providing data and technical support for solving environmental problems today and
building a science knowledge base necessary to manage our ecological resources wisely, under-
stand how pollutants affect our health, and prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for investigation
of technological and management approaches for reducing risks from threats to human health and
the environment. The focus of the Laboratory's research program is on methods for the prevention
and control of pollution to air, land, water, and subsurface resources; protection of water quality in
public water systems; remediation of contaminated sites and groundwater; and prevention and
control of indoor air pollution. The goal of this research effort is to catalyze development and
implementation of innovative, cost-effective environmental technologies; develop scientific and
engineering information needed by EPA to support regulatory and policy decisions; and provide
technical support and information transfer to ensure effective implementation of environmental
regulations and strategies.
This publication has been produced as part of the Laboratory's strategic long-term research
plan. It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
iv
Abstract
Field data from various copper monitoring studies and Lead and Copper Rule compliance data
are often inappropriate and misleading for reliably determining fundamental chemical relation-
ships behind copper corrosion control. To address this deficiency, a comprehensive solubility model
for copper in drinking water has been developed, that is qualitatively and quantitatively consistent
with available data for copper dissolution and precipitation. The concentration of Cu(I) is domi-
nated by Cu2O(s) or CuOH(s) solid phases, plus soluble aqueous ammonia and chloride complexes.
For young piping, the concentration of Cu(II) is mainly governed by Cu(OH)2(s) (cupric hydrox-
ide), rather than CuO(s) (tenorite) or Cu2(OH)2CO3(s)(malachite). Complexation of Cu(II) by DIC
and hydroxide ion is extremely important. Increases in DIC are predicted to cause significant
increases in copper solubility in the pH range of 7 to 10. Often, utilities must trade off increasing
cuprosolvency by DIC addition for ensuring adequate buffering intensity in the finished water.
Sufficient dosages of orthophosphate in the pH range of 6.5 to 7.5 may reduce cuprosolvency,
while sulfate may assist in controlling cuprosolvency under some chemical conditions, or may
interfere with the formation of cupric hydroxide films under mildly alkaline conditions. Dissolved
oxygen and chlorine residual play complicated roles in determining copper concentrations after
various standing times. Frequently, 48 to 72 hours are necessary to reach equilibrium levels of
copper in disinfected systems. Considerable uncertainties in much of the important thermody-
namic data for copper species still exist, and many research needs remain that are critical to im-
proving the understanding and control of cuprosolvency. This report covers a period from June
1991 to November 1993, and work was completed as of May 1995.
v
Table of Contents
Disclaimer....................................................................................................................................... ii
Foreword........................................................................................................................................iii
Abstract......................................................................................................................................... iv
List of Figures............................................................................................................................... vii
List of Tables ................................................................................................................................. ix
Acknowledgments ........................................................................................................................ x
Introduction .................................................................................................................................. 1
Historical Framework ............................................................................................................. 3
Study Objectives..................................................................................................................... 4
Copper Chemistry Overview-The Theory.................................................................................... 6
Thermodynamic Data Used for Modeling.............................................................................. 6
Oxidation Reactions of Copper ............................................................................................ 12
The Conceptual Role of Aqueous Complexation ................................................................. 15
Copper (I) Chemistry............................................................................................................ 15
Copper(II) Hydrolysis Reactions........................................................................................... 17
Copper(II) Complexation by Carbonate ................................................................................ 21
Passivating Cupric Oxide and Hydroxide Solids .................................................................. 24
The “Cupric Hydroxide Model" ............................................................................................. 26
The Malachite Problem ......................................................................................................... 29
Effect of Orthophosphate on Copper(II) Solubility............................................................... 35
Effects of Sulfate, Chloride and Ammonia on Copper(II) Solubility ................................... 42
Interrelationships of Copper (I) and Copper (II) ................................................................... 44
Complications of Dissolved NOM ........................................................................................ 48
Experimental and Field Evidence for the “Cupric Hydroxide Model” ....................................... 50
Experimental Systems ........................................................................................................... 50
Analytical and Data Reporting Procedures ...................................................................... 50
Coupon Experiments ....................................................................................................... 54
Single-Pass Pipe Experiments ......................................................................................... 58
Recirculation Solubility Experiments.............................................................................. 59
Data Analysis Approaches ..................................................................................................... 61
Experimental Results............................................................................................................. 64
vi
Table of Contents (Continued)
Oxidation Rates and Stagnation Curve Behavior ............................................................ 64
The Effect of pH and DIC on Cu(II) Solubility............................................................... 69
The Effect of Aging on Copper Levels in the Water ....................................................... 77
Conclusions ................................................................................................................................. 81
Significance of pH and DIC in Cuprosolvency Control........................................................ 81
Significance of pH and Orthophosphate in Cuprosolvency Control ..................................... 82
Implications for Controlling Highest Copper Exposures ...................................................... 82
Significance of Cupric Hydroxide Model for Demonstration Studies .................................. 83
Significance of Chlorination and Aeration on Copper Levels................................................ 84
Future Research Needs .......................................................................................................... 84
References ................................................................................................................................... 87
vii
List of Figures
Number Page
1. Effect of chloride complexation on CuOH(s) solubility, I=0.02, 25°C................. 18
2. Effect of ammonia complexation on CuOH(s) solubility, I=0.01, 25°C ............... 19
3. Three-dimensional surface plot of the effect of DIC and pH on copper
(II) solubility, assuming equilibrium with Cu(OH)2 having a large molar
surface (I=0.02, 25°C). .......................................................................................... 27
4. Contour diagram showing the effect of DIC and pH on copper(II) solubility,
assuming equilibrium with Cu(OH)2 solid having a large molar surface
(I=0.02, 25°C)........................................................................................................ 28
5 Copper(II) speciation assuming formation of solid Cu(OH)2;
a) DIC=4.8 mg C/L I=0.005, 25°C; b) DIC=96 mgC/L,
I=0.02, 25°C. ......................................................................................................... 30
6. Three-dimensional surface plot of the effect of DIC and pH on copper(II)
solubility for aged systems, assuming equilibrium with Cu2(OH)2CO3
and CuO (I=0.02, 25°C). ....................................................................................... 33
7 Contour diagram showing the effect of DIC and pH on copper(II) solubility,
assuming equilibrium with the stable solid phases Cu2(OH)2CO3
(malachite) and CuO (tenorite) at I=0.02 and 25°C. ............................................. 34
8. Copper(II) solubility comparison assuming different solids and solubility
constants, 25°C; a) DIC = 4.8mg C/L, I=0.005; b) DIC = 96 mg C/L, I=0.02 ..... 36
9. Effect of orthophosphate on copper(II) solubility at DIC = 4.8 mg C/L,
I = 0.005 and 25°C, assuming the formation of Cu(OH)2(s),
Cu2(OH)2CO3(s), and Cu3(PO4)2•2H2O................................................................. 40
10. Copper(II) solubility for differenct DIC levels with orthophosphate addition,
assuming formation of Cu(OH)2(s) and Cu3(PO4)2•2H2O(s) at 25°C:
a) DIC = 4.8 mg C/L, I = 0.005; b) DIC = 14.4 mg C/L, I = 0.005;
c) DIC = 48 mg C/L, I = 0.01; d) DIC = 96 mg C/L, I = 0.02. ............................. 41
viii
List of Figures (continued)
Number Page
11. EMF-pH diagram for copper in water containing carbonate, assuming
formation of cupric and cuprous hydroxide at DIC = 4.8 mg C/L, I = 0,
copper species concentrations of 1.3 mg/L, and 25°C. ......................................... 45
12. EMF-pH diagram for copper in water containing carbonate, assuming
formation of cupric and cuprous hydroxide at DIC = 96 mg C/L, I = 0,
copper species concentrations of 1.3 mg/L, and 25°C. ......................................... 46
13. Copper(II) solubility of different DIC levels compared to copper(I)
solubility, at I = 0.01, 25°C for: a) new pipe with Cu(OH)2(s);
b) aged pipe. .......................................................................................................... 47
14. Box plots of percent ion balance errors for complete water analyses from
the laboratory experimental runs reported. Boxes show mean (dark line), median
(light line), 25th and 75th percentiles (box borders), and the 10th and 90th
percentiles (error bars). Data outside these limits are marked (open circles).
Large errors represent missing analytes. ............................................................... 53
15. Relationship among depletion percentage for free chlorine residual,
dissolved oxygen, and copper concentration over varying stagnation
lengths, after 250 days of pipe use. ....................................................................... 65
16. Saturation index changes for three solids during stagnation of different
durations in copper pipe loop study. Pipe was aged approximately
250 days................................................................................................................. 66
17. Schematic illustration comparing metal concentrations that would be
observed after standing different amounts of time given different
controlling chemistry factors. ................................................................................ 68
18. Precipitation data of Patterson (1981) superimposed on two theoretical
models for Cu(II) solubility at low DIC concentrations. Models computed
for I = 0.01, and 25°C. Controlling solids were assumed to be: a)CuO(s)
and Cu2(OH)2CO3(s); b) Cu(OH)2(s). ................................................................... 70
19. Computed Saturation Indices for important solids in copper coupon study
runs 1-5; a) Run 1, pH = 8.5, 72 hour standing time; b) Run 2, ph 7.0,
72 hour standing time; c) Run 3, pH 7.5, 3 mg PO4/L, 24 hour standing
time; d) Run 4, pH 7.5, 0.5 mg PO4/L, 72 hour standing time; e) Run 5,
pH 7.5, 72 hour standing time. .............................................................................. 73
20. Comparison of theoretical and observed copper levels for coupon study. ............ 74
ix
List of Figures (continued)
Number Page
21. Computed saturation indices for three copper(II) solids in recirculation
solubility experiments with DIC=5 mgC/L at: a) pH 7.0; b) pH 8.0; and
c) pH 9.0. ............................................................................................................... 76
22. Distribution of adjusted copper leaching rates versus age of plumbing
for one litre samples from the residential metals study of the Contra Costa
Central Sanitary District. ....................................................................................... 79
List of Tables
Number Page
1. Internally-consistent Gibbs free energies of formation for copper and
related species at 25°C............................................................................................ 7
2. Equilibrium reactions in copper solubility programs, and correspond-
ing log K and log
β
values.................................................................................... 11
3. A selection of reported log
β
12 values for the reaction:
Cu2++2H2O
↔
Cu(OH)2°+2H+............................................................................... 20
4. A selection of reported log equilibrium constants (ss) for copper(II)
carbonate complexes. All values have been reported for 25°C and
I=0, or were extrapolated to I=0 by the Davies Equation for I ≤ 0.5. ................... 22
5. Reported minerals containing copper(II) and orthophospate[1]............................ 38
6. Analytical methods used for chemical analysis of water samples......................... 51
7. Summary statistics for the variation in background water quality during
the different runs of the coupon corrosion study................................................... 55
8. Water quality in the copper pipe study. ................................................................. 59
9. Water quality for different runs of recirculation study. ......................................... 60
10. Approximate water quality covering sites in the California wastewater
loading study. ......................................................................................................... 78
x
Acknowledgments
The authors are very grateful to the late Prof. Dr. Heinrich Sontheimer of the Engler-Bünte
Institüt, Karlsrühe, Germany, and Dr. R. Rhodes Trussell of Montgomery-Watson Engineers for
presenting intriguing hypotheses and observations that inspired and encouraged this effort. The
support of Dr. Marc Edwards of the University of Colorado (Boulder) who shared considerable
research information that assisted this study, and who provided considerable constructive editorial
review, is greatly appreciated. Thomas Sorg of USEPA also provided important review and edito-
rial commentary. The authors would also like to extend their thanks to personnel of several water
and waste water systems and their associates, who willingly provided useful information and moni-
toring data of importance to this investigation, especially Betsy Elzufon of Larry Walker Associ-
ates, Davis, CA, and the Central Contra Costa (California) Sanitary District. Analytical support for
the USEPA laboratory study samples was provided by Roger Rickabaugh, Steve Harmon, and
Greg George of Dyncorp/Technology Applications, Incorporated, and Keith Kelty, Lou Trombly,
Jim Doerger, Jim Caldwell, Ken Kropp, and Herb Braxton of the Drinking Water Research Divi-
sion, USEPA. Assistance with data compilation and presentation was contributed by Susan Schock,
Chris Keil, and Leslie Ostrozny of USEPA.
1
Introduction
The promulgation of the “Lead and Copper Rule” by the USEPA in 1991 has required approxi-
mately 80,000 water utilities nationwide to monitor for copper resulting mostly from the uniform
corrosion of copper plumbing materials.1-4 A complete breakdown of monitoring results covering
lead and copper levels and associated water quality parameters for the first two rounds of sampling
by all utilities has not been verified and made available for analysis. Therefore, the exact extent of
the problem is hard to quantify. However, in the first round of monitoring by the large water
systems (about 682), approximately 6 % exceeded the 1.3 mg/L action level according to a recent
study.5 Examination of data from 7,600 medium- and large-sized utilities entered into the Federal
Reporting Data System (FRDS) indicated the largest relative fractions of utilities reporting
exceedence of the action level were generally in the Northeast and Pacific Northwest regions of the
country.4 However, the highest relative fraction overall occurred in Nebraska (49%), and the states
of Minnesota (21%) and Iowa (16%) also had high copper levels in a relatively large fraction of
water systems of these sizes. The highest copper levels in the water for locations served by large
utilities appeared to be in the Southeastern and Western regions of the United States, in utilities
covering a considerable range of water qualities.5 Generally, the action level was exceeded more
frequently in water systems having ground water sources.4
The cuprosolvency (copper solubility) problem apparently increases with decreasing utility
size. For example, on the basis of data from the USEPA Office of Ground Water and Drinking
Water and two other reports are that for medium-sized systems in their first round of sampling, the
action level was exceeded by 87 utilities in Massachusetts, 57 in Florida, 32 in New York and
Illinois, 27 in Pennsylvania and Washington, and 26 in Minnesota.6,7 In the first round of regulatory
monitoring, approximately 485 medium-sized utilities exceeded the copper action level according
to these sources. For both rounds, a total of 600 of 7600 utilities exceeded the copper action level.4
Unfortunately, the regulatory monitoring data are of limited use for extracting details of copper
chemistry behavior and understanding potential copper passivation strategies. Several important
problems with such an approach are the following.
• Water quality parameter data are required only from utilities not meeting the Lead
and Copper Rule Action Levels, so the information is not reported or even not col-
lected by many utilities.1-3,8
• For medium-sized water systems, water quality parameter data were frequently,
collected at different times than the samples for lead and copper. In practice, many
utilitites first collected the source water and first-draw water samples, to see if they
2
met the 90th-percentile action levels for lead and copper. If they found they were
below the action level, reporting of such data was not required. Hence, it was often
not collected. If they were above the action level, they collected the water quality
data.
• Water quality data does not necessarily represent values at the initiation of water
stagnation for regulatory sampling. Geographic (within the system) location and
synchronization of water chemistry characterization with lead and copper levels is
critically important to assess chemical relationships behind corrosivity and treat-
ment options. Without employing these important sampling principles, making a
direct statistical or mechanistic link between background water chemistry condi-
tions and the corresponding lead and copper values becomes numerically possible
but scientifically meaningless. Therefore, using existing regulatory data for such
comparisons is only valid for systems having extremely consistent water qualities
over time, and throughout the distribution network.
• Data regarding some important background chemical constituents that might influ-
ence copper corrosion (eg. chloride, sulfate, silica, orthophosphate, etc.) are not
necessarily collected, particularly when corrosion inhibitors (phosphates, silicates)
are not employed.
• Stagnation times of monitoring samples are permitted to fall within a ten-hour pe-
riod (6-16 hours). Considerable scatter in tap water data resulting from differences
in standing time have been observed by Wagner.9
• Copper dissolution rates depend greatly on the age of the passivating films, so inter-
pretation of monitoring data must include adjustment for various plumbing ages to
correctly deduce differences in cuprosolvency resulting from chemical rather than
temporal factors.
• Surface films remaining on the interior pipe surface from the manufacturing process
can affect passivating films formed in the plumbing systems.10 The persistence and
impact of these films on subsequent copper dissolution into the water will depend
greatly on the entire usage pattern of the individual plumbing system.
• Lead and copper monitoring sites are purposely biased to represent locations tend-
ing to contain high levels of lead, such as sites with lead interior plumbing, lead
service lines, and high-lead solder installed after 1982.2 Consequently, in many
cases the collected sample volume may not have been exposed to copper plumbing.
• High copper levels that could considerably exceed the new 90th-percentile U. S.
action level may go undetected in the regulatory monitoring process. For copper,
the highest corrosion rates and solubilities will most often be associated with the
newest piping. These sites will normally be outside of the priority monitoring site
3
tiers, leading to the interesting possibility that a water supplier may pass the regula-
tory action level requirement, but may still have a copper solubility problem.
Nonetheless, several interesting gross-scale trends have been discerned for large water sys-
tems.4,5 There is a poor correlation between 90th-percentile lead and copper levels. Another trend is
that copper exceedences tended to be highest at very low alkalinities (<25 mg CaCO3/L) and in-
creasingly greater over 75 mg CaCO3/L. Finally, no action level exceedences were reported for
systems having a pH above approximately 8.
The data thus far suggest that cuprosolvency will be a major concern across the United States,
especially for smaller water suppliers that are less likely to employ corrosion control and use ground
water sources. Further, the poor correlation between reported 90th-percentile lead and copper levels
suggests that different control strategies for copper than those considered appropriate for lead may
need to be developed or employed by affected utilities. Understanding how copper will respond to
lead control measures and the results of other regulatory treatment requirements is therefore of
considerable interest. Indeed, a response that effectively controls lead corrosion might exacerbate
copper corrosion. Moreover, a utility must distribute aesthetically-pleasing water. A good example
of the conflicts between control of corrosion of iron mains and reducing copper corrosion rates has
been given for a study in Vancouver, BC.11
Historical Framework
While many previous papers and book sections have been published concerning copper corro-
sion, the focuses of most were on unusual corrosion phenomena or pitting. A summary of consid-
erable copper corrosion literature has been developed by Cruse and von Franquè.12
A review of the general literature on copper chemistry in drinking water and natural water
systems led to the conclusion that the basic cupric hydroxy-carbonate compound Cu2(OH)2CO3
(malachite) should form a passivating film over the pipe surfaces in many water systems, and
reduce the corrosion potential. The fact that many utilities reported soluble copper levels that far
exceeded predictions based on malachite solubility raised important questions about the accuracy
of this assumption.
The great influence of pH on copper corrosion has been widely accepted. Attack on copper by
“dissolved CO2” or “carbonic acid” has been reported by many investigators, although the mecha-
nism has been a matter of some debate.12,13 However, much confusion still exists in the drinking
water treatment field about what constitutes a corrosive water toward copper plumbing materials.
The confounding nature of the many variables impacting metal solubilities and leaching rates
makes comparing a large number of possible treatments a cumbersome and daunting task for indi-
vidual utilities considering experimental evaluations. It also obscures basic interrelationships, by
compounding intrinsic chemical variability and impacts with artifacts caused by even subtle differ-
ences in experimental testing protocols.
4
In attempting to address some of the data gaps for cuprosolvency control by utilities, a variety
of experimental systems have been constructed and operated in USEPA laboratories. During these
experiments, some perplexing data was generated that appeared to either contradict some “conven-
tional wisdom” on copper corrosion, or showed unexpected sensitivities to important water chem-
istry variables and experimental system operational protocols. These observations provided the
initial motivation to begin exploring the copper solubility issue in detail.
If an accurate equilibrium model for cuprosolvency were developed, it would provide utilities
and consultants with a practical tool for selecting and applying corrosion control programs that will
work over the long term. A sound theoretical and practical understanding of the important factors
affecting cuprosolvency would greatly obviate the need for what are often complicated, costly, and
sometimes misleading bench-top and pilot-plant experimental studies. Small and medium-sized
water systems frequently lack the technical sophistication, as well as the mechanical and financial
resources, needed to conduct complicated demonstration studies. For water systems able to con-
duct experimental treatment evaluations, an accurate model for cuprosolvency would reduce the
need for including fundamental tests (eg. pH and carbonate concentration effects) in their evalua-
tions. Utilities could then concentrate on determining the need for additional experimental evalu-
ations of inhibitor treatments or investigation of other unusual chemistry characteristics of particu-
lar water systems.
A better understanding of roles of pH and DIC (dissolved inorganic carbon) in cuprosolvency
would greatly improve the accuracy of regulatory “desk-top” corrosion optimization evaluations,
thereby providing a better assessment of long-term treatment impacts than those ascertained over a
short time-frame (ie. weeks, months). Finally, an accurate solubility model for copper would allow
many utilities to achieve the requirements of the Lead and Copper Rule without the need to conduct
the treatment studies.
A newly-emerging indirect constraint on copper corrosion control in drinking water is by waste-
water effluent guidelines and limits that are becoming increasingly stringent. Ironically, ambient
corrosion of domestic, commercial and institutional plumbing systems is now becoming a “con-
taminant” of wastewater that is becoming difficult to control through normal waste treatment pro-
cesses. The development of an accurate model for copper solubility will allow the evaluation of
treatment alternatives for optimizing the control of cuprosolvency beyond drinking water regula-
tory requirements, to achieve adequate levels at the end of wastewater treatment.
Study Objectives
This fundamental purpose of this initial phase of copper corrosion research was to tie together
into a comprehensive solubility model various aspects of both the aqueous and solid speciation of
copper under redox potential conditions typical of municipal drinking water systems. This model
development and presentation would clarify the interrelationships among copper solubility, pH and
the concentrations of various aqueous carbonate species in potable waters. The solubility model
developed, which will subsequently in this paper be called the "cupric hydroxide model," provides
5
a rational basis for applying pH and DIC adjustment for domestic plumbing corrosion control
through central water treatment. Further, the model also provides an essential foundation for future
research targeting the effects of other important solutes, such as chloride, sulfate, ammonia, chlo-
rine species, and "natural organic matter" (NOM) on copper corrosion. The model also helps ex-
plain the reduction in cuprosolvency brought about by aeration of many ground water supplies for
the removal of contaminants such as radon, VOC's, or iron.
To show at least initial evidence that the solubility model has promising practical validity,
appropriate experimental data from various other corrosion investigations performed by the Drink-
ing Water Research Division (DWRD) of the U.S. Environmental Protection Agency, and data
collected by other investigators were re-evaluated in the framework of the new model. New labora-
tory experiments in copper solubility were initiated, and furure research by DWRD with these and
other experimental systems will be used to further test the model and refine both the selection of
critical aqueous and solid species to be incorporated into the model, and constraints on the applica-
bility of the model to pipe loop corrosion tests and real water distribution systems covered by
USEPA regulations.
6
Copper Chemistry Overview—The Theory
Thermodynamic Data Used for Modeling
Table 1 gives a listing of a selected set of internally-consistent Gibbs Free Energies of forma-
tion ( ∆Gf°) for copper and related chemical species, along with their origin. Equilibrium constants
used in this work were computed for 25°C from ∆Gf° values from this table by the standard tech-
nique.14-16 Although some researchers have made predictions of copper corrosion behavior at el-
evated temperature17, adequate information is generally not available to characterize equilibrium
and solubility constants for many important copper compounds and complexes at temperatures
other than 25°C. Data are particularly lacking for the carbonate complexes (that dominate cupric
ion speciation above about pH 7), hydroxide solids, and phosphate species. Therefore, no attempts
at such modeling are presented in this paper. Research on Cu2(OH)2CO3 (malachite) solubility in
sea water has indicated that the net effect on copper solubility may be a significant decrease at low
temperature.18-20 but differences between sea water and most drinking waters affects the speciation
to a degree that quantitative extrapolation is unreliable. Part of the extrapolation problem results
from decreases in the solubility constants of some important copper solids as temperature increases,
analogous to the well-known case of calcium carbonate. However, the aqueous complexes that
dominate copper speciation do not necessarily follow the same equilibrium constant trends with
temperature as do the solids.
Computations of copper solubility in the following sections were made using a variety of soft-
ware on IBM-compatible personal computers. The software used included Lotus 1-2-3®† (Release
3.1+) for copper(I) computations, and Microsoft FORTRAN 5.0†† for copper (II) system calcula-
tions. A FORTRAN computer program, CU2SOL, was written specifically to compute the solubil-
ity of copper(II), following the pattern previously established with LEADSOL.16,21,22 Potential-pH
diagrams (“Pourbaix” diagrams) were computed using a version of the WATFIV computer pro-
gram of Froning, et. al.23 modified for use on the personal computers with the FORTRAN 5.0
(DOS) compiler. Graphical output was produced either manually or by importing ASCII-format
data files output from the PC programs or worksheet files from the spreadsheets. Graphing soft-
7
ware used included Freelance Graphics for DOS®†, SigmaPlot® for DOS and Windows††† (versions
5.01 and 2, respectively), PowerPoint®†† version 3.0, and Surfer®†††† for Windows packages.
SigmaPlot® for Windows and Surfer® for Windows were used for interpolation of copper concen-
trations in the construction of solubility “contour” diagrams.
TABLE 1. INTERNALLY-CONSISTENT GIBBS FREE ENERGIES OF FORMATION-COPPER AND
RELATED SPECIES AT 25°C
Species ∆Gf°kcal/mol ∆Gf°kJ/mol Reference
Cu(I) Aqueous Species
Cu+11.95 49.98 [1]
CuCl0-23.12 -96.72 [2]
CuCl2--58.29 -243.89 [3]
CuCl3--88.76 -371.35 [3]
Cu2Cl42- -115.71 -484.12 [3]
CuNH3+-2.44 -10.20 [4]
Cu(NH3)2+-15.46 -64.67 [4]
Cu(I) Solid Species
Cu(OH) -43.50 -182.00 [1]
Cu2O -35.01 -146.47 [5]
Cu(II) Aqueous Species
Cu2+ 15.65 65.48 [1]
Cu(OH)+-30.18 -126.28 [6]
Cu(OH)20-75.57 -316.21 [6]
Cu(OH)3--117.72 -492.55 [7]
Cu(OH)42- -157.14 -657.47 [8]
Cu2(OH)22+ -67.65 -283.03 [8]
Cu3(OH)42+ -151.49 -633.82 [8]
CuHCO3+-127.09 -531.73 [9]
CuCO30-119.84 -501.43 [9]
Cu(CO3)22- -251.19 -1050.98 [9]
CuCO3(OH)22- -206.00 -861.90 [10]
Cu(OH)CO3--161.43 -675.43 [11]
CuCl+-16.28 -68.10 [8]
CuCl20-46.95 -196.42 [12]
______________________
† Lotus Development Corporation, Cambridge, MA.
†† Microsoft Corporation, Redmond, WA.
††† Jandel Scientific, Inc., San Rafael, CA.
†††† Golden Software, Inc., Golden, CO.
8
TABLE 1 (Continued)
Species ∆Gf°kcal/mol ∆Gf°kJ/mol Reference
CuCl3--76.35 -319.44 [12]
CuHPO4+-250.14 -1046.57 [13]
CuH2PO4+-256.61 -1073.64 [13]
CuSO40-162.52 -692.53 [13]
CuNH32+ 3.80 15.92 [13]
Cu(NH3)22+ -7.21 -30.16 [13]
Cu(NH3)32+ -17.36 -72.65 [13]
Cu(NH3)42+ -25.72 -107.59 [13]
Cu(NH3)52+ -32.06 -134.14 [13]
CuNO3+14.97 62.63 [8]
Cu(NO3)2016.20 67.76 [8]
CuH2P2O7--468.47 -1960.08 [12]
CuHP2O70-463.16 -1937.88 [12]
Cu2P2O7-427.31 -1787.86 [12]
CuP2O72- -451.96 -1891.00 [1]
Cu(P2O7)26- -913.81 -3823.40 [1]
Cu(II) Solids
CuO -30.15 -126.16 [14]
CuO -30.62 -128.10 [14]
CuO -30.32 -126.84 [15]
Cu(OH)2-85.60 -358.16 [14]
Cu(OH)2-85.90 -359.41 [14]
Cu(OH)2-85.22 -356.56 [15]
CuCl2-56.29 -232.52 [13]
Cu2(OH)3Cl -160.09 -669.83 [13]
Cu4(OH)6Cl2 (atacamite) -320.51 -1341.00 [10]
Cu4(OH)6Cl2 (paratacamite) -320.27 -1340.00 [10]
Cu2(OH)2CO3 (malachite) -215.74 -902.68 [16]
Cu2(OH)2CO3 (malachite) -215.34 -900.96 [13]
CuCO3-123.67 -517.45 [12]
Cu3(CO3)2(OH)2 (azurite) -341.63 -1429.36 [12]
CuSO4 (chalcocyanite) -158.17 -661.80 [8]
Cu3(SO4)(OH)4 (antlerite) -345.75 -1446.60 [8]
Cu4(OH)6(SO4).H2O (langite) -488.53 -2044.00 [8]
CuSO4.5H2O -271.36 -1135.37 [12]
CuO.CuSO45.95 24.89 [12]
Cu4(OH)6SO4 (brochanite) -434.51 -1818.00 [10]
CuP2O7-448.11 -1874.91 [12]
Cu2P2O7-447.97 -1874.30 [8]
Cu3(PO4)2.2H2O -606.27 -2536.62 [12]
9
TABLE 1 (Continued)
Species ∆Gf°kcal/mol ∆Gf°kJ/mol Reference
Cu3(PO4)2-490.30 -2051.40 [12]
CuSiO3.H2O -1207.50 [10]
CuSiO3⋅2H2O -1443.90 [10]
Associated Dissolved Species
H+0
O2(g) 0
O2(aq) 3.92 16.40 [17]
OH--37.604 -157.34 [17]
H2CO3* -148.95 -623.21 [17]
HCO32- -140.28 -586.93 [17]
CO32- -126.19 -527.98 [17]
CO2(g) -94.257 -394.37 [17]
H3PO3-273.08 -1,142.57 [13]
H2PO4--269.80 -1,128.84 [13]
HPO42- -260.313 -1,089.15 [13]
H2O -56.690 -237.19 [17]
PO43- -243.48 -1,018.72 [13]
Cl--31.380 -131.29 [17]
F--67.34 -281.75 [17]
HSO4--180.66 -755.88 [17]
SO42- -177.95 -744.54 [17]
H4SiO40-312.6 -1,307.92 [17]
NO3--26.64 -111.46 [17]
HOCl0-19.1 -79.91 [17]
OCl--8.8 -36.82 [17]
Cl2(aq) 1.65 6.90 [17]
NH3(aq) -6.33 -26.48 [13]
References
1. Taken directly from: Wagman, D.D., et al., The NBS Tables of Chemical Thermodynamic Properties. Journal of Physical and
Chemical Reference Data, 1982. v. 11.
2. Cited in: Moffett, J.W. and R.G. Zika, Oxidation Kinetics of Cu(I) in Seawater: Implications for its Existence in the Marine
Environment. Marine Chemistry, 1983. 13: p. 239-251.
3. Computed from log
β
values given in: Fritz, J.J., Chloride Complexes of CuCl in Aqueous Solution. J. Phys. Chem., 1980.
84: p. 2241-2246.
4. Computed from conditional stability constants for I=2 mol /L tabulated on p. 301 of: Ringbom, A., Complexation in Analyti-
cal Chemistry. 1963, New York, NY: Wiley-Interscience, using ∆Gf° values from this table. Although not rigorously correct,
there should be little dependence of log
β
on ionic strength for these complexes because there is no change in charge. Hence,
these should provide reasonable estimates, given other data uncertainties and difficulties in correcting to I=0 from I=2 mol/L.
10
5. Computed from log
β
values given in: Baes, C.F., Jr. and R.E. Mesmer, The Hydrolysis of Cations. 1976, New York: Wiley-
Interscience.
6. Coputed from log
β
values given in: Paulson, A.J. and D.R. Kester, Copper(II) Ion Hydrolysis in Aqueous Solution. J. Solu-
tion Chem., 1980. 9(4): p. 269-277.
7. Computed from log
β
values given in: Vuceta, J. and J.J. Morgan, Hydrolysis of Cu(II). Limnol. & Oceanog., 1977. 22: p.
742-746.
8. Computed from log
β
values given in: Martell, A.E. and R.M. Smith, Critical Stability Constants. Vol. 5: First Supplement.
1980, New York, New York: Plenum Press.
9. Computed from log
β
values given in: Byrne, R.H. and W.L. Miller, Copper(II) Carbonate Complexation in Seawater.
Geochim. Cosmochim. Acta, 1985. 49: p. 1837-1844.
10. Either the only value tabulated, or the average of values tabulated in: Woods, T.L. and R.M. Garrels, Thermodynamic Val-
ues at Low Temperature for Natural Inorganic Materials: An Uncritical Summary. 1987, New York, New York: Oxford Univer-
sity Press.
11. Experimental log
β
data determined at the ionic strength of sea water of: Symes, J.L. and D.R. Kester, Copper(II) Interac-
tion with Carbonate Species Based on Malachite Solubility in Perchlorate Medium at the Ionic Strength of Seawater. Marine
Chemistry, 1985. 16: p. 189-211, extrapolated to infinite dilution (I=0) using activity coefficient data of Byrne and Miller [9] for
sea water, and estimating
γ
for the complex as 0.65.
12. Derived from data given in: Lindsay, W.L., Chemical Equilibria in Soils. 1979, New York, New York: John Wiley & Sons.
449, through accepting log
β
values, log K values, or computed ∆GR° and rederiving ∆Gf° by substituting values for other re-
actant or product species from this table.
13. Computed from log
β
values given in: Smith, R.M. and A.E. Martell, Critical Stability Constants. Vol. 4, Inorganic Ligands.
1976, New York: Plenum Press, using the Davies equation when necessary to extrapolate from values determined at I≤0.5 mol/L
to I=0.
14. Computed from log K or log
β
values given in: Schindler, P.W., Heterogeneous Equilibria Involving Oxides, Hydroxides,
Carbonates and Hydroxide Carbonates, in Equilibrium Concepts in Natural Water Systems. 1967, American Chemical Society:
Washington, DC., after correcting to I=0 by the Davies equation.
15. Computed from log K data given by: de Zoubov, N., C. Vanleugenhaghe, and M. Pourbaix, Copper, in Atlas of Electro-
chemical Equilibria in Aqueous Solutions. 1974, National Association of Corrosion Engineers: Houston, TX. p. 384-392.
16. Computed from log K data in: Symes, J.L. and D.R. Kester, Thermodynamic Stability Studies of the Basic Copper Carbon-
ate Mineral, Malachite. Geochim. Cosmochim. Acta, 1984. 48: p. 2219-2229.
17. From the table of ∆Gf° values in: Schock, M.R. and I. Wagner, The Corrosion and Solubility of Lead in Drinking Water, in
Internal Corrosion of Water Distribution Systems, 4, Editor. 1985, AWWA Research Foundation/DVGW Forschungsstelle: Den-
ver, CO.
Table 2 gives a listing of the fundamental equilibrium reactions incorporated into the Cu(I)
solubility spreadsheet and CU2SOL Fortran programs used to construct speciation and solubility
diagrams for this paper. These constants were computed from the ∆Gf° data from Table 1.
11
TABLE 2. EQUILIBRIUM REACTIONS IN COPPER SOLUBILITY PROGRAMS, AND
CORRESPONDING LOG K AND LOG
β
VALUES
Reaction log K or
β
at 25˚
Copper(I) Species
Cu˚
↔
Cu+ + e--8.760
Cu+ + Cl-
↔
CuCl-2.70
Cu+ + 2Cl-
↔
CuCl˚ 5.48
Cu+ + 3Cl-
↔
CuCl32- 4.81
2Cu+ + 4Cl-
↔
Cu2Cl42- 10.32
Cu+ + NH3˚
↔
CuNH3+5.9
Cu+ + 2NH3˚
↔
Cu(NH3)2+10.8
Cu2O(S) + 2H+
↔
2Cu+ + H2O -1.62
CuOH(S) + H+
↔
Cu+ + H2O 0.91
Copper(II) Species
Cu+
↔
Cu2+ + e--2.715
Cu2+ + H2O
↔
CuOH+ + H+-7.96
Cu2+ + 2H2O
↔
Cu(OH)2˚ + 2H+-16.24
Cu2+ + 3H2O
↔
Cu(OH)3- + 3H+-26.90
Cu2+ + 4H2O
↔
Cu(OH)42- + 4H+-39.56
2Cu2+ + 2H2O
↔
Cu2(OH)22+ + 2H+-10.58
3Cu2+ + 4H2O
↔
Cu3(OH)42+ + 4H+-20.76
Cu2+ + H+ + CO32-
↔
CuHCO3+12.13
Cu2+ + CO32-
↔
CuCO3˚ 6.82
Cu2+ + 2CO32-
↔
Cu(CO3)22- 10.60
Cu2+ + CO32- + H2O
↔
Cu(OH)CO3- + H+-4.25
Cu2+ + CO32- + 2H2O
↔
Cu(OH)2CO32- + 2H+-13.14
Cu2+ + 2H+ + PO43-
↔
CuH2PO4+21.09
Cu2+ + H+ + PO43-
↔
CuHPO4˚ 16.35
Cu2+ + SO42-
↔
CuSO4˚ 2.36
Cu2+ + NH3
↔
Cu(NH3)2+ 4.04
Cu2+ + 2NH3
↔
Cu(NH3)22+ 7.47
Cu2+ + 3NH3
↔
Cu(NH3)32+ 10.27
Cu2+ + 4NH3
↔
Cu(NH3)42+ 11.75
Cu2+ + 5NH3
↔
Cu(NH3)52+ 11.76
Cu2+ + Cl-
↔
CuCl+ 0.40
Cu2+ + 2Cl-
↔
CuCl˚ -0.12
CuO(s) + 2H+
↔
Cu2+ + H2O 7.98
Cu(OH)2(s) + 2H+
↔
Cu2+ + 2H2O 8.89
CuCO3(s)
↔
Cu2+ + CO32- -9.63
Cu2(OH)2CO3(s) + 2H+
↔
2Cu2+ + 2H2O +CO32- -5.48
Cu3(PO4)2 • 2H2O(s)
↔
3Cu3+ + 2PO43- + 2H2O -38.76
Cu3(PO4)2(s)
↔
3Cu3+ + 2PO43- -36.86
12
TABLE 2 (Continued)
Reaction log K or
β
at 25˚
Cu3SO4(OH)4(s) + 4H+
↔
3Cu2+ + SO42- + 4H2O 8.81
Cu4SO4(OH)6(s) + 6H+
↔
4Cu2+ + SO42- + 6H2O 15.38
Cu2(OH)3Cl(s) + 3H+
↔
2Cu2+ + Cl- + 3H2O 14.68
Cu4SO4(OH)6 • H2O + 6H+
↔
4Cu2+ + SO42- + 7H2O 17.34
Related Reactions
H2CO3*
↔
H+ + HCO3-6.352
HCO3-
↔
H+ + CO32- 10.329
H2O
↔
H+ + OH-13.995
H3PO4
↔
H+ + H2PO4-2.155
H2PO4-
↔
H+ HPO42- 7.207
HPO4-
↔
H+ + PO43- 12.346
NH4+
↔
H+ + NH3°-9.252
HSO4-
↔
H+ + SO42- 1.99
Oxidation Reactions of Copper
Copper metal from pipe and alloys is readily oxidized in contact with most drinking waters.
Copper may exist in water in either the monovalent copper(I) (cuprous) or divalent copper(II)
(cupric) valence states. The copper(III) and copper(IV) valence states are known to exist, how-
ever, they occur at high temperature and usually in the presence of high oxygen concentrations24
and are not expected in drinking water. No reports of their production under natural water and
water supply conditions have been located. In potable waters, copper metal may undergo the fol-
lowing two basic electrochemical transformations:
Cu(s) ↔Cu
+
+e
−
(1)
Cu
+
↔Cu
2+
+e
−
(2)
Because of the positive cell potentials for copper metal oxidation, copper pipe in water containing
dissolved oxygen will continue to corrode until all of the oxygen is depleted, or until precipitated
oxide films arrest the rate of corrosion.25 There is some evidence that the overall transformation
from Cu+ to Cu2+ is essentially the rate-limiting factor, with Cu+ existing essentially in reversible
equilibrium with the Cu metal at the pipe surface.26
In drinking waters, the oxidizing agents (electron acceptors) that will cause the corrosion of
metallic copper are predominantly dissolved oxygen and aqueous chlorine species.12,27,28 Follow-
ing are several possible net reactions for the corrosion of copper in drinking waters. These reac-
tions may be composed of several intermediate steps, any of which can be rate-controlling.
13
HOClo+H++Cu(s) ↔Cu2++Cl−+H2O
(3)
OCl
−
+2H
+
+Cu(s) ↔Cu
2+
+Cl
−
+H
2
O
(4)
Cu(s) +
12
O
2
+2H
+
↔Cu
2+
+H
2
O
(5)
The overall corrosion potential driving force for any of equations (3) through (5) may be com-
puted through the Nernst equation.14,15,29-31 For pure solids and in dilute aqueous solution, the ac-
tivities of the solid copper metal and that of water may be assumed to be equal to unity. Thus, the
expression (4), for example, becomes
E=E°−2.303 RT
2Flog
γ
Cl−Cl−
[]
γ
Cu2+Cu2+
[]
γ
OCl−OCl−
[]
γ
2
H
+
H
+
[]
2(6)
in Nernst form. After half-cell reactions are combined, their potentials added, and constants evalu-
ated, the resulting expression becomes:
E=1.38 −0.0296log
γ
Cl−Cl−
[]
γ
Cu2+Cu2+
[]
γ
OCl−OCl−
[]
γ
2
H
+
H
+
[]
2(7)
The symbols in equations (6) and (7) have their conventional meanings: E, the cell potential in
volts; E° the standard reduction potential for the overall reaction in volts; R the gas constant; T, the
temperature in Kelvins; and F, the Faraday constant. The number 2 in the denominator of (6)
reflects the 2 electron transfer for the oxidation reaction of copper metal to cupric ion in equation
(4). Similar Nernst equations can be derived and computed for alternate oxidants. Note also that
only the free ion activities are important to the equation, so complexation by carbonate, sulfate, etc.
must be computed to enable the isolation of the free hydrogen, chloride, and cupric ions that dictate
the voltage for this redox couple undergoing corrosion.
Many papers and texts have documented the impact of dissolved oxygen on copper oxidation
and dissolution rates. Several studies have also proven aqueous chlorine species have a significant
impact on the copper oxidation and corrosion rates.27,28,32-34 Free chlorine species (ie. HOCl°,
OCl-, Cl2) have not been conclusively shown to affect the equilibrium solubility of copper, other
than by influencing the valence state of the copper by its presence or absence.
Oxidants may have several other potential impacts on the observed copper levels in the water
and the nature of the passivating solids on the pipe. The effect of chlorine on the oxidation rate of
the copper metal might be alteration of the crystalline characteristics and porosity of the oxide
corrosion product film produced at the pipe surface, such as by reducing the formation of a protec-
tive Cu2O(s) underlayer by producing a high EH level, or by indirectly influencing scale structure
and conductivity through chloride formation as the chlorine is reduced.17 However, many of the
14
apparent effects of chlorine on copper solubility may merely result from accelerated corrosion
kinetics (rates), rather than changes in equilibrium conditions. That is, the rate of copper oxidation
and transport into solution may be accelerated by the oxidants, misinterpreted as representing a
greater equilibrium solubility. If the stagnation curve was controlled purely by ionic diffusion of
cupric or cuprous aqueous species, it would follow the model delineated by Kuch and Wagner,35
which would be independent of oxidant concentration above a threshold redox potential that causes
oxidation of the base copper piping.
The relative stabilities of Cu(I) and Cu(II) species in aqueous solution depend very strongly on
the nature of anions or other ligands present in the water. In pure water, cuprous ion is unstable in
the presence of an oxidizing agent such as oxygen, and cupric ion is formed by the disproportion-
ation reaction.12,24
2Cu+↔Cu(s) +Cu2+
(8)
Therefore, when cuprous species are present in water, they probably result from stabilization through
the formation of various aqueous complexes.
Specific research into the redox potential (ie., EH or pE) of drinking waters in general has not
been reported. Some indirect information to estimate the level for some systems may be gathered,
however. For disinfected systems, the redox potential will be dominated by the aqueous chlorine
(or other oxidant) species involved. Hypochlorous acid and hypochlorite ion have EH/pH stability
fields that are above the equilibrium line for water with oxygen in the atmosphere.14,30,31 This fact
explains the tendency for even pure water to lose its chlorine residual over time in plumbing.
Several studies have attempted to measure the redox potential (EH) of waters dosed with vary-
ing levels of chlorine, chloramines, ozone, or other disinfectants.36-39 The resulting EH values gen-
erally fell into the range of 0.6 to 1.2 V versus SHE (standard hydrogen electrode). The potentials
were a function of the individual or mix of disinfectant species, their concentrations, and time. As
expected, this EH range is higher than that attainable with typical dissolved oxygen levels in drink-
ing waters. Most surface water supplies will have undergone disinfection, so their redox potentials
will range from a low controlled by the dissolved oxygen in the system, to a high controlled by the
disinfection species.
For domestic or municipal water systems using wells, however, the redox potential could en-
compass a wider range. Other oxidation processes, such as greensand filtration or aeration (for
iron and manganese removal) may cause the resulting redox potential to remain higher and the
residual oxidant concentrations may be stable longer. Granular activated carbon (GAC) removes
many substances that consume oxidants. Other substances that can influence corrosion and lower
redox potentials include organic matter, nitrogen species, sulfur species, iron species, or even mi-
croorganisms.40,41
Considerable attention has been paid to factors governing the redox potential of natural sys-
tems, and problems inherent in its measurement.15,40,41 A precise range of redox potential values for
15
the wide variety of ground water conditions across the United States cannot be given, but some
limits can be estimated. In the pH range of 5 to 8, then, EH values could likely be in the neighbor-
hood of -0.2 V as a general lower limit.
Copper plumbing, therefore, can be exposed to a wide environment of EH-pH conditions. This
environment evolves, as oxidation and reduction reactions occur through corrosion and deposition
processes for different lengths of stagnation times, and in different diameters of pipes. Therefore,
all three normal copper valence states (metal, +1, +2) could reasonably occur under drinking water
conditions either geographically or with time.
The Conceptual Role of Aqueous Complexation
The inclusion of appropriate aqueous complexes in the chemical model is essential to enable
accurate prediction of metal solubilities. For copper, a more than 100-fold negative error in pre-
dicted equilibrium copper solubility is introduced by the neglect of the Cu(OH)2° complex, even if
other hydrolysis species are included.42 An even greater error is introduced in both the magnitude
and trend of solubility when only the free cupric ion (Cu2+) is assumed to be present. In virtually all
drinking waters, at least some DIC (dissolved inorganic carbonate) is present. Because Cu(II)
(cupric form) forms several very strong and some weaker aqueous complexes with carbonate spe-
cies (HCO3-, CO32-), carbonate complexation tends to become even more important that the hy-
drolysis reactions as the pH increases, especially above a pH of about 7.
Copper (I) Chemistry
In the absence of oxygen and added oxidants, water does not corrode copper. That is because
the immune region for copper corrosion extends well above the minimum EH for water stabil-
ity.12,29,43 With the addition of electron acceptors to the solution (eg. HOCl°, O2), either or both of
the oxide solids Cu2O (cuprite) or Cu(OH) may form. Given the variety of different crystalline
oxides of these solids that may form, differences in reported solubilities are significant enough to
cause prediction uncertainties for cuprous solubility of roughly 10 to 100. Additionally, the condi-
tions under which protective oxide layers form are poorly known.29 Identifying the solid is further
complicated by oxidation and dehydration that might accompany some of the normal sample prepa-
ration steps for analysis by X-Ray diffraction or other instrumental techniques. The transformation
from cuprous hydroxide to cuprous oxide at the pipe surface is also likely to be affected by hydro-
dynamic factors, and both the nature and concentrations of ions in solution. Some research has
indicated that Cu2O(s) may be formed below the surface of existing corrosion byproduct deposits
under oxidizing conditions, through rapid Cu(OH)2(s) dissolution followed by disproportionation
and redeposition of Cu2O(s)44. Hydrodynamic conditions are very important in this situation, which
might also affect the crystallinity of the Cu(OH)2(s) and its conversion to more stable CuO(s).45
This is most favored in fractures and crevices of the surface film, and helps explain the formation of
layered copper corrosion deposits that have cupric solids on the surface with Cu2O(s) underneath.
Le Gal La Salle, et. al. have also done electrochemical studies that indicate a cyclic transformation
of rapidly formed Cu+ and Cu2+ cations, compact Cu2O(s) layer, and then the reoxidation and for-
mation of a porous external layer of CuO(s) or Cu(OH)2(s).46
16
Cuprous ion forms very weak hydrolysis species (ie. CuOH°), if any. No species of this type
are generally reported in relevant reviews of copper chemistry,29,47-49 so they are not included in this
solubility model. The formation of bicarbonate and carbonate complexes (CuHCO3° and CuCO3-,
respectively) have been hypothesized to explain some trends in rates of oxidation by dissolved
oxygen, but no experimentally determined values or ones estimated from thermodynamic prin-
ciples are available.50
Despite these uncertainties, cuprous aqueous chemistry appears to be dominated by the forma-
tion of several stabilizing complexes. Notably, two are particularly significant in drinking water:
NH3 and Cl-. Cuprous ammine complexes can be formed either directly, or by reduction of cupric
ammine complexes by a route such as this:24
Cu(NH
3
)
42+
[]
+Cu(s) ↔2Cu(NH
3
)
2
+
[]
(9)
Also, complexes of Cu(I) are generally formed by direct interaction of ligands with copper(I) ha-
lides and reduction of Cu(II) compound, or reduction of Cu2+ in presence of the ligand.24 Rickard
has concluded from various complexation experiments and observations of geochemical mineral
assemblages that even in the presence of strong oxidants, ammonia can increase the solubility of
oxide, hydroxide, oxysulfate, oxycarbonate and oxychloride solids of copper.51,52 This results from
the stabilization of the cuprous form of copper in solution through the formation of the
diamminecuprous ion, [Cu(NH3)2]+. Similarly, in the presence of fairly high concentrations of
ammonium salts, the domain of stability of dissolved copper(I) species, and hence the domain of
corrosion, expands considerably.29,52 Severe attack of copper pipe attributed to the presence of high
concentrations of ammonia in the soldering flux has also been noted.53
Cuprous chloride complexes are not as strong as the ammine complexes, but they may be sig-
nificant because chloride concentrations are many times those of ammonia in most drinking wa-
ters. They have been shown to play an important role in retarding the rate of cuprous ion oxidation
in natural waters, particularly at neutral to alkaline pH’s.50,54-57 Presumably, this effect would also
apply to the drinking water environment under similar chemical conditions. The equilibrium solu-
bility of cuprous copper, ST,Cu(I), may be described as the sum of the concentrations of all dissolved
cuprous species:
ST,Cu(I)=Cu+
[]
+CuCl°
[]
+CuCl2−
[]
+CuCl32−
[]
+Cu2Cl42−
[]
+Cu(NH3)+
[]
+Cu(NH3)2+
[]
(10)
In which [ ] (brackets) indicate concentration in mol/L. The total solubility may then be computed
by standard techniques14,15,21,30,31. The concentrations of all aqueous species may be expressed in
terms of the concentrations of the metal and ligand and the formation constant for the complex,
valid for a particular temperature and ionic strength. Thus, equation (10) can be rewritten as:
17
S
T,Cu( I )
=Cu
+
[]
+β’
Cl,11
Cu
+
[]
Cl
−
[]
+β’
Cl,12
Cu
+
[]
Cl
−
[]
2
+β’
Cl,13
Cu
+
[]
Cl
−
[]
3
+
β’
Cl,24
Cu
+
[]
2
Cl
−
[]
4
+β
NH
3,
11
Cu
+
[]
NH
3
[]
+β’
NH
3
12
Cu
+
[]
NH
3
[]
2
(11)
in which
β
’ represents the conditional overall formation constant corrected for ionic strength and
temperature.
For the purpose of this calculation, the formation of the solid CuOH(s) is assumed to control the
activity of free cuprous ion at equilibrium as follows:
Cu
+
[]
=
K
s,CuOH
H
+
[]
γ
H
+
γ
Cu
+
(12)
Given that the activity of water and the solid are unity, the activity coefficient terms can be incorpo-
rated into a conditional solubility constant, K’s,CuOH, analogous to the conditional stability constants
used in expression (11). The resulting expression (13) may be substituted for each of the [Cu+]
terms in equation (11), to yield the final equation programmed for calculation.
Cu+
[]
=K’
s,CuOH H+
[]
(13)
Figures 1 and 2 show the effect of chloride and ammonia complexation on copper (I) solubility
at 25°C, assuming CuOH to be the solubility-controlling solid. If the solid Cu2O phase (cuprite)
controlled the solubility, it would lower the predicted Cu(I) solubility by a factor of about 60 to
100. However, it would not affect the trends of the solubility curves. Clearly, when oxidizing
power in the solution is absent, levels of chloride and ammonia found in many water supplies could
significantly increase soluble copper concentrations.
Copper(II) Hydrolysis Reactions
Major inconsistencies exist among most published hydrolysis and complexation formation con-
stants for the cuprous ion, particularly for the complex Cu(OH)2°. Reported values for log
β
have
been found ranging from -13.7 to “less than-17.3” for the reaction:
Cu
2+
+2H
2
O↔Cu(OH)
2o
+2H
+
(14)
To compound the problem, data either conflict or there are very few reports of any determina-
tions of stability constants for the possible complexes Cu(OH)3- and Cu(OH)42-. These complexes
could be of significance in waters of pH of approximately 9 and above. Thus, these values must be
treated with some skepticism. The confusion is exemplified by the critical evaluation of existing
data done by Baes and Mesmer.47 They could not select what they considered to be definitive
values for the formation constants of CuOH+, Cu(OH)2°, or Cu(OH)3-, all of which could be either
18
Figure 1. Effect of chloride complexation on CuOH(s) solubility, I=0.02, 25°C.
19
Figure 2. Effect of ammonia complexation on CuOH(s) solubility, I=0.01, 25°C.
20
directly or indirectly significant in drinking waters. They only chose to report limits on the likely
magnitudes of the respective log
β
constants (eg. “<-8,”, “<-17.3”, “<-27.8”), which unfortunately
encompasses almost all of the values reported for these species. Very little additional research has
been conducted in recent years. Table 3 indicates some of the reported log
β
values in the literature
for the Cu(OH)2° complex, and illustrates the uncertainty of any selected value.
TABLE 3. A SELECTION OF REPORTED LOG
β
12 VALUES FOR THE REACTION:
Cu2+ + 2H2O
↔
Cu(OH)2˚+ 2H+
Log
β
12 Reference
<-17.3 [1]
-16.22 [2]
-16.24 [3]
-14.70 [4]
-13.7 [5]
References
1. Baes, C. F. & Mesmer, R.E. The Hydrolysis of Cations. Wiley-Interscience, New York (1976).
2. Sunda, W. G. & Hanson, P.J. "Chemical Speciation of Copper in River Water". Chemical Modeling in Aqueous Systems,
ASC Symposium Series No. 93, American Chemical Society, Washington, D.C. (1979).
3. Paulson, A. J. & Kester, D. R. "Copper (II) Ion Hydrolysis in Aqueous Solution". J.Solution Chem. 9:4:269 (1980).
4. Mahapatra, S. and Subramanya, R.S. Proc. Indian Acad. Sci. 65:283 (1965).
5. Vuceta, J. and Morgan, J. J. "Hydrolysis of Cu(II)". Limnol. & Oceanog. 22:742 (1977).
The background solubility of cupric copper under oxidizing conditions resulting from the un-
avoidable hydrolysis reactions is completely defined by the simple total solubility expression (ST,OH),
which includes the free cupric ion concentration [Cu2+].
S
T,OH
=Cu
2+
[]
+CuOH
+
[]
+Cu(OH)
2°
[]
+Cu(OH)
3−
[]
+
Cu(OH)
4
2−
[]
+2Cu
2
(OH)
2
2+
[]
+3Cu
3
(OH)
4
2+
[]
(15)
After substituting rearranged complexation expressions from Table 2 in terms of concentra-
tions with equilibrium constants corrected for ionic strength and temperature, as was done for
Equation 11, the complete expression is:
21
S
T,OH
=Cu
2+
[]
+
β’
1,1
Cu
2+
[]
H
+
[]
+
β’
1,2
Cu
2+
[]
H
+
[]
2
+
β’
1,3
Cu
2+
[]
H
+
[]
3
+
β’
1,4
Cu
2+
[]
H
+
[]
4
+2
β’
2,2
Cu
2+
[]
2
H
+
[]
2
+3
β’
3,4
Cu
2+
[]
3
H
+
[]
4
(16)
assuming the activity of H2O is 1.
Copper(II) Complexation by Carbonate
Considerable research has been devoted to the study of the aqueous speciation of copper in
natural waters and seawater, in which the role of carbonate species was extremely important.19,20,51,58-
61 The concept of the significance of carbonate complexation was then applied to potable waters by
Schock, with indications that high DIC levels would aggravate cuprosolvency.22,62 Most of these
solubility models presumed that Cu2(OH)2CO3 (malachite) was the solid phase governing equilib-
rium solubility. Interestingly, though Pourbaix and colleagues did not originally consider carbon-
ate complexes in their pioneering work on EH-pH relationships involved in copper corrosion,29,63
they were ultimately added to the calculations.43 Unfortunately, few engineers and corrosion scien-
tists in the United States noticed the revision of the EH-pH diagrams, so much misinformation about
improved passivation through substantial DIC addition has continued to appear in articles, text-
books, and manuals.
The selection of the aqueous carbonate species to be incorporated into this solubility model
follows the selections of Byrne and Miller60 and Stiff.64 There is a considerable spread in the
reported constants for Cu(CO3)22- and CuHCO3+, as is shown in Table 4, which lists many of the
reported log
β
values for cupric bicarbonate, carbonate, and hydroxycarbonate complexes. Ex-
ploratory calculations showed that the CuCO3(OH)22- complex species reported in an uncritical
compilation of ∆Gf° values, may have an important impact on copper solubility at pH levels found
in many lime-softened or other high-pH water supplies. Therefore, it was included in this revised
model, pending speciation information to the contrary. The contribution to copper(II) solubility
from carbonate species, ST,CO3 is described by the expression:
ST,CO3=CuHCO3+
[]
+CuCO3°
[]
+Cu(CO3)2
2−
[]
+
CuCO3OH−
[]
+Cu(CO3)2(OH)2
2−
[]
(17)
22
TABLE 4. A SELECTION OF REPORTED LOG EQUILIBRIUM CONSTANTS (SS) FOR COPPER (II) CARBONATE
COMPLEXES. ALL VALUES HAVE BEEN REPORTED FOR 25˚C AND I=0, OR WERE EXTRAPOLATED
TO I=0 BY THE DAVIES EQUATION FOR I≤0.5
Log
β
†
CuHCO3+CuCO3˚ Cu(CO3)22- CuCO3OH-Reference
12.41 ———[1]
12.13 6.82 10.6 -4.5 [2]
— 6.86 10.71 — [3]
12.53 ———[4]
— 6.71 9.01 — [5]
— 6.75 — — [5]
— 6.73 9.83 — [6]
— 6.74 10.23 — [7]
— 6.80 — — [8]
— 6.34 — — [9]
— 6.77 10.11 — [10]
12.43 ———[11]
11.63 — — — [12]
— — — -4.25 [13]
— 6.75 10.69 — [14]
13.83 6.89 — — [15]
13.03 — — — [16]
†For the reactions
Cu2+ + H+ + CO32-
↔
CuHCO3+ + H2O
Cu2+ + CO32-
↔
CuCO3˚
Cu2+ + 2CO32-
↔
Cu(CO3)2-2-
Cu2+ + H2O + CO32-
↔
CuCO3OH- + H+
References
1. Bauman, J.E., Jr. Thermodynamic Measurements of Carbonate Equilibria Involving Metal Ions. in Proceedings of the
Workshop on Techniques for Measurement of Thermodynamic Properties. 1981. Albany, OR: Bureau of Mines Information
Circular/1981 IC 8853.
2. Byrne, R.H. and W.L. Miller, Copper(II) Carbonate Complexation in Seawater. Geochim. Cosmochim. Acta, 1985. 49: p.
1837-1844.
3. Bilinski, H., R. Huston, and W. Stumm, Average of 2 values given in: Determination of the Stability Constants of some
Hydroxo and Carbonato Complexes of Pb(II), Cu(II), Cd(II), and Zn(II) in Dilute Solutions by Anodic Stripping Voltammetry and
Differential Pulse Polarography. Analytica Chimica Acta, 1976. 84: p. 157-164.
4. Fouillac, C. and A. Criaud, Carbonate and Bicarbonate Trace Complexes: Critical Reevaluation of Stability Constants.
Geochem. Jour., 1984. 18: p. 297-303.
5. Ernst, R., H.E. Allen, and K.H. Mancy, Characterization of Trace Metal Species and Measurement of Trace Metal Stability
Constants by Electrochemical Techniques. Water Research, 1975. 9: p. 969-979.
6. Schindler, P., M. Reinert, and H. Gamsjäger, Löslichkeitskonstanten und Freie Bildungsenthalpien von Cu2(OH)CO3
(Malachit) und Cu3(OH)2(CO3)2 (Azurit) bei 25•C. Helvetica Chim. Acta, 1968. 51(2): p. 1845-1856.
7. Sunda, W.G. and P.J. Hanson, Chemical Speciation of Copper in River Water, in Chemical Modeling in Aqueous Systems, Ch.
8, E.A. Jenne, Editor. 1979, American Chemical Society: Washington, DC. p. 147-180.
8. Stiff, M.J., Copper/ Bicarbonate Equilibria in Solutions of Bicarbonate Ion at Concentrations Similar to Those Found in
Natural Water. Wat. Res., 1971. 5: p. 171-176.
23
9. Scaife, J.F., The Solubility of Malachite. Canadian Journal of Chemistry, 1957. 35: p. 1332-1340.
10. Silman, J.F.B., The Stabilities of Some Oxidized Copper Minerals in Aqueous Solutions at 25˚ C and 1 Atmosphere Total
Pressure. 1958, PhD dissertation, Harvard University:
11. Mattigod, S.V. and G. Sposito, Estimated Association Constants for Some Complexes of Trace Metals with Inorganic
Ligands. Soil Science Society of America Journal, 1977. 41: p. 1092-1097.
12. Sylva, R.N., The Environmental Chemistry of Copper(II) in Aquatic Systems. Wat. Res., 1976. 10: p. 789-792.
13. Symes, J.L. and D.R. Kester, Thermodynamic Stability Studies of the Basic Copper Carbonate Mineral, Malachite. Geochim.
Cosmochim. Acta, 1984. 48: p. 2219-2229.
14. Turner, D.R., M. Whitfield, and A.G. Dickson, The Equilibrium Speciation of Dissolved Components in Freshwater and
Seawater at 25•C and 1 atm. Pressure. Geochimica et Cosmochimica Acta, 1981. 45: p. 855-881.
15. Zuehlke, R.W. and D.R. Kester, Ultraviolet Spectroscopic Determination of the Stability Constants for Copper Carbonate
and Bicarbonate Complexes up to the Ionic Strength of Seawater. Marine Chemistry, 1983. 13: p. 203-226.
16. Zirino, A. and S. Yamamoto, A pH-Dependent Model for the Chemical Speciation of copper, Zinc, Cadmium, and Lead in
Sea Water. Limnology and Oceanography, 1972. 17: p. 661-671.
In an analogous manner to the cupric hydrolysis reactions above, the concentrations of each of
these complexes can be expressed in terms of only [Cu2+], [H+], and [CO32-], as follows:
Cu
2+
+CO
32−
+H
+
↔CuHCO
3+
(18)
Cu
2+
+CO
32−
↔CuCO
3°
(19)
Cu
2+
+2CO
32−
↔Cu(CO
3
)
2
2−
(20)
Cu
2+
+CO
32−
+H
2
O↔CuCO
3
OH
−
+H
+
(21)
Cu
2+
+Co
32−
+2H
2
O↔CuCO
3
(OH)
2
2−
+2H
+
(22)
Making the rearrangement and corresponding substitutions of equations (18) through (22) into
equation (16) based on reactions written by this convention yields equation (23):
ST,CO3=
β
’1,1,1 Cu2+
[]
H
+
[]
CO32−
[]
+
β
’
1,0,1 Cu2+
[]
CO32−
[]
+
β
’
1,0,2 Cu2+
[]
CO32−
[]
2
+
β
’
1,−1,1 Cu2+
[]
CO32−
[]
H
+
[]
+
β
’
1,−2,1 Cu2+
[]
CO32−
[]
H
+
[]
2(23)
in which the formation constants (
β
') are corrected for ionic strength and temperature, and the
negative numbers in the subscript for hydrogen ion indicate hydroxide ions in the complex. The
24
total solubility of copper(II) in pure water containing carbonate as the only complexing family of
ligands is then given by ST,Cu(II):
ST,Cu(II)=ST,OH +ST,CO3(24)
Passivating Cupric Oxide and Hydroxide Solids
The quantitative prediction of copper levels in drinking water, made much more important by
recent U. S. Environmental Protection Agency regulations,1-3 relies heavily on the solubility and
physical properties of cupric oxide, hydroxide and basic carbonate solids that comprise most scales
in water supplies. Because of considerable uncertainty in the values for the solubility constants of
all of these minerals, the tendency for prolonged existence of thermodynamically metastable phases
such as Cu(OH)2(s), and the possibly slow formation rate of Cu2(OH)2CO3(s) (malachite), quantitation
becomes highly problematic.
The differences among several solubility constants for CuO(s) and Cu(OH)2(s) have been dis-
cussed by de Zoubov, et. al., who selected two constants that they felt best agreed with actual
copper solubility data.29 In their critical evaluation of hydrolysis behavior of almost all important
environmental metals, Baes and Mesmer47 selected somewhat different values, which were an out-
growth of the research of Schindler, et. al. on particle size effects and the solubility of oxides and
hydroxides.31,65,66 This intriguing work by Schindler et. al. on the relationship of the solubility and
solubility constant for CuO(s) and Cu(OH)2(s) to a parameter called “molar surface” (units of m2)
may provide another avenue of important insight into the discrepancies among the reported ther-
modynamic constants.65,66 Essentially, the molar surface is related through various parameters to
the surface area of particles of a solid, such that as the size of the particles decreases, the molar
surface increases. Increases in the molar surface are physically related to the thermodynamic prop-
erty Gibbs free energy, which governs the intrinsic solubility of the solid and its stability relative to
other crystalline forms of the same solid.31,65,66 Schindler, et. al. found the log equilibrium constant
for the solubility reaction
Cu(OH)
2
(s)+2H
+
↔Cu
2+
+2H
2
O
(25)
to vary from 8.92 to 9.13 at I=0.2, as the molar surface increased from 0 to 4570 m2. If these
constants are corrected to zero ionic strength, using the Davies equation,15,41,67 the range is 8.67 to
8.89, respectively. Similarly, they found the log equilibrium constant for the tenorite solubility
reaction
CuO(s ) +2H
+
↔Cu
2+
+H
2
O
(26)
to vary from 7.89 to 8.27 at I=0.2, as the molar surface rose from 0 to 4340 m2. Correcting for ionic
strength using the Davies equation,15,41,67 the range is 7.64 to 7.98. Particle size effects may there-
fore have significantly contributed to scatter in some of the reported experiments that determined
equilibrium constants for these solids.
25
Particle size effects may result in a transition in solubility control from one hydrated solid to a
dehydrated form as crystal growth occurs. The difference between the Cu(II) dibasic solids CuO(s)
and Cu(OH)2(s) is essentially one water molecule of hydration.31 The solubility constant determi-
nations (and, therefore, ∆Gf° determinations), show that at equilibrium, even large particles of
cupric hydroxide can never be stable relative to cupric oxide. With smaller particle sizes, the molar
surface of the solid phase is correspondingly larger, so there is a point of crystallite size where
cupric oxide is thermodynamically unstable relative to the hydroxide.31,65,66 In practical situations
these differences in particle size and solubility could easily change actual equilibrium copper levels
by a factor of more than two.
The thermodynamic measurements and arguments do not predict anything about the reaction
rates of the conversion and crystal growth of the solids. For example, metastable
γ
-FeOOH(s) is
known sometimes to take millions of years to convert to the thermodynamically-stable coarse-
grained Fe2O3(s) in sediments.31 Reported log Ksp values for freshly-precipitated “Fe(OH)3” vary
by 3 orders of magnitude.15
Adeloju and Hughes suggest that the formation of solid Cu(OH)2(s) is kinetically favored over
the formation of CuO(s) solid.17 Many researchers have observed significant Cu(OH)2(s) to be
produced by corrosion at anodic locations in electrochemical experiments.44,45,68-71 These experi-
mental observations, coupled with the levels of copper observed in many sampling programs and
pipe rig experimental systems, supplement the argument that at least for relatively new plumbing
systems, the use of Cu(OH)2(s) in the solubility model is more realistic than CuO(s). Metastability
may be particularly important for controlling copper in plumbing systems that do not have thick,
adherent aged coatings on them.
The transitioning dehydration process from freshly-precipitated, amorphous bluish-green cu-
pric hydroxide, through a brownish color into black CuO has been studied by Patterson, et. al. for
alkaline precipitation of copper from dilute cupric nitrate solutions.72 They found that long-term
aging (about a month) experiments corresponded well to the mechanism proposed by Schindler, et.
al. for growth in crystallite size until cupric oxide became more stable than cupric hydroxide.65,66
In assessing the probable accuracy or reliability of solubility constants, one unavoidable inter-
action is the selection of the stability constants of the background hydrolysis species, eg. CuOH+,
Cu(OH)2°, Cu(OH)3-, etc., used in deriving the solubility constant from experimental data. While
other interfering species (such as carbon dioxide, ammonia, chloride) may be excluded from ex-
perimental systems, nothing can be done to avoid hydroxide complexes, except operating at low
pH. That is clearly not a viable solution for relevant study of oxide and hydroxide solids, because
of likely differences in kinetics, surface properties, and speciation differences resulting from high
concentrations of metal. As noted previously, Baes and Mesmer indicated problems with several
formation constants for cupric hydroxide complexes, and they recommended limiting values for
these three complexes, rather than precise numbers.47 Therefore, only a limited amount of confi-
dence can be placed in the reported Ksp for any cupric oxide or hydroxide solid.
26
The “Cupric Hydroxide Model”
Using the expression of equation (24), a variety of solubility diagrams were computed using the
CU2SOL computer program for 25°C, assuming several concentrations of DIC and different ionic
strengths. Initially, the diagrams were computed assuming the free cupric ion activity was fixed by
equilibrium with a Cu(OH)2 solid with the highest molar surface (see Tables 1 and 2) by the expres-
sion:
Cu
2+
[]
=K’
s,Cu(OH )
2
H
+
[]
(27)
in which log Ks,Cu(OH)2 was 8.89. Because the maximum ionic strength used in any of the calcula-
tions was sufficiently low, the activity of water and of the solid were assumed to be unity for all
calculations. The solubility constant for the solid with the highest molar surface was initially
chosen for the model because it probably best represents the initial cupric oxide solid formed in the
first few months to years of exposure of pure copper pipe to drinking water.
Figure 3 illustrates a three-dimensional surface of the effect of DIC and pH on cupric hydroxide
solubility. Alternate selections of solubility constants for Cu(OH)2(s), or even CuO(s), would yield
the same trend, just shifted up or down. To provide a more quantitative estimate of the trends of
solubility, a contour diagram was prepared by fitting a grid to copper solubility calculations from
DIC 0 to 150 mg C/L and pH 6 to 11, and contouring the interpolated surface, shown in Figure 4.
Contour diagrams were constructed using the software packages SigmaPlot® for Windows (2.0)
and Surfer® for Windows. The default contouring by SigmaPlot® with logarithmic scaling of inter-
vals was found to be adequately accurate for most purposes, although predicted equilibrium con-
centrations read from such diagrams should not be considered to be very precise and be
overinterpreted. When using some other software, such as Surfer®, careful application of the avail-
able grid interpolation and contouring algorithms was necessary, to avoid mathematical artifacts
and misleading interpretations of solubility behavior. After some experimentation and hand-checking
of computed values from CU2SOL versus those obtained after gridding and contouring, the best
results were obtained using the minimum curvature gridding method, a grid size of 101 by 101,
normal search method, and a specified maximum error of less than or equal to 0.0005 of the surface
relief. A maximum error of ±0.02 in log(mg Cu/L) for some pH values below DIC of 5 mg/L was
still found. However, above that DIC, diagram interpolation errors are all less than 0.001 for
log(mg Cu/L) for the entire pH range of the figure. Discrepancies between the displayed interpo-
lated values and the raw computed data were somewhat greater for the diagrams with the logarith-
mic scaling displayed. However, the convenience of the direct reading of concentrations from such
diagrams was judged to outweigh the value of the higher display precision, and these diagrams
were selected for this report.
An increase in copper solubility with lower pH and higher DIC is evident from the figures.
Above a pH of approximately 9.5, an upturn in solubility is predicted, caused by carbonate and
hydroxide complexes increasing Cu(OH)2(s) solubility. In the pH range of approximately 7 to 9,
significant increases in copper solubility are predicted from the addition of even small amounts of
2
27
Figure 3. Three-dimensional surface plot of the effect of DIC and pH on copper (II) solubility,
assuming equilibrium with Cu(OH)2 having a large molar surface (I=0.02, 25°C).
28
Figure 4. Contour diagram showing the effect of DIC and pH on copper(II) solubility, assuming
equilibrium with Cu(OH)2 solid having a large molar surface (I=0.02, 25°C).
29
carbonate, although maximum solubility remains less than about 0.3 mg/L. The largest relative
impact on increasing copper solubility occurs where the contour lines are most nearly parallel to
the pH axis, which is in the first approximately 15 mg/L of DIC. The effect remains considerable
even above pH 9, but the overall solubility is so low that the enhancement is not of much practical
significance.
To understand the roles of the different aqueous species better, Figures 5a and 5b were plotted.
These show the predicted aqueous speciation for two concentrations of DIC, 4.8 and 96 mg C/L (4
x 10-4 mol/L and 8 x 10-3 mol/L) assuming the formation of the Cu(OH)2 solid with high molar
surface.66 These DIC levels generally represent waters of moderate to very high alkalinity, respec-
tively.
Clearly, carbonate complexation dominates copper(II) solubility in most drinking waters over
pH 7, and is the key to developing effective cuprosolvency reduction strategies. In the pH and
carbonate concentration ranges of most drinking water, the figures clearly show that complexes
with carbonate species are the dominant ones. At a pH of approximately 9.5, the complex
CuCO3(OH)22- becomes important in enhancing the solubility of Cu(OH)2(s). The prediction of
increased solubility by waters of increasing DIC is supported by the results of many research and
field studies.73-79
Computations over the ionic strength range of 0.001 to 0.02 were performed to assess the sen-
sitivity of the solubility calculations to this parameter. Because of the dominance of uncharged
aqueous species over much of the pH/DIC range, the effect is generally negligible for practical
purposes. At low DIC concentrations and at low pH (eg. DIC=5 mg C/L at pH 6.5) the effect is
largest, amounting to approximately 2 mg/L. The impact is rapidly reduced to differences on the
order of only one to several parts per billion by pH 8, for all DIC levels checked. The effect is less
at higher DIC's because of the prevalence of the CuCO3° complex.
A study by KIWA tested the impact of several water quality variables on carefully exhumed
copper pipe, having well-developed scale layers.76 They found the maximum stagnation copper
levels (CuMAX) were approximately linearly related to DIC, pH and sulfate concentration by the
equation
CUMAX =0.52DIC −1.37 pH +2SO42−
[]
+10.2 (28)
in which CUMAX is in mg Cu/L, but DIC and sulfate concentrations are in mmol/L. The signs in
the equation show that DIC and sulfate both tend to increase copper solubility, but increased pH
significantly reduces solubility. Their finding is in good agreement with the predictions of this
model.
The Malachite Problem
According to previous corrosion and natural water modeling studies, the solid most likely to be
controlling the solubility of Cu(II) in drinking water should be the basic cupric carbonate mineral
30
Figure 5. Copper(II) speciation assuming formation of solid Cu(OH)2; a) DIC=4.8 mg C/L
I=0.005, 25°C; b) DIC=96 mgC/L, I=0.02, 25°C.
31
32
Cu2(OH)2CO3(s) (malachite). Although basic copper chlorides and sulfates, such as langite,
posnjakite, brochantite and atacamite may be formed in some circumstances, their significance in
uniform corrosion of copper has not been adequately established. Some of these solids, and others
like CuCl2, CuCl and Cu2(OH)3Cl, may be important in pitting processes, however. Complicating
the problem is that traditional “eyeball” identification of malachite by its blue-green color is ex-
tremely unreliable, because almost all cupric hydroxysulfates, hydroxycarbonates, hydroxychlorides,
and even fresh cupric hydroxide can be some shade of blue-green.
There are considerable differences among reported values for the solubility constant of
Cu2(OH)2CO3(s) (malachite). For the reaction written as:
Cu
2
(OH)
2
CO
3
(s)+2H
+
↔2Cu
2+
+2H
2
O+CO
32−
(29)
the most widely-reported values for log K range from -5.16 to -6.20. One study, including some
discussion of temperature and redox potential dependence of copper corrosion, used a value that
corresponds to -3.99 in the form of equation (29),17 an extreme difference. The most recent careful
experimental work by Symes and Kester20 leads to a value of log K = -5.48, which has been used for
most of the modeling reported here. Little compelling evidence currently exists for the selection of
this value over that of Schindler, et. al.80 of -5.18. Thus, the uncertainty in the computed copper
concentration in equilibrium with malachite is at least about a factor of 2, aside from uncertainties
in aqueous speciation, until further experimental data focusing on this problem is generated.
Figures 6 and 7 show solubility surface and contour diagrams for copper(II), corresponding to
equilibrium with either Cu2(OH)2CO3(s) (malachite) or CuO(s) (tenorite), whichever is thermody-
namically stable at a given pH/DIC point. These diagrams were constructed in the same manner as
Figures 3 and 4. The malachite solubility constant from Symes and Kester was used for both
Figures 6 and 7.20 The solubility constant used for CuO(s) (tenorite) represented the result for the
thermodynamically most stable dibasic cupric oxide solid with the smallest molar surface.66
Cu2(OH)2CO3 (malachite) is thermodynamically more stable than any cupric hydroxide or oxide
solid in particular pH and DIC ranges. The selection of these two solids was used to provide a
simulation of what might happen when recrystallization of cupric hydroxide or slow precipitation
of malachite may have taken place over some considerable time. It also provides a representation,
for comparison, of the solubility model historically most often applied to drinking water, adjusted
for improvements in the understanding of the aqueous speciation of copper(II).
Several important contrasts in cuprosolvency behavior between the case represented by Figures
6 and 7 as opposed to the assumptions behind Figures 3 and 4 should be noted. The “trough”
towards the lower pH values and higher DIC values is caused by the Cu2(OH)2CO3 (malachite)
stability field. If Cu2(OH)2CO3 (malachite) is present and capable of forming, below a pH of about
6.5 the addition of DIC is predicted to decrease cuprosolvency, but increase cuprosolvency above
about pH 7. There is a small transition zone between these values where the first approximately 5
mg/L of DIC should slightly reduce copper solubility, but additional carbonate would decrease it or
have essentially no effect. Below a pH of about 7, equilibrium cuprosolvency becomes essentially
insensitive to DIC above approximately 30 mg C/L. Malachite formation would enable attainment
33
Figure 6. Three-dimensional surface plot of the effect of DIC and pH on copper(II) solubility for
aged systems, assuming equilibrium with Cu2(OH)2CO3 and CuO (I=0.02, 25°C).
34
Figure 7. Contour diagram showing the effect of DIC and pH on copper(II) solubility, assuming
equilibrium with the stable solid phases Cu2(OH)2CO3 (malachite) and CuO (tenorite) at I=0.02
and 25°C.
35
of 1.3 mg/L after stagnation above approximately pH 6.5 for all DIC levels. This is in stark contrast
to the effect of pH and DIC when only cupric hydroxide is formed, where a pH of over 7 would be
necessary to stay under 1.3 mg/L for long stagnation times at very low DIC levels, and over 7.5 for
systems with high DIC.
Figures 8a and 8b illustrate the impact on predicted copper(II) concentrations produced by
some different choices for the solubility constants for Cu(OH)2(s) (cupric hydroxide) and
Cu2(OH)2CO3(s) (malachite). Figure 8a shows that at a fairly low DIC concentration of 4.8 mg C/
L (4.0 x 10-4 M) the pH boundary between these two solids can shift from approximately 8.8 to 9.9,
depending on the combination of values chosen. At the assumed temperature of 25°C, reasonably
representative of normal drinking water temperatures, the impact becomes less as carbonate con-
centration increases, and the thermodynamic stability domain of Cu2(OH)2CO3(s) (malachite) rela-
tive to Cu(OH)2(s) (cupric hydroxide) widens (Figure 8b). During recrystallization and aging,
cupric hydroxide will slowly convert to CuO(s) (tenorite). This will cause the pH of transformation
from malachite stability to that of tenorite to occur at lower values than with cupric hydroxide.
In hot water systems the oxide or the hydroxide solid becomes increasingly stable relative to
the basic carbonate.17 The absence of a complete set of formation enthalpies or temperature func-
tions for the various hydroxide and carbonate complexes involved in copper(II) solubility prevent
the development of reliable similar diagrams for the 0 to 55°C temperature range of interest to
potable water studies.
The previous focus of most copper corrosion research has been on pitting failures and corrosion
rate reduction to lengthen the service life of the plumbing systems. Therefore, the significance of
malachite formation as a control of copper levels in the first few months or years of service has
been given little attention. If malachite formation is kinetically inhibited, or is just relatively slow
under normal plumbing usage conditions, the sensitivity of copper solubility to pH will be consid-
erably increased and much higher solubilities are generally expected. This can be seen by exami-
nation of Figures 6 and 7, including malachite and tenorite, in contrast with Figures 3 and 4 for
cupric hydroxide.
Some evidence for the natural slow formation of malachite has been presented by researchers in
different areas. For example, Fruchter, et. al. found that equilibrium with CuO(s) instead of mala-
chite was a much more plausible model for explaining the levels of copper found in fly-ash pore
fluids and leachates81. From simple mass-action considerations, some increase in the rate of mala-
chite formation with increases in DIC would be highly plausible. However, specific research into
the rates of formation of malachite under differenc pH, DIC and flow conditions that would be
pertinent to passivation film formation has not been published.
Effect of Orthophosphate on Copper(II) Solubility
Little information on the solubility of copper orthophosphate solids is available, and almost no
progress has been made in identifying solids and determining solubility constants since the absence
of reliable thermodynamic data was pointed out in the major review by Rickard in 1970.51 A
36
Figure 8. Copper(II) solubility comparison assuming different solids and solubility constants,
25°C; a) DIC = 4.8mg C/L, I=0.005; b) DIC = 96 mg C/L, I=0.02
37
38
tabulation of naturally-occurring copper orthophosphate minerals is given in Table 5.82 The solid
lebethenite, Cu2(PO4)OH(s), has been observed in some hot water copper heat exchanger tube
deposits83. Unfortunately, no solubility or ∆Gf° data has been located for any of these minerals. In
tabulations of ∆G°f data or tables of solubility constants (see Table 1), only the solids Cu3(PO4)2(s)
and Cu3(PO4)2.2H2O(s) are listed. Neither of these solids is listed as a naturally-occurring mineral,
and their existence in corrosion deposits or passivation films in drinking water pipes is yet to be
proven. Because they represent the only solubility estimates available, they are used in these
modeling calculations.
TABLE 5. REPORTED MINERALS CONTAINING COPPER (II) AND ORTHOPHOSPHATE[1]
Name Formula
Libethenite Cu2PO4OH
Cornetite Cu3PO4(OH)3
Reichenbachite Cu5(PO4)2(OH)4
Ludjibaite Cu5(PO4)2(OH)4
Pseudomalachite Cu5(PO4)2(OH)4•H2O
Nissonite Cu2Mg2(PO4)2(OH)2•5H2O
Zapatalite Cu3Al4(PO4)3(OH)9•4H2O
Turquoise CuAl4(PO4)4(OH)8•5H2O
Sieleckite Cu3Al4(PO4)2(OH)12•2H2O
Planerite (Cu,Ca)Al6(PO4)4(OH)8•H2O
Hentschelite CuFe2(PO4)2(OH)2
Andrewsite (Cu,Fe2+)3Fe3+6(PO4)4(OH)12
Chalcosiderite Cu,Fe3+(PO4)4(OH)8•4H2O
Phosphofibrite KCuFe3+15(PO4)12(OH)12•12H2O
Reference
1. Clark, A.M., Hey’s Mineral Index. 1993, London: Chapman and Hall.
Orthophosphate has been reported48 to form two weak complexes with cupric ion, CuH2PO4+
and CuHPO4°. The fraction of Cu(II) solublity contributed by orthophosphate complexes is, there-
fore:
SCu(II),PO4=CuH2PO4+
[]
+CuHPO4°
[]
(30)
Substituing appropriate equlibrium constant expressions as before yields the following expression:
ST,PO4=
β
’1,2,1 Cu2+
[]
H
+
[]
2
PO43−
[]
+
β
’
1,1,1 Cu2+
[]
H
+
[]
PO43−
[]
(31)
When orthophosphate is present, then, the total solubility expression for copper(II) will be:
ST,Cu(II)=SCu(II ),OH +SCu(II ),CO3+SCu(II ),PO4(32)
39
Previous modeling calculations with orthophosphate projected a negligible effect22 or only a
small reduction in copper solubility for systems with low carbonate concentrations, and at pH
levels around neutral.84 These predictions resulted from the assumption that Cu2(OH)2CO3 (mala-
chite) is the controlling phase for copper levels in the water, as is illustrated in Figure 9.
Two major problems with the solubility data constrain the accuracy of predictions made for
orthophosphate effects on copper(II) solubility. First, given little experimental confirmation of the
solubility constants, relatively large errors (factor of 2 or more) are conceivable. While the error
could be in either direction, if the constants used here are biased high, more benefit in solubility
reduction by orthophosphate would be possible in drinking water situations for many water chem-
istries and chemical dosages. A second scenario is the hypothesis that at least short-term copper
levels are controlled by Cu(OH)2(s), rather than by Cu2(OH)2CO3(s). Then, a very different picture
emerges, where the predicted stability field for Cu3(PO4)2.2H2O(s) (the less soluble of the two
copper orthophosphates having reported solubility constants) is much larger. This reevaluation is
shown in Figures 10a through 10d, representing four different concentrations of DIC and four
levels of orthophosphate. Note the extremely important interaction among pH, DIC and ortho-
phosphate concentration in governing both the levels of soluble copper(II), and the pH of minimum
solubility.
While the minimum of solubility for the copper orthophosphate solid in waters of low to mod-
erate DIC is indicated to be approximately at pH 7.5 (eg. Figures 10a and 10b), progressively
larger differences in solubility between the hydroxide and orthophosphate solid occur as pH goes
down toward 6. Thus, orthophosphate addition for the control of lead and copper corrosion in a
low-DIC water at the frequent target pH of 7.5 will have a significant impact on reducing
plumbosolvency, but it may not improve copper control beyond that afforded by raising the pH to
7.5. For water systems not encountering lead material problems, however, doses of orthophosphate
at 1-5 mg PO4/L may enable attainment of the 1.3 mg Cu/L action level trigger for copper in new
plumbing systems at a pH below the approximately 7.2 to 8 range that otherwise might be neces-
sary for sufficient cuprosolvency reduction. In contrast, for systems in which substantial aging of
Cu(OH)2(s) to CuO(s) has taken place, or in which a substantial film of a Cu2(OH)2CO3(s) (mala-
chite) has developed, orthophosphate addition would not necessarily provide reduction of
cuprosolvency (see Figure 9). This phenomenon may explain why some treatments that look prom-
ising in pipe loop studies do not work as well when applied to an entire distribution system that has
many sites with older copper plumbing.
Two primarily electrochemical and gravimetric studies have shown some reduction in copper
corrosion rates by the addition of 1-5 mg PO4/L orthophosphate at pH 7.5 to 828,77. Another study
reported reductions in copper concentrations, but the orthophosphate level was approximately 950
mg PO4/L.85
The solubility mechanism modeled in this research probably not the only way in which ortho-
40
Figure 9. Effect of orthophosphate on copper(II) solubility at DIC = 4.8 mg C/L, I = 0.005 and
25°C, assuming the formation of Cu(OH)2(s), Cu2(OH)2CO3(s), and Cu3(PO4)2•2H2O.
41
Figure 10. Copper(II) solubility for differenct DIC levels with orthophosphate addition, assum-
ing formation of Cu(OH)2(s) and Cu3(PO4)2•2H2O(s) at 25°C: a) DIC = 4.8 mg C/L, I = 0.005;
b) DIC = 14.4 mg C/L, I = 0.005; c) DIC = 48 mg C/L, I = 0.01; d) DIC = 96 mg C/L, I = 0.02.
42
phosphate affects cuprosolvency. Other roles orthophosphate might play in copper corrosion could
be by causing alteration of the nature or growth rate of passivating films, or the kinetics of oxida-
tion/reduction reactions at the copper pipe surface. For example, visual and X-Ray diffraction of
copper pipe surfaces exposed to water with DIC of 5 mg C/L at pH 7, 8, and 9 in ongoing U. S. EPA
studies indicate substantial differences in appearance and mineralogy with and without the pres-
ence of 3 mg PO4/L orthophosphate. Pipe specimens from the systems having the orthophosphate
were almost free of crystalline copper(II) solids. Those experimental systems, as well as identical
ones for 10 mg C/L also indicate a drastic reduction in the rate of free chlorine residual depletion in
the presence of 3 mg, especially below pH 8. This implies that orthophosphate also plays a critical
role somewhere in the oxidation process of copper metal to cupric ion.
A German study has reported 3 mg PO4/L orthophosphate at pH 7.2 and DIC of about 75 mg C/
L allows higher copper levels upon stagnation in aged (426-512 days) pipe than without orthophos-
phate,84,85 which also indicates that orthophosphate may interfere with the normal corrosion scale
oxidation and aging processes. Under some chemical conditions, orthophosphate could preserve
higher copper levels that might otherwise be reduced when a stable malachite film formed.
Effects of Sulfate, Chloride and Ammonia on Copper(II) Solubility
Some minor enhancement of solubility may be produced by the formation of the CuSO4° aque-
ous complex in waters of high sulfate concentration, but it is not a very strong complex. At least
one value has been reported for a formation constant for a Cu(SO4)22- complex, but its existence
remains uncertain. Therefore, aqueous sulfate complexes are not likely to be significant influences
on cuprosolvency in potable waters.
Copper forms a wide variety of basic sulfate solids, as are shown in Tables 1 and 2. Only some
of the mineral species have either solubility or Gibbs free energy data available for them. The
problem is compounded because there is some discrepancy in the X-Ray analysis, mineralogy and
thermodynamic data literature on the naming and stoichiometry of some of the solid phases. There
is good agreement that the formula Cu4(OH)6SO4 applies to the monoclinic mineral brochantite.
Hydrated forms of that mineral for which official mineral names exist include: Cu4(OH)6SO4.H2O
(posnjakite, monoclinic crystal system), Cu4(OH)6SO4.2H2O (langite, orthorhombic crystal sys-
tem), and Cu4(OH)6SO4.2H2O (wroewolfenite, monoclinic crystal system).88 Unfortunately, in com-
pilations of Gibbs free energies of formation, the mineral name “langite” is ascribed to
Cu4(OH)6SO4.H2O by one important primary data source,89 and to both Cu4(OH)6SO4.H2O and
Cu4(OH)6SO4.1.3H2O by another.90 Each water of hydration contributes -56.69 kcal to the ∆Gf°
value for the compound, so the computed solubility constant will vary significantly, thus adding
confusion to modeling and data analysis.
The kinetic constraints on the formation of these solids in water systems are largely unex-
plored. Calculations with the CU2SOL model show that even relatively low levels of sulfate (such
as 20-30 mg/L) may cause supersaturation of one or more basic cupric sulfate solids in slightly acid
to slightly alkaline pH’s. Thus, in the pH range of 6 to 8 at low to moderate DIC levels, and in the
43
absence of Cu2(OH)2CO3 (malachite) formation, calculations show that basic copper sulfate solids
may keep copper solubility below the levels that would otherwise occur through passivation by
cupric hydroxide.
Especially above a pH of approximately 8, the persistence of metastable basic cupric sulfate
solids when the hydroxide or oxide should be thermodynamically favored is also possible, and is
supported by X-Ray analysis in this and other studies.91 Extensive studies of phase relationships of
cupric hydroxide and other precipitates in the presence of sulfate indicated that the ratio of equiva-
lents of cupric sulfate to alkali in a mixture was important in defining the solid produced92. For
example, if the ratio was equal to or greater than 1.33, the precipitate found was the basic salt
4CuO.SO3.3H2O, identical with the mineral brochantite. If the ratio was less than 1, the precipitate
was pure cupric hydroxide. Other mixes of solids were found at ratios between 1 and 1.33.
Clearly, some relevence to drinking water plumbing systems is likely. Metastable basic sulfate
solids may limit the amount of cuprosolvency reduction possible by pH and DIC adjustment, espe-
cially above pH 8. However, a detailed exploration of the solubility ramifications of sulfate is
beyond the scope of this investigation, so only various aspects of it are considered where appropri-
ate.
A consensus exists that chloride forms at least three weak complexes with cupric ion, CuCl+,
CuCl2°, and CuCl3-. The complexes CuOHCl° and Cu2Cl42- have also been reported by various
researchers, but at typical drinking water concentrations of chloride, they are unlikely to be very
significant. Ammonia complexation constants for cupric ion are strong, and the complexes formed
are CuNH32+, Cu(NH3)22+, Cu(NH3)32+, and Cu(NH3)42+. The additional chloride, sulfate and ammo-
nia species of copper(II) may be combined into an additional term, SCu(II),add
SCu(II),add =CuCl+
[]
+CuCl2°
[]
+CuCl3−
[]
+CuSO4°
[]
+CuNH32+
[]
+
Cu(NH3)2
2+
[]
+Cu(NH3)3
2+
[]
+Cu(NH3)4
2+
[]
(33)
to complete the copper(II) solubility model as follows.
ST,Cu(II)=SCu(II ),OH +SCu(II ),CO3+SCu(II ),PO4+SCu(II),add (34)
However, these additional complexes seem to have little predicted effect on overall copper(II)
solubility compared to carbonate, which is usually present in much higher concentration than am-
monia and chloride. Enhancement of copper solubility during periods of chloramination with con-
siderable excess ammonia present, followed by a decrease in solubility when a period of breakpoint
chlorination was practiced was observed for Champaign, IL tap water.93 Whether this is caused by
cuprous ion ammine complexes, cupric ion ammine complexes, or some other mechanism has not
been determined.
44
While some species, such as chloride, dissolved organic materials, and sulfate may not directly
enhance equilibrium copper(II) solubility, they may indirectly influence it by modifying the crystal
structure, morphology, and compactness of the solids formed at the pipe surface.79,94,95
Interrelationships of Copper(I) and Copper(II)
Figures 11 and 12 are EH-pH (EMF-pH or “Pourbaix”) diagrams for systems having relatively
low (4.8 mg C/L = 4 x 10-4M) and relatively high (96 mg C/L = 8 x 10-3M) levels of dissolved
carbonate. For these diagrams, the thermodynamic activities of the aqueous species are assumed to
be equal to their concentrations. Aqueous copper species activities are assumed to be at the action
level of 1.3 mg Cu/L. Copper(I) solubility is assumed to be controlled by Cu2O(s) (cuprite).
Copper(II) solubility is assumed to be controlled by the metastable Cu(OH)2(s) high molar surface
solid in these figures. These diagrams are in considerable contrast to many published previously,
because of the inclusion of the significant cupric carbonate and hydroxycarbonate complexes. The
diagrams reflect increasing solubility of copper(II) with DIC.
When transformations from copper(II) to copper(I) are considered, as may take place when
oxidizing agents in the water are depleted, an interesting feature emerges. Figure 13a shows solu-
bility curves for copper(II) being controlled by Cu(OH)2(s) (cupric hydroxide) at three levels of
DIC superimposed on solubility curves of the two possible copper(I) solids, Cu(OH)2(s) and Cu2O(s)
(cuprite). Ammonia and chloride are assumed to be absent. Depending on the pH and DIC levels,
major drops in copper solubility could take place after depletion of oxygen and chlorine, as has
been suggested by recent research.86,87 Essentially as an inverse of equation (8), copper metal and
cuprous ions in the film layer immediately on the surface of the pipe may serve as reducing agents
to convert solution or interstitial cupric ions and fresh cupric hydroxide surficial solids into CuOH(s)
(cuprous hydroxide) or Cu2O(s) (cuprite), by a path such as:
Cu(s) +Cu2+↔2Cu+
(35)
2Cu
+
+H
2
O↔Cu
2
O(s)+2H
+
(36)
The net result of such a reaction process would be a decrease in soluble copper levels as Cu(II)
(cupric) aqueous species are reduced to less-soluble Cu(I) (cuprous) forms.
For example, for a DIC of 96 mg/L at pH 8.5, copper solubility could ultimately decrease from
about 1 mg/L to a virtually undetectable level (approximately 0.002 mg/L) from this phenomenon.
If Cu2(OH)2CO3 (malachite) and CuO (tenorite) are the controlling scale-forming solids, as might
be the case with aged copper pipe, the concentration change would be much less, as is shown in
Figure 13b. Interestingly, the amount of concentration change is predicted to depend on pH and
DIC for both newer and older copper surfaces, and whether or not CuOH(s) or Cu2O(s) (cuprite)
most readily forms.
In addition to the cited German research, previous research by Tronstad and Veimo and KIWA
indicated the possible occurrence of this effect,73,76 as has some research in Japan75. Thus, copper
45
Figure 11. EMF-pH diagram for copper in water containing carbonate, assuming formation of
cupric and cuprous hydroxide at DIC = 4.8 mg C/L, I = 0, copper species concentrations of 1.3
mg/L, and 25°C.
46
Figure 12. EMF-pH diagram for copper in water containing carbonate, assuming formation of
cupric and cuprous hydroxide at DIC = 96 mg C/L, I = 0, copper species concentrations of 1.3
mg/L, and 25°C.
47
Figure 13. Copper(II) solubility of different DIC levels compared to copper(I) solubility, at I =
0.01, 25°C for: a) newpipe with Cu(OH)2(s); b) aged pipe.
48
levels at the tap are likely to behave much differently under some chemistry and water usage sce-
narios than lead (which usually follows classical diffusion-controlled stagnation profiles) when
oxidant concentrations are depleted under service conditions. This also has considerable impact on
the design and conduct of corrosion optimization studies for copper, both by pipe loop and electro-
chemical testing. Careful interpretation of the meaning and consistency of standing copper samples
must be done, and additional data such as chlorine residual, dissolved oxygen, and other back-
ground water chemistry parameters may need to be collected before and after standing in different
parts of the distribution system, to enable proper conclusions to be drawn from sampling data.
Complications of Dissolved NOM
Several investigators have looked at interactions between Cu2+ and dissolved natural organic
matter (NOM). NOM is a diverse collection of material that is mainly composed of humic or fulvic
substances, and that is generally characterized empirically by statistical models that describe bind-
ing of metals by the formation of one or several theoretical ligands under a specific set of experi-
mental conditions (such as pH and ionic strength). These are numerical constructs that can not
always be used for comparison of data for different water supplies or NOM sources. Nonetheless,
some features of interaction with copper can be examined.
Some studies suggest that while copper can form complexes with NOM ligands (sometimes
called DOM for “dissolved organic matter”), at concentrations of copper typical of drinking wa-
ters, copper speciation is more likely dominated by hydrolysis or carbonate complexes.96-98
Cu2+ ion may also bind with adsorbed organic material containing appropriate functional groups.
The binding with adsorbed organic matter seems to be stronger than direct binding with surface
sites on several materials tested.99 Copper present as an organic complex may bind preferentially
with adsorbed organic material.99,100 These studies suggest that some reduction in copper concen-
tration may be caused by adsorbed organic material acting as either a diffusion barrier, or as a less-
soluble corrosion film.
Some other studies suggest that NOM may play a major role in the aqueous speciation of cupric
ion, particularly when carbonate concentrations are low.101,102 Organic ligands produced by marine
diatoms and during diatom blooms have been shown to strongly complex copper, though usually at
low copper concentration.103,104 Unsaturated organic ligands were also shown in experiments at
pHs generally lower than drinking water to increase the dissolution rate of copper metal in the
presence of cupric ion, by a complicated interaction affecting the electron transfer rate between
Cu(s) and Cu2+, and by stabilizing the Cu(I) state by complexation.105
The significance of NOM to cuprosolvency relative to drinking water concentrations of copper
and competing non-metals and ligands has not been conclusively determined, though it is an area
under active investigation by some research groups in the United States. Research into copper
plumbing pitting has indicated that some NOM may actually alleviate the propensity of a water to
cause pitting attack, and possibly alter some scale formation characteristics of uniform copper
corrosion.91 Any effect on cuprosolvency will likely be stronger in untreated surface water supplies
49
than in ground waters with very low TOC or coagulated and filtered surface waters106,107,346 . In
addition to pH and ionic strength, which have been widely acknowledged to be important in com-
plexation studies with NOM, consideration must also be given to the considerable role that cupric
carbonate, hydroxide, and hydroxide/carbonate complexes must play, particularly as the pH in-
creases above 7.
In drinking water systems, the presence of extensive amounts of copper metal piping as a source
creates a considerably different environment than natural water or aquifer systems where NOM/
copper complexation has been studied most extensively. Further, depending on source water and
disinfection conditions, the redox potential of drinking waters can vary over a wider range than
natural systems, and a variety and quantity of solids may exist that are also not present in ground
waters, lakes and sea water. Because of the uncertainty in the role of NOM, and the lack of
availability of routine characterization information and corresponding binding constants, no at-
tempt was made to model its impact in this study. The systems used for confirmation of the solubil-
ity model were chosen to be generally free of this complication. Lack of correspondence between
cuprosolvency predictions and appropriately chosen field water samples may be at least partly the
result of NOM effects, and that should be borne in mind when data is evaluated.
50
Experimental and Field Evidence for the “Cupric Hydroxide Model”
Experimental Systems
Analytical and Data Reporting Procedures
Chemical analytical procedures used by DWRD for the experiments used in this report are
listed in Table 6, along with the observed detection limits. Because the experiments described in
this report took place over several years, there were some changes in analysis procedures over that
time. The analysis of most metals was switched from a flame atomic absorption spectroscopy
(AAS) system to a simultaneous inductively-coupled plasma spectroscopy (ICP) system during
August 1993, after comparability was statistically established. This change also improved the
detection limits and low-level precision for the copper analyses. Potassium was always determined
by flame AAS, and low-level lead determinations were always done by matrix-modified graphite
furnace AAS. The analysis of Si by ICP (reported as SiO2) was substituted for the automated
colorimetric silicate procedure in November, 1993. The analysis of total sulfur by ICP replaced the
automated turbidimetric procedure in January, 1994. This was primarily implemented to reduce
the analyst workload and improve analytical throughput, rather than to improve the accuracy of the
analysis. Because the presence of sulfide or other reduced sulfur species is chemically impossible
or the concentrations negligible in DWRD experiments, for all calculations the ICP analysis of total
sulfur is interpreted as sulfate.
The pH measurement procedure for the laboratory experiments followed the closed-system
technique previously developed at EPA.108,109 Triplicate samples were carefully collected in 15 mL
glass vials, which were sealed with no air space by caps having conical polyethylene liners until
analysis. The measurement process was made more precise and accurate by linking the pH meter
output to a personal computer through an RS-232 port, and monitoring stabilization through a
BASIC program. A sample pH was considered stable when three successive 10-second measure-
ments showed an estimated standard deviation of less than 0.001 pH. Standardization of the pH
meter was accomplished either directly with pH buffers obtained from NIST†, or with commer-
cially-prepared buffers accurate to within 0.005 pH that were directly traceable to NIST buffers.
Samples were analyzed at least in triplicate. The final pH almost always reflects replication of the
final two samples to within ± 0.01 pH.
† National Institute of Standards and Technology, Gaithersburg, MD.
51
TABLE 6. ANALYTICAL METHODS USED FOR CHEMICAL ANALYSIS OF WATER SAMPLES
Detection
Analysis Method Method Number Reference Limit (mg/L)
Metals
Calcium AA-Flame 7140 EPA10.1
Magnesium AA-Flame 7450 EPA12.0
Sodium AA-Flame 7770 EPA13.0
Potassium AA-Flame 7610 EPA10.25
Iron AA-Flame 7380 EPA10.05
Copper AA-Flame 7210 EPA10.02
Lead GFAAS 7421 EPA10.002
Zinc AA-Flame 7950 EPA10.01
Manganese AA-Flame 7460 EPA10.01
Calcium ICAP 200.7 EPA20.01
Magnesium ICAP 200.7 EPA20.025
Sodium ICAP 200.7 EPA20.025
Copper ICAP 200.7 EPA20.003
Lead ICAP 200.7 EPA20.02
Zinc ICAP 200.7 EPA20.001
Manganese ICAP 200.7 EPA20.0004
Silicon (as SiO2) ICAP 200.7 EPA20.053
Sulfur (as SO4) ICAP 200.7 EPA20.045
Aluminum ICAP 200.7 EPA20.025
Iron ICAP 200.7 EPA20.002
Anions
Chloride Automated Potentiometric Titration 4500-Cl- D. Std. Methods31.0
Fluoride Automated Standard Additons — Orion4<0.1
Potentiometric ISE 340.2 EPA50.10
Orthophosphate Automated Colorimetric I-2601-85 USGS60.02 (as PO4)
Total Phosphate Automated Colorimetric I-2600-85 USGS60.05 (as PO4)
Nitrate -N Automated Colorimetric A303-5173-00 Alpkem70.02 (as N)
Silicate Automated Colorimetric A303-5220-13 Alpkem70.4 (as SiO2)
Sulfate Automated Turbidimetric A303-5220 Alpkem7~6.0 (as SO4)
Total Alkalinity Automated Potentiometric 2320 B.4.6. Std. Methods 3~0.3 (as CaCO3)
Titration to Equivalence Point
Others
Dissolved Oxygen Winkler (Azide Modification) 4500-0 D. Std. Methods30.50
Ammonia Automated Colorimetric 350.1 EPA50.03
Total Inorganic Carbon Coulometric Titration D513-92 ASTM8< 0.5
Total Chlorine DPD Colorimetric 8167 Hach90.02
Free Chlorine DPD Colorimetric 8021 Hach90.02
pH Closed-System Electrometric — EPA (DWRD)10 —
1 RREL SW846, Sept. 1986.
2 USEPA, “Methods for the Determination of Metals in Environmental Samples,” EPA-600/4-91-010 (1994).
3 “Standard Methods for the Examination of Water and Wastewater,” 18th Edition (1992).
4 Orion Research, Inc., Boston, MA.
5 USEPA, “Methods for Chemical Analysis of Water and Wastes,” EPA-60014-79-020 (1983).
6 Modified from methods for Determination of Inorganic Substances in Water & Fluvial Sediments, U.S.
Geological Survey Open-File Report, (85-495) 1985.
7 Alpkem Research, Inc., Clackamas, OR.
8 “1994 Annual Book of ASTM Standards,” section 11, volume 11.01 Water (I).
9 Hach Company, Loveland, CO.
10 Drinking Water Research Division, USEPA, Internal Method. References: Journal AWWA 72:5:304 (1980);
Schock & Lytle, Proc. AWWA WQTC (1994).
For the recirculation solubility and late runs of the coupon study experiments, DIC was ana-
lyzed directly using a coulometric method.110 This has frequently been found to be more accurate
and precise than deriving the concentration from two measured quantities (pH and total alkalinity),
52
each with an analytical uncertainty. The combined method and sampling standard deviation for
DIC during the course of this study was computed using pooled data from analyses of duplicate
samples, and was determined to be ±0.1 mg C/L, with no statistically-significant concentration
dependence over the experimental range of 5 to 70 mg C/L.
Quality assurance practices for the instrumental analyses followed the usual documented
Inorganics and Particulates Control Branch standard operating procedures, that include require-
ments for analysis of duplicates and spikes of samples comprising more than 10% of the sample
load, and verification of instrument calibration and some interference checking through external
certified reference standards at multiple times during each analytical run. The exact location and
frequency of different types of quality control spikes, standards, blanks, and duplicates, along with
accuracy requirements, are specified in those documented procedures for each type of analysis.
They cannot be generalized because the precision and accuracy expectations vary with the type of
instrument used and the levels of the analyte encountered in the different experiments.
One general indicator of the completeness and accuracy of the analyses performed in each
experiment is a check of the percent ion balance error.111 Computations of ion balance errors on all
samples used in the computation of Saturation Indices and which were used for solubility model
evaluation were made using fully speciated results from the WATEQX model112. These calcula-
tions are summarized in box plot form in Figure 14. The analyses are shown to be extremely
complete, and the biases well below ± 5% for almost all analyses for all three DWRD studies
covered in this report.
Data reporting for tabulations and statistical evaluation followed the recommendations of the
ASTM standard practice for low-level data (D4210-83).113 In this practice, actual instrument re-
sults, including negative values, are reported and used in statistical calculations, to avoid inaccura-
cies brought about by artificial data truncation commonly used by environmental laboratories. After
computations and statistical summaries were completed, concentrations below observed analytical
detection limits were replaced by the detection limit. When analytical procedures were changed
during an experimental run, the most conservative method detection limit that was applicable dur-
ing that run was used in the tabulation.
X-Ray diffraction analysis was performed on corrosion deposits from pipe samples, whenever
sufficient sample volume was available. Deposits scraped from the pipe sections were finely ground
by synthetic ruby or agate mortar and pestle, to pass through a 200-mesh sieve (approximately 75
µm) whenever possible. Samples were then suspended as a slurry with amyl acetate, and deposited
on zero-background quartz plates by disposable pipettes for mounting in the powder diffractometer.
The instrument used was a Scintag XDS-2000 theta-theta diffractometer with a copper X-Ray
tube operated at 45 kV and 33 mA. Scans were usually over the range of 5 to 60 degrees 2-theta,
with 0.03 degree step sizes that were held for 3 seconds each. Pattern analysis, performed with the
software provided by the instrument manuacturer, generally followed procedures outlined by
ASTM.114
53
Figure 14. Box plots of percent ion balance errors for complete water analyses from the
laboratory experimental runs reported. Boxes show mean (dark lines), median (light line), 25th
and 75th percentiles (box borders), and the 10th and 90th percentiles (error bars). Data outside
these limits are marked (open circles). Large errors represent missing analytes.
54
Coupon Experiments
Even though the experiments were not designed for this purpose, relevent data for copper cor-
rosion was obtained from a multi-year evaluation of the effect of pH, orthophosphate concentration
and metal composition (eg. lead percentage) on the leaching of metals from brass in Cincinnati tap
water.115 In the study, “control” coupons of pure copper (CDA 122) were included in the investiga-
tion. The basic test system consisted of a 100 gallon (379 L) heavy-gauge polyethylene reservoir
tank with a floating lid, that fed a manifold of 12 30-mL teflon™ cells containing standardized 1-in
x 2-in x 0.125-in (2.5-cm x 5.1-cm x 0.32-cm) metal coupons†. Coupons were cleaned using a
procedure devised to remove cross-contamination from manufacturing, plus chemically cleaning
the surface, to assure uniformity of behavior.115 Chemical cleaning steps used a surfactant soak, a
brief 30% HCl soak, and a final acetone rinse. Between cleaning steps, speicmens were thoroughtly
rinsed with ultra-pure deionized water.
A pump containing a non-metallic chamber fed the system with test water through all PVC
plastic and teflon™ tubing. Rotameters and needle valves were used to adjust the flow rates for
uniformity. Teflon™ three-way valves were used to allow draining of the cells for daily water
exchange or sampling.
In the first several experimental runs, the reservoir tank was filled daily from the tap water
source. The pH was adjusted by manually dosing the tank water with 6 N HCl or 8 N NaOH. When
orthophosphate was tested, it was checked daily and adjusted if necessary by dosing with a stan-
dardized solution of 0.1 M Na2HPO4. The water quality in the tank was found to be chemically
stable enough in later runs that the tank only had to be refilled one time per week. However, the
free chlorine residual and pH were checked daily and adjusted to meet target values, if necessary.
In all experiments, water chemistry samples were taken from the reservoir daily to provide a back-
ground characterization of the water quality.
Each day, water was drained from all cells, and the manifold cells were flushed with a total of
approximately 10 gallons (38 L) of reservoir water, corresponding to approximately 0.8 gallons (3
L) per cell. At the beginning of the runs, samples from each cell were collected daily for analysis.
During later run stages, cells were sampled 2 to 3 times per week. Standing times for the coupons
were 24 hours during the week, and 72 hours over the weekend. Because of the limited cell volume
and air contact during sample collection, sensitive parameters such as pH, chlorine residual and
dissolved oxygen could not be analyzed on these stagnation samples. Sample volume was also
inadequate to enable estimation of dissolved versus colloidal or particulate material by filtration.
† Metal Samples, Inc., Munford, Alabama.
55
The five experiments (Runs 1-5 in subsequent discussions) covered the following conditions.
1. The effect of the background tap water distributed by the Cincinnati Water Works Rich-
ard Miller Water Treatment Plant (California, OH), with an unadjusted pH of approxi-
mately 8.5;
2. The effect of the same background tap water quality, but at pH adjusted to 7.
3. The effect of 3 mg PO4/L disodium orthophosphate at pH adjusted to 7.5.
4. The effect of 0.5 mg PO4/L disodium orthophosphate at pH adjusted to 7.5; and,
5. The effect of background tap water quality with pH adjusted to 7.5.
The background water qualities for each run of this “Coupon Study” are summarized in Table
7. Note that several chemical constituents fluctuated widely over the course of some of the indi-
vidual runs, and from one run to another. Inability to maintain a constant background water chem-
istry to isolate the most significant chemical variables put some important contraints on the ability
to use data from this experimental study to confirm the accuracy of the solubility model.
TABLE 7. SUMMARY STATISTICS FOR THE VARIATION IN BACKGROUND WATER QUALITY DURING
THE DIFFERENT RUNS OF THE COUPON CORROSION STUDY
RUN #1
Analyte N Min Max Mean Std. Dev. 95% CI Median
Lead, µg/L 70 <0.002 11.9 1.4 2.4 0.7 0.48
Calcium, mg/L 64 34.2 54.8 42.4 4.6 1.2 40.9
Copper, mg/L 54 <0.02 0.03 0.00 <0.02 <0.02 <0.02
Iron, mg/L 57 <0.05 0.07 <0.05 <0.05 <0.05 <0.05
Potassium, mg/L 54 2.3 4.4 3.5 0.45 0.5 3.6
Magnesium, mg/L 64 9.1 016.6 13.5 1.8 0.5 13.5
Manganese, mg/L 43 <0.01 0.03 <0.01 0.01 0.01 <0.01
Sodium, mg/L 64 14.8 44.5 33.4 6.8 1.8 34.7
Zinc, mg/L 58 <0.01 0.39 0.03 0.06 0.02 0.01
Alkalinity, mg CaCO3/L 60 43.5 74.1 60.2 5.6 1.4 60.4
Sulfate, mg SO4/L 46 104.0 130.0 116.3 7.9 2.3 115.0
Chloride, mg/L 32 39.0 43.8 40.3 1.2 0.4 39.9
Silica, mg Sio2/L 5 2.4 2.8 2.6 0.2 0.3 2.8
Nitrate, mg N/L 54 <0.02 1.1 0.77 0.17 0.04 0.73
Ammonia, mg N/L 54 <0.03 <0.03 <0.03 0.00 0.00 <0.03
Orthophosphate, mg PO4/L 25 <0.02 0.37 <0.02 0.08 0.08 <0.02
Dissolved oxygen, mg/L na* na* na* na* na* na* na*
Total inorganic carbon, mg C/L na* na* na* na* na* na* na*
Free chlorine, mg CI2/L 68 0.65 3.1 2.0 0.56 0.14 2.00
pH, pH units 67 8.04 8.79 8.53 0.18 0.04 8.54
*na - not analyzed
56
Run #2
Analyte N Min Max Mean Std. Dev. 95% CI Median
Lead, µg/L 93 <0.002 14.1 0.68 2.12 0.44 0.50
Calcium, mg/L 91 25.6 51.2 33.5 6.3 1.3 33.3
Copper, mg/L 90 <0.02 2.3 0.05 0.37 0.08 <0.02
Iron, mg/L 89 <0.05 0.66 <0.05 0.13 <0.05 <0.05
Potassium, mg/L 91 1.6 3.2 2.0 0.39 0.08 1.96
Magnesium, mg/L 91 6.0 12.8 8.1 1.7 0.36 8.05
Manganese, mg/L 87 <0.01 0.03 <0.01 0.02 0.17 <0.01
Sodium, mg/L 91 3.8 53.9 15.7 7.4 1.6 14.0
Zinc, mg/L 89 <0.01 0.07 <0.01 0.03 0.01 <0.01
Alkalinity, mg CaCO3/L 92 21.7 52.2 34.9 6.4 1.3 33.9
Sulfate, mg SO4/L 92 52.8 108.9 68.2 10.7 10.7 66.2
Chloride, mg/L 82 21.4 46.0 29.6 5.6 5.6 28.0
Silica, mg SiO2/L 88 4.3 7.7 6.0 0.73 0.73 6.1
Nitrate, mg N/L 70 <0.02 2.0 1.1 0.30 0.30 1.2
Ammonia, mg N/L 83 <0.03 <0.03 <0.03 0.11 0.11 <0.03
Orthophosphate, mg PO4/L 76 <0.02 0.31 0.02 0.05 0.05 <0.02
Dissolved oxygen, mg/L 41 6.0 10.2 8.9 0.72 0.72 9.0
Total inorganic carbon, mg C/L 23 10.9 14.1 12.5 1.21 1.21 12.4
Free chlorine, mg CI2/L 92 1.0 2.2 1.4 0.28 0.28 1.3
pH, pH units 93 6.9 7.1 7.0 0.05 0.05 7.0
*na = not analyzed
Run #3
Analyte N Min Max Mean Std. Dev. 95% CI Median
Lead, µg/L 66 <0.002 2.0 0.29 0.49 0.12 0.10
Calcium, mg/L 63 34.5 44.5 40.0 2.6 0.65 40.3
Copper, mg/L 61 <0.003 0.02 0.003 0.01 <0.03 <0.003
Iron, mg/L 64 <0.002 0.17 0.01 0.03 0.01 0.01
Potassium, mg/L 64 <2.00 3.4 3.0 0.42 0.11 3.1
Magnesium, mg/L 64 <0.023 11.1 9.7 1.4 0.36 9.9
Manganese, mg/L 61 <0.0004 0.01 <0.0004 <0.0004 <0.0004 <0.0004
Sodium, mg/L 64 16.3 26.2 21.3 1.1 1.1 22.8
Zinc, mg/L 61 <0.001 0.01 <0.001 <0.001 <0.01 <0.001
Alkalinity, mg CaCO3/L 58 43.1 58.4 52.2 3.0 0.79 52.6
Sulfate, mg SO4/L 56 64.6 93.6 78.6 7.5 2.0 78.8
Chloride, mg/L 59 0.04 36.5 30.8 5.0 1.3 31.7
Silica, mg SiO2/L 63 3.7 8.5 6.1 1.1 0.28 5.6
Nitrate, mg N/L 43 <0.02 1.1 0.85 0.33 0.10 0.96
Ammonia, mg N/L 57 <0.03 <0.03 <0.03 <0.03 <0.03 <0.03
Orthophosphate, mg PO4/L 57 2.5 3.3 2.8 0.16 1.3 2.8
Dissolved oxygen, mg/L 51 5.5 10..8 8.7 0.76 0.21 8.8
Total inorganic carbon, mg C/L 30 11.2 14.8 13.7 0.76 0.28 13.7
Free chlorine, mg CI2/L 68 0.69 2.3 1.4 0.49 0.12 1.2
pH, pH units 68 7.4 7.6 7.5 0.05 0.01 7.5
*na = not analyzed
57
Run #4
Analyte N Min Max Mean Std. Dev. 95% CI Median
Lead, µg/L 54 <0.002 1.3 0.13 0.40 0.00 0.10
Calcium, mg/L 59 27.7 39.6 34.3 3.0 0.93 34.7
Copper, mg/L 59 <0.003 0.22 0.005 0.03 0.008 <0.003
Iron, mg/L 60 <0.002 0.03 0.01 0.01 0.003 0.004
Potassium, mg/L 58 1.6 2.4 2.1 0.25 0.08 2.1
Magnesium, mg/L 58 7.1 10.7 9.0 0.91 0.36 9.0
Manganese, mg/L 56 <0.0004 0.01 <0.0004 <0.0004 <0.0004 <0.0004
Sodium, mg/L 58 8.8 19.8 13.8 2.5 0.77 13.3
Zinc, mg/L 60 <0.001 0.01 0.003 0.006 0.008 <0.001
Alkalinity, mg CaCO3/L 60 32.1 56.2 45.6 6.7 1.9 44.9
Sulfate, mg SO4/L 59 54.1 79.6 67.3 5.8 2.6 67.1
Chloride, mg/L 60 15.0 28.3 21.1 4.1 1.1 22.2
Silica, mg SiO2/L 59 4.6 10.2 6.5 0.10 0.40 6.7
Nitrate, mg N/L 59 0.79 1.8 1.16 0.29 0.08 1.0
Ammonia, mg N/L 40 <0.03 0.09 <0.03 0.03 <0.03 <0.03
Orthophosphate, mg PO4/L 59 0.30 0.54 0.46 0.05 <0.02 0.47
Dissolved oxygen, mg/L na* na* na* na* na* na* na*
Total inorganic carbon, mg C/L 57 8.2 14.4 11.2 1.9 8.0 11.1
Free chlorine, mg CI2/L 53 0.98 1.5 1.3 0.11 1.26 1.2
pH, pH units 53 7.4 7.7 7.5 0.07 0.07 7.5
*na = not analyzed
Run #5
Analyte N Min Max Mean Std. Dev. 95% CI Median
Lead, µg/L 78 <0.002 <0.002 <0.02 <0.002 <0.002 <0.002
Calcium, mg/L 81 27.7 47.4 38.9 6.5 1.4 42.5
Copper, mg/L 80 <0.003 0.01 0.004 <0.003 <0.003 <0.003
Iron, mg/L 81 <0.002 0.05 0.01 0.007 <0.002 0.01
Potassium, mg/L 54 1.0 6.1 3.9 0.82 0.22 3.9
Magnesium, mg/L 81 6.7 15.6 11.3 3.1 0.68 13.4
Manganese, mg/L 81 <0.0004 <0.004 <0.0004 <0.0004 <0.0004 <0.0004
Sodium, mg/L 81 9.0 43.1 25.1 10.4 2.3 29.0
Zinc, mg/L 81 0.01 0.02 0.01 0.03 <0.001 0.01
Alkalinity, mg CaCO3/L 80 17.9 94.3 57.2 22.4 5.0 63.4
Sulfate, mg SO4/L 69 43.8 136.3 100.8 23.4 5.6 106.8
Chloride, mg/L 74 24.0 49.5 35.0 5.8 52.0 35.1
Silica, mg SiO2/L 61 1.3 5.8 3.0 1.6 0.35 2.8
Nitrate, mg N/L 34 0.69 1.5 1.05 0.23 0.08 1.1
Ammonia, mg N/L 76 <0.03 <0.03 <0.03 <0.03 0.06 <0.03
Orthophosphate, mg PO4/L 31 <0.02 0.05 0.02 <0.02 <0.02 <0.02
Dissolved oxygen, mg/L 21 5.1 10.6 8.4 1.6 0.72 8.65
Total inorganic carbon, mg C/L 64 13.4 23.7 17.6 3.4 31.7 16.0
Free chlorine, mg CI2/L 72 0.68 1.4 0.94 0.15 0.04 0.92
pH, pH units 72 7.4 7.7 7.5 0.06 0.06 7.6
*na = not analyzed
58
Single-Pass Pipe Experiments
In a different study designed to investigate the effect of chlorine residual and silicate corrosion
inhibitors on copper leaching from rigid copper tubing, a test pipe rig was constructed in one of the
USEPA AWBERC facility pilot plants. This study will subsequently be referred to as the “Copper
Pipe Study.” Some data applicable to compare to the predicitions of this cuprosolvency model are
available from this previous investigation.
Cincinnati tap water (Miller WTP) was passed through a series of two small cartridge carbon
filters to remove chlorine and reduce TOC, and fed into three heavy-duty polyethylene plastic 200-
L (53-gal) reservoir tanks with floating lids. Each reservoir tank had appropriate chemicals added
manually to adjust the water quality to target values. Test water contained in each reservoir was
continuously recirculated by a magnetic drive pump system. All pump components in contact with
the water were constructed of nonmetallic materials.
Each reservoir was used as the initial water feed source for a set of triplicate 8-ft (2.4-m) straight
lengths of 0.5-in (1.27-cm) ID type M hard-drawn copper pipe. Plumbing connections were made
of 0.5-in (1.27-cm) ID Schedule 80 PVC pipe and fittings. Pipe sections were chemically cleaned
in the same manner as described previously for the coupons to assure reproducibility of surfaces in
contact with the water.
Each single test pipe section of each triplicate set was isolated from the other two by PVC check
valves. Plastic sampling taps were installed at the end of each pipe section, and before the check
valves. Water was pumped through the pipes by a pressure-actuated pump. For intervals of flush-
ing after different stagnation periods, flow was triggered by a solenoid valve attached to each of the
three test systems, controlled by an electro-mechanical pin timer. After the solenoid valves opened,
all pipe systems discharged into a common drain.
The normal mode of operation was to collect samples routinely after 8 hours of stagnation.
However, the protocol was varied during the course of the experiments to examine copper levels
after various standing times, especially in relation to the depletion of free chlorine residual and
dissolved oxygen. Approximately 31.5 litres of test water passed through each pipe section per day
(8.3 gpd).
Of the several experiments conducted thus far, one is particularly relevant to the issues of this
investigation, and will be covered here. The water quality summary for this experiment is summa-
rized in Table 8.
59
TABLE 8. WATER QUALITY IN THE COPPER PIPE STUDY
Analyte Mean Std. Dev.
Lead, mg/L <0.002 0.000
Calcium, mg/L 36.8 1.4
Copper, mg/L <0.003 0.007
Iron, mg/L 0.003 0.010
Potassium, mg/L 2.8 0.3
Magnesium, mg/L 10.8 2.5
Manganese, mg/L <0.0005 0.002
Sodium, mg/L 21.0 3.3
Zinc, mg/L <0.001 0.007
Alkalinity, mg CaCO3/L 46.4 6.3
Sulfate, mg SO4/L 70.3 7.6
Chloride, mg/L 25.5 7.4
Silicia, mg SiO2/L 5.8 1.1
Nitrate, mg N/L 1.1 0.2
Ammonia, mg NH3/L <0.03 0.02
Phosphate, mg PO4/L <0.10 0.01
Total inorganic carbon, mg C/L 10.98 1.72
Free chlorine, mg CI2/L 0.4 0.0
pH, pH units 7.6 0.1
Temperature, degrees Celsius 25.† --
Ionic strength (computed) 0.0050†† 0.0006
† Approximated
†† Computed by WATEQX
Recirculation Solubility Experiments
New experimental systems were designed, constructed and put into operation by DWRD to
specifically address the development and confirmation of equilibrium solubility models for copper(II).
As of the time of this report, only one experimental run has been completed, and the data analyzed.
A set of recirculation experiments was performed to examine the interrelationships of pH and DIC
on cuprosolvency under highly-controlled conditions. These systems are patterned after those origi-
nally developed for USEPA asbestos-cement pipe studies and some experiments on lead and galva-
nized pipe corrosion.16 Several important modifications were made.
The major modification was the replacement of the stainless steel tanks with 200-L (53-gal)
heavy duty polyethylene tanks. All parts of the recirculation system connection plumbing were
made of PVC plastic or tygon.™ A magnetic drive centrifugal pump continually circulates water
through the 4-ft (1.2-m) length of 0.5-in (1.27-cm) ID type M hard-drawn copper piping at a rate of
approximately 3785 litres per day (1000 gpd). The pipe sections mounted in the recirculation
systems were chemically cleaned before the start of the experimental runs by the same procedure as
previously described for the coupons and pipes used in the single-pass pipe loop systems.
60
Water was synthesized for the first run of these experiments (reported here) by mixing approxi-
mately 3 parts of mixed-bed deionized Cincinnati tap water with 1 part of untreated Cincinnati tap
water. DIC was adjusted at the beginning of the run with NaHCO3. Precise chlorine residual
(redox potential) control was not possible in these systems, but the free chlorine residual was moni-
tored at least two times per week and sodium hypochlorite solution was added as necessary to
attempt to maintain a continual chlorine residual of between 0.2 and 0.9 mg Cl2/L. The pH was
generally stable, but was adjusted if necessary once per week, when samples were collected for
metals and DIC analysis. The pH of the tanks was controlled within ± 0.1 pH of the target value
weekly, but was usually within ± 0.05 pH. Some absorption of CO2 from the air occurred in these
low-DIC experiments, even with the floating lids. Therefore, the data analysis approach must take
this into account.
Complete water analyses were performed on water samples at the beginning, end, and once per
month. Filtration through a 0.1 µm polycarbonate filter membrane with all-plastic apparatus was
performed once per month. The filtration procedure followed that described previously for lead
studies, except the sample volume was reduced to conserve water in each of the tanks.62 The
soluble metal concentrations were statistically indistinguishable from the unfiltered samples, af-
firming the use of the unfiltered samples for the saturation index calculations for these systems.
For clarity, the filtered sample data points are not shown in the figures.
Data from experiments for DIC=5 mg C/L and pH 7, 8 and 9 are reported here. A partial
general summary of the background water quality for these three experimental runs is given in
Table 9. The concentrations of chloride and sodium increased over time, from the addition of
sodium hypochlorite for disinfection, and either HCl or NaOH for pH adjustment.
TABLE 9. WATER QUALITY FOR DIFFERENT RUNS OF RECIRCULATION STUDY
Tank 1 Tank 2 Tank 3
Analyte Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
Aluminum, mg/L 0.02 0.03 0.03 0.03 0.06 0.10
Lead, mg/L <0.002 0.000 <0.002 0.000 <0.002 0.000
Calcium, mg/L 12.5 0.2 12.5 0.2 11.1 0.3
Iron, mg/L <0.004 0.00 <0.004 0.00 <0.004 0.00
Potassium, mg/L 0.8 0.1 0.8 0.1 1.2 0.2
Magnesium, mg/L 3.8 0.1 3.9 0.1 3.7 0.1
Manganese, mg/L <0.0005 0.00 <0.0005 0.00 <0.0005 0.00
Sodium, mg/L 15.8 4.4 18.0 5.2 19.7 3.8
Zinc, mg/L 0.01 0.00 0.01 0.00 <0.001 0.00
Sulfate, mg SO4/L 32.6 1.1 33.5 0.1 30.7 1.1
Chloride, mg/L 20.0 5.4 20.1 6.4 17.7 5.6
Silica, mg SiO2/L 0.8 0.1 0.8 0.0 0.7 0.1
Nitrate, mg N/L 0.3 0.0 0.4 0.0 0.3 0.0
Ammonia, mg NH3/L <0.03 0.02 <0.03 0.00 <0.03 0.01
Phosphate, mg PO4/L <0.10 0.00 <0.10 0.00 <0.10 0.00
Total inorganic carbon, 5.1 0.5 5.2 0.3 5.3 0.3
mg C/L
Free chlorine, mg Cl2/L 0.6 0.3 0.5 0.3 0.3 0.1
pH, pH units 7.1 0.1 8.0 0.1 9.0 0.0
Temperature, degrees Celsius 25.6 1.1 25.7 1.2 24.8 0.7
Ionic strength (computed) 0.0024 0.0002 0.0025 0.0002 0.0024 0.0001
61
Data Analysis Approaches
Experimental data has been analyzed in several ways to try to identify factors controlling the
equilibrium solubility of copper in the experimental systems. As pointed out by Williams, second-
ary minerals formed by exposure to natural water systems, such as in oxidation zones around ore
deposits, marine corrosion products, and archaeological artifact corrosion products, may serve as
“metal ion buffers” that control concentrations in the water.116 Frequently, these secondary miner-
als are oxide, hydroxide, hydroxy-sulfate, hydroxy-carbonate, or hydroxy-chloride compounds.
These kinds of natural environments are similar in many respects to the conditions of exposure of
plumbing materials in drinking water systems. Hence, many computational tools and approaches
used by geochemists to investigate natural processes involved in governing metal levels are also
highly applicable to corrosion research. Various applications of geochemical modeling programs,
and some constraints on their use have been reviewed by Basset and Melchior.117 Similar applica-
tion of a geochemical modeling program based on the original WATEQ and WATEQ2 codes to
investigate lead corrosion was previously described.62
The concentrations of all possible detectable background water constituents cannot be pre-
cisely controlled at all times to desired levels in the different types of corrosion and metal leaching
experiments used in the DWRD program. Therefore, the data analysis approach to test for solubil-
ity controls on copper levels must be robust enough to compensate for this experimental difficulty,
and for differences in several constituents from experiment to experiment. Therefore, in addition to
graphical analysis, complete background water chemistry information was combined with copper
and other metal leaching data and was evaluated with a modified version of the computerized
equilibrium aqueous speciation model WATEQX.112 Modifications by the authors to WATEQX
included expansion of the size of the aqueous and solid species arrays to accommodate copper,
lead and zinc species of interest in USEPA corrosion research studies.
The first step in employing WATEQX for the data analysis was the adjustment of the thermo-
dynamic data base to be consistent with the best known set of aqueous species, solid species, and
their equilibrium constants to characterize the system under study. A critically-reviewed set of
equilibrium constants and temperature functions for major potable water constituents has been
published by Nordstrom, et. al.118, among others. Other selections of species and equilibrium con-
stants used in WATEQX are given in Table 2.
The WATEQX program was used to compute the ionic strength of the solutions, to derive the
DIC concentration from titrimetric alkalinity determinations (carbonic acid equivalence point) and
pH as a check, and to provide calculations of the aqueous speciation of the various metals, ligands,
62
and other constituents in the test waters. Correction is made by the program for non-carbonate
contributions to alkalinity, such as HPO42- and NH3(aq).14,30,31,110
The computer program was also used to determine the ion activity products and saturation
states for various solids of interest14,15,31,62,117. For example, for the pure Cu2(OH)2CO3 (malachite)
dissolution reaction given in Equation (29), the activity product (IAPmal) expression is defined as:
(37)
wherei ( gamma) represents the activity coefficient for ion i (i=Cu2+, CO32+, or H+) and the paired
brackets [ ]i represent the mol/L concentration for ion i. The assumption is made that the activity of
water and of the pure solid is unity. At equilibrium, the ion activity product of any solid s (IAPs) is
equal to the solubility constant Ks0 for the reaction written exactly the same way as the IAP expres-
sion. Therefore, a measure of the saturation state of any solid s may be defined as the “Saturation
Index,” or
(38)
where equilibrium is indicated by a value for SIs of 0. Thus, if SIs > 0, the solution is in an oversatu-
rated state with respect to that solid, so a tendency toward precipitation to re-establish equilibrium
is indicated. Conversely, if SIs < 0, a tendency toward dissolution is indicated.
Plummer has described the use of the Saturation Index to attempt to identify mineral equilib-
rium processes controlling the concentration of several metals in ground water systems119. In the
context of these recirculation solubility, pipe, and coupon tests, the values for SI should approach
(but not exceed) zero from the negative direction. Systematic positive values for SI beyond those
attributable to analytical imprecision or error may indicate: possible errors in one or more equilib-
rium constants for copper species; very slow precipitation kinetics of passivating films on the pipe;
the formation of mixed solids of higher solubility than pure solids; or complexing species omitted
from the aqueous model that enhance metal solubility.
An example of the use of the SI to follow a water chemistry process is as follows. Suppose a
water is supersaturated with a solid, such as calcium carbonate in water leaving a lime softening
plant. The SI value for that solid will start out at some positive value dictated by the water chem-
istry at the end of the process, and will then decrease toward zero as precipitation takes place.
Because there is some ambiguity to the interpretation of the Saturation Index in some systems,
mineralogic observations such as X-Ray, petrographic or chemical scale analysis and the use of
mass-balance computations are necessary to corroborate hypothesized chemical processes and solid
phases.119
IAPmal =
γ
Cu2+
2Cu2+
[]
2
γ
CO3
2−CO3
2−
[]
γ
H
+
2
H
+
[]
2
SIs=log10 IAPs
Ks0
63
A particular advantage of the Saturation Index approach is that it is most sensitive to only a
small number of analytes, plus the accuracy of the equilibrium constants. Emphasis can then be
applied to optimizing analytical precision and accuracy of those particular analytes, and in their
experiemental control. For example, the only direct variables involved in computing SI for the
copper hydroxide and hydroxycarbonate minerals are pH, copper, and carbonate ion. Imprecision
in the estimation of SI results from the analytical uncertainties in determining pH, DIC, and copper
and with the uncertainty associated with the solubility constant Ks0. For many corrosion systems,
the uncertainties in Ks0 and errors in pH are the largest sources of error in the SI calculations.
Additional uncertainty in interpretaion of the computed SI is caused by kinetic restraints governing
whether or not a solid dissolves or precipitates, solid impurities, coprecipitation, crystal poisoning,
or the interaction of complexes not included in the computation of IAPs. The effects of ionic strength
and competitive complexation of other metals and ligands are secondary, and of small magnitude.
Therefore, the SI approach automatically compensates for changes in sodium or chloride concen-
trations, for example, as occur in the recirculation solubility experiments, or small weekly differ-
ences in pH or DIC. The approach similarly compensates for variations in calcium concentration,
sulfate concentration, pH or other variables from tank filling to tank filling in the other experiments
cited in this report.
After the speciation and SI calculations were performed, special FORTRAN programs were
written to extract saturation indices and other specific information from the WATEQX output, and
reformat it for graphical analysis by other software.
64
Experimental Results
Oxidation Rates and Stagnation Curve Behavior
Because of the observations in the European studies of copper levels that reached maxima and
then decreased upon further standing, a detailed investigation was deemed necessary. The reported
behavior was in contrast reported the experience with lead pipe, where stagnation curves were
usually found to be explained by a model that assumed lead levels were controlled by radial diffu-
sion up to the point of saturation with a passivating solid16,35. Also, examination of data collected
from the EPA coupon study indicated that copper levels in samples standing over weekends for 72
hours were systematically higher than those taken after 24 hours.
During initial experiments with the copper pipe loop study, chlorine residuals (Table 7) were
found to deplete in a matter of hours. As the experiment continued for several months, the rate of
chlorine depletion became lessened. After approximately 250 days of operation, the triplicate
copper pipe loops receiving the pH-adjusted Cincinnati tap water were sampled at different stagna-
tion intervals over several weeks to determine an empirical stagnation curve. Free chlorine re-
sidual and dissolved oxygen concentrations were monitored, in addition to copper concentration.
Insufficient sample volume was available to analyze pH.
Figure 15 illustrates changes in the concentrations of copper, free chlorine residual, and dis-
solved oxygen (DO) for a series of stagnation experiments. Because the starting concentrations of
DO and chlorine residual were not always the same, their decreases are expressed as a percentage
of the original feed water concentrations. Although chlorine was greatly consumed by reaction
with the piping, some was still present up to at least 72 hours. The DO data indicates that, as
expected, the stronger oxidant (hypochlorous acid and hypochlorite ion) was mostly consumed
first, before the oxygen depletion started. The copper concentrations did not essentially stabilize
until after at least 48 hours.
Computations of the Saturation Index for three important copper(II) solids using the WATEQX
computer program and complete water chemistry data (see Table 7) are shown in Figure 16. The SI
approach was used to compensate for minor differences in water chemistry from one stagnation
period to the next. The water is shown to closely approach equilibrium saturation with Cu(OH)2(s)
(cupric hydroxide), ie. SI=0, from the undersaturated (negative) side. The calculation assumed the
KS0 value for reaction (25), representing the cupric hydroxide solid with the high molar surface.
Evidence from previous studies argued that it is the likliest form of cupric hydroxide to initially
precipitate. The curves for the SI values for CuO(s) (tenorite) and Cu2(OH)2CO3(s) (malachite) are
65
Figure 15. Relationship among depletion percentage for free chlorine residual, dissolved
oxygen, and copper concentration over varying stagnation lengths, after 250 days of pipe use.
66
Figure 16. Saturation index changes for three solids during stagnation of different durations in
copper pipe loop study. Pipe was aged approximately 250 days
67
appreciably above equilibrium values, indicating significant thermodynamic supersaturation. Be-
cause the copper concentration had essentially stabilized, the calculations are not consistent with
the formation of malachite being the mechanism for controlling the copper(II) level in solution in
these experiments. Further, precipitation of malachite or tenorite was not indicated by turbid water,
erratic copper concentrations indicitive of colloidal dispersion, a dropping SI from positive values
toward zero, or by dropping copper levels.
The empirical stagnation curve data was then compared to the predictions of the Kuch and
Wagner diffusion model.35 The maximum copper(II) concentration used as input into the equation
was that calculated for the average water quality by the CU2SOL computer program. The empiri-
cal and theoretical curves are displayed in Figure 17, with error bars representing standard devia-
tions of replicate samplings. While the theoretical solubility could not be expected to match ex-
actly, the shape of the observed stagnation curve significantly differs from the diffusion-based
model. The initial limb of the stagnation curve for the experimental system (over the first 0-2 hr,
particularly) is of a much more shallow slope than that produced by the curve representing the
assumption of a simple radial diffusion process. This discrepancy indicates that there appears to be
a reaction rate-limiting step for copper oxidation, that is not applicable to lead. This limiting step
slows the attainment of the equilibrium solubility level. Subsequent examination showed that the
internal pipe surface had little surficial film developed on it, but action of an extremely thin and
effective diffusion barrier can not be precluded.
There are several important implications concerning cuprosolvency that result from these studies.
• In water supplies maintaining oxidizing disinfectant residuals, copper oxidation and
dissolution into the water will take place until that residual is depleted. However,
copper levels will stop climbing after saturation equlibrium is reached with a passi-
vating solid (eg. Cu(OH)2). If dissolved oxygen is present in addition to the oxidiz-
ing disinfectant, continued oxidation and dissolution will take place until the oxy-
gen is then consumed, or passivation occurs.
• The profile of copper concentration versus time in oxidizing systems will depend
upon the water chemistry and the relative stabilities of copper(I) and copper(II)
aqueous species. The copper concentration in the water may continue to rise for
many hours, perhaps substantially beyond the “overnight standing” period normally
used for sampling, depending on the operative kinetics in a given system. Com-
plexation of either copper(I), copper(II) or both may be important in these systems.
• In undisinfected water systems having low levels of dissolved oxygen, such as many
closed ground water systems, complicated copper versus time profiles may result,
as the initial small quantity of oxygen is consumed. This behavior is consistent with
the observation of short-term copper concentration maxima, followed by decrease
in concentration, reported in several studies.73,75,76,86,87,120 Copper(I) aqueous chemi-
cal speciation and solid reactivity may become extremely important in these sys-
tems.
68
Figure 17. Schematic illustration comparing metal concentrations that would be observed after
standing different amounts of time given different controlling chemistry factors.
69
• Different parts of a distribution system, particularly ones fed by a combination of
different water sources (such as wells and conventionally-treated surface water),
may exhibit different stagnation profiles.
• Significant variability in copper levels from site to site in the same water system
will likely be found in water after the normal number of hours of stagnation found in
most sampling programs. This results from the interaction of aqueous complex-
ation and oxidation chemistry of copper, plus effects of the existence or develop-
ment of passivating films in the pipe.
The Effect of pH and DIC on Cu(II) Solubility
Although studies are not complete, data collected thus far from some experimental systems can
be used to test aspects of the “cupric hydroxide model.” Approaching solubility equilibrium from
both oversaturation and undersaturation should be a useful tool for examining the validity of the
solubility predictions.
Patterson has reported on several studies of the use of the carbonate and hydroxide systems for
the removal of copper and other heavy metals from wastewater, which provide a reasonable test for
this solubility and speciation model.121 Precipitation experiments were performed over 24-hour
periods at constant pH and DIC. Carbonate concentrations ranged from about 2 mg C/L (10-3.8 mol/
L) to 8 mg C/L (10-3.2 mol/L), over the pH range of 6 to 13. Figure 18a shows the reported precipi-
tation data superimposed on the theoretical solubility curves for the formation of crystalline CuO(s)
(tenorite) and Cu2(OH)2CO3(s) (malachite), as would be predicted by the traditional copper model-
ing approach. For the purposes of the calculations, the ionic strength was assumed to be 0.01, and
the temperature to be 25°C. This solubility model condsiderably underestimates the observed
copper concentrations throughout the experimental pH range. In contrast, Figure 18b shows the
reported precipitation data superimposed on the theoretical solubility curves for the formation of a
fresh Cu(OH)2(s) (cupric hydroxide) precipitate. Except for data at both extremes of DIC in the pH
range of 9.5 to 11 where the solubility data is systematically slightly underestimated, the theoretical
curves match the data reasonably well. Even over this small range of DIC, the trend of increased
solubility with DIC can be discerned in the pH range up to approximately 9.5. The data clearly
indicate a higher solubility than would be predicted on the basis of assuming Cu2(OH)2CO3(s)
(malachite) formation.
The cause for the discrepancy between pH 9.5 and 11 is not known, and could be caused by
either experimental factors or speciation model inaccuracies. Previously, problems with uncertain-
ties surrounding stability constants for a suite of hydroxide, carbonate and hydroxycarbonate com-
plexes for copper(II) were noted, and the discrepancy occurs in the pH range that would be ex-
tremely sensitive to such errors, particularly in the formation constants of the complexes Cu(OH)2°,
Cu(OH)3-, CuCO3OH-, and CuCO3(OH)22- which are known to be among the most highly uncertain.
Data from the five experimental runs of the USEPA coupon study can also be the basis for a
check on the validity of the cupric hydroxide model. The complete water chemistry analyses (Table
70
Figure 18. Precipitation date of Patterson (1981) superimposed on two theoretical models for
Cu(II) solubility at low DIC concentrations. Models computed for I = 0.01, and 25°.
Controlling solids were assumed to be: a)CuO(s) and Cu2(OH)2CO3(s); b; Cu(OH)2(s)
71
72
6) plus copper values from 72-hour standing samples (except Run 3) were used to compute the
Saturation Indices for several plausible solids that could conceivably control copper solubility.
Again, within the limits of anlaytical and thermodynamic data error, the solid whose Saturation
Index is consistently the closest to zero for the long equilibration times of the experiments repre-
sents the most plausible candidate for controlling the copper level in the water. The results are
summarized in Figures 19a-19e. Because of the small sample volume and handling, pH could not
be determined on the stagnation sample, and was assumed to remain unchanged upon standing for
the SI calculations. This will introduce some error, but likely not enough to be critical to this
exercise. Two of the three experimental runs without orthophosphate addition (Run 2 and Run 5),
plus the low level 0.5 mg/L orthophosphate experiment (Run 4) are extremely consistent with the
hypothesis of control by Cu(OH)2(s) (cupric hydroxide), and not at all consistent with control by
malachite formation.
The case of the 3 mg/L orthophosphate dosage (Run 3) is somewhat ambiguous with respect to
either control by Cu(OH)2(s) or Cu3(PO4)2.2H2O(s). At the pH of the experiment, which was se-
lected on the basis of exploring lead solubility relationships, there would only be a small difference
in copper level if the orthophosphate solid formed instead of the hydroxide. The least-soluble
literature solubility constant for Cu3(PO4)2.2H2O(s) was used for the calculation (Table 2), and it is
highly uncertain. The data fall virtually midway between literature solubility constant values for
Cu3(PO4)2(s) and Cu3(PO4)2.2H2O(s). Therefore, the data are inconsistent with malachite forma-
tion, but they do not clearly fit a model based on cuprosolvency control by cupric hydroxide passi-
vation as opposed to cupric orthophosphate passivation. None of the coupons were covered by
enough scale to enable good microscopic, elemental or X-Ray diffraction analysis of the films to
confirm the proposed mechanisms. A hypothesis of orthophosphate acting as an inhibitor of the
copper oxidation rate or as a creator of a thin film possibly of amorphous crystal structure can not
be precluded.
The first experiment (Run 1) reflects a large discrepancy with any solubility model based on pH
and DIC. The high concentration of sulfate in this system (Table 6) may have provided an unfore-
seen complication in this study, possibly interfering with the formation of a tight and adhering
cupric hydroxide/oxide film79,122. Another possible influence is that there was decreased pH buffer-
ing in this system relative to the other runs because of the pH being nearer the carbonate system
buffer intensity minimum14,15,30. A large pH increase resulting from hydroxyl ion production during
copper oxidation could put the solution into the field where the model may tend to underestimate
the solubility as a result of errors in formation constants or presumed aqueous speciation.
Figure 20 shows the data from the five experimental runs superimposed on a solubility dia-
gram, with curves corresponding to the different average experimental DIC and orthophosphate
concentrations. The figure also shows the computed solubility line for one set of Gf° data for
Cu4(OH)6SO4.H2O, which is in much closer agreement to the observed data for Run 1. Increases in
general corrosion rates for copper under oxidizing conditions caused by sulfate ion have been
reported,95,122 but the applicability of those investigations to this system has not been firmly estab-
lished. As previously noted, Weiser, et. al. have reported the precipitation of various mixtures of
73
Figure 19. Computed saturation indices for important solids in copper coupon study runs 1-5; a)
Run 1, pH = 8.5, 72 hour standing time; b) Run 2, pH 7.0, 72 hour standing time; c) Run 3, pH
7.5, 3 mg PO4/L, 24 hour standing time; d) Run 4, pH 7.5, 0.5 mg PO4/L, 72 hour standing time;
e) Run 5, pH 7.5, 72 hour standing time.
74
Figure 20. Comparison of theoretical and observed copper levels for coupon study. The wide
dashed line indicates the predicted solubility of a hydrated cupric hydroxysulfate in comparison
to cupric hydroxide or cupric orthophosphate phases.
75
cupric hydroxide and a basic cupric sulfate salt in various proportions, depending upon the ratio of
cupric sulfate salt to alkali (NaOH) in a solution,92 and the sulfate term in the empirical equation
(27) for copper levels after stagnation from KIWA studies76 leads to higher copper levels with
higher sulfate concentrations. More investigation needs to be done to determine if hydroxide and
basic sulfate salt coprecipitation occurs on pipe surfaces, or if cupric oxide or hydroxide formation
is inhibited in the presence of high sulfate concentrations, leading to solubility control by copper
hydroxy-sulfate minerals.
The most directly-applicable data collected thus far for establishing verification of modeled
pH/DIC effects were generated from the USEPA recirculation solubility experiments described
above. A background water quality data summary for the experiments with DIC=5 mg C/L at pH 7,
8 and 9 is given in Table 8. Figures 21a-21c show the saturation indices for malachite, tenorite and
cupric hydroxide versus time. Clearly, the data for all three systems are much more consistent with
copper control by fresh cupric hydroxide, than either Cu2(OH)2CO3 (malachite) or CuO (tenorite).
Approximately every fourth data collection included metal samples filtered through polycarbonate
0.1 µm filters (datapoints not shown). The soluble metal concentrations were statistically indistin-
guishable from the unfiltered samples, affirming the use of the unfiltered samples for the Saturation
Index calculations for these systems.
Interesting results were obtained from X-Ray diffraction and Energy-Dispersive X-Ray analy-
ses of the deposits formed on these experimental pipes. Consistent with the oxidation multi-layer
theory discussed in a previous section, both Cu2O(s) (cuprite) and broadened peaks corresponding
to finely-crystalline or poorly-ordered CuO(s) (tenorite) were found on all pipe specimens in these
systems where chlorine residuals were constantly maintained and dissolved oxygen levels were
above 6 mg/L. The cuprite layer seemed to be better-developed at pH 9 than at pH 7. At pH 7, a
trace of malachite was found, but not on either of the other two pipes. Indeterminable quantities of
several other solids were found, however, even in these relatively pristine systems. Significant
peaks were found in the pH 8 and pH 9 pipes for posnjakite, Cu4(OH)6SO4.H2O(s), even though
sulfate levels were only approximately 30 mg/L. This mineral has also been reported in some
copper pipes in a German study of pipe corrosion in a hospital.123 Diffraction peaks likely corre-
sponding to the solid Cu(Cl,OH)2.2H2O (calumetite) were found, particularly at pH 8 and 7. Quali-
tative elemental analysis confirmed the presence of S, Cl and also Al on the pipe surfaces. A large
peak and a secondary peak apparently consistent with the solid CuAl4SO4(OH)12.3H2O
(chalcoalumite) were found on the pH 7 specimen, but only a corresponding minor peak was found
on the pH 8 sample. From the qualitative elemental analysis, the Al concentration on the pipe
appeared higher at pH 7, consistent with the general trend in solubility of many aluminum miner-
als, and a decrease in aluminum in the water during the pH 7 experiment. The presence of alumi-
num on the pipe is also noteworthy because of the low Al concentration in the water in all three
experimental systems, which suggests possiby a strong role for aluminum in the formation of natu-
ral diffusion barriers in plumbing and distribution systems.
Some exploratory bench-top experiments were conducted to determine if cupric hydroxide on
the pipe surface would be stable enough to be detected by the X-Ray analysis procedure, because
76
Figure 21. Computed saturation indices for three copper (II) solids in recirculation solubility
experiments with DIC=5 mgC/L at: a) pH 7.0; b) pH 8.0; and c) pH 9.0.
77
none was clearly found on the experimental pipe surfaces (except for the related calumetite). Light
blue-green cupric hydroxide was precipitated from sodium hydroxide solution by addition of cu-
pric nitrate solution. The precipitate was removed by filtration, and within several hours of expo-
sure to air, where significant moisture was present on the filter paper, the solid had converted into
black cupric oxide with broad diffraction peaks, like that found in the pipe scale samples. These
observations agree well with the aging phenomena found in a study of copper precipitation from
alkaline solution previously reported.72 They also show that a similar change in the mineralogy of
the pipe deposit would likely occur during the pipe removal, scraping and sample mounting pro-
cesses. Therefore, the analyses of the pipe samples from these experimental systems are consid-
ered to be consistent with the saturation index calculations showing cuprosolvency control by cu-
pric hydroxide.
The Effect of Aging on Copper Levels in the Water
An implication of considerable importance that results from the cupric hydroxide model is that
given identical stagnation times, equivalent water usage patterns, and the absence of unrelated
mineral deposition producing diffusion barriers, standing copper levels should decrease over time.
The aging process consists of both the recrystallization and dehydration of Cu(OH)2(s) to CuO(s)
described previously,31,65,66,72 and the slow formation of Cu2(OH)2CO3(s) in the lower pH range
below the CuO(s) stability boundary. Wagner has indicated this trend is evident in a German
corrosion study published in 1988.9
In the United States, most Lead and Copper Rule monitoring programs do not produce samples
that represent a broad spectrum of plumbing ages, or samples that are appropriately targeted for the
copper pipes. However, a recent study in California conducted by Larry Walker Associates, Inc. for
the Contra Costa Central Sanitary District was completed to estimate the impact of residential
plumbing as a source of metal contamination of water received by the sewage treatment plants†.
Approximately sixty homes were sampled in the district, of which approximately thirty residences
were fed by water supplied by the Contra Costa Water District (CCWD), and the rest by water
produced by the East Bay Municipal Utility District (EBMUD). The residential plumbing systems
ranged in age from 1 to 72 years. A summary of general water qualities of the areas sampled is
provided in Table 10.
† "Contra Costa Central Sanitary District Residential Metal Study," B. Brandenberg and T. Ross of CCCSD; B. Elzufon, C. Malone and T.
Grovhoug of Larry Walker Associates, Inc., Davis, California, engineering study report, July-August 1993.
78
TABLE 10. APPROXIMATE WATER QUALITY COVERING SITES IN THE CALIFORNIA
WASTEWATER LOADING STUDY.
Parameters*CCWD (Avg.) EBMUD (mix)
pH (units) 8.5 8.9
Total Alkalinity (as CaCO3)70 25
DIC (computed as C) 17 6
Calcium 16 8
Magnesium 18 1
Sodium 91 2
Potassium - 10.5
Chloride 142 5
Sulfate - 3
Fluoride 0.93 0.9
Nitrate (as N) - 0.1
Silica - 6
Conductivity (µS) 731 70
* mg/L unless noted otherwise
The sampling protocol used was as follows. The tap to be sampled was flushed the night prior
to sampling, and the time recorded. In the morning, the resident turned on the tap prior to any other
water usage, and discarded the first 500 mL. The next 1000 mL was immediately collected for
analysis, so that the copper plumbing was targeted by the sample. The time of this sampling was
also recorded. Standing times ranged from 6 to 14 hours, with an average of 7.65. However, 83%
of the samples were between 6.2 and 9.1 hours. Because of the observations of slow oxidation and
solubilization rates of copper in the EPA studies described in an earlier section, and the nearly
linear increase observed repeatedly for the first 10-15 hours in disinfected systems, the copper
leaching amounts can be approximately normalized for comparison by dividing each of the con-
centrations by its standing time to convert them into rates. Figure 22 shows the resulting distribu-
tion of copper leaching rates, expressed as micrograms of copper per hour, for the different plumb-
ing ages.
Two aspects of the data distribution are notable, though caution must be exerted to avoid
overinterpreting the data because of its limitations and assumptions used. First, based on pH and
DIC considerations shown in Table 9, reference to Figure 4 predicts that for fresh copper piping the
water from CCWD would be considerably more cuprosolvent at equilibrium than the water from
EBMUD, even though neither would really be considered actually corrosive toward copper. This
appears to be borne out by the distribution of copper leaching rates, particularly for residences
newer than about 20 years old. Second, there is a downward trend in both sets of data, indicating
that copper leaching into the water lessens with aging of the piping in the homes, up to some point
where the distribution becomes considerably more random. This point appears to occur in the 20-
35 year range, but cannot be determined very reliably from this dataset without more stagnation
curve profile information to validate the assumptions. For example, there is no direct quantitative
predictor of copper leaching rate based on computed equilibrium solubility, and vice versa. If
79
Figure 22. Distribution of adjusted copper leaching rates versus age of plumbing for one litre
samples from the residential metals study of the Contra Costa Central Sanitary District.
80
water in contact with pipe at one or more of these sites reached the equilibrium copper concentra-
tion before it was drawn, then dividing the sampled concentration by the standing time will bias the
resulting computed rate to be erroneously low. If this phenomenon tended to occur more readily in
older pipes with well-developed passivating films of low solubility (such as malachite), the slope
of the aging trend would decrease too rapidly.
Although the study data cannot be shown to prove the validity of the cupric hydroxide model
for young copper plumbing systems, they do indicate behavior that is consistent with assumptions
of the model, and logical predictions from it.
81
Conclusions
Significance of pH and DIC in Cuprosolvency Control
This research has shown conclusively that trends in copper concentrations resulting from uni-
form corrosion are predictable for a range of appropriate water qualities. Given that field condi-
tions frequently do not reflect equilibrium conditions, some discrepancy between the model predic-
tions and the results of normal tap water monitoring programs is understandable. Nonetheless,
water quality objectives for copper corrosion control by pH and DIC adjustment can be predicted
without the necessity of complicated experimental and field studies, by using the described equilib-
rium chemistry model.
The relationship between DIC and copper(II) solubility has been shown to be very complex.
DIC can play several significant roles as follows, depending upon its concentration, other water
chemistry factors, age of plumbing, water flow amount and pattern of use.
• In new plumbing or at high pH, where cupric solubility is controlled by either cupric
hydroxide or cupric oxide, DIC complexes dominate copper speciation above ap-
proximately pH 6.5, resulting in increased cuprosolvency. The solubility enhance-
ment effect is strongest in the DIC range of 0 to 20 mg C/L, but the predicted soluble
copper at equilibrium is still below about 1 mg/L.
• DIC serves to control the buffer intensity in most water systems, therefore, suffi-
cient DIC is necessary to provide a stable pH throughout the distribution system for
corrosion control of copper (and lead). In treatment practice, the increase in DIC to
ensure pH control through buffering will probably need to be offset by increasing
pH to maintain lowered cuprosolvency.
• Possibly offsetting the solubility enhancement of copper(II) by carbonate complexes
is the possibility that moderate DIC levels would logically accelerate the formation
of passivating Cu2(OH)2CO3(s) (malachite) films in the pH/DIC region where it
would be thermodynamically stable. Thus, enhancement of conditions that would
hasten the formation of malachite, CuO (tenorite), or both, relative to cupric hy-
droxide would result in a net lower copper(II) solubility, even with some additional
carbonate complexation.
82
• In the region of pH from approximately 8.5 to 10, this solubility model occasionally
systematically predicts slightly lower copper(II) solubility than has actually been
observed in some well-controlled pipe loop and precipitation studies. The discrep-
ancies are most likely the result of formation constant errors, inadequacies of under-
standing copper(II) aqueous complexation with hydroxide and carbonate, or the
presence of metastable copper solids associated with anions other than carbonate or
hydroxide.
Significance of pH and Orthophosphate in Cuprosolvency Control
Because cupric hydroxide solubility has been shown to be a better estimate of cuprosolvency
tendencies in many water systems than previously presented models based on Cu2(OH)2CO3 (mala-
chite) equilibrium, orthophosphate in sufficient dosage is now predicted by the models to have an
ability to further reduce copper levels in the pH range of approximately 6.5 to 7. The necessary
dosage is hard to quantify, but the calculations suggest that 3-5 mg PO4/L orthoposphate may be
necessary to achieve substantial improvements in cuprosolvency over the DIC/pH system at ap-
proximately pH 8, but perhaps only 1-3 mg PO4/L at approximately pH 7.
While orthophosphate in sufficient quantity may decrease cupric ion solubility relative to the
formation of cupric hydroxide solid, the calculations suggest that for pH above approximately 6.5
the equilibrium solubility of Cu2(OH)2CO3 (malachite) is still lower. Therefore, lower solubility
levels may ultimately be achieved by aging in the absence of orthophosphate. The interesting ex-
periments of Werner, et. al.86,87 on aged copper pipe possibly suggest a complicated role for ortho-
phosphate. These experiments, though performed in the initial presence of oxygen but in the ab-
sence of a chlorine residual, showed that at a high DIC level where formation of malachite would
be most likely, higher copper levels are produced by the presence of orthophosphate at pH 7.2. This
possibly indicates inhibition of the normal transformation that would take place from cupric hy-
droxide to malachite with aging. More experimentation with orthophosphate is clearly warranted.
More investigation is also still needed to ascertain objectively and consistently if any synergis-
tic cuprosolvency reduction effect exists from the use of blended phosphates rather than orthophos-
phate, as has been suggested by one research project.124
Implications for Controlling Highest Copper Exposures
The distinct aging process noted for recrystallization and development of cupric hydroxide,
cupric oxide, and cupric orthophosphate films raise an important caution for determining the high-
est exposures to copper in drinking water. While the Lead and Copper Rule2 was specifically
intended to bias the sampling site selection towards locations with high relative risks for lead expo-
sure, these sites do not generally correspond to commensurately high risks for copper exposure,
which include: newest construction and remodeling, areas with unstable pH, and dead ends.
83
Changes in solubility of copper as pipes age, and films build up, has been indicated in several
studies and by practical experience. This investigation has shown how such a phenomenon may
have a firm chemical basis that has heretofore been unappreciated. In practice, when attempting to
predict the impact of different water treatment scenarios on cuprosolvency, a fruitful strategy may
be to apply different solubility models depending on the general age of the plumbing systems
involved. For example, for older neighborhoods where Cu2(OH)2CO3 (malachite) and CuO (tenorite)
have had sufficient time to form an integral part of the passivating film, a better prediction of pH,
DIC and orthophosphate dosing impacts would be obtained assuming their presence (ie. Figures 6,
7, 9 and 13b). Conversely, for areas predominantly of new construction, the “cupric hydroxide
model” would be more applicable (ie. Figures 3,4,10, and 13a).
Examination of copper concentrations in coupon cells and pipe loop systems after various stand-
ing times show that the redox conditions of the water play a crucial role in defining stagnation
profiles for copper plumbing. These experiments are complementary to those reported in the Euro-
pean studies, by showing that under oxidizing conditions resulting from both chlorine residuals and
the presence of several mg/L of dissolved oxygen, copper concentrations continue to rise to satura-
tion levels of passivating cupric solids. This almost linear increase in concentration frequently
extends far beyond the 6-16 hour standing time requirement of regulatory sampling.
Significance of Cupric Hydroxide Model for Demonstration Studies
Utilities may encounter some complications in projecting pipe loop or coupon leaching
cuprosolvency data obtained in demonstration studies to behavior across the whole distribution
systems. In the absence of significant concentrations of orthophosphate, this research indicates
that for the short timeframes of most experimental studies conducted to satisfy Lead and Copper
Rule requirements (6 months to 2 years in most cases), cupric hydroxide will usually be the most
important solid phase, rather than CuO(s) (tenorite) or Cu2(OH)2CO3(s) (malachite). Both pH and
DIC effects will be magnified in the experimental systems, relative to significantly aged piping in
the distribution system.
Because of the slow oxidation rate relative to diffusion rates, it is unlikely that copper concen-
trations are at “equilibrium” when samples are taken from experimental pipe rig systems operated
to allow “overnight” standing times, unless the systems have been operated long enough that the
passivating films are well-developed, perhaps for years.
When orthophosphate dosing is tested, overestimates of its effectiveness could be obtained for
distribution system areas having old copper plumbing with well-developed passivating films. The
experimental systems may also be very sensitive to minor dissolved oxygen, chlorine residual,
sampling, and stagnation time fluctuations, because of the highly non-equilibrium nature of copper
behavior in the 6-16 hour time period.
84
Significance of Chlorination and Aeration on Copper Levels
The presence of oxidizing agents appear to increase the rate of copper dissolution, and promote
the formation of copper(II) aqueous and solid species, rather than copper(I) species. Confusing
data may be obtained from water systems that employ oxidative processes for different reasons. A
system never having used aeration or disinfection that institutes such a process may encounter
much higher copper levels than previously, because of the difference in solubility of copper metal
or copper(I) as opposed to copper(II) (see Figures 11-13). Sampling of water from copper piping
done under nonequilibrium conditions will also likely show fluctions depending on the levels of
oxidants in the system.
Future Research Needs
The absence of fundamental equilibrium constant and solubility constant data for important
aqueous species and solids involved in passivation film formation is the most significant constraint
on the ability to extend the model predictions in a more quantitative way to systems using corrosion
inhibitors, or with significant concentrations of other chemical species (eg. orthophosphate,
polyphosphate species, ammonia, natural organic matter, etc.) that would measurably impact
cuprosolvency. The reliability of the carbonate and hydroxide complexation predictions are also
weakened for pHs above approximately 8.5 because of the lack of adequate equilibrium constants
and experimental confirmation. The most comprehensive and reliable research into copper hy-
drolysis and carbonate complexation has focused on either systems with much higher copper con-
centrations than drinking water, where many polymeric cationic hydroxide complexes form (eg.
Cu2(OH)22+, Cu3(OH)42+, etc.), or under sea water conditions. The general uniformity of the pH of
sea water amongst the major oceans (about 8.0-8.3) and the high salinity make many of these
studies too restricted for drinking water application.
To most effectively address the data gaps, and to provide the most useful and practical informa-
tion to water utilities and consultants investigating cuprosolvency control, future research should
depart from the traditional “field study” orientation, and address several critical information needs
in the laboratory. These important needs are:
• Analyze pipe deposits from copper plumbing systems of various ages, to determine
the pH/DIC and time-dependent stability domains of Cu(OH)2 (cupric hydroxide),
CuO (tenorite), and Cu2(OH)2CO3 (malachite). Deposits should also be examined
to determine the water chemistry domains responsible for forming cupric
hydroxysulfate solids, such as Cu4SO4(OH)6.H2O, which may be much more impor-
tant in controlling cuprosolvency than has traditionally been suggested. These analy-
ses must include compound specific methodologies, because conventional elemen-
tal analysis techniques (such as x-ray fluorescence, energy-dispersive x-ray analy-
sis on an SEM, or ICP analysis of dissolved coatings) can not readily and accurately
differentiate among carbonate solids and oxide/hydroxide solids. Such differentia-
tion is critical in understanding the mechanism of passivation, and to developing a
quantitative model for cuprosolvency treatment.
85
• Analyze pipe deposits from copper plumbing systems in utilities using phosphate-
containing inhibitors, to determine the copper solids present. This is particularly
important for the pH range of approximately 6 to 8. Using this information, the
applicability of the existing solubility data for Cu3(PO4)2.2H2O can be assessed, and
new solubility experiments can be conducted to obtain the necessary constants and
temperature functions for the actual scale minerals found in real systems. Once
again, compound-specific analytical techniques are required for this research to be
highly useful.
• Determine if a relationship exists between DIC concentration and rate of malachite
or tenorite formation in the pH range of 7 to 9, that could be used to accelerate the
reduction of cuprosolvency.
• Conduct laboratory experiments to determine for the first time the solubility of im-
portant cupric orthophosphate solids that have been found in nature and which may
be important in plumbing systems (see Table 5).
• Refine the formation constants for the uncharged and anionic hydrolysis complexes
of Cu(II), such as Cu(OH)2°, Cu(OH)3- or other possible species. Further investiga-
tion of the hydrolysis of Cu(I) also appears to be needed. These determinations
should to be made at appropriate ionic strengths to enable extrapolation to I=0 for
modeling, and should encompass the temperature range of 5-55°C.
• Refine the formation constants for copper(II) carbonate and mixed carbonate/hy-
droxide complexes (eg. Cu(CO3)22-, CuCO3OH-, CuCO3(OH)22-) under the same con-
ditions as the hydroxide complex experiments suggested above.
• Investigate the effect of chloride and sulfate on the mineralogy of passivation films
formed, and their properties as oxidation barriers.
To gain further understanding of copper chemistry in drinking water and its likely response to
various water treatment changes in the future, several other research areas need considerable atten-
tion. Some of these areas are:
• Investigate the effect of free ammonia and chloramines on the speciation of the
dissolved copper, and the oxidation rate. In conjunction with this activity, analyzing
standing water samples for both Cu(I) and Cu(II) would help elucidate the complex-
ation and oxidation chemistry.
• Investigate the characterization of NOM present after typical drinking water treat-
ment and disinfection, and its impact on copper oxidation rates, aqueous speciation
and passivation film formation.
86
• Investigate the relative affinity and conditional stability constants for copper with
various polyphosphate species present in formulations used for iron and manganese
control, to determine the optimum conditions and mechanisms for the reduction of
copper dissolution. Also, investigate a possible pH-dependent tendency of
polyphosphate and silicate species to interfere with the formation of normal cuprous
and cupric hydroxide/oxide films.
• Investigate the “stagnation curve” behavior of copper in plumbing systems in rela-
tion to oxygen and chlorine levels, and different conditions of cumulative water
flow and relative flow/stagnation cycling. The impact of these factors on the miner-
alogy and permeability of passivation films needs to be determined, particularly in
relation to the replacement of cupric hydroxide films with malachite.
• The research presented here suggests malachite to be relatively unimportant in con-
trolling copper(II) levels in the short-term (months to years). Therefore, some re-
search in natural water systems may need to be reexamined for the role of basic
cupric sulfates or adsorption as significant controls of cupric ion activity under some
circumstances, instead of malachite as has often been assumed.
Some of the latter topics are currently under investigation by different research groups. How-
ever, the time constraints of drinking water and waste water regulations create a need for more
immediate activity and support in these research areas. Research areas beyond those listed here
will probably also be found to be important as more studies are done.
87
References
1. Lead and Copper. Final Rule Correction. Fed. Reg. 56:135:32112 (July 15, 1991).
2. Lead and Copper. Final Rule. Fed. Reg. 56:110:26460 (June 7, 1991).
3. Lead and Copper. Final Rule Correction. Fed. Reg. 57:125:28785 (June 29, 1992).
4. Marshall, W. Copper in Drinking Water: What the Lead and Copper Rule Tells Us and What
It Doesn’t Tell Us. Proc. AWWA Water Quality Technology Conference, San Francisco, CA,
Nov. 6-10, 1994
5. AWWA. Initial Monitoring Experiences of Large Water Utilities Under USEPA’s Lead and
Copper Rule. Water Industry Technical Action Fund, AWWA, Denver, CO, (1993).
6. Anon. “Copper Creating Headaches for Florida, Washington Utilities”. Waterweek,
March 29 (1993).
7. Anon. “Massachusetts Medium Systems Top EPA’s Latest Copper List”. Waterweek, June 7
(1993).
8. U. S. Environmental Protection Agency. Lead and Copper Rule Guidance Manual. Volume
I: Monitoring. American Water Works Association, Denver, (1991).
9. Wagner, I. “Kupfer im Trinkwasser von Hausinstallationen aus Kupfer”. Wasser. Abwasser,
129:11:690 (1988).
10. Gilbert, P. T. “Dissolution by Fresh Waters of Copper from Copper Pipes”. Water Treatment
and Examination, 15:165 (1966).
11. Millette, L. & Mavinic, D. S. “The Effect of pH Adjustment on the Internal Corrosion Rate of
Residential Cast-iron and Copper Water Distribution Pipes”. Can. Jour. Chem., 15:79 (1988).
12. Cruse, H. & von Franqué, O. “Corrosion of Copper in Potable Water Systems”. Ch. 5 In:
Internal Corrosion of Water Distribution Systems. AWWA Research Foundation/DVGW
Forschungsstelle, Denver, Colorado, (1985).
13. Stumm, W.& L., C.R. “Technical and Legal Aspects of Copper Tube Corrosion”. Jour. AWWA,
59:3:401 (1967)
88
14. Snoeyink, V. L. & Jenkins, D. Water Chemistry. John Wiley and Sons, New York, (1980).
15. Stumm, W. & Morgan, J. J. Aquatic Chemistry. Second Edition. John Wiley & Sons, Inc.,
New York, New York, (1981).
16. Schock, M. R. & Wagner, I. “The Corrosion and Solubility of Lead in Drinking Water”. Ch.
4 In: Internal Corrosion of Water Distribution Systems. AWWA Research Foundation/DVGW
Forschungsstelle, Denver, CO, (1985).
17. Adeloju, S. B. & Hughes, H. C. “The Corrosion of Copper Pipes in High Chloride-Low
Carbonate Mains Water”. Corros. Sci., 26:10:851 (1986).
18. Mor, E. D. & Beccaria, A. M. “Effect of Temperature on the Corrodibility of Copper and Zinc
in Synthetic Sea Water”. NACE Corros., 31:8:275 (1975).
19. Paulson, A. J. & Kester, D. R. “Copper(II) Ion Hydrolysis in Aqueous Solution”. J. Solution
Chem., 9:4:269 (1980).
20. Symes, J. L. & Kester, D. R. “Thermodynamic Stability Studies of the Basic Copper Carbon-
ate Mineral, Malachite”. Geochim. Cosmochim. Acta, 48:2219 (1984).
21. Schock, M. R. “Response of Lead Solubility to Dissolved Carbonate in Drinking Water”.
Jour. AWWA, 72:12:695 (1980).
22. Schock, M. R. Treatment or Water Quality Adjustment to Attain MCL’s in Metallic Potable
Water Plumbing Systems. Plumbing Matrials and Drinking Water Quality: Proceedings of a
Seminar, Cincinnati, OH, May 16-17, 1984, (1985).
23. Froning, M. H., et al. “An Improved Method for Calculation of Potential-pH Diagrams of
Metal-Ion-Water Systems by Computer”. Corros. Sci., 16:371 (1976).
24. Cotton, F. A. & Wilkinson, G. Advanced Inorganic Chemistry. Fifth. John Wiley & Sons,
Inc., New York, (1988).
25. Obrecht, M. F., et al. Integrated Design of Field Test Panel Pilot Unit for Investigating Pitting
Corrosion of Copper Water Tube by Potable Water Supplies. NACE Fourth International
Congress on Metallic Corrosion, (1972).
26. Mattsson, E. “Corrosion of Copper and Brass: Practical Experience in Relation to Basic Data”.
Br. Corros. J., 15:1:6 (1980).
27. Atlas, D. C., J. & Zajicek, O. T. “The Corrosion of Copper by Chlorinated Drinking Waters”.
Water Res., 16:693 (1982).
28. Reiber, S. “Copper Plumbing Surfaces: An Electrochemical Study”. Jour. AWWA, 81:7:114
(1989).
89
29. de Zoubov, N., Vanleugenhaghe C. & Pourbaix, M. “Copper”. Sect. 14.1 In: Atlas of Elec-
trochemical Equilibria in Aqueous S