Corrosion Science 42 (2000) 1801±1822
www.elsevier.com/locate/corsci
Copper corrosion in distribution systems:
evaluation of a homogeneous Cu2O ®lm and a
natural corrosion scale as corrosion inhibitors
Abhijit Palit a, Simo O. Pehkonen b
a
Environmental Engineer Analyst, The Cadmus Group Inc., 1901 N. Fort Myer Drive, Suite 1016,
Arlington, VA 22201, USA
b
Department of Chemical and Environmental Engineering, National University of Singapore, 10 Kent
Ridge Crescent, Singapore 119260
Received 4 August 1999; accepted 20 January 2000
Abstract
The aim of this work is to assess the performance of dierent corrosion scales as eective
copper corrosion inhibitors using synthetic waters and distribution system waters. The
major objective of the study was to evaluate the stability of an arti®cially synthesized Cu2O
®lm and naturally formed heterogeneous corrosion scales under the in¯uence of a number
of synthetic and real waters, using Electrochemical Impedance Spectroscopy (EIS). Cu2O
coatings were synthesized by using a special ¯ame aerosol deposition process. Based upon
the EIS data, conclusions regarding the stability of the dierent scales under typical water
quality conditions, (which are commonly encountered in water distribution systems), are
drawn. 7 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Copper; EIS
1. Introduction
The promulgation of the ``Lead and Copper Rule'' by the USEPA in 1991 led
to the reduction in the action level for copper to 1.3 mg/l in ®nished drinking
waters. This forced nearly 80,000 water utilities nationwide to monitor for copper,
which arises mostly from the uniform corrosion of copper plumbing materials [1],
and initiated signi®cant research to control copper corrosion in distribution
0010-938X/00/$ - see front matter 7 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 0 - 9 3 8 X ( 0 0 ) 0 0 0 2 4 - X
1802
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
systems. Regulatory monitoring data have been of limited use in the
understanding of copper chemistry in water conduits. Some of the major reasons
for this are:
. Concentration data of some important background chemical constituents that
might in¯uence copper corrosion (e.g., chloride, sulfate, silica, orthophosphate
etc.) may not have been necessarily collected.
. Although copper dissolution rates are highly dependent upon the age of the
passivating ®lms, adjustments for dierent plumbing ages may not have been
conducted, which is very important to correctly deduce dierences in
cuprosolvency resulting from chemical, rather than temporal factors.
. The collected water quality data may not be representative of the entire
distribution system.
Much research has been carried out in the past several years in the area of
copper corrosion. Both, weight loss methods and electrochemical techniques have
been utilized to assess the rate of copper corrosion. Feng et al. [2,3] have recently
conducted very thorough experiments in order to understand the rate limiting
processes of copper corrosion. They have carried out these studies by utilizing a
sophisticated spectroscopic tool of X-ray Photon Spectroscopy (XPS) for the ®lm
characterization as well as several electrochemical techniques in order to gain
insight into the important elementary processes controlling copper corrosion. The
processes include charge transfer, diusion of copper ions through copper oxide
®lm and diusion of copper ions in solution. Feng et al. [2,3] concluded that
under their experimental conditions, the rate limiting process (which controls the
overall corrosion rate) was the diusion of copper ions through the oxide ®lm.
Feng et al. [2,3] conducted Electrochmical Impedence Spectroscopy (EIS) and
XPS studies to understand the eect of pH, immersion time and rotation speed of
the electrode on copper corrosion and ®lm growth. Although much insight into
copper corrosion can be obtained from the detailed experiments and discussion by
Feng et al. [2,3], their studies are limited in terms of the dierent water quality
conditions under which corrosion ®lm development can occur in real distribution
systems. The only water quality parameter, which was varied in their studies, was
the pH. However, it must be stated that the focus of the study by Feng et al. [2,3]
was not on the eects of numerous water quality parameters on the ®lm
development and the rate limiting processes of copper corrosion.
Copper may exist in water in the monovalent (cuprous) or divalent (cupric)
states. In potable waters, copper undergoes the following electrochemical
transformations:
a Cu s $ Cu e ÿ ;
b Cu $ Cu 2 e ÿ ;
E 0 ÿ0:52 V
E 0 0:16 V
Due to the potentials for reactions (a) and (b) being much smaller than that of
the oxygen reduction reaction (O2(g) + 4H+ + 4eÿ t 2H2O; E 0 = 1.23 V),
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1803
copper pipes carrying water with dissolved oxygen continue to corrode until all
the oxygen is depleted or a precipitated copper oxide ®lm arrests the rate of
corrosion. There is some evidence that the transformation from Cu+ to Cu2+ is
the rate limiting factor during copper corrosion, with Cu+ existing in reversible
equilibrium with Cu metal at the pipe surface [4]. In drinking waters, the
predominant electron acceptors (i.e., oxidants for copper) are dissolved oxygen
and aqueous chlorine or chloramine species [5±7].
Some progress has been made in understanding the copper chemistry as far as
inorganic components in drinking waters are concerned [8]. It is now widely
believed that inorganic carbon species (i.e., bicarbonate, carbonate and dissolved
carbon dioxide), ammonia, sulfate and chloride are the main inorganic species
aecting the chemistry of copper corrosion. Evidence exists to suggest that
although chloride is very aggressive initially, it has long-term bene®cial eects and
may also counteract the deleterious eects of bicarbonate under certain conditions
[9]. Long-term copper levels have been observed to be governed by the
precipitation of solid phases like cupric hydroxide Cu(OH)2, malachite
(Cu2(OH)2CO3), tenorite (Cu2O) etc., each one being in the most
thermodynamically stable form at speci®c pH and DIC (dissolved inorganic
carbon) ranges [10]. However, a thorough understanding of the mechanisms of
copper corrosion as a function of chloride and DIC dosages at dierent pH and
background water quality conditions is lacking.
1.1. Copper (I) chemistry
Cuprous ion forms a weak and unstable hydrolysis species (CuOH0). On
account of this and the diculties in detecting and quantifying it, this species is
considered quite irrelevant in most reviews of copper chemistry [11,12]. The
0
formation of CuCOÿ
3 and CuHCO3 has been hypothesized to explain some trends
in oxidation rates of Cu(I) by dissolved oxygen. However, their kinetic and
thermodynamic data are not well documented [13]. Chloride can react with the
cuprous ion to form CuCl, which subsequently hydrolyzes to form Cu2O(s) [14].
Studies utilizing X-ray diraction and other surface analysis techniques support
the formation of a Cu2O(s) scale in the presence of chloride under various water
quality conditions [15,16]. The aqueous chemistry of cuprous ion may also be
dominated by the formation of several stabilizing complexes with the NH3(aq)
ligand. The cuprous ammine complexes can be formed directly or by a reduction
of cupric ammine complexes by a reaction such as [17]:
2
Cu s 4 2 Cu NH3 2
Cu NH3 4
1.2. Copper (II) chemistry
The most important Cu(II) species governing its solubility are the hydrolysis
species: Cu(OH)+, Cu(OH)02, Cu(OH)ÿ
3 etc. The stability constants of these species
1804
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
are quite unreliable and large dierences exist between the reported values. Baes
and Mesmer [12] have recommended both upper and lower limits for these
complexes, rather than precise values. Carbonate complexation dominates Cu(II)
solubility in most drinking waters at pH values greater than 7.3 [1,15,18]. The
hydroxy-carbonate complex, CuCO3(OH)2ÿ
2 , becomes important in enhancing
Cu(OH)2(s) solubility at pH values greater than 9.0. The prediction of increased
solubility of copper(II) by waters of increasing DIC is supported by the results of
many laboratory and ®eld studies [18ÿ21]. However, according to Schock et al.
[1], DIC regulates cuprosolvency in conjunction with pH and the dominant solid
phase under the corresponding exposure conditions. For example, malachite
formation would enable attainment of copper levels less than 1.3 mg/l (i.e., the
USEPA action level) after stagnation, below a pH of 6.5 for all DIC levels. The
quantitative prediction of copper levels in drinking waters relies heavily on the
solubility and physical properties of cupric oxide, hydroxide and basic carbonate
solids, which comprise most scales in drinking water supplies. Due to considerable
uncertainty in the solubility constants of all these minerals, the tendency for
prolonged existence of thermodynamically metastable phases like Cu(OH)2(s) and
the possible slow rate of formation of Cu2(OH)2CO3(s) (malachite), quanti®cation
becomes dicult. The dierences among several solubility constants for CuO(s)
and Cu(OH)2(s) have been discussed by de Zoubov et al. [11], who selected two
constants, which they felt agreed best with actual copper solubility data. Adeloju
and Hughes [10] suggest that the formation of Cu(OH)2(s) is kinetically favored
over the formation of the CuO(s) solid. Many researchers have observed
signi®cant Cu(OH)2(s) being produced by corrosion of copper at anodic locations
in electrochemical experiments [22,23]. These experiments, along with the copper
levels observed in many sampling programs and pipe rig systems, support the
argument that for relatively new plumbing systems, the use of Cu(OH)2(s) in the
solubility models is more realistic than the use of CuO(s).
Thus, although Cu2O(s) was the oxide of choice in the current study, additional
studies with some of the aforementioned Cu(II) solid phases can be carried out.
This allows a better understanding of the stability of the various dominant copper
oxide species under dierent water quality conditions.
The current study attempts to build on the earlier ®ndings and relate the
stability of the copper corrosion ®lms to special water quality parameters (i.e.,
pH, DIC and chloride concentrations etc.) and discusses the probable chemical
processes that could explain the observed EIS results.
The objective of the present study is to evaluate the stability of an arti®cially
synthesized, relatively homogeneous, cuprous oxide (Cu2O) ®lm on a fresh copper
coupon using synthetic and real ambient waters. Furthermore, it is our objective
to compare the performance of the Cu2O ®lm to naturally formed corrosion scale
as copper corrosion inhibitor and its capacity to protect the underlying copper
metal. EIS was used to study the stability of the ®lms under dierent water
quality conditions. The synthetic waters used in the second phase of the study
consisted of several combinations of typical distribution system chloride and DIC
concentrations at pH values of 7.0 and 8.5. Finally, the performance and stability
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1805
of the arti®cially synthesized cuprous oxide ®lm was compared to that of naturally
scaled copper coupons under typical distribution system water qualities.
2. Experimental
2.1. Electrochemical (DC) measurements
Corrosion currents were used to quantify the instantaneous corrosion rates. To
estimate the three key parameters: the corrosion currents, the corrosion potentials
and the Tafel slopes, potentiodynamic polarization scans were carried out using a
Gamry AC/DC corrosion testing machine with the CMS 100 software. A threeelectrode con®guration was adopted with a platinum auxiliary electrode and a Ag/
AgCl reference electrode. A summary of the parameter settings for the
potentiodynamic polarization scans is given in Table 1. For every set of
measurements to estimate the Tafel parameters by potentiodynamic polarization
scans, the average values obtained from three concurrent scans (not diering from
one another by more than 10%) are reported. As an additional quality control
measure for selected samples, three polarization resistance scans were conducted
as well to estimate Rp (polarization resistance). The Tafel slopes ba and bc were
calculated from the potentiodynamic scans using commonly used equations.
Substituting the values of Rp (from the polarization resistance scans) and ba and
bc into the expression Rp ba bc = 2:3 Icorr ba bc , yielded an independent
estimate of Icorr. If this estimate of Icorr was not within 20% of that obtained from
the potentiodynamic scans, all the scans in that batch were repeated until all
quality control requirements (i.e., within 20%) were satis®ed. The Rp values from
the Tafel scans were compared to those from the EIS measurements (discussed
later), as an additional quality control measure.
2.2. Synthesis of Cu2O coatings
A ¯ame aerosol generation and deposition technique was used for the synthesis
Table 1
Parameter settings for potentiodynamic scans
Parameter
Value
Voltage scan range
ÿ250 mV to +250 mV relative to
open circuit voltage (OCV)
1 mV/s
2s
0.78 cm2
15 s
O
600 s
Scan rate
Sampling time
Sample area
Auto-ranging time (delay)
Conditioning
Total scan time
1806
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
of the Cu2O coating. The apparatus consists of a honeycomb premixed ¯ame
burner, the detailed design parameters of which is described elsewhere [24,25]. It
has features of a uniform temperature pro®le over the burner cross-section and a
steady one-dimensional ¯ow ®eld. Fig. 1 is a simpli®ed schematic of the premixed
¯ame aerosol reactor system. Particle-free dry oxygen (RH = 0%) was premixed
with methane and subsequently ignited by the burner to produce the ¯ame. In
some of the initial trials, the air was bubbled through heated copper acetate
solution (temperature maintained at 508C) before being premixed with methane in
order to provide adequate copper precursor (in the form of air-stripped copper
acetate) for the formation of the cuprous oxide aerosol. However, this step was
found to be unnecessary since the copper substrate (coupon) itself could be used
as a precursor material. As a result, during the subsequent syntheses, this step was
eliminated. The operating conditions that had to be optimized during the
syntheses were: CH4 and O2 ¯ow rates, ¯ame temperature, actual ¯ame jet initial
diameter, the distance of the coupon from the ¯ame and the number and the
speed of passes of the copper substrate over the ¯ame. Identi®cation of the oxide
coating was conducted by X-ray diraction using a Siemens D500 Diractometer.
A con®rmation of the identity and elemental composition of the oxide ®lm was
obtained by conducting an SEM analysis of the ¯ame prepared Cu2O samples
using a Hitachi S-4000 SEM with EDAX capability. The standardization of the
cuprous oxide coating synthesis procedure resulted in excellent reproducibility of
the ®lm morphology and thickness.
2.3. Electrochemical Impedance Spectroscopy (EIS)
EIS is a technique to analyze the response of corroding electrodes to small
amplitude alternating potential signals of widely varying frequency. The equivalent
circuit for the oxide coated metal surface is shown in Fig. 2. Using the CMS 100
application software, reasonable initial values for the circuit components were
plugged in and a number of iterations performed in order to get a close ®t
Fig. 1. A simpli®ed schematic of the pre-mixed ¯ame aerosol reactor system.
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1807
between data points and the curves logjZj and y versus log frequency). If a good
®t was not obtained, then additional circuit components known as Warburg
impedances would be added. These components simulate the diusion-limited
processes. From the Nyquist plot data one can conclude that most of the
corrosion reactions were not under diusion control. Fig. 3 shows a typical
Nyquist plot. The stability of the coating is considered to be poor if Cdl increases
with time along with a corresponding decrease in Rpo.
EIS measurements were also carried out on fresh copper surfaces and naturally
scaled copper coupons, after exposure to test solutions for 0 and 72 h. The
naturally scaled coupons (after 10 weeks of exposure to synthetic and real waters)
were tested using Cincinnati tap water and ozonated/bio-®ltered water from Lake
Blu (IL) water treatment plant as electrolytes.
The impedance of the equivalent circuit, Z o, may be expressed in terms of
real Z 0 o and imaginary Z 00 o components as follows:
Z o Z 0 o Z 00 o
A summary of the parameter settings for the EIS scans is given in Table 2. The
impedance behavior of the copper electrode was expressed as a Bode plot logjZj
versus log o). The EIS spectrum data was then ®tted and an estimate of the
pertinent resistances and capacitances obtained.
Fig. 2. An electrical circuit analog of the EIS system.
1808
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
Fig. 3. A typical Nyquist plot for fresh copper surface (t = 0 h, pH = 8.5, DIC = 4.9 mg/l, chloride
= 25 mg/l), one time constant.
3. Results and discussion
The purpose of this study was to investigate the eectiveness of an arti®cially
synthesized Cu2O(s) ®lm in arresting the copper corrosion process under speci®c
water quality conditions. The rationale behind this is to eventually isolate all the
identi®ed oxide phases from the heterogeneous natural pipe scale and to quantify
the eectiveness of each of these oxide phases as copper corrosion inhibitors.
After the identi®cation of the best corrosion inhibiting copper oxide, this
information could then be used to control the water quality, the temperature and
Table 2
Parameter settings for EIS scans
Parameter
Value
AC voltage amplitude
10 mV for a fresh electrode
20 mV for a coated electrode
0 mV vs. open circuit potential
5
5000±0.005 Hz
5.0 cm2
8.3 g/cm3
31.5 g/eq
100 s
DC voltage
Points per decade
Frequency range
Sample area
Sample density
Equivalent weight
Initial delay
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1809
pipe hydrodynamics, in order to favor the deposition of a speci®c copper scale
that is the best copper corrosion inhibitor under a given set of conditions.
In the current study, the eects of chloride and DIC concentrations at pH
values 7 and 8.5 were investigated. The results from synthetic Cu2O were
compared to those from naturally scaled copper surfaces under controlled water
quality conditions. The performance of the naturally scaled and arti®cially coated
surfaces under the in¯uence of real waters (i.e., ozonated/bio-®ltered Lake
Michigan water and Cincinnati tap water) was also investigated and compared to
the results obtained from the synthetic waters. EIS measurements were carried out
initially and after 72 h of exposure, to study the stability of the ®lms and evaluate
the development of corrosion scale over the fresh copper surfaces.
For the synthesis of the Cu2O coating, several trials had to be carried out to
establish the optimum operating parameters of the ¯ame aerosol reactor. X-ray
diraction studies indicated that the ®nal coatings were not contaminated with
unwanted oxide phases such as Cu(II) oxides, which could easily form under
oxygen-rich ¯ame conditions. To con®rm the XRD ®ndings, SEM (magni®cation
of 25,000 and electron beam accelerated at 20 kV) and EDAX analyses were
conducted on the coated surfaces to ascertain the morphology and elemental
composition of the coatings (Figs. 4±6)). The SEM analyses involved several
sample preparation steps, including sputtering the sample with a 50 nm thick gold
Fig. 4. An SEM photograph (25,000 magni®cation) of the arti®cially synthesized cuprous oxide
coating.
1810
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
layer, the details of which are described elsewhere [26,27]. The copper to oxygen
ratios for the samples (atomic percentages) were in the vicinity of 8:1, indicating
that the only copper oxide present in the coating was Cu2O. EDAX analyses were
conducted at four or ®ve dierent locations of the sample, and the average of the
atomic ratio values was reported. Occasionally, some of the samples were found
to be contaminated with traces of zinc and silicon (<2%). However, this was not
believed to be signi®cant enough to aect our results. To estimate the thickness of
the coatings, EDAX analyses were conducted along the vertical cross-section of
the samples. By observing a change in the elemental composition along the
vertical cross-section and using the scale on the SEM micrograph, the thickness of
the coatings was estimated. For each sample, the coating thickness was measured
at three dierent locations and the average value was reported. Under the
operating conditions of the aerosol reactor, the Cu2O coating thickness in the
copper samples was found to vary between 0.70 and 0.92 mm (i.e., the
reproducibility of the coating thickness was quite good). Moreover, the thickness
Fig. 5. An EDAX spectrum of the cuprous oxide coating.
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1811
Fig. 6. An SEM photograph (11,000 magni®cation) of the vertical surface of the cuprous oxide
coated copper coupon. EDX analyses were conducted to determine the coating thickness.
of the naturally formed scales (i.e., upon exposure of the copper coupons to
simulated drinking water solutions: Scale A1 of 0.88 mm thickness and Scale B2 of
0.71 mm thickness) matched closely with those of the ¯ame synthesized Cu2O
coatings. Synthetic solutions containing dierent combinations of chloride and Ct
concentrations (Table 3) were used for the EIS experiments.
Typical Bode plots are shown in Fig. 7 (a single time constant) and Fig. 8 (two
time constants). The following parameters were determined from the Bode plot
using appropriate software for parameter ®tting (CMS 100) and techniques
discussed elsewhere [28]:
1.
2.
3.
4.
5.
1
coating capacitance, CC
electrical double layer capacitance, Cdl
polarization resistance, Rp
coating resistance, Rpo
resistance of the solution, Rs
Scale A: Scale developed after 10 weeks of exposure to a pH 7.0 solution with Ct = 250 mg/l (DIC
= 49 mg/l) and Clÿ = 25 mg/l.
2
Scale B: Scale developed after 10 weeks of exposure to ozonated/bio-®ltered Lake Michigan water
whose pH was adjusted to 7.
1812
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
Table 3
The composition of synthetic solutions used for EIS experiments
ÿ
Solution pH Ct = [COÿ2
3 ] + [HCO3 ] + [H2CO3]
no.
(mg/l)
Clÿ
Na+
Buer
(anion only) (cation only)
(mg/l)
(mg/l)
1
2
3
4
5
6
7
8
125
25
125
25
125
25
125
25
7.0
7.0
7.0
7.0
8.5
8.5
8.5
8.5
a
250
250
25
25
250
250
25
25
175.3
110.5
90.4
25.6
175.3
110.5
90.4
25.6
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
(M)
(M)
(M)
(M)
(M)
(M)
(M)
(M)
MOPSa
MOPS
MOPS
MOPS
boric acid
boric acid
boric acid
boric acid
MOPS refers to 4-morpholinepropanesulfonic acid (pKa = 7.2).
In the ®eld of corrosion and coating performance science and engineering, a
coating is considered to be performing satisfactorily if it has high resistance and
low capacitance, even after prolonged exposure to the corroding medium. If the
coating breaks down under exposure, it may result in the entrainment of water
within its structure. As a result, its capacitance goes up because water has a very
Fig. 7. A Bode plot for fresh copper surface (t = 0 h, pH = 8.5, DIC = 4.9 mg/l, chloride = 25 mg/
l), one time constant.
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1813
Fig. 8. A Bode plot for cuprous oxide coated surface (t = 0 h, pH = 8.5, DIC = 4.9 mg/l, chloride =
125 mg/l), two time constants.
high dielectric constant. In general, degradation of the coating may take place by
any one or a combination of the following mechanisms: (a) delamination of the
coating (i.e., the breakage of the adhesive forces (physical/chemical) between the
coating and the interface, resulting in the coating layer being stripped o from the
metal surface); (b) physical breakdown or dissolution of the coating at certain
locations, resulting in seepage of the aqueous electrolyte within the coating.
Eventually, parts of the underlying metal surface get exposed to the electrolyte
directly. The electrolyte then tries to form a layer between the coating and the
interface and breaks down the adhesive forces at the interface. Mechanism (a)
usually takes place under conditions of very high hydrodynamic shear. Mechanism
(b) appears more likely under laminar ¯ow or quiescent conditions (like the
conditions in the present study).
3.1. Eect of DIC and chloride concentrations at pH 7
Table 4 summarizes the parameters estimated from the EIS analyses for the
1814
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
Table 4
EIS parameters for fresh and Cu2O coated copper surfaces at pH 7, at t = 0 and 72 h of exposurea
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
Ct = 250 mg/l,
Clÿ = 25 mg/l
Ct> = 25 mg/l,
Clÿ = 25 mg/l
Ct = 250 mg/l,
Clÿ = 125 mg/l
Ct = 25 mg/l,
Clÿ = 125 mg/l
(288)b
363 (213)
13,437 (11,004)
(2580)b
37 (49)
33 (47)
53,430 (51,380)
22,167 (19,364)
(237)
209 (195)
6350 (9050)
(1955)b
173 (288)
166 (257)
3500 (4230)
2805 (1204)
(346)b
525 (282)
18,838 (7700)
(2190)b
53 (70)
49 (69)
45,500 (28,550)
15,270 (10,460)
(355)b
363 (282)
18,360 (6611)
(1762)b
421 (487)
407 (388)
1880 (10,300)
1010 (1241)
a
The numbers in parentheses refer to those at t = 72 h (i.e., at the end of the exposure period) while
those outside are for t = 0 h (i.e., initially). Bold refer to the Cu2O coated surface.
b
The measurement is not applicable.
fresh and Cu2O coated surfaces at pH 7. The EIS spectrum for the fresh copper
surfaces shifts from a one time constant case (initially) to a two time constant case
(after 72 h) in all the cases, due to the development of a corrosion scale. For the
fresh copper surface at Ct = 250 mg/l, increasing the Clÿ dosage to 125 mg/l
results in a 30% increase in corrosion rate after 72 h of exposure. This can be
inferred from the respective Rp values (since Rp is inversely proportional to the
corrosion rate). The higher chloride dosage also results in higher values of coating
and double layer capacitances (i.e., Cc and Cdl) after 72 h of exposure. At a Clÿ
dosage of 25 mg/l and Ct = 250 mg/l, the Cu2O coating is extremely stable even
after 72 h of exposure and results in a decrease of the corrosion rates by nearly a
factor of 5. Cc does not increase appreciably upon exposure. At a Clÿ dosage of
125 mg/l, there is a signi®cant reduction in Rp for the ¯ame synthesized Cu2O
after 72 h of exposure. However, the increase in Cc is very low indicating that the
increase in corrosion rates was likely due to local defects in the coating and not
due to the seepage of the electrolyte through the microstructure of the coating.
For the Ct = 250 mg/l and Clÿ = 125 mg/l case, the ¯ame synthesized Cu2O
coating resulted in the reduction of corrosion rates by a factor of 3.6 relative to
the fresh copper surface. However, at a Ct concentration of 25 mg/l, the Cu2O
coating was found to be extremely unstable in the presence of Clÿ ions at
concentrations as low as 25 mg/l. In fact, the corrosion rates were observed to be
50 to 60% higher than those obtained with the fresh copper surfaces. Increasing
the Clÿ concentration to 125 mg/l resulted in an increase in the initial corrosion
rate by a factor of nearly two for the ¯ame synthesized Cu2O. Finally, for the
Cu2O coating, the increase in the coating capacitance after 72 h of exposure was
quite signi®cant for the Ct = 25 mg/l cases.
3.2. Eect of DIC and chloride concentrations at pH 8.5
Table 5 summarizes the parameters estimated from the EIS analyses for the
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1815
Table 5
EIS parameters for fresh and Cu2O coated copper surfaces at pH 8.5, at t = 0 and 72 h of exposurea
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
Ct = 250 mg/l,
Clÿ = 25 mg/l
Ct = 25 mg/l,
Clÿ = 25 mg/l
Ct = 250 mg/l,
Clÿ = 125 mg/l
Ct = 25 mg/l,
Clÿ = 125 mg/l
(107)b
360 (79)
11,050 (41,600)
(5692)b
247 (259)
273 (281)
6249 (4864)
2012 (1448)
(76)b
263 (63)
6500 (55,379)
(10036)b
195 (187)
182 (178)
6652 (6549)
2070 (1992)
(115)b
502 (94)
16,200 (39,560)
(5722)b
234 (363)
251 (341)
5380 (6049)
2585 (1838)
(91)b
410 (72)
13,300 (40,270)
(8094)b
289 (241)
252 (213)
7330 (12,427)
2112 (2802)
a
The numbers in parentheses refer to those at t = 72 h (i.e., at the end of the exposure period) while
those outside are for t = 0 h (i.e., initially). Bold refer to the Cu2O coated surface.
b
The measurement is not applicable.
fresh copper coupon and the ¯ame synthesized Cu2O at pH 8.5. For the fresh
copper surface with a Clÿ concentration of 25 mg/l, the corrosion rate after 72 h
of exposure decreased by 23% as the Ct decreased from 250 to 25 mg/l. The lower
Ct concentration also results in a more compact natural scale as is indicated by a
signi®cant increase in Rpo from 5692 to 10,036 O, and a reduction in Cc from 107
to 76 mF. Increasing the Clÿconcentration to 125 mg/l results in a minor increase
in corrosion rates for the Ct = 250 mg/l case and a 27% increase for the Ct = 25
mg/l case, after 72 h of exposure. For the fresh copper surface at Ct = 250 mg/l,
the resistance of the scale developed after 72 h was found to be independent of the
Clÿ dosage. At pH 8.5, the ¯ame synthesized Cu2O coating was observed to be
extremely unstable, irrespective of the Clÿ and DIC concentrations. In fact, the
corrosion rates after 72 h were slightly higher or insigni®cantly lower than those
measured initially, except for the case of Ct = 25 mg/l and Clÿ = 125 mg/l. The
initial corrosion rates for the Cu2O case were comparable to those for the fresh
copper surface for one case (Ct = 25 mg/l, Clÿ = 25 mg/l) and signi®cantly
higher than the fresh copper surface for the other three cases. For the Cu2O
coated surfaces, the change in corrosion rates after 72 h was not very signi®cant,
except for the case of Ct = 25 mg/l and Clÿ = 125 mg/l, where the corrosion rate
decreased by a factor of 1.7. However, at pH 8.5 for the fresh copper surface, the
reduction in corrosion rates after 72 h was very signi®cant for all four cases. The
corrosion rates decreased by a factor of nearly 8.5 for the Ct= 25 mg/l and Clÿ
= 25 mg/l condition. A possible explanation for the instability of the Cu2O
coating at pH 8.5 and its ability to aggravate the corrosion process is as follows.
At pH 8.5, the DIC speciation shifts signi®cantly to CO2ÿ
3 , which diuses through
the defects in the coating and reacts with the free ionic Cu2+ species at the
interface [18,29]:
DG0f ÿ675:4 kJ=mol 30
Cu 2 CO32ÿ H2 O $ CuCO3 OH ÿ H
1816
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
2ÿ
Cu 2 CO32ÿ 2H2 O $ CuCO3 OH 2 2H
DG0f ÿ861:9 kJ=mol 31
These reactions result in the release of protons, which decreases the pH
signi®cantly over small pockets at the coating/metal interface. This results in the
solubilization of the coating as follows: Cu2O(s) + 2H+ t 2Cu2+ + H2O. As a
result, not only does the coating gets disturbed at a number of locations, but
several Cu2+ ions are generated which can subsequently react to form
and further reduce the pH. Hence, the Cu(II)
CuCO3OHÿ and CuCO3(OH)2ÿ
2
complex formation and the Cu2O(s) dissolution processes are synergistic in
nature. For the ¯ame synthesized Cu2O coatings, an examination of the Cc values
for the pH 8.5 cases reveals that there is very little change after 72 h, except for
the case of Ct = 250 mg/l and Clÿ = 125 mg/l (Table 5). This indicates that the
attack by CO23, and the resulting increase in acidity, which lead to the dissolution
process must be quite localized. Even after 72 h of exposure, the electrolyte does
not get entrained within the coating microstructure to result in a general failure.
However, the general failure of the coating may occur after a longer exposure.
The defects and pores in the microstructure of the Cu2O ®lm are extremely
detrimental for its proper performance as a corrosion inhibitor at pH 8.5. At low
DIC and high chloride concentrations, the hydrolysis of CuCl to form Cu2O(s)
becomes dominant. As a result, it is hypothesized that the rate of deposition of
Cu2O exceeds the rate of its dissolution by the processes described above.
Experimental results show an increase in Rp from 7330 to 12,427 O for the case of
Ct = 25 mg/l and Clÿ = 125 mg/l at pH 8.5 (i.e., corrosion rates decrease by a
factor of 1.7 after 72 h of exposure). This clearly supports the hypothesis
proposed earlier.
3.3. Comparison of the performance of naturally scaled coupons and synthetic Cu2O
®lms
The stability of dierent ®lms was evaluated upon the exposure to (a) ozonated/
bio-®ltered Lake Michigan water (pH 7.34) and (b) Cincinnati tap water (pH
7.29). The chloride and DIC concentrations of the two water samples are shown
in Table 6. The conditions for the formation of the two natural scales were
speci®ed earlier (refer to Scales A and B). A summary of the EIS parameters is
given in Table 7
3.3.1. Solution 1: ozonated/bio-®ltered Lake Michigan water
Scale A resulted in a very large reduction of the initial corrosion rate relative to
the fresh copper surface, the ¯ame synthesized Cu2O coated surface and scale B.
However, the performance of scale A deteriorated rapidly during the 72 h
exposure to solution 1 (Rp decreased from 56,000 to 19,300 O and Cc increased
from 37 to 113 mF). The changes in the values of Cc and Cdl for the Cu2O and
scale B cases after 72 h were very minor compared to those for the case of scale
A. This indicates that scale A deteriorates much more rapidly than the other two.
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1817
Table 6
DIC and chloride concentrations of water samples
Water sample
pH
DIC (mg/l)
Chloride (mg/l)
Lake Michigan, ozonated/bio-®ltered water
Cincinnati tap water
7.34
7.29
24.0
19.4
40.8
10.6
At the end of the 72 h exposure period, the corrosion rate for the Cu2O coating
was comparable to that for scale A.
3.3.2. Solution 2: Cincinnati tap water
Initially, scale A was found to be an excellent corrosion inhibitor. However,
similar to the earlier case, scale A was extremely unstable during the exposure
period and resulted in a sharp increase in the corrosion rate (Rp decreased from
57,174 to 25,800 O and Cc increased from 37 to 92 mF). However, scale B was
found to be the most stable and resulted in the lowest corrosion rates after 72 h of
exposure. The corrosion rate for scale B (after 72 h of exposure) was only 50% of
the rates for scale A and the synthetic Cu2O coating.
The performance of the Cu2O coating upon exposure to Cincinnati tap water
was much better than that upon exposure to ozonated/bio-®ltered lake Michigan
water. From the synthetic water experiments it was concluded that, at a DIC to
chloride ratio of 1.96 at pH 7 (i.e., Ct = 250 mg/l, Clÿ = 25 mg/l), Cu2O coating
Table 7
EIS parameters for fresh, Cu2O coated and naturally scaled Cu surfaces using (a) solution 1: ozonated/
bio-®ltered Lake Michigan water, (b) solution 2: Cincinnati tap water, at t = 0 and 72 h of exposurea
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
Fresh copper
Cu2O coated
Scale A
Scale B
Sol. 1
Sol. 2
Sol. 1
Sol. 2
Sol. 1
Sol. 2
Sol. 1
Sol. 2
±b
(68)
60
(58)
5070
(7345)
±b
(10,400)
±b
(42)
79
(36)
8225
(6900)
±b
(20,106)
93
(119)
125
(112)
16,480
(17,904)
4342
(3636)
118
(112)
95
(90)
24,600
(30,262)
4970
(5568)
37
(113)
40
(85)
56,000
(19,300)
19,894
(6370)
37
(92)
40
(85)
57,174
(25,800)
19,000
(7416)
46
(60)
58
(55)
11,500
(12,640)
10,200
(3800)
50
(40)
61
(35)
32,100
(60,000)
12,300
(17,706)
a
The numbers in parentheses refer to those at t = 72 h (i.e., at the end of the exposure period) while
those outside are for t = 0 h (i.e., initially).
b
The measurement is not applicable.
1818
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
was very stable and did not result in an increase in the corrosion rates
signi®cantly (Table 4). Using Cincinnati tap water with a comparable DIC to
chloride ratio and a slightly higher pH (i.e., 1.94, pH 7.29), the Cu2O coating was
once again found to be very stable during the 72 h exposure period (Rp increased
by nearly 25% and there was very little change in Cc, Table 7). However, the Rp
values for the Cincinnati tap water were signi®cantly lower. This indicates that
besides the three key parameters: pH, DIC and chloride concentrations, other
organic and inorganic species and water quality parameters can signi®cantly
in¯uence the instantaneous corrosion rates. However, the deterioration of the
Cu2O coating may be a strong function of the DIC to Clÿ ratio at a given pH,
and is almost independent of the concentration of the other inorganic and organic
species.
To con®rm this hypothesis, additional control experiments were conducted to
assess the variation of coating capacitance, Cc, as a function of the DIC to Clÿ
ratio. The pH was maintained at approximately 7.0 and the chloride concentration
was kept ®xed at 25 mg/l. The DIC adjustment was carried out by using freshly
prepared 2 M stock solution of NaHCO3. To assess whether the other
background water quality parameters aected the results, control experiments
were conducted using Cincinnati tap water whose DIC to Clÿ ratio had been
adjusted to certain speci®c values. Table 8 provides a summary of the typical
values of pertinent background water quality parameters for Cincinnati tap water.
All wet chemistry analyses for assessing water quality parameters were conducted
using protocols described in Standard Methods for the Examination of Water and
Waste water [32].
Fig. 9 shows that irrespective of the background water quality conditions, Cc
was found to be a strong function of the DIC to Clÿ ratio at pH 7. At pH 7, Cc
values were found to be the lowest in the vicinity of a DIC to Clÿ ratio of 2. The
plots for the synthetic water and the Cincinnati tap water overlap greatly,
Table 8
Pertinent water quality parameters for Cincinnati tap water at 258C (avg.2standard deviation)
Parameter
Value
pH
Alkalinity (mg
CaCO3/l)
DO (mg/l)
Conductivity (mS)
ORP (mV)
DIC (mg/l)
Chloride (mg/l)
Sulfate (mg/l)
TOC (mg/l)
DOC (mg/l)
BDOC (mg/l)
UV254 (cmÿ1)
7.2920.03
46.525.8
7.020.1
21523
30225
19.421.2
10.620.6
21.421.7
1.120.02
0.920.07
0.220.1
0.2720.02
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1819
Fig. 9. Cu2O coating capacitance after 72 h as a function of DIC/chloride ratio at pH 7 (chloride = 25
mg/l).
although at any given DIC to Clÿ ratio, the Rp values with the Cincinnati tap
water were signi®cantly lower than those with the synthetic water (not shown in
Fig. 9 (i.e., the corrosion rates were greatly dissimilar). Therefore, although earlier
studies [33,34] have concluded that chloride (up to rather high concentrations)
decreases the corrosion rate of copper, our results in this study indicate that the
interplay between chloride and DIC is very interesting and complicated.
Furthermore, both chloride and DIC should be considered in future studies,
which attempt to predict and fully understand copper corrosion in distribution
systems.
3.4. Eect of buer concentration on corrosion behavior
The choice of the buer solutions was primarily dictated by their inertness and
has practically no in¯uence on the corrosion chemistry. Our objective was to limit
1820
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
Table 9
Comparison of EIS results at two dierent buer concentrations, for cuprous oxide coated samplea
EIS
parameters
Cc (mF)
Cdl (mF)
Rp (O)
Rpo (O)
pH 7.0, Ct = 250 mg/l and Clÿ = 25 mg/l
pH 8.5, Ct = 250 mg/l and Clÿ= 25 mg/l
0.01 (M) MOPS
37 (49)
33 (47)
53,430 (51,380)
22,167 (19,364)
0.01 (M) boric acid
247 (259)
273 (281)
6249 (4864)
2012 (1448)
0.1 (M) MOPS
47 (59)
48 (61)
43,900 (42,101)
18,210 (15,480)
0.1 (M) boric acid
290 (317)
320 (345)
5600 (4344)
1807 (1300)
a
The numbers in parentheses refer to those at t = 72 h (i.e., at the end of the exposure period) while
those outside are for t = 0 h (i.e., initially).
the buer concentration to the bare minimum without, at the same time,
compromising our ability to maintain a fairly constant pH over the exposure
period. We agree that buer solutions increase the conductivity of the test
solution and lead to potentially higher observed corrosion rates (due to the
migration contribution). We were aware of this while performing the experiments.
In order to get an idea about the eects of variable buer concentrations (and
hence solution conductivity), a few control experiments have been performed
(both at pH 7.0 and 8.5) at a buer concentration of 0.1 M (instead of 0.01 M,
which was used in most of the experiments). Table 9 summarizes the EIS
experimental data for the two dierent buer concentrations. The 10-fold increase
in the buer concentration led to a nearly 18% increase in the corrosion rate at
pH 7 and a 10% increase at pH 8.5. However, one must remember that the main
focus of our discussion in this paper is not to study the absolute values of
corrosion rates, but to evaluate the relative performance of the synthesized and
natural scales under dierent pH conditions and at dierent chloride and DIC
concentrations. It is certain that the corrosion rate will depend on the solution
conductivity. However, by keeping the buer concentration constant in all our
experiments, we can almost completely eliminate the dierences observed in
corrosion rates due to conductivity/migration, when comparing the eects of
chloride and DIC concentrations at the two pH values studied. This will provide a
common baseline to compare the eects of chloride and DIC on scale stability at
the two pH values. Hence, we feel that although conductivity/migration
contribution is an important issue, it is even more important to try to establish a
common baseline from which to compare the eects of chloride and DIC.
4. Conclusions
The performance and the stability of the Cu2O scale are strongly dependent
upon the pH, the DIC and the chloride concentrations. At pH 7, the coating was
found to be very stable at a DIC to chloride ratio in the vicinity of 2. However, at
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
1821
pH 8.5, due to the shift in the DIC speciation toward the CO2ÿ
and the
3
consequent development of small pockets of high H+ concentration, the
performance of the Cu2O coating deteriorated very fast. Under the in¯uence of
real waters, the performance of the Cu2O coating was found to be comparable to
that of the two naturally developed scales only under certain conditions. Although
the instantaneous corrosion rates are strongly dependent on several water quality
parameters and inorganic and organic species, the deterioration of the Cu2O ®lm
upon exposure at a given pH seems to depend most strongly on the DIC to
chloride ratio.
The present study by no means answers all the questions regarding ®lm stability
in distribution systems, though it provides an excellent foundation for additional
studies in this critical area of research. The performance of other Cu(I) and Cu(II)
mineral phases that have been identi®ed in the water distribution corrosion scales
needs to be quanti®ed under the in¯uence of a variety of real and synthetic
waters. The eect of temperature, hydrodynamic shear and disinfectant residuals
on the solubility, stability and morphology of these scales need to be ascertained
in order to optimize the performance of the ®lms, when they are exposed to
dierent water quality conditions.
Acknowledgements
The authors wish to thank Prof. Pratim Biswas of University of Cincinnati for
his generosity in using the aerosol reactor for cuprous oxide synthesis and Mr.
Michael Schock of the USEPA Drinking Water Laboratory (Cincinnati, Ohio) for
his useful comments and input to the scope of the research described herein.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
M. Schock, D. Lytle, J. Clement, EPA Report: EPA/600/R-95/085 (1995).
Y. Feng, W.-K. Teo, K.-S. Siow, K.-L. Tan, A.-K. Hsieh, Corrosion Sci. 38 (1996) 369.
Y. Feng, W.-K. Teo, K.-S. Siow, A.-K. Hsieh, Corrosion Sci. 38 (1996) 387.
E. Mattson, Brass Corros. 15 (1980) 6.
D.C. Atlas, J. Coombs, O.T. Zajicek, Water Res. 16 (1982) 693.
H. Cruse, O. von Franque, Internal Corrosion of Water Distribution Systems, AWWA Research
Foundation/DVGW Forschungsstelle, Denver, Colorado, 1985.
S. Reiber, JAWWA 81 (1989) 114.
M. Edwards, M. Schock, T. Meyer, JAWWA 88 (1996) 81.
T.E. Meyer, The eect of inorganic anions on copper corrosion. A Ph.D. thesis at University of
Colorado at Boulder, 1994.
S.B. Adeloju, H.C. Hughes, Corrosion Sci. 26 (1986) 851.
N. de Zoubov, C. Vanleugenhaghe, M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous
Solutions, 2nd ed., National Association of Corrosion Engineers, Houston, TX, 1974 (Section
14.1).
C.F. Jr. Baes, R.E. Mesmer, The Hydrolysis of Cations, Wiley±Interscience, New York, 1976.
F.J. Millero, M. Izaguirre, V.K. Sharma, Mar. Chem. 22 (1987) 79.
J.N. Alhajji, M.R. Reda, British Corrosion J. 31 (1996) 125.
1822
A. Palit, S.O. Pehkonen / Corrosion Science 42 (2000) 1801±1822
[15] A. Palit, S.O. Pehkonen, Copper corrosion: the eect of speci®c water quality parameters,
Student Paper Competition of the 91st Annual Meeting of the Air & Waste Management
Association, 1998.
[16] Y. Feng, K.S. Siow, W.K. Teo, K.L. Tan, A.K. Hsieh, Corrosion 53 (1997) 389.
[17] F.A. Cotton, G. Wilkinson, Advanced Inorganic Chemistry, 5th ed., Wiley, New York, 1988.
[18] M. Drogowska, L. Brossard, H. Menard, J. App. Electrochem. 24 (1994) 344.
[19] C.H.J. Elzenga, H.J. Boorsma, Corrosion of copper pipes in various types of drinking water,
International Committee on Corrosion and Protection of Underground Pipelines, 1981.
[20] S. Yamauchi, Dissolution of copper from copper water tubes and new copper alloys showing
anti-cuprosolvency, in: Proc. International Symposium on Corrosion of Copper and Copper
Alloys in Buildings, Tokyo, Japan, 1982.
[21] M.M., Benjamin Chemistry of Corrosion Inhibitors in Potable Water. AWWA Research
Foundation, Denver, CO, 1990.
[22] D.W. Shoesmith, et al., Electrochem. Acta 22 (1977) 403.
[23] Y. Suzuki, H.J. Akahoshi, Electrochem. Soc. 57 (1989) 983.
[24] H. Chang, P. Biswas, J. Colloid Interface Sci. 153 (1992) 157.
[25] G. Yang, H. Zhuang, P. Biswas, Nanostructured Materials 7 (1996) 675.
[26] C.O. Wells, Scanning Electron Microscopy, McGraw-Hill, New York, 1974.
[27] L.B. Gabriel, SEM: A User's Manual for Material Science, American Society of Metals
publication, Metal Park, OH, 1985.
[28] S.W. Tait, An Introduction to Electrochemical Corrosion Testing for Practicing Engineers and
Scientists, PairODocs Publications, Racine, Wisconsin, 1994.
[29] M. Drogowska, L. Brossard, H. Menard, J. Electrochem. Soc. 139 (1992) 39.
[30] L. Millette, D.S. Mavinic, Canadian Jour. Chem. 15 (1988) 79.
[31] P.T. Gilbert, Water Treatment and Examination 15 (1966) 165.
[32] A.D. Eaton, L.S. Clesceri, A.E. Greenberg, Standard Methods for the Examination of Water and
Wastewater,, 20th ed., United Book Press, Baltimore, MD, 1996.
[33] A. Elfstrom Broo, B. Berghult, T. Hedberg, Corrosion Science 39 (1997) 1119.
[34] M. Edwards, T. Meyer, J. Rehring, JAWWA 86 (1994) 73.