(EngineeringEBookspdf) Corrosion of Copper PDF

The Corrosion of
Copper and I t s Alloys
A Practical Guide for Engineers
Roger Francis
I N T E R N A T I 0 N A L
THE CORROSION SOCIETY
1440 South Creek Drive

Houston, Texas 77084
I N T E R N A T I O N A L
THE CORROSION SOCIETY
@ 20 10 by NACE International
All rights reserved
Printed in the United States of America
Library of Congress Cataloging in Publication Data
Francis, Roger, Engineer.

The corrosion of copper and its alloys : a practical guide for engineers I
Roger Francis.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-57590-225-8 (hardback)
1. Copper-Corrosion. 2. Copper alloys Corrosion. I. Title.
TA48O.C7F73 2010
620.1'82234~22 2010005806
ISBN: 978-1-57590-225-8
Neither NACE International, its officers, directors, or members thereof

accept any responsibility for the use of the methods and materials discussed
herein. The information is advisory only and the use of the materials and
methods is solely at the risk of the user.
This book, or any parts thereof, may not be reproduced in any form without permission
of the copyright owners.
NACE International
1440 South Creek Drive
Houston, Texas 77084
http:llwww.nace.org
Preface
When I left university in 1973, I went to work for BNF Metals Research
Association in Oxfordshire, as a junior research scientist in the corrosion
department. (It had just moved to Oxfordshire from central London.) A year
later, the organization changed its name to BNF Metals Technology Centre.
The organization was founded in the 1920s and was set up to support the
nonferrous metals industry. It carried out research and development work
on production, processing, and corrosion of all nonferrous metals. By far the
bulk of the work was carried out on copper and its alloys. The organization
was a membership-based one and all members received copies of research
reports relevant to their industry. Some, but not all, of the results of its
research were published.
I left the organization in 1991 and less than two years later, the
organization went bankrupt. Unfortunately, most of the library, including
its collection of research reports going back to the first one in 1924, was
consigned to a skip. When I left BNF, I had copies of quite a few reports to
which I had contributed. I thought it a shame that all this knowledge should
be lost, especially as there was (and is) no obvious successor organization
with the practical expertise in copper and its alloys. Hence, I conceived the
idea of an engineering review of the corrosion of copper and its alloys to
present an up-to-the-minute snapshot, including as much of this unpublished
data as possible.
During my quest for information, photographs, and case histories, I
contacted friends and ex-BNF colleagues, many of whom had data that I
thought was lost and who also gave me other unpublished data. I would
like to take this opportunity to thank them all. In particular, I would like to
thank Hector Campbell, David Davies, Bob Kain, Carol Powell, and Phillip
Munn. I would also like to thank Angela Vesey of CDA, who gave me such
xi
xii PREFACE
good access to their records, including old BNF reports. A lot of data also
came from old reports of the International Copper Research Association,
now the International Copper Association. Finally, the contribution of the
staff at TWI to the chapter on joining is gratefully acknowledged. Any
differences between the information as you gave it to me and as it appears
in print are my fault. I hope you feel the end result meets your expectations.
Finally, I must thank my brother Ian, who did the original drawings.
Contents
Preface xi
1 Introduction 1
Reference 4
2 Copper and Its Alloys 5

2.1 Coppers 5
2.2 Brasses 7
2.3 Bronzes 8
2.3.1 Tin Bronzes and Gunmetals 9
2.3.2 Silicon Bronze 9
2.3.3 Aluminum Bronze 10
2.4 Copper-Nickel and Nickel-Copper Alloys 11
2.5 Nickel Silvers 13
References 13
3 Film Formation and Properties 15

3.1 Copper and the Brasses 15
3.2 Aluminum Brass 17
3.3 Copper-Nickel Alloys 18
3.4 Aluminum Bronze 22
3.5 Tin Bronzes 22
3.6 Ferrous Sulfate 23
3.7 Chlorine 25
3.8 Sulfide 27
References 29
V
vi CONTENTS
4 General Corrosion 33
4.1 Natural Waters 33
4.1.1 Seawater and Fresh Water 33
4.1.2 Cuprosolvency 40
4.1.3 Rosette Corrosion 43
4.1.4 Plumbosolvency 43
4.1.5 Disinfection 44
4.2 Acids 45
4.2.1 Sulfuric Acid 45
4.2.2 Hydrochloric Acid 46
4.2.3 Nitric Acid 47
4.2.4 Hydrofluoric Acid 48
4.2.5 Phosphoric Acid 49
4.2.6 Organic Acids 49
4.3 Alkalis 50
4.4 Other Chemicals 52
4.5 Methods of Prevention 53
References 54
5 Pitting Corrosion 57
5.1 Type I Pitting 57
5.1.1 Copper 57
5.1.2 Water Composition 62
5.1.3 Mechanism 67
5.1.4 Copper Alloys 69
5.2 Type I1 Pitting 70
5.3 Type I11 Pitting 74
5.4 Aluminum Brass 78
5.5 Alloy 400 78
5.6 Formicary Corrosion 80
5.7 Rosette Corrosion 82
5.8.1 Type I Pitting 83
5.8.2 Type I1 Pitting 89
5.8.3 Type I11 Pitting 90
5.8.4 Alloy 400 90
5.8.5 Formicary Corrosion 90
5.8.6 Rosette Corrosion 91
References 91
Contents vii
6 Crevice Corrosion 95
6.1 Mechanism 95
6.2 Fresh Water 97
6.3 Seawater 97
6.4 Avoidance of Crevice Corrosion 108
References 108
7 Erosion Corrosion and Erosion 111

7.1 Erosion Corrosion 112
7.1.1 Fresh Water 112
7.1.2 Seawater 116
7.1.3 IrodFerrous Sulfate 134
7.1.4 Effect of Chlorine 139
7.1.5 Sulfides 143
7.2 Erosion 149
7.3 Prevention of Erosion Corrosion and Erosion 156
7.3.1 Erosion Corrosion 157
7.3.2 Erosion 160
References 161
8 Cavitation 165
8.1 Cavitation Test Methods 167
8.2 Performance of Copper Alloys 168
8.3 Avoidance of Cavitation 171
References 171
9 Dealloying 173
9.1 Dezincification 173
9.2 Dealuminification 179
9.3 Dealloying of Other Copper Alloys 184
9.4 Hot Spot Corrosion 184
9.4.1 Classical Hot Spot Corrosion 185
9.4.2 Sulfide-Induced Hot Spot Corrosion 185
9.4.3 Ammonia-Induced Hot Spot Corrosion 186
9.5 Avoidance of Dealloying 188
References 191
viii CONTENTS
10 Stress Corrosion Cracking 195

10.1 Ammonia 196
10.2 Mercury 206
10.3 Nitrites 209
10.4 Sulfur Dioxide 21 1
10.5 Other Environments 213
10.6 High-Strength Copper-Nickel Alloys 216
10.7 Test Methods and Testing 218
10.8 Prevention of Stress Corrosion Cracking 22 1
References 222
11 Corrosion Fatigue 225

11.1 Brasses 226
11.2 Copper-Nickels 227
11.3 Nickel Aluminum Bronze 233
11.4 Prevention of Corrosion Fatigue 234
References 235
12 Hydrogen Embrittlement 237

12.1 Copper and Its Alloys 237
12.2 Nickel-Copper Alloys 238
12.3 Avoidance of Hydrogen Embrittlement 24 1
References 24 1
13 Fouling and Microbially Influenced Corrosion 243

13.1 Macrofouling 243
13.2 Microfouling and Microbially Influenced Corrosion 25 1
13.3 Bacterial Toxicity 257
References 263
14 Galvanic Corrosion 267

14.1 Conditions Necessary for Galvanic Corrosion 268
14.2 Factors Affecting Galvanic Corrosion 269
14.2.1 Electrode Potential 269
14.2.2 Potential Variability 27 1
14.2.3 Electrode Efficiency 272
14.2.4 Electrolyte 275
Contents ix
14.2.5 Area Ratio 276

14.2.6 Aeration and Flow Rate 279
14.2.7 Metallurgical Condition and Composition 280
14.2.8 Bimetallic Corrosion without Physical
Contact 282
14.2.9 Stifling Effects 283
14.3 Alloy Groups 283
14.4 Galvanic Corrosion in Seawater 285
14.5 Galvanic Corrosion in Fresh Water 287
14.6 Galvanic Corrosion in the Atmosphere 29 1
14.7.1 Materials 295
14.7.2 Insulation and Separation 295
14.7.3 Paints and Coatings 297
14.7.4 Cathodic Protection 298
14.7.5 Inhibitors 299
14.7.6 Design 299
References 300
15 Underground Corrosion 303

15.1 Long-Term Exposure Tests 304
15.2 Copper Alloys 311
15.3 Galvanic Corrosion 312
15.4 Methods of Prevention 3 14
References 3 15
16 Atmospheric Corrosion 3 17
16.1 Factors Affecting Atmospheric Corrosion 317
16.2 Atmospheric Aggressivity 321
16.3 Corrosion Rates Outdoors 323
16.4 Galvanic Corrosion 326
16.5 Corrosion Indoors 327
References 333
17 Corrosion in Waters 335

17.1 Fresh Water 335
17.2 Seawater 337
17.3 Swimming Pools 342
X CONTENTS
17.4 Desalination 343

17.4.1 Multistage Flash 343
17.4.2 Multiple Effect Distillation 346
17.5 Oils and Oily Waters 348
References 350
18 Joining and the Corrosion of Joints 353

18.1 Soldering 353
18.2 Brazing 355
18.3 Welding 355
References 358
Appendix A: Nominal Compositions of Some Commonly Used

Copper Alloys 361
Appendix B: Laboratory Tests to Evaluate the Erosion Corrosion
Resistance of Copper Alloys 365
B. 1 Test Methods 365
B.2 Recirculating Versus Once-Through Testing 368
Index 37 1
Color plate section follows p. 388.
CHAPTER 1
Introduction
Mankind first began using copper domestically some 6,000 to 10,000 years
ago. Copper trinkets that date to the 9th millennium BC have been found
in Iran. Copper is soft; however, when smelted with tin, it forms bronze, a
harder, stronger metal. The period of widespread use of bronze is termed
the Bronze Age, which began in northern Europe about 1500 BC. The fact
that we are discovering copper and copper alloy implements that are thou-
sands of years old, and in good condition, testifies to the metal’s excellent
corrosion resistance.
Bronze enabled the manufacture of weapons, tools, vessels, vases, and
so on. The Romans were the first to use bronze for building ties in construc-
tion. Brass is an alloy of copper and zinc, which is less strong than bronze
but widely used by the ancient Greeks and Romans for personal ornamen-
tation and decorative metalwork. Many brass artifacts from this era are in
good condition, again showing the metal’s excellent resistance to corrosion.
During the Middle Ages, bronze was used for bells and cannon. The
decorative use of copper on buildings also came to be appreciated. Fig-
ure 1.1 shows the copper clad roofs on the towers of the medieval cathedral
in Bamberg, Germany.
The industrial revolution brought about big changes, with increased
production, lower levels of impurities, and a wider range of alloys. At this
time, the high electrical conductivity of copper and its use for both wires and
lightning conductors were demonstrated. This period also saw the invention
of the brass stamping press, which made brass forgings cheap and plentiful.
Copper sheathing was also being used on wooden-hulled ships, particularly
by the Royal Navy, to prevent attack of the timbers by marine borers. The
cladding also reduced marine fouling, enabling higher water speeds.
1
2 THE CORROSION OF COPPER AND ITS ALLOYS
FIGURE 1.1 Copper cladding on the towers of Barnberg Cathedral, Germany. (See color
section.)
The latter half of the 19th century heralded the commercial production
of electricity using copper wire for generators and transmission lines. A
century later, the electronics era made extensive use of copper for printed
circuit boards. Figure 1.2 shows copper wire being used in a modern, high-
efficiency electric motor.
Copper has also been shown to be essential for human health, and the
World Health Organization recommends a daily intake of 1-3 mg of copper
to maintain health.
Copper is ever present in the modern world, for instance, as an elec-
trical conductor in domestic appliances and electronics and as a vehicle
for carrying our drinking water. Copper and its alloys are widely used for
piping, pumps, and valves handling fresh water, seawater, and brines, and
they are also used in heating systems. Copper alloys have also found nu-
merous applications in industry, where corrosion resistance is important for
handling more aggressive fluids.
Their high thermal conductivity and corrosion resistance means that
copper alloys are widely used for heat exchanger tubes in power stations,
Introduction 3
FIGURE 1.2 Copper wires being used in the construction of a modern, high-efficiency
electric motor. (Photograph courtesy of CDA.) (See color section.)
desalination plants, oil and gas production, and so on. Corrosion resistance
and the pleasing appearance of copper and its alloys mean that they are
popular for fittings inside and outside buildings as well as for decoration.
The foregoing demonstrates the combination of properties that makes
copper and its alloys so widely used today. The aim of this book is to draw
together the large body of information on the corrosion of copper and its
alloys into a single volume in a form that is useful to engineers who are
trying to select the most suitable alloy for a specific application.
The next chapter classifies the commonly used copper alloys by group
and discusses the merits and properties of each group. The corrosion resis-
tance of copper and its alloys is dependent on the film that forms when they
are first exposed to the service environment. Chapter 3 describes the current
thinking about the composition and nature of the films on the commonly
used alloys, what is needed to make them protective, and what can cause
the formation of less protective films.
Chapters 4 to 14 describe different types of corrosion and how these
can affect copper and its alloys. In particular, each chapter ends with a
section on how to avoid that form of corrosion, both in new plants and in
service. Chapters 15 to 17 deal with the specific environments of under-
ground corrosion, atmospheric corrosion, and corrosion in waters. These
chapters discuss the factors that affect corrosion and the types of corrosion
that may occur, referring readers back to Chapters 4 to 14. The final chapter
discusses corrosion issues associated with the three most common methods
of joining copper alloys: soldering, brazing, and welding.
In addition, there are two appendixes. Appendix A summarizes the
compositions of the commonly used alloys by group. Alloys are mostly
referred to by their UNS numbers,' but national standards have been used
for alloys for which no UNS number exists. Appendix B reviews the differ-
ent test methods used to determine erosion corrosion resistance and assists
the reader in comparing results from the wide variety of methods reported
in the literature.
If this document is read from beginning to end, a certain amount
of repetition is noticeable. This is because most readers will want to dip
into sections relevant to their current problem and will not wish to keep
following references to other chapters. Where the point at issue requires
more lengthy discussion, references to other chapters are unavoidable.
Reference
1. Metals and Alloys in the UniJed Numbering System (Warrendale, PA: SAE
International and West Conshohocken, PA: ASTM International, 2004).
CHAPTER 2
Copper and Its Alloys
In this chapter, the composition and important properties of the common

copper alloys are discussed. This chapter does not pretend to be an in-
depth discussion of the metallurgy and physical properties of copper alloys;
rather, it is intended to acquaint engineers with the important alloys and
their properties. More detail can be found in the references throughout this
chapter. The composition of all the alloys referred to in the text can be
found in the tables in Appendix A. These tables can then be used as a ready
reference when reading any of the subsequent chapters.
2.1 Coppers
The coppers are essentially 99.85% copper or higher, with other elements
being mostly residuals from manufacturing. A major use of copper is as
an electrical conductor, and high-conductivity copper contains a maximum
of 0.04% impurities, although it may contain about 0.05% oxygen. When
heated in a reducing environment, these coppers can become brittle, and
for this reason, oxygen-free high-conductivity copper is recommended.
Copper is widely used in sheet, strip, and foil form, and may be either
tough pitch copper or phosphorus-deoxidized copper (-0.0 I % P). Although
tubing may be made from either form of copper, it is more commonly made
of phosphorus-deoxidized material.
Copper is not very strong in the annealed form, as shown in Table 2.1.
Where higher strength is required, copper may be cold worked, which
greatly increases the 0.2% proof stress at the expense of ductility. For
instance, the typical 0.2% proof stress of copper is around 70 MPa, with
55% elongation. When heavily cold worked, copper’s proof stress increases
5
TABLE 2.1 Typical mechanical properties of copper and some of its alloys at room temperature
in the annealed condition
0.2% Proof stress Tensile strength
Class Common name UNS no. (MPa) (MPa) Elonaation (%)
Copper Phosphorus-deoxidized c12200 70 325 60
(PdO) copper
Arsenical comer C14200 100 300 50
Brass Commercial bronze c22000 80 265 55
Cartridge brass C26000 120 325 70
Aluminum brass C68700 150 360 60
Hot stamping brass C38000 180 400 35
Manganese bronze cz114(A) 195(B) 460(B) 15(B)
Cu-Nil 90110 copper-nickel C70600 140 320
Ni-Cu 70130 copper-nickel C7 1500 170 400
Alloy 729 C72900 760") 900(O
m Marinel C72420 700(') 930(c)
Alloy 400 NO4400 200 500
Alloy K-500 NO5500 800'9 1,OOO(C)
Tin Bronze Phosphor bronze c51000 120 340 60
LG4 gunmetal C93410 140 270 20
Silicon Bronze Silicon bronze C65500 100 340 40
Aluminium Alloy D C6 1400 230 540 40
Bronze Alloy E C63300 290 710 20
Cast nickel C95800 270 660 15
Aluminum bronze
Nickel Silver 63110127 C74500 110 350 65
63120117 NS 108(A) 140 380 50
'A) British Standard quoted as no UNS number.
'B) Minimum mechanical properties.
(c) Typical properties in heat-treated condition.
Copper and Its Alloys 7
to more than 320 MPa, with an elongation of about 4%. An alternative

way to gain a more modest increase in strength is to use arsenical copper
(-0.4% As), which also increases the creep strength at moderately elevated
temperatures.
2.2 Brasses
Copper alloys containing zinc, commonly in concentrations from 5 % to

40%, are known as brasses. Other elements may be added to modify
properties such as strength, machinability, or corrosion resistance. The
strength of brasses increases as the zinc content increases, as shown in
Table 2.1, although greater increases can be obtained by adding additional
elements.
Brasses containing up to about 37% zinc are single-phase alpha al-
loys, whereas alloys containing more zinc are two-phase alloys, with both
alpha and beta phases. The quantity of the beta phase increases as the zinc
content (or alloying additions with a high zinc equivalence) increases. Some
elements, such as silicon and aluminum, have a high zinc equivalence, and
so the beta phase will form at substantially lower zinc contents than in sim-
ple binary alloys. Aluminum brass, with 2% aluminum, has a zinc content
of 22%, rather than 30%, to retain its all-alpha structure, which can then be
protected from dezincification by arsenic additions.
Before the discovery of arsenic's ability to inhibit dezincification,
it was discovered that tin reduced the rate of dezincification of the alpha
phase. This is why alloys such as admiralty brass and naval brass contain
1% tin. The tin addition also improves the machinability of naval brass. The
tin addition to admiralty brass has been observed to improve its corrosion
resistance in moderately polluted seawater, compared with 70/30 arsenical
brass or aluminum brass.' Aluminum was found to increase the resistance
of alpha brass to erosion corrosion, but at concentrations significantly above
2%, an increased susceptibility to localized pitting occurs, particularly in
polluted seawater.*
As discussed in detail in Chapter 9, arsenic is a required addition to
many single-phase brasses to inhibit dezincification. Most specifications
require 0.02-0.06 wt% arsenic. However, in the 1980s, the German DIN
standard 17660 restricted arsenic to 0.02-0.035 wt%. The effect of arsenic
on both intergranular corrosion and stress corrosion cracking (SCC) was
reviewed by Campbell, who concluded that arsenic contents in the range
0.035-0.06 wt% caused no significant increase in susceptibility to either
form of c o r r o ~ i o nIn
. ~ fact, the presence of manganese or iron impurities
can tie up part of the arsenic so that it no longer acts as an inhibitor for
dezin~ification.~ Hence alloys with arsenic at the lower end of the range may
contain insufficient free arsenic to maintain dezincification resistance, and
an arsenic content of 0.035 wt% should be regarded as a minimum, rather
than a maximum. Because dezincification-resistant (DZR) brass often uses
large quantities of scrap in the melting process, it is necessary to increase the
arsenic range for this alloy to 0.02-0.15 wt%, with a typical concentration
being 0.08 wt%.
The brasses traditionally used for hot stamping of fittings, valve bod-
ies, etc., are duplex because all alpha-phase alloys tend to crack at forging
temperatures. However, these duplex fittings are susceptible to dezincifica-
tion in some waters (see Chapter 9). DZR brass was invented to provide a
brass that is two-phased at hot stamping temperatures but can be converted
to an all-alpha structure by a simple, poststamping heat treatment, typically
500°C for 2 hours.4 The presence of arsenic then makes the alloy resistant
to dezincification.
Lead is a common addition to brasses, usually 1 % to 2%, that gives
improved machinability. In addition, lead helps to make sand castings
pressure-tight. Bronze is probably one of the most misused terms in metal-
lurgy. Traditionally, bronzes are alloys of copper and tin. However, some
brasses have been termed bronzes. Thus, 90/10 brass is sometimes called
commercial bronze. Another widely used term is manganese bronze, for
the family of high-tensile brasses. These alloys contain about 38% zinc and
0.5% to 5% manganese, with optional additions of iron, aluminum, and
lead. These are duplex brasses that are heat treated to give moderate to high
strength, with strength increasing as the beta content of the alloy increases.
Because these alloys are duplex, they are susceptible to dezincification; and
so service in aggressive environments, such as seawater, should be avoided
unless the alloy is galvanically protected.
2.3 Bronzes
Traditionally, the bronzes are alloys of copper and tin, with smaller additions
of other elements for specific properties. The alloys known as silicon bronze
and aluminum bronze have been so named because they have many simi-
larities with the tin bronzes with regard to metallurgical properties. Other
alloys called bronzes, such as manganese bronze, commercial bronze, etc.,
are, in fact, brasses and are discussed in Section 2.2.
2.3.1 Tin Bronzes and Gunmetals

The wrought tin bronzes usually contain small additions of phosphorus used
as a deoxidant in the melting process, and are sometimes called phosphor
bronzes. These alloys have very good corrosion resistance, particularly
in seawater, which increases as the tin content increases. The higher-tin
bronzes have very good resistance to polluted seawater (see Chapter 7). The
wrought phosphor bronzes are often used for corrosion-resistant springs.
When the tin bronzes are produced as castings, it is common to add
some zinc, and then the alloys are traditionally known as gunmetals because
of their early use for cannon barrels. To ensure good pressure tightness of
castings, such as in valve bodies and pump casings, it is usual to add
lead. This is why 10% phosphor bronze (C90700) may be used for pump
impellers, but it is rarely used for pump cases because the alloy contains
no lead. Admiralty gunmetal, also with 10% tin, offers the same corrosion
resistance and can be made pressure-tight if lead is at the upper level of its
specification range (-2%).
LG2 gunmetal is a very old alloy that has been cast for over 200 years.
However, it is difficult to cast in thick, complex sections, and for such cast-
ings, LG4 gunmetal or M bronze is recommended.
If the lead content is increased further, to approximately lo%, the
leaded tin bronzes make very good bearing alloys. For such applications,
there is a long history of successful use in pumps and other rotating ma-
chinery.
-
Additions of tin in excess of 10% result in the precipitation of the
delta phase. This makes the alloy even stronger, but also significantly more
brittle; thus, there are few commercial uses of such alloys.
2.3.2 Silicon Bronze

There are a number of silicon bronzes, but by far the most common is a
silicon bronze that contains approximately 3% silicon and 1% manganese
and produces an alloy with moderate strength. When manganese bronze is
used for fasteners, it is susceptible to dezincification in seawater, and many
other waters, as well as to stress corrosion cracking. Silicon bronze offers
a more corrosion- and SCC-resistant alloy, which has a similar strength to
manganese bronze. The mechanical properties in Table 2.1 are for annealed
material, but in the as-manufactured condition, the 0.2% proof stress can
be 400 MPa, while still retaining 20% elongation. The stress corrosion
cracking resistance of silicon bronze is discussed in more detail in Chap-

ter 10.
2.3.3 Aluminum Bronze

The copper-aluminum alloys combine high strength with good corrosion
resistance and are widely used in both wrought and cast forms, particularly
nickel aluminum bronze (NAB). For a detailed discussion of their metal-
lurgy, properties, and corrosion resistance, the reader is referred to Harry
Meigh's book on this class of copper alloy.5 An extensive review of the
corrosion resistance of aluminum bronzes was prepared by Campbell.'
Up to about 8% aluminum, the alloy is single-phase. With higher
aluminum content, the beta phase is formed, which has good corrosion
resistance and superior erosion corrosion resistance to the alpha phase.
However, if these two-phase alloys are cooled too slowly, the gamma-2
phase can form, and this phase is highly susceptible to corrosion. However,
iron contents of 2% or more suppress the formation of the gamma-2 phase
and refine the grain structure.
When the aluminum content is 9% to lo%, additions of nickel and iron
of 4% to 5% produce an alloy with high strength and corrosion resistance.
The structure of the alloy is complex and was explained by Lorimer et al.7
A simplified diagram of the structure is shown in Figure 2.1. As the alloy
cools, the martensitic beta phase transforms to alpha and kappa phases.
Figure 2.1 shows four kappa phases, all of which are enriched in iron and/or
nickel. If the alloy cools too quickly, such as occurs during welding or with
thin sections, not all the beta phase may transform. The residual beta phase
(labeled "martensite" in Figure 2.1) is anodic to the alpha phase and it will
rapidly corrode in However, a short heat treatment will ensure
that all the beta phase transforms to alpha and kappa. For castings, heating
to 675" f 25°C for 4-6 hours followed by air cooling is recommended. For
wrought products, a higher temperature of 725" f 25°C is preferred. For
this reason, NAB should not be used in the as-welded condition in seawater,
and Royal Navy specifications forbid it (see Chapter 18).
The kappa I11phase can cause corrosion problems if it becomes anodic
to the alpha phase because it is present in long, thin stringers. The origin
of this problem and the best ways of avoiding it are discussed in Chap-
ter 9.
Alloys with very high manganese content (> lo%), such as CMAl
(see Table AS), have an alpha-beta structure, but the beta phase has a
very different composition to the beta phase of binary copper-aluminum
KN U ICIy K, KnI martensite
FIGURE 2.1 Schematic of the structure of nickel aluminum b r o n ~ e . ~
alloys, and it is more susceptible to corrosion. However, these alloys have

good cavitation resistance and a high fatigue strength.
Generally, the corrosion resistance of the aluminum bronzes increases
as the aluminum and other alloying additions increase. Hence, NAB, in
both wrought and cast forms, is highly corrosion-resistant, but it is very
susceptible to erosion corrosion in polluted seawater (see Chapter 7): in such
applications, high-tin bronzes are preferred. In addition to clean seawater,
NAB has excellent resistance to corrosion in a wide range of aggressive
chemicals, including hydrofluoric acid (HF) and fluorine.6
2.4 Copper-Nickel and Nickel-Copper Alloys

Some common copper-nickel alloys are shown in Table A.2. Nickel and
copper form a continuous solid solution, so all copper-nickel compositions
are single phase, unless other elements are added. Manganese is usually
present because it is used as a deoxidant in the melting process. Manganese
also acts as a solid solution strengthener and increases the resistance to
erosion corrosion.8 However, iron is five times more efficient at increasing
the resistance to erosion corrosion, and this is why alloys such as 90/10 and
70/30 copper-nickel contain iron.8 With 30% nickel, iron is soluble at up to
several percent; but with 10%nickel, the solubility limit of iron at room tem-
perature is about 1%. Hence, commercial 90/10 copper-nickel, with 1.5%
or more iron, must be quenched from the annealing temperature (-900OC)
to keep the iron in solution. If this is not done, the corrosion resistance
decreases significantly. Hence, it is important to check the quality of 90/10
copper-nickel with a simple magnetic permeability gauge to ensure that the

iron is in solution. This is discussed in more detail in Chapters 7 and 17.
Copper-nickel alloys, particularly 90/10 and 70/30 copper-nickel, are
widely used in marine applications because of their excellent resistance to
both fouling and corrosion. They are used for heat exchanger tubes, piping
for seawater and firewater systems, and cladding for larger steel items, such
as water boxes.
There have been numerous evaluations of other alloying additions to
copper-nickel alloys to improve their corrosion resistance. Kirk reviewed
a large body of mostly unpublished data and there is a summary of this
and other works in Chapter 7.9 Although some of the alloying additions
conferred benefits, there was usually a downside. For example, an increase
in the resistance to erosion corrosion was often accompanied by increased
sensitivity to crevice corrosion or stress corrosion cracking. Only one alloy-
ing element conferred increased strength and resistance to erosion corrosion
without a significant negative impact, and that was chromium.
Table A.2 shows the composition of two wrought alloys and one
cast alloy containing chromium. The cast alloy is an INCO experimental
alloy, but a similar composition (with -2% Cr) is used by the Royal Navy
to replace NAB castings that have suffered corrosion problems. Although
the author knows of no other users of the cast alloy, Alloy 722 (with 0.5%
Cr) has been used for heat exchanger tubes, as an alternative to 90/10 or
70/30 copper-nickel, where there are suspended solids in the water (see
Chapter 7).
The strength of the common copper-nickel alloys is only moderate,
and there are applications that require much higher strength, combined
with corrosion resistance. One way to boost the strength of copper-nickel
alloys is to add tin (as in Alloy 729). This alloy displays spinodal strength-
ening through the development of submicroscopic chemical composition
fluctuations. Table 2.1 shows the much higher strength obtainable with
Alloy 729 compared with single-phase copper-nickel alloys. An alternative
method of increasing the strength is to add aluminum and niobium (as in
Alloy C72420) and heat treat to precipitate the gamma prime phase. After
heat treatment, such alloys have similar strength to Alloy 729, as shown in
Table 2.1. The principal use of these high-strength alloys is for marine fas-
teners, where high strength and corrosion resistance are required. Although
both high-strength alloys have good resistance to corrosion and erosion
corrosion in seawater, they can suffer SCC under some circumstances, in
some fluids. This is discussed in more detail in Chapter 10.
Alloy 400 is a nickel-copper alloy with good resistance to corrosion

and excellent resistance to erosion corrosion in seawater. However, the
alloy is susceptible to crevice corrosion in clean seawater (see Chapter 6)
and deep pitting in polluted seawater, despite its high nickel content (see
Chapter 7). It is not galvanically compatible with high-alloy stainless steels
and nickel-chrome-molybdenum alloys in seawater, but it can be safely
coupled to copper alloys (see Chapter 14).
For higher-strength applications, Alloy K-500 is available, which has
additions of aluminum and titanium so that gamma prime is formed during
heat treatment. Table 2.1 shows the large increase in strength that is possible
with this alloy compared to Alloy 400. Alloy K-500 behaves more or less
like Alloy 400 in terms of corrosion.
2.5 Nickel Silvers

The name of this group of alloys is a little confusing because, although they
contain nickel, they contain no silver. The name comes from their appear-
ance. This class of copper alloy has excellent tarnish resistance, and they
are essentially single-phase brasses with a high nickel content, as shown
in Table A.6. Because they are not inhibited, these alloys will suffer dez-
incification, particularly in high-chloride waters. However, their resistance
to atmospheric corrosion is excellent, although they may suffer SCC in the
presence of moist ammonia (NH3). Higher nickel content equates to greater
resistance to SCC. Normally, alloys with 18% nickel have good resistance
to tarnishing, and there is no real need for the higher-nickel alloys, unless
the atmosphere is particularly aggressive.
Nickel silvers have excellent spring properties, and have been widely
used in electrical equipment for both springs and electrical connectors. The
higher-nickel alloys are somewhat stronger than the lower-nickel ones, as
shown in Table 2.1, although greater increases in strength can be obtained
by moderate cold working.
References
1. R.S. Bem, P.T. Gilbert, BNF Metals Research Association Report, A1095,
September 1955.
2. R.S. Bem, BNF Metals Research Association Report, A 1 132, December 1956.
3. H.S. Campbell, BNF Metals Technology Centre, Miscellaneous Publication,
MP600. December 1982.
4. J.E. Bowers, P.W. Oseland, G.C. Davies, “Development of a Hot-Stamping Brass

Resistant to Dezincification,” BK Corros. J . 13,4 (1978): p. 177.
5 . H.J. Meigh, Cast and Wrought Aluminium Bronzes, Properties, Processes and
Structures (London, U.K.: Institute of Materials, 2000).
6. H.S. Campbell, “Aluminium Bronze Alloys Corrosion Resistance Guide,” Copper
Development Association U.K. Publication, 80, July 1981.
7. G.W. Lorimer, F. Hasan, J. Iqbal, N. Ridley, “Observation of Microstructure and
Corrosion Behavior of Some Aluminium Bronzes,” BK Corros. J. 2 1 , 4 (1 986):
p. 244.
8. G.L. Bailey, “Copper-Nickel-Iron Alloys Resistant to Seawater Corrosion,” J. Inst.
Met. 79,5 (1951): p. 243.
9. W.W. Kirk, “Evaluation of Alloying Effects on Corrosion Behaviour of
Copper-Nickel Alloys in Seawater,” INCRA Report, PRJ382, April 1986.
CHAPTER 3
Film Formation and Properties
In many environments, copper and its alloys corrode at low rates, which is
due to the protective nature of the films that form. These films are not simple,
and the mechanisms by which protection is conferred are often complex.
In some cases, they are still not well understood. The films and their prop-
erties for some common copper alloys are reviewed in this chapter, along
with the effects of some common additions to waters and some common
pollutants.
3.1 Copper and the Brasses

At ambient temperature, the principal corrosion product on copper in aque-
ous solution is cuprous oxide (Cu20). In its pure form, it is bright red, but
small quantities of impurities, such as iron, manganese, and nickel, can
change its color to orange, dark red, brown, or nearly black. In hot waters,
cupric oxide can also form, usually as a black layer on top of the cuprous
oxide. The ratio of the two oxides cannot be used to infer the operating tem-
perature because small quantities of impurities can affect the temperature
at which cupric oxide first forms and also the ratio of the two oxides.
Diffusion through cuprous oxide is relatively easy because of the
high number of cation vacancies in the lattice.' North and Pryor showed
that the rate of corrosion is controlled by the combined electronic and ionic
resistance of the Cu20.' They also showed that on exposure to water, the
defect structure of cuprous oxide changes. This explains why air-formed
cuprous oxide films on copper do not confer protection in water. The air-
formed film is broken down, and then a new film forms with a modified
structure.'
15
When copper is first placed in water, the initial corrosion rate is

high but decreases as the cuprous oxide film forms and thickens2 The
effect of chloride ions in the water is to increase the corrosion rate, both
initially and after a Cu20 film has formed. Once a stable cuprous oxide
film forms, it slowly but continually dissolves at the film-water interface,
and new oxide forms at the metal-film interface. The copper that is slowly
released from the oxide can accumulate there, if the water flow is not too
turbulent, until the solubility of copper in the water is exceeded and copper
hydroxychloride (Cu2Cl[OHj3JtH20) deposits on the surface of the oxide,
if there is a significant chloride content in the water. This is usually as
one of the atacamites, for example, paratacamite ( C U ~ [ O H ] ~ C ~ . ~In H~O).
hard waters of low chloride content, it is more common to get a layer of
basic copper carbonate, for example, malachite (CuCO3.Cu[OH]2). In soft
waters, it is common to get basic copper sulfate deposited over the cuprous
oxide. These green outer films are usually not very protective and are easily
removed if the water turbulence increases.
The other anions that are commonly found in waters are sulfate, bicar-
bonate, and nitrate. Sulfate is strongly adsorbed onto cuprous oxide, which
can affect the nature of the film and the way that it breaks down.2 This is
discussed in more detail in Chapter 5. Nitrate is not adsorbed onto copper
oxide, and although it can affect the pitting resistance of copper, its effect is
much less than that of sulfate. Shalaby et al. found that bicarbonate had little
effect on the corrosion of ~ o p p e rThey
. ~ added bicarbonate as the sodium
salt, while Ives and Rawson investigated both sodium and calcium Salk2
They found that sodium bicarbonate (NaHC03) did not adsorb onto cuprous
oxide, whereas calcium bicarbonate (Ca[HC03 12) adsorbed strongly. Bi-
carbonate is usually present in waters as calcium salt and this explains why
Type I pitting occurs in hard and moderately hard waters (see Chapter 5).
This is because the presence of adsorbed bicarbonate makes film breakdown
more difficult so that it occurs at isolated spots, rather than at numerous sites.
Mor and Beccaria studied the effect of temperature on corrosion and
found that the corrosion rate of copper decreased at higher temperatures in
synthetic ~ e a w a t e rThis
. ~ suggests that an oxide with fewer defects forms
at higher water temperatures, following the model of North and Pryor.’
One property of cuprous oxide that is often not recognized is its
photoelectric sensitivity. While this is of little concern inside pipes and
closed tanks, it can influence corrosion where the copper is exposed to
sunlight.’ Depending on its electrochemical potential and the wavelength
of light to which it is exposed, cuprous oxide can be either photopositive
or ph~tonegative.~ Wilhelm et al. suggested that illumination during initial
film growth could affect how protective the oxide was in service.’
Film Formation and Properties 17
When brasses corrode, they also form cuprous oxide. Zinc tends to
be detected in the film as well-more so as the zinc content of the brass
increases. It has not been determined whether zinc substitutes for copper
in the oxide matrix or whether zinc compounds are formed for low zinc
brasses. If the chloride content of the film is high, such as in seawater, many
of the zinc compounds that might form are soluble.
Beccaria and Poggi investigated the corrosion of admiralty brass in
seawater at 40°C.6 After 15 days, they identified both cuprous and cupric
oxides, zinc oxide (ZnO), and zinc sulfate (ZnSo4.7H20) as well as copper
hydroxychloride. They also found a small quantity of tin oxide (SnO).
Admiralty brass has been observed to perform better than both 70130 brass
(C26000) and aluminum brass in slightly polluted water. This was always
believed to be due to the presence of the 1% tin in the alloy. The detection
of tin oxide in the film on admiralty brass shows that tin is active in the
corrosion process.
Albaya et al. also studied the films on admiralty brass, but they did
not find compounds other than cuprous oxide and copper hydro~ychloride.~
However, their tests were in sodium chloride (NaCl), rather than natural
seawater, and they only used x-ray diffraction to identify the corrosion
products. Beccaria and Poggi used more sophisticated techniques that do
not require the corrosion product to be crystalline.6
3.2 Aluminum Brass

Aluminum brass behaves somewhat differently to other brasses, and the
films that form on aluminum brass can vary substantially. In some fresh
waters, sodium chloride solution, and synthetic seawater, the film on alu-
minum brass is cuprous oxide, as seen on 70/30 brass. However, in certain
fresh waters, aluminum brass has been observed to pit.' When this hap-
pened, the film on the nonpitted surfaces was identified as a mixed spinel of
zinc and aluminum (ZnO.A1203).' The nature of this film and the absence
of copper suggest that there was organic material or microbes in the water
that are chelating the copper and preventing it from becoming incorporated
into the film.
In natural seawater, the films on aluminum brass are different
again. Usually, there is an outer orange/brown deposit of hydrated
iron oxide (FeOOH), which is formed either from corroding iron or
steel, or by the regular addition of ferrous sulfate (FeS04.7H20) to
the seawater. Beneath this is a thin, soft film, usually white in color,
which is a mixed basic hydroxy carbonate of magnesium and aluminum
of the hydrotalcite (Mg6A12[OH] 16C03.4H20) family of compounds
(Mg6A12C03[OH] 16 .4H20).This compound was identified by several lab-

oratories, all investigating a spate of heat exchanger tube failures in very
large crude carriers in the early 1970s. The nature of the film and the mech-
anism by which it protects the metal surface was the subject of extensive
work by Castle and his coworkers at the University of Surrey, in the United
Kingdom.9- I
Hydrotalcite contains two metal cations: one a minor element in the
alloy and the other a minor constituent of seawater. For this compound to
form, both copper and zinc must be chelated by microbial action to prevent
their incorporation into the film. Further work by Castle et al. showed
how aluminum brass reacts with the various biological layers that form on
aluminum brass in natural seawater.I4-l5
The mechanism by which hydrotalcite protects the surface involves
its ability to vary its composition widely from the nominal composition. As
the pH decreases, hydrotalcite sheds magnesium and takes up aluminum.
As the pH increases, the reverse happens.”13 Thus, the hydrotalcite is able
to buffer the pH at the film surface and prevent accelerated attack.
Hydrotalcite is a soft compound with little mechanical strength, and
so it cannot resist turbulent water flow. Castle and Parvisi showed that the
hydrotalcite is not only present as a film on the surface, but it is also present
in the pores in the FeOOH layer above it.” Thus, the hydrated iron oxide is
able to provide an erosion-resistant network within which the hydrotalcite
is fixed.
3.3 Copper-Nickel Alloys
The two most studied copper-nickel alloys are 90/10 (C70600) and 70/30
copper-nickel (C7 1500).These studies have used sodium chloride solution,
synthetic seawater, and natural seawater. The results in all three media are
similar, but the films formed in natural seawater have some properties that
those formed in synthetic solutions do not.
North and Pryor showed that cuprous oxide forms on copper-nickel
alloys, and the resistance of this film is improved by the incorporation of
nickel into vacant lattice sites and as a substitute for copper ions.’ The
nickel content of the film increases as the nickel content of the alloy in-
creases. However, later work has shown that the films on copper-nickel
alloys are more complex than this in solutions with significant chloride
concentrations. Ijsseling et al. and Drolenga et al. were among the first
to detect a thin layer next to the metal surface that was rich in copper
and chloride ion^.'^-'^ Detailed surface analysis by Kato et al. and Castle
and Parvisi showed that it is this thin layer of copper chloride at the metal
-50
w -100
0
rn
>
=.-E
-
m
C
al
-150
2 -200
-250
-300
0 10 20 30 40 50 D
Time (days)
FIGURE 3.1 Potential vs. time for 90/10 copper-nickel at various seawater
temperatures.20
surface that provides the high level of cathodic polarization seen with 90/10
' ~ - ' ~high resistance to corrosion of 90/10 copper-
~ o p p e r - n i c k e l . ' ~ ~The
nickel is achieved primarily by inhibition of the cathodic process: the re-
duction of dissolved oxygen. In contrast, the outer layer of cuprous oxide
enriched in nickel and iron contributes only a small amount of anodic polar-
ization. The iron in 90/10 copper-nickel is essential for resistance to erosion
corrosion, and Castle and Parvisi suggested that under turbulent water con-
ditions, the iron forms an outer layer of hydrated iron oxide (FeOOH),
which provides a mechanical defense in the same way that ferrous sulfate
additions do for aluminum brass (Section 3.2).13
The enrichment of the oxide layer with nickel and iron appears to be
temperature-dependent. '33'6,20 At ambient temperature, the oxide is mostly
enriched with nickel, while at elevated seawater temperatures (around
40"C), an iron-rich film develops. Ijsseling et al. showed that there is a poten-
tial change at elevated seawater temperatures, compared with that observed
at 10"C.20At 20°C the potential moved more positive after -25 days, fin-
ishing at around -100 mV SCE (saturated calomel electrode) (Figure 3.1).
At 40"C, the potential moved steadily more negative, finishing at about -
280 mV SCE. The film at 40°C gave the lowest corrosion rate, and it appears
that the changes in the film at 20°C were different from those at 40°C.
16
14
-300 -250 -200 -150 -100 -50 0

Potential (mV SCE)
I--1-0 -Fe -Ni -CuI
FIGURE 3.2 Concentration of elements in the film on 90/10 copper-nickel vs. p0tentia1.I~
(See color section.)
The potential at which 90/10 copper-nickel is first exposed has a strong

effect on the nature of the protective film. Blundy and Pryor demonstrated
that nickel enrichment is strongly a function of the potential.*' More detailed
work by Castle and Parvisi (Figure 3.2) showed that the chloride-rich layer
formed only over a nairow potential range, while nickel and iron enrichment
occurs at somewhat more positive potentials. l 3
Less work has been carried out on the films that form on 70130 copper-
nickel, but Castle and Parvisi suggest that the films are similar in nature
to those on 90/10 copper-ni~ke1.l~ It is certainly well established that the
cuprous oxide film is enriched in nickel.'-22 Castle and Parvisi failed to
find any iron enrichment in the film, although their exposure time was only
3 hours.I3 However, 90/10 copper-nickel showed iron enrichment after the
same length of time in seawater. Both Efird and Francis showed that over a
period of days, the potential of 70/30 copper-nickel moves considerably in
the positive d i r e ~ t i o n . However,
~ ~ - ~ ~ this potential change is a function of
temperature, and at temperatures of 10°C and below, no change of potential
occurred (Figure 3.3).24Simple film analysis showed that the films on tubes
with positive potentials were enriched in iron, and sometimes in nickel as
-2504 . I . , . . . . . I . , . . .
0 1 2 3 4 5 6 7 6
Time (days)
FIGURE 3.3 Potential vs. time for 70/30 copper-nickel in seawater at various
temperature^.^^
well. Francis postulated that iron was continuously dissolving from the alloy
through the film, and an iron-rich film formed when the iron concentration
at the film-liquid interface exceeded the solubility limit.24
Efird and Francis also examined the 66/30/2/2 Cu/Ni/Fe/Mn alloy
(UNS C71640) and found a similar effect of temperature on potential
enn~blement.~ The
~-~ films
~ were very thin, and the simple analysis that
was carried out by both workers showed no iron enrichment in the films
with positive potentials. However, they argued that more detailed analysis
by techniques such as x-ray photoelectron spectroscopy (XPS), might re-
veal iron enrichment. Castle and Parvisi suggest that the protective layer
on 70/30 copper-nickel is similar to 90/10 copper-nickel and contains an
inner layer rich in copper and chloride ions, but they present no detailed
analysis. l 3
Ross reported tests on 90/10 copper-nickel in treated seawater at
elevated temperatures, as is found in parts of multistage flash (MSF) de-
salination plants.25 Copper oxide was the primary corrosion product, but
there was also a considerable concentration of magnesium found in the
film at both 60°C and 107°C. Some of this may have been magnesium
hydroxide; but pyroaurite (Mg6Fe2C03[OH] 16.4H20), which is related to
hydrotalcite, as seen on aluminum brass (Section 3.2), was also detected on

some samples. It was not determined whether this compound acts as a pH
buffer, as hydrotalcite does. Ross also found deposits of calcium carbonate
(CaC03) on the oxide films, which seemed to be a factor further reducing
corrosion by acting as a barrier.25Carbonate scaling does not usually occur
in MSF plants because the water is treated to prevent this, as the scaling
causes a reduction in plant efficiency.
3.4 Aluminum Bronze
The air-formed film on aluminum bronze is mainly alumina (Al2O3), and

this gives the alloys very good corrosion resistance in air up to 600°C or
higher.26 In seawater, alumina has been found in the films on aluminum
bronze, but this was after a short exposure to seawater, when alumina
from the air-formed film could still be expected on the metal s ~ r f a c e . ~ ~ - ~ *
Francis and Maselkowski carried out tests on cast nickel aluminum bronze
(C95800) for periods of 1-12 weeks in natural seawater.29The films were
analyzed after 4 and 12 weeks’ exposure at 10°C. Samples were exposed
at both 0.1 and 2.5 m/s flow velocity. After 4 weeks, there was a thin,
greenish outer layer, which was presumed to be copper hydroxychloride.
Beneath this was a black layer, which was tentatively identified as a deriva-
tive of Fe3A1 by x-ray diffraction. The black layer was slightly enriched in
aluminum compared to the base metal but was greatly enriched in nickel
and iron. The results after 12 weeks were a little more confused because
of the much thicker layer of copper hydroxychloride. However, not only
was the film enriched in aluminum, nickel, and iron, but there was also a
small amount of magnesium present. It is possible that this was present
as hydrotalcite, as was found on aluminum brass (Section 3.2). However,
it could not be identified by x-ray diffraction. The exact nature of the
films and the method by which they confer protection were not deter-
mined, but unpublished work by Castle on NAB also found iron and nickel
enrichment in the film and a lower level of aluminum than in the base
metal.30
3.5 Tin Bronzes

These alloys have received even less attention than the aluminum bronzes
in terms of the nature of the protective films, but their excellent resistance to
corrosion in seawater is well recognized. Atsumi et al. examined the films
on a commercial 8% tin bronze alloy ((252100) exposed in seawater.” From

an energy-dispersive x-ray analysis of the film, they concluded that the very
high tin content of the film suggested the presence of a tin oxide as a distinct
phase. Oxides of tin are both robust and stable, and the presence of a layer
of essentially tin oxide would explain the good corrosion resistance of the
tin bronzes.
3.6 Ferrous Sulfate

Throughout the first half of the 20th century, admiralty brass (C44300) and
aluminum brass (C68700) were the most commonly used alloys for tubes in
seawater-cooled heat exchangers. In the 1950s, there was a move to replace
bare carbon steel and cast iron components with more corrosion-resistant
alloys or to apply coatings to increase reliability. Following this, there was
an increase in the number of heat exchanger tube failures due to erosion
corrosion. In 1961, Bostwick described how daily additions of ferrous
sulfate to the cooling water replaced the iron-rich films that had previously
been formed by corroding steel, with a large reduction in the number of
tube failures.32Shortly after, Lockhart described a similar success story at
a U.K. power station.33Since that time, additions of ferrous sulfate solution
(typically 1 mg/l Fe2+ for 1 h/d) have been made at many plants to solve
heat exchanger corrosion problems.
As described in Section 3.2, the FeOOH layer that is formed by ferrous
sulfate additions appears to offer mechanical protection in turbulent water
to the more fragile films adjacent to the metal. The use of iron additions to
combat the corrosion of copper alloy heat exchanger tubes was surveyed
for U.K. power stations, and the published literature was examined to
determine when ferrous sulfate dosing would be effective, and for which
alloys.34
Although intended originally for brass tubes, ferrous sulfate has oc-
casionally been used successfully to solve corrosion problems with 90110
and 70/30 copper-nickel. The original corrosion problem that iron addi-
tions were used to overcome was erosion corrosion (impingement attack),
but they have also been used to try and solve other corrosion problems
with varying degrees of success. In Japan, ferrous sulfate dosing solved
problems of localized attack of aluminum brass tubes that had a cathodic
film of manganese dioxide (MnOz) on them.35 Manganese usually is not
present at high concentration in seawater. However, this film was produced
by the oxidation of manganese in solution in seawater by chlorine, added
to control fouling in the local seawater.
Ferrous sulfate dosing has also solved corrosion problems caused

by ammonia (NH3) pollution. The author used ferrous sulfate dosing for 2
months at the start-up of a MSF desalination plant to form more protective
films on 90/10 copper-nickel tubes. The water was intermittently polluted
with ammonia, which had caused low-temperature hot spot attack in the
heat exchangers when the first unit was started. These were retubed, and
ferrous sulfate dosing was started from new at a level of 1 ppm Fe2+ for
1 hour every day. The plant has now been in operation since 1994, with
no further failures. At another MSF plant in the Middle East, the ammonia
pollution was almost continuous because the water intake was near the out-
fall of a fertilizer plant. Daily ferrous sulfate dosing throughout its life was
recommended for this plant as it was tubed with aluminum brass. There had
been failures of 9 0 4 0 copper-nickel tubes in an earlier plant on the same
site. The plant has now been in operation since 1998, with no problems.
Ferrous sulfate does not work in water polluted with hydrogen sul-
fide (HzS), unless protective films can first be formed in clean water.36 A
U.K. power station, operating in a tidal estuary, used ferrous sulfate dos-
ing to prevent corrosion of 70/30 copper-nickel tubes, where H2S was an
intermittent pollutant at certain states of the tide.34
All these experiences suggest that the initial formation of a hydrated
iron oxide layer, under relatively clean seawater conditions, enables a more
protective film to form beneath, at least in some cases. The mechanism by
which this occurs is not known but could simply involve the FeOOH layer
acting as a barrier to harmful species in the bulk water.
It is well documented that chlorinehypochlorite dosing. to control
fouling interferes with the efficacy of ferrous sulfate The rec-
ommendation is that if the chlorination is continuous, it should be turned
off while ferrous sulfate dosing is taking place. If the chlorination is inter-
mittent, then it should be staggered so that it does not occur during ferrous
sulfate dosing.
Service experience with Baltic and Arctic waters suggests that alu-
minum brass does not perform quite as well as might be expected, and some
companies prefer copper-nickel alloys for heat exchangers. The reason for
this may be connected to the fact that ferrous sulfate dosing is common
with aluminum brass to improve protective film formation and corrosion
resistance. Effertz and Fichte examined the formation of iron-rich films on
copper alloys due to ferrous sulfate dosing and found that the thickness of
the film was proportional to the quantity of FeOOH colloid produced by
adding ferrous sulfate to the cooling water.39They found that the colloid
yield doubled for every 10°C increase in temperature and that the yield at
low temperatures (around 5°C) was very low. Hence, the iron-rich films
on aluminum brass will not be very thick and protective in cold water, in-
creasing the susceptibility to erosion corrosion. This explains the superior
performance of copper-nickel alloys, which do not normally require fer-
rous sulfate dosing to ensure good corrosion resistance. Much of Effertz and
Fichte’s work was carried out with river water, but they found no difference
in FeOOH colloid formation in fresh water and seawater.
3.7 Chlorine
Chlorine is added to seawater cooling systems to control macrofouling
such as by weeds and shellfish. It also has the effect of keeping micro-
fouling (slime) at acceptable levels. Chlorine is usually generated elec-
trolytically in a bypass line and is then injected into the seawater intake,
although in some land-based plants, tanks of sodium hypochlorite (NaOCl)
may be used instead. Chlorine is a powerful oxidizing agent, and it re-
acts rapidly with organic material in seawater. Hence, the concentration
decreases rapidly from the point of dosing. To control fouling, it is only
necessary to have a very small residual (for example, 0.1 mg/L) of free chlo-
rine at the o~tfall/discharge.~~The chlorine content at the inlet will depend
on the organic activity of the water (biological chlorine demand) and the
length of piping from intake to outfall. For large, land-based plants, chlorine
concentrations from 5 to 25 mg/L may be necessary at the seawater intake
at the height of the fouling season, while a ship may manage with 1 mg/L
or less.
Chlorine interferes with the formation of FeOOH films from ferrous
sulfate dosing. It oxidizes the FeOOH colloid, making it more difficult for
it to be attracted to the tube wall. Hence, ferrous sulfate dosing is usually
not very effective with continuous chlorination of the water, unless the
chlorination is stopped during the ferrous sulfate dosing.38
Chlorine also affects film formation on copper-nickel alloys, if the
concentration is high enough. As the loss of corrosion resistance is seen
most under turbulent water conditions, it is presumed that the chlorine
prevents the formation of a mechanically protective, iron-rich layer.
Francis reported the effects of continuous chlorine additions to sea-
water on the films on aluminum brass that was also dosed with ferrous
sulfate.41 In clean seawater and seawater containing 1 mg/L chlorine, a
white film of hydrotalcite formed, as usual. However, with the addition of
4 mg/L chlorine, the film was thin and golden, and XPS analysis showed it
to be predominantly monovalent copper and chloride. The film also showed
\
\
-6001
0.01 0.1 1 10
Current Density (pA/cmZ)
-2hl NO CI --- 60dl No CI - -6Od/ 4mglL CI I

FIGURE 3.4 Cathodic polarization curves for aluminum brass in ~eawater.~’
a high degree of cathodic polarization (Figure 3.4) compared with the film
produced after exposure to clean seawater. The inhibition of the cathodic
reaction by a film rich in chloride and copper was seen for 90/10 copper-
nickel (Section 3.3), and it appears that high concentrations of chlorine can
cause its formation on other copper alloys. Unfortunately, the copper chlo-
ride layer is not mechanically strong, and it is easily disrupted in turbulent
water. The presence of high concentrations of chlorine would prevent the
formation of a FeOOH layer to give mechanical protection.
Francis et al. also investigated the films formed on cast nickel alu-
minum bronze in seawater at 10°C containing 0.5 mg/L chlorine.29Their
results in clean seawater were presented in Section 3.4. In chlorinated sea-
water, the films were generally of the same appearance as in clean seawater.
While the films formed in chlorinated seawater were also enriched in nickel
and iron compared with the base metal, enrichment was not found to the
same extent as in clean seawater. In addition, magnesium and chloride were
present in small but significant quantities in the films formed in chlorinated
seawater, although their compounds were not established.
3.8 Sulfide
When hydrogen sulfide is present in seawater, it is almost always pro-
duced by the action of sulfate-reducing bacteria (SRB). These bacteria
need anaerobic conditions to thrive and are normally inactive in aerated
seawater. When the water is stagnant or access to oxygen is restricted, for
example, in bottom mud, the aerobic bacteria gradually consume the dis-
solved oxygen. Once the oxygen is gone, the anaerobic bacteria, including
SRB, become active. They reduce the sulfate in seawater to H2S as part
of their metabolic processes. In deaerated seawater containing H2S, the
corrosion rate of copper alloys is low, similar to the corrosion rate in clean,
aerated seawater.42 However, a mixture of deaerated seawater containing
H2S and aerated seawater is very aggressive to most copper alloy^.^^-^^ This
can happen when stagnant bottom water is mixed with aerated top water by
tidal action or incorrect siting of the suction intake.
This mixture of aerated seawater containing H2S is not stable, as the
dissolved oxygen and the H2S react together:
2H2S + 0 2 + 2H20 + 2 s (3.1)
However, this reaction is slow, with a half-life of 15 to 20 minutes, which is
more than enough time for the reacting mixture to pass through most cooling
systems. If the H2S and 0 2 have time to fully react, say, in a holding tank or
basin, then the elemental sulfur that is produced is still aggressive to copper
alloys, but not as much as H2S.43
Eiselstein et al. explained the corrosion of copper-nickel alloys using
the Evans diagram.43 The effect of sulfide is to move the anodic curve
electronegatively, but this does not cause a large corrosion rate in deaerated
water because the cathodic reaction is hydrogen ion reduction. However,
when dissolved oxygen (DO) is present, the cathodic reduction of DO is the
cathodic reaction, and this occurs at much more electropositive potentials.
The result is a very high corrosion rate. As corrosion by sulfides usually in-
volves localized pitting under aerated conditions, corrosion rates exceeding
10 m d y are p0ssible.4~
The protection of copper-nickel alloys by a thin layer rich in copper
and chloride adjacent to the metal surface was described previously.'3 Cas-
tle and Parvisi also showed that there is a difference in the properties of this
film, depending on whether it is formed in synthetic solutions or natural
seawater. Films formed in synthetic solutions showed no significant loss
of cathodic polarization when the solution was deaerated (Figure 3.5A);
however, films formed in natural seawater showed a substantial loss of
-200
-220
I
-240.
G
0
v)
-260-
-.-m
Y
.- --------
c
f -280-
c
0
0.
-300 -
-320.
-340
Current Density (pA/cm2)
I-Aerated --- Deaerated I
--- I
0.01 0.1 1 10 100 1,000
Current Density (pA/crn2)

-Aerated --- Deaerated
FIGURE 3.5 Cathodic polarization curves for 90/10 copper-nickel after exposure to
(A) sodium chloride solution and (B) natural ~ e a w a t e r . ' ~
polarization when the solution was deaerated (Figure 3.5B). This was ob-
served for both 90/10 and 70/30 copper-nickel. Clearly, there is an organic
compound present in natural seawater that aids the cathodic polarization
reaction and that is deactivated in deaerated seawater. Castle and Parvisi
presented some preliminary results of identifying the type of organic com-
pound that causes this effect.13
The presence of sulfides in seawater is well known to cause accelerated
corrosion of copper alloys, and the outer layer is usually a thick, porous
layer of copper sulfide (CuS). Beneath this is a thin layer of cuprous oxide,
containing both nickel and sulfur. These layers are clearly not protective,
and the copper chloride layer is absent. Corrosion rates are observed to be
highest where the water is most turbulent, although the attack is corrosion,
rather than erosion corrosion. This presumably means that the transport of
oxygen and sulfide to the metal surface is controlled by its availability in
solution, rather than by diffusion through the sulfide-oxide film. Kato et al.
observed what appeared to be enhancement of the cathodic reaction; that is,
depolarization, by films on 90/10 copper-nickel formed in sodium chloride
solution containing H2S4
When the conditions are restored to clean seawater, the copper sulfide
gradually spalls off, and the normal chloride and oxide films are slowly re-
stored on copper-nickel alloys.42It can take days or weeks before corrosion
rates decrease to the levels normally found in clean seawater.
Nickel aluminum bronze is particularly susceptible to attack by sul-
fides. Schussler and Exner observed that unlike the copper-nickel alloys,
NAB did not reform protective films in clean seawater following exposure
to sulfides, even after 50 days.45
The high-tin bronzes (Sn >7%) are well known to have good resis-
tance to corrosion in sulfide-polluted ~eawater.~' Although the films that
form on these alloys in clean and polluted seawater have not been studied in
detail, it is known that tin oxides are thermodynamically more stable than
tin sulfides. Hence, the more protective oxide should form, provided that
there is some DO present to aid its formation.
References
1. R.F. North, M.J. Pryor, Corros. Sci. 10 (1970): p. 297.
2. D.J.G. Ives, A.E. Rawson, J. Electrochem. SOC. 109 (1962): p. 462.
3. H.M. Shalaby, EM. Al-Kharafi, A.J. Said, Brit. Corros. J. 25 (1990): p. 292.
4. E.D. Mor, A.M. Beccaria, Corrosion 31 (1975): p. 275.
5. S.M. Wilhelm, Y. Tanizawa, C.-Y: Liu, N. Hackerman, Corros. Sci. 22 (1982):

p. 791.
6. A.M. Beccaria, G. Poggi, Analyst 111 (1975): p. 275.
7. H.C. Albaya, J.B. Bessone, P.M. Taberner, O.A. Cobo, Corrosion 30 (1974):
p. 437.
8. H.S. Campbell, BNF Metals Technology Centre Miscellaneous Publication,
MP574, August 1972.
9. J.E. Castle, D.C. Epler, Surf: Sci. 53 (1975): p. 286.
10. D.C. Epler, J.E. Castle, Corrosion 35 (1979): p. 451.
11. J.E. Castle, M.S. Parvisi, “Minor Alloying Elements and the Marine Corrosion of
Copper Alloys,” International Colloquium on Choice of Material for Condenser
Tubes and Plates and Tube and Tightness Testing (Avignon, France: SFEN, 1982).
12. Z.B. Luklinska, J.E. Castle, Corros. Sci. 23 (1983): p. 1163.
13. J.E. Castle, M.S. Parvisi, Corros. Prev. Control (February 1986): p. 5.
14. J.E. Castle, D.C. Epler, “The Development of Inorganic and Biological Fouling
Layers on Copper Based Alloys,” Progress in the Fouling of Industrial Plant
Conference (Nottingham, U.K.: Institute of Corrosion, 1981).
15. J.E. Castle, M.S. Parvisi, A.H.L. Chamberlain, Microbial Corrosion, Book 303
(MET SOC, 1983), p. 36.
16. F.J. Ijsseling, J.M. Kroughman, L.J.P. Drolenga, “The Corrosion Behaviour of the
System CuNilOFe/Seawater: The Protective Layer of Corrosion Products,” Fifth
International Congress on Marine Corrosion and Fouling (Barcelona, Spain:
1980), p. 146.
17. L.J.P. Drolenga, F.J. Ijsseling, B.H. Kolster, Werkst. Korros. 34 (1983): p. 167.
18. C. Kato, B.G. Ateya, J.E. Castle, H.W. Pickering, J. Electrochem. SOC.127 (1980):
pp. 1890, 1897.
19. C. Kato, H.W. Pickering, J. Electrochem. SOC. 131 (1984): p. 1219.
20. F.J. Ijsseling, L.J.P. Drolenga, B.H. Kolster, Brit. Corros. J. 17 (1982): p. 162.
21. R.G. Blundy, M.J. Pryor, Corros. Sci. 12 (1972): p. 65.
22. G.E. McGuire, A.L. Bacarella, J.C. Griess, R.E. Clawing, L.D. Hulett,
J. Electrochem. SOC.125 (1978): p. 1801.
23. K.D. Efird, Corrosion 33 (1977): p. 347.
24. R. Francis, Brit. Corros. J. 18 (1983): p. 35.
25. R.W. Ross, CORROSION/77, paper no. 94 (Houston, TX: NACE International,
1977).
26. H.S. Campbell, Aluminium Bronze Alloys Corrosion Resistance Guide,
Publication 80 (Potters Bar, U.K.: Copper Development Association, 1981).
27. A. Schussler, H.E. Exner, Corros. Sci. 34 (1993):p. 1793.

28. B.G. Ateya, E.A. Ashour, S.M. Sayed, Corrosion 50 (1994):p. 20.
29. R. Francis, C.R. Maselkowski, BNF Metals Technology Centre Report, R506,
May 1985.
30. J.E. Castle, University of Surrey, correspondence to author, 1985.
31. T. Atsumi, K.Sudo, K. Nagata, Sunzitonzo Light Met. Tech. Rep. 26 ( 1985): p. 13.
32. T.W. Bostwick, Corrosion 17 (1961): p. 12.
33. A.M. Lockhart, Proc. Inst. Mech. Eng. 179 (1964-1965): p. 459.
34.R. Francis, INCRA Project, 289,January 1979.
35. S. Sato, M.Okawa, Sumitorno Light Met. Tech. Rep. 17(1976):p. 17.
36. S. Sato, T. Nosetani, Y . Yamaguchi, K. Onda, Sumitorno Light Met. Tech. Rep. 16
(1975): p. 23.
37. R. Francis, M P 21 (1 982): p. 44.
38. E.B. Shone, G.C. Grimm, Trans. Inst. Marine Eng. 98 (1989,Paper 11.
39. P.H. Effertz, W.Fichte, VGB Kruftswerktech. 57 (1977): p. 1 16.
40.F. Inui, Bull. MESJ 1 (1973):p. 21.
41.R. Francis, Corros. Sci. 26 (1986):p. 205.
42. B.C. Syrett, Corros. Sci. 21 (1981): p. 187.
43.L.E. Eiselstein, B.C. Syrett, S.S. Wing, R.D.Caligiuri, Corrm Sci. 23 (1983):
p. 223.
44.C. Kato, H.W. Pickering, J.E. Castle, J. Electrochem. Soc. 13I (1984):p. 1225.
45. A. Schussler, H.E. Exner, Corros. Sci. 34 (1993):p. 1803.
CHAPTER 4
General Corrosion
General corrosion involves more or less uniform metal dissolution from a

surface. The rate at which this occurs varies with pH, temperature, dissolved
oxygen (DO), and the presence of aggressive ions such as halides and
sulfides etc. In the following section, data have been collected from a
variety of common environments found in service, including the effect of
environmental variables.
4.1 Natural Waters
4.1.1 Seawater and Fresh Water

Seawater typically contains about 19,000 mglL chloride, but this may vary
from 5,000 to 10,000 mg/L when dilution with fresh water occurs and may
rise up to 26,000 mg/L when concentrated by evaporation (for example,
the Arabian Gulf). The corrosion rates of some common copper alloys in
seawater from several different sources are shown in Table 4.1 ,I-' in which
most of the data was collected at local ambient temperature (10" to 25"C),
but that in reference 5 was obtained at 40°C, hence the slightly higher
corrosion rates compared with the other references.
All the data in Table 4.1 show the low general corrosion rate of
copper alloys in seawater. The corrosion rate of pure copper is some-
what higher but still does not exceed 0.1 mm/y under the most aggressive
conditions.
Efird and Anderson carried out exposure tests on 90/10 and 70l30
copper-nickel for up to 14 years, under quiet, flowing (0.6 mls), and tidal
33
TABLE 4.1 General corrosion rate of some common copper alloys in seawater
at ambient temperature(A)
Corrosion rate (um/v)
Alloy Ref. 1 Ref. 2 Ref. 3 Ref. 4 Ref. 5 Ref. 6 Ref. 7
Copper - 15-100 10-40 50-100 - - 40
Brass - 10-30 8-13 2-7 50-300 20 50
90/10Cu-Ni t25 3-15 12 - 70 18 40
70/30Cu-Ni <25 3-15 6 - 40 25 25
NAB 55 25-50 5-10 1.3-2 - - 60
Gunmetal 18 25-50 - - - - 2540
Alloy400 t25 t25 - - 8 13 25
(A)
Most of these data were collected at local ambient temperatures ( 1 O-ZS'C), but the data from
reference 5 were obtained at 40°C (hence the slightly higher corrosion rates compared with the
other references). References 1 4 are data as collected, whereas reference 7 is design data taken
from the U.K. Ministry of Defence Guidelinesfor Naval Service. The higher numbers in reference
7 are probably because of the high tidal water velocity at the U.K. test site that was used to generate
data for the guidelines.
conditions.' The metal loss with time is shown for 90/10 and 70/30 copper-
nickel in Figures 4.1A and 4.1 B, respectively. Although the initial rate of
corrosion was highest, it was still only about 12 pm/y for both alloys-
not a high value. After 14 years, the rate for 90/10 copper-nickel in all
three environments was the same, at 1.3 pm/y. For 70/30 copper-nickel,
the rate was 0.8 to 1.9 pm/y. These are all very low figures, and the
data show that the film becomes more protective with time, with steady
state not being reached for several years. The results also show that the
films form most slowly under flowing conditions. Over the 14-year expo-
sure, none of the samples of either alloy showed any significant localized
corrosion.
As the temperature increases, so does the corrosion rate of copper
alloys, but not dramatically. Aluminum brass and copper-nickel alloys have
been used successfully for many years in seawater-cooled heat exchangers,
where discharge temperatures of aerated seawater may be up to 50 or 60°C.
Data have been published for higher temperatures, but these are mostly at
reduced DO levels, as are found in multistage flash (MSF) and multiple
effect distillation (MED) desalination plants. Table 4.2 shows data from
May et al. at 41 and 86°C. The corrosion rates at 41°C were generally
higher than those at ambient temperature (Table 4. I), except for Alloy 400.
The value for admiralty brass at 41°C seems anomalously high, but other
authors also report increased corrosion rates for this alloy in warm seawater.
Increasing the temperature from 41 to 86°C has little effect on the corrosion
60 i
0 4 ’ . , . . . . r . . . , . . .
0 2 4 6 8 10 12 14
Time (years)
1-4- Quiet +Flowing *Tidal I

(A)
0 2 4 6 8 10 12 14
Time (years)
I--+- Quiet +Flowing -f- Tidal I

(B)
FIGURE 4.1 Effect of time on the corrosion of copper-nickel alloys in ambient tempera-
ture seawater: (A) 90/10 copper-nickel; (B)70/30copper-nickel.8
TABLE 4.2 Effect of temperature on the

corrosion of various copper alloys in fully
aerated seawater5
Corrosion rate (pm/y)
Alloy 41°C 86°C
Admiralty brass 305 48
Aluminum brass 48 64
90/10 Cu-Ni 71 30
70/30 Cu-Ni 38 36
Alloy 400 8 10
rate, except for admiralty brass, whose rate reduces to levels similar to
the other copper alloys. The reason for this is probably that although the
kinetics of corrosion will be more rapid at higher temperatures, the reduced
oxygen concentration at 86°C will slow down the cathodic reaction, which
is the reduction of DO.
Copper alloys are widely used in MSF and MED desalination
plants, where temperatures can be high and oxygen concentrations are
low. Schreiber et al. showed the effect of DO concentration at 107°C
on copper and copper-nickel alloys (Figure 4.2).9 The corrosion rates
0.5
I
0 10 20 30 40 50 60 70 80 90
Dissolved Oxygen (ppb)
I +Copper -W- 90/10 Copper-Nickel +70/30Copper-Nickel I

FIGURE 4.2 Effect of DO on the corrosion of copper and some alloys in ~ e a w a t e r . ~
General Corrosion 37
2 015 Temp. - 104°C -

C pH - neutral
.-0In Time - 9Odays
0 50 100 150 200

Dissolved oxygen (ppb)
Al Brass -90/10 Copper-Nickel __ 70/30Copper-Nickel
FIGURE 4.3 Effect of DO on the corrosion of some copper alloys.'0
are high compared with those of Oldfield et al., at 104°C (Figure 4.3)."
The difference may be due to the longer exposure time used by Oldfield
et al. as corrosion rates decrease with time, or it may be that the veloc-
ity in Schreiber et al.'s tests was higher, which also increases corrosion
rates.
Figure 4.4 shows the effect of temperature on the corrosion rates of
copper alloys with 200 ppb DO; the corrosion rates for the copper-nickel
alloys are very low at all temperatures. Figure 4.5 shows the effect of
temperature measured in a plant in Texas, USA.6 The results show the
opposite effect to that shown in Figure 4.4, with corrosion rates decreasing
with increasing temperatures. The reason for this is the variable oxygen
concentration in a real plant, with the highest oxygen content at the lowest
temperatures, and vice versa, whereas the data in Figure 4.4 are at a constant
oxygen concentration.
Ross investigated the effect of water composition, pH, and tem-
perature on the corrosion of copper-nickel alloys over a range of DO
concentrations." Ross found that there was no indication of protec-
tive film formation on copper-nickel alloys at 32°C and pH < 6 with
0 2 >500 ppb. Reducing the DO to 100 ppb or increasing the pH >6 im-
proved the formation of protective films. This is explained by the Pourbaix
-.- I
0.25 l O x y g e n - 200ppb
pH - neutral /‘
-2 0.2
Time
I
- 90days
I 1
‘
/’
--E
a
m
p! 0.15
c
.-0m
g
5 0.1
0
0.05
0
70 75 80 85 90 95 100 1G5 110
Temperature (“C)
~ Al Brass -90/10 copper-nickel-70/30 copper-nickel I

FIGURE 4.4 Effect of temperature on the corrosion of some copper alloys.10
0.2
Oxygen - 4 to 600 ppb

0.15 PH - 6.2 to 7.8
h Vel. - low
2
E
Time - 156days
E
Y
m
2
c
0.1
c
.-u)
e
5
0
0.05
C
50 60 70 80 90 100 110 !O
Temperature (“C)
I -t- Al-Brass -D- Comer-Nickel I

FIGURE 4.5 Effect of temperature on the corrosion of copper alloys.”
CUO
0.4 cue*-
%
2 0.2
P
--g2
u)
o
-0.2
C
m
c
2 -0.4 \
-0.6
-0.8
-1
0 2 4 6 8 10 12 14 16
PH
FIGURE 4.6 Pourbaix diagram for seawater (copper-chloride-water system).'*
diagram for copper in seawater (Figure 4.6).12 This shows that cuprous
oxide (Cu20) is only stable down to pH 5 at normal potentials, but this is
-
under equilibrium conditions. Under flowing conditions, Ross's observation
that protective films did not form below pH 6 is not surprising, and copper
and its alloys are not normally recommended for continuous service in fresh
water or seawater at pH < 6.
Ross also found a synergistic effect between pH, DO, and total al-
kalinity (bicarbonate) on the corrosion of copper-nickel alloys at elevated
temperatures in the range 54 to 77°C.'' At pH 27.5, increasing oxygen con-
centration (> 100 ppb) accelerated corrosion when the bicarbonate was low
(25-50 mg/L). When the bicarbonate was 65-90 mg/L and the pH 2 7.5,
corrosion was inhibited regardless of the oxygen concentration. At higher
seawater temperatures (107"C), protective film formation appeared com-
plete, irrespective of bicarbonate and DO concentration, when the pH 2 7.5.
In fresh waters, the lower chloride content, typically <200 mg/L,
means that the corrosion rate of copper and its alloys is even lower than
for seawater. Corrosion rates of 1 to 10 p d y are typical of copper in fresh
water at temperatures from 10 to 60°C after a year or more.
Vik et al. reported the results of a monitoring exercise on copper pipes

in Oslo, Norway.13 The supply water is a soft, surface-derived type with the
following characteristics: pH 5.0 to 7.0; total alkalinity 3 to 30 mg/L (as
CaC03); calcium 1 to 5 mg/L. This is clearly an aggressive water, in which
the formation of a protective film on copper tubes will be difficult. Over a
period of 46 weeks, the corrosion rate at both 0.17 and 0.6 m/s was 76 f
26 p d y . While higher than normally seen in fresh waters, the corrosion
rate was still not excessive.
4.1.2 Cuprosolvency
The slight general dissolution of copper that takes place in some fresh
waters has little effect upon the copper pipes or tanks but may affect water
quality. Water containing 1.5 mg/L or more copper can cause green stains
on sanitary fittings. Smaller concentrations of copper can cause accelerated
corrosion or pitting of other metals downstream of copper pipes. As little
as 0.1 mg/L copper is sufficient to increase the corrosion rate of galvanized
steel, and higher concentrations can cause pitting, particularly in hot waters.
Francis describes such a case from a block of flats in Denmark.I4 For this
reason, it is always recommended that galvanized steel be used upstream
of copper pipes, but never downstream. Copper concentrations as low as
0.02 mg/L can cause pitting of aluminum in some waters.15
Normally, the copper concentration drops rapidly in the first few days
as a protective film forms on the copper surfaces. However, in some waters,
high concentrations of copper can persist. These waters are generally soft
and have a low pH (associated with a high free carbon dioxide [CO2]
content). Cuprosolvency is more of a problem in hot waters as copper
dissolves more readily at higher temperatures; however, some hard waters
that are cuprosolvent when cold may be less of a problem when hot because
the precipitation of calcium carbonate scale reduces copper dissolution.
It has been suggested that under some conditions, bicarbonate in fresh
waters can increase cuprosolvency, but this has never been verified by other
workers.16 Figure 4.7 shows the appearance of a typical copper pipe in
a cuprosolvent water. The tube has a thick green deposit that is loosely
adherent, with little copper oxide beneath it.
A number of authors have reported the conditions that cause cuprosol-
vency. Gilbert found that all the cuprosolvent waters that he investigated had
a pH from 4.9 to 6.9. I 7 Nuttall stated that cuprosolvent waters have a pH < 6,
while the EEC mandatory limits for potable water are 6.5 < pH < 9.5."
Hence, waters within these limits would not be expected to cause
- .-
FIGURE 4.7 Appearance of films on copper tube exposed in a cuprosolvent water. (See
color section.)
cuprosolvency. Acidic soft waters, whose pH has been increased, say, by

caustic soda additions, are not buffered, and some slight corrosion can
cause the local pH to drop to low values, allowing the water to become
cuprosolvent in those regions.
However, there are other ways that high copper levels can be pro-
duced. Long periods of stagnation, particularly when piping is new, have
been observed to result in poor film formation and high copper levels.''
Maselkowski reviewed reported cases of cuprosolvency, including several
for which there had been long periods of stagnation after a new piping
system had been filled.20These problems are usually associated with large
buildings that have long pipe spurs that are used infrequently and involve
soft waters of low pH.
The levels of copper that are permitted in waters vary; for instance, in
the United Kingdom, the permitted level is 2 mg/L. It is important to realize
that copper is essential for human health; the World Health Organization
recommends a daily intake of 2.0 to 3.0 mg of copper. Gilbert found levels
of 2.9 to 5.3 mg/L of copper in water that had been standing 0~ernight.l~ In
Norway, good water has 5 0.1 mg/L copper when flowing, while 0.2 to 0.3
mg/L is acceptable, but copper >0.3 mg/L is unacceptable. For standing
water, copper concentrations > 1.O mg/L are unacceptable. Figure 4.8 shows
the copper concentration monitored by Vik et al. in Oslo tap water over a pe-
riod of 42 weeks from 1987 to 1988.13The pH varied from 5.0 to 7.0, and the
total alkalinity varied from 3 to 30 mg/L (as calcium carbonate [CaCO,]).
It can be seen that the flowing water was acceptable most of the time, but
the standing water was unacceptable most of the time. This demonstrates
the increasing risk of cuprosolvency in acidic, stagnant water.
P
6 0.6-
0.4-
.o.
Not Acceptable Water
\. o-
6
9. 0 -0.
-. - - o. + .
0 0. * * , Acceptable Water ,
0.2- o--..
0 ‘o‘ ‘0.0- 0 -0
Good Water
07
Time (weeks)
- -0 - Standing Water Flowing Water
FIGURE 4.8 Copper concentration in Oslo tap water during 1987-1988.13
A similar problem is often referred to as blue water. This is caused

when an insoluble copper corrosion product, believed to be a form of copper
hydroxide (Cu[OH]2), is formed but does not deposit on the tube walls. It
is carried along in water, giving it a blue or bluish green color. Oliphant
lists some of the causes of blue waterJ9:
1. Waters with bicarbonate contents t 2 0 mg/L (as CaC03), provided

that other factors are present
2. Waters with pH > 9, where copper hydroxide is more stable and is
more likely to be formed
3. Microbiological films that inhibit access of DO to the metal surface
and prevent the formation of copper oxide
4. Water stagnation during the early Iife of piping
Critchley et al. compared the dissolution of copper from pipes exposed to

sterile water with pipes exposed to sterile water to which bacteria from a
problem water had been added.21Much higher rates of copper dissolution
occurred on tubes exposed to the water with bacteria. This suggests that
bacteria can contribute to the problem of blue water and that its causes need
not be solely chemical, as already mentioned by Oliphant.
FIGURE 4.9 Appearance of a copper tube carrying so-called blue water. (Photograph
courtesy of P. Munn.) (See color section.)
Munn noted that blue water can sometimes occur with waters that
would not normally be thought to cause this problem.22 He detected
large quantities of phosphorous in the blue-green deposit on the tubes-
presumably copper phosphate. A typical example is shown in Figure 4.9.
These waters are being dosed with phosphates to reduce plumbosolvency
(see next section), although the waters are not obviously at high risk to
cause this. It appears that the presence of phosphates interferes with the
formation of protective films. Sometimes the phosphate corrosion products
can cause significant blockage of the pipe.22
4.1.3 Rosette Corrosion

Although this is really a type of general c o r r o ~ i o n , it
' ~ has some of the
appearances of pitting and is associated with the use of aluminum protector
rods to prevent Type I pitting of copper water cylinders and tanks. Hence,
the topic is discussed in more detail in Chapter 5.
4.1.4 Plumbosolvency
Copper alloys are frequently used for fittings in potable water systems; the
most commonly used ones are forging brasses (e.g., C38000 and CZ132)
and LG2 gunmetal (C83600). All these alloys contain lead (2 to 5%) to
improve machinability and to produce pressure-tight fittings. The concern

is that in some waters, unacceptable quantities of lead may leach from
fittings.
Campbell discussed the water compositions that are likely to lead to
high dissolution rates of lead.15 Hard waters, and those that produce scale
(positive Langelier Index), are the least likely to cause leaching of lead,
whereas waters with high nitrate and/or chloride contents lead to increased
dissolution. The worst waters for lead dissolution are soft waters of low
pH or nonscaling waters with a high dissolved C02 content. The quality of
these waters can be improved by additions of lime (Ca[OH]2).
Paige and Covino investigated lead dissolution in distilled water under
static conditions as a function of both time and temperature for a number of
commonly used copper alloys.23They found that most of the lead dissolved
in the first 24 hours, and it exceeded the U.S. limit of 15 pg/L under all con-
ditions. This is hardly surprising given the aggressive nature of distilled wa-
ter toward lead dissolution, as defined by Campbell.” They also found that a
pretreatment in acetic acid (CH3COOH) to remove lead smeared by machin-
ing did not significantly affect lead dissolution. Increasing the water tem-
perature either had no effect on lead dissolution or reduced it only slightly.
Tests by Oliphant showed that there was no correlation between the
lead content of the alloy and lead release into natural, fresh waters.24
Oliphant also reported test results to the draft standard DD 256:2002 for
a range of leaded alloys.’’ A round-robin evaluation of this test method
found wide variation in the results obtained from laboratory to laboratory,
and there is currently no European standard test for lead release from copper
alloy fittings.
The limits on lead release in the United Kingdom were 50 pg/L up to
2003, when the limit was reduced to 25 pg/L. From 2013, the limit is to be
reduced to 10 WgL. Oliphant found that all the alloys met the 25 pg/L limit,
but compliance with the 10 pg/L limit might be difficult for some alloys.’’
4.1.5 Disinfection
When potable water systems are installed in large public buildings, it is
common to disinfect the tubes with a strong solution of sodium hypochlo-
rite (NaOCl), at anywhere from 20 to 100 mg/L free chlorine. Francis and
Saxton investigated the effect of strong disinfection regimes on the subse-
quent performance of copper tubes.25 Tests for up to 30 days under both
stagnant and flowing conditions (up to 3 d s ) showed no effect of a pretreat-
ment of 100 mg/L chlorine for 2 hours or 20 mg/L chlorine for 16 hours.
The tests included both a hard well water and a soft, surface-derived water
at room temperature and at 60°C. Under continuous chlorine dosing at 1
mg/L, there was no effect on copper corrosion at room temperature under
any flow condition. At 60°C, there was a marked increase in general corro-
sion rates in both waters, an increase in the susceptibility to Type I1 pitting
under slow flows in the soft water, and an increased susceptibility to erosion
corrosion at high flow velocities. The effect of chlorine on Type I1 pitting
is discussed in more detail in Chapter 5.
Bond et al. reported the results from tests lasting up to 12 months
using the same disinfection regimes, flow rates of 0.1 and 2 m/s, and
ambient and 60°C temperatures.26 After 12 months, the corrosion rates of
all the tubes were in the range 1 to 3 pdy-a very low rate. The corrosion
rates after 4 months were somewhat higher, but all were (0.01 mm/y-still
a low corrosion rate. Type I1 pitting was not seen in the hot soft water with
continuous chlorine dosing because the soft water used was different from
the water used in the 30-day tests and did not support Type I1 pitting. The
conclusion was that disinfection of copper tubes under these regimes had no
detrimental effect on the performance of copper tubes in subsequent service.
4.2 Acids
The data in this section are taken from Rabald,27 and those seeking more
detailed data or corrosion rates in acids other than those mentioned here are
urged to consult Rabald’s book.
4.2.1 Sulfuric Acid

Sulfuric acid (H2S04) is the most commonly used of the mineral acids,
and copper has moderate resistance to dilute acid, but the corrosion rate
increases when the solution is aerated or oxidizers are present, and also
when the temperature is increased. The following example in 6% sulfuric
acid shows the effect of aeration:
Corrosion rate (mm/y)
Temperature
(“C) Static Aerated
60 0.45 1.18
80 0.61 1.51
90 0.61 1.67
Brass is not recommended for use in sulfuric acid under any conditions,
but tin bronzes (without zinc) have been used at room temperature from
-
0.125 0.25 mm/y
0 10 20 30 40 50 60 70 80 90 100
Acid Concentration (wt%)
FIGURE 4.10 Corrosion rate of nickel aluminum bronze in sulfuric acid.27
1% to 80% acid. The corrosion rate varies from 0.04 mm/y (deaerated) to
0.4 mm/y (aerated).
Aluminum bronze has good resistance from 20 to 90°C up to 80%
acid concentration. Corrosion rates are in the range 0.08 to 0.14 mm/y, but
this increases under strong aeration. Under nonoxidizing conditions, nickel
aluminum bronze (C63200) has good resistance to sulfuric acid over a wide
range of temperatures and acid concentrations: as shown in Figure 4.10.
The copper-nickel alloys are inferior in performance to both NAB and tin
bronzes in sulfuric acid.
Alloy 400 has some uses in sulfuric acid, particularly at room tem-
perature, when air is not present, as shown in Table 4.3. The corrosion rate
increases in the presence of air, particularly at elevated temperatures.
4.2.2 Hydrochloric Acid

In acid concentrations less than 30%, copper corrodes in the range 0.08
to 0.85 m d y , provided that the solution is free of air or other oxidizers.
With oxidizers present or with stronger acid, the corrosion rate increases
substantially.
Brass should not be used in hydrochloric acid (HCl) under any con-
ditions, but tin bronzes corrode at 0.33 to 2.5 mm/y in dilute acid in the
TABLE 4.3 Corrosion of Alloy 400 in sulfuric acid2’

Acid
concentration (%) Temperature (“C) Deaerated Aerated
2 20 0.12 0.80
2 20 0.19 1 .0
10 20 0.15 0.90
30 20 0.10 0.4 1
50 20 0.09 0.30
70 20 0.05 0.28
80 20 0.02 0.27
90 20 2.8 3.9
5 50 0.38 1.7
5 80 0.37 2.9
5 90 0.20 2.5
5 100 0.05 0.4
25 260 Unsuitable Unsuitable
temperature range 15 to 100°C. Aluminum bronze is similar to tin bronze,

and oxidizers greatly affect corrosion rates. The following data are from
Rabald27:
Temperature Acid concentration Corrosion rate

(“C) (wt%) (mm/Y) Comment
20 3.6 0.18 -
100 15 2.29 -
100 30 0.57 -
20 30 0.57 -
20 30 1.71 0.8% chlorine
20 30 4.70 1 % FeCll
This indicates that aluminum bronze would not normally be suitable for
service in HC1, but it has been used for hooks in pickling/descaling baths
for steel, where the galvanic couple with steel prevents corrosion.
The copper-nickel alloys corrode at about 0.3 mm/y, provided that
there is no aeration or oxidizers. Alloy 400 has some uses in hydrochloric
acid, but corrosion rates increase dramatically in the presence of oxidizers
(e.g., oxygen, iron, copper). Table 4.4 shows some typical corrosion rates.
4.2.3 Nitric Acid

Copper and its alloys have unacceptably high corrosion rates in nitric acid
(HN03) under all conditions, as does Alloy 400, and should not be used.
TABLE 4.4 Corrosion of Alloy 400 in hydrochloric acidz7

Acid Temperature Corrosion
concentration (%) ("C) rate (mmly) Comments
0.5 100 0.73 Unaerated
1.o 30 0.08 Deaerated
1.o 70 0.13 -
2.0 27-50 1.80
3.6 60 0.90 Unaerated
5.0 30 0.18 Deaerated
5.0 30 2.10 Aerated
5.0 70 1.20 Deaerated
5.0 70 1.30
30.0 30 6.7 Aerated
4.2.4 HydrofluoricAcid
In liquid hydrofluoric acid (HF) at 15 to 20"C, copper corrodes at -0.5
m d y , but this increases under agitation. Brasses perform similarly to
copper.
Tin bronzes corrode at 0.5 m d y in liquid HF at 15 to 40"C, rising to
1.5 m d y at 80 to 90°C. The aluminum bronzes have low corrosion rates in
HF, for example, 0.1 1 m d y in 40% HF at 20°C. This corrosion resistance
has resulted in the use of nickel aluminum bronze for the nozzles that spray
HF to produce the frosting of light bulbs and also for the acid collection
trays beneath the nozzle^.^
The 70/30 copper-nickel alloy also has good resistance to HF. In
liquid HF, the corrosion rate varies from 0.18 to 0.25 r n d y in the tem-
perature range 15 to 90°C. In 98% HF at 38"C, the corrosion rate is 0.05
mdy.
Alloy 400 is, in general, very resistant to attack by HF, but its corrosion
rate increases when oxidizers or reducing sulfur compounds are present.
Following are some typical corrosion rates in pure HF:
Acid concentration Temperature Corrosion rate

("/.I ("C) (mm/Y)
30 21 0.203
48 21 0.003
70 21 0.003
93 21 0.152
100 150 0.076
TABLE 4.5 Corrosion of nickel aluminium bronze and 15%

copper-nickel in phosphoric acid *’
Acid NAB 15% Cu-Ni

concentration
(”/.) 15°C 50°C 75°C 15°C 50°C 75°C
20 0.06 0.10 0.25 0.61 0.17 0.58
40 - - - 0.03 - 0.11
60 0.01 0.01 0.00 0.01 - 0.03
4.2.5 Phosphoric Acid

As with other mineral acids, the corrosion rate of copper is moderate at low
acid concentrations and temperatures. The presence of oxidizers greatly
accelerates attack. The brasses perform similarly to copper, while the tin
bronzes are more corrosion-resistant. For example, bronze with 8.5% tin
corrodes in 80% acid at 0.1 mm/y at 60”C, at 0.21 r n d y at 110°C, and at
0.6 m d y at 120°C.
Nickel aluminum bronze is very resistant to corrosion in phosphoric
acid, whereas copper-nickel alloys are somewhat less resistant, as shown in
Table 4.5.27
4.2.6 Organic Acids

Copper can have low corrosion rates in acetic acid, provided that air and
oxidizers are excluded. Following are some typical corrosion rates at 20°C:
Acid concentration
(%) Deaerated Aerated
6 0.03 0.58
50 0.08 1.84
99 0.05 0.05
The corrosion rate also varies with temperature, as shown in the following,
for solutions in contact with air:
Acid concentration
(”/) 25°C Boiling
20 0.40 1.67
60 0.14 0.3 1
100 0.20 6.74
Brass can also be acceptable in acetic acid if air and oxidizers are excluded.
The bronzes have low corrosion rates in acetic acid, with tin bronze cor-
roding at 0.08 mm/y in 33% acid at the boiling point and aluminum bronze
corroding at 0.06 mm/y under the same conditions.
Alloy 400 has good resistance to acetic acid at 20°C at all concentra-
tions. The corrosion rate is 0.08 mm/y when little air is present, rising to
0.6 mm/y with strong aeration and agitation. The greatest corrosion rate
occurs in 80% acetic acid. At elevated temperatures, corrosion rates can
rise up to 0.61 mm/y.
In formic acid (HCOOH),the performance of copper and its alloys is
similar to its performance in acetic acid, with high corrosion rates when air
or oxidizers are present. Alloy 400 is less sensitive to the presence of air
than other copper alloys, but the corrosion resistance is also similar to the
corrosion resistance in acetic acid.
In 50% citric acid (C6Hg07.H20), the corrosion rate of copper in-
creases from 0.57 mm/y at 20°C to 0.69 mm/y at 70°C in aerated solution.
When deaerated, the corrosion rates are 0.10 and 0.12 mm/y, respectively.
Brass performs similarly to copper in citric acid, as do the tin bronzes,
although the latter are less sensitive to aeration.
Aluminum bronze is very resistant to citric acid, with a corrosion rate
of -0.03 mm/y in 33% to 36% acid from 20°C to the boiling point; 70/30
copper-nickel also has a low corrosion rate in all strengths of citric acid,
typically, 0.04 mm/y at room temperature and 0.18 mm/y at 60°C.
4.3 Alkalis
The most common alkali is sodium hydroxide (NaOH) (caustic soda), and
copper corrodes at modest rates from room temperature to 200°C up to
75% concentration, provided that air is excluded and there are no oxidizers.
Brasses have been used at room temperature in up to 33% caustic soda
with corrosion rates 10.13 mm/y. Tin bronzes are somewhat better than the
brasses under the same conditions, with a corrosion rate of S O . 10 mm/y.
Aluminum bronzes are satisfactory in dilute solutions at around room
temperature, but they corrode rapidly at elevated temperatures or in strong
solutions; 70/30 copper-nickel has been used at up to 75% concentration
from 2 to 175°C.In 65% to 75% caustic, the corrosion rate is 0.10 mm/y at
150 to 175°C.
Alloy 400 has excellent resistance to caustic soda and is frequently
used for numerous components in its manufacture. Figure 4.1 1 shows the
isocorrosion curve (0.1 mm/y) for Type 3 16L stainless steel (UNS 53 1603),
200
180
160
140
60
40
I
0 10 20 30 40 50 60 70
Caustic Concentration (wt%)
FIGURE 4.11 lsocorrosion curve (0.1 mm/y) for 316LSS and alloy 400 in caustic soda.
and it can be seen that Alloy 400 can be used up to the boiling point, even
in strong solutions.
The performance of copper and its alloys in calcium hydroxide
(Ca[OH]z) is very similar to its performance in sodium hydroxide.
Ammonia (NH3) is very alkaline and forms cuproammonium com-
plexes, which tend to accelerate corrosion rates. Copper is not suitable for
use in ammonia, and brasses are liable to suffer stress corrosion cracking
(SCC; see Chapter 10). The tin bronzes are much more resistant to SCC;
they corrode at -0.53 r n d y in 5% aerated ammonia at room temperature.
Aluminum bronzes are not so resistant as the tin bronzes, with a corrosion
rate of 0.82 r n d y under the same conditions.
Ammonia was once used as a method of pH control in boiler feed water
in power stations, and it would carry over with the steam into the condensers.
This caused SCC of brass condenser tubes in some power stations, but
copper-nickel tubes performed much better. Caruso and Michels carried
out tests to study the effect of ammonia on some common condenser tube
alloys.’* In the fog test, the alloys were exposed to an ammoniacal mist that
condensed onto tubes through which cooling water at 12°C was passed. The
mist was produced from distilled water containing 1,000 mg/L ammonia at
TABLE 4.6 Corrosion in ammonia fog and spray tests

for some copper alloy condenser tubes 28
Average corrosion rate (mmly)
Alloy Fog test Spray test

Phosphorized copper 0.17 > 10
Admiralty brass 0.28 13.2
Aluminum brass 0.10 4.3
Al-bronze (C60800) 0.18 NT(~)
90/10 Cu-Ni 0.10 0.05
70/30 Cu-Ni 0.01 t0.01
Alloy 400 0.01 <0.01
(A)
NT,not tested
30°C, and the test lasted 100 days. In the spray test, a solution containing
2,000 mg/L ammonia was sprayed onto tubes at 21°C so that it cascaded
over several tubes mounted one above the other. The test lasted 170 days,
and the results of both tests are shown in Table 4.6. It can be seen that
while the corrosion rates of copper, the brasses, and aluminum bronze were
high, the copper-nickel alloys performed better, with 70/30 copper-nickel
and Alloy 400 performing the best.
Rabald reported the corrosion rate of Alloy 400 in 60% ammonia
at 20°C as 0.2 mm/y without air, rising to 0.6 mm/y with aeration and
agitation.27
4.4 Other Chemicals

The comments given in this section are only general; the reader requiring
data for a specific compound is urged to consult Rabald's
Corrosion rates in neutral salts (6< pH<8) are generally low (the spe-
cific example of chlorides was discussed previously). Less aggressive salts,
such as sulfates and nitrates, cause lower corrosion rates than chlorides.
Metal salts with a low pH, such as iron, copper, and manganese, pro-
duce higher corrosion rates, particularly if the anion is chloride; however,
the dissolution rate will be higher when the pH is below the stability range
for cuprous oxide-typically, pH 5.0-5.5. However, these corrosion rates
need not be excessive if the temperature is close to ambient, concentrations
are low, and there is little agitation. Remember that the commercial ver-
sions of these salts often contain some free acid, which makes them more
aggressive than the pure salt.
Alkaline salts, such as sodium silicate (Na2Si03), phosphate

(Na2P04), and carbonate (Na2C03), are not as aggressive as the strong
alkalis, such as sodium and calcium hydroxide. Copper and its alloys have
acceptable corrosion resistance over a wide range of concentrations and
temperatures. Note that the presence of oxidizers tends to increase cor-
rosion rates, although not necessarily to an unacceptable level. Testing in
specific environments is advised, especially when more than one compound
is present.
The other important group of salts are those that can form complexes
with copper such as the acetates, citrates, and tartrates. General corrosion
in solutions of these salts is not usually very high, but over certain potential
ranges, it is possible to get SCC, particularly with brasses. Tin bronzes and
the copper-nickel alloys are much more resistant to SCC in these environ-
ments; susceptibility to SCC is discussed in more detail in Chapter 10.
4.5 Methods of Prevention
Low levels of corrosion in fresh water and seawater can be maintained,

provided that flow velocities are low enough to avoid erosion corrosion
(see Chapter 7) and pollutants are not present (see Chapter 3). These low
corrosion rates are maintained at elevated temperatures, provided that the
pH is >6.
Where cuprosolvency is a problem, water treatment to increase the
pH and bicarbonate content will significantly improve matters. Where the
problem has been exacerbated by a long period of stagnation soon after
first filling, methods need to be adopted to encourage the formation of
protective cuprous oxide. Maselkowski described a case of cuprosolvency
examined by BNF Metals Technology Centre at a school in Torquay, in the
United Kingdom.20 The copper pipes had been left full of stagnant water
for some time before the school was occupied. Copper levels were mainly
in the range of 3 to 4 mg/L, with excursions to 6 to 8 mg/L. The problem
was solved by allowing water to run slowly but continuously through the
system, particularly long spurs, for about 3 weeks. Thereafter, the copper
concentration in the water returned to more normal levels.
Problems of blue water also require water treatment to encourage the
formation of protective oxide films. If microbial action is also contributing
to this problem, the water treatment will be more complicated and may
involve biocides. Shock treatments with sodium hypochlorite (NaOC1) did
not hinder protective film formation on copper tubes, and the oxidizing
power of hypochlorite should discourage bacterial action, while encourag-

ing oxide f ~ r m a t i o n . * ~ * * ~
Where plumbosolvency is a problem, water treatment with lime to
increase the pH has been found to reduce lead dissolution rates. Where
water treatment is not possible, then a switch from brass compression
fittings to copper fittings, soldered with lead-free solder (see Chapter 18),
is an alternative.
When selecting an alloy to handle an aqueous chemical solution, it
is important to pick one of the more resistant alloys. As was described
previously, the resistance of copper alloys varies according to the chemical,
the pH, the temperature, and the degree of aeration. This latter variable is
most important as copper alloys will resist corrosion well in a wide variety
of chemicals when the solution is deaerated or of low oxygen content, but
will corrode rapidly when the solution is aerated or there are oxidizers
present.
As with waters, it is important to keep flow velocities down to avoid
erosion corrosion problems.
Avoid the selection of copper alloys for environments in which they
are susceptible to SCC, unless stresses are very low or are compressive.
SCC can lead to premature failure, despite a low general corrosion rate (see
Chapter 10). Again, select the more resistant alloys for specific environ-
ments.
Where multiple chemicals are present, particularly when the pH or
the redox potential is uncertain, it is advisable to carry out corrosion tests to
ascertain an alloy’s suitability. When a copper alloy is connected to a less
noble material, such as carbon steel, it will often be galvanically protected
and will not corrode, even if the environment is aggressive (see Chapter
14). The use of nickel aluminum bronze hooks for lowering carbon steel
plates into hydrochloric acid pickling solution is a typical example (see Sec-
tion 4.2.2).
References
1. B. Todd, “Factors in the Choice of Materials for Marine Engineers,” Trans. Inst.
Marine Eng. 80 (1968): p. 161.
2. INCO, “A Guide to the Selection of Marine Materials,” 2nd ed. (INCO, 1973).
3. EM. Reinhart, “Corrosion of Metals in Hydrospace, Part IV, Copper and Copper
Alloys,” Naval Civil Engineering Laboratory Technical Note, N-961, April 1968.
4. C.W. Hummer, C.R. Southwell, A.L. Alexander, Matel: Prof. 7 (1968): p. 41.
5. T.P. May, E.G. Holmberg, J. Hinde, “Seawater Corrosion at Elevated

Temperatures,” Dechema-Monographian Bank, 47, 1962, p. 253.
6. E.M. Mattson, R.M. Fuller, “A Study of Materials of Construction in Distillation
Plants,” Joint Progress Report by Office of Saline Water and INCO, No. 163,
October 1965.
8. K.D. Efird, D.B. Anderson, M P 14 (1975): p. 37.
9. C.F. Schreiber, 0. Osborn, F.H. Coley, M P 7 (1968): p. 20.
10. J.W. Oldfield, G. Swales, B. Todd, “Corrosion of Metals in Deaerated Seawater,”
Second Middle East Corrosion Conference (Bahrain: BSE-NACE, 198I),
p. 48.
1 1. R.W. Ross, CORROSION/77, paper no. 94 (Houston, TX: NACE International,
1977).
12. G. Bianchi, P. Longhi, Corros. Sci. 13 (1973): p. 853.
13. E.A. Vik, T.L. Nilsen, P. Hallberg, “Corrosion Monitoring Techniques Used in
Water Supply Systems,” 1 1th Scandinavian Corrosion Congress (Stavanger,
Norway, 1989).
14. R. Francis, Galvanic Corrosion-A Practical Guide for Engineers (Houston, TX:
NACE International, 2001).
15. H.S. Campbell, “Corrosion of Metals in the Water Supply Industry,” BNF
Miscellaneous Publication, MP544, October 1968.
16. L. Trostad, R. Veimo, J. Inst. Met. 66 ( 1940): p. 17.
17. P.T. Gilbert, Proc. Soc. Water Treat. Exam. 15 (1966): p. 165.
18. J. Nuttall, “Copper,” in Corrosion and Related Aspects CfMaterials for Potable
Water Supplies, P. McIntyre, A.D. Mercer, eds. (London U.K.: Institute of
Materials, Minerals, and Mining, 1993), p. 65.
19. R. Oliphant, “Causes of Copper Corrosion in Plumbing Systems,” Foundation for
Water Research Report, FWR0007, May 2003.
20. C.R. Maselkowski, BNF Research Report, S495/2, February 1984.
21. M. Critchley, R. Taylor, R. O’Halloran, M P 44 (2005): p. 56.
22. P. Munn, “Effect of Phosphates on Corrosion of Copper in Domestic Hot and Cold
Water Systems,” U.K. Corrosion (Sheffield, U.K.: Institute of Corrosion, 2007).
23. J.I. Paige, B.S. Covino, Corrosion 48 (1992): p. 1040.
24. R. Oliphant, “Contamination of Potable Water by Lead from Leaded Copper
Alloys,” Second International Conference on Water Pipeline Systems, Conference
Series Publication 10 (BHR Group, 1994).
25. R. Francis, H. Saxton, BNF Metals Technology Centre Research Report, R658/6,
January 1991.
26. S. Bond, R. Francis, H. Saxton, BNF Metals Technology Centre Research Report,
R658/12, August 1992.
27. E. Rabald, Corrosion Guide, 2nd ed. (New York, N Y Elsevier, 1968).
28. L. Caruso, H.T. Michels, M P 20 (1981): p. 35.
CHAPTER 5
Pitting Corrosion
When copper and its alloys are immersed in water and form a film of
cuprous oxide (CuzO), it is usually protective, and corrosion rates are low.
However, if this film breaks down locally, or a nonprotective film forms,
particularly in localized areas, then rapid local attack can occur. Once this
develops, the pits can be very narrow and may propagate rapidly. Failures
by pitting attack may occur in a few months or only after 10 or more years.
There are several different types of pitting attack that can occur on copper
and its alloys, and they are discussed in the following sections, along with
the actions necessary to prevent them.
5.1 Type I Pitting
5.1.1 Copper
In the late 1940s and 1950s, there was a great increase in failures in the
United Kingdom by pitting of copper piping carrying fresh water. This
corresponded to a time of massive rebuilding following the Second World
War, but also to a change by the major copper tube manufacturers from open
hearth annealing to bright annealing. Under the reducing conditions during
bright annealing, any residual drawing lubricant could crack, leaving thin
carbon films on the surface. The film was not visible to the naked eye, but it
was suggested as early as 1950that these films could play apart in the pitting
mechanism.' In the 1960s and 1970s, failures were seen across Europe, as
all the copper tube manufacturers adopted bright annealing. Today, Type I
pitting, as it came to be called, has been seen on all continents.
Figure 5.1 shows a typical Type I pit in a copper pipe. The green
mound over the pit is basic copper carbonate and calcium carbonate
57
FIGURE 5.1 Typical appearance of Type I pitting. (Photograph courtesy of H.S.

Campbell.) (See color section.)
(CaC03), with a red crystalline layer of cuprous oxide beneath it. Be-
neath the oxide is usually a thin layer of cuprous chloride (CuCl), but it is
not always visible to the naked eye. The metal surrounding the pit also has
a cuprous oxide film, often with some basic copper carbonate above it, but
less than is seen above the pit (Figure 5.1).
Type I pitting occurs in cold, hard well waters, and failures have
occurred in as little as 3 months. Type I pitting does not occur in warm
waters, probably because other copper compounds become stable. Pitting
does occur at elevated temperatures, which is discussed in the following
sections on Types I1 and I11 pitting. Type I pitting does not occur in surface-
derived waters, and Campbell suggested that this is because surface-derived
waters contain traces of organic compounds, some of which could act as
“natural inhibitors.”’
Early work showed that the cuprous oxide film that formed in surface-
derived waters was a dull brown color and adherent.’-3 The oxide formed
in well waters that caused pitting was red and flaked off easily. Waters
containing natural inhibitor fluoresced under ultraviolet light, while pitting
waters showed little or no fluorescence. The fluorescence can be quantified
in a fluorometer against a solution of sodium chloride (NaCI) in distilled
water.4 An alternative way of assessing a water is to pass a current of 10
@cm2 between two copper wires, one of which is in the water under test
and the other of which is in the same water plus sodium hydroxide (NaOH)
Pitting Corrosion 59
and sodium ~ h l o r i d e After

.~ 14 days, the nature of the oxide on the wire
in the test water will be as described previously for pitting and nonpitting
waters, that is, red and flaky or brown and adherent, respectively.
The early work showed that the inhibitor was a negatively charged,
acidic colloidal substance that was stable up to 100°C and could be extracted
with chl~roform.'-~ It was thought that the inhibitor was possibly a delta
lactone. BNF Metals Technology Centre revisited this area in the 1980s and
extracted organic material from activated charcoal used at a water treatment
works in the London area. This was separated by chromatography into
six major groups, the smallest of which contained the inhibitor. Repeated
separations of this part of the extract always showed the inhibitor to be
in the smallest fraction. Eventually, the smallest fraction was too small
for further separation or analysis, but when added to a pitting water, the
inhibitor concentration was 5 1 mg/L, and it completely prevented pitting.
Similar work by Edwards et al. also showed the effectiveness of the natural
inhibitor in small q ~ a n t i t i e sIt. ~is clear that a very large sample of organic
material will need to be extracted to identify the inhibitor, and this work
still remains to be done.
There is increasing attention by water companies to water treatment,
including the removal of organic material. It is becoming increasingly
common to pass water through activated charcoal or similar substances that
remove much of the organic material present in surface-derived waters,
including the natural inhibitor. Campbell reports several instances in which
such treatments have turned nonpitting waters into pitting water^.^
The copper tube that is usually used for fresh water transport is tough
pitch or phosphorous deoxidized (PDO), and both types are susceptible to
Type I pitting. Devroey and Depommier found in field tests that pdo copper
was slightly more resistant to pitting than tough pitch.6 Sometimes copper
also has small additions of arsenic, and field trials found that arsenical
coppers were more susceptible to pitting than nonarsenical coppers.'
Copper tubing is produced in three conditions: hard, annealed, and
half-hard. Hard tube requires no annealing and so should contain no carbon
films. Half-hard tube is produced by cold finishing annealed tube. Devroey
and Depommier compared uncleaned PDO copper tubes in all three tem-
pers, exposed for 1 year in a pitting water.6 The results, in Table 5.1, show
that soft tubes were more susceptible to Type I pitting compared with half-
hard tubes with a similar carbon content. Hard tubes were virtually immune
to pitting, with only very shallow attack on all tubes.
Cornwell et al. carried out extensive field trials in the United Kingdom
using two pitting waters and one nonpitting water.8 They found that bright
TABLE 5.1 Effect of copper tube temper on the susceptibility

to Type I pitting of uncleaned tubes6
Maximum pit depth
(after 1 year)
Temper No. of tubes t0.lmm 0.1-0.45 mm >0.45 mm
Soft 13 3 (23.1%) 2(15.4%) 8(61.5%)
Half-Hard 62 41 (66%) 13(21%) S(13%)
Hard 33 33 (100%) 0 0
annealed tubes with no postanneal internal cleaning pitted readily in both

pitting waters, with high tube potentials, while tubes that had been internally
cleaned suffered no pitting and had low potentials. Both types of tube had
low potentials in a nonpitting water, and no pitting was visible after testing.
Transferring pitting tubes to a nonpitting water resulted in much lower
potentials and the cessation of pitting, while the transfer of uncleaned tubes
from a nonpitting water to a pitting water resulted in pit initiation and high
potentials. Cornwell et al. defined three potential zones:
1. No pitting, E < 100 mV SCE (saturated calomel electrode)
2. Pitting possible, 100 < E < 170 mV SCE
3. Pitting propagating, E > 170 mV SCE
Work by Pourbaix et al. found similar thresholds for pitting when comparing
thermodynamic and potential eq~ilibria.~-"
The work of Cornwell et al. as well as the work of other laborato-
ries around Europe clearly showed that tubes that were properly cleaned
internally were much less likely to pit, even if uncleaned tubes did.' This
demonstrated the importance of carbon films in Type I pitting of copper
tubes. Other possible causes of Type I pitting are described later. Type I
pitting has also occurred in copper tanks and low-pressure hot water cylin-
ders (in the United Kingdom), and these do not contain carbon films. Pits
form at the bottom of tanks and cylinders because in a hot water cylinder,
the water is cold in the bottom region. Lucey attributed the pitting to the
formation of streaming corrosion products during early exposure such that
large quantities of cuprous chloride are produced on the vessel sides and
stream down to the lower regions. l4 This stimulates corrosion there and
enables a pit to form, when otherwise, it would not.
Edwards et al. have taken this argument further and have suggested
that any environment where cuprous chloride is produced at a rate that
exceeds the rate of its oxidation or hydrolysis to copper oxide or other
copper salts can encourage the formation of Type I pits.5 This need not be in
FIGURE 5.2 Type I pitting at a solder flux run. (Photograph courtesy of P. Munn.) (See
color section.)
tanks or cylinders, but could be in tubes, such as when the water is stagnant
for a long time after first filling. The pipe shown in Figure 4.7 was suffering
from cuprosolvency due to excessive stagnation after installation. However,
a few pits had also formed under the loose corrosion products, and one of
them was sufficiently deep to cause penetration and water leakage. Hence,
the carbon film is just one way in which the metal surface can be shielded
to allow the formation of sufficient cuprous chloride for pitting to initiate.
Other means by which high levels of cuprous chloride may be pro-
duced locally have been examined. Riedl and Klimbacher reported Type I
pitting associated with flux residues from soldering, even though the tube
was free of carbon films.15An example is shown in Figure 5.2. Flux residues
can be acidic and frequently contain high levels of chloride, added as an
activator, and so the production of higher than normal quantities of cuprous
chloride locally is perfectly understandable. This is sometimes regarded
more as a form of crevice corrosion (see Chapter 6), but there have been
cases in which pitting under flux runs has occurred that was associated with
pitting waters, while none has occurred in a nonpitting water. It clearly
depends on the level of residues left in the flux as well as on the water
composition.
Munn reports an increased tendency for Type I pitting in waters
that are on the borderline of supporting pitting when they are dosed with
phosphates.16 Phosphates are added to waters by water companies to re-
duce lead dissolution from old lead pipes. This is frequently done, even
FIGURE 5.3 Type I pitting in a water with high phosphate. (Photograph courtesy of
P. Munn.) (See color section.)
with hard waters that do not have a great tendency to be plumbosolvent.

Munn attributes the problem to the formation of loosely adherent copper
phosphate as well as the more normal basic carbonate. Figure 5.3 shows a
typical pipe with a layer giving a strong signal for phosphorus in the scan-
ning electron microscope and having Type I pits underneath it. The porous
phosphate layer is able to shield the metal surface and encourages an in-
creased concentration of cuprous chloride, which increases the likelihood
of Type I pitting.
Munn also suggested that pitting could result in pipes left stagnant
after a high chlorine disinfection treatment if the chlorine was inadequately
washed out. High levels of chlorine do cause substantial copper dissolution,
initially, under stagnant conditions (see Chapter 4).
5.1.2 Water Composition

Not all hard well waters will support Type I pitting, and a number of
workers have investigated the water compositions that will cause pitting.
Devroey and Depommier looked at a large number of waters from Belgium
and Luxembourg and concluded that the most important factors governing
Type I pitting were the calcium carbonate saturation index, the pH, and the
organic content, as determined by fluorescence. l 7 Aggressive waters had
a positive saturation index, a pH slightly above 7, and very slight or no

fluorescence.
A survey of waters in the United States concluded that pitting waters
contained over 5 mg/L dissolved carbon dioxide (CO2) (89% had between
10 and 40 mg/L).I8 The pH was in the range 7.0-7.8, and the dissolved
oxygen (DO) content was high (10-12 mg/L). The pitting waters also had
a high sulfate to chloride ratio, usually 3 or greater.
Lucey constructed a nomogram, originally based on 17 U.K. water
compositions, that considered six factors in determining a pitting propensity
rating, or PPR.19This nomogram was later refined using data from 103U.K.
waters2’ The factors that were included were sulfate, chloride, nitrate,
sodium, pH, and DO. Increasing sulfate, sodium, and DO increased the
pitting propensity, while increasing chloride, nitrate, and pH decreased it.
While sufficient chloride to produce cuprous chloride is required, large
chloride concentrations decreased the pitting propensity.
More recent work by Edwards et al. suggested that increasing sul-
fate promotes pitting, while increasing bicarbonate (which is linked to
pH) decreases pitting pr~pensity.~ Additional work showed that increas-
ing chloride decreased the pitting propensity, while nitrate increased it.2’
The finding with nitrate is the only one that disagrees with Lucey’s
findings.
Jacobs et al. considered the role of sulfide on pitting in fresh waters.22
This is uncommon in fresh waters in Europe, but it is prevalent in some
areas of the United States, such as Florida, where concentrations from 0.0 1
to 0.3 mg/L are possible. They found that sulfide could depolarize the
cathodic reaction and increase corrosion rates. It was suspected that pitting
problems in several U.S. states where sulfide was present were exacerbated
by its presence.
Lucey’s nomogram has no basis in theory and was developed empiri-
cally from the known behavior of a number of waters; its usefulness lies in
being able to predict the performance of other waters. Campbell reported
the successful use of the nomogram for waters not included in compiling
the nomogram, including some Icelandic water^.^
The nomogram is shown in Figure 5.4A, and Figure 5.4B shows how
it is used. A little explanation will be helpful. First, a horizontal line is
drawn from the chloride concentration on line PK to the curve PQ. Where it
intersects, a vertical line is drawn (K’L). The sulfate concentration is fixed
on line DE, and a horizontal line is drawn to the curve RS. The intersect,
G, is moved up or down, according to whether the nitrate is in excess of the
sodium or vice versa (point H). Line FH is then extended to the “effective
'F ' w
(A)
1 LEFT I I CENTRE
"I r
FIGURE 5.4 Lucey's nomogram for Type I pitting: (A) the nomogram; (B)how it is used.I4
sulfate” axis (point J). The intersect of effective sulfate and chloride is
point L. A line is then drawn from the DO concentration through point L
to line VW (point N). A horizontal line is then drawn from point N until
it intersects the curve of the pH of the water (curve XZZ). The curve is
then followed to its intersection with YZ, where the PPR can be read off.
This is a complicated procedure, and it was incorporated into a simple
computer program to make its use easier. The nomogram produces a PPR
that indicates the likelihood of failure, as follows:
1. No failures, PPR < 0
2. Borderline, failure unlikely (10 to 20 years), 0 < PPR < 1
3. Moderately aggressive, failure in 3 to 10 years, PPR = 2; and
4. Very aggressive, failure in less than 3 years, PPR > 2
Note that PPR values very much greater than 2 do not necessarily mean
that the time to failure will be drastically reduced below 3 years.
The nomogram has a number of limitations and should only be used
within a certain range of compositions. For example, it cannot handle
chloride contents greater than -50 mgL, and it also suggests that a water
with zero chloride would be very aggressive, while it is fairly certain that
pitting would not O C C USimilarly,
~.~ waters with a very low pH appear to
be aggressive, when in fact, increased general corrosion would occur. One
area in which the nomogram underestimates the PPR is for waters with low
sulfate contents (<10 mg/L). There were insufficient waters of this type in
the original sample to take proper account of this. The inclusion of sodium
seems unusual, but following publication of the nomogram, several Type I
failures were seen in base exchange softened waters, when untreated water
was not aggressive. Hence, increasing the sodium content can increase
susceptibility to Type I pitting.
One area in which the nomogram has been useful is in predicting
the effect of mixing waters. This is not uncommon across Europe, and the
nomogram can show the PPR at different mixing ratios. For many waters,
the curve is smooth, almost linear, from 100%of the water with the highest
PPR to 100% of the water with the lowest PPR. In some cases, however,
mixtures can be more aggressive than either of the waters alone. One such
case was reported by Lucey, and pitting failures occurred with a mixture
close to the worst case scenario (Figure 5.5).17 The author has also found
the nomogram useful for this type of analysis on several occasions.
Lucey’s nomogram does not appear to consider whether the water
is surface-derived or from a well. However, it has been established that
surface-derived waters frequently contain humic acid, which can hold in
1-
solution concentrations of calcium carbonate well above the equilibrium

quantity. Consequently, these waters have an unusually high, stable pH.
This is true for many surface waters, and a high pH gives a low PPR value
on Lucey’s nomogram.
A study in 1980 tried to rationalize the differences between waters
causing Type I pitting in the United States and the United Kingdom.23
One hundred eighteen water compositions from the United Kingdom and
38 from the United States were used in an automatic interaction detector
computer program. Using only the U.K. waters, the critical parameters were
CO2 content, sulfate-chloride ratio, sulfate, and DO. Using only waters from
the United States, the most important parameters were carbonate hardness
followed by C02, nitrate (NO3), and temperature. Taking all the waters
together, total alkalinity, pH, DO, and calcium were the most important
parameters. One possible cause of the discrepancies was changes in the
water compositions of the U.S. waters between the original failures and the
water sampling years later.
One factor that has turned out to be important is which cations are
associated with which anions. This came to light when research at BNF
Metals Technology Centre was initiated to manufacture a synthetic pitting
water.24All the waters were tested using the BNF Metals Technology Cen-
tre accelerated pitting test cell, which is described in Section 5.8. Initial
thought was that a water with a moderately high bicarbonate content (150
to 200 m g L ) and a sulfate-chloride ratio of approximately 2 would initiate
pitting. Tests with a synthetic water made solely of sodium salts performed
just like distilled water, with no pitting. As pitting waters all contain cal-
cium salts, a water was made with calcium chloride (CaC12), sodium sulfate
(Na2S04), and sodium bicarbonate (NaHC03). This, too, performed just
like distilled water. Finally, a synthetic water was made from the salts
as they are present in natural pitting waters, that is, calcium bicarbonate,
sodium sulfate, and sodium chloride. This water behaved just like an ag-
gressive pitting water. All these waters had a similar pH (7.3 to 7.5), and
no explanation of why the addition of bicarbonate as calcium bicarbonate
is necessary has been offered. It is possible that it was the C02 content that
made the water aggressive, but the calcium bicarbonate solution was purged
with air to remove the excess C02 and increase the pH into the desired
range.
5.1.3 Mechanism
It is not the intention to go into a full discussion of the mechanism of Type
I pitting, but a brief review of the current position will be useful.
The first detailed study of the pitting of copper was published by
May,25 and this was followed by work by Pourbaix and his colleagues at
CEBELCOR and Lucey at BNF Metals Research Association. In 1967,
Lucey published a theory of pitting corrosion based on a detailed examina-
tion of service failures and extensive laboratory experiments. l 4
The theory requires an excess cuprous chloride concentration at the
pit initiation site. All previous theories postulated a large cathodic area
surrounding the pit, which is the anode. The production of hydroxyl ions
at the cathode should, in hard waters, result in the precipitation of cal-
cium carbonate scale on the cathodic areas. Lucey found, however, after
examining a large number of service failures from Type I pitting, in both
pipes and tanks, that there was no more carbonate scale around the pit
than at areas remote from any pitting. There was, however, a large amount
of calcium carbonate in the mound above the pit. This suggested that the
cathodic reaction was taking place principally above the pit, and not on the
surrounding surface. Sosa et al. have suggested that the white product was
not carbonate and that the cathode was really in the area around the pit.26
However, Lucey painstakingly removed the white crystals from many pit
MSK CWRlC SALTS

AND CALCIUM CARBONATE
CUPROUS OXlCf MEMBRANE
CRYSTALLINE CUPROUS OXIDE

C" COPPER
\ CUPROUS CHLORIM
FIGURE 5.6 Schematic of Type I pitting according to Lucey.14
mounds, and they all effervesced when tested with cold, 10% sulfuric acid
(H2S04), showing the presence of carbonate.
Lucey also showed that there is a cuprous oxide membrane across the
mouth of the pit, and he demonstrated that it acts as a bipolar membrane,
with the lower surface being anodic and the upper surface being ~ a t h o d i c . ' ~
Cuprous chloride within the pit is anodically oxidized to cupric on the
undersurface of the membrane, and the cupric ions thus produced can
further attack the copper metal within the pit to reform cuprous ions. The
principal cathodic reaction on the upper surface of the membrane is the
reduction of cupric ions formed from cuprous ions by reaction with the
DO in the water. The cuprous ions for this reaction come partly from
diffusion through the pores in the oxide membrane from within the pit and
partly by recycling cuprous ions produced by the cathodic reduction on the
upper surface of the membrane. A mound forms above the pit, which is
composed partly of basic copper salts and partly of calcium carbonate, by
reaction between the cupric ions, oxygen, and calcium bicarbonate. Fig-
ure 5.6 shows a schematic of the structure of a corrosion pit and indicates
the nature and site of each of the reactions involved.14
Lucey's theory was revolutionary in that it rejected the classical small
anode surrounded by a large cathode theory. Sosa et al. have questioned
Lucey's model and have repeated some of his experiments.26They claimed
that they have demonstrated that pitting in copper follows the classical
model, with a large cathode around a small anode. However, their model
does not explain the crystals of calcium carbonate seen in the mounds above
the pits, They suggested that the white product was not calcium carbonate,
but it is hard to think of another possible compound that is white, crystalline,

and effervesces vigorously in 10% sulfuric acid.
Sosa et al. suggested that it is a copper ion concentration cell that
provides the driving force for pitting.26 Lucey's model requires the forma-
tion of a cuprous oxide membrane across the mouth of the pit, and one
way of producing this is a copper ion concentration cell, as described in the
next chapter, on crevice corrosion. The author has cross-sectioned pitting
at crevices on copper alloys and seen copper oxide membranes across the
pit mouth; however, once established, the crevice (the source of the copper
ion concentration gradient) can be removed, and the corrosion continues,
with a mound of basic copper salts on top of the pit, just like a Type I pit.
Hence, a copper ion concentration gradient is one way of establishing a
pit. As Edwards et al. suggested, anything that creates an excess of cuprous
chloride may initiate ~ i t t i n g a; ~carbon film is just one possible route for
this to happen. Once the pit is established, Lucey's model describes what
is found in service very well. The mechanism of Type I pitting is described
in more detail in the literature cited previously.
5.1.4 Copper Alloys

There are very few reports of pitting failures due to Type I pitting for
copper alloys, and all of the ones that do exist are for heat exchanger
tubing. Breckon and Baines reported some in the 1950s in marine heat
exchanger^.^^ Retief reported failures due to the presence of carbon films
of both admiralty brass and 904 0 copper-nickel tubes in heat exchangers in
South Africa handling brackish water.28 Marsden reported similar failures
with 90/10 copper-nickel in a recirculating water containing 100 to 1,000
mg/L chloride and 200 to 1,700mg/L sulfate, also believed to be caused by
carbon films in the tubes.29
Grassiani reported failures of admiralty brass tubes in a recirculating
cooling water containing the f~llowing:~'
100-250 mg/L chloride

250450 mg/L sulfate
200-300 mg/L bicarbonate
15-30 mglL silicon
130-160 mg/L calcium
40-70 mg/L magnesium
pH=7
In addition, the water was dosed with a variety of chemicals, including

chromate (5 mgL), phosphonates, polysacrylates, and azoles.
Heavy carbon films were found in the corrosion products and also
on new tubes of both admiralty brass and 90/10 copper-nickel. Strong acid
cleaning was required to remove the carbon films from new tubes.
Sat0 et al. carried out a study of the effect of surface finish on the
corrosion of aluminum brass tubes carrying seawater.” They concluded that
surface finish had no effect on long-term corrosion behaviour. However,
they did not characterize the carbon contamination level by any of the
recognized tests, and they did not include heat transfer in their tests.
Shone and Grimm reported experiences with corrosion over 25 years
with the Shell tanker fleet.32 This included failures of 66/30/2/2 Cu-Ni-
Fe-Mn heat exchanger tubes due to deleterious films from the manufacturing
process. Shone et al. did not specifically call the films carbon, but it seems
likely that this is what they were.
The author conducted tests on a range of copper heat exchanger
alloys with various surface finishes and also heat transfer.33 All tests were
conducted with once-through seawater for 3 months. The only tubes that
suffered pitting were those with a heavy carbon film. Tubes with carbon
films that would be classed as unsatisfactory for fresh water service showed
no pitting and looked just like the tubes with clean bores. The pitting
in the tubes with heavy carbon films was shallow without heat transfer
(0.04 mm), but on the tubes with heat transfer, the maximum depth of
attack was 0.2 mm on one tube and 0.73 mm on a duplicate. Both tubes had
lots of pits in the zone subject to heat transfer. It is clear that carbon films
can cause pitting in copper alloy tubes in seawater, but the level of carbon
to cause pitting is much higher than for fresh water, and heat transfer is also
required.
5.2 Type II Pitting

In the early 195Os, Campbell conducted a survey of pitting corrosion failures
of copper tubes in the United Kingdom. About 65% of the failures were in
cold, hard well waters-what later came to be called Type I pitting attack.
A further 30% of failures were in soft, hot waters of low pH. This type of
corrosion is now called Type ZZpitting, and it is mostly found in countries that
have predominantly soft waters such as Scandinavia and Japan. However,
failures have also been seen in the United Kingdom, Germany, and Canada.
Type I1 pitting is slower than Type I, and failures usually occur after 8 to
12 years.
Type I1 pitting is characterized by deep pits of small cross section
containing very hard, crystalline cuprous oxide and basic copper sulfate.
FIGURE 5.7 Typical appearance of Type II pitting. (Photograph courtesy of H.S. Camp-
bell.) (See color section.)
The surface between the pits is a nearly black layer of water-formed oxide
beneath a thin layer of silt deposited from the water.4 Typical Type I1 pitting
is shown in Figure 5.7. Sato et al. investigated a large number of failures of
copper hot water pipes from the Tokyo area of Japan and found similar pits
to those described by Campbell.34 They also found a thin film of cuprous
chloride in the base of the pit, although this was not visible to the naked
eye. A schematic of a Type I1 pit, according to Sat0 et al., is shown in Fig-
ure 5.8.34
Mattson and Fredriksson examined failures of copper tubes in Swe-
den and concluded that Type I1 pitting would occur when the pH t7.4, the
bicarbonate-sulfate ratio is less than 1, and the water temperature is 60°C
or more.35 Campbell reported data from Canada, Germany, and the United
Kingdom that follow this trend.4 A variation of Type I1 pitting can occur
when the water contains mangane~e:~ as little as 30 pg/L has caused prob-
lems, although the manganese concentration is usually higher. The black
film between the pits then contains large quantities of manganese dioxide
(Mn02), which is a well-known cathodic depolarizer. The manganese diox-
ide appears to stimulate Type I1 pitting in waters that might otherwise be
regarded as b ~ r d e r l i n e These
.~ often have a pH >7.4 because they have
been treated, but they have a low bicarbonate content, so any corrosion
could rapidly lower the local pH below 7.4.
Kasahara et al. investigated failures of copper pipes from recirculating
hot water systems in the Tokyo area of Japan, with a variety of supply
Brown deposit
Amorphous
Yellowish green deposit

Amorphous Greenish blue mound
Black film
CUCl
FIGURE 5.8 Schematic of Type II pitting according to Sat0 et al.34
waters.36All of those that had a bicarbonate-sulfate ratio less than 1 formed

a corrosion product layer that contained large quantities of orthosilicates,
with pitting under the orthosilicate layer. Kasahara et al. believed that the
silicate layer played a part in the initiation of pitting. However, the silicate
layer was also seen in waters with a bicarbonate-sulfate ratio greater than
I, and pitting did not occur in these tubes.
In the 1980s, concern about Legionnaires’ disease led many large
buildings in Japan to increase the chlorination of their recirculated hot
water supply. Chlorine residuals of 1 to 3 mg/L were common, and there
was a big increase in the failures by Type I1 pitting. Some of these were in
waters in which the bicarbonate-sulfate ratio exceeded I , and research was
undertaken to investigate the effects of chlorine on Type I1 pitting.
Baba et al. reported that residual chlorine in hot water was a cause
of pitting in copper pipes.37 They also showed, in accelerated tests, that
residual chlorine concentrations of a few milligrams per liter in potable

water, along with a high water temperature and flowing conditions, af-
fected the electrode potential of copper tubes.38 The potential rose to levels
in excess of 150 mV SCE, which is above the pitting potential at 60°C.
This is not dissimilar to the potential of 170 mV for Type I pitting, allow-
ing for differences in pH and temperature. After several days' exposure, at
this potential, numerous small pits appeared on the tube surface, whereas
under the same conditions, but with chlorine-free water, the tube potential
remained below 80 mV SCE, and no pits were observed, even after pro-
longed exposure. Baba et al. concluded that chlorine, a strong oxidizing
agent, plays an important role in the initiation of Type I1 pitting.
Suzuki et al. showed that the rise in the tube potential occurred in
two stages: the first stage reflected the formation of a cuprous oxide layer
at potentials up to 60 mV SCE, while the second stage, from 60 mV to 150
mV SCE, reflected an absorption reaction of the hypochlorite ions with the
cuprous oxide layer.39 They demonstrated that hypochlorous acid (HOCl)
and hypochlorite ions, produced by the chlorination of the water, acceler-
ated the formation of the cuprous oxide layer, and anions, such as C10-,
interchanged with OH-, increasing the electrode potential of the copper.
Baba et al. conducted further laboratory tests to reexamine the critical
potential for the onset of pit growth on copper tube in flowing synthetic
hot water by a potentiostatic method.40 They also reexamined the effect of
chlorine on the potential of copper tubes. The polarization tests were not
successful in that no increase in current occurred even when pits formed,
so no critical potential could be identified. However, examination of the
tube surfaces confirmed the value of 150 mV SCE as the critical one for the
initiation of pitting in flowing hot water. The change in corrosion potential,
indicating the initiation of pitting, in water at pH 7 containing 3, 5 , and 10
m g L chlorine was observed to occur after 80,60, and 37 hours, respectively.
The structure and morphology of the pits that formed under these conditions
were the same as that of Type I1 pitting described previously.
Francis and Saxton also investigated the effect of chlorine on copper
tubes in Sheffield water, a soft, surface-derived water with a bicarbonate-
sulfate ratio less than 1, at 60°C.4' In nonchlorinated water, tubes that had
been shock dosed with either 100 mg/L chlorine for 2 hours or 20 mg/L
chlorine for 16 hours developed protective films, similar to those on tubes
that had not been shock dosed. None of these tubes showed any pitting.
However, tubes that were continuously dosed with 1 mg/L chlorine devel-
oped pits in the 30-day tests. The pits were broader than the normal Type I1
pits described by Mattson and Fredriksson and C a m ~ b e l l . However,

~?~~
Hamamoto et al. also reported a similar pit morphology in an independent
study with a Japanese soft water with a continuous chlorine residual of 1
mg/L.42 Chlorine not only modified the shape of the pits, but also changed
the rate of propagation. Type I1 pitting normally does not lead to leakage in
less than 8 years, but both Francis and Saxton and Sat0 et al. found more
rapid penetration rates in chlorinated water, with failures in less than 3 years
in some case^.^^,^'
There has been less research to determine the mechanism of Type I1
pitting, but Mattson and Fredriksson carried out tests in an artificial pit cell
and investigated the effects of pH and water composition on the oxide that
formed on a copper anode.35At pH values from 7 to 10, three types of green
corrosion product formed over a period of 4 days. Type A was adherent and
formed in all waters containing bicarbonate but free of sulfate and chloride.
In waters with 10 mg/L sulfate or more, a Type B crust formed, which was
less adherent. This was identified as basic copper sulfate. In waters with
chloride, and also in waters with sulfate and 280 mg/L bicarbonate, a Type
C crust formed, which was adherent. The greatest corrosion occurred with
a Type B crust, and if this is poorly adherent and has cracks or defects in
it, the crust will act as a screen to enable the establishment of a copper ion
concentration cell, in a similar manner to Lucey’s model for Type I pitting
(see Section 5.1.3). This would then create the conditions for pit initiation,
with hydrolysis and pH reduction under the crust.
5.3 Type 111 Pitting

The occurrence of Type I11 pitting was not really recognized until the 1970s
and 1980s, when there were a large number of failures, many in large
public buildings such as hospitals, offices, and hotels. There was concern
about the occurrence of a new type of pitting and why it had not been seen
previously. However, when Campbell reviewed a large number of copper
tube failures by pitting in the United Kingdom in the early 195Os, 5 % of
them were neither Type I nor Type I1 attack. Most of the 5% were in soft
waters of low pH, either in warmer cold water pipes or in cooler hot water
pipes. It seems likely that these were Type I11 pitting, but at that time, the
preponderance of the other two types of pitting meant that it was lost in the
statistics.
The author reviewed Type I11 pitting failures and found that they had
been reported in Germany, Sweden, Australia, Saudi Arabia, the United
FIGURE 5.9 Typical appearance of Type 111 pitting. (Photograph courtesy of H.S. Camp-
bell.) (See color section.)
Kingdom, Japan, and the Republic of Ireland!3 Some of these were in cold
water pipes, where they were in a warm environment and water tempera-
tures were 25°C or more. Others were in hot water pipes, but far removed
from the heater, with water temperatures typically 550°C. All the waters
were soft, with a low pH, but other factors also appeared to be contri-
buting.
Figure 5.9 shows typical Type I11 pitting, with broad deposits of basic
copper sulfate covering multiple pits filled with cuprous oxide. Where
the multiple pits perforate the tube, the typical pepper-pot appearance of
Type I11 failures is created. The dark film between the pits is mainly cupric
oxide, often with deposits of organic material above it.
The first instance of Type I11 pitting was reported by von Franque,
and about the same time, an epidemic of failures occurred in one part
of S ~ e d e n ! ~ - ~The
~ failures in Sweden concerned cold water pipes in
domestic properties, and the failures appeared to coincide with a change in
water supply. Up to 1970,the water came from the Skallsjo works, in Floda,
but thereafter, increasing amounts of water from the Oxsjo works, in Lerum,
were added. The water compositions are shown in Table 5.2; the main
difference is the low bicarbonate content of the Oxsjo water, particularly as
a raw water. This is because the source was heavily affected by acid rain, and
the water had to be treated with sodium hydroxide to increase the pH. Even
so, the bicarbonate-sulfate ratio for the Oxsjo water was less than 1, while
TABLE 5.2 Water composition from two Swedish sources

from 1970 to 198045
Concentration(rngll)
Skallsjo (Floda) Oxsjo (Lerum)
Factor Raw Supply Raw Supply
PH 6 8.3 6.3 7-9
Total Hardness (CaC03) 28 28 8 10
Bicarbonate (HC03) 16 48 1 10
Chloride 16 13 12 12
Sulphate 10 13 17 18
that of the Skallsjo water was greater than 1. Tests in both waters showed
that it was the Oxsjo water that caused the Type I11 pitting.45 Linder and
Lindman suggested that a natural inhibitor, as discussed in Section 5.1 for
Type I pitting, might be present in Skallsjo water and was absent in Oxsjo
water.45 However, they did not conduct any tests to positively identify the
inhibitor’s presence.
In the 1980s, there was a series of failures in large buildings, mostly
hospitals, in several countries around the world. Campbell et al. summa-
rized the cases that they had investigated from several countries.46Because
the pitting was somewhat of a cross between Type I and Type I1 pitting,
Campbell et al. called it Type 11/2 pitting, but its appearance is clearly the
same as that reported by other authors for Type I11 pitting.
Interestingly, Campbell et al. reported that none of the domestic prop-
erties adjacent to the hospitals had any pitting problems.46 This suggests
that stagnation andor slow flow may play a part in pit initiation, as hospitals
frequently have long runs of pipe, in which water flow is low or intermit-
tent, with long periods of stagnation. All the waters had similar composi-
tions:
Total hardness (as CaC03) 2 5 4 0 mg/L

Total alkalinity (as CaC03) 10-20 m g L
Chloride 8-20 mg/L
Sulfate 10-30 mg/L
In most of the waters, the bicarbonate-sulfate ratio was less than 0.5, and
Type I1 pitting was seen in the same waters at temperatures of 60°C or more.
However, where the leaks occurred, the water temperature was i50°C, with
a pH from 7 to 7.5.
Campbell et al. found that tubes with pits also contained polysaccha-
ride films.46 This is an exopolymer produced by many bacteria as part of
their metabolic processes, and Campbell et al. identified three species of

pseudomonad in the film on failed tubes. These bacteria were found in all
the tubes that were suffering pitting and also caused pitting when injected
into a sterile water of the same composition, when no pitting occurred in
the sterile water.46
Fischer et al. described tests carried out on the biofilms (polysaccha-
rides) removed from failed copper tubes from a German hospital.47 These
carried water at approximately 50°C and showed the typical pepper-pot ap-
pearance of perforations. The biofilm was present between the metal and the
oxide-hydroxide layer. Pitting occurred where the biofilm was disrupted.
This was believed to then cause a copper ion concentration cell to become
established due to the high copper ion concentration in the exopolymeric
layer. Peroxide and hydroperoxides were detected in the biofilm, and these
may have provided a more rapid cathodic reaction than the reduction of
DO, thus accelerating the corrosion process. The biofilm also contained
localized concentrations of amino acids, particularly glycine (CzH502N)
and gluconic acid ( C ~ H I ~ Owhich~ ) , are both complexants for copper and
which would increase local dissolution rates, thereby helping to establish
the copper ion concentration gradient.
It is thought that biofilms are not essential for Type I11 pitting, but they
can be a contributing factor. Nuttall reported that of the hospital failures in
the United Kingdom, biofilms were only detected in approximately 50% of
the failed tubes.48 This does not mean that there was no biological activity
in those tubes, only that the tests used detected no films. He also stressed
that the presence of humic acid in the organic material in the water could
be a contributory factor.44 This was reiterated by Bond, who conducted
tests on copper tubes in a Scottish hospital that was experiencing problems
with Type I11 pitting in the hot water circuit.49 Bond operated one rig on
the normal hot water supply, and the other received water that was filtered
to 0.5 pm. This reduced the incoming solids content considerably but did
not produce a sterile water, which requires finer filtering. The water flow
was 0.1 m / s for 6 hours, followed by 6 hours’ stagnation, with a typical
water temperature of approximately 50°C when the water was flowing,
decreasing to about 25°C after 6 hours’ stagnation. The tubes in the rig
receiving filtered water suffered less pitting than the ones receiving the
normal hot water supply, but the pitting was not totally suppressed. This
suggests that the bacteria producing the biofilm were the critical component,
and filtering the water just reduced their food supply. The role of biofilms
in the localized corrosion of copper tubes carrying fresh water was recently
reviewed by Kee~il.~O
Siederlarek et al. studied the formation of films on copper in waters

simulating those that were present in a German hospital that had suf-
fered numerous Type I11 pitting f a i l ~ r e s .As
~ ~with
, ~ ~studies of Type I and
Type I1 pitting, they found that the type of oxide that formed influenced the
propensity for pitting. Waters with a significant sulfate content had a high
propensity for pitting. They also found that water flow conditions signifi-
cantly affected the nature of the film. The propensity for pitting increased
when the water was stagnant or the flow was intermittent, with long periods
of stagnation. These results mirror those reported for Type I and Type I1
pitting.
5.4 Aluminum Brass

Aluminum brass is not usually used to carry fresh water, but there are
some instances of this. Campbell reports pitting of aluminum brass tubes
carrying drinking water on board ship.4 The pits were similar in structure to
Type I pits in copper, apart from the presence of zinc corrosion products in
the mounds. However, the pits were smaller and more numerous than Type
I pits in a copper tube. The surface of the tube around the pits carried a film
that was identified by x-ray diffraction as a mixed zinc-aluminum oxide of
the spinel type Zn0.A1203.The presence of a film free of copper suggests
the presence of bacteria that are able to chelate copper.
This type of pitting has received little attention. The author was unable
to reproduce it in the laboratory using alternate flow conditions at ambient
temperature using both a hard well water and a soft, surface-derived water.
The principal corrosion product in both cases was cuprous oxide.
5.5 Alloy400
Alloy 400 has been observed to pit in very slow moving or stagnant seawater.
Figure 6.1B shows crevice attack on Alloy 400 panels exposed to quiescent
seawater, but pitting is also visible away from the crevice region.52 At
higher seawater velocities, no pitting occurs. This is thought to be due to a
shortage of oxygen to keep the passive film intact at very low water flow
rates. The pitting is exacerbated when sulfides are present in the seawater.
This is thought to be due to sulfide inhibiting the formation of the passive
oxide film and causing less protective sulfide films to form instead (see
Chapter 3). Pitting can occur even at higher water flow rates if sufficient
sulfide is present in the water.
100
iij -100
0
v)
>
-.-
E
m
E -200
m
I
-300
-400
10 20 30 40 50 60 70
Time (days)
I Natural seawater Seawater + Img/L CI I

FIGURE 5.10 Corrosion potential vs. time data for Alloy 400 in seawater.52
Klein et al. investigated the crevice corrosion resistance of Alloy

400 in quiescent natural and chlorinated (1 m g L ) seawater at ambient
temperature, but they also found that the alloy pitted outside the crevice
region, as shown in Figure 6.1B.s2 They found that the pitting was most
severe in natural seawater and was less severe in seawater containing 1 mg/L
chlorine. They related this to the pitting potential of Alloy 400, as previously
reported by Tipton and Kain, who found that the pitting potential was in the
range -20 mV to +10 mV SCE at temperatures from 20 to 28°C.s3 Klein
et al. plotted the potential of their Alloy 400 samples at various times of
exposure up to 60 days, as shown in Figure 5.10. Pitting was seen on all the
samples for which the potential exceeded the pitting potential.
Gallagher et al. found pitting in a seawater piping system in Alloy
400 that had seen mostly natural seawater due to the chlorination system
working only intermittently. They investigated the potential in natural sea-
water and found that it exceeded the pitting potential soon after expo~ure.’~
Samples exposed in seawater with 0.5-0.8 mg/L chlorine had potentials
that were just below the pitting potential and did not suffer attack.
This suggests that a biofilm of some kind forms on Alloy 400 in
natural seawater, similar to that found on high-alloy stainless steels. This is
sufficient to raise the potential into the pitting zone. A level of chlorine that is
just sufficient to prevent the biofilm formation keeps the potential below the
pitting potential and free of attack. However, chlorine is a powerful oxidizer,
and excess chlorine will also elevate the potential above the pitting potential.
The variable results of Klein et al. suggest that the chlorine demand of the
water varied during their tests so that sometimes there was excess chlorine
sufficient to cause pitting, and sometimes there was not. It appears that
chlorine concentrations around 0.5 m g L offer the safest fouling control
with Alloy 400.
5.6 Formicary Corrosion

Formicary corrosion was named because the corrosion takes the form of
tunnels of corrosion products. In cross section, the attack has the appearance
of an ants' nest; hence the name. The surface around the leaks is grayish
brown or bluish purple, and the pinhole leaks are extremely tiny so that they
can only be seen under a microscope. The attack may occur either from
the inside or from the outside of the tubes, depending on which side the
aggressive conditions necessary for formicary corrosion are present.
Formicary corrosion was first described by Keyes in the 1960s and
by Edwards et al. in the 1970s, but it was not properly understood until
extensive research was undertaken in Japan in the 1 9 8 0 ~ . ~ ~ - ~ '
Yamauchi et al. and Notoya et al. examined over 20 service failures by
formicary corrosion of copper refrigeration and air-conditioning t ~ b e s . ~ ~ - ~ ~
They had all failed on final test, prior to dispatch, or soon after entering
service. Most of the failures were from the inside, but a few were from the
outside. They appeared to be associated with the presence of carboxylic
acids, either formic or acetic (HCOOH or CH3COOH). The internal fail-
ures were mostly associated with formic acid, while the external failures
were from sushi bars, where they use a lot of vinegar (acetic acid). The car-
boxylic acids inside the tubes were thought to be produced during the final
cleaning operation, and the corrosion initiated during storage. The service
failures suggested that formicary corrosion requires oxygen, moisture, and
a carboxylic acid.
Notoya et al. carried out tests on copper exposed in desiccators con-
taining 100 mL of a 1 % v/v solution of various carboxylic acids for 125
days.58 The attack on the samples exposed over formic acid is shown in
Figure 5.1 1 and was similar to that seen in service failures. Figure 5.11A
shows the presence of voids due to attack within the metal, whereas Fig-
ure 5.1 1B shows microcracks in the copper oxide within the pits. Somewhat
B
FIGURE 5.1 1 Microsections showing formicary corrosion: (A) presence of voids;
(B)microcracks in oxide.
broader attack occurred with acetic acid, while the attack with propionic
and n-butyric acids was much finer. No attack at all was seen with n-valeric
and n-pentanoic acids. Clearly, the simpler the acid structure, the greater
was the attack, and this is reflected in the dissociation constants for the
acids, with the relative strength being greatest for formic acid.
Notoya et al. proposed a mechanism for the attack, whereby copper
forms a complex with the acid, as cuprous formate or a~etate.~' This is
then oxidized to give cuprous oxide and cupric formate or acetate. Because
cuprous oxide has a larger volume, it will tend to crack and hence be less
protective, permitting solution to contact the metallic copper beneath. The
cupric formate, or acetate, can then be reduced to the cuprous form by
reaction with copper dissolved from the metal. The pitting is fine because
the cracks in the cuprous oxide permit attack at the pit base, but the oxide
on the pit sides acts as the cathode.
While at BNF Metals Technology Centre, Campbell investigated the
failures of copper pipes placed in troughs in a concrete floor in a new
warehouse. The pipes leaked due to external corrosion after about a year.
Only the cold water pipes were attacked, whereas the hot water pipes
were free of corrosion. Both sets of pipes were insulated with a compound
containing phenol formaldehyde. The cold water pipes were damp when
the coating was applied because they already contained water from final
pressure testing. The combination of moisture and air led to the oxidation
of the formaldehyde to formic acid, which caused the formicary corrosion.
The hot water pipes were unattacked because the hot water dried the pipes
before the insulation was applied.
So far, all the failures discussed have been associated with copper
tubes and pipes, but Olszewski and Corbett reported the failure of 90/10
copper-nickel tubes used in a heat exchanger to cool hydrogen, with seawa-
ter inside the tubes.59 The tubes leaked after 8 months in service from the
inside, and the attack had the typical appearance of formicary corrosion.
It seems unlikely that significant quantities of carboxylic acids would be
present in the seawater, and it seems more likely that the attack occurred
during storage of the tubes. This may have involved the use of an insula-
tion or packing compound that contained formaldehyde. Clearly formicary
corrosion is not confined to copper; copper alloys can also be susceptible.
5.7 Rosette Corrosion
This type of attack is a relatively recently recognized phenomenon and is

only found in the United Kingdom and other regions that use low-pressure,
copper hot water cylinders.60Although most of the water in these cylinders

is hot, the very bottom region never gets warm and can suffer Type I
pitting in localities with a supply water having a high pitting propensity. To
combat this, Lucey suggested fitting a small aluminum anode to the base of
the cylinder, which protects the copper by depressing the potential during
the early part of its life. Once the aluminum is consumed, the copper has
formed a protective oxide film and is no longer susceptible to Type I pitting.
Unfortunately, cylinders with aluminum protector rods have been fitted in
many areas where the water does not support Type I pitting, in the mistaken
belief that the aluminum protects against all corrosion.
In some waters, corrosion has occurred beneath the aluminum corro-
sion products on the cylinder base. The aluminum corrosion products vary
in color from bluish white to light green. Underneath the deposit, it is a
vivid brick red and magenta color, consisting of large platelets of cuprous
oxide separated by deep trenches-hence the name Rosette attack. It was
originally thought to be a type of pitting, but it is more properly accelerated
general corrosion, due to the aggressive conditions beneath the deposits. In
some cases, this can lead to perforation and leakage.60 Figure 5.12 shows a
typical example of Rosette corrosion.
One possible cause of the attack is that under some conditions, there
is aluminum chloride (AlC13) as well as AlOOH in the corrosion product
deposit. Aluminum chloride hydrolyses readily and could produce very low
pH conditions beneath the deposit to exacerbate corrosion.
Another possibility is the fact that the galvanic current between the
copper and aluminum is sufficiently high that it is possible to reduce nitrate
to ammonia (NH3) and sulfate to sulfide.60Both these anions are aggressive
to copper and would increase attack. In some parts of the United Kingdom,
there has been a general increase in the nitrate content of the water due to
increased agricultural runoff. This increase, and the ability of the nitrate
ion to be reduced to ammonia, may explain why it has taken over 30 years
of the use of aluminum protector rods for the problem of Rosette corrosion
to be noticed.60

5.8.1 Type I Pitting
As described previously, one of the main causes of Type I pitting is carbon
film from the manufacturing process. Carbon films are less of a problem
nowadays, as most manufacturers use water-soluble drawing lubricants, and
FIGURE 5.12 Typical appearance of Rosette corrosion. (Photograph courtesy of P.

Munn.) (See color section.)
some also bright anneal in an atmosphere that is slightly oxidizing, rather

than reducing. This has the effect of oxidizing any cracked drawing lubricant
to carbon dioxide. Other manufacturers clean the tube bore by internal grit
blasting or by pickling. However, not all manufacturers are so assiduous,
and the European standard for copper tubes, EN 1057,61specifies that tubes
must be free of deleterious films in the bore. The older British Standard
for copper alloy heat exchanger tubes, BS 2871 Part 3,62also includes this
stipulation. The standards do not say how to check for such films. Shone and
Grimm3’ suggested the use of a simple electrochemical test, but it has been
found that the results for tubes on the borderline are not very r e p r ~ d u c i b l e . ~ ~
An alternative, which the author has used on several projects, is the detection
of carbon by heating a sample such as in the LECOt-type test. In this test,
a half-section of tube (abraded on the outside surface) is heated in a stream
of oxygen, and the CO2 produced at different temperatures is determined.
t Trade name.
By heating at increasing temperatures, the carbon due to various sources

can be isolated:
200°C solvents
650°C greases
850-900°C carbon films
Only the carbon measured at the highest temperature is used for assessing
the tube quality. The pass-fail criterion that has been adopted for copper
tubes for use in fresh waters is carbon 5 0.2 mg/dm2. Tubes that fail this
test can be cleaned by internal grit blasting or by pickling and then retested.
The test should be applied to several samples from each batch of tubes.
The author has also specified this test for copper alloy heat exchanger
tubes carrying seawater. Although research has shown that higher levels of
carbon are required to cause pitting, compared with fresh waters, the level
of carbon of 50.2 mg/dm2 was still applied, and the manufacturer had no
trouble meeting this standard. Clearly the test is only required for annealed
and half-hard tubes that are heat treated after drawing.
It is also clear that long periods of stagnation, after tubes are first
filled with water, may lead to pitting, and this should be avoided. The
number of long, rarely used legs should be kept to a minimum, and these
should have a slow flow of water through them for 2 to 3 weeks to form
protective films. This was found to work at a school in Torquay, United
I n g d o m , whose problems were investigated by BNF Metals Technology
Centre.
Munn16 has raised the question of phosphates interfering with pro-
tective film formation and possibly encouraging pitting. This is an area that
clearly requires more research.
Where low-pressure hot water cylinders are used, it is best to use one
fitted with an aluminum protector rod in waters that have a pitting propensity
rating greater than 1.@Oliphant recommends the use of a modified design
of cylinder, where the heating coil is much lower in the body.6" This makes
sure that even water in the very bottom of the cylinder is heated above 40°C
so that Type I pitting does not occur and an aluminum protector rod is not
required.
It is clearly important to know whether a particular water composition
will support Type I pitting. The nomogram of Lucey, described in Section
5.1.2, was a good start, but it has deficiencies and is not widely available.
The nomogram cannot cope with waters in which the chloride exceeds
57 mgL, and it underestimates the aggressivity of waters containing less
than 10 m g L sulfate.65
In an attempt to get around these problems, Shaw and Lucey devel-

oped a simple test in which a sample of the water could be evaluated.66
The test cell consists of an acrylic block with 12 copper anodes and a sin-
gle copper cathode (Figure 5.13). A glass rod created a crevice over one
anode, and a constant current was passed through the cell for 100 hours.
The greater the current passed by the shielded electrode after 100 hours, the
greater was the propensity to cause Type I pitting. Figure 5.14 shows the
correlation between the current passed by the shielded anode and the PPR,
calculated from Lucey ’s nomogram. There is a good correlation between
the current and the PPR such that Shaw and Lucey suggested the following
aggressivity rating:
Proportion of total
current carried by
shielded anode (“h) Aggressivity
<20 Nonpitting
20-5 0 Borderline (no failures expected)
50-70 Fairly aggressive (PPR = 1-2)
>70 Very aggressive (PPR > 2)
The BNF test cell is a little cumbersome to use in the field, and the anode
block needs regrinding after every test. Ansuini developed a simpler test
cell using an etched copper circuit board for the electrodes, designed such
that a new board could be plugged into the cell for each test.67 The board
is connected to a battery-operated power supply that also provides a visual
display of the PPR, as shown in Figure 5.15. The cell is easy to take to
site and connect to a slow flow of water through the cell (Figure 5.15).
The results with this compact unit correlated well with the BNF data, and
the PPR was found to hardly change after 50 hours of t e ~ t i n g . The ~~-~~
test equipment is still available and is recommended for assessing water
aggressivity toward Type I pitting.
Work has also been carried out to look at water composition mod-
ification to avoid Type I pitting. The first consideration must be to avoid
the use of activated charcoal or similar filters that can remove the naturally
occurring inhibitor from surface-derived waters that might support Type I
pitting. If water blending is possible, then mixing a surface-derived water
(with the natural inhibitor) with an aggressive well water can prevent Type
I pitting.
Depommier examined four water treatments to try to reduce the ag-
gressivity of Brussels tap water.68These were as follows:
Pitting Corrosion a7
Cathode
:ir
--- Copper Anodes
FIGURE 5.13 Schematic of BNF test cell for Type I pitting.
0 5 10 15
Current (FA)
FIGURE 5.14 Relationship between current on shielded anode and the pitting propensity
rating from Lucey’s nomogram.65
FIGURE 5.15 Picture of test cell by Ansuini with portable power source.67 (Photograph
courtesy of F. Ansuini.)
1. Add NaOH to increase pH to 8;

2. Add Ca(OH)2 to increase pH to 8;
3. Partially soften and add NaOH to increase pH to 8; and
4. Partially soften and pass through a Magno-Do1 filter (DIN 19621)
A test rig was constructed that contained copper tubes from a variety of
manufacturers, including some with carbon films. Depommier found that
neutralizing the COz with NaOH after partially softening (treatment 3) was
the only treatment that was successful. Even this treatment was not 100%
successful, and some half-hard tubes with a heavy carbon film still pitted.
He concluded that removal of the carbon films (by internal grit blasting or
pickling) or annealing in a slightly oxidizing atmosphere offered a more
reliable solution to problems of Type I pitting.
5.8.2 Type II Pitting

Mattson and Fredriksson recommended the following measures for pre-
venting Type I1 pitting35:
1. Reduce the water temperature to t60"C;
2. Keep the water pH >7;
3. Increase the bicarbonate content >70 mg/L; or
4. Increase the bicarbonate-sulfate ratio to greater than 1
If the water temperature is reduced too much, there is an increasing risk of
Type I11 pitting. In practice, a water temperature of -55°C is recommended
as the best compromise; however, it is important to also consider regulations
for the control of Legionella. In the United Kingdom, the Health and Safety
Executive Document L8 requires all hot water systems to operate at 55°C
or greater to prevent the growth of Legionella.
Bicarbonate can be added to water by acidifying a sidestream with
carbon dioxide, passing the water through dolomite limestone, blending it
back into the main supply, and adjusting the pH. In severe cases, all the
water might require treatment through a limestone bed, which increases
costs.
Where the manganese content of a water is high and is contributing
to Type I1 attack, as described in Section 5.2, it may be necessary to
include a manganese filter, as well, to significantly reduce the manganese
concentration. Commercial units to do this are available.
Where recirculating hot water is being dosed with chlorine to prevent
Legionnaires' disease, there is a clear risk of increasing the susceptibility
to Type I1 pitting, as described in Section 5.2. Hamamoto et al. carried out
tests and showed that the propensity for Type I1 pitting was not increased,
provided the chlorine residual did not exceed 0.25 mg/L.69
Kasahara et al. investigated the use of ultraviolet irradiation of hot
water circuits to prevent the growth of ba~teria.~' They found that bacteria
could be controlled by ultraviolet irradiation and a chlorine residual of
0.2 mg/L, compared with a residual of 5 to 10 mg/L without ultraviolet
treatment. At the low chlorine concentration with ultraviolet treatment, no

pitting occurred in hot Tokyo water.
5.8.3 Type 111 Pitting

Although the large outbreaks of Type I11 pitting in the 1980s led to a great
deal of research, there is less agreement on methods of prevention. Where
bacterial action does not appear to be important, Linder and Lindman rec-
ommended increasing the bicarbonate content of the water to >70 mg/L.45
This can be done as described in Section 5.8.2. In addition, increasing
the pH above 8 is also beneficial as it removes most of the excess carbon
dioxide.
Where microbial action contributes to the problem, Fischer et al. re-
viewed the methods adopted in a German hospital to reduce the incidence
of Type I11 ~ i t t i n g . ~With
' the hot recirculated water system, it was rec-
ommended that the water temperature be maintained in the range 55 to
60°C. The bacteria do not thrive at 55"C, but Type I1 pitting may occur if
the temperature goes above 60°C. In addition, dead legs should be avoided
wherever possible, and the water should be pumped to avoid stagnant,
cooler zones, where bacteria might thrive.
Cleaning of the corroded pipes with dilute acid did not prevent reini-
tiation of pitting on refilling with water." Irradiation with ultraviolet light
did not prevent bacterial growth. Filtering of the water reduced bacterial
growth, but did not prevent it, as was confirmed by B0nd.4~
Increasing the bicarbonate content of the water reduced the incidence
of attack. The higher the bicarbonate content, the better, provided that the
bicarbonate-sulfate ratio exceeded 1.71
5.8.4 Alloy 400

Alloy 400 is most susceptible to pitting in quiescent water, and if the flow
rate can be kept to 1 m/s or more, the alloy will perform well. In the presence
of chlorine, it is important to keep the chlorine concentration to around
0.5 mg/L to avoid initiating pitting. In the presence of sulfide, pitting may
occur at any velocity if the sulfide concentration is sufficiently high; for
such cases, an alternative material is recommended.
5.8.5 Formicary Corrosion

Most of the incidences of formicary corrosion have involved packing
or insulation that contained formaldehyde and the presence of a moist,
oxidizing atmosphere, which could oxidize the formaldehyde to formic

acid. The problem can be avoided by the use of packing or insulation ma-
terials that do not contain formaldehyde. Similarly, compounds that can
oxidize to form acetic acid should also be avoided. In addition, packing
should be done so that the contents remain dry until required, by the use of
desiccants or water-absorbing gels.
5.8.6 Rosette Corrosion

This problem is best avoided by the use of the new design of hot water
cylinder that heats the water above 40°C even at the cylinder base.60 Even
in an aggressive water, an aluminum protector rod is not necessary with
such a design.60 Where this type of cylinder is not available, then one
with an aluminum protector rod should only be used in waters known to
cause Type I pitting. Although magnesium anodes have been used in some
countries, aluminum is now the preferred material because of its lower cost
and longer life.
References
1. H.S. Campbell, J. Inst. Met. 77 (1950): p. 345.
2. H.S. Campbell, J. Appl. Chem. 4 (1954): p. 633.
3. H.S. Campbell, Proc. Soc. Water Treat. Exam. 3 (1954): p. 100.
4. H.S. Campbell, “A Review: Pitting Corrosion of Copper and its Alloys,” BNF
Miscellaneous Publication, MP574 August 1972. (This is an expanded version of
a paper presented at the U R Evans International Conference on Localized
Corrosion, Williamsburg, VA, USA, December 1971.)
5. M. Edwards, J.F. Ferguson, S.H. Reiber, J. Am. Water Works Assoc. 87 (1994):
p. 74.
6. P. Devroey, C. Depommier, unpublished report, Visseries et Trefileries Reunies
S.A., 1965.
7. H.J. Boorsma, C.H.J. Elzenga, H. Nijholt, unpublished report, KIWA.
8. F.J. Cornwell, G. Wildsmith, P.T. Gilbert, Brit Corros. J. 8 (1973): p. 202.
9. M. Pourbaix, J. van Muylder, P. van Laer, N. de Zoubov, CEBELCOR Rapport
Technique, 125, 1965.

14. V.F. Lucey, Brit. Corros. J. 2 (1967): p. 175.
15. R. Riedl, J. Klimbacher, Int. J. Muter: Product Tech. 4 (1989): p. 159.
16. P. Munn, “Effect of Phosphates on the Corrosion of Copper in Domestic Hot and
Cold Water Systems,” U.K. Corrosion Conference (Sheffield, UK: Institute of
Corrosion, 2007).
17. P. Devroey, C. Depommier, unpublished report, Visseries et Trefileries Reunies
S.A., 1962.
18. M.F. Obrecht, W.E. Sartor, J.M. Keyes, 4th International Congress on Metallic
Corrosion (Amsterdam, Holland, 1969), p. 576.
19. V.F. Lucey, BNF Metals Research Association Research Report, A1692, June
1968.
20. V.F. Lucey, BNF Metals Research Association Research Report, RRA1838,
December 1972.
21. M. Edwards, J. Rehring, T. Meyer, Corrosion 50 (1994): p. 366.
22. S. Jacobs, S. Reiber, M. Edwards, J. Am. Water Works Assoc. 90 (1998):
p. 62.
23. R.B. Diegle, W.E. Berry, INCRA Project Final Report, 272, April 1980.
24. P.D. Goodman, V.F. Lucey, C. Maselkowski, The Use of Synthetic Eiwironmeizts
for Corrosion Testing STP 970 (Philadelphia, PA: ASTM International, 1988),
p. 165.
25. R. May, J. Inst. Met. 32 (1953): p. 65.
26. M. Sosa, S. Patel, M. Edwards, Corrosion 55 (1999): p. 1069.
27. C. Breckon, J.R.T. Baines, J. Inst. Marine Eng. 67 (1955): p. 1.
28. R. Retief, Brit. Corros. J. 8 (1973): p. 264.
29. D.D. Marsden, M P 17 (1978): p. 9.
30. M. Grassiani, “The Effect of Carbon Residue Films on Localised Corrosion of
Brass Condenser Tubes,” International Congress on Metallic Corrosion, vol. 4
(Toronto, Canada, 1984), p. 206.
3 1. S. Sato, K. Nagata, S. Yamauchi, Sumitomo Light Met. Tech. Rep. 20 ( 1 979):
p. 107.
32. E.B. Shone, G.C. Grimm, Trans. Inst. Marine Eng. 98 (1985): Paper I 1.
33. R. Francis, BNF Metals Technology Centre Research Report, R461/4, October
1985.
34. S. Sato, T. Minamoto, K. Seki, H. Yamamoto, T. Takizawa, S. Yamauci, Y.

Hisamatsu, I. Suzuki, T. Fuji, T. Kodama, H. Baba, K. Nawata, “Case Studies on
Pitting Corrosion Failures of Copper Tubes in Hot Waters,” International
Symposium on Corrosion of Copper and Copper Alloys in Buildings (Tokyo,
Japan: CDA, 1982).
35. E. Mattson, A.M. Fredriksson, Brit. Corros. J. 3 (1968): p. 246.
36. K. Kasahara, S. Komukai, Y. Konishi, T. Fujiwara, “An Investigation of Cases of
Pitting Corrosion in Copper Pipes of Centralised Hot Water Systems,” Annual
Report of the Tokyo Gas Research Institute, 1986, p. 213.
37. H. Baba, T. Kodama, T. Fuji, Y. Hisamatsu, Y. Ishikawa, Boshoku Gijutsu (Corros.
Eng.) 30 (1981): p. 113.
38. H. Baba, T. Kodama, T. Fuji, Y. Hisamatsu, Boshoku Gijutm (Corros. Eng.) 30
(1981): p. 161.
39. I. Suzuki, Y. Ishikawa, Y. Hisamatsu, Corros. Sci. 23 (1983): p. 1095.
40. H. Baba, T. Kodama, T. Fuji, Trans. Natl. Res. Znst. Met. 28 (1986): p. 42.
January 1991.
42. T. Hamamoto, M. Kumagai, K. Kawano, S. Yamauchi, Sumitorno Light Met. Tech.
Rep. 28 (1987): p. 16.
43. R. Francis, BNF Metals Technology Centre Research Proposal, PR963/6, January
1990.
44. 0. von Franque et al., Werkst. Korros. 26 (1975): p. 4.
45. M. Linder, E.-K. Lindman, “Investigation of Type I11 Pitting Corrosion in Copper
Pipes,” Ninth Scandinavian Corrosion Congress (Stockholm, Sweden: Swedish
Corrosion Institute, 1983).
46. H.S. Campbell, A.H.L. Chamberlain, P.J. Angel], “An Unusual Form of
Microbially Induced Corrosion in Copper Water Pipes,” in Corrosion and Related
Aspects of Materials for Potable Water Supplies, P. McIntyre, A.D. Mercer, eds.
(London: Institute of Materials, 1993), p. 222.
47. W. Fischer, I. Hanssel, H.H. Paradies, Microbial Corrosion, vol. 1 , Conference
held in Sintra, Portugal (New York, N Y Elsevier, 1998), p. 300.
48. J.L. Nuttall, ‘Topper,” in Corrosion and Related Aspects of Materials,for Potable
Water Supplies, P. McIntyre, A.D. Mercer, eds. (London, U.K.: Institute of
Materials, 1993), p. 65.
49. S. Bond, BNF Metals Technology Centre Research Report, R688/6, December
1991.
50. C.W. Keevil, Water Sci. Technol. 49 (2004): p. 91.
51. H. Siederlarek, D. Wagner, M. Kropp, B. Fussinger, I. Hanssel, W.R. Fischer,
“Effects of Water Composition and Operating Conditions on the Corrosion
Behaviour of Copper in Potable Waters” in Corrosion and Related Aspects of

Materials for Potable Water Supplies, P. McIntyre, A.D. Mercer, eds. (London,
U.K.: Institute of Materials, 1993), p. 122.
52. P.A. Klein, R.A. Hays, R.J. Ferrara, R.M. Kain, CORROSION/91, paper no. 509
(Houston, TX: NACE International, 1991).
53. D.G. Tipton, R.M. Kain, CORROSIONI80, paper no. 36 (Houston, TX: NACE
International, 1980).
54. P. Gallagher, A. Nieuwhof, R.J.M. Tausk, “Experiences with Sea Water
Chlorination on Copper Alloys and Stainless Steels,” in Marine Corrosion of
Stainless Steels-Chlorination and Microbial Effects, European Federation of
Corrosion Publication 10 (London, U.K.: Institute of Materials, 1993), p. 73.
55. J.M. Keyes, “Pinhole Pitting of Copper Tube in Storage,” International Copper
Research Association Symposium on Pitting Corrosion of Water Tubes (Brussels,
Belgium, 1965).
56. J.O. Edwards, R.I. Hamilton, J.B. Gilmour, MP 16 (1977): p. 18.
57. S. Yamauchi, K. Nagata, S. Sato, B. Shimono, Sumitorno Light Met. Tech. Rep. 25
(1984): p. 1.
58. T. Notoya, T. Hamamoto, K. Kawano, “Unusual Corrosion in Copper Tubes,” Step
into the 90’s Conference (Queensland, Australia, 1989).
59. A. Olszewski, R. Corbett, M P 46 (2007): p. 52.
60. R. Oliphant, “Causes of Copper Corrosion in Plumbing Systems,” Foundation for
Water Research Review, FR/R0007, May 2003.
61. EN 1057, “Copper and Copper Alloys, Seamless, Round Copper Tubes for Water
and Gas in Sanitary and Heating Applications” (Brussels, Belgium: EN).
62. BS 2871 Part 3, “Specification for Copper and Copper Alloys. Tubes. Tubes for
Heat Exchangers” (London, U.K.: BSI).
63. S. Bond, R. Francis, BNF Metals Technology Centre Research Report, R660/8,
November 1991.
65. D. Shaw, V.F. Lucey, BNF Metals Technology Centre Research Report, A1956,
June 1979.
66. D. Shaw, V.F. Lucey, INCRA Project Final Report, 302, 1984.
67. F. Ansuini, ICA Project Final Report, 442, July 1991.
68. C. Depommier, ATB Metull. 14 (1984): p. 387.
69. T. Hamamoto, M. Kumagai, K. Kawano, S. Yamauchi, Sumitorno Light Met. Tech.
Rep. 28 (1987): p. 70.
70. K. Kasahara, S. Komukai, T. Fujiwara, Corros. Eng. 37 (1988): p. 361.
71. W.R. Fischer, D. Wagner, H. Siedlarek, M P 34 (1995): p. 50.
CHAPTER 6
Crevice Corrosion
6.1 Mechanism
Crevice corrosion occurs when part of the metal surface is shielded from
the bulk environment. This may be because of deposits, or the crevice may
be engineered such as those on a flange joint or under a washer on a bolt.
Unlike the oxygen concentration cell that is required to initiate crevice
corrosion on stainless steels, crevice corrosion of copper alloys depends
on a copper ion concentration cell. As the metal corrodes, copper ions are
released into solution, and in the bulk environment, they are either swept
away or they precipitate as a copper compound such as copper oxide (Cu20)
or copper hydroxychloride ( C U ~ C I [ O H ] ~ . ~ HIn~the
O )creviced
. area, there
is a shortage of reactants to form copper compounds, and the copper ion
concentration increases, making the metal in the creviced region cathodic
to the metal outside the crevice. Corrosion occurs at the anodic region just
outside the crevice, often along much of its length.
Figure 6.1A provides an example of crevice corrosion on 70/30
copper-nickel after exposure in seawater, followed by cleansing. Its appear-
ance is similar to that of most copper alloys experiencing crevice corrosion.
Apart from the location just outside the crevice, the appearance is very sim-
ilar to pitting corrosion in seawater (see Chapter 5). In the example, most of
the crevice sites are where barnacles were attached to the plate during the
60-day exposure to slowly moving seawater. Figure 6.1B shows two plates
of Alloy 400 that had been exposed for 6 months in quiescent seawater with
'
nonmetallic annular crevice formers, as described by Oldfield. Like 70/30
copper-nickel, the attack is just outside the crevice, but pitting has also
occurred on openly exposed surfaces. Alloy 400 is susceptible to this type
of attack in stagnant or slowly moving seawater, as described in Chapter 5.
95
FIGURE 6.1 Appearance of crevice corrosion on (A) 70/30 copper-nickel and (B) Alloy
400, exposed to quiescent seawater. (Photographs courtesy of R.M. Kain.) (See color
section.)
Crevice Corrosion 97
The concentration of copper ions in the crevice provides the cathodic

reaction, and redeposited, metallic copper is often seen in the shielded area
when crevice corrosion is occurring, that is,
Cu+ + e + Cu. (6.1)
The supply of copper ions for this reaction is limited and will soon be-
come exhausted. However, once crevice corrosion has initiated, it is self-
sustaining, and if the crevice is removed, the corrosion will continue under
the accumulated corrosion products. The corrosion now looks somewhat
similar to a Type I pit, as described in Chapter 5 , with a mound of green,
basic copper salts on top of the pit and the cathode above the pit. The copper
ion concentration that was required to initiate attack is no longer necessary.
Although crevice corrosion is seen quite often in corrosion tests, it is rarely
a cause of service failure (see Section 6.4).
6.2 Fresh Water
Copper pipe is commonly used for fresh water distribution, particularly

in domestic premises. Upland waters and some river waters have a high
organic content and/or solids burden, and this can settle in stagnant or
slow-moving water regions.2 Soft waters and waters of low pH are also
more likely to cause corrosion of cast iron water mains, and thus deposits
may also contain iron corrosion products. These deposits create a shielded
area, and crevice corrosion can initiate on copper surfaces at the edge of
such deposits. However, propagation rates are low, and penetration of a
I-mm-thick copper tube usually takes 10 years or more if the attack is not
stifled.2
Flux residues from soldering may also cause crevice corrosion, but
this can be much more rapid.2 This is because older-style fluxes contain
ammonium (NH4CI) and/or zinc chloride (NaC12), and this means that
conditions beneath a flux deposit can be very acidic, hence the high rate of
corrosion. Modern fluxes for potable water pipes must be free of chlorides
and water-soluble and so are less corrosive; however, some flux residues
dissolve only slowly, and crevice corrosion like that seen with other deposits
may occur.
6.3 Seawater
There is more data on crevice corrosion of copper alloys in seawater
and modified seawater than in any other medium. Kain reviewed the test
TABLE 6.1 Depth of crevice corrosion after 180 davs in natural seawater'
Crevice Max. depth (mm)

Alloy Condition former Finish Flow(A) A B
M Cast Gasket Phonograph 0.34 0.32
Bronze Cast Gasket Phonograph 0 0
NAB Cast Gasket Phonograph 0.60 t0.0 1
Cast Gasket Ground <0.0 1 <0.01
70130 Cast Gasket Phonograph <0.0 I 0.2s
Cu-Ni Cast Gasket Phonograph 0.05 tO.O I
70130 Wrought PTFE Ground 0.10 0.04
Cu-Ni Wrought PTFE Ground 0.17 0.11
90110 Wrought Gasket Phonograph I .05 1.28
Cu-Ni
Alloy Wrought PTFE Ground Q 0.67 0.69
400 Wrought PTFE Ground F 0.2s 0.27
(A1
Q, quiescent; F, flowing (1.6 rn/s).
methods available to study crevice corrosion, including the parameters that

should be considered and the various methods of evaluating the results.3
The U.S. Navy conducted a study of the crevice corrosion resistance
of newer alloys, and as part of that study, they included the copper alloys
currently used in U.S. naval vessel^:^ M bronze (C92200), cast nickel alu-
minum bronze (NAB), 90/10 and 70/30 copper-nickels, and Alloys 400
and K-500. Tests were conducted in both flowing seawater and quiescent
conditions using crevice assemblies consisting of rubber gasket material
under metal washers and also polytetrafluorethylene (PTFE) crevice form-
ers. Surface finish of the test samples was also a variable that was examined.
Some of the results are presented in Table 6.1 ; they show that crevice cor-
rosion was much deeper under quiescent conditions and was negligible
for M bronze, NAB, and 70/30 copper-nickel under flowing conditions.
Note that the crevice geometry and, in particular, the cathode to anode
ratio, were different in flowing and quiescent conditions. The cathode area
on the crevice assemblies exposed to quiescent seawater was much larger
than that used on the samples in flowing conditions, and this would be
expected to produce more severe crevice corrosion, as seen in Table 6.1.
Although there was deep attack on 90/10 copper-nickel under quiescent
conditions, no data were generated under flowing conditions. The results
for 70/30 copper-nickel show that a PTFE crevice former creates a more
severe crevice than one with a rubber gasket. This is also the case for stain-
less steels, but the reason for this, a smaller crevice width, does not explain
the reactions on copper alloys, where the crevice region is the ~ a t h o d e . ~

It is possible that the greater depth of attack with a PTFE crevice former
was due to an earlier transition to a mechanism in which the reduction of
dissolved oxygen (DO) was the cathodic reaction. The depth of attack was
generally more severe for Alloy 400 than the copper-rich alloys, with the
exception of 90/10 copper-nickel.
The author conducted a series of tests of copper alloy condenser
tubes utilizing the Campbell condenser tube test rig, which contains both
crevices exposed to cold water and crevices exposed to warm water (see
Section B.2 in Appendix B).6-8The alloys under test were aluminum brass,
904 0 copper-nickel, 70/30 copper-nickel, 66/30/2/2 copper-nickel-iron-
manganese, and Alloy 722. In tests investigating the effects of chlorine and
ferrous sulfate additions to seawater, there was no significant crevice corro-
sion in either region at all chlorine concentrations up to 4 mg/L continuous.6
When ammonia (NH3) was present, corrosion occurred in the warm water
crevice.’ As this was more akin to dealloying than crevice corrosion, it is
described in detail in Chapter 9.
When sulfide was added to simulate polluted waters, the effects were
variable.8 At low levels of sulfide (0.01 mg/L), there was an increased ten-
dency for crevice attack to occur outside the crevice on all alloys, presum-
ably because the sulfide hindered the formation of protective films outside
the crevice. At higher concentrations of sulfide (0.03 and 0.1 mg/L), there
was no classical crevice corrosion, but general corrosion and pitting oc-
curred in the crevice. It appears that the sulfide in the seawater prevented
the establishment of a copper ion concentration cell by causing copper
sulfide (CuS) to precipitate.
In an investigation of the variability of seawater on corrosion, Phull
et al. reported the results of corrosion tests of 90/10 copper-nickel exposed
at 14 seawater sites around the world.’ The test panels were bolted to the
supporting frame with a nylon bolt and washer, which created four crevices
on each panel. The results are summarized in Table 6.2. It should be noted
that samples from seven of the sites showed no crevice corrosion at any
exposure time. Also, the 3- and 5-year results from California could not
be quantified because erosion corrosion had occurred in addition to crevice
corrosion. The depth of crevice corrosion was very variable, with half the
sites reporting none. While pitting away from the crevices was seen on
panels exposed at all the sites, it was often quite shallow. The variability of
the attack shows that it is influenced not so much by seawater conditions
(temperature, DO, etc.) as by local conditions such as sulfides, deposits,
and so on.
TABLE 6.2 Maximum depth of crevice corrosion on 90/10 copper-nickel

after up to 5 years’ exposure in natural seawater9
Max. depth (mm)
Exposure time
Site(A) (years) A B
Wrightsville Beach, 0.5 0 0
NC, East Coast, USA 1.o 0 0
3.0 0 0
5 .o 0.05 0.01
Freeport, Texas, 0.5 0 -
Gulf Coast. USA 1 .o 0.02 0.07

3.0 0.62 0.52
5.0 0.16 0.7 1
Port Hueneme, 0.5 0 0
California, West Coast, 2.0 1.3 0.8
USA 3.0 (B) (B)
5.0 (B) (B)
Peru 5.0 0.10 0.33

Australia 1.o 0 0.26
3.0 0.33 2.06
5.0 1.01 0.10
Denmark 0.5 0.1 1 0.05
1.5 0.20 0.30
3. I 0.3 1 0.30
5.0 0.36 0.49
United Kingdom 0.5 0 0
1.o 0 0
3.0 0.40 0.24
5.0 0.60 0.55
(A)
No crevice corrosion found on panels exposed at Ocean City, Maryland; Key West, Florida;
Hawaii; Japan; Italy; and Sweden (two sites).
(B)
Depth not reported because of erosion corrosion damage at crevice site.
Klein et al. examined the effect of chlorine additions to seawater on

the crevice corrosion of 70/30 copper-nickel and Alloy 400 using vari-
ous crevice formers on tube and plate under both flowing and quiescent
conditions.l o Replicate samples were exposed to natural seawater, seawater
with 1 mg/L chlorine (to simulate a normal antifouling regime), or seawa-
ter with 90 m g L chlorine (to simulate a concentrated sodium hypochlorite
[NaOCl] line). The results for 70/30 copper-nickel and Alloy 400 under
quiescent conditions are shown in Tables 6.3 and 6.4, respectively. Note
that there were cathode to anode area ratio differences between samples ex-
posed to quiescent seawater and flowing seawater, as described previously?
which explains the more severe attack under quiescent conditions.
TABLE 6.3 Maximum depth of crevice corrosion of 70/30 copper-nickel under

quiescent conditions with different crevice formers"'
Depth of attack (mm)
Material Crevice Chlorine
form former (mg/L) 45/60 days(A) I 80 days
Tube Vinyl 0 0.01 0.07
sleeve 1 0.0 1 0.03
90 0.05 0.05
Buna N 0 0.01 0.08
sleeve 1 0.01 0.01
90 0.08 0.13
O-ring 0 0.01 0.07
1 0.01 0.03
90 0.03 0.07
Plate Polyacetal 0 0.01 0.02
annulus 1 0.01 0.02
90 0.15 0.07
PTFE 0 0.01 0.02
annulus 1 0.01 0.01
90 0.06 0.06
Nylon 0 0.01 0.05
annulus 1 0 0.02
90 0.09 0.10
PTFE 0 0.0 1 0.0 1
O-ring 1 0.01 0
90 0.01 0.06
Rubber 0 0.01 0
O-ring 1 0.01 0
90 0.01 0
(A)
Forty-five days for plate and 60 days for tube samples.
The results for 70/30 copper-nickel show that very little crevice cor-
rosion occurred in natural seawater, and this was reduced even more in
the presence of 1 mg/L chlorine. However, 90 mg/L chlorine increased the
attack to the same as that seen in natural seawater or more. Despite the dif-
ferences in product forms and crevice geometry, there was no discernible
difference between the results for any of the crevice formers on tube or
plate.
With Alloy 400, there was deep attack in natural seawater, which was
reduced in the presence of 1 mg/L chlorine, but there was still a significant
depth after 180 days' exposure. The addition of 90 mg/L chlorine made the
conditions more aggressive-about the same as seen in natural seawater.
This is presumably because small additions of chlorine reduce the potential
below the pitting potential, while high levels of chlorine take it back above
TABLE 6.4 Maximum depth of crevice corrosion of Alloy 400 under quiescent
conditions with different crevice formers"'
Material Crevice Chlorine
form former (mg/L) 45/60 days(A) 180 days
Tube Vinyl 0 0.27 0.56
sleeve 1 0.01 0.20
90 0.20 0.24
Buna N 0 0.23 0.08
sleeve 1 0.01 0.1 1
90 0.25 0.26
O-ring 0 0.27 0.49
1 0.02 0.25
90 0.3 1 0.38
Plate Pol yacetal 0 0.4 1 0.76
annulus I 0.08 0.34
90 0.36 0.67
PTFE 0 0.41 0.68
ann u1us 1 0.07 0.30
90 0.35 0.70
Nylon 0 0.38 0.68
annulus 1 0.07 0.3 1
90 0.33 0.65
PTFE 0 0.10 0.23
O-ring 1 0.08 0.19
90 0.12 0.20
Rubber 0 0.08 0.19
O-ring I 0.0 I 0.05
90 0.06 0.15
(A)
Forty-five days for plate and 60 days for tube samples.
this potential." There was no real difference between the different crevice
formers on tube samples, but the O-rings on plate samples were a little
less severe than the annular crevice formers. One observation for Alloy 400
was that the rate of propagation of crevice corrosion in natural seawater
decreased with time, while it was more or less constant in chlorinated
seawater. Another observation was that deep pitting occurred on Alloy 400
in natural seawater, but there was hardly any in chlorinated seawater. Deep
pitting also occurred with 90 mg/L chlorine. The effect of chlorine on the
pitting behavior of Alloy 400 is discussed in Chapter 5.
Under flowing conditions, the results were a little different, as shown
in Table 6.5. For 70/30 copper-nickel, there was deeper attack in flowing,
natural seawater than under quiescent conditions, presumably because there
TABLE 6.5 Maximum depth of crevice corrosion of 70/30 copper-nickel

and Alloy 400 tubes under flowing (1.6 m/s) conditions“’
Crevice Chlorine
Alloy former (mglL) 60 days 180 days 360 days
70130 Vinyl sleeve 0 0. I - 0.36
Cu-Ni 1 tO.O1 - 0.02
90 0.48 - Perforated
Buna N 0 0.01 0.21 -
sleeve 1 tO.O1 0.02 -
90 0.7 I .5 -
Nylon 0 to.01 - 0.28
compression I tO.O1 - 0.03
fitting 90 0.27 - 1.01
Alloy 400 Vinyl sleeve 0 0.37 0.63
1 0.1 0.45
90 1.7 - Perforated
Buna N 0 0.36 I .4 -
sleeve 1 0.1 1 0.27

90 1.7 1.95 -
Nylon 0 0.05 0.43
compression 1 0.02 0.56
fitting 90 0.04 0.17
was a more steady supply of DO to the cathode to prevent polarization. In

seawater with 1 mg/L chlorine, the attack was reduced to very low levels.
This suggests that the reduction of chlorine to chloride is a less efficient
cathodic reaction than the reduction of DO in the presence of a biofilm.”
With 90 mg/L chlorine, the crevice corrosion was much deeper than in
natural seawater, with perforation, or near perforation, of the tubes. Again,
there were no discernible differences between the different crevice formers
for 70/30 copper-nickel.
The results followed the same trend for Alloy 400 as for 70/30 copper-
nickel, but with much deeper attack. Even though the depth of attack with
1 mg/L chlorine was reduced, the depth of attack was still significant. The
nylon compression fitting appeared to be a less aggressive crevice former
for this alloy compared with the sleeves.
Kain et al. examined the possible use of 70/30 copper-nickel for
reverse osmosis (RO) desalination plants.13 Samples were exposed under
three separate conditions. The first was filtered seawater (- 19,000 mg/L
chloride), as found in the high-pressure section of an RO pilot plant. The
second was water with 200 to 2,000 m g L chloride, as found after partial
TABLE 6.6 Maximum depth of crevice corrosion on 70/30copper-nickel

exposed in various parts of a reverse osmosis plant"
Medium Chloride (mglL) Attacked region Depth of attack (mm)
Filtered seawater 19,000 All sites <0.01
Second-pass 200-2,000 PCA 0.04
brine Rubber gland 0.04
20.6-mm O-ring t0.01
23.8-mm O-ring 0.08
Third-pass t0.02 PCA <o.o I
permeate 23.8-mm O-ring tO.0 1
Rubber gland 0
20.6-mm O-ring 0
treatment of the seawater. The third condition was water with <0.2 m g L
chloride, as found in the final stage of water treatment. The tests lasted for 90
days, with O-ring crevice formers on tubes and acrylic crevice assemblies
(PCA) for plate samples, as described by Oldfield.' All the samples were
exposed in empty RO membrane housings.
The results, in Table 6.6, show the depths of attack at the various
crevice sites. There was negligible crevice corrosion at all the sites, in both
filtered seawater and the final permeate. In the second-pass permeate, with
moderate chloride concentrations, there was deeper attack at most of the
crevice sites. It is not clear why this intermediate concentration of chlorides
should be more aggressive in terms of crevice corrosion initiation. Another
fact that comes out of Table 6.6 is that the looser O-ring produced deeper
attack than the tighter O-ring. This is probably because the cathode under
an O-ring is much smaller for a tight O-ring and so cannot produce an ion
concentration cell to supply much current to create deep attack.
Kain and Weber examined the effects of lay-up conditions on the
crevice corrosion resistance of 70/30 copper-nickel and Alloy 400 under
RO ~0nditions.l~ The samples were tubes and plates creviced as described
earlier.I3 The exposure cycle was 500 hours of flowing seawater, followed
by 500 hours in either stagnant, filtered seawater or stagnant lay-up solution,
which consisted of 1% sodium metabisulfite in water containing 1,000m g L
chloride. The samples were examined after one and three cycles of exposure,
and Figure 6.2 shows the appearance of the 70/30 copper-nickel samples
after one and three cycles. After the first cycle, the tube was largely covered
in a black sulfide film, and the crevice corrosion where the O-rings were
located is clearly visible. After three cycles, there were also some green
and brown corrosion products, with the crevice attack at the O-rings again
still clearly visible. The depths of attack are shown in Table 6.7.
FIGURE 6.2 Appearance of 70/30copper-nickel tubes with O-ring crevices after one and
three cycles of 500 hours in flowing seawater followed by 500 hours of stagnant seawater.
(Photographs courtesy of R.M. Kain.) (See color section.)
TABLE 6.7 Depth of crevice corrosion after exposure to alternating flowing

and stagnant c~nditions'~
Crevice
Alloy No. of cycles former(A) Bisulfite lay-up Seawater lay-up
70130 1 PCA'B' tO.O1 to.01
Cu-Ni Rubber gland tO.O1 tO.0 1-0.02
O-ring <0.01 tO.0 1 -0.0 1
3 PCA'B' 0.05 tO.O1
Rubber gland <0.01-0.27 tO.0 1-0.08
O-ring 0.02-0.15 tO.0 1-0.03
Alloy 400 1 PCA(B' 0. I4 0.28
Rubber gland t0.01-0.32 tO.0 1-0. I5
O-ring 0.0 1-0.30 0.05-0.30
3 PCA(B' 0.22 0.24
Rubber gland 0.01-0.46 0.01-0.29
O-ring 0.15-0.56 0.04-0.40
(A)
PCA is on plate; gland and O-ring are on tubes.
(B)
PCA, perspex crevice assembly.'
The attack on the 70/30 copper-nickel was generally shallow, but

some deeper pits developed over three cycles with the bisulfite solution
during lay-up. The attack was deeper on Alloy 400, but there was no
significant difference between depths after one and three exposure cycles.
This suggests that the bulk of the attack occurs early during exposure and
the rate of further propagation is low. There was no particular difference in
the severity of attack on Alloy 400 with either lay-up solution. There was
also no particular difference between the different crevice formers for both
alloy 400 and 70/30 copper-nickel. The stagnant seawater lay-up did not
produce much sulfide, which might be expected to have a deleterious effect
on the resistance of these alloys to crevice corrosion.14
Wharton and Stokes investigated crevice corrosion of NAB in both
3.5% sodium chloride (NaC1) solution and natural seawater.15 In sodium
chloride, they found conventional crevice corrosion outside the crevice to a
modest depth, but much more severe attack along the kappa I11 phase in the
acidic region within the crevice. This is discussed in more detail in Chapter
9. Wharton and Stokes also found selective attack of the kappa I11 phase at
the base of the crevice corrosion pit, where the acidic conditions make the
kappa I11 anodic to the alpha phase (Figure 6.3).15 The attack in seawater
was similar, but much deeper, with the crevice attack spreading horizontally
away from the crevice for some distance. This is because, once established,
the crevice attack is similar to Type I pitting, with the cathode above the
FIGURE 6.3 Cross sections of NAB after 6.5 months' immersion in 3.5% sodium chloride
solution: (A) attack at the edge of the crevice; (B) detail from Figure 6.3A showing attack
of kappa 111 phase at base of pit. (Photographs courtesy of J.A. Wharton.)
mound. This favors sideways propagation in high-conductivity media like

seawater.
6.4 Avoidance of Crevice Corrosion
Although crevice corrosion of copper alloys is a well-recognized phe-

nomenon, it rarely causes failures. In over 30 years as a corrosion engineer,
the author has never seen a failure of a copper alloy due to crevice corro-
sion, although crevice corrosion has been observed on some components
that failed for other reasons. Where silt or deposits may cause problems,
the fitting of an adequate filtration system is recommended. The exception
to this is NAB, but the failures are due to selective phase attack of the kappa
I11 phase (see Chapter 9), rather than true crevice corrosion.
When thin material is being employed in a critical application, it is best
to avoid a design that creates a crevice. Where a crevice is unavoidable, the
evidence is that very tight crevices are less likely to cause crevice corrosion
with copper alloys. Hence, greasing screwed joints or using mastic or
gaskets containing a copper corrosion inhibitor (e.g., benzotriazole) can be
considered for critical applications.
References
1. J.W. Oldfield, Int. Met. Rev. 32 (1987): p. 153.
2. J.L. Nuttall, “Copper,” in Corrosion and Related Aspects of Materials for Potable
Water Supplies, P. McIntyre, A.D. Mercer, eds. (London, U.K.: Institute of
Materials, Minerals, and Mining, 1993): p. 65.
3. R.M. Kain, “Seawater Crevice Corrosion Testing of Stainless Steels, Ni-Base and
Cu-Ni Alloys: Perspectives on Methodologies and Interpretation of Results,”
EUROCORR 2006 (Maastricht, Holland: NCC/European Federation of Corrosion,
2006).
4. D.M. Aylor, R.A. Hays, R.M. Kain, R.J. Ferrara, CORROSION/99, paper no. 329
5. R. Francis, The Selection of Materials for Seawater Cooling Systems-A Practical
Guide for Engineers (Houston, TX: NACE International, 2006).
6. R. Francis, MP 2 1 (1982): p. 44.
9. B. Phull, S. Pikul, R.M. Kain, “Seawater Corrosivity Around the World; Results
from Five Years of Testing,” ASTM STP 1300 (1997): p. 34.
10. P.A. Klein, R.A. Hays, R.J. Ferrara, R.M. Kain, CORROSION/91, paper no. 509
(Houston, TX: NACE International, 199 1).
11. P. Gallagher, A. Nieuwhof, R.J.M. Tausk, “Experiences with Seawater
Chlorination on Copper Alloys and Stainless Steels,” in Marine Corrosion of
Stainless Steels-Chlorination and Microbial Effects, European Federation of
Corrosion Publication 10 (London, U.K.: Institute of Materials, Minerals, and
Mining, 1993).
13. R.M. Kain, W.L. Adamson, B. Weber, “Corrosion Coupon Testing in Natural
Waters: A Case History Dealing with Reverse Osmosis Desalination of Seawater,”
ASTM STP 1300 (1997): p. 122.
14. R.M. Kain, B.E. Weber, CORROSION/97, paper no. 422 (Houston, TX: NACE
15. J.A. Wharton, K.R. Stokes, Electrochim. Acta 53 (2007): p. 2463.
CHAPTER 7
Erosion Corrosion and Erosion
Copper and its alloys have good resistance to corrosion at moderate water
flows. However, when the local velocity is high enough, the local shear stress
becomes sufficient to remove the corrosion products. Metal dissolution
occurs, followed by film formation, which is again disrupted by the water
flow. This is termed erosion corrosion, or impingement attack, because
it often occurs where water flow impinges on the metal surface. This is
an electrochemical process because the application of cathodic protection
prevents attack.
The attack usually takes the form of pits that are often undercut on the
downstream side. Where the attack in a tube is severe, the attack often has
the appearance of horseshoe imprints, with the horse walking upstream, as
shown in Figure 7.1. The leading edge of the attack is frequently undercut
due to the swirling action of the water.
When solids are present in the water, then erosion can occur at veloc-
ities at which the metal loss would be low in the absence of solids. Section
7.1 discusses erosion corrosion of copper and its alloys, whereas Section
7.2 deals with erosion in the presence of solids. Section 7.3 offers some
advice on mitigating attack due to both of these corrosion mechanisms. The
data presented come from a variety of sources, using several different test
methods, from flow loops to high-velocity jet test rigs. The different meth-
ods of carrying out such tests and their relevance to service performance
are discussed in Appendix B.
111
FIGURE 7.1 Erosion corrosion of 15-mm copper pipe after a 90-degree elbow. (Photo-
graph courtesy of P. Munn.) (See color section.)
7.1 Erosion Corrosion
7.1.1 Fresh Water

In fresh waters, the most commonly used copper alloys are copper and
brass (both single-phase and duplex), with gunmetal being used mostly for
pumps and valves and, occasionally, for fittings. Other copper alloys have
only limited use.
There has been little published data on erosion corrosion in fresh
waters, but this type of attack is sometimes thought to be more common
in soft waters, where calcium carbonate (CaC03) cannot form. Myers and
Obrecht carried out extensive studies in a sodium zeolite-softened well
water, whose composition is shown in Table 7.1.' They tested tubes made
of copper, red brass (C23000), admiralty brass, and 90/10 copper-nickel
over a range of temperatures and velocities. Tests were conducted in three
waters: fully softened, a blended water with intermediate hardness, and
softened water with reduced dissolved oxygen (DO) and increased carbon
dioxide (CO2) concentrations. There is insufficient space to present all their
findings here, but some of their data and the main conclusions are included.
Erosion Corrosion and Erosion 113
TABLE 7.1 Composition of raw well water used

for erosion corrosion tests'
Concentration (mg/L)
Calcium As CaC03 2 10-250
Magnesium As CaC03 100-120
Sodium As Na 6-9
Bicarbonate As CaC03 300-325
Sulphate AS SO4 14-24
Chloride As C1 6-8
Free carbon dioxide As CO2 10-40
Oxygen AS 0 2 &I 2
Silica As S O 2 8-12
Total hardness As CaC03 3 10-370
Total alkalinity As CaC03 300-3 25
PH 6.8-7.5
Figures 7.2 and 7.3 show the maximum annual metal loss as a function
of velocity and temperature, respectively, for copper tube in fully softened
water. The data show that there is an increasing loss of metal with an increase
of velocity, but this is not excessive up to 60°C. At higher temperatures there
is a much greater metal loss with increasing velocity. At 9 3 T , the metal
0.35
-
h
0.3
--
0
0.25
--E
5
.-
0.2
0.15
0
n
.-
; 0.1
I
3
0.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Velocity (mls)
FIGURE 7.2 Maximum penetration rate vs. velocity for copper tube in softened water.'
- 0.3
x
P
0.25
J ,* 2.5 m/s
0.05
04
0 10 20 30 40 50 60 70 80 90 100
Temperature ("C)
FIGURE 7.3 Maximum penetration rate vs. temperature for copper tube in softened
water.'
loss is less than at 77"C, which is probably due to the low level of DO in
the water at the higher temperature. The rate of metal loss is then controlled
by the availability of DO to sustain the cathodic reaction, the reduction
of DO.
Figures 7.4 and 7.5 show the effect of velocity on red brass, admiralty
brass, and 90/10 copper-nickel at 10°Cand 60"C, respectively. These graphs
show that red brass behaves similarly to copper, while admiralty brass
suffers less attack at higher temperatures, and 90/10 copper-nickel suffers
even less. On the basis of these test data, Myers and Obrecht recommended
safe operating velocities for the alloys that they tested.' A summary of these
is shown in Table 7.2, which suggests that copper and red brass are suitable
TABLE 7.2 Recommended safe maximum velocities for copper and its alloys
in fresh water'
Alloy UNS no. Temperature range ("C) Max. velocity (m/s)
Copper c12200 0-60 1.2
Red brass C23000 0-60 1.2
Admiralty brass C44300 0-77 3.8
90/10 copper-nickel C70600 0-93 3.8
0.3
E 0.25
0
1
2 0.2 -
.-
.-s
E
g
1
0.15 - A
-.--_ ...-_ _ _
0
n
0.1 -
.-X -- -
=\
-
0 = - - - - - - - - - - -
2 0.05 - A x + *
07
- *-
- ---- ---A
I+ Red Brass - -W - Admiralty Brass -A- 90/10 Cu-Ni I

FIGURE 7.4 Effect of velocity on the erosion corrosion rate in soft water at 10°C.’
o o :
0.3
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Velocity (mk)
1- Red Brass - -E - Admiraltv Brass -* 90/10 Cu-Ni I

FIGURE 7.5 Effect of velocity on the erosion corrosion rate in soft water at 60°C.’
TABLE 7.3 The corrosion rates of some common alloys as a function

of velocity in seawater5
0-0.6 mls
Alloy UNS no. General Deepest pit 8 m/s 40 m/s

Carbon steel 0.075 2.0 1.o 4.5
Cast iron - 0.550 4.9 4.4 13.3
Gunmetal C83600 0.018 0.325 1.80 I .33
Admiralty gunmetal C90500 0.028 0.25 0.875 1.08
Ni-Al-bronze C96320 0.055 1.125 0.225 0.975
90/10 Cu-Ni C70600 t0.025 0.200 0.45 0.825
70/30 Cu-Ni C71500 <0.025 0.250 0.125 1.48
316L SIS S3 1603 t0.025 1.80 t0.025 0.005
Alloy 400 NO4400 t0.025 1.30 0.015 0.010
at moderate velocities up to 60°C. Oliphant made similar recommendations,

but suggested that whereas the maximum water velocity in hot water should
be 1.0 d s , in cold water, it can be increased to 2.0 m k 2 In pipes that
experience only intermittent flow, these maximum velocities can be relaxed,
provided that the off time is sufficiently long to enable film repassivation
to occur.
One location where erosion corrosion is commonly seen in copper
piping systems carrying fresh water is immediately after sharp bends or
elbows. If the fitting has been soldered into place, sometimes a protruding
solder bead increases the local turbulence. The erosion corrosion is always
on the copper and never on the solder, although the potential of solder is
significantly more electronegative than that of copper.3 Francis and Camp-
bell believed that this is because tin oxide is more stable and mechanically
strong than copper oxide (CuzO) in fresh waters. In unpublished work car-
ried out at BNF Metals Technology Centre, they exposed samples of copper
coated with solder in the May jet test at 10 m/s with 3% added air in a hard
borehole water containing 100 mg/L ~ h l o r i d eNo
. ~ measurable erosion cor-
rosion had occurred on the solder after 30 days at 15°C. These results
demonstrated the high resistance of solders to erosion corrosion in fresh
water.
7.1.2 Seawater
Copper and its alloys are more susceptible to erosion corrosion in seawater
compared with fresh water. This means that alloys that are more resistant
than copper and brass are usually chosen. Table 7.3 shows the effect of
70/30 Brass Admiralty Brass Al-Brass 90/10 Cu-Ni 70/30 Cu-Ni
FIGURE 7.6 Depth of attack of some wrought copper alloys in the jet test at 9.3 mls
and 18"C6
velocity on the corrosion rate of some common copper alloys, with the
rates for carbon steel, cast iron, Alloy 400, and Type 316L stainless steel
. ~ rate of attack of the cop-
(SS) (UNS 53 1603) provided for c ~ m p a r i s o nThe
per alloys increases with velocity, whereas those of Alloy 400 and 3 16L SS
remain low. However, both these alloys are susceptible to localized pitting
and/or crevice corrosion at low velocities. The corrosion rate of NAB in Ta-
ble 7.3 is not as low as might be expected. However, the composition tested
(C96320) contains only 2.5% iron, whereas the more common C95800
contains 4.0% to 4.5% iron and is more resistant to erosion corrosion (see
the following discussion).
Bog6 carried out jet impingement tests4 on a range of copper alloys at a
jet velocity of 9.6 m/s using recirculated seawater. The results for a selection
of common, wrought copper alloys are shown in Figure 7.6. The results
show the poor resistance of 70/30 brass and admiralty brass to erosion
corrosion and the excellent resistance of aluminum brass and the copper-
nickel alloys.
In later work, Bog evaluated the resistance of some cast copper alloys
to erosion corrosion by the same method at 20°C.' The results, given in
Table 7.4, show the benefit of increasing tin content on the performance
of gunmetals and bronzes and the poor performance of high-tensile brass
TABLE 7.4 Depth of attack in jet impingement tests of some cast

alloys at 9.6 m/s jet velocity and 20°C’
Alloy UNS no. Mean depth(mm)
Mn-bronze C86400 0.5
Gunmetal (LG2) (5% Sn) C83600 0.23
Gunmetal (LG4) (7% Sn) C92410 0.05
Gunmetal (Gl) (10% Sn) C90500 0.02
Leaded tin bronze (3% Pb) C92900 0.02
NAB C95800(*’ 0.0
(A)
Aluminum slightly below specification minimum.
(manganese bronze). The best performance came from nickel aluminum

bronze (NAB; C95800).
The copper-nickel alloys have good resistance to erosion corrosion,
and variations in their composition have been studied to optimize it. Bailey,
in his description of the development of 90/10 copper-nickel, showed that
the maximum resistance to erosion corrosion is found when the iron content
is 1.5 to 2.5 wt% (Figure 7.7).’ Efird looked at the synergistic effect of
nickel and iron on the susceptibility to erosion corrosion.9 He obtained
similar results to Bailey for 90/10 copper-nickel, and he found the optimum
iron content for 70/30 copper-nickel to be 0.8 to 1.2 wt%.
0 1 2 3 4 5
Iron Content (A%)
FIGURE 7.7 Effect of iron on the erosion corrosion resistance of 90/10 copper-nickel in
the jet test at 5 m/s8
These figures are interesting because the specifications (for example,

ASTM B 111 lo) for 90/10 and 70/30 copper-nickel (C70600 and C7 1500,
respectively) are 1.0% to 2.0% and 0.4% to 1.0%, respectively. Where
resistance to erosion corrosion is required, it is recommended that 90110
copper-nickel have a minimum iron content of 1.5 wt% and that 70/30
copper-nickel have a minimum iron content of 0.7 wt%.
The iron content of 90/10 copper-nickel exceeds the solubility limit
at room temperature. Bailey demonstrated that the iron must be in solid
solution to confer the best erosion corrosion resistance.8 This is achieved
commercially by annealing in the range 850 to 900°C and then cooling
’
rapidly to room temperature (for example, water quenching).’ If 90/10
copper-nickel is heated such that the iron comes out of solution, then the
erosion corrosion resistance decreases.*
In unpublished work at BNF, Guyoncourt and Harper aged copper-
nickel alloys for 16 hours at 645°C and also investigated cold rolling, either
before or after aging.” In jet impingement tests at 10°C and 9.5 m/s jet
velocity, the depth of attack was 0.11 mm in both the solution-annealed
and aged conditions for 90/10 copper-nickel. However, the area of attack
was much greater on all the aged samples. The most interesting effect of
aging was a fourfold increase in the general corrosion rate. With Alloy 722,
there was no attack in the solution-annealed or aged conditions and only
very slight attack (0.01 mm) on the cold-rolled samples. Alloy 722 contains
more nickel and less iron than 90/10 copper-nickel, and so the fact that it
does not lose corrosion resistance after thermal aging is not surprising.
Other research has also showed that aged 90/10 copper-nickel does
not have a significantly different erosion corrosion resistance compared
with solution-annealed material. Appel et al. conducted tests on pipes with
seawater flows of 3.3-7.5 m/s.13 The pipes had varying magnetic perme-
abilities from 1.O (solution annealed) to 1.6 (fully aged). Not only was there
no significant difference in erosion corrosion resistance between solution-
annealed and aged pipes, but also, there was no real difference in the general
corrosion rate. This is different from the results of Guyoncourt and Harper
and suggests that the nature or distribution of the precipitates is critical to
the subsequent performance.” Guyoncourt and Harper showed electron mi-
crographs with the size and distribution of the precipitates, but Appel et al.’s
paper showed only optical micrographs that do not properly characterize
the precipitates.
Although laboratory tests on aged 90/10 copper-nickel do not show
a large increase in susceptibility to erosion corrosion, service failures have
occurred. Frick et al. reported failures of a 90/10 copper-nickel piping
system in the North Sea after less than 12 months in service.14 The water
FIGURE 7.8 Selective corrosion of parent 90/10 copper-nickel pipe because most of the
iron was not in solution. (See color section.)
velocity did not exceed 2.2 d s , well within the design limits for this alloy,
and the iron content of the failed samples was 1.5% to 1.7%, within the
specification for 90/10 copper-nickel. The high rates of corrosion that were
seen in the vicinity of welds were attributed to much of the iron not being
in solution. This was confirmed by magnetic permeability measurements.
This is the reverse of what would be expected in welding, as the weld
metal and heat-affected zone (HAZ) cool quickly and tend to hold the iron
in solution. Corrosion of the welds suggests that they were welded with
incorrect filler metal.
The author reported a similar series of failures from an oil platform
in the North Sea.'' The system was suffering a very high rate of general
corrosion plus erosion corrosion after the bends, even though the water
velocity was only -3 m/s. The total iron content of the pipe was 1.7%. The
clue to the cause of the problem was at the welds: not only was the 70/30
copper-nickel weld metal unattacked, but neither was the parent metal in
the HAZ. Figure 7.8 shows a half-section of pipe, and the cut edge clearly
shows the thicker metal in the weld and HAZ. This was because the iron was
not in solution in the pipe, but the weld thermal cycles in the HAZ had re-
solutionized the iron and restored the corrosion resistance. The weld metal
was not affected because iron is more soluble in the 70/30 copper-nickel
filler metal. Determination of the magnetic permeability with a simple de-

vice, such as the Severnt magnetic gauge, showed the parent metal to have
a value of >1.2, whereas the weld metal and HAZ were <1.05. It is sus-
pected that the parent metal suffered severe general corrosion because the
water velocity was close to the recommended limit, and as the weld and
HAZ became increasingly proud, they increased turbulence downstream
sufficiently to cause erosion corrosion. It was recommended by Yorkshire
Imperial Metals that 90/10 copper-nickel should have a magnetic perme-
ability of 5 1.1 to give optimum corrosion resistance.
Kirk carried out an extensive review of the effect of alloying addi-
tions to copper-nickels on their corrosion resistance.16 His review includes
a large quantity of data from the LaQue laboratory that were previously
unpublished and includes other forms of attack in addition to erosion cor-
rosion. Only the highlights are discussed here; anyone requiring additional
data is referred to the original report.
The work of Efird, on the effect of nickel and iron and the way that they
are interrelated, was described previo~sly.~ Kirk pointed out that several
workers have shown that manganese acts in a similar way to iron, but at only
about 20% of the efficiency; that is, 1 % Mn is equivalent to 0.2% Fe.16 This
means that manganese supplements the benefit of iron additions, particu-
larly when the iron is on the low side. Tests have shown that 70/30 copper-
nickel alloys with high iron contents (>2%) have good resistance to erosion
corrosion but increased susceptibility to localized attack such as crevice cor-
rosion. Where components are to be used almost continuously with high wa-
ter velocities, this can be of benefit. Kirk cites the use of a cast 70/30 pump
impeller with -5% iron that gave excellent performance for many years."
Figure 7.9 shows the effects of various alloying additions on 90/10
copper-nickel at 2.4 and 4.6 m/s velocity. The tests lasted 6 years, which
gives an indication of long-term performance. Figure 7.10 shows similar
data for 70/30 copper-nickel alloys at the same velocities.
Figure 7.9 shows that increasing the iron content in 90/10 copper-
nickel to 2% had no significant effect at 2.4 m/s and was detrimental at
4.6 m/s. This is surprising, and one wonders if all the iron was in solution in
the 2% iron alloy. The addition of 2% aluminum was beneficial, particularly
at the lower velocity, but only in the aged condition. Additions of niobium
gave a small improvement at the lower velocity, but only in the solution-
annealed condition. The addition of 0.5% molybdenum was detrimental at
both velocities.
Tradename.
0.5
0.4
-
-E
E
Y
03
1
;
r
; 0.2
E
.-z
I
0.1
0 I I I I
C70600 2% Fe 2% Al (SA) 2% Al 0.5% Nb 0.5% Nb 0.5% Mo
(Aged) PA) (Aged)
1
0 1 Year 0 6 year I
I
1.2
1
1
-g
-
Y
08
0 0.6
rP
0"
s
!. 0.4
I
0.2
0
C70600 2% Fe 2% Al (SA) 2% Al 0.5% Nb 0.5% Nb 0.5% Mo
(Aged) PA) (Aged)
0 1 Year 0 6 year
(B)
FIGURE 7.9 Effect of alloying additions on the erosion corrosion of 90/10 copper-nickel:
(A) 2.4 m/s; (B) 4.6 m/s.16
0.4
0.3
-
-
Y
E
E
$ 0.2
5
n
:
2
z
.-
0.1
0
C71500 1.2% Fe C71640 5% Fe 2%Al 2%Al 0.5%Nb 0.5% Nb 0.5% Mo
PA) (Aged) (SA) (Aged)
1 0 1 Year O 6 v e a r I
1
1.2
I'
0.8
Y
P
b
5 0.6
n
n
:.5
m
0.4
0.2
0 I I I I I I I I
C71500 1 2% F e C71640 5% Fe 2% Al 2% Al 0 5% Nb 0 5% Nb 0 5% Ma
(SA) (Aged) FA) (Aged)
101Year meyear
(5)
FIGURE 7.10 Effect of alloying additions on the erosion corrosion of 70/30copper-nickel:

(A) 2.4 m/s; (B)4.6 m/s.16
TABLE 7.5 Strength of 70/30copper-nickel with niobium and silicon
0.2% Proof Tensile Elongation

Alloy stress (MPa) strength (MPa) (”/.)
C7 1500 (sol. ann.) 145 393 43
+
0.5% Si 1.5% Nb (aged) 398 706 24
Figure 7.10 shows the effect of increasing iron content (0.5% to 5%)
on the performance of 70/30 copper-nickel. At 2.4 d s , the best perfor-
mance was achieved with 1.2% iron, but at 4.6 d s , the 5% iron alloy
offered the best performance. As for 90/10 copper-nickel, the addition of
2% aluminum gave a small increase in performance, particularly at 4.6 m/s,
but only in the aged condition. The addition of 0.5% niobium produced no
significant change in corrosion resistance in either the solution-annealed or
aged condition. The addition of 0.5% molybdenum decreased the erosion
corrosion resistance at both velocities.
Additions to 70/30 copper-nickel to improve both its strength and hot
ductility were investigated at the BNF Metals Research Association during
the 1 9 6 0 ~ . ’ ~The
- ’ ~ most attractive additions were niobium and silicon.
Additions of silicon alone gave poor ductility and weldability; the further
addition of niobium gave high strength, toughness, and weldability. The
optimum composition was an alloy of Cu/30Ni/O.SSi/1.3Nb/l.2Fe/lMn.
Figure 7.1 1 shows the depth of attack in jet tests at 10.7 d s , with small
additions of hydrogen sulfide (H2S) to increase the aggressivity. The results
show that the alloy with niobium and silicon additions in the aged condition
gave similar performance compared with the standard alloy (C7 1500). In
clean seawater, there was no difference between the two alloys. In seawater
polluted with large quantities of HzS, the modified alloy was slightly infe-
rior. The mechanical properties of the wrought alloy in the aged condition
compared with those of C71500 are shown in Table 7.5. These indicate the
much greater strength of the modified alloy, similar to that of NAB.
Townsend also investigated the effect of additions of niobium and
silicon to 90/10 copper-ni~ke1.l~ The addition of 0.5% silicon and 0.3%
niobium gave a large increase in strength in the aged condition, while not
significantly affecting the resistance to erosion corrosion. He also investi-
gated the effect of adding 2% aluminum and found, like Kirk, that it gave
good results in the aged condition but increased the susceptibility to crevice
corrosion.16
Townsend also investigated the effects of zirconium additions to 70/30
copper-ni~ke1.I~ With 0.25% zirconium, the hot ductility improved, and
C71500 As Cast Sol. Ann. Aged Sol. Ann. Aged

Wrought
Cast
FIGURE 7.11 Depth of attack on 70/30 copper-nickel with niobium and silicon in jet test
at 10.7 m/s additions.’*
there was no change in the resistance to erosion corrosion in clean seawater.

However, in polluted seawater, the attack was much worse on the alloy with
zirconium additions compared with conventional 70/30 copper-nickel.
One element that has a big effect on erosion corrosion when it is
added to copper-nickel alloys is chromium. Kirk summarized the test re-
sults from a number of studies.16 The results from all the studies were
two alloys, one a 70/30 copper-nickel with 3% chromium (C71900) and
a Cu/l5Ni/0.8Fe/0.7Cr/0.6Mn alloy (C72200). Both these alloys can be
cold worked to give a higher strength than C71500, while the cast ver-
sion of the 70/30 alloy with chromium additions is also stronger in the
as-cast condition. In jet impingement tests, Anderson and Efird compared
some chromium-containing alloys with conventional copper-nickel alloys,
as shown in Figure 7.12.20The results show the significant improvement
in resistance to erosion corrosion with chromium additions, particularly to
the 70/30 copper-nickel alloys. IN768 is a cast copper-nickel alloy with the
nominal composition Cu/30Ni/l.6Cr/0.7Fe/0.6Mn.
A particular problem can occur in heat exchanger tubes when a foreign
body becomes lodged, causing local high turbulence. This can increase
local flow velocities by up to a factor of 5. Polan et al. reported the results
0.6
0.5
04
-E2
E
1
0.3
0
5
B 02
0.1
0
C70600 C71500 C72200 C71900 IN768
(A)
1.6
1.4
12
-E 1
-2
E
2-
1
08
06
0
0.4
02
0 I I I I
C70600 C71500 C72200 C71900 IN768
1-
(B)
FIGURE 7.12 Depth of attack of some copper-nickel alloys containing chromium at a jet
velocity of (A) 6.8 m/s and (B)15.3 m/s?O
08
07
-E 0.6
5 0.5
5.-0.4
Y
m
0
50 0.3
B
0.2
0.1
0
C70600 C71500 C72200 C71640
0 2.1 mls W 3.4 m/s 0 4.3 mls
FIGURE 7.13 Maximum depth of attack around a blockage of various copper-nickel alloys
after 12 months in seawater.*' (See color section.)
of tests with artificial blockages.21 Figure 7.13 shows the results at three
different nominal flow velocities. Although Alloy 722 with chromium was
better than 90/10 or 70/30 copper-nickel, the best performance came from
C7 1640 (66/30/2/2 Cu-Ni-Fe-Mn). Note, however, that there was some
crevice corrosion at the blockage with both Alloy 722 and C7 1640, while
there was none with 90/10 or 70/30 copper-nickel.
Kirk reported the results of tests of the 70/30 copper-nickel alloy with
3% chromium and additions of beryllium.'6 At a level of 0.15%, the rate
of attack increased, but at levels of 0.2% and 0.5%, it decreased. How-
ever, beryllium additions also made welds susceptible to stress corrosion
cracking.
When zinc is added to copper-nickel in substantial quantities, the
nickel silvers are produced. These alloys have very poor resistance to ero-
sion corrosion in seawater compared with 90/10 and 70/30 copper-nickel. l6
Additions of 0.9% aluminum to nickel silvers improved the resistance to
erosion corrosion, whereas manganese additions (3% to 7%) lowered the
resistance and also made the alloy susceptible to stress corrosion cracking.
In Japan, during the 1960s, there were problems in power stations,
with accelerated corrosion of copper alloy condenser tubes due to polluted
water containing sulfides. To combat this, AP bronze was developed, a
0.1
-
E 0.07.
E
0.06 .
0
3 0.05.
2
0.04.
5P
0.03
0.02 -
0.01 - 0. * - _ -
- , .
-2 0 2 4 6 8 10 12 14 16
Temperature (“C)
1- +- - 90/10 CU-NI +70/30 CU-NI I
FIGURE 7.14 Depth of attack in jet test at various seawater ternperat~res.~~
Cu/SSn/lAl/O. 1% alloy, which also has good resistance to corrosion in

clean seawater. Nosetani et al. described the results of model condenser
tests, and in clean seawater, at normal flow velocities (2.0-2.7m/s), the
corrosion rates of aluminum brass and AP bronze were similar.22
In an investigation of the microstructure of NAB, Michels and Kain
carried out jet impingement tests on cast NAB (C95800) with aluminum
contents from 8.1% to 9.4% and in a range of heat treatment conditions.*’
Despite the range of microstructures produced, the depth of attack at both
4.6 m/s and 9.1 m / s jet velocity was very similar in all conditions. This
was in contrast to the effect of microstructure on localized corrosion, as
discussed in Chapter 9.
Besides composition, there are a number of physical factors that can
affect the susceptibility of copper alloys to erosion corrosion. One of these
is temperature. Mills reported the effect of seawater temperature from - 1
to 15°C in the jet test at 9 m / ~ . The
* ~ results, shown in Figure 7.14, show
an increase of attack on 70/30 copper-nickel below 10°C, whereas 90/10
copper-nickel was unaffected. This has been attributed to the slower forma-
tion of iron-rich films at lower temperatures and was discussed in detail in
Chapter 3.
Campbell reported jet test results for 70/30 copper-nickel and NAB
(Table 7.6), which show deeper attack at 10°C compared with 15°C for
both alloys.25
TABLE 7.6 Erosion corrosion in jet impingement test for NAB and
70/30copper-nickel vs. seawater temperature’’
Attack at jet
Temperature (“C) Alloy Diameter (mm) Depth (mm)
Cast NAB 10 0.12
10 (C95800)
70/30 Cu-Ni 14 0.10
(C7 1500)
Cast NAB 0 0
15 (C95800)
70/30 Cu-Ni 2.1 0.05
(C71500)
The potential of 90/10 copper-nickel in natural seawater is about

-200 mV SCE (saturated calomel electrode). However, coupling to more
noble metals can increase the potential. Rowlands and Angel1 investigated
depth of attack as a function of the distance from the center of an impinging
jet.26They found that at 20 d s , little attack occurred at - 100 mV Ag/AgCl
(silver/silver chloride), but deep attack occurred as the potential increased
up to 0 mV (Figure 7.15).
The effect of coupling 90/10 copper-nickel to other copper al-
loys when it is undergoing erosion corrosion was investigated by LaQue
1.8
1.6
-
E
1.4
E. 1.2
Y
B 1
5
0 08
5
06
cl
0.4
0.2
DistancelNozzle diameter
FIGURE 7.15 Depth of attack of 90/10 copper-nickel as a function of potential and

distance from the center of an impinging jet after 1,000 hours.26
1.2
E
-
E
Y
0.8
9
’c 0.6
s
P
n 0.4
0
0 10 20 30 40 50 60 70 80 I
Area of Cathode (cm2)
+90/10 Cu-Ni +70/30 Cu-Ni --AT Alloy 400
FIGURE 7.16 Effect of copper alloy cathodes on the depth of erosion corrosion of 90/10
copper-ni~kel.~~
(Figure 7.16).27The results show that coupling a 90/10 copper-nickel anode

(high velocity) to large cathodic areas of 90/10 copper-nickel (low veloc-
ity) had no effect, and coupling to 70/30 copper-nickel only increased the
depth of attack slightly. However, coupling 90/10 copper-nickel to Alloy
400, which has a significantly more electropositive potential and is also an
efficient cathode, produced a significant increase in attack as the cathodic
area was increased.
Campbell and Carter investigated the erosion corrosion of cast man-
ganese bronze propellers on marine vessels.28 They found that a high ve-
locity jet of seawater alone did not induce much erosion corrosion. By the
use of dyes and high-speed photography, they demonstrated that air bub-
bles within a certain size range induced additional turbulence into the water
jet. The increased turbulence produced much greater attack on manganese
bronze, typical of that seen in service.
Besides external effects, sometimes corrosion products themselves
can influence erosion corrosion. When copper alloys are first immersed,
they often corrode quite rapidly until a protective film forms (see Chapter
3 ) . Cheung and Thomas reported erosion corrosion studies of 90/10 copper-
nickel in recirculating, synthetic ~eawater.’~ In the first tests, the corrosion
TABLE 7.7 Surface shear stress and critical velocity for erosion
corrosion of some copper alloys at 20°C”
Alloy Critical shear stress (MPa) Critical velocity (m/s)
Copper 9.6 1.3
Al-brass 19.2 2.4
90/10 Cu-Ni 43.1 4.2
70130 Cu-Ni 47.9 4.4
Alloy 722 296.9 10.9
rates were very low, much lower than any published data. When filters were
fitted to remove the high levels of copper corrosion products in the seawater,
the corrosion rates increased to more normal levels. Such filters were stan-
dard equipment on the BNF jet test when used with recirculated seawater.
The effect of initial surface condition was examined by Sat0 et al.,
who tested aluminum brass condenser tubes finished by bright annealing
(the standard finish), annealing in air, bright annealing after pickling, and
pickling after bright annealing.” Although electrochemical tests showed
differences in the films on these surfaces, after a few days in seawater at
either 2 d s or 7.7 ds, there was no discernable difference between the
corrosion rates in any of the tubes. Instead, corrosion rates were greatly
affected by both entrained solids in the water and sulfide pollution.
Much of the testing of copper alloys for resistance to erosion corrosion
has been conducted using the jet test, where a high-velocity water jet
impinges on the sample at 90 degree^.^ However, in many applications, the
flow is more or less parallel to the surface such as in a pipe. Efird conducted
parallel flow tests on several copper alloys to determine the critical shear
stress at which the protective film broke down and accelerated metal loss
occurred.31 Efird found that the shear stress was not only a function of
the bulk velocity, but also of the water temperature.3’ Table 7.7 shows
the critical velocities and shear stresses for several common copper alloys,
normalized to 20°C. The results show that the critical velocity and shear
stress increase with alloy content. Most notable is the substantial increase
in critical shear stress for Alloy 722. It should be noted that this only applies
to parallel flow and takes no account of local turbulence produced by bends
or surface irregularities, which can increase local velocities.
Efird’s tests were conducted in small-diameter pipes, and it is often
argued that flow is smoother and more uniform in larger-diameter pipes.”
Kirk carried out tests with NPS 4 pipe in 90/10 copper-nickel at velocities
from 2.4 to 7.3 m/s.32 The system included straight pipes and long radius
bends, all welded together. After 16 weeks, there was no significant erosion
TABLE 7.8 Erosion corrosion rates of various alloys in seawater

at very high jet velocities.
Average Velocity Mean corrosion
Alloy temperature (“C) (mls) rate (mmly)
~
Grey cast iron(A) 28 42 13.3

Aust. cast iron 19 39 2.19
90/10 Cu-Ni 9 39.6 0.56
16 40 0.66
27 39 0.84
70130 Cu-Ni 9 42 1.17
16 40.5 1.83
27 40.5 2.67
661301212 16 40.5 2.34
Cu-Ni-Fe-Mn
70/30 Cu-Ni 11 40.5 0.36
(5% Fe)
LG2 gunmetal 25 40.5 1.35
Admiralty gunmetal 20 38 0.76
(A)
Ten days only.
corrosion at any velocity, except downstream of welds. The maximum depth

of attack in these regions was 0.23-0.25 mm at all velocities. Clearly the
extra turbulence caused by the weld bead was sufficient to rupture the pro-
tective film. The results showed that the critical shear stress was not passed
in a larger-diameter pipe, except where local turbulence was increased.
Some studies have looked at very high water velocities because even
when the film has been damaged, metal loss rates may not be very high.
The LaQue Laboratory carried out very high-velocity jet tests for 30 days
on a range of alloys. The jet velocity was mostly around 40 m/s, although
a few tests were conducted at lower velocities. Table 7.8 shows a summary
of the data for copper alloys compared with austenitic cast iron and gray
cast iron. The results for the copper-nickel alloys show increasing metal
loss with temperature. The corrosion rates for 90/10 copper-nickel were
significantly lower than those for 70/30 copper-nickel, suggesting that at
very high velocities, the iron content is more important than the nickel
content. Certainly the corrosion rate for 70/30 copper-nickel with 5% iron
was very low. The rate for C7 1640(Cu/30Ni/2Fe/2Mn) was higher than that
for the other copper alloys. This is unusual as this alloy has exceptionally
good resistance to erosion corrosion (see the previous discussion). It is
possible that the iron was not all in solution in the samples used for these
tests. The results for the gunmetals show decreasing metal loss as the tin
Next Page
16
14
-2
E 12
E
c
0
2 '
C
'5 08
e
5
06
04
02
0
0 5 10 15 20 25 30 35 0
Velocity (mls)
I-90110 Cu-Ni -70/30Cu-Ni -Copper I

FIGURE 7.17 Corrosion rate of some copper alloys in high-velocity tests in ~eawater.~'
content of the alloy was increased, as seen at lower velocities (see the
preceding discussion).
A study was carried out comparing high-velocity erosion corrosion
test results using a multivelocity parallel flow rig, a rotating disc test, and
the BNF jet test.33 In the multivelocity rig, velocities of 2.9-36.6 m/s were
investigated. Some of the results are shown in Figure 7.17. The surprising
result is the low weight loss of copper compared with the copper-nickel
alloys. The reason for this is the degree of protection conferred by the film.
The film on pure copper was poorly protective, and the attack was spread
more or less uniformly over the sample, whereas the film on 70/30 copper-
nickel was very protective and only broke down in a few places, where
attack was deep. The results for 90/10 copper-nickel were intermediate
between these two extremes. The importance of the protectiveness of the
film, as well as the velocity, was brought out by Cotton in his discussion of
the preceding work.34 The author also described the change from narrow,
deep attack to shallow, broad attack, as the film becomes less p r ~ t e c t i v e . ~ ~
Previous Page
In some unpublished work, the US.Navy looked at high flows in

an NPS 2 piping system with 90/10 copper-nickel piping and bends and
bronze valves and flanges. The flow velocities were 2.4, 4.6, and 9.1 m/s.
Where orifice plates produced cavitation downstream, there was severe
attack, particularly at the highest velocity. Where noncavitating orifices
were fitted, more normal erosion corrosion was seen downstream. At 9.1
m / s , there was heavy roughening in turbulent areas and light to moderate
roughening elsewhere. Leakage occurred in the most turbulent area within
6 months. At 4.6 m / s , there were no failures over the 8 months of testing, but
there was heavy roughening in the most turbulent areas. Attack was worst
on or after bends. There was also attack of the bronze valves, but this did
not lead to failure because of the higher wall thickness. However, the depth
of attack was greater on the bronze than on the 90/10 copper-nickel. The
attack on the piping at 2.4 m / s was more normal, with no significant attack
even in the turbulent zones. The conclusion was that high water velocities
cannot be sustained for extended periods in copper alloy piping systems
without leaks due to erosion corrosion.
However, there are occasions when copper alloys may experience
intermittent high velocities. This was investigated by Melton using a small
anode experiencing a high-velocity water jet coupled to a larger cathode
in gently moving seawater, both of 90/10 ~ o p p e r - n i c k e l The
. ~ ~ currents
between the anode and cathode, measured on freshly abraded samples over
a range of velocities from 4 to 10 m / s , stabilized at levels equivalent to
a metal loss of 1.0-1.2 mrdy. However, for samples that were prefilmed
in slow-moving water, the current between the anode and cathode was
effectively zero, even with a jet velocity of 19 m / s . These tests only lasted
a few days, but the results show that once a protective film forms on 90/10
copper-nickel, it can survive high seawater velocities, at least for a few days.
This is borne out by the experience with 90/10 copper-nickel in off-
shore firewater systems. These generally operate at velocities up to 10 m/s,
but they are only tested for about 1 hour per week and remain stagnant for
the rest of the time. There have been no failures due to erosion corrosion
reported, either from cold waters, such as in the North Sea, or from tropical
waters, where pipe wall temperatures may reach 60 or 70°C under stagnant
conditions when exposed to direct sunlight.
7.1.3 Iron/ Ferrous Sulfate

Throughout the first half of the 20th century, admiralty brass (C44300)
and aluminum brass (C68700) were the most commonly used alloys for
0.6
(c
5 0.3
Q
d 0.2
0.1
0
Al-Brass 90/10 Cu-NI 70/30 Cu-Ni 66/30/2/2Cu-Ni
I O N o F e W+Fe I
FIGURE 7.18 Effect of iron dosing (1 mg/L for 1 h/d) on the depth of attack at 9 m/s.35
tubes in seawater-cooled heat exchangers. In the 1950s, there was a move

to replace bare carbon steel and cast iron components with more corrosion-
resistant alloys or to apply coatings to increase reliability. Following this,
there was an increase in the number of heat exchanger tube failures due
to erosion corrosion. In 1961, Bostwick described how daily additions of
ferrous sulfate to the cooling water replaced the iron-rich films that had
previously been formed by corroding steel, with a large reduction in the
number of aluminum brass tube failures.” Shortly after, Lockhart described
a similar success story at a U.K. power station.’8 Since that time, additions
of ferrous sulfate solution (typically I mg/L Fez+ for 1 hour/day) have been
made at many plants to solve heat exchanger corrosion problems.
The use of iron additions to combat corrosion of copper alloy heat
exchanger tubes was surveyed among U.K. power stations as well as the
published literature to determine when ferrous sulfate dosing would be ef-
fective and for which alloys.39Ferrous sulfate was clearly efficacious on alu-
minum brass at normal flow velocities (2 m/s), and a number of laboratories
conducted tests at higher velocities and included other alloys. Figure 7.18
shows the effect of ferrous sulfate dosing ( 1 mg/L Fe2+ for 1 h o d d a y )
on a number of copper alloys tested in the Campbell condenser tube ap-
paratus at a jet velocity of 9 m / ~ . ’The
~ results show that even at this high
-
E
E
Y
8
-4
0.1
5Q
:
0.01
Velocity (mls)
I--C No Fe *O.Olmg/l Fe t.0.03mgll Fe *0.05mgll Fe I

FIGURE 7.19 Effect of continuous iron dosing rate and velocity on the corrosion of alu-
minum brass.40
velocity, ferrous sulfate dosing significantly reduced the depth of attack on

the copper-nickel alloys, but not on aluminum brass. The reason for the
anomalous result with aluminum brass is the nature of the protective film.
At 9 m/s, ferrous sulfate dosing produces a more protective film, but it
cannot prevent attack under the jet. Hence, the attack is narrow and deep.
Without ferrous sulfate dosing, the film is not so protective, and the attack
is more widespread and shallow.34
Initially, iron was added as ferrous sulfate solution on an intermittent
basis. However, iron can also be added continuously by the use of electri-
cally stimulated soft iron or mild steel anodes. Iron added continuously at
0.04 mg/L Fe2+ is roughly equivalent to 1 mg/L Fe2+ for 1 hour/day. Sato
et al. investigated the effect of a range of continuous iron dosing levels on
the corrosion of aluminum brass tubes at various velocitie~.~'The results
(Figure 7.19) show that iron dosing levels as low as 0.03 mg/L Fe2+reduced
the depth of attack at velocities up to 3.7 d s . Sato also showed that ferrous
sulfate dosing reduced the attack around lodged obstructions in aluminum
brass heat exchanger tubes, but it did not totally stop it.4'
Nosetani et al. showed that the duration of the test also affected
the perceived benefit of ferrous sulfate additions.42 Figure 7.20 shows the
0.09-
*. *
0.08-
0.07-
. *I.
006-
0.05-
’.
004-
*.
0.03-
‘*
0.02- - - - - - - .- - - - * - ’ - - - _ . -_ - - .
0.01 - I
- - - - _- A
0,
- -+ - 1,000hr (int) -m- 1,000hr (cont) - -k - 6,000hr (int) 6,000hr (cont)
FIGURE 7.20 Effect of time and manner of iron dosing on corrosion of aluminum brass
at 2 m/s?’
corrosion rate of aluminum brass tubes at a flow rate of 2 m/s with both
continuous and intermittent iron dosing. The continuous iron dosing levels
have been converted to the equivalent amount of iron dosed for 1 hodday.
The results show that the corrosion rate was much lower at all dosing levels
after 6,000 hours, compared with 1,000 hours’ exposure. This is because
the protective films become more efficient as they grow and growth rates
are slow. The results also show that continuous low-level dosing with iron
is more protective than the equivalent dosed intermittently. The same effect
as 1 m g L Fe2+ for 1 hour/day can be achieved with approximately half
that equivalent level dosed continuously.
Although intended originally for brass tubes, ferrous sulfate has oc-
casionally been used successfully to solve corrosion problems with 904 0
and 70/30 copper-nickel. The original corrosion problem that iron additions
were used to overcome was erosion corrosion (impingement attack), but
they have also been used to try to solve other corrosion problems with vary-
ing degrees of success. In Japan, ferrous sulfate dosing solved problems of
localized attack of aluminum brass tubes that had a cathodic film of man-
ganese dioxide (MnO;?)on them.43This film was produced by the oxidation
of manganese in solution in the seawater by chlorine, added to control

fouling.
Ferrous sulfate dosing has also solved corrosion problems caused by
ammonia (NH3) pollution. The author used ferrous sulfate dosing for 2
months at the start-up of a multistage flash (MSF) desalination plant to
form more protective films on 90/10 copper-nickel tubes. The water was in-
termittently polluted with ammonia, which had caused low-temperature hot
spot attack in the heat exchangers when the first unit was started. These were
retubed, and ferrous sulfate dosing was started from new at a level of 1 ppm
Fe2+ for 1 hour every day. The plant has now been in operation since 1994
with no further failures. At another MSF plant in the Middle East, the ammo-
nia pollution was almost continuous because the water intake was near the
outfall of a fertilizer plant. Daily ferrous sulfate dosing throughout its life
was recommended for this plant as it was tubed with aluminum brass. There
had been failures of 90/10 copper-nickel tubes in an earlier plant on the same
site. The plant has now been in operation since 1998 with no problems.
Ferrous sulfate does not work on aluminum brass tubes in water
polluted with H2S, unless protective films can first be formed in clean
water.44 Hack, Gudas, and their colleagues investigated the effect of iron
additions on the corrosion of 9040 and 70/30 copper-nickel in polluted
~ e a w a t e r . They
~~-~ found
~ that intermittent ferrous sulfate dosing had little
effect, but continuous iron additions were more efficient, particularly with
70/30 copper-nickel. The iron additions worked best when the sulfide pol-
lution was at a low continuous level. A U.K. power station, operating in
a tidal estuary, used ferrous sulfate dosing to prevent corrosion of 70/30
copper-nickel tubes, where H2S was an intermittent pollutant at certain
states of the tide.39
It is well documented that chlorine/hypochlorite dosing to control
fouling interferes with the efficacy of ferrous sulfate This is
because of the way in which ferrous sulfate acts. When added to water, it
forms a colloid of FeOOH, whose charge increases the smaller the particle
size. The colloid is attracted to the tube wall by the local electric field,
so the greater the charge on the colloid, the more will be attracted to the
surface. Chlorine causes very rapid growth of the colloid so that the charge
is much weaker and is less likely to be attracted to the tube wall. Figure 7.2 I
shows the increasing pit depth with chlorine concentration for 90/10 copper-
nickel under turbulent, high-velocity water conditions.?' It also shows that
with no chlorine, ferrous sulfate additions suppressed attack, but this effect
was cancelled when there was more than 0.5 mg/L chlorine in the water.
The recommendation is that if the chlorination is continuous, it should be
I /-
07 -
06-
E
E
05-
2B 04-
0
5$ 03-
P
:+
0
0 1 2 3 4 5
Chlorine (rnglL)
I-tNoFe ++Fe 1
FIGURE 7.21 Effect of ferrous sulfate and chlorine on the depth of attack on 90/10 Cu-Ni
at 9 r n / ~ . ~ ~
turned off while ferrous sulfate dosing is taking place. If the chlorination
is intermittent, then it should be staggered so that it does not occur during
ferrous sulfate dosing.
7.1.4 Effect of Chlorine

Most cooling waters contain living organisms that can colonize metals. In a
cooling system, this can lead to blockage if macrofouling (weed and shell-
fish) is not controlled. Copper and its alloys are very resistant to macrofoul-
ing (see Chapter 13), but microfouling by slimes and bacteria can impede
the efficiency of heat exchangers. One of the most common methods of
fouling control is the use of chlorine additions. These may be added as
hypochlorite solution, or chlorine may be generated electrolytically from
brine or seawater. Chlorine is a powerful oxidizing agent and could have a
detrimental effect on copper alloys.
Sat0 carried out tests on a range of copper alloy heat exchanger tube
materials in a Japanese power station.48The water flow rate was nominally
2 d s , and a range of chlorine levels were investigated over several months.
The results, given in Figure 7.22, show that the corrosion rate increased as
200,
0 2 4 6 8 10 12
Chlorine (mglL)
--t Al Brass +90/10 Cu-Ni -A? 70/30Cu-Ni
FIGURE 7.22 Effect of chlorine on copper alloy condenser alloys at 2 r n / ~ . ~ ~
the chlorine concentration increased. However, what is of more concern is

localized corrosion caused by chlorine.
In the 1970s, failures began occurring in the aluminum brass heat
exchanger tubes of several oil supertankers. The attack took the form of
erosion corrosion at scratch lines (Figure 7.23). All the tankers with failures
had continuous chlorine dosing equipment as well as ferrous sulfate dosing.
Tankers with only ferrous sulfate dosing were not experiencing problems.
BNF undertook a program of research to investigate the p r ~ b l e m . ’The ~
initial results showed that chlorine could prevent the formation of the normal
protective film, particularly after local damage, leading to accelerated local
attack (see Chapter 3). Figure 7.21 shows the results of tests on 90/10
copper-nickel, whereas Figure 7.24 shows the results for aluminum brass
and 66/30/2/2 Cu-Ni-Fe-Mn. High levels of chlorine accelerated attack,
particularly for aluminum brass, whereas the 66/30/2/2 Cu-Ni-Fe-Mn alloy
was hardly affected. Figure 7.24A shows that the attack on tubes also dosed
with ferrous sulfate was deeper than on those with chlorine alone. This
effect is the same as that discussed for tests on copper alloys at very high
velocities, where attack with a more protective film is localized, but very
FIGURE 7.23 Erosion corrosion at a scratch line on an aluminum brass tube, where
repassivationdid not occur due to overchlorination. (See color section.)
deep, while the attack with a poorly protective film is more widespread and
shallow. Further testing was conducted to identify safe chlorine levels for
the common copper alloys.35 These were as follows:
All alloys When ferrous sulfate dosing is being used, switch off con-
tinuous chlorination during the dosing period
Aluminum brass Intermittent chlorination preferred (e.g., 1 m g L for 1 hour
every 12 hours), but up to 0.5 mg/L continuous with ferrous
sulfate dosing
90/10 Cu-Ni Up to 0.5 mg/L continuous
70130 Cu-Ni Up to 1 .O mg/L continuous
66/30/2/2Cu-Ni-Fe-Mn Up to 2.0 mglL continuous
Nore: These chlorine concentrations are measured in the feed piping or inlet water box.
When the water is polluted with sulfide, the additional presence of chlorine
can produce a large acceleration of attack at sites experiencing high water
velocities. This is discussed in detail in the next section.
Chlorine (mg/L)
0.5
0.4
-E 0.3
-
X
E
3
;
5 0.2
P
d
0.1
Chlorine (mg/L)
t N o F e -+Fe
(B)
FIGURE 7.24 Depth of attack in Campbell condenser tube test rig vs. chlorine level at 9
m/s: (A) aluminum brass; (B)66/30/2/2
c~-Ni-Fe-Mn.~~
The effect of chlorine and flow rate on the corrosion of NAB has
also been i n ~ e s t i g a t e d .Cast
~ ~ NAB (C95800) was exposed in quiescent
seawater (-0.1 m/s) and also in parallel flow rigs at 2.5 m/s in both natural
seawater and seawater with a chlorine residual of 0.5 mg/L. All the tests
were at 10°C. In the first 4 weeks of exposure, the corrosion rates in
-
natural seawater at 0.1 and 2.5 m / s were -80 and 155 pm/y, respectively.
In chlorinated seawater, the corrosion rates were -30 and -75 pm/y.
Although the corrosion rate increased with velocity, the rates in chlorinated
seawater were significantly lower. However, after 12 weeks’ exposure, there
were no significant differences between the corrosion rate in natural and
-
chlorinated seawater at 0.1 m/s (-25 pm/y). At 2.5 m/s, the corrosion rates
in natural and chlorinated seawater were 10 and -40 pm/y, respectively.
The corrosion rates had decreased with exposure time, but more so for the
samples in natural seawater. It should be noted that even in chlorinated
seawater, the corrosion rate is not excessive.
When the chlorine levels are higher (e.g., -5 mg/L) and velocities are
also high, then accelerated erosion corrosion of NAB can occur. The author
has seen failure by erosion corrosion of the first stage of an NAB multistage
seawater lift pump on an offshore platform. The pump was situated inside
a steel caisson, and strong hypochlorite solution was injected close to the
pump intake. Failure occurred in less than 12 months.
7.1.5 Sulfides
When hydrogen sulfide is present in seawater, it is almost always pro-
duced by the action of sulfate-reducing bacteria (SRB). These bacteria
need anaerobic conditions to thrive and are normally inactive in aerated
seawater. When the water is stagnant or the access of oxygen is restricted,
for example, in bottom mud, the aerobic bacteria gradually consume the
DO. Once the oxygen is gone, the anaerobic bacteria, including SRB, be-
come active. They reduce the sulfate in seawater to H2S as part of their
metabolic process. In deaerated seawater containing H2S, the corrosion
rate of most metals is low.
During the 1940s and 195Os,there was a marked increase in the num-
ber of failures of copper alloy heat exchanger tubes in both power stations
and ships operating with polluted cooling water. BNF began an extended
test program utilizing the jet test.4 H2S was added either continuously or
intermittently to the tests using recirculated seawater. Bem and Gilbert
reported the results of tests on a range of alloys under a variety of operating
condition^.^' Both low-level, continuous dosing of sulfide and intermittent,
high-level dosing with H2S produced deep attack on all the common copper
alloys, similar to that seen in service. The results for 90/10 copper-nickel
showed that an alloy with 1% iron performed better than one with 2% iron.
Similarly, 70/30 copper-nickel with 0.8% iron suffered less attack than
66/30/2/2 Cu-Ni-Fe-Mn. Aluminum brass suffered severe attack under all
conditions. Bem and Gilbert also conducted tests in which the sulfide was
all oxidized to elemental sulfur by the DO in the seawater, that is,
2H2S + 0 2 + 2H20 + 2s.
All the alloys were still attacked more than in clean seawater, but there was
less attack than in water containing H2S, particularly for the copper-nickel
alloys. As before, those with the lower iron content suffered less attack.
In later tests, Bern showed that if the H2S was present in deaerated
seawater, there was no erosion corrosion on any copper alloy, and the
general corrosion rates were This demonstrated the importance of
the presence of oxygen with the H2S to cause accelerated attack.
The work at BNF was not published in the open literature, and further
work was undertaken in the 1970s in the United States, following the
premature failure of 90/10 copper-nickel piping on a U.S. Navy vessel.52
Gudas et al. showed that accelerated corrosion occurred on copper-nickel
alloys that were first exposed to deaerated polluted water and then clean
seawater, but not to the same extent as seen in the service failure.52 When
they investigated the effect of low levels of sulfide in aerated seawater, the
severe attack seen in service was r e p r o d ~ c e dThe
. ~ ~ tests also showed that
velocity was a critical factor and the higher the velocity, the deeper the
attack. On aluminum brass, this sometimes takes the form of accelerated
pitting corrosion, rather than the smooth water-swept pits more normally
associated with erosion corrosion. The mechanism of sulfide attack was
discussed in Chapter 3.
Lee et al. carried out further studies on several heat exchanger alloys
as a function of sulfide concentration and velocity.54 The results, given
in Figure 7.25, show the isocorrosion curves (0.1 mm/y) as a function of
velocity and sulfide. This is quite a high corrosion rate for a condenser tube
alloy in that an 18 SWG tube would have lost half its thickness in 6 years-a
relatively short life. The results show that 90/10 copper-nickel (in this case,
with 1.1% iron) performed the best. At sulfide levels below 0.01 mgL,
70/30 copper-nickel was next best, with aluminum brass being the worst.
However, at higher sulfide concentrations, aluminum brass was slightly
better than 70/30 copper-nickel. At a typical heat exchanger velocity of
2 m/s, 90/10 copper-nickel was satisfactory up to 0.01 mg/L sulfide. These
-Al-Brass - - 90/10 Cu-Ni - - - - 70/30Cu-Ni

~
FIGURE 7.25 lsocorrosion curves (0.1 mm/y) as a function of velocity and sulfide
sulfide levels appear low, but they are continuous in aerated water. Under
intermittent pollution, the alloys can tolerate more sulfide, especially when
a protective film is first formed in clean seawater.
A number of methods have been attempted to ameliorate the effects of
sulfide attack. Rowlands recommended a pretreatment of copper-nickel al-
loys with a hot solution of sodium dimethyldithiocarbamate.ss Christensen
conducted trials on aluminum brass as well as both 90/10 copper-nickel and
70/30 copper-nickel tubes in both the treated and untreated condition^.^^
The tubes were first exposed for 5 weeks in a heavily polluted Danish
harbor, followed by exposure in a heat exchanger at 1.8 m/s with clean sea-
water for several weeks. The overall conclusion was that the pretreatment,
far from being beneficial, resulted in increased corrosion of all three alloys
compared with untreated tubes.
The work of Gudas, Hack, and their colleague^,^^-^^ who investigated
the effects of ferrous sulfate dosing in polluted water, was described in
Section 7.1.3. They found that continuous iron dosing was better than
intermittent dosing, and it worked better on 70/30 copper-nickel than on
90/10 copper-nickel. In no case did iron/fen-ous sulfate dosing completely
suppress sulfide attack; rather, it only reduced it.
TABLE 7.9 Depth of attack in jet impingement test at 20°C for some cast
copper alloys7
Allov UNS no. Clean Seawater Seawater with sulfide
Mn-bronze C86400 0.5 0.42
Gunmetal (LG2) (5% Sn) C83600 0.23 0.07
Gunmetal (LG4) (7% Sn) C924 10 0.05 0.04
Gunmetal (GI) (10% Sn) C90500 0.02 0.02
Leaded tin bronze (3% Pb) C92900 0.02 0.03
NAB C95800‘*’ 0.0 0.14
IAl
Aluminum slightly below specification minimum.
INCRA sponsored work to determine whether ferrous sulfate dosing

combined with sponge ball cleaning could mitigate the effects of heavy
sulfide pollution.57 The experimenters found that sponge ball cleaning plus
0.18 mg/L Fe2+ continuously (equivalent to 4.3 mg/L Fez+ for 1 h o d d a y )
minimized the corrosion of 90/10 copper-nickel in the presence of 0.08
mg/L sulfide, but it did not prevent all accelerated attack. Joseph and Ham-
mond also found that excessive sponge ball cleaning accelerated attack
and recommended that the rate of sponge ball cleaning and the iron dose be
optimized by measuring the heat transfer efficiency of the heat exchanger.57
Many attempts have been made to reduce the attack in polluted waters
by alloying. Bog carried out jet impingement tests on a range of cast copper
alloys, the results of which are shown in Table 7.9.7 The results show the
poor performance of NAB in polluted water as well as the poor performance
of manganese bronze. It is clear that the resistance to sulfides increases as
the tin content of the bronzes increases, with the 10% tin alloys performing
best. The benefits of tin additions to copper alloys were explored at BNF
in the 1950s. Bem reported on trials with the tin brasses containing about
17% zinc and 4% to 6% tin.58The best performance in polluted water was
obtained with a brass containing 6% tin and from a straight tin bronze
containing 7.5% tin. These alloys were never developed commercially in
the United Kingdom, but Sumitomo Light Metals experienced similar cor-
rosion problems with copper alloy heat exchanger tubes in polluted waters
around Japan and introduced AP bronze in the 1970s. This alloy has a
nominal composition of Cu/8Sn/lA1/0. I Si, and test data in polluted water
were reported by Nosetani et al. (Table 7.10).*’ The results show similar
performance of aluminum brass and AP bronze in clean seawater and a big
increase in corrosion rate for aluminum brass in polluted water, while the
corrosion rate of AP bronze is virtually unchanged. The laboratory data
TABLE 7.10 Corrosion rates of aluminum brass and

AP bronze heat exchanger tubes at 2 m/sZ2
Alloy Corrosion rate (rnm/y)
~ ~~
Clean seawater 0.05 mglL sulfide

Aluminum brass 0.08 0.23
AP bronze 0.07 0.08
and over 10 years of successful operating experience with AP bronze in

Japanese power stations were reviewed by Atsumi et al.59
The effects of sulfide and chlorine together were investigated by
the author.60 The results for aluminum brass, 90/10 copper-nickel, and
66/30/2/2 Cu-Ni-Fe-Mn are shown in Figures 7.26,7.27, and 7.28, respec-
tively. It is clear that low levels of chlorine reduce the effects of sulfide on the
corrosion of aluminum brass and 90/10 copper-nickel, but higher chlorine
levels resulted in accelerated attack compared with sulfide alone, leading
to tube perforation at the higher sulfide levels. The 66/30/2/2 Cu-Ni-Fe-Mn
alloy was severely attacked by sulfide, and any dosing of chlorine further
increased the corrosion. Alloy 722 performed similarly to the 66/30/2/2
Wall Thickness
--___----
1.2 -* -ri-
- _ - -
1 -_ . -_- . - - - 1
-
A
E
E A-.
Y
o 0.8 -
0 0.02 0.04 0.06 0.08 0.1 0.12

Sulfide (mglL)
I--t NoCI -m- 0.25mglLCI - -A - 0.5malLCI I

FIGURE 7.26 Effect of chlorine and sulfide on the erosion corrosion of aluminum brass at
7 m/s60
1.4 I
Wall Thickness
1.2 -1
* -----_
- _- - - - - _ _
1 - ---- --
-
E
E
Y
, ---A
0 0.8 -
B
2
' 0.6
5P
-
*
d ,
5 0.4 - A
4
I+ No CI -m- 0.25 rnglL CI - -A - 0.5 mglL CI I

FIGURE 7.27 Effect of chlorine and sulfide on the erosion corrosionof 9/10 copper-nickel
at 7 rn/s60
1.4
Wall Thickness
1.2
- 1 .-
.-
-0
E
E
Y
0.8
* + -
4
r
' 0.6
r,
P
d
5 0.4
ii
0.2
I
0 I
0 0.02 0.04 0.06 0.08 0.1 0.12
Sulfide (mglL)
&NoCI -W- 0.25rnalLCI - -k - 0.5rnalLCI I

FIGURE 7.28 Effect of chlorine and sulfide on the erosion corrosion of 66l3Ql2l2
Cu-Ni-Fe-Mn at 7 m/s.60
Cu-Ni-Fe-Mn alloy, although the attack was not quite so severe; 70/30
copper-nickel was similar in performance to 90/10 copper-nickel, except
that the corrosive effects of sulfide were reduced even with a 0.5 m g L
chlorine addition.
In the 1950s, a power station in Havana, Cuba, was experiencing
severe corrosion of both aluminum brass and copper-nickel heat exchanger
tubes due to polluted water. Fouling was also a problem, and chlorine dosing
was carried out. When the chlorine dosing was stopped, the rate of tube
failure decreased. These experiences mirror the test results with 0.5 m g L
chlorine.
The results of Francis demonstrate not only the severe effect of sul-
fide, but also the possibility of reducing its effects with low-level chlorine
dosing.60 This is discussed in more detail in Section 7.3.
7.2 Erosion
Erosion is the mechanical removal of the protective film and, usually, the
underlying metal as well, most commonly by solids suspended in the
liquid. In appearance, it is somewhat similar to erosion corrosion, often
with a scored effect. Figure 7.29 shows a severely eroded gunmetal im-
peller operated in water containing high levels of suspended solids. With
erosion corrosion, it can be suppressed by applying cathodic protection,
whereas the mechanical effect of erosion cannot. However, it has been
shown that where an alloy is corroding as well as eroding in a fluid, there
is also a synergistic effect, which increases the metal loss.6' The more the
alloy is corroding, the greater the synergistic effect.
In erosion, the angle of incidence of the solids is important; the greater
the angle of incidence, the more severe the attack. Most test rigs use a 90-
degree angle of incidence, whereas there are many service applications in
which the angle can be much lower, for example, 30 to 45 degree.
The suspended solids are usually present as silt or sand. These are
usually introduced into cooling water when the area around the suction
intake is relatively shallow and the bottom is stirred up by storms or tidal
action. It can also occur when settling ponds are not dredged on a regular
basis. Silt is fine material up to -50 vm diameter, whereas sand usually
varies from 50 to 250 Fm diameter, or more.
Silt can be present in large quantities (hundreds of milligrams per
liter), but it is less erosive than sand because of its fine nature. Sand is more
erosive the larger its particles, and a small amount of coarse sand may do
FIGURE 7.29 Severe erosion of a gunmetal impeller due to a high level of suspended
solids. (See color section.)
0.25 1
0.2
-2 0.15
--
E
a2
2
.-gIn 0.1
e
b
0
0.05 - .'
,
i0
-50 prnsand - - - 250 pm sand
FIGURE 7.30 Erosion of aluminum brass after pretreating in seawater with ferrous
more damage than a larger quantity of silt. Sand and silt incursions are
often intermittent, and so tubes that have had time to form protective films
could be more resistant to erosion by sand. Sat0 et al.showed that aluminum
brass tubes that were prefilmed by ferrous sulfate dosing could resist at-
tack by 50-pm-diameter sand, but not 250-pm-diameter sand, as shown in
Figure 7.30.62 This is clearly of value where the solids loading is intermit-
tent and there is time to form a protective film.
Corrosion by entrained sand gives rise to four distinct phases of attack,
as shown in Figure 7.31 for aluminum brass.41The flow rate was 2 m/s with
50- to 100-pm-diameter sand particles:
Stage I: Low sand content has no significant effect on corrosion
rate.
Stage 11: The rate of attack increases as the sand content increases;
here the protective film is being increasingly removed as
the sand content rises.
Stage 111: The rate of attack is more or less constant with sand
content; here the rate of film formation is much lower
than the rate of film removal.
Stage IV The rate of attack rises sharply with an increase in sand
content; electrochemical effects are minor, and the main
damage is by mechanical removal of metal (pure erosion).
This phase is rarely seen in practice.
Copper-nickel alloys formulated to increase the resistance to sand
erosion, such as 66/30/2/2 Cu-Ni-Fe-Mn, were superior to aluminum brass,
whereas 90/10 copper-nickel differed little from aluminum brass.4'
Parker and Roscow reported the effect of varying silt quantities on the
erosion rate of condenser alloys exposed at power stations along the Severn
estuary in the United Kingdom.63The strong tidal action in the estuary stirs
up silt from the bottom, but the quantity at the water intake varies from
50 mg/L near the ocean to -5,000 mg/L high up the estuary. The mean silt
content vs. erosion rate was plotted for a number of common condenser
alloys, all exposed at -2 m / s velocity. The results, in Figure 7.32, show
that aluminum brass and 70/30 copper-nickel suffered severe attack as the
silt level increased. The 70/30 copper-nickel alloy with 2% iron and 2%
manganese (C7 1640), for increased erosion resistance, was better. A special
70/30 copper-nickel alloy with 3% iron and 3% manganese exhibited further
improvement, but this alloy was more susceptible to localized attack during
shutdowns, when stagnant conditions prevailed. Only titanium gave good
results at the highest silt levels.
000
FIGURE 7.31 Effect of sand content on erosion of aluminum brass at 2 m/s by sand up
to 0.1 mm diameter.62
0.8
0.7
0.6
0.5
0.4
0.3
0.2
01
0
10 100 1,000 10,000
Silt Content (rnglL)
FIGURE 7.32 Metal loss of some heat exchanger alloys in silt-laden seawater.63
04 I
0 1 2 3 4 5 6
Iron Content (wtX)
FIGURE 7.33 Effect of iron in a Cu/l5Ni alloy on erosion by sand.62
The effect of alloying on erosion resistance has been studied by several

workers. Sat0 et al. showed the effect on erosion of iron additions to a Cu-
15% Ni alloy.41High iron contents significantly reduced the metal loss, as
shown in Figure 7.33. This graph also shows data for 90/10 copper-nickel
and Alloy 722. These points show that the addition of 5 % nickel to 90/10
copper-nickel had only a small effect on erosion resistance. However, the
0.5% chromium addition present in Alloy 722 significantly improved the
erosion resistance, compared with a chromium-free alloy with the same
iron content.
Gaffoglio reported Sumitomo data for several copper heat exchanger
alloys as afunction of the silt or sand particle diameter.64The results showed
a steady increase in erosion rate, with a particle diameter up to 100 pm for
all the alloys, except aluminum brass. This followed the same trend as the
copper-nickel alloys up to 40 pm diameter, followed by a sudden increase in
erosion rate with larger-diameter particles (Figure 7.34). This suggests that
a transition from stage I to stage 11, as shown in Figure 7.3 1, had occurred.
The higher-alloyed copper-nickels showed the best performance, but more
rapid erosion rates would be expected with these alloys with sand particles
of 250 pm and larger.
1 10 100 1000
Mean Sand Diameter (pm)
-Al-Brass - 90/10 Cu-Ni -Alloy 722 66/30/2/2 Cu-Ni-Fe-Mn
FIGURE 7.34 Erosion rate of various alloys as a function of sand
A recent has investigated the interaction of velocity, tempera-

ture, and sand content in synthetic seawater, using a recirculating erosion rig
of the type described by Dallas et a1.66The results for cast NAB (C95800),
shown in Figure 7.35, show that an increase in sand content from 50 to
500 mg/L had less effect than an increase in seawater temperature from 18
to 50°C. The effect of temperature on erosion corrosion in seawater has not
previously been reported.
Another recent study, in the United States, investigated the effect of
sand and sulfide on the corrosion of copper alloy heat exchanger tubes.67The
flow rate was 2 d s , sand was added at 200 m g L (particle diameter 50-250
pm), and sulfide was added at 0.2 m g L . The results, given in Figure 7.36,
show that sand increased the corrosion rate by a factor of 2 to 3, but sulfide
increased the corrosion rate by much more: 45 times for aluminum brass,
19 times for 70/30 copper-nickel, and nine times for 90/10 copper-nickel.
When sand and sulfide were present together, the corrosion rate decreased
dramatically for aluminum brass, but there were no significant changes
to the rates for the copper-nickel alloys. This suggests that the corrosion
products on aluminum brass in polluted seawater are cathodic to the bare
0.1 1
0 5 10
Velocity (mls)
15 20 25
1- 1- "- 18"Cl 50 mglL sand - 18"C/ 500 mglL sand -5O"Cl 50 mglL sand/
FIGURE 7.35 Erosion loss of NAB as a function of velocity, sand content, and
temperat~re.~~
1.8
1.6
1.4
1.2
P
E
-
5 1
a,
0.8
.-0
u)
2
W
0.6
0.4
0.2
0
No Addition + Sand + Sulfide + Sand and Sulfide
I 0 Al-Brass W 90l10 Cu-NI 0 70l30 Cu-Ni I

FIGURE 7.36 Effect of sand and sulfide on the erosion of some common condenser
TABLE 7.11 Condenser tube plate corrosion rates with and without sand
injection in seawater at 2 m/s6’
Alloy
Water (plate/tube) No sand +200 mg/L sand
Clean Al-Bronze/Ti 0.820 0.423
seawater Muntz/Ti 2.830 1.540
Muntz/Al-Brass 0.364 0.295
Muntd90/10 Cu-Ni 0.460 0.460
Muntd70/30 Cu-Ni 0.762 0.671
Sulfide polluted
seawater Al-bronze/Ti 1.570 0.824
Muntz/Ti 6.080 4.530
Muntz /Al-brass 2.410 2.300
Muntd 90110 Cu-Ni 2.930 2.020
Muntz/70/30 Cu-Ni 3.340 3.570
metal, exacerbating attack. The removal of some of this film by the sand is
probably the reason for the reduced corrosion rate (see Chapter 3 ) . Ault et al.
also observed that the corrosion rate decreased with time to a steady value
in clean seawater, but when sand was present, there was a slow but steady
increase in the corrosion rate over a period of 90 days.67The corrosion rate
approximately doubled between day 10 and day 80.
Besides investigating condenser tubes, the study also examined tube
plate corrosion.67 The authors investigated aluminum bronze, Alloy D
(C61400), coupled to titanium tubes, and muntz metal coupled to tita-
nium and the three copper condenser tube alloys. The results, given in
Table 7.1 1, show that sand had little effect on the corrosion of muntz metal
coupled to any of the copper alloy tubes. However, the corrosion rate of both
muntz metal and aluminum bronze decreased in both clean and polluted
water when sand was added (reduction of 20% to 50%). The sand appeared
to be having its biggest effect on the current from the tubes. This sug-
gests that the sand was removing, or partially removing, a highly cathodic
film.
7.3 Prevention of Erosion Corrosion and Erosion

The following sections discuss the best ways to avoid corrosion at high
water velocities or in turbulent water. Recommendations for the avoidance
of erosion corrosion and erosion are separated into two sections, for clarity.
7.3.1 Erosion Corrosion

The best way to avoid erosion corrosion of copper and its alloys in fresh
water is to design to the velocities given in Section 7.1.1. In large buildings,
such as hospitals and office buildings, it is necessary to use a pumped system
for the hot water so that the feed to any tap from the loop is relatively short.
Some engineers specify a smaller-diameter pipe for the return leg of the
loop, assuming that much less water will be returning. Campbell pointed
out that failures have occurred in some installations because hot water use
was not that great, and hence, water velocities in the return leg were too
high, causing failures by erosion corrosion.68
The best erosion corrosion resistance for the copper-nickel alloys is
achieved with a minimum of 1.5% iron for 90/10 copper-nickel and 0.7%
iron for 70/30 copper-nickel, as described in Section 7.1.2.
Previously, the nature of erosion corrosion and the limitations of
copper alloys were discussed. However, what are important to the engineer
are the safe design velocities for components in a flowing system.
The work of Efird was described previously, in which he determined
the critical velocity at which the shear stress was sufficient to rupture the
film.31However, his results take no account of any turbulence in the system,
which can increase local velocities.
In long lengths of larger-diameter pipe (NPS 3 or 4 and greater), BS
MA1 8 recommends velocities as follows69:
Aluminum brass 3m/s
90AO copper-nickel 3.5m/s
70/30 copper-nickel 4m/s
This takes no account of restrictions or sharp bends, where local

velocities may increase by a factor of 5 or 10. BS MA 18 does recommend
that the safe velocity be decreased as the pipe diameter decreases below NPS
3 because of increased turbulence. The document recommends velocities
in the range 1.0-1.3 m/s for 19-mm outside diameter heat exchanger tube.
This is clearly too low as there is excellent experience at around 2 m/s with
copper alloys.
Table 7.12 shows some typical design velocities that have been used
for copper alloys in clean seawater. Note that impeller tip speed is not a
guide to water velocities in a pump because the water is also moving, and
it is relative velocity that is important. In piping systems with sharp bends,
the lower velocity in Table 7.12 is preferred. It is more important to keep
velocities low in thin-walled components, such as heat exchanger tubes (for
TABLE 7.12 Typical velocity limits for copper alloys

in clean seawater
Maximum velocity (m/s)
Alloy Heat exchanger Pipe (>NPS 3)
Admiralty brass 1.0-1.5 -
Aluminum brass 1.5-2.0 2.0-2.5
90/10 copper-nickel 2.0-3.0 3.0-3.5
70/30 copper-nickel 2.5-3.5 3.04.0
NAB - 3.5-4.5
which the typical wall thickness is 1.2 mm), compared with castings for
pumps and valves (with a typical thickness > 15 mm), where a little erosion
corrosion can sometimes be tolerated.
The velocities in Table 7.12 are meant as guidelines and must be used
in conjunction with good design practice. The important thing to note is that
there is a definite ranking order with copper alloys. The copper-nickel alloys
have superior resistance to erosion corrosion compared with the brasses,
and NAB has similar resistance to 70/30 copper-nickel. Table 7.12 does
not include Alloy 722 or 66/30/2/2 Cu-Ni-Fe-Mn, which can be ranked
equivalent to or better than NAB. However, these two alloys are more often
selected when there are suspended solids in the water (see Section 7.3.2).
The successful use of copper alloys and the avoidance of erosion
corrosion depend on the quality of the design. Utilization of the velocities
in Table 7.12 alone is not sufficient. It is important to avoid designs that
increase local turbulence. Hence 5-D bends are preferred to 3-D bends and
3-D bends are preferred to elbows. Traditional 90 degree elbows produce a
lot of turbulence just downstream of the bend, which can lead to erosion cor-
rosion. Smoothing out the flow with a radiused bend minimizes this effect.
Similarly, in heat exchangers, it is important not only to use a suitable
velocity for the tubes, but also to use a water box design that distributes
water evenly to all the tubes, without raising turbulence. Shone and Grimm
warned against just increasing water box dimensions proportionately when
designing a large heat exchanger.47To avoid turbulence, water boxes need
to be deep, particularly if sacrificial anodes must be accommodated. Also,
the water inlet should not be immediately opposite the tube plate because
this tends to force most of the water down the central core of tubes. An
angled inlet, or one at right angles to the tube plate, is preferred, with
internal baffles, if necessary, to ensure uniform water flow to all tube^.^"
Esso demonstrated the consequences of poor design in the 197Os,
when it built a model of a heat exchanger designed for a very large crude
carrier. With the inlet facing the tube plate and a shallow water box, there
was good water flow in the central tubes, but in the outer ring of tubes,
there was a steady flow of water in the opposite direction. The importance
of water box design was clearly demonstrated.
Caruso discussed the design of water boxes, including the problems
of a water box that is too deep, multipass heat exchangers, and feeding two
heat exchangers from a common water In heat exchangers, it is also
important to avoid entrained air, which can lead to t~rbulence.~" If there is
inlet end erosion corrosion due to excessive turbulence, it can be mitigated
in some cases by nylon tube inserts, some five or six tube diameters long.
It is important to taper the trailing edge of the insert to avoid it raising
turbulence there. If this is not done, the erosion corrosion is merely moved
from the tube inlet to the end of the insert.6s Erosion corrosion can also
occur on the steam side of a condenser if the inlet velocity of the steam is
too high. This is usually prevented by inserting a baffle at the inlet to spread
the steam
Page described the problems that can arise from the positioning of
condensers in power stations.71When the final condenser is high up and the
waste cooling water flows down from the discharge water box, it is possible
to create pressure conditions such that symphonic air release occurs in the
condenser. The release of air increases turbulence and can cause erosion
corrosion.
The ability of ferrous sulfate dosing to prevent erosion corrosion was
discussed in Section 7.1.3. It is almost always required with brass tubes,
particularly aluminum brass, and its use with copper-nickel alloys should
be considered, depending on water composition.
Where it is important to protect aluminum brass tubes immediately
on exposure to seawater, Sumitomo developed APF-coated t ~ b e s . APF ~~-~~
is a coating of hydrated iron oxide held in place with a polymeric binder. In
unpublished work, it was shown that the coating conferred good resistance
to erosion corrosion in turbulent water, but the coating was slowly eroded
away and had to be reinforced with regular ferrous sulfate dosing.
Where chlorination is carried out, to prevent fouling, it is important to
switch off the chlorine/hypochlorite when ferrous sulfate dosing is ongoing.
Safe chlorination levels to prevent acceleration of erosion corrosion for the
common condenser alloys are given in Section 7.1.4.
When sulfides are present in the water, either intermittently or con-
tinuously, it is important to select alloys carefully. Both 90/10 and 70/30
copper-nickel are more resistant than aluminum brass and 66/30/2/2 Cu-Ni-
Fe-Mn, and ferrous sulfate dosing will help if the pollution is intermittent.
For more heavily polluted waters, tin bronze tubes, such as AP bronze, are
better. Similarly, the high tin cast copper alloys are preferred to NAB for
pumps and valves. The 10% tin bronzes are best, but it can be difficult to
produce pressure-tight castings in these alloys. Hence, the 10% tin bronzes
are used for pump impellers and valve internals, while LG4 gunmetal is
used for pump cases and valve bodies. Even in clean seawater, if the pump
or valve is to spend significant time full of stagnant seawater, where H2S
may be produced, the high-tin bronzes are preferred to NAB.
Chlorine reacts with pollutants in seawater, such as HzS, with the
following reaction:
HIS + Cl2 + 2HC1+ S. (7.2)
Elemental sulfur is less corrosive toward copper alloys than H2S, and hence,
corrosion will be reduced.74 However, if there is excess chlorine, then the
corrosion rate can be greatly accelerated with some copper alloys such
as 90/10 copper-nickel, as described in Section 7.1.5. Cigna et al. used
the fact that oxidation of H2S reduces the aggressivity of seawater toward
copper alloys to solve corrosion problems at a Belgian power station.75The
tidal, brackish cooling water to four power stations caused severe corrosion
of the aluminum brass heat exchanger tubes after 1-2 years at two of the
stations. The corrosion was so severe that the stations were retubed in 70/30
copper-nickel. The third station had its aluminum brass tubes acid cleaned,
and measures were installed to prevent further accelerated attack. Water
analysis showed that H2S was present in the aerated water, and its presence
reduced the redox potential. By controlling the redox potential with small
doses of chlorine, Cigna et al. were able to prevent corrosion by sulfides
without accelerating corrosion by overdosing with chlorine.
7.3.2 Erosion
In waters containing suspended solids that may cause erosion, it is important
to keep velocities low and to avoid turbulence by good design, as described
in the previous section. Where the main problem is silt (particles up to
50 pm diameter), the data of Parker and Roscow suggest that aluminum
brass, 90/10 copper-nickel, and 70/30 copper-nickel may be used up to
-100 mg/L silt content without undue acceleration of corrosion.63 Alloy
722 and 66/30/2/2 Cu-Ni-Fe-Mn are more resistant and may be used up to
-400 mg/L silt content.
Where the problem is coarser particles of sand ( 2100 pm diameter),
the data of Ault et al. suggest that levels up to 100-200 mg/L sand will
roughly double the corrosion rate.67 This is surprising as it suggests that
sand is less of a problem than silt, whereas the reverse is actually correct.
Note that the data of Ault et al. were obtained with a recirculating test loop,
whereas those of Parker et al. were obtained from power station experiences
in the Severn estuary, United Kingdom. Hence, it would be prudent to limit
the use of 90110 and 70130 copper-nickel to a maximum of around 50 m g L
sand, whereas Alloy 722 and 661301212 Cu-Ni-Fe-Mn could be used up to
approximately 100 mg/L sand. Note that any source of local turbulence will
increase local erosion rates.
Any methods of reducing sand and silt contents are to be preferred.
Filters and screens are effective, provided that there is sufficient pumping
pressure to maintain flow as these block up. It is important to ensure that
filters are cleaned regularly and are not just bypassed when the filters get
blocked.
A common alternative is to use settling tanks. When the first fills up, a
second is used while the first is dredged to remove accumulated solids. It is
important that this is maintained. The author investigated erosion problems
with copper alloy piping at a plant that were caused by the postponing
of settling tank dredging to save money. The costs of shutdowns and pipe
replacements, due to the ingress of sand, were more than the savings realized
by not dredging the ponds.
References
1. J.R. Myers, M.F. Obrecht, MP 12 (1973): p. 17.
2. R.J. Oliphant, “Causes of Copper Corrosion in Plumbing System,” Foundation for
Water Research Report, FRR0007, May 2003.
3 . R. Francis, Galvanic Corrosion-A Practical Guide for Engineers (Houston; TX:
4. R. May, R.W. de Vere Stacpoole, J. Inst. Met. 77 (1 950): p. 33 I .
5. B. Todd, Trans. Inst. Marine Eng. 80 (1968): p. 161.
6. S. Bog, “The Resistance to Impingement Attack by Seawater of Copper Alloy
Tubes,” Fifth Scandinavian Corrosion Congress (Copenhagen, Denmark: 1968).
7. S. Bog, “The Resistance of Copper Alloys to Different Types of Corrosion in
Seawater,” Seventh Scandinavian Corrosion Congress (Trondheim, Norway:
1975).
8. G.L. Bailey, J. Inst. Met. 79 (1951): p. 243.
10. ASTM BILL, “Standard Specification for Copper and Copper-Alloy Seamless
Condenser Tubes and Ferrule Stock” (West Conshohocken, PA: ASTM).
11. K. Nagata, S. Sato, Sumitorno Light Met. Tech. Rep. 22 (1 981): p. 22.
12. D.M.M. Guyoncourt, S. Harper, BNF Metals Technology Centre Research Report,
RRA1913, February 1977.
13. H.G. Appel, H. Urselmann, M. Tappen, W. Beckmann, M. Hecht, M. Jasner,
“Suitability of CuNi 90/10 for Seawater Service,” EUROCORR 2001 (Riva del
Garda, Italy: European Federation of Corrosion, 200 1 ).
14. J.P. Frick, L.R. Scharfstein, T.M. Parill, G. Haaland, J. Met. (March 1985):
p. 28.
15. R. Francis, The Selection of Muteriuls,for Seawnter Cooling Systerizs: A Practical
Guidefor Engineers (Houston, TX: NACE International, 2006).
16. W.W. Kirk, INCRA Report, PRJ 382, April 1986.
17. D.W. Townsend, British Non-Ferrous Metals Research Association Research
Report, A1546, October 1965.
18. H.S. Campbell, D.W. Townsend, British Non-Ferrous Metals Research
Association Research Report, A1547, October 1965.
19. D.W. Townsend, British Non-Ferrous Metals Research Association Research
Report, A 1671, December 1967.
20. D.B. Anderson, K.D. Efird, “The Influence of Chromium on the Corrosion
Behaviour of Copper-Nickel Alloys in Seawater,” Third International Congress on
Marine Corrosion (NBS, 1972), p. 264.
21. N.W. Polan, M.A. Heine, J.M. Popplewell, C.J. Gaffoglio, CORROSIONW2,
paper no. 58 (Houston, TX: NACE International, 1982).
22. T. Nosetani, S. Sato, K. Kazama, Y. Yamaguchi, T. Yasui, Siirizitoriio Light Met.
Tech. Rep. 22 (1971): p. 85.
23 H.T. Michels, R.M. Kain, CORROSION/05, paper no. 233 (Houston, TX: NACE
24. D.J. Mills, BNF Metals Technology Centre Research Report, RRAl899,
September 1975.
25. H.S. Campbell, Aluminiunz Bronze Alloys Corrosion Resistririce Giii~le,
Publication 80 (Potters Bar, U.K.: Copper Development Association, 198 I ).
26. J.C. Rowlands, B. Angell, AMTE, Holton Heath, correspondence to author, 1978.
27. EL. LaQue, Marine Corrosiorz (New York, NY John Wiley, 1975).
28. H.S. Campbell, V.E. Carter, J. Inst. Met. 91 (1961-1962): p. 362.
29. W.K. Cheung, J.G.N. Thomas, “The Effect of Dissolved Copper on the
Erosion-Corrosion of Copper Alloys in Flowing ASTM Seawater,” ASTM STP
970 (1988): p. 190.
30. S. Sato, K. Nagata, S. Yamauchi, Sunzitonzo Light Met. Tech. Rep. 20 (1979): p. 32.
32. W.W. Kirk, INCRA Project Report, 396, July 1987.

33. R.J. Ferrara, J.P. Gudas, “Corrosion Behaviour of Copper Base Alloys with
Respect to Seawater Velocity,” Third International Congress on Marine Corrosion
(NBS, 1972), p. 285.
34. R.J. Ferrara, J.P. Gudas, “Corrosion Behaviour of Copper Base Alloys with
Respect to Seawater Velocity,” Third International Congress on Marine Corrosion,
J.B. Cotton, Discussion of Reference 32, ibid (NBS, 1972), p. 297.
35. R. Francis, M P 21 (1982): p. 44.
36. D.G. Melton, INCRA Project Report, 370, February 1986.
37. T.W. Bostwick, Corrosion 17 (1961): p. 12.
38. A.M. Lockhart, Proc. Inst. Mech. Eng. 179 (1964-1965): p. 459.
39. R. Francis, INCRA Project Report, 289, January 1979.
40. S. Sato, T. Nosetani, Surizitoiizo Light Met. Tech. Rep. 1 I (1970): p. 27 I .
41. S. Sato, K. Nagata, “Factors Affecting Corrosion and Fouling of Condenser Tubes
of Copper Alloys and Titanium,” Conference on Corrosion in the Power Industry
(Montreal, Canada: NACE International, 1977).
42. T. Nosetani, S. Sato, K. Onda, Y. Yamaguchi, S. Ando, S ~ ~ n i i f o r iLight
io Met. Tech.
Rep. 15 (1974): p. I .
43. S. Sato, M. Okawa, Siirnitoino Light Met. Tech. Rep. 17 (1976): p. 17.
44. S. Sato, T. Nosetani, Y. Yamaguchi, K. Onda, Siirizitoriio Light Met. Tech. Rep. 16
(1975): p. 23.
45. J.P. Gudas, H.P. Hack, D.W. Taylor, CORROSION/78, paper no. 23 (Houston, TX:
46. H.P. Hack, J.P. Gudas, D.W. Taylor, CORROSION/79, paper no. 234 (Houston,
TX: NACE International, 1979).
47. E.B. Shone, G.C. Grimm, Traizs. bzst. Marine Ens. 98 ( 1985): paper 1 I
48. S. Sato, Surnitonzo Light Met. Tech. Rep. 3 (1962): p. 276.
49. R. Francis, C.R. Maselkowski, BNF Metals Technology Centre Research Report,
R506, May 1985.
50. R.S. Bem, P.T. Gilbert, British Non-Ferrous Metals Research Association
Research Report, A 1095, September 1955.
5 I . R.S. Bem, British Non-Ferrous Metals Research Association Research Report,
Al132, December 1956.
52. J.P. Gudas, G.J. Danek, R.B. Niederberger, CORROSION/76, paper no. 76
53. J.P. Gudas, H.P. Hack, D.W. Taylor, CORROSION/78, paper no. 93 (Houston, TX:
54. T.S. Lee, H.P. Hack, D.G. Tipton, “The Effect of Velocity on Sulphide-Induced
Corrosion of Copper Base Condenser Alloys,” Fifth International Congress on
Marine Corrosion and Fouling (Barcelona, Spain: 1980).
55. J.C. Rowlands, J. Appl. Chem. 15 (1965): p. 57.
56. C. Christensen, “The Influence of Polluted Seawater on the Corrosion of Copper
Alloys,” Seventh Scandinavian Corrosion Congress (Trondheim, Norway, 1975).
57. M.W. Joseph, F.W. Hammond, INCRA Project, 349, Report A (1983, Report B
(1986).
58. R.S. Bern, British Non-Ferrous Metals Research Association Research Report,
A1233, March 1959.
59. T. Atsumi, A. Ogiso, K. Nagata, S. Sato, Sumitomo Light Met. Tech. Rep. 29
(1988): p. 23.
61. A. Neville, T. Hodgekiess, Surf: Eng. 12 (1 996): p. 303.
62. S. Sato, T. Nosetani, Y. Yamaguchi, K. Onda, Sumitonzo Light Met. Tech. Rep. 16
(1975): p. 23.
63. J.G. Parker, J.A. Roscow, “The Service Performance of Condenser Tube Materials
in Cooling Waters Containing Appreciable Solids,” Proceedings of the Eighth
International Congress on Metallic Corrosion (Mainz, Germany: 198I), p. 1285.
64. C.J. Gaffoglio, Power Eng. 86, 8 (1982): p. 60.
65. A. Neville, University of Leeds, correspondence to author, 2005.
66. J. Dallas, T.A. McConnell, “Design and Testing of a Protoype Slurry Pump for
FGD Plants,” Solids Pumping Conference (London: Inst. Mech. E., 199I).
67. J.P. Auk, G.A. Gehring, B.C. Syrett, M P 34 (1995): p. 63.
68. H.S. Campbell, “Erosion Corrosion in Copper Plumbing Systems,” Building
Services Journal, October 1984, p. 62.
69. BS MA18, “Salt Water Piping Systems in Ships” (London: BSI, 1973).
70. L. Caruso, “Cut Surface-Condenser Tube Problems by Paying Close Attention to
Design,” Power, August 1981.
71. G.G. Page, N. Z. J. Sci. 26 (1983): p. 415.
72. K. Nagata, S. Sato, Sumitomo Light Met. Tech. Rep. 27 (1986): p. 18.
73. T. Nosetani, Y. Hotta, S. Sato, Sumitomo Light Met. Tech. Rep. 28 (1987): p. 29.
74. B.C. Syrett, D.D. Macdonald, S.S. Wing, Corrosion 35 (1979): p. 409.
75. R. Cigna, K. De Ranter, 0. Fumei, L. Giuliani, Brit. Corros. J. 23 (1988): p. 190.
CHAPTER 8
Cavitation
Cavitation is a corrosion phenomenon that can result in the loss of large

amounts of metal over a relatively small area. It occurs when there is
a sudden drop in pressure, usually at a high flow velocity. This creates a
series of vapor cavities that may be up to 1 mm in diameter.' These cavities,
or bubbles, migrate along the pressure gradient, until the external pressure
is sufficient to cause their collapse. The energy release when they collapse
can be sufficient to mechanically remove metal from the adjoining surface.
This is termed cavitation damage and often presents a jagged, honeycomb
appearance, as shown in Figure 8.1
Cavitation can occur on ships' propellers, where the bubbles form
near the hub and migrate outward along the blades. Cavitation damage is
usually seen about one-third of the way along the blade. This should not be
confused with propeller wastage, which occurs at the tips of the propeller
blades. The latter is erosion corrosion, and it should not be confused with
cavitation damage, which has sometimes been termed cavitation erosion.
Campbell discusses the differences in detai1.2-3
Cavitation can also occur in pumps, particularly when they are being
operated under conditions well removed from the design optimum. Again,
cavitation damage occurs toward the eye of the impeller, while any damage
at the outer edge of the impeller is usually erosion corrosion. Cavitation is
also seen after partly throttled valves or orifice plates, where the pressure
drop is large and the flow is often high. Vibration may also cause cavitation,
and damage to the cylinder liners of large marine diesel engines has been
attributed to cavitation.
Cavitation proceeds broadly in four stages, as shown in Figure 8.2.4
The first stage is incubation, and the mass loss is small. However, pitting
165
FIGURE 8.1 Cavitation damage, showing typical jagged, honeycomb appearance. (See
color section.)
Stage 1 : Stage2 : Stage 3 Stage 4
Time
FIGURE 8.2 Schematic showing the four stages of metal loss during cavitation4
166
Cavitation 167
will occur in ductile materials, whereas brittle materials will show cracking
or pitting and cracking. In stage 2, the process accelerates, and significant
metal loss occurs, which gradually spreads over the whole cavitation area.
In the third stage, the metal loss rate is almost constant, corresponding to
metal loss over the whole cavitation zone. In the fourth stage, the metal loss
rate decreases, either because the loss of metal is now so great that the local
flow conditions have changed and the cavitation intensity is reduced, or
because the metal surface is now protected by a layer of stagnant water. Not
all these four stages may show easily during any specific test for cavitation
re~istance.~
The resistance to cavitation shows a strong correlation with the hard-
ness of the material. However, the direct correlation between the two shows
a lot of scatter, and it is clear that material strength and ductility are also
important.'
The foregoing description of cavitation suggests that it is purely me-
chanical and, as such, the application of cathodic protection would have
no effect. However, in corrosive fluids, it has been observed that there is a
synergistic effect between cavitation and corrosion, similar to that between
erosion and corrosion, as described in Chapter 7. Hence, the total mass loss
is given by
MTotal = MCav + MCorr f MSyn (8.1)

where
M T =~total~ mass
~ ~loss
Mcav = mass loss due to cavitation
Mcor = mass loss due to corrosion
Msyn= synergistic mass loss
Wood and Fry described cavitation tests of pure copper and 70/30 copper-
nickel in seawater using a cavitation t ~ n n e lBy
. ~ separating each of the cav-
itation and corrosion effects, they demonstrated that there was a synergistic
effect for both materials. Furthermore, the magnitude of the synergistic
metal loss rate was approximately the same for both materials.
8.1 Cavitation Test Methods
There are a number of methods of studying cavitation, and it is impor-

tant to understand the differences between the different techniques when
comparing data derived by these methods. The simplest test method is based
on a 20k Hz oscillator, in which the material under test is fastened to the end
of the oscillating arm. This creates cavitation of high intensity, resulting

in significant damage in a short time. It is useful for comparing materials,
but the data cannot easily be related to hydrodynamic cavitation, say, in a
pump. The test has been standardized and is described in ASTM G32.6
Another test, which is hydrodynamic, involves a disc rotating at high
speed in the fluid of interest. A hole or pin is positioned near the rim of
the disc, and this induces cavitation bubbles. In a variation of this test, a
plate with a hole in it is placed in a high-velocity laminar fluid flow, again
inducing cavitation bubbles.
In the cavitating jet test, a long orifice nozzle is submerged in liquid,
and the liquid is pumped at high pressure through the nozzle, which causes
cavitation bubbles to form. These impinge on a sample placed close to the
jet. This test has been standardized and is described in ASTM G134.7 The
advantage of this test is that it is hydrodynamic, and also, it enables the
cavitation number (a)and the flow velocity to be varied, where
CT = ( P d - Pu)/0.5(px V2), (8.2)
where
Pd = static pressure in the downstream chamber
P, = vapor pressure
p = liquid density
V =jet velocity
It is relatively easy to calculate o values for some service applications in
which cavitation is occurring and so relate experimental data to service
performance.
The final test method is a cavitation tunnel, in which the fluid is
accelerated by a venturi and cavitation is induced by an obstruction, often a
cylinder, at right angles to the flow. The specimen may be the obstruction,
or it may be a panel that is flush mounted into the tunnel wall near the
obstruction. This reproduces many cavitating service environments well,
but testing is costly and time consuming.
Note that many reports on cavitation testing quote volume metal
loss, rather than mass loss. This is because material densities can vary
significantly, and volume loss gives a clearer idea of the scale of the damage.
8.2 Performance of Copper Alloys
It was stated previously that material hardness has a significant effect

on cavitation resistance. Hence, pure copper will have relatively low
Cavitation 169
TABLE 8.1 Results of ultrasonic cavitation tests in fresh waterX

Alloy UNS no. Cavitation rate (rnm3/h)
Manganese bronze C86550 4.7
Admiralty gunmetal C90600 4.9
Mn-Al-bronze CMA 1(*) 1.5
Al-bronze C95200 0.8
NAB C95800 0.6
Alloy K-500 (cold drawn) NO5500 2.8
Alloy K-500 (aged) NO5500 1.2
Type 316 stainless steel S3 1600 1.7
(A) No UNS number, BS1400 designation quoted
resistance, while harder copper alloys, such as manganese bronze and alu-
minum bronze, will be significantly better.
Campbell describes the results of some ultrasonic cavitation tests
(ASTM G32) in fresh water.' The results, given in Table 8.1, show that the
resistances to cavitation of manganese bronze and admiralty gunmetal were
much less than the resistances of the aluminum bronzes. The aluminum
bronzes were superior not only to other copper alloys, but also to alloy
K-500 and Type 316 stainless steel (SS) (UNS 531600). Some similar tests
in 3% sodium chloride (NaC1) (Table 8.2) again showed nickel aluminum
bronze (NAB) to be superior to manganese bronze and Type 321 SS (UNS
S32100).'
Angel1 shows the results of tests on a range of copper and other alloys
obtained in seawater in a cavitation tunnel.' The data were generated at a
velocity of 40 m/s and a cavitation number (a)of 0.08, which represents
severe cavitation conditions. The results, given in Figure 8.3, show the
range of metal loss rate for several alloys in some of the categories such as
nickel alloys. It is clear that the ranking is somewhat different to the ranking
obtained in the ultrasonic cavitation test. One reason is that Angell's data
were obtained under real hydrodynamic cavitation conditions; a second is
that Angell's tests were in natural seawater, where there would have been
a significant contribution from corrosion and hence also the synergistic
effect, as discussed previ~usly.~ The aluminum bronzes were significantly
TABLE 8.2 Depth of attack after ultrasonic cavitation tests in 3%

sodium chloride solutions
Alloy UNS no. Depth of attack (mrn) Test time (h)
Manganese bronze C86550 0.280 6
NAB C95800 t0.025 7
Type 321 stainless steel S32100 0.305 7
TABLE 8.3 Depth of cavitation after testing in seawater for 30 days

Alloy Velocity (m/s) Temperature (“C) Depth (mm)
Manganese bronze 122 27 0.05
Mn-Al-Bronze I22 27 0.08
70/30 copper-nickel 121 6 0.56
90/10 copper-nickel 130 8 0.03
90/10 copper-nickel 128 27 0.03
Alloy 400 121 26 >I
Alloy K-500 132 6 0.23
Alloy K-500 121 27 0.03
Type 3 16 stainless steel 122 9 0.02
Type 3 16 stainless steel 122 12 0.25
more resistant to cavitation than the other copper alloys and were on a par
with the nickel alloys, but they were not as resistant as the austenitic and
duplex SS. Note that with a different cavitation number and flow velocity,
the relative resistance of materials can change.
The LaQue Laboratory, in the United States, carried out unpublished
cavitation tests on a wide range of alloys. The materials were tested as flat
plates with a small hole 25 mm from the leading edge in a high-velocity
laminar seawater flow for 30 days. The results, shown in Table 8.3, are
difficult to compare with Angell’s data because the latter were expressed as
Cast Iron
Gunmetals
Copper-Nickels
Manganese Bronze
Nickel Alloys
i
Austenitic Stainless Steel
Duplex Stainless Steel

1I
0
I
i 2 3 4
I
Volume Loss Rate (mmYh)
FIGURE 8.3 Cavitation data obtained in seawater with a cavitation tunnel at 40 m/s and
a cavitation number of 0.08.’
Cavitation 171
volume loss rate, whereas the LaQue data are expressed as depth.’ Hence,
a shallow depth could represent a high metal loss if the attack was over
a wide area. Similarly, deep attack might represent a low metal loss if it
were very localized. Depending on the component in question, the depth of
attack may be critical in some cases, whereas the extent of the attack may
be important in others.
The LaQue data show Alloy 400 to have a low resistance to cavitation,
whereas Alloy K-500, which is stronger and harder, has better resistance.
The results also show 70/30 copper-nickel to be inferior to the 90/10 alloy
and also much worse than manganese bronze and manganese aluminum
bronze (CMA1). Unfortunately, aluminum bronze was not included in these
tests. The results for Type 316 SS show the variability possible in this kind
of test. It is not clear whether the results for Alloy K-500 show a real effect
of temperature or whether they just represent typical test-to-test variation.
8.3 Avoidance of Cavitation

Where cavitation occurs in service, one option is to upgrade to a more
resistant alloy. NAB is the most cavitation-resistant of the copper alloys,
and it has given good service in pumps and valves around the world.
Where the cavitation is intermittent, perhaps only under certain operating
conditions, this is probably the simplest solution, although not necessarily
the lowest-cost solution.
Where the cavitation is more or less constant, or it is intermittent
but particularly severe, a design change is really the best solution. If
the problem is with a valve or orifice plate, then decreasing the pressure
more gradually will prevent cavitation. This might mean replacing a single
orifice plate with a series of spaced plates, each producing a smaller pres-
sure decrease than the original plate. Where the problem is in a pump, then
the options are to change the operating conditions nearer to its design condi-
tions or to replace the pump with one more suited to the changed operating
conditions.
References
1. B. Angell, “Cavitation Damage,” in Corrosion, 3rd ed., L.L. Shreir, R.A. Jarman,
G.T. Burstein, eds. (Oxford: Butterworth Heinemann, 1994), p. 8: 197.
2. H.S. Campbell, Report to the Inter Service Metallurgical Research Council,
ISMET 2945, CED 526, August 1962.
3. H.S. Campbell, V.E. Carter, J. Inst. Met. 90 (1961-1962): p. 362.

4. P.A. Lush, C M E J . 34 (1987): p. 31.
5. R.J.K. Wood, S.A. Fry, J. Fluids Eng. 11 1 (1989): p. 271.
6. ASTM G32, “Standard Test Method for Cavitation Erosion Using Vibratory
Apparatus” (West Conshohocken, PA: ASTM).
7. ASTM (3134, “Standard Test Method for Erosion of Solid Materials by a
Cavitating Liquid Jet” (West Conshohocken, PA: ASTM).
Publication 80 (Potters Bar, U.K.: Copper Development Association, 198I).
CHAPTER 9
Dealloying
Dealloying occurs when a component of an alloy is selectively dissolved.

When one element is selectively removed from a metal, for example, zinc
from brass, it is termed dealloying. Sometimes one phase in a multiphase
alloy is corroded; this is known as selective phase attack. Since the two are
closely related, they are treated together in this chapter.
9.1 Dezincification
Probably the most well-known form of dealloying is the selective removal of

zinc from brass, which is known as dezincijication. Lucey showed that both
zinc and copper are dissolved together as the anodic reaction, and copper is
'
redeposited at the advancing dezincified front as the cathodic reaction. This
leaves behind a spongy mass of copper with the same shape as the original
material, but with no significant mechanical strength. Dezincification has
been observed to form in layers over the whole surface, but it can also form
plugs in localized regions. Lucey also showed that dezincification can be
stimulated when crevice corrosion is occurring. '
Dezincification can lead to a number of problems. In certain soft
waters, particularly hot waters, blockage of brass fittings can occur. They
become filled with a voluminous, but hollow, white corrosion product, not
unlike meringue (Figure 9.1). This is very different to hard water scale of
calcium carbonate (CaC03), which is hard and compact.
If dezincification carries on for several months or years, then it can
penetrate the wall of a brass fitting or a brazed joint. Usually, leakage is
slow, unless the soft, residual copper plug is disturbed. Over a period of
time, the mechanical strength of a component decreases significantly with
173
FIGURE 9.1 Meringue dezincification blocking a duplex brass fitting. (Photograph cour-
tesy of Copper Development Association U.K.) (See color section.)
time as dezincification propagates. Figure 9.2A shows a brass wire from a

toilet cistern in a soft water area. The fracture face has the typical coppery
appearance associated with dezincification. The microsection of the wire
(Figure 9.2B) shows that a large part of the cross section of the wire had
dezincified.
Single-phase, alpha brasses containing up to about 15% zinc do not
suffer from dezincification. This includes cast alloys, such as the gunmetals,
which are all immune to dezincification. Single-phase brasses containing
15% to -36% zinc will dezincify, but this can be prevented by the addition
of small quantities of arsenic. This is why specifications for alloys like
aluminum brass (C68700) require an addition of 0.02 to 0.06% arsenic.
Dealloying 175
(B)
FIGURE 9.2 Dezincification of brass wire: (A) fractured wire; (B) microsection showing
extent of dezincification. (See color section.)
Brasses containing more than -36% zinc also contain beta phase such
as muntz metal. It is not possible to make an alloying addition that will
prevent the dezincification of the beta phase. Included in this category are
the high-tensile brasses or manganese bronzes, which dezincify faster as
the beta phase content increases.
In the 1950s and 1960s, much effort was made to find an alloying
addition to prevent dezincification of brasses containing beta phase. The
reason for this effort is that the presence of beta phase makes a brass easy to
forge or hot stamp, whereas all alpha brasses may crack on forging. Weldon
investigated a wide range of alloying additions to brasses but found none
that totally prevented dealloying.* Reducing the beta phase content slowed
-
the attack down, but after a period of time, attack occurred on the alpha
phase as well. Weldon did find that the addition of 1% tin slowed down
dezincification of a 60/40 brass. This is why naval brass (containing I % tin)
is often preferred to muntz metal for heat exchanger tube plates in waters
that cause dezincification.
The problem of preventing dezincification in brasses used for forgings
or die-casting was finally solved by BNF Metals Technology Centre in the
1 9 7 0 Because
~~ of the shape of the copper-zinc equilibrium diagram, a
brass with -36% zinc will be two-phase at 800°C (forging temperature)
but all alpha phase at 500°C. A brass with this composition, with additions
of lead for machinability, became dezincification-resistant brass, or DZR
brass. DZR brass can be hot stamped at 800°C but is 100% alpha after
a 2-hour heat treatment at 500°C. Because forging brass often contains
impurities that can tie up arsenic, DZR brass requires a slightly higher
arsenic content than brasses for heat exchanger tubes to ensure freedom
from dealloying, typically 0.06 to 0.10%. DZR brass can also be die-cast
if -0.5% aluminum is added, which necessitates a slightly reduced zinc
content. Both these DZR alloys are now included in European Standards.
Campbell and Francis exposed components made of both DZR brass and
rolled naval brass (the duplex brass most resistant to dezincification) in
Portland Harbor seawater for one year: On examination, there was no
dealloying of the DZR brass, whereas the naval brass was dezincified to
a depth of 0.10-0.15 mm. In several places, the alpha phase between the
attacked beta phase was also dezincified. On the basis of these data, Lloyds
Register approved DZR brass for through hull fittings on yachts and small
craft.
Lucey showed that the effect of arsenic in a single-phase brass is to
increase the potential above that at which copper can precipitate, as the
cathodic reaction in dezin~ification.~ Instead, the cuprous chloride (CuC1)
Dealloying 177
hydrolyzes to form cuprous oxide (CuzO), which further acts to prevent

dissolution. With beta brass containing arsenic, the potential remains below
that at which copper can redeposit, thus explaining why arsenic does not
prevent dezincification of beta brasses.
The rate of dezincification can be greatly increased if the brass is
coupled to a large area of slightly more noble material such as copper.
Hence, when brazing copper, it is usual to use alloys that are not susceptible
to dezincification such as Cu-Ag-P. The acceleration occurs because the
brazed joint has a small area compared to the large surrounding cathodic
area of copper (see Chapter 18).
Dezincification can also occur externally on duplex brass fittings
used underground, if the ground water is aggressive. Hence, it is necessary
to select materials for fittings to be used underground with care where
dezincification may occur.
Not all waters will cause dezincification. Turner examined a large
number of waters to determine which ones would cause meringue
dezincification.6 The results in Figure 9.3 show that high-chloride, low-
temporary hardness waters with a pH > 8 caused meringue dezincification.
The laboratory tests compared well with the known behavior of a number
of waters from around the United Kingdom, as also shown in Figure 9.3.
Note that these composition limits apply only to meringue dezincification.
Dealloying can still occur with waters whose composition lies a little be-
low the curve in Figure 9.3, but the corrosion products will not block the
fittings. All waters with a high chloride content (such as seawater) cause
dezincification, but the zinc corrosion products are soluble, and blockage
does not occur. Hence, dezincification-resistant materials are required for
seawater and similar brackish waters.
Dezincification is also more likely in hot waters than in cold. This is
not just because the kinetics are greater at higher temperatures. Waters with
a pH below 8, but which fall within the shaded area of Figure 9.3, are likely
to cause dezincification when hot because heating usually causes a rise
in pH.
When waters are base exchange softened, the chloride-bicarbonate
ratio does not change; rather, only the sodium-calcium ratio changes. Hence,
it would be expected that such waters would still cause dezincification.
However, this is not the case because fully base exchange-softened waters
rarely have a pH >8, that is, a pH sufficiently high to cause dezincification.
In central heating systems, duplex brass fittings are used without
problems, despite the high temperatures. This is because the dissolved
oxygen (DO) content falls to a very low level a few hours after first
CLITHEROE A
a
PITSFORD :
GAINSBAOROUGH
20 .
OLDHAM
A LANCASTER
A LEEDS
10. 0 0 0 0
AHEREFORD
0 0 0 0 0 0 0
0 0 0 0 0
0
0 10 20 30
o Laboratory experiments in which meringue

dezincification did not take place
Laboratory experiments in which meringue
dezincification took place
D Public water supplies that are known not

to produce meringue dezincification
A Public water supplies that are known
to produce meringue dezincification
lool / GUERNSEYA
0 0
0
.
=
rn
A
GLOUCESTER
(Newent)
..
E
e
&I 6 0 - HA INGFIELD.
P LINCOLN
-b
c
. --
..
A PENZANCE
0 0
" 40 SWINDON
RTHAMES
0
ACIRENCESTER
20
P o A
%l'AFFORDSHIRE STROUD
n
I ,
0 50 100 150 200 250

Temporary Hardness (mgiL CaC03)
FIGURE 9.3 Influence of water composition on meringue dezincification.6

Dealloying 179
starting the system. Without DO in the water, for the primary cathodic
reaction, dezincification does not occur, and duplex brasses can be used,
even if the water composition is liable to cause dezincification in aerated
waters.
9.2 Dealuminification
Aluminum bronzes, like brasses, can be two-phase. Dealloying of single-

phase bronzes (up to -8% aluminum) is unknown, but aluminum contents
above 9.1% will cause formation of the gamma 2 phase in binary aluminum
bronzes if they are cooled too slowly. Gamma 2 phase is not selectively
attacked in free-flowing fresh water or seawater but dealloying can occur
under crevice corrosion, under deposits or marine growths, or under the
influence of galvanic corrosion. However, additions of 2% or more of iron
suppress the formation of gamma 2 phase, resulting in an alpha-beta struc-
ture. Unlike brasses, the beta phase in aluminum bronzes is not detrimental,
but rather improves the resistance to erosion corrosion and ~avitation.~
Manganese stabilizes the beta phase, and in copper-manganese-
aluminum alloys (containing typically 12% Mn and 8.5% Al), the beta
-
phase has a different composition to that formed in alloys containing only
1% manganese. In copper-manganese-aluminum alloys, the beta phase
will dealloy in a similar manner to the beta phase in brass. The dealloy-
ing is more likely to occur under stagnant or shielded conditions. As the
beta phase is continuous in these alloys, the attack can be severe, and the
copper-manganese-aluminum alloys should be avoided in seawater when
shielded areas or static conditions are l i k e l ~ . ~
In the nickel aluminum bronzes, alpha and beta phases form at high
temperatures, and the beta phase transforms to alpha and kappa on slow
cooling.' If the alloy is cooled quickly, then residual beta phase can be
retained, and this phase is highly anodic to the alpha phase. Welds are
particularly sensitive to this as they usually cool very quickly, leaving some
beta phase in the weld metal. It is not unknown for a seam-welded pipe that
has not been postweld heat treated to unzip in seawater as the beta phase
is corroded away (see Chapter 18). The transformation to alpha and kappa
for castings can be achieved by a final heat treatment at 675°C for 6 hours,
followed by air cooling. A higher temperature of 725°C is recommended
for wrought product^.^
Even when properly heat treated, nickel aluminum bronze (NAB)
can suffer selective phase corrosion. This is sometimes on the alpha phase
adjacent to the lamellar kappa I11 phase, but it is seen more frequently on
the kappa I11 phase. Because the phase is present in long, thin stringers,
penetration rates can exceed 1 mm/y." The reason for this was explained by
Rowlands, who showed that the alpha phase is anodic to the kappa I11 phase
in normal seawater at pH 8." However, at pH 3, there is a polarity reversal,
and kappa I11 becomes anodic to the alpha phase. Hence, shielded areas,
where the local pH can decrease to 3 , are more likely to see preferential
attack of the kappa I11 phase. Rowlands also demonstrated that if NAB is
exposed immediately to flowing seawater, a protective film forms, and no
selective phase attack occurs. However, if the NAB is exposed to very slow
water flows (-0.1 m/s) or to alternating stagnant and flowing conditions
(with stagnant first), then localized attack of the kappa I11 is likely. Heat
treatment at 675°C did not prevent this happening, but it did delay the
initiation of attack.
When the attack is on alpha phase adjacent to kappa 111, this usually
occurs in the heat-affected zone of a weld, and heat treatment does not seem
to affect attack in this case."
Maselkowski and Francis exposed in seawater plain and welded
coupons of cast NAB that had been heat treated at 675°C and air cooled.12
Some were in natural seawater with a nominal flow rate of 0.15 m/s, while
others were in similar tanks that also contained 0.25 mg/L chlorine. The
potentials were monitored with time over a period of 7 months, and the
results are shown in Figure 9.4. Over the first 70 days or so, the potential
- -
moved electropositively from -220 mV to - 150 mV Ag/AgCl (sil-
-
verhilver chloride). The initiation of selective phase attack was marked by
a sharp increase in potential from - 150 mV to -50 mV. Welded material
was more susceptible to corrosion than plain material. The addition of 0.25
m g L chlorine retarded the initiation of selective phase attack, particularly
for plain material, but after 150 days, all the samples in both tanks were
suffering selective phase attack.
Michels and Kain investigated the effect of composition and heat
treatment on the selective phase attack of NAB.13 They tested an alloy
with 9.4% aluminum to C95800 and a similar alloy with 8.1% aluminum.
The second alloy actually falls outside the minimum aluminum content
for C95800 (8.5%). These alloys were exposed in quiescent seawater for
6 months. The alloy with 8.1% aluminum was much more susceptible to
selective phase attack, and the attack was also deeper. Material heat treated
at 675°C for 6 hours generally suffered less attack, but was not immune.
It has been suggested that the resistance of NAB to selective phase
attack is improved if the nickel content exceeds the iron content by 0.5%
or more.7 The nickel exceeded the iron content for both alloys tested by
Michels and Kain, but selective phase attack still occurred.
1- - Welded -Plain 1
(B)
FIGURE 9.4 Potential-time curves for nickel aluminum bronze (NAB) in quiescent sea-
water at ambient temperature: (A) natural seawater; (B) chlorinated seawater."
TABLE 9.1 Depth of attack of nickel aluminum bronze coupled to

superduplex stainless steel after 60 days in seawaterI4
Couple alloy Area ratio Chlorine (mg/L) Crevice Nut Open surface
SD(*) 1:l 0 0.27 0.03 1.19
SD 1:l 0 0.35 0.01 0.29
SD I:1 0 0.27 0 0.22
SD 1:l 0.7 0 0 0
SD 1:l 0.7 0 0 0
SD 1:l 0.7 0 0 0
NAB 1:l 0 0 0 0
NAB 1:l 0 0.01 0 0
NAB 1:l 0 0 0 0
NAB 1:l 0 0 0 0
NAB 1:l 0.7 0 0 0
NAB 1:l 0.7 0 0 0
NAB 1:l 0.7 0 0 0
NAB 1:l 0.7 0 0 0
SD 10:1 0 0.35 0 0.46
SD 10:1 0 0.36 0 0.47
SD 10:l 0 0.39 0 0.27
SD 10:l 0.7 0.20 0.01 0
SD 1O:l 0.7 0.22 0 0
SD 10:l 0.7 0.49 0.01 0
(A)
SD = superduplex.
When NAB is galvanically coupled to a more noble material, such as

a high-alloy stainless steel (SS), then selective attack can be stimulated. The
author tested cast NAB that had been heat treated, coupled to both itself and
a superduplex SS,I4 with cathode-anode ratios of 1:1 and 10:1. The couples
were exposed in slowly refreshed tanks of natural seawater or seawater with
0.7 mg/L chlorine at -23°C for 60 days. Figure 9.5 shows the appearance of
some of the samples on removal. The main form of corrosion was selective
phase attack, and there was more of it on the samples exposed to natural
seawater. The attack took the form of both crevice corrosion and pitting on
open surfaces. Table 9.1 shows the maximum depth of attack of the NAB
for each couple. The attack was all along the kappa I11 phase. Because
the exposure was only for 60 days, no attack was seen on NAB coupled
to itself as this takes some 70 or more days to initiate.12 The results in
Table 9.1 clearly show the severe attack on NAB in natural seawater, both
on open surfaces and in crevices. With chlorine additions, open-surface
Dealloying 183
(6)
FIGURE 9.5 Appearance of NABhperduplex stainless steel couples after 63 days’ ex-
posure: (A) natural seawater; (6) chlorinated ~eawater.’~
(Photographs courtesy of RA
Materials.) (See color section.)
pitting was eliminated, and crevice attack only occurred at the higher area
ratio. This is because the cathodic reaction is less efficient in the presence
of chlorine compared with natural ~eawater.'~
The laboratory results are borne out by service experience. NAB
valves in a superduplex SS piping system on a North Sea oil platform
showed severe selective phase attack. This platform was cooled by natural
seawater. A similar platform, but using chlorinated seawater, has had few
problems with its NAB valves in the superduplex SS pipe work.15
9.3 Dealloying of Other Copper Alloys

DenickeliJCication is a term that has been used to describe the selective
phase attack of copper-nickel alloys. This has mainly been seen with 70/30
copper-nickel in heat exchangers, where process-side temperatures are high
and water flows are 10w.l~Such conditions can occur in refineries, where
process-side temperatures above 150°C are common. The problem is cured
by increasing the water flow to remove local hot spots. This type of attack
is rare, and because local high temperatures are required, it is discussed in
more detail in Section 9.4.
Reports of dealloying of other copper alloys are even rarer, and all
have occurred where local temperatures were high. There are a few reports
of dealloying of tin in cast tin bronzes in hot brine or steam and dealloying
of silicon from silicon bronze in high-temperature steam, containing acid
spe~ies.'~
9.4 Hot Spot Corrosion
Hot spot corrosion is characterized by pitting on the cooling-water side

of copper alloy heat exchanger tubes at sites of local high temperature.
Redeposited copper is often seen in the corrosion products.
A number of service failures were described by Breckon and Gilbert,
who showed that all the copper alloys commonly used in heat exchangers
could suffer from this type of attack.'* Failures in some cases occurred in as
little as a few months. BNF Metals Technology Centre examined a number
of failed heat exchanger tubes from service where hot spot corrosion was
believed to be the cause. Research was also undertaken to establish the
conditions under which it could occur. The results led the author to believe
that hot spot corrosion can be subdivided into three different types. l 9
Dealloying 1a5
9.4.1 Classical Hot Spot Corrosion

This type of corrosion occurs in essentially clean, unpolluted waters, where
the temperature on the product side of the heat exchanger is >13O"C. The
corrosion is characterized by pitting on the cooling-water side of the tubes
at sites of local high temperature, often with redeposited copper inside the
pits. Bem and Campbell conducted a number of tests in clean seawater
on the commonly used copper alloys and pure copper.*' Only pure copper
and 70/30 copper-nickel showed any appreciable attack at 130°C. Breckon
and Gilbert carried out tests from 90 to 170°C, but they did not report
detailed results, although they said that attack occurred in most cases.I8
However, initiation of this type of attack is a function of both flow velocity
and temperature, as was demonstrated by Boden in his experiments on
hot spot corrosion of pure copper in sodium chloride (NaCl) solutions.*'
Boden concluded that in aerated solutions, an oxygen concentration cell
is established, and only under deaerated conditions would thermogalvanic
effects become important in the initiation of hot spot corrosion.
9.4.2 Sulfide-Induced Hot Spot Corrosion

A number of the service failures reported by Breckon and Gilbert occurred
in water polluted with hydrogen sulfide (HzS), and the authors comment
that the attack may be exacerbated in polluted waters.I8 The effect of sulfide
was studied in more detail by Bem and Campbell, who showed that up to
a temperature of 7 5 ° C the rate of attack was very low, and that it greatly
increased between 75 and 100"C.20Above 100"C, there was no significant
change in the rate of attack with temperature. The critical temperature for
the acceleration of attack between 75 and 100°C has not been determined.
More recently, work has been carried out by Lloyd et al. and Shone
and Grimm on 66/30/2/2 Cu-Ni-Fe-Mn (C71640).22-23Some of the tankers
operated by Shell had suffered perforations due to corrosion near the outlet
end of both steam condensers and oil coolers. The corrosion appeared
to be similar to classical hot spot corrosion, but the nominal tube wall
temperature was only 60°C. Lloyd et al. tested tubes of the same alloy
in a model heat exchanger cooled by an estuarine water containing 0.1
mg/L sulfide.22When the tubes were tested at steam temperatures of 40 to
60"C, superficial corrosion occurred at cooling-water velocities of 2.0 and
0.25 m/s. In a third test, the water velocity was reduced to 0.14 d s , and
the steam side of the condenser was partitioned so that half the tubes at the
seawater inlet received steam at 50°C, while the other half received steam at
92°C. With such a low cooling-water velocity, air bubbles became trapped
in the upper part of each tube. Corrosion similar in appearance to that seen
in the service failures had occurred to a depth of 0.1 mm in the upper half
of the tubes toward the outlet end. Lloyd et al. concluded that the presence
of moving air pockets along the tubes would result in transient localized
temperatures that would exceed those seen in areas in contact with cooling
water. These results confirm the findings of Bem and Campbell that sulfide
reduces the temperature at which hot spot corrosion can occur.2o
The appearance of the pits is very similar to their appearance in
classical hot spot corrosion in that the pits often contain redeposited copper,
as shown in Figure 9.6A. To differentiate between classical and sulfide-
induced hot spot corrosion, it is necessary to test the corrosion products for
sulfides.
9.4.3 Ammonia-Induced Hot Spot Corrosion

The ability of ammonia (NH3) to induce hot spot corrosion was suspected
at BNF from an examination of service failures. The author examined
the effects of ammonia on the corrosion of copper alloys in seawater and
observed corrosion, similar to hot spot corrosion, occurring at very low flow
rates and only when heat transfer was simultaneously taking place.24Pitting
occurred on all the copper alloys tested when 2 m g L ammonia was present
in the water and there was a temperature difference of as little as 10°C
between the outer tube wall and the cooling water. Because the corrosion is
associated with low flows and very little heat transfer is required, the attack
might also be described as ammonia-induced deposit attack. However, in
the absence of a temperature differential, no localized corrosion occurred.24
The pits observed by Francis contained mostly copper oxide, with a
variable amount of redeposited copper distributed throughout the oxide, as
shown in Figure 9.6B.24 However, tubes that have failed in service in the
presence of ammonia often contain substantially more redeposited copper.
This is not surprising as the quantity of redeposited copper is likely to de-
pend on the temperature as well as the ammonia concentration and water
flow rate. From the preceding observations, it seems likely that there is no
lower temperature limit for hot spot corrosion in the presence of ammonia,
provided that there is some heat transfer and an as yet undetermined mini-
mum concentration of ammonia present. Both the laboratory tests and the
service failures indicate that ammonia-induced hot spot corrosion is only
likely to occur when there is little or no cooling water flow, for example, in
so-called dead areas in oil coolers.
Dealloying 187
(A) (A = copper; B = parent metal)
( B)
FIGURE 9.6 Microsections showing redeposited copper during hot spot corrosion:
(A) sulfide-induced; (B) ammonia-induced.lg
A low product-side temperature usually indicates that the presence

of ammonia has aided the initiation of hot spot corrosion. Ammonia can
sometimes be detected in the corrosion products within the pits, but because
of its volatile nature, its retention within the pit depends very much on the
tube history between removal and examination in the laboratory.
The effect of mixtures of ammonia and sulfide on the initiation of hot
spot corrosion has not been studied in detail, but Francis has reported results
obtained with 1 mg/L ammonia and 0.1 mg/L sulfide in aerated seawater.25
All the copper alloys tested suffered pitting in the same area of the test
samples and to the same extent, as was observed with 2 mg/L ammonia
alone.24 In both cases, the temperature difference between the tube outer
-
wall and the seawater was 10°C. Hot spot corrosion was not observed
with either pollutant added singly at that concentration.
Thus, it appears that the effect of ammonia in addition to sulfide is to
reduce the concentration of pollutants and the temperature at which pitting
can occur. Ammonia is often present in polluted harbor waters, along with
hydrogen sulfide, and service failures have been observed in the oil coolers
from ferries, where oil temperatures can vary from 50 to 80°C and water
flow rates are often low.
9.5 Avoidance of Dealloying

The avoidance of dezincification lies in the selection of dezincification-
resistant alloys for use in waters, or underground, where dealloying may
occur. The following guidelines should be followed:
0% to 15% zinc will not dezincify

15% to 36% zinc will resist dezincification if arsenic is added
>36% zinc will dezincify, although 1% tin will slow it down
Note that some elements have a high zinc equivalence such as silicon
and aluminum. Silicon has a zinc equivalence of 10; that is, 1% silicon
is equivalent to 10% zinc. Hence, it is important that these elements be
controlled when a single-phase, alpha structure is required that can be
inhibited with arsenic.
The author investigated the failure of some die-cast brass water meter
bodies that had dezincified in an aggressive water in the United Kingdom.
The bodies were supposed to be in DZR brass to BS EN 1982 (1988),26
but the silicon content was 0.23%, well in excess of the 0.02% maximum
permitted by the standard. The aluminum content was 0.3%, well within the
Dealloying 189
FIGURE 9.7 Half section of die-cast brass water meter body with fracture around base
due to dezincification. (See color section.)
permitted limit, but the silicon content was enough to make the brass two-
phase, with significant beta content. The plastic filter in the meter created
a crevice, just in a region where the meter body was thinnest. The dezinc-
ification in this area weakened the body so much that fracture occurred
(Figure 9.7).
It is advisable, when procuring dezincification-resistant alloys, to
specify testing to BS EN I S 0 6509 (1995).” This will demonstrate resis-
tance to dezincification and can also show if DZR brass has not been given
a final heat treatment to remove any residual beta phase.
With aluminum bronzes, there is no risk of dealuminification with
single-phase alloys such as C61400. The formation of gamma 2 phase can
be avoided in two-phase aluminum bronzes by ensuring that they contain a
minimum of 2% iron. With NAB, it is advisable to specify that the nickel
content must exceed the iron content by 0.5% or more. This is written in
some standards, but not all. This measure does not prevent selective phase
attack, but it does reduce the risk. As a final operation, that is, after hot
forming or welding, it is recommended that components be heat treated for
6 hours, followed by air cooling. The temperature should be 675 f25°C for
TABLE 9.2 Relative susceptibility to hot spot corrosion of some

common copper alloys (See color section.)
Susceptibility to hot spot attack
Alloy UNS no. Classical Sulfide Ammonia
Admiralty C44300 Erratic results Erratic results NT(*'
brass
Aluminum C68700
brass
90110 Cu-Ni C70600 Intermediate Intermediate
70130 Cu-Ni C7 1500

C7 1500
661301212 C7 1640
Cu-Ni-Fe-Mn
Alloy 722 C72200
(A)
NT = not tested.
castings and 725 f 25°C for wrought products. This ensures freedom from
residual beta phase and also improves the mechanical properties slightly.
To prevent selective phase attack of the kappa I11 phase in NAB,
it is advisable to put the component into flowing water soon after first
immersion so that a protective film forms quickly. This seems to confer
a high resistance to subsequent dealloying.'O It is interesting to note that
industry, which usually puts materials into service as quickly as possible,
reports few problems with selective phase corrosion of NAB, while the
Royal Navy, whose vessels spend a long time with intermittent flow in
fitting-out basins, has had numerous problems.
Dealloying in other alloys appears to be associated with high temper-
atures and low water flows. Hence, dealloying can be avoided by keeping
water flows high, 2 1.5 mls.
The relative susceptibility of the copper condenser alloys to hot spot
corrosion is shown in Table 9.2. Generally, aluminum brass has the best
resistance, particularly when ferrous sulfate (FeS04.7H20) dosing is being
carried out. The best way of preventing hot spot corrosion of all types is to
increase the water flow. A minimum of 1.5 m / s should ensure that hot spot
corrosion cannot initiate. The problems of hot spot corrosion in tankers, de-
scribed by Lloyd et al., were also remedied by increasing the cooling-water
flow rate.22In large tankers, this is achieved by switching from scoop feed
to pump feed when traveling at slow speed. Where increasing the flow is
not possible, sometimes a modification is possible to eliminate dead or
Dealloying 191
slow-flow areas. Francis quotes two cases in which ammonia-induced

hot spot corrosion occurred in heat exchangers tubed with 70/30 copper-
In both cases, remedies were effected by small changes in the heat
exchanger design and mode of operation, to eliminate slow-flow or dead
areas. This should be equally effective in combating classical or sulfide-
induced hot spot corrosion.
Francis showed that tubes that were dosed with iron did not suffer hot
spot corrosion, even with 2 mg/L ammonia in the water.24The author used
ferrous sulfate dosing for 2 months at the start of a multistage flash (MSF)
desalination plant to form more protective films on 90/10 copper-nickel
heat exchanger tubes. The water was intermittently polluted with ammonia,
which had caused hot spot corrosion in the heat rejection section when
the first unit was started. This was retubed, and ferrous sulfate dosing was
started from new at a level of 1 mg/L Fe2+ for 1 hour per day. This was
continued for 60 days, until protective films were judged to have formed.
The plant has now been in operation since 1994 with no further failures. At
another MSF plant in the Middle East, the ammonia pollution was almost
continuous because the water intake was near the outfall of a fertilizer plant.
Daily dosing of ferrous sulfate throughout its life was recommended for
this plant, as it was tubed with aluminum brass. There had been failures
of 90/10 copper-nickel tubes in an earlier plant on the same site. The MSF
plant has now been in operation since 1998 with no problems.
Where sulfide or ammonia and sulfide are present, ferrous sulfate dos-
ing may not be successful. An alternative could be to use low doses of chlo-
rine or sodium hypochlorite (NaOCl) at the cooling-water inlet to increase
the redox potential. This technique was used by Cigna et a]. to solve cor-
rosion problems due to sulfide in a Belgian power station.28The aluminum
brass tubes had suffered corrosion during the first year of operation due to
sulfides. The tubes were acid cleaned to remove the harmful films, and the
redox control system was instituted. There have been no further corrosion
problems. Some initial work would be required to identify the optimum
redox potential for each specific application.
References
2. B.A. Weldon, BNF Metals Research Association Research Report, A1 148, April
1957.
3. J.E. Bowers, P.W.R. Oseland, G.C. Davies, Brit. Corros. J. 13 (1978): p. 177.
4. H.S. Campbell, R. Francis, “Dezincification Resistant Brass, CZ132, in Seawater,”

The Metallurgist, April 1983.
6. M.E.D. Turner, Proc. Soc. Water Treat. Exam. 10 (1961): p. 162.
8. G.W. Lorimer, F. Hasan, J. Iqbal, N. Ridley, Brit. Corros. J. 21 (1986): p. 244.
9. R. Francis, The Selection of Materials for Seawater Cooling Systems-A Practical
Guide for Engineers (Houston, TX: NACE International, 2006).
10. J. Rowlands, “Studies of the Preferential Phase Corrosion of Cast Nickel
Aluminium Bronze in Seawater,” Eighth International Congress on Marine
Corrosion and Fouling (Mainz, Germany, 1981).
11. E.A. Calpan, G. Rose, Brit. Corros. J. 14 (1979): p. 160.
12. C.R. Maselkowski, R. Francis, BNF Metals Technology Centre Research Report,
R506 (Supplementary Report), July 1985.
13. H.T. Michels, R.M. Kain, CORROSION 2003, paper no. 262 (Houston, TX:
16. A Guide to the Selection of Marine Materials, 2nd ed. (INCO, 1973).
17. R. Steigerwald, Metallurgically Injuenced Corrosion, ASM Handbook, vol. 13
(Materials Park, OH: ASM International, 1987), p. 130.
18. C. Breckon, P.T. Gilbert, “Corrosion of Condenser Tubes Under Conditions of
Local High Temperatures,” Proceedings of the First International Congress on
Metallic Corrosion (London, U.K.: Butterworths, 1961), p. 625.
20. R.S. Bem, H.S. Campbell, “Laboratory Hot Spot Corrosion Tests of Condenser
Tube Alloys,” Proceedings of the First International Congress on Metallic
Corrosion (London: Butterworths, 1961), p. 630.
21. P.J. Boden, “Experimental Investigation and Evaluation of the Hot Spot Corrosion
of Copper” Proceedings of the Second International Congress on Metallic
Corrosion (Houston, TX: NACE International, 1963), p. 771.
22. K.A. Lloyd, R.A. Shaw, E.B. Shone, “Effects of Operating Temperatures and
Seawater Flow Rates on the Corrosion of 70/30 Copper-Nickel Heat Exchanger
Tubes” Proceedings of the Symposium “Corrosion in Heat Transfer Conditions”
(London, U.K.: SCI, 1982), p. 65.
Dealloying 193
23. E.B. Shone, G.C. Grimm, “25 Years Experience with Seawater Cooled Heat
Transfer Equipment in the Shell Fleets,” Trans. Inst. Marine Eng. 98 ( 1985):
paper 11.
26. BS EN 1982, “Copper and Copper Alloys: Ingots and Castings” (London, U.K.:
BSI, 1988).
27. BS EN IS0 6509, “Corrosion of Metals and Alloys: Determination of
Dezincification Resistance of Brass” (London, U.K.: BSI, 1995).
28. R. Cigna, K. De Ranter, 0. Fumei, L. Giuliani, Brit. Corros. J. 23 (1988): p. 190.
CHAPTER 10
Stress Corrosion Cracking
Stress corrosion cracking (SCC), or emironinentally assisted cracking, re-

quires a specific corrosive environment and a simultaneous stress for it to
occur. The environments that can cause SCC are often specific to a metal or
alloy, for example, the cracking of titanium in methanol. There are several
environments that can cause SCC of copper and its alloys, of which the best
known is ammonia (NH3). This and others of commercial significance are
reviewed in this chapter.
The stress to cause cracking can be an externally applied load due to
service conditions or an internal stress due, for instance, to welding or cold
working. SCC is not a new phenomenon; so-called season cracking of brass
cartridge cases and other brass items has been known for over 1.50 years.
The environment that caused the cracking was airborne ammonia, and the
stress was the residual stress in the deep-drawn cartridge cases.
There are a number of reviews of SCC of copper alloys that are useful,
particularly where they evaluate apparently conflicting results. Whittaker
reviewed SCC data up to 196.5 for the International Copper Research As-
sociation (INCRA) in the United Kingdom (now the International Copper
Association, New York).’ Campbell updated this review in 1973, while
Ledheiser did an update for INCRA in 19851.”~McEvily produced an atlas
of stress corrosion curves in 1990, but there is a paucity of data for some
alloys and some environment^.^ Whittaker and Ledheiser both commented
on the shortage of data in many areas and, particularly, for many cast cop-
per alloys, and also some common wrought ones. In the 20+ years since
Ledheiser’s review, there has been little work done to fill this gap with
engineering data.
There are some general comments about SCC that are worth making
before discussing specific environments. The first is that there are a number
195
of environments that can cause SCC of copper alloys, and the alloy ranking
order with regard to SCC resistance can change with the environment.
When SCC occurs under natural atmospheric conditions, it is common for
ammonia, sulfur dioxide (SO*),and chloride all to be present, and the effects
of each one are not easily separated. In the atmosphere, a high humidity and
a low temperature favor cracking because condensation is more likely. Rates
of reaction increase with temperature, and this can increase the risk of SCC.
However, this is more important under immersed conditions, rather than
atmospheric exposure, in which higher temperatures increase the rate of
evaporation and decrease the time of wetness. Cracking may also correlate
with rainfall because a thin film of moisture is necessary for corrosion.
However, high corrosion rates mean that SCC is unlikely because it usually
occurs where a protective film suffers local breakdown. I
The nature of the films that form is also very important. Under normal
atmospheric conditions, the more hygroscopic a film is, the more water it
will retain, and the more likely it is that SCC can initiate. I The orientation of
the samples can also affect the susceptibility to SCC. Samples cut transverse
to the rolling direction can perform differently from those cut longitudinally.
This is because the metallurgical properties (for example, strength) vary
with the rolling direction.
Although there is a large body of data published on the SCC resistance
of copper and its alloys, much of it has been generated under a wide range
of test conditions, stressing, and so on. This makes comparison of data from
different laboratories very difficult.
10.1 Ammonia
Copper and its alloys are susceptible to SCC in moist ammonia vapor and
also in the presence of certain ammonia compounds, although there are no
known cases of SCC of the single-phase copper-nickel al10ys.~Figure 10.1
shows a typical intergranular SCC crack in dezincification-resistant (DZR)
brass after 20 hours in Mattsson’s solution. (Tests in Mattsson’s solution
are discussed in Section 10.7.) In addition, amines can also cause SCC,
either by direct attack of the metal or by breakdown of the amines to release
’
ammonia. All of these are common in service environments. Amines can be
found as additives to boiler feed water or as additives to concrete. Ammonia
is present in some fertilizers and cleaners and can also be produced by the
decomposition of some organisms, both in the soil and in the sea.
Pure copper is immune to SCC by ammonia, but commercially pure
grades of copper, such as phosphorus-deoxidized (PDO) copper, with
Stress Corrosion Cracking 197
FIGURE 10.1 Typical stress corrosion crack in DZR brass exposed to Mattsson’s solution
for 20 hours.
maybe 0.04% phosphorus, and arsenical copper (0.3% As), are suscep-
tible.2,6
PDO copper hot water pipes in a district heating scheme failed by
SCC where they were laid in foamed concrete with an organic foaming
agent, which released ammoniacal compounds. This agent is no longer
used, but SCC failures still occur occasionally with copper pipes laid in
concrete floors. Ammonia is always detectable on the surface of the tubes.
The source could be latex cements used for flooring or the decomposition
of amines that were used in curing the concrete.2
The effects of alloying elements on SCC resistance were studied by
Thompson and Tracey using a season ~ r a c k e rThis
. ~ is a sealed chamber at
constant temperature (35°C) with a controlled atmosphere of 80% air, 16%
100,000
\
\
\
-+
-
I
w
*A. .
100
- - - _ I
I - - -
Zinc Content (%)
I-+ 34 MPa -M- 69 MPa - -A- - 103 MPa I

FIGURE 10.2 Time to SCC failure versus zinc contenL7
ammonia, and 4% water vapor. They alloyed copper with various concen-
trations of zinc, phosphorus, arsenic, antimony, silicon, nickel, aluminum,
and tin. Figures 10.2 and 10.3 show the time to failure vs. alloying content
for zinc and tin. Increasing zinc additions clearly reduce the SCC resis-
tance, particularly above 10 to 15% zinc. Shreir et al. commented that
failures of brass with up to 10 or 15% zinc are not common.’ However,
low-zinc brasses can suffer SCC in ammonia if the stresses are high enough.
Additions of tin increase the SCC resistance, which supports Campbell’s
comment that he had never seen an ammonia SCC failure of a gunmetal
(5% to 10% tin).2 Note that gunmetals are used as castings, often with thick
sections, so that stresses will normally be low.
Thompson and Tracey’s data for the other elements generally show a
decrease in the SCC resistance with a small alloying addition, followed by
a restoration of SCC resistance at higher concentration^.^ For details, the
reader is referred to Whittaker’s review because his graphs contain data not
included in Thompson and Tracey’s original paper.’
Brasses are commonly used alloys of copper, and Whittaker com-
mented that the addition of most third elements has no great beneficial
100,000
-
u)
10,000
-
.-
E
2
-.-m
U
0
L
E 1,000
F
100 I
1 2 3 4 5 6 7 a 9
Tin Content (%)
-69 MPa - + -138MPa

FIGURE 10.3 Time to SCC failure versus tin content?
effect.’ The results from various authors vary as to whether the effect is
beneficial or detrimental, and to what degree. The two exceptions are sil-
icon and nickel, both of which have been observed to increase the SCC
resistance of single-phase brasses.
More recently, the effects of arsenic and tin additions to 70/30 brass
were investigated in Mattsson’s solution at pH 7.3 (for details of Mattsson’s
solution, see Section 10.7).*Both elements increased the SCC resistance,
and this was ascribed to the effects of the arsenic and tin on the nature of
the protective film.
In some unpublished work at BNF Metals Technology Centre, the
SCC resistance of DZR brass was compared with that of a two-phase leaded
hot stamping brass in ammoniacal environment^.^ The purpose was to assess
the suitability of DZR brass for pipe fittings for use underground. The
test samples were forged, three-quarter-inch female couplings, into which
taper plugs were screwed to a torque of 38 Nm. Tests were conducted in a
simulated ground water containing from 1,000to 10,000mg/L chloride and
500 mg/L ammonium (NH4+) ions. Small, shallow cracks were seen in the
DZR brass after 28 days with 10,000mg/L chloride, but not at lower chloride
2.5 - t
4 Failed
h
E
vE 2-
5
Q
0
% 1.5-
2
0
E
;. 1 ~
f
0.5 -
07
0 5 10 15 20 3
ExposureTime (hours)
I + Duplex Brass H DZR Brass
FIGURE 10.4 Crack depth vs. exposure time in Mattson’s s o l ~ t i o n . ~
concentrations. With 780 mg/L chloride and 1,000 mg/L ammonium ions,
there was also some shallow cracking of the DZR brass. No cracking of the
duplex brass occurred in any environment, but dezincification occurred in
all of them.
Tests were also conducted in Mattsson’s solution at pH 7.5 for various
times. Figure 10.4 shows the crack depth versus exposure time. The results
show that cracking occurs quickly in DZR brass, but once the immediate
stress is relieved, cracking stops. In contrast, cracking was a little slower
to initiate in the duplex brass, but once started, the rate of propagation
increased dramatically. This was presumably due to the higher internal
stresses in the duplex alloy. These results suggest that the DZR brass would
be satisfactory for plumbing fittings for use underground.
All-beta brass is extremely prone to SCC in the presence of ammonia,
the cracking usually being transcrystalline. Two-phase alphdbeta brasses
are comparatively less susceptible to SCC by ammonia, and when they do
crack, it is usually transgranular through the beta phase.* This is why high-
tensile brasses should not be used for critical applications where ammonia
might be present. Although these alloys are intended to be two-phase, it is
possible to finish with an all-beta microstructure that cracks readily.2
TABLE 10.1 Time to failure for some aluminium bronzes

in ammonium sulfate and copper sulfate at various
DH values?
Time to failure (hours)
PH Cu/4AI Cu/8AI
4.0 NF(*) NF
5.8 NF NF
6.8 NF NF
7.3 NF NF
8.3 37.5 21
10.2 37.5 21
11.2 15.5 3.8
12.1 2.2 I .6
(A)
NF = no failure.
Campbell commented that tin bronzes (7% to 10% tin) occasionally

crack in the presence of ammonia, but gunmetals do not.' This probably
has to do with stresses. Cast gunmetals generally experience low stresses in
service due to their thickness, whereas wrought tin bronzes are sometimes
highly stressed.
The aluminum bronzes are less susceptible to SCC by ammonia than
brasses.2 This was also shown by the tests on binary Cu-A1 alloys in the
season cracker, where 5% aluminum alloys gave the same resistance to SCC
as ~ o p p e rLedheiser
.~ presented data on two aluminum bronzes that were
constantly strained (0.33 per second) in a solution of ammonium sulfate
([NH4]2S04) and copper sulfate (CuSO4) over a range of pH values.3 The
results (Table 10.1) show that cracking only occurred in solutions with an
alkaline pH and that cracking occurred more readily as the pH increased.
In addition, the Cu/4Al alloy was slightly more resistant to SCC than the
Cu/8Al alloy. However, this final finding was contradicted by other work,
in which Cu/l.5A1, Cu/4Al, and Cu/7Al alloys were tested and the Cu/4Al
alloy was the most susceptible to SCC."
Atmospheric exposure tests at two sites in the United States showed
the relative resistance to SCC of some common copper alloys and the high
resistance of nickel aluminum bronze (NAB; Table 10.2)." All the alloys
were cold rolled and exposed as U-bend specimens, so the stresses were
high.
Silicon bronze (Cu-3%Si-l %Mn) has been used to replace high-
tensile brass because of its high corrosion resistance and virtual immunity
to SCC by ammonia.' The other copper alloys that are essentially immune
to SCC by ammonia are the single-phase copper-nickel alloys (for example,
C70600 and C7 1500). Note that if the iron content exceeds 1 % and it is
TABLE 10.2 Time to failureof some common copper alloys

in atmospheric exposure tests' '
Time to failure
Temper
Alloy (% cold rolled) New Haven Brooklyn
70130 brass 50 35-47 days 0-23 days
Leaded duplex brass 50 51-136 days 70-104 days
Admiralty brass 40 5 1-95 days 4 1-70 days
Aluminium brass 40 2 2 1 4 9 5 days 3 11-362 days
Aluminium bronze 40 >8.5 years >8.5 years
(Cu/9.7A1/3.7Fe)
not all in solution, then 90/10 copper-nickel can be susceptible to SCC by

ammonia.12 If aluminum is added to increase the strength of these alloys,
it makes them liable to intergranular SCC in ammoniacal environments.2
The high-strength copper-nickel alloys are discussed in more detail in Sec-
tion 10.6.
During the 196Os, BNF conducted a series of tests on heat exchanger
alloys in seawater containing ammonia (added as ammonium chloride
[NH4C1]>.l3Samples of both aluminum brass and 90/10 copper-nickel
tubing were held in tension at various stresses, while seawater containing up
to 0.75% ammonium ions was circulated around them. The results (Table
10.3) show, not surprisingly, that aluminum brass is susceptible to SCC
under these conditions. The threshold stress with 0.75% ammonium ions
was -50% of the 0.1% proof stress, but this increased to >75% of the 0.1%
proof stress'when the ammonium ion content was reduced to 0.1%. As ex-
pected, 90/10 copper-nickel did not suffer any SCC, even at 95% of the 0.1%
proof stress. All of the cracking originated in a shielded area beneath a PVC
seal. This is not surprising as both Campbell and the author have reported
the concentration of ammonia in areas of localized corrosion on copper
l4
The nature of the cracking is affected by several parameters. Cracking
is mostly intergranular in straight alpha brasses, although it is transgranular
in aluminum brass and mixed mode in admiralty brass. Besides composi-
tion effects, the nature of the cracking can also be affected by the stacking
fault energy, as changed by cold work, as well as the environment, in terms
of the composition and P H . ~It is well documented that increasing the
pH of Mattsson's solution can change the cracking from intergranular to
transgranular for many copper alloys.'.2 The grain size can also be impor-
tant. Shreir et al. reported increased susceptibility to SCC of large-grained
alpha b r a ~ s e s . ~
TABLE 10.3 Results of SCC tests in seawater plus ammonia"

Alloy NH4+(%) Stress (Yoof 0.1% PS) Life (weeks)
Aluminium brass 0.75 95 4
95 4
85 8
85 >34
75 7
75 8
50 38
50 >95
30 >52
30 >95
0.1 95 32
95 >40
75 >40
75 >40
0 95 > 80
95 > 80
90/10 copper-nickel 0.75 95 >52
95 > 52
0.1 95 > 65
95 > 65
75 >40
75 >40
Clearly the higher the stress, the greater is the susceptibility to SCC.
The effect of cold work is somewhat varied. With some alloys, it can increase
susceptibility, whereas in others, it can reduce it.' This is because cold work
can produce more than one effect on the microstructure, depending on the
alloy. Whittaker described tests on a Cu-Ni-Sn age-hardening alloy.' Cold
working before aging the alloy significantly increased the SCC resistance
to ammonia vapor, as shown in Figure 10.5. Cold working after aging
produced only a small increase in SCC resistance.
The Chase Brass Company produced a large number of stress vs.
time to failure curves for many alloys using data from a season cracker.
Most of the alloys were tested in several different heat treatment conditions.
Whittaker reproduced much of these data in his review, but unfortunately,
many of the alloys tested are no longer in common use, and there are no
data for some recent copper alloys.' The data for some common alloys that
were included are summarized in Table 10.4; Figure 10.6 shows the data
for silicon bronze, which shows the increase in the threshold stress for SCC
obtained in the cold-worked condition.
TABLE 10.4 Some results from exposure tests by Chase Brass in a

season cracker’
Nominal Loss i n . Endurance
Alloy composition (wt%) Temper UTS (Yo) stress (MPa)
Commercial brass Cull OZn Hard 41 276
Annealed 80 55
Cartridge brass CuI30Zn Hard 99 14
Annealed 96 21
Muntz metal Cul4OZn Hard 98 14
Annealed 95 21
Admiralty brass CuI30Znll Sn Hard 98 14
Annealed 96 34
Copper-nickel CuI30Ni Hard NC(A’ -
Annealed NC -
Silicon bronze cu13si Hard 49 345
Annealed 76 172
(A1
NC = no cracking.
z
u .1l
-
.g 300 1
n
2 200 -“4
---_ -
-._
. . . . . . .A. . . . . . . . . . . . . . . . . . A7-b
100 . A-
07
I 75% Reduction 50% Reduction A 0% Reduction I

FIGURE 10.5 Applied stress vs. time to failure for Cu-Ni-Si alloy as a function of cold
working prior to age hardening.’
7nn J
- 300 -
.P
P
2 200 I
1
.....I
1
hm...
'-...ii.... ...................
'00 - I -.....
0 1
0 200 400 600 800 1000 1 00
Time to Failure (hours)
+ Cold Worked 1 Annealed
FIGURE 10.6 Stress vs. time to failure in a season cracker for 3% silicon bronze.'
It is well known that copper and ammonia readily form complexes.

In tests on the SCC resistance of brass, it was found that 3-6 g/L of
copper in ammonia greatly increased the corrosion rate and shortened the
time to f a i 1 ~ r e . lClearly
~ copper will frequently be present in service due
to dissolution of the parent metal, unless it has already formed a very
protective film.
The mechanism of SCC in ammoniacal environments has been linked
to the films that form. A film formation and rupture mechanism has often
been suggested for the propagation of SCC in copper alloys. Hoar and Roth-
well refined this model when they recalculated the potential pH diagram for
copper, ammonia, and water, taking the oxidation of ammonia to nitrite into
account.16They showed that at pH 7.3 and 11.3, there can either be a film of
cuprous oxide (Cu20) or a soluble cuproammonium complex, depending
on the potential. They suggested that active attack of the yielding metal at
the crack tip produces the soluble complex, whereas a protective cuprous
oxide film forms on the static metal at the crack sides. This mechanism
seems more plausible than a simple brittle film rupture theory.
One final feature of SCC in ammoniacal environments is the role of
localized pitting. Campbell observed SCC from the base of pits in aluminum
brass tubes carrying drinking water.2 Ammonia was readily detected in the
TABLE 10.5 Copper alloys tested by Goodman et al. in mercurous

nitrate”
Alloy Description UNS equivalent
CZ114 Manganese bronze C67600 (lower Mn)
DGS 1043 NAB C63200
CZ106 70130 brass C26000
CZ132 DZR brass No equivalent
cz121 Leaded duplex brass (free cutting) C38500
cz122 Leaded duplex brass (forging) C37700
corrosion product in the pits, whereas only very small amounts could be
detected in the bulk water. The author has seen a similar failure of admiralty
brass heat exchanger tubes, handling a brackish water. In this case, the
pitting was caused by carbon films (see Chapter 5 ) , but the leakage was due
to SCC from the base of the pits. As before, ammonia was detected in the
corrosion products in the pits, but not in the products on the open surfaces.
Clearly ammonia can concentrate in pits, presumably by a mechanism
involving the formation of cuproammonium complexes.
10.2 Mercury
Copper and its alloys are all susceptible to SCC by mercury and its com-
pounds. The attack is liquid metal embrittlement, but the threshold stress
for cracking, compared with the 0.2% proof stress, varies markedly from
alloy to alloy. Mercury is used as the aggressive species in the mercurous
nitrate test, which is discussed in more detail in Section 10.7. In this test,
samples are exposed in a solution containing 10 g/L mercurous nitrate and
10 mL/L nitric acid (HN03) at room temperature.
Mercury and its compounds are the only species that will cause SCC
of single-phase copper-nickel alloys when they are in the correct heat
treatment condition.
Goodman et al. investigated alternatives to the mercurous nitrate test
for the detection of excessive residual stresses.I7 As part of this work, they
looked at the applied stress vs. time to failure for some common copper
alloys in the mercurous nitrate test. Table 10.5 lists the alloys tested and
their approximate UNS equivalents, if they exist. The results, plotted in
Figure 10.7, show that all of the alloys have a fairly clear threshold stress,
which can be related to the mechanical properties of the alloy. Figures 10.8
and 10.9 show the stress ratio vs. the 0.2% proof stress and the tensile
strength, respectively. In these figures, the threshold stress was taken as the
450
-2 400
rm
m2 350
5
0
.--
300
4
250
200
+ DGS1043 CZ114 A CZ132 0 CZ121 A CZ122 .P 70/30 brass1
FIGURE 10.7 Applied stress vs. time to failure in a mercurous nitrate test for some stress-
relieved copper alloys.” (See color section.)
A = Annealed
2.5
70130 Brass (A)
DGS 1043 (SR)
CZ132 (SR)
0 50 100 150 200 250 300 350 400
0.2% Proof Stress (MPa)
FIGURE 10.8 Stress ratio vs. 0.2% proof stress for some copper alloys in the
mercurous nitrate test.17
CZ122 (A)
0.8 -
-.z
al
0.5 -
C
?tn
u)
0.4 .
'=
E
r
0
0.3 .
SR = Stress-Relieved
A Annealed
t 0.2
E
0.1 .
0,
FIGURE 10.9 Stress ratio vs. tensile strength for some copper alloys in the mercurous
nitrate test.17
stress at which cracking occurred after 0.5 hours because this is the standard
exposure time in the mercurous nitrate test for many copper alloys. Once
cracking occurs in a tensile sample in mercurous nitrate, failure follows
shortly after because the crack propagation rate is high. Figures 10.8 and
10.9 show a strong correlation between the mechanical properties and the
threshold stress ratio, except for CZ132 (DZR brass), which consistently
fails to follow the trend. DZR brass is a leaded, single-phase brass, and it
is not clear why it does not follow the trend when 70/30 brass (also single-
phase) does. The threshold stress is around the 0.2% proof stress for the
strongest alloys, whereas it is twice that or more for the very ductile alloys.
Goodman et al. also investigated the effect of lowering the mercury
concentration to determine whether similar results to the standard test could
be ~ b t a i n e d . The
' ~ results, given in Table 10.6, show that similar threshold
stresses could be achieved for some alloys, but with significantly longer
exposure times. The 70/30 brass showed unusual results in that the alloy
was more susceptible to cracking with 10 mg/L mercurous nitrate than with
100 mg/L. Goodman et al. were unable to explain these anomalous results.
However, they show that even small quantities of mercury compounds can
cause cracking of brasses in a relatively short time.
TABLE 10.6 Effect of reducing the mercury concentration

on the exposure time for cracking of some brassesI6
Exposure time (hours) Threshold stress (MPa)
HgN03 (mglL) &IN03 (WlL)
Alloy 0.01 0.1 10.0 0.01 0.1 10.0
CZ 121 (stress relieved) 5 - 0.5 360 - 340
70/30 brass (annealed) 2.5 4 0.5 246 240 246
CZ114 (annealed) - 22 0.5 - 355 355
10.3 Nitrites
Despite the fact that SCC of copper alloys by nitrites has been known for
well over 25 years, it is rarely mentioned in most corrosion textbooks.
This is possibly because the effect of nitrites was first discovered in the
laboratory, and it was thought to have little practical relevance. However,
this is far from the case.
Failures of copper heating pipes by SCC led to the DIN specification
for coatings and insulation forbidding the use of all materials containing
nitrites. Slusser et al. reported the failure of gas cylinder rupture discs in
90/10 brass (C22000) by SCC from the outside.18They showed that nitrites
readily caused SCC of this brass and several other copper alloys that were
being considered as replacements for C22000. Their work is described in
detail later.
A failure examined by the author a few years ago also serves as a
useful example of the role of nitrites. It concerns an electronics package in
the North Sea that was encased in coated carbon steel (CS) and was attached
to the flotation buoy by several nickel-aluminum bronze (NAB) bolts. To
avoid galvanic corrosion of the CS at defects in the coating, the operators
elected to coat the NAB bolts adjacent to the steel. This is in accord with best
practice to avoid galvanic corrosion. However, the bolts all failed within a
few months, the package sank, and retrieval was an expensive operation.
The bolts failed by SCC because the coating method chosen was heat shrink
tubing. It is common practice with heat shrink tubing to add nitrite inhibitors
to limit corrosion when it is used with CS. However, some of the nitrite
reacted with the copper in the NAB to form ammonia, and the combination
of nitrite and ammonia caused SCC of the NAB at relatively low stresses.
The solution was to use nitrite-free heat shrink tubing.
Whittaker reported some Russian work on simple binary alloys that
produced cracking in 10-30 min in a copper nitrite solution.' However,
it is the later work of Parkins and his coworkers that really defined the
susceptibility of copper and brasses to SCC in nitrite solutions.'"-'"
Ammonia does not readily crack copper, but nitrites do; this suggests
that nitrites are more potent in terms of SCC." Parkins and colleagues
used the slow strain rate technique and defined the potential ranges at
which SCC could occur and also the minimum concentration of nitrite
to cause Rebak et al. looked at the mechanism of SCC by
nitrites for a 37% zinc alpha b r a s 2 ' They investigated a range of solution
concentrations and pH values and concluded that SCC only occurred where
passivity breakdown had occurred.
When copper or brass is abraded into 1M sodium nitrite solutions,
ammonia can be produced by the reduction of nitrite.2' When Slusser
et al. examined failed brass rupture discs, they found ammonia present
after exposure to sodium nitrite." Copper cracked under these conditions,
whereas it did not crack in the presence of ammonia alone. Hence, the
ammonia was not the cause of the SCC. This does suggest that in service
failures where ammonia is detected, nitrite may also be involved. As nitrite
is readily oxidized to nitrate in the atmosphere, it is useful to check for all
three species when examining service failures.
Wu et al. examined cyclic loading of 70/30 brass in sodium nitrite
solutions. If the film cleavage mechanism is correct, then the crack growth
rate should be dependent on the frequency of cyclic loading.23 They found
no such correlation and concluded that the film cleavage model did not apply
to 70/30 brass in nitrite solutions. It seems more likely that the mechanism
of SCC involves the formation of cupronitrite complexes at the crack tip,
similar to the mechanism proposed for SCC of copper alloys by ammonia.''
Slusser et al. tested gas cylinder rupture discs at a pressure of 2,000
psi (138 bar)." Every day, the disc was depressurized, sprayed with test
solution, and then repressurized. Most tests were conducted for 25 cycles,
or until failure occurred. Their results are summarized in Table 10.7, which
shows that nitrites readily caused cracking of 90/10 brass and copper, but
the susceptibility decreased as the temperature increased. They selected
either beryllium copper or nickel 201 as their first choices for replacement
of 9040 brass for the rupture discs.
Goodman et al. investigated alternatives to the mercurous nitrate
test." They tested a solution of 1M sodium nitrite in 0.2% nitric acid
and the same solution with an addition of 5 g/L copper nitrate. This addi-
tion was to form the cupronitrite complex more readily and thus promote
SCC more rapidly, as was observed with copper and ammonia, described
in Section 10.1. The addition of copper nitrate shortened the time to failure
TABLE 10.7 Results of SCC tests on gas cylinder rupture discs at 2,000 psi
(138 bar)'*
Result (temperature "C)
90110 brass Copper Cu-Ni-Si Cu-Be
Test solution (C22000) (C10100) (C72500) (C17200)
1% NaN02 F4C(A);FlOC; F6C; F8C; NF22C NF22C
NF24CCB' F11C
1% NaN02 (pH 8-9) F4C; F8C; F4C; F8C; NF22C NF22C
FlOC; NFI5C
F11C; F9C;
F18C
1% NaN02 +1% NF34C - - -
Na2P04
1% NaN02 1% + NF39C - - -
NaSi03
1% NaN02 1% + F16C; F8C; - - -
NaS04 NF15C
+
I % NaN02 1% NaCl NF98C - - -
+
Detergent NaN03 NF39C - - -
Spray disinfectant NF39C - - -
(contains NaN02)
(*)F = failure.
( B ) =~ no~ failure.
under constant load compared with the copper-free solution for alloys
CZ114, CZ121, and 70/30 brass. The results for CZ122 are shown in
Figure 10.10, where it can be seen that the addition of copper nitrate totally
changed the shape of the stress vs. time to failure curve. However, even
with a copper nitrate addition, no cracking was observed with DGS 1043
(NAB) after 139 hours at a stress of 450 MPa (1.4 x 0.2% proof stress).
This is surprising in view of the failure of NAB bolts described previously.
Despite the fact that CZ122 readily cracked in nitrite solutions, CZ121 only
cracked after 54.7 hours at a stress of 340 MPa (1.3 x 0.2% proof stress).
However, the CZ122 was exposed at much higher stresses before rapid
cracking occurred (-1.8 x 0.2% proof stress). This study used commer-
cially produced alloys and did not include any effects of grain size, phase
distribution, impurities, and so on, on SCC.
10.4 Sulfur Dioxide
Whittaker reported that the Russians were the first to demonstrate that
some copper alloys can suffer SCC in moist atmospheres containing sulfur
Solution 1 : 1 M sodium nitrite in 0.2% nitric acid

Solution 2: 1M sodium nitrite + 5glL cupric nitrite in 0.2% nitric acid
270
0 2 4 6 8 10 12 14

A Solution 1 Solution 2
FIGURE 10.10 Applied stress vs. time to failure of annealed CZ122 in nitrate solution^.'^
dioxide.' The attack is much slower than that in the presence of ammonia,
and there is no hard quantifiable data on threshold stresses. Whittaker pre-
sented previously unpublished data on U-bends of cold-rolled brass strip.'
The results showed that cracking only occurred when the SO2 content in
the test atmosphere was between 0.05% and 0.5%. At lower SO:! concen-
trations, there was no attack, and at higher SO2 concentrations, the attack
became general corrosion everywhere.
Campbell described the work of Bobylev and Manzienko, who tested
some copper-zinc-silicon alloys in moist sulfur They obtained
similar results in the laboratory to atmospheric exposure tests, but the
laboratory tests were of long duration. Unfortunately, their results do not
properly show the effect of silicon because their control alloys, without
silicon, had a very different copper content.
Slusser et al. tested a number of alloys as gas cylinder rupture discs
(the tests were described in the previous section). One of the environments
that was sprayed onto the discs was 1% sodium sulfite (Na2SO3). This
produced cracking of 90AO brass (C22000) at I'C, but not at 22°C or 38°C.
No cracking occurred on copper (C10100), beryllium-copper (C 17200), or
copper-nickel-tin (C72500), although these alloys were only tested at 25°C.
The production of SO2 atmospheres in the laboratory with a known

SO:! concentration is difficult because it is not possible to calculate this.
Cotton, in a private communication to BNF, reported that his laboratory had
determined a method by trial and error.25They found that a concentration
of 0.5% was most useful for determining susceptibility to SCC, but it was
possible to produce a range of SO:! concentrations in an 8-in desiccator
(volume of 6 L), as shown:
SO2 (%) Na2S03(g) + 2%v/v H2S04 (mL)

0.1 0.07 + 10
0.5 0.35 + 10
I .o 0.7 + 10
Cotton also pointed out that the chemicals should be mixed after the desic-
cator is sealed.
10.5 Other Environments

There have been many studies of SCC of copper alloys, mostly brasses, in
a range of environments. One of the most common is chloride, and copper
alloys are generally resistant to SCC in chloride solutions. The exception
is all-beta brass, which Campbell described as being very susceptible to
chloride SCC.' Whittaker reported the results of an extended U S . Navy test
of copper alloys in both full immersion seawater and a marine atmosphere.'
The only alloys to suffer SCC were a cast beta brass and a nickel silver
(Cu-27Zn- 18Ni).
Attempts to modify the composition of manganese bronze to increase
its resistance to chloride SCC were reviewed by Whittaker.' The chloride
SCC resistance of Cu-33Zn-4AI was greatly improved by additions of
zirconium or titanium. A cast manganese bronze (BS 1400-HTB3) resisted
chloride SCC in seawater when 1.0% to 2.5% nickel was added to the alloy.
Without the nickel addition, both samples failed in under an hour.
Sat0 and Nagata reported the failure of aluminum brass tubes in pure,
high-temperature steam.26 They showed that at steam temperatures from
120 to 21O"C, aluminum brass suffered intergranular corrosion that devel-
oped into cracking under an applied stress. They found that cracking was
worse with a commercial aluminum brass compared with a simple labora-
tory Cu-Zn-A1 ternary alloy. They ascribed the difference to the presence
of arsenic in the commercial alloy (added to prevent dezincification). They
also found that the coarser the grain size, the worse was the cracking. An
I I
E 100 1 I
t c(-p"--.' HCUO,
>
E 00
- -100
v
A OA
.-
-
L
0
2 -300
A OA
cu
-500
."" I
4 5 6 7 8 9 10 11 12 13 14 5
Solution pH
0 Hydroxide 0 Hydroxide
A Formate A Formate w Acetate 0 Acetate
FIGURE 10.11 Potential pH diagram showing the domain of SCC failure for 70/30brass
in various chemical solutions.27
alloy of 70/30 brass with 0.03% phosphorus was totally resistant to cracking
in this environment.
Cracking of single-phase aluminum bronze (C6 1400) has also oc-
curred in high-pressure steam service." The addition of 0.25% tin to
this alloy completely eliminated the susceptibility to SCC in this envi-
ronment. The modified alloy is designated C61300. Campbell commented
that C61400 is also susceptible to SCC in hot brines and recommended the
use of the tin-modified alloy for this service."
Campbell reported tests on 70/30 brass in both citrate and tartrate
solutions, both of which produced cracking.2 Citrate produced the more
rapid cracking. Parkins and Holroyd investigated 70/30 brass at various po-
tentials in acetate, formate, and tartrate solutions.27They demonstrated that
intergranular cracking could occur in all three solutions when the potential
and pH were such that cuprous oxide (Cu20) was the stable phase. At more
positive potentials, the cracking was transgranular, with mixed-mode crack-
ing at the cuprous oxide domain boundary. The results are summarized in
Figure 10.11. At very positive potentials, the corrosion was either pitting or
general attack. Parkins and Holroyd used electrochemical techniques to de-
termine the potentials where cracking was likely to These potentials
-
0.1
0'12
0.08
5
-
4-
m
0
0.06
0
.-c
0
0
2 0.04
U
.....................................
.,
0.02
0
0 50 100 150 200 250 300 350 400
1- P<O.OOl% -m- P=O.OI% -A- P=0.35% 1

FIGURE 10.12 Arsenic content vs. time to failure of aluminum brass in a seawater-citrate
solution.29
were at the boundary of active and passive regions, where the scan speed
of a potentiodynamic sweep caused a change in the threshold potential for
the change from active to passive. It is presumed that the mechanism in all
three of these solutions involves the formation of complexes with copper,
in a similar manner to that of copper and ammonia, as described in Sec-
tion 10.1.
In the early 1980s, concern was expressed about the differing ranges
of arsenic permitted in arsenical brasses by various standards and its possi-
ble effect on susceptibility to SCC. Most national standards permit 0.02%
to 0.06% arsenic, but the ASTM standard B111" permits up to 0.1%,
while the German DIN standard 1766029restricts the maximum arsenic to
0.035%. Campbell reviewed the data on arsenic and SCC resistance and
concluded that there was no significant difference between the SCC suscep-
tibility with 0.03% arsenic compared with 0.06% at typical phosphorous
levels of 0.01%.30Figure 10.12 shows the data of Torchio for aluminum
brass, obtained in a solution of sodium chloride (NaCI) and sodium citrate
(Ca6H5Na3072H20), with cuprous chloride (CuCl) additiom3'
Campbell also reviewed the effect of arsenic content on intergranular
corrosion.30 The data show that phosphorus has five times the effect that
arsenic does on intergranular corrosion, and it is much more important to

keep phosphorus levels low, rather than arsenic levels. In the arsenic range
of commercial interest, 0.03% to 0.1 %, no significant intergranular corro-
sion was observed on stressed samples exposed to ammoniacal solutions
containing copper.
Veld reported the failure of 90/10 copper-nickel overhead condenser
tubes in a steam cracker.32The environment contained both hydrogen sul-
fide (H2S) (1-2 wt%) and ammonia (1-2 wt%) at 120°C maximum and
had a pH of 9-9.5. Failure of the cold-worked tubes was by intergranular
corrosion after just a few months of operation. The authors argued that
solution-annealed tubes would have been more resistant to H2S under these
conditions. An adjacent condenser had been in service without problems
for 20 years, but this one was tubed with aluminum brass.
In alkaline solutions, from lo-' to lOP5Nsodium hydroxide (NaOH),
Whittaker reported cracking of a beta brass at all concentrations and crack-
ing of a duplex brass at 10-3N and lOP5N concentrations of sodium
hydroxide.' No cracking was seen with an all-alpha brass after 50 days
at all concentrations. Parkins and Holroyd carried out SCC tests of a 70/30
brass in caustic solutions at pH 12 and 13 using the slow strain rate test.'7
The results were the same as those found for acetate solutions, reported
previously. Where cuprous oxide was the stable phase, the cracking was in-
tergranular, and at more positive potentials, the cracking was transgranular.
The results are summarized in Figure 10.1 1.
Whittaker also presented some results showing that a range of brasses
can crack when exposed over nitric acid.' Alloys with 70% copper or less
were susceptible to SCC.
10.6 High-Strength Copper-Nickel Alloys
The strength of single-phase copper-nickel alloys is relatively low, and

there are two main methods of alloying to increase the strength. The first
is to alloy with tin, which increases the mechanical properties by a spin-
odal mechanism. The alternative is to add aluminum, which increases the
strength by the formation of a gamma prime (Ni3Al) precipitate. Both types
of alloy have been used where high strength and corrosion resistance are
required, for example, in marine fasteners.
Campbell commented that high-strength copper-nickel alloys contain-
ing aluminum have been observed to suffer SCC in ammonia environments,
but he gave no details.' In the early 1990s,there were a series of failures of a
copper-nickel-aluminum alloy, mostly as fasteners, on a number of offshore
TABLE 10.8 Summary of the test environments and

results from SSRT tests for C72420, high-strength
copper-nickel alloy”
Environment Result
Amine paints and lubricants Cracking
Mercury Cracking
Ammonia Cracking
Natural seawater (no CP), T 2 50°C Cracking
MoS2 - containing lubricants Cracking
(except those with a heavy paraffin base)
Seawater with CP (all temperatures) No cracking
Natural seawater (no CP), T < 50°C No cracking
platforms. These were all in the same alloy, C72420. Eight of these were
described by Andersen et al., who noted that SCC had been caused by a
variety of species.33 In some cases, the aggressive element was not identi-
fied. Those that were included mercury (from a cap seal), hydrogen sulfide
(from the decomposition of a molybdenum disulfide-containing lubricant),
and amines (from paints and greases). The various operators were con-
cerned with the wide variety of chemicals that could cause SCC of this
alloy, all of which are commonly found on offshore platforms.
A series of slow strain rate tests were undertaken in a wide range of
chemicals found on offshore platforms.33The environments and the results
are summarized in Table 10.8. One surprising result was that some of the
greases containing bentone or lithium compounds were the most aggressive
chemicals causing SCC of C72420. As the production of bentone involves a
quaternary amine compound, which can decompose to form ammonia, this
is not surprising. Cracking in the presence of lithium compounds has not
previously been reported. This paper highlights the need to check carefully
all the chemicals that will come into contact with critical components to
ensure that none of the ingredients can cause SCC.
Andersen et al. suggested that it might be the presence of grain
boundary precipitates that was making the C72420 alloy so susceptible
to SCC.33 This was investigated further by Tuck, who studied the effect
of various environments on a copper-nickel-tin alloy (C72900), a copper-
nickel-aluminum alloy (Werk. no. 2.1504), and a copper-nickel-aluminum-
manganese-niobium alloy (C72420).’4
In hydrogen charging tests at - 1 .O V SCE (saturated calomel elec-
trode), at room temperature, none of these alloys showed any signs of
embrittlement. However, in NACE TM017735 tests (for sulfide SCC), the
TABLE 10.9 Details of some common tests used to assess the SCC resistance
of copper and its alloys
Duration
Test Standard Solution Type(A) (hours)
Mercurous nitrate ASTM B154 10g/L HgN03 I 0.5'"
10ml/L nitric acid
Mattsson ASTM G37 55.5g/L (NH4)2S04 I 24
12.6gL CuS04.5H20
2N NaOH")
Eichorn - 12.5 wt% ammonia solution A 48
Aebi I S 0 6957 2M NH4CI NaOH(D) A 24
(A)
A = atmospheric; I = immersion.
(B)
Twenty-four hours for Cu-Ni alloys.
(C)
Sufficient to adjust to pH 7.5.
ID)
Sufficient to adjust to pH 10.
copper-nickel-tin alloy was brittle, and the copper-nickel-aluminum alloy

showed secondary cracking. Only Alloy C72420, showed no signs of SCC
in this environment. Similar results for all three alloys were found in 10%
ammonium chloride at 82°C.
In a series of tests on Alloy C72420, Tuck found that overaging,
which increases the amount of gamma prime precipitate, also increases the
susceptibility to SCC.34Heat treatments that increased the grain size also
increased the amount of gamma prime on grain boundaries and increased
susceptibility to SCC. Finally, the iron content was also critical and, where
it was not all in solution, the susceptibility to SCC increased. Tuck sug-
gested heat treatment to keep the grain size small and the alloy underaged.
He also suggested limiting the maximum iron content to 0.85% to increase
the SCC resistance (the UNS composition permits up to 1.2% iron).
10.7 Test Methods and Testing

A number of methods are commonly used in the laboratory to assess the
susceptibility of copper alloys to SCC. The main ones are summarized in
Table 10.9, where it can be seen that apart from the mercurous nitrate test,
the tests are based on ammonia. However, these tests do not all have the
same purpose, and it is important to be aware of their limitations before
doing any testing.
The mercurous nitrate test (ASTM B 154") is designed to determine
whether there are excessive residual stresses in a copper alloy. As such, it
is a production test, with the main requirements that it be simple and rapid.
For all copper alloys, except the copper-nickel alloys, a 30-min exposure is
sufficient, and cracking can be seen with the naked eye. The copper-nickel
alloys are not so susceptible to SCC by mercury, and a 24-hour exposure
is required. This is the one big advantage of the mercurous nitrate test; it
is the only medium that will produce SCC of copper-nickel alloys. The
disadvantages of the test are that the chemicals are hazardous and disposal
of spent solution is difficult and expensive. However, it remains the only
test that can be used with copper-nickel alloys.
Because of its purpose, the threshold stress for cracking is high for all
alloys. l7 Goodman et al.’s attempts to find a replacement for the mercurous
nitrate test using one based on nitrites (see Section 10.3) were not successful
because there was no solution that would work equally well on all, or even
most, copper alloy^.'^ The exposure time required varied with the alloy,
and a low-power microscope was needed to see the cracking.
Whittaker commented on the high stresses needed for an alloy to fail
the mercurous nitrate test.’ While an alloy that fails the test clearly should
not go into service, there is no guarantee that an alloy that passes the test
will not fail by SCC in service, particularly if ammonia is present. The same
comment is made by Shreir et a1.6
Mattsson et al. compared the results of tests in mercurous nitrate,
Mattsson’s solution, and the Eichorn test with long-term atmospheric ex-
posure tests.37 The samples were deep-drawn duplex brass cups, and some
were stress relieved at 250°C for 2 hours. These cups passed the mercurous
nitrate test, but many failed within a year during atmospheric exposure.
The Eichorn test cracked all the samples, irrespective of stress re-
lief heat treatment, although cups stress relieved at 275°C or greater did
not crack in the atmospheric exposure test. Clearly the Eichorn test is ex-
ceedingly aggressive. The results for the tests in Mattsson’s solution, at
pH 7.5, showed a better correlation with the atmospheric exposure tests.
However, some of the samples stress relieved at 275°C and higher tem-
peratures cracked in Mattsson’s solution but did not crack in the 1-year
atmospheric exposure test. Clearly Mattsson’s solution is somewhat more
aggressive than atmospheric exposure.
The Aebi test was designed to be less aggressive than the Eichorn test
and is controllable by adjusting the pH with sodium hydroxide. Mattsson
et al. evaluated the Aebi test as the basis for a standard test to assess the
SCC resistance of copper alloys.38 They compared the results with those
from 1-year atmospheric exposure tests at different locations in Sweden.
The test worked well with brasses, but more work would be required to
demonstrate its suitability with other copper alloys, for example, bronzes.
At pH 9.5, there was a good correlation between most alloys in the Aebi
test and the atmospheric exposure tests. The test was more aggressive at
pH 10.5, although it was not as severe as the Eichorn test. Premoistening
the samples increased the severity of the test to close to that of the Eichorn
test. Testing at pH 10 gave similar results to the mercurous nitrate test when
applied to brasses.
Mattsson not only showed the importance of pH in this test, but also
showed the need for good temperature control (&l"C). Although some
samples cracked very quickly in the Aebi test, 24 hours was felt to be the
optimum exposure time to obtain results that were reproducible. The test
has since been adopted as an International Organization for Standardization
method as IS0 6957.'9
Thompson and Tracey's season cracker was described in Section 10.1,
but this test requires complex and expensive equipment, and the testing time
is much longer than in the Aebi test. However, the results from the season
cracker do reflect severe atmospheric exposure fairly well.
It has been pointed out by several workers that the addition of copper
ions to a test solution that can form complexes with copper (for example,
ammonia) makes the test more aggressive.'-lSThis can obviously shorten
testing times when trying to reproduce a service failure.
The method for obtaining a sulfur dioxide atmosphere to test for SCC
was described in Section 10.4.
Parkins reviewed the important factors to be taken into account when
trying to simulate, in the laboratory, service environments that have caused,
or may cause, SCC.40Not only is the solution composition important, but
the potential and the pH are also important. He also pointed out the use-
fulness of Pourbaix diagrams and electrochemical testing for identifying
regions where SCC may occur. The choice of material for testing is very
important because small changes in composition can strongly affect SCC
susceptibility in some environments. The surface finish can also be impor-
tant in some circumstances, as can the sample orientation with respect to
the rolling direction.'
Whittaker pointed out the need to include unstressed samples as con-
trols. Some alloys suffer intergranular corrosion without undergoing SCC. '
He also said that, in atmospheric exposure tests over a solution (for example,
Aebi), the height of the samples above the liquid affects the aggressivity.
The closer the samples are to the liquid surface, the more aggressive is the
vapor and the greater the risk of SCC. Hence, it is important to maintain
a constant sample height over the solution in laboratory tests to get good
reproducibility.
TABLE 10.10 A guide to the susceptibility of copper alloys to SCC(A’

Environment
Alloy Ammonia Chloride Mercury Nitrite SO2
Copper R R S S
90/10 brass M R S S S
70/30 brass S R S S S
60/40 brass S R S S
High-tensile brass S M S
Beta brass S S S
Phosphor bronze R R S
Silicon bronze R R S
Aluminium bronze M R S M
High-strength Cu-Ni M R S
90/10 Cu-Ni R R S R
70/30 Cu-Ni R R S R
(A)
S = high risk of SCC; R = high resistance to SCC; M = may suffer SCC. Empty cells are
unknown.
10.8 Prevention of Stress Corrosion Cracking

Where it is known in advance that an environment may cause SCC, it is
advisable to select a material that is resistant in that environment. To this
end, it is useful to know the relative resistance of copper alloys to SCC
in various environments. Table 10.10 gives some guidelines on the relative
resistance of some common alloys in some common environments that can
cause SCC of copper alloys. Remember that it is not only composition that
affects the susceptibility to SCC. The applied stress (as a function of the
tensile strength) and the degree of cold work also affect the susceptibility to
SCC. Generally, the higher the applied stress, the greater is the susceptibility
to SCC. As mentioned previously, cold work can increase SSC susceptibility
for some alloys and can decrease it for others. Note that there are significant
blanks in Table 10.10, where data are not available. Do not assume that a
blank means that an alloy is suitable.
An example of changing alloy is the use of manganese bronze (high-
tensile brass) for wall ties, particularly for concrete panels in large buildings.
The alloy was selected to replace CS, which had a relatively short life due
to corrosion. Unfortunately, there were numerous failures of manganese
bronze ties due to SCC, either by ammonia or chloride^.^' As mentioned
in a previous section, manganese bronze is very susceptible to SCC when
the beta phase content is high. This can be caused by poor alloying control
and/or heat treatment. The solution was to use silicon bronze or aluminum
bronze. Both of these have given excellent service.
However, there are sometimes reasons why changing material is not

possible, and then methods of minimizing risk must be adopted. One way is
to reduce the stress, and Mattsson et al. showed that annealing at 300°C for
2 hours greatly reduces the internal stresses.37The atmospheric exposure
test results showed much better performance for stress-relieved samples
than for as-drawn samples.
Cathodic protection, either impressed current or by means of a sacri-
ficial anode, can depress the potential sufficiently that SCC cannot occur.
Separating the metal from the environment is an alternative preventative
method that may be attempted in a number of ways.
One way is to use inhibitors, but it is important to test these to ensure
that they are satisfactory for any specific environment. Whittaker presented
results from a brass exposed to a 0.5% moist sulfur dioxide atmosphere.’
The untreated material cracked, but material treated with benzotriazole
remained shiny and unattacked. However, benzotriazole had no effect on
samples exposed to moist ammonia, in which both treated and untreated
samples cracked. Whittaker suggested that this was because the inhibitor
was soluble in ammonia.
The other way of separating the metal and the environment is to
use a coating. Not only must the coating be free of defects, where the
environment can penetrate to the metal, but the coating must also resist
mechanical cracking on stressing of the base metal.
Laub investigated the use of electroplated coatings on a Cd35Zn
brass.42 His testing included both stressing of the substrate and the effect
of heat treatment. The best coatings were nickel and nickel plus silver
(6 pm of each), heat treated for 1 hour at 350°C after coating.
Laub also investigated the use of absorbents as a method of preventing
SCC.43Silica gel was found to be very useful in the presence of ammonia,
and it was recommended as a packing agent for use with brass. Coatings
of metallic sulfides, water-repellent layers, and air-drying paints showed
little success in preventing SCC?4 Campbell reported that coating with an
air-drying acrylic-ester lacquer containing phenylthiourea prevented SCC
of brass in both ammonia and sulfur dioxide atmospheres.2
References
1. J.A. Whittaker, “A Survey on the Stress Corrosion of Copper Based Alloys,”
International Copper Research Association Report, 1966.
2. H.S. Campbel1,J. Inst. Met. 101 (1973): p. 232.
3. H. Ledheiser, “Stress Corrosion Cracking of Copper Alloys,” International Copper
Research Association Report, 1985.
4. A.J. McEvily Jr., Atlas of Stress Corrosion and Corrosion Fatigue Curves
(Materials Park, OH: ASM International, 1990).
5. L.L. Shreir, R.A. Jarman, G.T. Burstein, eds., Corrosion, 3rd ed. (Oxford, U.K.:
Butterworth Heinemann, 1994), p. 150.
6. L.L. Shreir, R.A. Jarman, G.T. Burstein, eds., Corrosion, 3rd ed. (Oxford, U.K.:
Butterworth Heinemann, 1994), p. 4:63.
7. D.H. Thompson, A.W. Tracey, Trans. Am. Inst. Min. Met. Eng. 185 ( 1 949): p. 100.
8. B.C. Syrett, R.N. Parkins, Corros. Sci. 10 (1 970): p. 197.
9. R. Francis, BNF Metals Technology Centre Research Report, A1969, June 1980.
10. A.W. Blackwood, N.S. Stoloff, Trans. ASM 62 (1969): p. 677.
1 1. H.S. Campbell, Aluminium Bronze Alloys Corrosion Resistance Guide,
Publication 80 (Potters Bar, U.K.: Copper Development Association, I98 I).
12. J.M. Popplewell, Corros. Sci. 13 (1973): p. 593
13. H.S. Campbell, BNF Metals Research Association Report, A1577, February 1966.
15. E.N. Pugh, W.G. Montague, A.R.C. Westwood, Truns. ASM 58 (1965): p. 665.
16. T.P. Hoar, G.P. Rothwell, Electrochem. Acta 15 (1970): p. 1037.
17. P.D. Goodman, V.F. Lucey, J.J.B. Ward, BNF Metals Technology Centre Research
Report, R503/4, August 1985.
18. J.W. Slusser, S.W. Dean, D.M. Drummer, CORROSION186, paper no. 330
19. J. Yu, J.H. Holroyd, R.N. Parkins, “Application of Slow Strain Rate Tests to
Defining the Stress Corrosion Crack Initiation in 70/30 Brass,” ASTM STP 82 I
(l984), p. 288.
20. L.A. Benjamin, D. Hardie, R.N. Parkins, Brit. Corros. J. 23 (1988): p. 89.
21. R.B. Rebak, R.M. Carronza, J.R. Galvele, Corros. Sci. 28 (1988): p. 1089.
22. R.C. Newman, G.T. Burstein, J. Elecrrochem. Soc. 127 (1980): p. 2527.
23. D. Wu, H.S. Ahluwalia, H. Cai, J.T. Evans, R.N. Parkins, Corros. Sci. 32 (1991):
p. 769.
24. A.V. Bobylev, Y.P. Manzienko, Sou J. Non-Ferrous Met. 7 (1966): p. 97.
25. J.B. Cotton, IMI, correspondence to BNF Metals Research Association, July
1967.
26. S. Sato, K. Nagata, Sumitomo Light Met. Tech. Rep. 15 (1974): p. 73.
27. R.N. Parkins, N.J.H. Holroyd, Corrosion 38 (1982): p. 245.
28. ASTM B 111, “Standard Specification for Copper and Copper-Alloy Seamless
Condenser Tubes and Ferrule Stock” (West Conshohocken, PA: ASTM).
29. DIN 17660 (Brussels, Belgium: DIN).

30. H.S. Campbell, BNF Metals Technology Centre Miscellaneous Publication,
MP600, December 1982.
31. S. Torchio, Corros. Sci. 21 (1981): pp. 59,425.
32. A. Veld, MP 45,4 (2006): p. 52.
33. 0. Andersen, M.W. Joosten, J. Murali, D.E. Milliams, CORROSION/96, paper
no. 78 (Houston, TX: NACE International, 1998).
34. C. Tuck, CORROSION 2005, paper no. 462 (Houston, TX: NACE International,
2005).
35. NACE TM0177, “Laboratory Testing of Metals for Resistance to Sulfide Stress
Cracking in Hydrogen Sulfide (HzS) Environments” (Houston, TX: NACE
International).
36. ASTM B 154, “Standard Test Method for Mercurous Nitrate Test for Copper
Alloys” (West Conshohocken, PA: ASTM).
37. E. Mattsson, S. Lindgren, S. Rask, G. Wennstrom, “Stress Corrosion in Brass,” in
Current Corrosion Research in Scandinavia (Helsinki, Finland: Scandinavian
Corrosion Congress, 1964).
38. E. Mattsson, R. Holm, L. Hassel, “Ammonia Test for Stress Corrosion Resistance
of Copper Alloys,” ASTM STP 970 (1988): p. 152.
39. IS0 6957, “Ammonia Test for Stress Corrosion Resistance” (Geneva, Switzerland:
ISO).
40. R.N. Parkins, “Synthetic Solutions and Environment Sensitive Fracture,” ASTM
STP 970 (1 988): p. 132.
4 1. Guidance Note on the Security of the Outer Leuf of Large Concrete PuneIs of
Sandwich Construction (London: Institution of Structural Engineers, 1989).
42. H. Laub, Metalloberflache 20 (1966): pp. 413,453.
43. H. Laub, Metalloberjache 21 (1967): p. 587
44. H. Laub, Metalloberjache 20 (1966): p. 493
CHAPTER 11
Corrosion Fatigue
Fatigue occurs when a material is subject to a repetitive cyclic stress, usually

at stresses below that at which failure would occur under constant load. As
the peak applied stress is reduced, the number of cycles to failure increases
until a threshold stress is reached at which failure does not occur, even after
a large number of cycles (>lo8),as shown schematically in Figure 11.1.
When the material is also exposed to a corrosive environment, failure often
occurs after fewer cycles than in air, and this is termed corrosion .fatigue.
With many materials, there is no clear threshold stress for corrosion fatigue
(Figure 1 1.1).
Corrosion fatigue is not just a function of the peak stress; it is also
affected by the stress range, which is often defined by the R value, where
minimum stress
R= (1 1.1)
maximum stress
Corrosion fatigue is also influenced by the frequency of the stress cycle and
the amount of strain induced under the stress.
Corrosion fatigue has several phases, the first of which is initiation,
followed by propagation. Propagation may proceed differently as the crack
develops. The stresdnumber of cycles to failure curves, or S/N curves
(for example, Figure 11. l), show the total life to failure, usually for a
plain sample. Other tests take precracked samples and look at the effect
of various factors on crack propagation. Unfortunately, there is no simple
relationship between S/N data and crack propagation data for many practical
applications.'
Corrosion fatigue cracks propagate by the action of local strains de-
forming or damaging the protective film, followed by corrosion of the
underlying metal. Hence, the environment does not need to be particularly
225
* Fatigue
I...................................
Fatigue
\
\
.................................................................................................................
I
Corrosion I -
1 -
1 - - ,
Fatigue
Cycles (log scale)
FIGURE 11.1 Schematic comparing the typical stress vs. number of cycles for fatigue
and corrosion fatigue.
corrosive for corrosion fatigue to occur.' The mechanisms of corrosion

fatigue are discussed in detail by Scott, but corrosion fatigue cracks are
often blunt ended compared with stress corrosion cracks.' Propagation is
by a series of crack growth and arrest steps, which produce a classic series
of curves on the fracture face, sometimes called bench marks. Figure 1 1.2
shows a fatigue failure of a pump shaft with the curves radiating from the
left-hand corner of the groove in the half on the left-hand side, where the
crack initiated.
Scott commented that surprisingly little has been published on corro-
sion fatigue of copper alloys.' The following sections group the available
data by alloy type.
11.1 Brasses
Jaske et al. reported the results of corrosion fatigue tests of manganese

bronze in seawater.* The environment had little effect on fatigue crack
propagation, although the frequencies were quite high.
Campbell carried out corrosion fatigue tests on aluminum brass in
both seawater and seawater containing 0.75% ammonia (NH3) (added as
Corrosion Fatigue 227
FIGURE 11.2 Typical appearance of a fatigue fracture, including beach marks. (See color
section.)
ammonium chloride [NH4Cl]), using reverse bend tests at low frequency

(<1Hz).3 The results, in Figures 1 1.3 and 1 1.4, show the life as a function
of the mean stress and the alternating stress. The corrosion fatigue limit
in seawater was f 8 5 MPa for zero mean stress, falling to f 5 5 MPa with
a mean stress of about 100 MPa. In seawater with 0.75% ammonia, the
corrosion fatigue limit decreased, being f 7 0 MPa with zero mean load.
This decreased sharply to about 4 ~ 3 0MPa for mean stresses >40 MPa, but
this, then, remained constant up to the 0.1% proof stress.
Work described in the previous chapter showed that aluminum brass
is susceptible to stress corrosion cracking (SCC) in seawater containing
ammonia, so the reduction in corrosion fatigue strength in the presence
of ammonia is not surprising. There appears to be a fatigue limit in the
presence of ammonia of about 30 MPa, irrespective of mean stress.
11.2 Copper-Nickels
Spiedel suggested that in seawater and aerated salt solutions, the copper-
nickel alloys have a corrosion fatigue strength in the range 75-125 MPa.4
180
Pass = z 50 Million cycles

160 +.....
-.._0.1% proof
140 -
......stress
*%.
........
-2 120~ -.. ....%.
I,
" .......
....
.....
?!
a ......
cn 800
.
"
n %. a w
.-
L
0 0
m
$ 60-
0 \o ......... -
m
...'..
I
z
40 '..
.......
....
20 -
.......
... %.
O T . , . . . . . . . . . . . . . . . . .
0
10 Pass Fail I
FIGURE 11.3 Corrosion fatigue of aluminum brass in ~ e a w a t e r . ~
180
Pass :> 50 Million cycles
160 +........ 0.1% proof
140
........
stress
.%.
.%.
2 120.
z$ looi
"........
rn
E! e....
arn ...... - 0
-
.-
8o
60\
*w..
..........
1
a,
c
2 40 a .%. rn- 1
.%.
20
0 0 0 0 ....... 0
...........
0
1 Strain Range (Yo) -

b l.O\ \0.8 \0.6 \C
1 10 100 1,000 11 100
Life (h)
FIGURE 11.5 Life vs. strain frequency for 70/30 copper-nickel in seawater at
10-1 5oc.5
Bowers reported the results of high-strain, reverse bend tests on 70/30

copper-nickel (C71500) over a range of frequencie~.~ The results, given in
Figure 11.5, show that the life increases as the frequency decreases, but
not in a linear manner. At very low frequencies, the curves show a marked
departure from linearity, with an increase in life, but not as much as would
be expected from the data at higher frequencies. The introduction of a
notch, with a theoretical stress concentration factor of 2.35 or a weld with a
matching filler, gave similar reductions in fatigue life compared with plain
samples, as shown in Figure 11.6 for a frequency of 10 cycledh.
Townsend conducted Vibraphore tests on 70/30 copper-nickel alloys
at frequencies in the range 50-100 H z . ~These results in air would not be
expected to be very different in seawater because of the high corrosion
resistance of these alloys and the fact that corrosion has a lesser effect
on corrosion fatigue at higher frequencies. Figure 11.7 shows the data
for 70/30 copper-nickel (C7 1500) and a high-strength version containing
0.54% silicon and 1.21% niobium. This alloy was discussed in Chapter 7 in
terms of its erosion corrosion resistance. Figure 11.5 shows that although
the alloy with silicon and niobium had a higher strength, its fatigue limit
was only a little better than the conventional C7 1500 alloy.
1.2,
1 -
0.8 -
Y
a,
0
m
%.- 0.6 .
2
5
a,
I! 0.4 -
v)
0.2 .
Life (cycles)
-Plain 0 Notched A Welded
FIGURE 11.6 High strain fatigue curves for 70/30copper-nickel in seawater at 10-1 5°C
at 10 ~ y c l e s / h . ~
3
300
250
2
5 200
2
5
.-p 150
L
II - . A AA
2f
A A A
100 L
010,000
50 I 100,000 1,000,000
Life (cycles)
10,000,000 100,
00,000
1 A 70/30 Cu-Ni W 70/30Cu-Ni + Si+Nb 1

FIGURE 11.7 Fatigue of 70/30 copper-nickel alloys in air at 50-1 00 H z . ~
a 120
....
P ....
%.
5 .%
.........
;1 0 0 -
2
z .
a
.-0 80- '....

-
-- -- -
-
+a .%.
....
-
E
al
a
60~'
40.-
-
-----
I I I I I I I .
I -1..
........
&....I
...
-.*...?
20 -
.%.
'%.
'%_
o - . 1 . 1 , I , r , z , I . I . I ..
FIGURE 11.8 Corrosion fatigue of 90/10 copper-nickel in seawater with a r n r n ~ n i a . ~
In the previous section, the results of Campbell on the corrosion

fatigue resistance of aluminum brass in seawater and seawater containing
ammonia were d e ~ c r i b e d Campbell
.~ also carried out tests on 90/10 and
70/30 copper-nickel, the results of which are shown in Figures 11.8 and
11.9. All the test conditions below the lines suffered no corrosion fatigue
failure after 50 million cycles. The data for 90/10 copper-nickel show that
0.75% ammonia produced a big reduction in the corrosion fatigue strength.
Although the loss in corrosion fatigue strength was less with 0.1% ammonia,
it was only with 0.01% ammonia that no effect on corrosion fatigue strength
compared with clean seawater was seen. This is surprising because in the
previous chapter, data were presented showing that 90/10 copper-nickel
did not increase its susceptibility to SCC in seawater containing 0.75%
ammonia. This suggests that the protective film on 90/10 copper-nickel
may be more easily disrupted by a cyclic stress than by a constant one.
70/30 copper-nickel proved to be even more susceptible to corrosion
fatigue in seawater containing ammonia than 90/10 copper-nickel. There
was a large decrease in the corrosion fatigue strength with 0.1% ammo-
nia, and there was still a significant decrease with 0.01% ammonia (Fig-
ure 11.9). 70/30 copper-nickel also had very good resistance to SCC in
160
140
1
.........
..... 0.1% proof
-2 120~
....... stress
.-..
B3 ..........
-1..
100-
.......
zern ....
.......
.-
c
80 -
. I -- I I I I - &...
-
-4
60- ....
.........
.......
40 -
- - - - - - - - +......
20 -
.....
01 . . . . . . . . . . . . . . . . . . . .
Mean Stress (MPa)
I -No Ammonia - 0.1% Ammonia -- 0.01% Ammonia I

FIGURE 11.9 Corrosion fatigue of 70/30 copper-nickel in seawater with arnrn~nia.~
seawater containing ammonia, and its effect on the corrosion fatigue

strength suggests that the protective film is more easily disrupted by cyclic
stresses than by constant ones.
In the 1980s, when 90/10 copper-nickel was first being considered for
cladding for the legs of offshore platforms, concern was expressed about
corrosion fatigue of the cladding. Harvey et al. reported tests on 90/10
copper-nickel in 3.5% sodium chloride (NaCl) at 15 H z . ~They used an R
value of 0.4, with axially loaded compact tension specimens and applied
anodic currents from 0 to 1,800 pA/cm2. However, normally, the copper-
nickel would be welded to the carbon steel (CS) legs, and Harvey et al.
also conducted fatigue tests on 90/10 copper-nickel coupled to an equal
area of CS. As the anodic current increased, the number of cycles to failure
decreased, with the biggest decrease in the range 0 to 300 pA/cm' (Fig-
ure 11.10). However, when coupled to CS, the number of cycles to failure
increased from -2.5 million to -4 million. The cathodic protection con-
ferred by the CS significantly reduced the effect of corrosion on fatigue life.
These results demonstrate that the use of 90/10 copper-nickel for cladding
offshore platform legs poses no corrosion fatigue risk.
0.5 -
*
O ~ . , . , , , . , , , , , , ~ , , . , ,
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000
Anodic Current Density (pA/cm*)
FIGURE 11.10 Cycles to failure for 90/10 copper-nickel as a function of anodic current
density.’
11.3 Nickel Aluminum Bronze

Because of its high strength, nickel aluminum bronze (NAB) is often chosen
for applications in which the alloy may be subject to high strains at low-
frequency cyclic stressing. Campbell summarized the data generated at BNF
Metals Research Association that was originally presented by Bowers.’-’
The test method was the same as that used for copper-nickel alloys, as
described in the previous section. The data are summarized in Figure 11.11,
where it can be seen that the life at low frequencies is much greater for cast
NAB than for wrought 70/30 copper-nickel (Figure 1 1.5). The excellent
resistance of NAB to corrosion fatigue at high frequencies is due mainly
to its high strength, whereas its resistance at low frequencies is due to a
combination of high strength and good resistance to corrosion.’ Bowers
pointed out that at frequencies below 10 cycles/h, there is little increase
in fatigue life, and a 10-fold reduction in the frequency only doubles the
corrosion fatigue life.’
Bentley and Duquette examined the corrosion fatigue of cast NAB
(C95800) in air and 0.5 M sodium chloride at 20 H z . Under
~ freely corroding
-1
10 100
1 10 100 1,000 10,000 100,000

..
I
1.01 1.000
Life (hours)
FIGURE 11.11 High-strain, low-cycle fatigue data for cast NAB in seawater at 32"C.5.8
conditions, there was a reduction in the threshold fatigue stress compared

with air, but only of the order of 15 MPa. Under applied currents, the
corrosion fatigue behavior was variable. With small cathodic currents, the
corrosion fatigue life increased beyond the corrosion fatigue life seen in air.
If the cathodic current was increased to 1.O mA/cm2, there was an increase
in corrosion fatigue life of 38% above the corrosion fatigue life seen in
air. The authors postulated that this was due to a thin alumina film on the
NAB, but alumina films have never been detected on NAB in seawater (see
Chapter 3).
Anodic currents t 0 . 2 mA/cm2 increased the fatigue life compared
to anodic currents under freely corroding conditions. In the anodic current
range 0.2-1 .O mA/cm2, the corrosion fatigue life was decreased because
of the creation of etching notches by anodic dissolution. However, at even
higher anodic current densities, the fatigue life was restored because the
notches were blunted by anodic dissolution and cracks did not form.
11.4 Prevention of Corrosion Fatigue
Where corrosion fatigue may occur, it is important to use good design

practice at the outset, using known corrosion fatigue properties of the alloy
being considered and adhering to the requirements of the design code that
is in force. It is also important to avoid designs that increase the risk of

localized corrosion, such as crevices, which can act as stress raisers.
However, there are occasions when the stresses are higher than an-
ticipated and cracks are detected on inspection. If the component must be
replaced, then a change to a more fatigue-resistant alloy will be needed, or
perhaps a change in design that reduces the stresses will have to be made.
The aluminum bronzes, and NAB in particular, offer a significant increase
in corrosion fatigue resistance compared with most other copper alloys.
Where the corrosion is more severe than anticipated, a corrosion
prevention method may be an alternative to a change of alloy. The beneficial
effects of even small amounts of cathodic protection on 90/10 copper-nickel
and NAB were described previously, where corrosion fatigue lives in excess
of those in air were possible. Where this is impractical, an organic coating
may be adequate. Where this is to be applied in situ, it is important that the
surface preparation requirements are adhered to. A poorly applied coating
may induce localized corrosion more severe than that which led to the initial
problem. In some cases, a metallic coating may be preferred, applied either
by electroplating or thermal spray. Such coatings usually contain minor
defects, and it is important that the metal chosen for the coating be anodic
to the copper alloy substrate to prevent localized corrosion at defects.
In closed-loop systems, corrosion inhibitors can be used to prevent
corrosion, remembering that the alloy chosen should still have an adequate
resistance to fatigue in air to give a satisfactory life.
References
1. P.M. Scott, “Corrosion Fatigue,” in Corrosion, 3rd ed., L.L. Shreir, R.A. Jarman,
G.T. Burstein, eds. (Oxford, U.K.: Butterworth Heinemann, 1994), p. 8: 144.
2. C.E. Jaske, J.H. Payer, V.S. Balint, Corrosion Fatigue of Metals in Marine
Environments (New York, N Y SpringedBatelle, 1981).
3. H.S. Campbell, BNF Metals Research Association Research Report, A1577,
February 1966.
4. M.O. Spiedel, “Influence of Environment on Fracture,” in Proceedings ofthe
Fifth International Conference on Fracture, vol. 6 (New York, N Y Pergamon,
1982), p. 2685.
5. J.E. Bowers, “The High Strain Fatigue Properties in Seawater of 70/30
Cupro-nickels and Cast Aluminium Bronze,” BNF Metals Research Association
Research Report, May 1974.
6. D.W. Townsend, BNF Metals Research Association Research Report, A 1546,
October 1965.
7. D.P. Harvey, T.S. Sudershan, M.R. Louthan, R.E. Swanson, “Corrosion Fatigue
Behaviour of Copper-Nickel Cladding for Marine Structures,” Coatings and
Bimetallics for Aggressive Environments (Hilton Head, SC: ASM International,
1985).
Publication 80 (Potters Bar, U.K.: Copper Development Association, 198 1).
9. R.M. Bentley, D.J. Duquette, International Copper Research Association Project
Report, 241, November 1979.
CHAPTER 12
Hydrogen Embrittlement
When carbon steel (CS) is used underground or subsea, it is common to

apply cathodic protection (CP) to prevent the steel from corroding. The po-
tential is usually in the range -850 to -1,040 mV SCE (saturated calomel
electrode), and any other metals connected to the CS will also receive this
level of CP. At these potentials, hydrogen evolution is the main cathodic re-
action, and if the hydrogen diffuses into the metal, it can become embrittled.
Materials well known to suffer hydrogen embrittlement (HE) include high-
strength steels, ferritic and duplex stainless steels, and some nickel alloys.
The following sections examine the resistance of copper alloys to HE.
12.1 Copper and Its Alloys

When HE occurs, hydrogen atoms are created as part of the cathodic re-
action. Instead of combining to form hydrogen molecules and evolving as
gas, the atomic hydrogen diffuses into the metal. Where it accumulates or
becomes trapped, embrittlement occurs. Copper and its alloys have a very
low hydrogen adsorption coefficient, making hydrogen entry into copper
very difficult. Hence, copper and its alloys are usually regarded as being
immune to HE.
High-strength copper alloys (for example, UNS C72420) have been
used for subsea fasteners on cathodically protected structures. In the 1980s,
BNF Metals Technology Centre tested two of these alloys following failures
of Alloy K-500 bolts (see the next section).'-2 One alloy was Hiduron 191
(C72420), and the other was Marine1 (also C72420, but with a niobium
addition to further increase strength).
The test method, which was described in detail by Campbell and
Francis, involved constant load tests of tensile samples exposed to natural
237
seawater containing 5 mg/L sulfide (as hydrogen sulfide [HzS]) to poison the
hydrogen recombination r e a ~ t i o nThe
. ~ samples were polarized to - 1,040
mV SCE for 60 days, after which they were removed, tensile tested, and
examined for indications of embrittlement.
Samples of Hiduron 191 were tested at 90% of the 0.2% proof stress
and showed no signs of embrittlement after 60 days' exposure.' Marinel
was tested at 75% of the 0.2% proof stress and showed no susceptibility to
embrittlement in either the standard aged condition or in material that was
aged and cold reduced
Marinel has also been tested using the slow strain rate test, and no
embrittlement was seen under CP in seawater at 20 and 50°C.4 Tuck also
reported no embrittlement of C72420 under cathodic charging in slow strain
rate tests.5
12.2 Nickel-Copper Alloys

The two most common nickel-copper alloys are Alloy 400 (UNS N04400)
and K-500 (UNS N05500). The latter is a high-strength version of Alloy 400
and was widely used for corrosion-resistant fasteners subsea in the North
Sea during the 1970s. However, in the early 1980s, there were failures of
several K-500 bolts subsea in a major North Sea oil field. The failures
were identified by BNF Metals Technology Centre as being due to HE. The
failures were reproduced in the laboratory, and a program of research was
undertaken to investigate the effects of various factors on the embrittlement
of K-500 and also to test some alternative alloy^.^
Efird reported K-500 bolt failures by HE from another North Sea oil
field, and he suggested that the problem could be overcome by keeping
the hardness of the K-500 below HRC35.6 However, Francis and Campbell
tested K-500 with hardnesses of 336 HV and 296 HV, the latter well below
HRC35, and found that both samples embrittled equally.' Stress was also
not important, as unstressed samples also embrittled. Figure 12.1 shows a
typical fracture face of K-500 after 60 days' exposure and tensile testing.
The embrittled area around the perimeter of the sample is clearly visible, as
are the secondary cracks along the rest of the gauge length. Embrittlement
was observed to begin at potentials of --750 mV SCE, and it became more
severe as the potential became more negative.
Tests on samples that were aged and cold worked showed a similar
depth of embrittlement to normally aged material. Campbell and Francis
also investigated the effect of time on the properties of cathodically charged
Hydrogen Embrittlement 239
FIGURE 12.1 Scanning electron microscopy image of fracture face of Alloy K-500 after
60 days’ hydrogen charging in seawater and tensile t e ~ t i n g . ~
K-500.3 Figure 12.2 shows the elongation and reduction of area for expo-
sures up to 200 days. The results show that the elongation decreased fairly
steadily over the period, but the reduction in area decreased rapidly over
the first 60 days and was almost constant thereafter. Figure 12.3 shows the
depth of embrittlement for both stressed and unstressed samples, and it can
be seen that embrittlement increased almost linearly with time over the 200
days. Francis and Campbell used these data to estimate the time to failure of
a 1.25-in (3 1.75-mm) diameter K-500 bolt under CP.’ With a stress around
the 0.2% proof stress, failure was estimated to occur in 2 to 4 years, while
at 50% of the 0.2% proof stress, failure would occur in 10 to 1 1 years. For
subsea structures, the design life is usually at least 25 years, and this means
that K-500 is not a suitable material for the bolting.
50 30
25
20
s
Y
C
15 2
m
u1
0
E
10
0
0 50 100 150 200
Charging Time (days)
FIGURE 12.2 Reduction of area and elongation of K-500 after hydrogen ~ h a r g i n g . ~
oiy
0 50 100 150 200
ChargingTime (days)
1- Stressed - -0- - Unstressed I

FIGURE 12.3 Depth of embrittlement of K-500 vs. charging time.3
Hydrogen Embrittlement 241
Erlings et al. observed HE of K-500 drill collars on CS downhole

tubulars in a magnesium chloride packer fluid.7 The hydrogen evolution
occurred on the K-500 as the cathodic reaction in the galvanic couple, with
the CS under downhole conditions.
Alloy 400 was tested by Price and colleagues and was also shown to
be very susceptible to HE-more than other nickel alloys
12.3 Avoidance of Hydrogen Embrittlement

All the common copper alloys appear to be immune to HE and can be used
freely in equipment to which CP is applied. It is clear from the data in the
previous section that Alloys 400 and K-500 should not be used in situations
where they are under stress and exposed to hydrogen evolution from CP or
galvanic coupling, for example subsea marine bolting.
References
1. R. Francis, H.S. Campbell, BNF Metals Technology Centre Research Report,
R538/1, June 1986.
2. R. Francis, H.S. Campbell, BNF Metals Technology Centre Research Report,
R611/14, March 1990.
3. H.S. Campbell, R. Francis, Brit. Corros. J. 30 (1995): p. 154.
4. 0. Andersen, M.W. Joosten, J. Murali, D.E. Milliams, CORROSION/96, paper no.
78 (Houston, TX: NACE International, 1996).
5. C.D.S. Tuck, CORROSION 2005, paper no. 462 (Houston, TX: NACE
6. K.D. Efird, M P 24,4 (1985): p. 37.
7. J.G. Erlings, H.W. de Groot, J.M.M. van Roy, MP 25, 10 (1986): p. 28.
8. C.E. Price, R.K. King, “The Embrittlement of Monel400 by Hydrogen and
Mercury as a Function of Temperature,” International Conference and Expo on
Fatigue Corrosion Cracking, Fracture Mechanics, and Failure Analysis (Salt Lake
City, UT: ASM International/ASTM, 1986): p. 8 I .
9. C.E. Price, J.A. Morris, “The Comparative Embrittlement of Nickel Base Alloys
by Hydrogen and Mercury,” Corrosion of Nickel Base Alloys (Columbus, OH:
ASM International/lNCO, 1985): p. 69.
CHAPTER 13
Fouling and Microbially Influenced

Corrosion
This chapter is concerned with the effect of biological attachments onto

copper and copper alloy surfaces, principally in waters. The effects may be
mechanical and/or corrosion related and may occur on both the macro and
micro scale. Not all biological effects are detrimental, and some beneficial
effects are described in the following sections.
13.1 Macrofouling
When almost any solid material is immersed in seawater, marine growths
will gradually attach themselves over a period of weeks or months. These
can take the form of shellfish, for example, mussels, or plants, such as sea-
weeds; they are termedfouling. More correctly, such attachment of marine
growths is termed macrofouling, whereas inicrofouling is the colonization
of metal surfaces by bacteria and will be discussed in the next section. Note
that some macrofouling only occurs at certain times of the year (for exam-
ple, mussels), whereas other types can occur all year round, (for example,
limnoria and tunicates). Macrofouling can also occur in fresh water, but it
is usually much less of a problem.
Macrofouling can cause a number of problems, one of which is the
increase of weight on offshore structures and also the increase in the stresses
on the structure under wave action. Fouling in heat exchangers can block
the tubes, dramatically reducing heat transfer, and can cause severe erosion
corrosion at partial blockages. Severe fouling can also significantly reduce
the water flow in piping systems. Finally, the environment beneath the
fouling can be very different to the one in bulk seawater, causing rapid or
localized corrosion.
243
Copper is toxic to many organisms, and it is well able to resist fouling

in seawater-so much so that the Royal Navy began cladding the immersed
parts of wooden warships in copper in the 18th century. However, it was
not until the second half of the 20th century that the fouling of copper and
its alloys was studied in detail.
Efird carried out exposure trials on flat panels at the LaQue Labora-
tory, North Carolina, over a period of 5 years.' The materials were carbon
steel (CS), copper (C1 lOOO), and 90/10 and 70/30 copper-nickels. The ap-
pearances of each material after 3, 9, 18, 36,48, and 60 months' exposure
are shown in Figures 13.1-13.4. CS showed increasingly heavy fouling
with time (Figure 13.1), but this sloughed off from time to time, when
the fouling was heavy and the corrosion products beneath were not adher-
ent enough to the metal surface. After 60 months, part of one panel had
completely corroded away. With copper (Figure I3.2), there was no signif-
icant fouling after 9 months; rather, there was just a green layer of copper
hydroxychloride ( C U ~ C I [ O H ] ~ - ~ over
H ~ Oa) thin layer of copper oxide.
After 36 months, there was some attachment of fouling, and 60% of the sur-
face was covered after 60 months. The appearance of the 90/10 copper-
nickel panels (Figure 13.3) was somewhat similar to copper, with no
macrofouling after 9 months and 70% coverage after 60 months. The 70130
copper-nickel panels were also similar in appearance (Figure 13.4). There
was no significant difference in the fouling behavior of pure copper and the
copper-nickel alloys. A close examination of Figures 13.2-1 3.4 reveals that
after 36 months, the fouling was attached to the Alloy 400 mounting bolts.
After 48 months, it was still attached to the bolts but was beginning to en-
croach onto the panels. Only after 5 years was the fouling well established
on the panels.
The corrosion rate of all three copper alloys was in the range 2.5-
7.5 pm/y. The copper metal release from the alloys over this range of
corrosion rates is too low to cause the fouling resistance observed. In fact,
the corrosion rate, and hence, the copper release rate, for 70/30 copper-
nickel is about one-third that of pure copper, but the fouling resistance is
about the same. It is commonly stated that a minimum corrosion rate of 18
pm/y is required to release enough copper to suppress fouling.'
Over many years, copper and its alloys lose and regain their fouling
resistance, as shown by the periodic attachment of fouling and its sloughing
off. This was observed over 14 years at the LaQue Laboratory.2 Efird
suggested that in addition to the release of copper, the copper oxide layer was
also toxic.' After it became covered with a layer of copper hydroxychloride,
the fouling could attach, but as this layer is poorly adherent, the fouling
FIGURE 13.1 The S·year sequence of fouling on CS in seawater (exposure periods from left to right): (top) 3
months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months. 1 (See color section.)
FIGURE 13.2 The 5-year sequence of fouling on copper in seawater (exposure periods from left to right): (top) 3
months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.' (See color section.)
FIGURE 13.3 The 5-year sequence of foul ing on 90/10 Cu-Ni in seawater (exposure periods from left to right) : (top)
3 months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.1 (See color section.)
FIGURE 13.4 The 5-year sequence of fouling on 70/30 Cu-Ni in seawater (exposure periods from left to right): (top)
3 months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months. 1 (See color section.)
Fouling and Microbially Influenced Corrosion 249
would slough off when it became too heavy. The copper oxide would
then be exposed, and the surface would resist fouling until the copper
hydroxychloride layer reformed.
The mechanism was also supported by the results from exposures of
half-coated panels of 90/10 copper-nickel, where fouling was seen up to
the boundary of the coating, but not on the metal. If fouling resistance was
conferred by copper release alone, then some resistance to fouling on the
coating adjacent to the metal should have been seen.
Copper-containing antifouling paint was also evaluated, and though
it had an effect, it was not as effective at resisting fouling as solid metal.'
Copper and copper-nickels were also very resistant to fouling at the water
line compared with plastics and other inert materials.' Even when copper,
or its alloys, did foul, the corrosion rate was no different to that on material
exposed in flowing seawater, where fouling was much less; that is, the
fouling was having no significant effect on the corrosion rate.'
The antifouling properties of 90/10 copper-nickel mean that it has
applications where resistance to fouling is essential. In many cases, solid
copper-nickel would be uneconomical, and a number of methods of apply-
ing copper-nickel coating onto steel have been examined. The most common
are hot-rolled, clad-steel sheets, adhesive-backed foil, and granules embed-
ded in rubber. The latter are cylinders approximately 1 mm diameter and
1 mm long, with a density such that approximately 30% of the exposed sur-
face is metal and the total coating thickness is about 3 mm. Campbell et al.
exposed these three product forms as well as expanded 90/10 copper-nickel
mesh bonded onto a rubber base (EPDM) in Langstone Harbor, United
Kingdom, for up to 8 years.3 All four products showed excellent resistance
to fouling. The performance of the granules in rubber is surprising as only
30% of the surface is metal. However, the particles are small and close
together, so perhaps this small area of rubber is prevented from fouling by
copper release from the metal.
One of the first modern uses of copper-nickel to prevent fouling
was for the hulls of small ships. The earliest of these was the Asperidu
ZZ, launched in 1968, which is still afloat and has excellent resistance to
fouling. Between 1968 and 1991, 10 ships were built with solid hulls,
mostly of 90/10 copper-nickel, and these have resisted fouling well. Not
only does the copper-nickel reduce downtime for the removal of fouling,
but also, a cleaner hull means that the boats are more economical on fuel.
The average fuel savings were calculated at 25% in 1977 and are probably
greater now with higher fuel costs.4 Corrosion on all these vessels has been
minimal.
Although the 70/30 copper-nickel welds on the 90/10 copper-nickel

hulls are slightly more noble than the base 9 0 4 0 metal, they were no more
prone to fouling. Even when fouling does attach on copper-nickel, it is very
easy to remove, compared with fouling on CS and inert material^.',^
Some seven boats have been built with copper-nickel-clad steel hulls,
using 2 mm of 90/10 copper-nickel. Most of these have performed well,
but some had poor fouling resistance. In some cases, this was thought to be
due to the boats being tied up for long periods in stagnant and/or polluted
water, where the toxic copper oxide film would not form properly.
Another method of obtaining antifouling is to weld copper-nickel
sheets onto steel hulls. Twelve large panels were welded onto the hull of
the 16-knot oil tanker Arco Texas and examined after 2 years. The tanker
had made seven trips through the Panama Canal, and the panels were intact
and performing well, even though several showed mechanical damage from
forceful impacts on the sides during passage through the canal.4
Adhesive 90/10 copper-nickel foil comprises sheets that are 0.15-
0.3 mm thick and can be molded to curved surfaces. This technique has
mostly been used on fiberglass-hulled vessels using an overlapping wall-
papering technique, which also confers excellent resistance to fouling.”‘
Unfortunately, this product was no longer available at the time of writing.
A more detailed evaluation of the use of copper-nickel for ships’ hulls
and its performance up to 1997 is contained in the work of Powell and its
appendi~es.~
If copper-nickel is used to protect steel structures offshore, the areas
that are of chief concern are the splash zone, the water line, and the fouling
zone, which usually extends several meters below the water line. If the
copper-nickel is welded onto the steel, it will be galvanically coupled to it
and corrosion will be reduced, which would be expected to reduce fouling
resistance. Powell and Michels reported the results of a 10-year trial in
which the fouling resistance of bare steel piles was compared with that of
piles with directly welded copper-nickel cladding, piles with copper-nickel
cladding insulated from the steel with concrete, and piles with rubber-
insulated lad ding.^ The results (Table 13.1) show no significant difference
between using concrete and rubber as an insulator. Both of these resisted
fouling well. Although directly welded copper-nickel cladding lost some
of its fouling resistance, it had only half the fouling seen on bare steel.
Copper-nickel cladding was first used offshore on platform legs on the
Morecambe Bay Gas Platform in the United Kingdom in the 1980s.After 20
years, the 4-mm-thick cladding was in good condition, and divers reported
that removal of fouling from the directly welded cladding was much easier
TABLE 13.1 Biofouling mass on copper-nickel-sheathed

steel pilings after 10 years’ exposure5
Exposure Biofouling Coverage
Piling (years) (kg/m2) (“4
Bare steel 5 18.0 100.0
10 12.0 100.0
Concrete insulated 5 0.36 1.9
10 0.40 1.2
Directly welded 5 7.95 44.3
10 4.43 36.8
Rubber insulated 5 0.26 I .4
10 0.55 4.5
10 0.35 2.8
10 0.62 5.3
than from bare steel. Since that time, the rubber coating embedded with
copper-nickel granules has also become available, and several offshore
platforms have used this for protection of structural legs, crossibracing,
and ~ i p i n g . ~ - ~
90/10 copper-nickel has been used for seawater piping for many years,
and its resistance to fouling means that other antifouling measures may not
be required, or only intermittently. Albaugh reported experience from the
Gulf of Mexico, where 8-in (203-mm) CS piping was severely fouled in
just 22 months of operation, while 90/10 copper-nickel piping showed no
fouling.6 It was clear that 6-in (152-mm) 90/10 copper-nickel piping would
be adequate for the water flow because of the alloy’s resistance to fouling.
Albaugh carried out a 20-year life cycle cost analysis and showed that 8-in
copper-nickel piping was 73% of the cost of CS, while the cost of 6-in
copper-nickel piping would be only 65% that of steel.6
13.2 Microfouling and Microbially Influenced Corrosion

For microbially influenced corrosion (MIC) to occur, it is necessary for
bacteria to colonize the metal surface. Despite the toxicity of copper, all
organisms require it to sustain them, usually in small quantities. Some
bacteria are more copper-tolerant than others, and it would be expected that
they would colonize the metal surface first.
Costerton et al. reviewed the formation of biofilms on metal surfaces.’
Once the first bacteria colonize a surface, they produce extracellular material
that typically consists of polysaccharide polymers. As the biofilm grows
BULK SEAWATER
ANAEROBIC AEROBIC
BACTERIA BACTERIA
FIGURE 13.5 Schematic of biofilm on an immersed metal surface. (See color section.)
and thickens, the aerobic bacteria consume the oxygen near the metal-
biofilm interface, and anaerobic bacteria become active. In the outer layer
of the biofilm, oxygen can still penetrate, and aerobic bacteria are active.
This is shown schematically in Figure 13.5. The secondary colonization of
the biofilm can promote or inhibit the activity of the primary bacteria, de-
pending on the type.7 Corrosion only occurs once an effective microcolony
is established and a potential difference is established between different
areas on the metal surface. The growth and development of biofilms are
affected by the temperature, pH, water velocity, and surface roughness.8
Most bacteria thrive best within a limited range of temperature and pH,
although these can vary significantly from species to species. Bacteria do
not adhere well to metal surfaces if the water flow is high or the surface is
very smooth.
Chamberlain and Garner followed the development of a biofilm on
90/10 copper-nickel in natural seawater and found that after 4 weeks, the
surface was colonized by bacteria and diatoms.’ After 30 weeks, Ectocar-
pus, a copper-tolerant species, was present, and after this, other species
were able to form secondary colonies. After 9-10 months, sloughing of the
biofilm occurred, exposing fresh metal. Chamberlain and Garner concluded
that the high copper concentration at the metal surface initially prevented
colonization and enabled the protective oxide film to grow. The subsequent
colonization by Ectocarpus and other species prevented sulfate-reducing
bacteria (SRB) from colonizing the metal surface and accelerating corro-
sion.
Bacteria cause corrosion not by directly reacting with the metal, but
because some of their metabolic by-products are aggressive to the metal.
There are three common types of reaction that lead to corrosion: one of these
is iron-oxidizing bacteria, which mainly affect steel and cast iron; a second
type is sulfur-oxidizing bacteria, which produce sulfuric acid (H2S04) as a
by-product, and any metal that is attacked by sulfuric acid can suffer MIC
by these bacteria; the third type is SRB, and this has received the most
study. These reduce sulfate in the environment to sulfide as part of their
metabolic processes. Unlike the first two types, SRB are only active under
anaerobic conditions.
Not all bacteria are detrimental to metal surfaces. In Chapter 3, the
development of protective hydrotalcite (Mg6A12C03[OH]16.4H20) films
on aluminum brass in seawater was described. These films are rich in
aluminum and magnesium, minor constituents of the alloy and seawater,
respectively, and they can only form if there are bacteria present to chelate
the copper and prevent it from forming a copper oxide film, as happens in
3% sodium chloride (NaC1) s o l ~ t i o n .The
' ~ hydrotalcite is able to buffer
the pH at the metal surface, as described in Chapter 3. Not only does the
hydrotalcite form on the metal surface, but it also forms within the biofilm,
growing as the biofilm grows."
Although the role of biofilms in corrosion has been discussed almost
as long as corrosion has been studied, work on copper alloys was slow
to take off, presumably because of the known toxicity of copper ions. In
the 1930s and 1940s, a series of research programs were conducted at
BNF Metals Research Association that included elements of MIC. Many of
these elements were summarized by Rogers in a 1948 paper that is seldom
referenced." Following are some of the findings from this review.
Bacteria isolated from various estuarine waters showed a range of
copper tolerance. Some could only tolerate 10 mg/L or so, but others could
tolerate in excess of 100 m g L copper, and one species was found to tolerate
2,000 mg/L copper. Bacteria with a high copper tolerance were always
found in waters where MIC of copper alloys was occurring.
Cultures isolated from an estuarine water greatly increased the cor-
rosion of 70/30 brass in stagnant solution, more so than occurred under
inert deposits under sterile conditions. The addition of organic acids that
are found in seaweeds and can act as bacterial food increased pit depths
by a factor of 2-10. The addition of peptone to the water strengthened the

protective film so that some bacteria could not break it down, although the
most aggressive still did.
Rogers postulated that the acceleration of attack was caused by an
increase of hydrogen acceptors and the establishment of alternative oxygen
and reduction systems during bacterial metabolism, that is, depolarization
of the cathodic reaction." Samples of seawater that caused rapid pitting of
copper alloys contained an organic sulfur compound that could be reduced
to a mercaptan and oxidized back to a disulfide very rapidly. Hence, this
compound could act as an oxygen carrier, depolarizing the cathodic reac-
tion, the reduction of dissolved oxygen (DO), and hence accelerating the
rate of pitting.
It was also found that pigments could cause accelerated corrosion of
copper alloys in seawater. Melanin was one such pigment, which is pro-
duced by the action of the enzyme tyrosinase on tyrosine. It was shown that
bacteria producing tyrosinase could accelerate corrosion in direct propor-
tion to the amount of tyrosine present.
Tests were also conducted using organic sulfur compounds that are
known to be metabolic by-products or breakdown products. Additions of
cystine or glutathione greatly increased the depth of impingement attack on
70/30 brass when added to sterile seawater, compared with that in sterile
seawater alone. The acceleration of attack was similar to that seen when
inorganic hydrogen sulfide (H2S) additions were made to the seawater.
In a separate series of atmospheric exposure tests of brass plates, deep
pits occurred when molds formed on the metal panels. Rogers postulated
that this was due to organic acids, which are produced during the growth
of many molds."
The composition of bacterial exopolymeric films (usually polysac-
charides) can vary, with some being aggressive toward copper. Jolley et al.
showed that some exopolysaccharides could accelerate the corrosion of cop-
per and anchor the colony to the surface.12-14Geesey and colleagues showed
that the acidic polysaccharides were the most However, the
effect of acidic polysaccharides was variable, with some being much more
aggressive toward copper than others.
In Chapter 5 , Type I11 pitting of copper was discussed, as was the
possible involvement of bacteria in the mechanism. Fischer et al. described
tests carried out on the biofilms removed from failed copper pipes from a
German hospital. l 7 These carried water at approximately 50°C and showed
the typical pepper-pot appearance of perforations. The biofilm was present
between the metal and the oxide-hydroxide layer. This implies the presence
of a species very tolerant of copper that can form a biofilm prior to the
formation of the oxide layer. Pitting occurred where the biofilm was dis-
rupted. This was believed to be due to a copper concentration cell, due to the
large amount of copper in the exopolymeric layer. In addition, peroxide and
hydroperoxides were detected, and these may have provided a more rapid
cathodic reaction than the reduction of DO, thus accelerating the corrosion
process.
The phenomenon of blue water was discussed in Chapter 4, and it
occurs in soft, fresh waters when the water is left stagnant in pipes for
an extended period (days or longer) so that a protective oxide film does
not form. Critchley et al. compared the dissolution of copper from pipes
exposed to sterile water with the dissolution of pipes exposed to sterile water
to which an addition of bacteria from a problem water had been made."
Much higher rates of copper dissolution occurred on tubes exposed to
the water with bacteria. This suggests that bacteria can contribute to the
problem of blue water and that it need not be solely chemical.
Much of the work in seawater has been concerned with the effect of
sulfur compounds, particularly hydrogen sulfide, produced by SRB. Davis
et al. carried out tests on copper, 90/10 copper-nickel, and a single-phase
aluminum bronze (C60800) in four environments: aerated seawater, aer-
ated sterile seawater, deaerated seawater, and deaerated sterile seawater. l 9
Samples in aerated environments corroded more than samples in deaerated
ones, and biological activity accelerated the formation of the oxide film in
aerated water and the sulfide film in deaerated water. Biological activity was
clearly accelerating chemical and electrochemical processes. When sam-
ples were transferred from deaerated to aerated conditions, the corrosion
rate increased, and the increase was greater in water with bacteria present.
This demonstrates the importance of bacteria in seawater to the corrosion
of some common copper alloys.
The nature of the protective film on 90/10 copper-nickel, where a thin,
chloride-rich layer beneath the copper oxide causes cathodic polarization,
was discussed in Chapter 3. This film occurs whether the samples are
exposed to natural seawater or 3% sodium chloride solution. However, the
cathodic polarization is lost if the seawater is changed from aerated to
deaerated, while it is not lost with 3% sodium chloride solution.*()This
suggests that anaerobic bacteria are contributing to the loss of cathodic
polarization, probably due to the production of sulfur compounds.
Chamberlain and Gamer investigated SRB and found that some
species were particularly copper-tolerant and could colonize 90/10 copper-
nickel surfaces, as opposed to being secondary colonizers of the biofilm.g
After 14 days' exposure, substantial quantities of sulfur were detected in
the film on the metal surface. When samples were moved from deaerated to
aerated seawater, the same acceleration in corrosion was seen as occurs with
inorganic sulfide additions. This showed that the sulfide does not have to be
produced elsewhere when 90/10 copper-nickel suffers attack. Chamberlain
and Garner postulated that the variable performance of 90/10 copper-nickel
under polluted conditions in the literature could be due, at least in part,
to variable colonization by SRB, depending on the local availability of
copper-tolerant species.’
Little et al. exposed samples of welded 90/10 copper-nickel piping
in natural seawater with and without SRB additions.2’ The welded areas
contained crevices and irregularities and were favored sites for SRB colo-
nization. This explains why service failures of piping are often downstream
of the welds and close to them.
In an unpublished review, Tuthill described some service failures due
to MIC.22 In two of these, 90/10 copper-nickel was used to replace Alloy
400 that failed by MIC, thought to be SRB. However, in another plant,
which used a natural, fresh water for cooling, 90/10 copper-nickel failed,
which was thought to be due to MIC, whereas 70/30 copper-nickel was
successful. This suggests that 90/10 copper-nickel can suffer MIC due to
SRB when particularly aggressive bacteria are present. However, Tuthill
knew of no cases of MIC failure of 70/30 copper-nickel.22
Nicklin described MIC corrosion problems associated with 70/30
copper-nickel heat exchanger tubes in Royal Navy submarines.” The fail-
ures were localized pitting, believed to be due principally to SRB introduced
during U.K. docking. All submarine coolers had the same type of bacteria
present, but only the ones in service in warmer waters suffered corrosion
problems.
Range1 et al. reported failures of 70/30 copper-nickel heat exchanger
tubes that handled an estuarine cooling water and were also dosed with fer-
rous sulfate ( F e S 0 4 . 7 H ~ 0 )Although
.~~ there were black spots on the tubes,
these were not sulfides. They identified both iron- and sulfur-oxidizing bac-
teria on the tube surface, and it was thought that one of these was the cause
of the problem. No detailed study of the effect of these bacteria on copper
alloys has been published.
Alloy 400 has been observed to suffer from severe attack in the
presence of sulfides in seawater. Gouda et al. studied Alloy 400 exposed to
SRB cultures and also stagnant seawater from the Arabian Gulf at 19, 30,
and 50°C.25 They found that SRB readily colonized the metal surface and
that corrosion took the form of localized pitting. Attack was most severe
when exposure alternated between deaerated and aerated conditions, as was
found for other copper alloys. They also found that attack was most severe
at 30”C, with less at 19 and 50°C. This shows that particular bacteria have
an optimum temperature range for high activity.
Similar results were reported by Little et al. for Alloy 400 in the Gulf
of Mexico, and the LaQue Laboratory, in North Carolina, found pits up to
1.07 mm deep after 3-year exposures in stagnant, natural sea~ater.’“~’
Lee et al. carried out tests on Alloy 400 with both inorganic sulfides
and SRB.28The corrosion with inorganic sulfides was uniform, whereas
corrosion with SRB took the form of pitting. They concluded that this was
because the sulfide concentration varied across the metal surface, depending
on local SRB activity, and this generated local concentration cells. A similar
reaction with inorganic sulfides could only occur if the flow was sufficiently
turbulent to mechanically disrupt the sulfide film.
In his review of service failures due to MIC, Tuthill recorded three
service failures of Alloy 400 heat exchanger tubing due to SRB.” Two of
these were solved by replacing the tubes with 90/10 copper-nickel, and the
other case was solved by replacement with 70/30 copper-nickel.
The nature of biofilms on metals is very variable, and Dexter and
LaFontaine reported a case in which biological activity increased galvanic
corr~sion.’~ They tested 6% Mo austenitic stainless steel (SS) coupled to
an equal area of copper in natural seawater. In half of the couples, the
SS had been preexposed to seawater for 4 weeks, whereas the others had
freshly cleaned panels. The galvanic current was approximately 2 FA for
the freshly cleaned samples, whereas the ones with preexposed SS had a
galvanic current of 500-1,000 PA. The copper panels coupled to preexposed
SS also suffered a higher weight loss than those coupled to freshly exposed
SS, but only 2-3 times greater. The reason for the acceleration was the
biofilm that forms on SS in aerated seawater and depolarizes the cathodic
reaction, the reduction of DO. Without this film, SS are not very efficient
cathodes.30 Although the galvanic current increased greatly with biofilm-
coated SS, the weight loss increased by much less. This was because even
if the SS is not a very efficient cathode, the copper will still be corroding
in natural seawater, and the initial rate will be quite high, until a protective
film has formed.
13.3 Bacterial Toxicity
In previous sections, the toxicity of copper toward microbes has been de-
scribed in terms of reducing fouling on metal surfaces. However, copper is
very toxic toward some bacteria that can be very harmful to human beings.
TABLE 13.2 Population of E. colion various

pipe surfaces with ti mez9
E. coli counthnitial f.coli (%)
Material 5h 24 h
PVC 68 75
Polybutylene 93 92
Cross-linked PVC 89 99
Copper 1 0
Wells described tests using different materials and tap water containing Es-
cherichia ~ o l i . The
~ ' water was dechlorinated and circulated through tubes
of three different plastics and copper. The E. coli population was deter-
mined at intervals, and the results are shown in Table 13.2. These clearly
show a rapid and dramatic reduction in the E. coli population in water pass-
ing through copper tube, whereas that going through the polymer tubing
showed very little reduction.
Following the demonstration of toxicity to E. coli, Wells did further
tests on Legionella pneumophila in tap water.32The tests were carried out
in a similar manner to those on E. coli, but the high pH of the water (pH
9-10) caused a rapid reduction of L. yneumoyhila even in glass. The re-
sults in Table 13.3 were obtained using water with a biological buffer to
keep the pH closer to neutral. It can clearly be seen that there was no dif-
ference between the glass control system and PVC tubing, whereas both
metal piping systems showed dramatic reductions in Legionella popula-
tions after only 24 h Extended tests, for up to 6 months, showed no signif-
icant biofilm on the copper tube, whereas the polybutylene tube contained
fungi.
West et al. also compared the toxicity of copper pipes to Legionella
and other bacteria compared with several polymeric materials and a glass
control.33 They found that copper was particularly toxic to Legionella,
TABLE 13.3 Growth of Legionella in Trizma-buffered water in contact

with different tubinu materials")
Legionella count
Tubing 0 days 1 day 4 days 8 days
Glass (control) 17,500 66,000 290,000 4,900,000
Copper 17,500 < 10 <I0 < 10
Galvanized steel 17,500 < 10 < 10 < 10
PVC 17,500 64,000 280,000 4,300,000
1.00E+08
0 10 20 30 40 50 60 70 80 90 100
Time (min)
I U I E 7 +1E6 +1E5 *1E4 -1E3 -@ 1E2 I

FIGURE 13.6 Effect of innoculum size of EMRSA-16 on the time for total kill when
exposed to copper.32 (See color section.)
reducing or halting its growth, whereas more copper-resistant, but non-

harmful, bacteria continued to grow. Polybutylene was readily colonized
by Legionella, and the balance of the bacterial population was unchanged
in polybutylene tubing.
Hospital-acquired infections are a current concern, particularly the
so-called superbugs such as methicillin-resistant Stuplzylococcus ciureiis
(MRSA). About 80% of infectious diseases are spread by touch, and a hand
contaminated with influenza A virus, for example, will contaminate the
next seven surfaces that are touched.
A study was carried out to compare the survival rates of MRSA as
a function of innoculum size on copper plates, as shown in Figure 13.6.'4
The results show that as many as 10 million MRSA are inactivated in 90
min. Further work has compared the viability of MRSA on copper, some
copper alloys, and Type 304 SS (UNS S30400).'5 The results in Figure 13.7
show that the MRSA count decreases to very low levels in just 90 min on
copper and in 270 min on 80/20 brass, whereas there was a much smaller
reduction on nickel silver and very little on the SS. The viability of MRSA is
clearly reduced as the copper content of the alloy increases. An 18-month
trial began in April 2007 in a U.K. hospital to compare copper surfaces
1.00E+08
-
-I
E
1.00E+06
-5
L
n
c)
1.00E+04
0
0
.-m
0)
c
m
1.00E+02
1.OOE+OC
0 50 100 150 200 250 300 350 400
I +Comer - Time (min)
Brass -m- Nickel Silver -A- 304 SS
FIGURE 13.7 MRSA viability on some metals at 20°C.33 (See color section.)
1
against traditional materials in controlling hospital-acquired infections such

as MRSA.
In 2008,275 copper alloys were registered with the U S . Environmen-
tal Protection Agency as antimicrobial materials that kill these bacteria. The
list includes copper, brasses, and bronzes, and these alloys offer a range of
properties that make them suitable for antimicrobial surfaces.
Where the toxicity of copper and copper alloy surfaces to fouling is not
sufficient, additional methods of fouling control must be considered. One
of the most well-known methods of preventing fouling, either macro or
micro, is to dose either intermittently or continuously with chlorine or
hypochlorite (OCl-).36-38 Chlorine is readily generated electrolytically in
seawater or brines and is the most common way of controlling fouling.
Dosing may only be needed in the fouling season, which is site-specific,
and the dose may be varied as fouling becomes more or less severe. The
effect of chlorination on materials and the chemical reactions involved were
reviewed by Goodman.39
One effect of chlorine is to reduce the resistance of copper alloys to

erosion corrosion if the chlorine dose is too high.4" This is discussed in
detail in Chapter 7. If ferrous sulfate dosing is being carried out to help
with film formation on heat exchanger tubes, it is important to turn off
the chlorine dosing when ferrous sulfate is being added as it prevents the
formation of a protective hydrated iron oxide (FeOOH) d e p ~ s i t . ~ '
Where a water contains small amounts of hydrogen sulfide, such as
from SRB activity, chlorine additions can be used to prevent accelerated
corrosion.42 By using a redox probe in the cooling-water line, chlorine
can be injected whenever the redox potential drops, but excessive chlorine
dosing can be avoided by stopping dosing before the redox potential gets
too high. This method was used to prevent accelerated attack of aluminum
brass condenser tubes at a Belgian power station and to avoid a costly
retube, as occurred at a sister tat ion.^'
However, chlorine does not prevent accelerated attack of all alloys.
Gouda et al. found that chlorine additions reduced sulfide pitting due to
SRB of Alloy 400 but did not prevent it."
Chlorine dosing is less damaging if it is carried out intermittently,
rather than c o n t i n u ~ u s l yThe
. ~ ~most
~ ~ ~suitable dosing concentration and
frequency depend on the local conditions and must be determined for each
plant.
In fresh waters, with much lower chloride concentrations than seawa-
ter, copper alloys are more tolerant of chlorine. Tuthill et al. presented data
showing that copper and a number of commonly used copper alloys were
not affected by up to 2 mg/L chlorine added continuously, provided water
velocities were not excessive.a
When potable water systems are installed in large public buildings,
it is common to disinfect the tubes with a strong solution of hypochlo-
rite, at anywhere from 20 to 100 mg/L free chlorine. Francis and Saxton
investigated the effect of strong disinfection regimes on the subsequent
performance of copper tubes.45Tests up to 30 days under both stagnant and
flowing conditions (up to 3 d s ) showed no effect of a pretreatment of 100
mg/L chlorine for 2 h or 20 mg/L chlorine for 16 h. The tests included both
a hard well water and a soft, surface-derived water at room temperature
and at 60°C. Under continuous chlorine dosing at 1 mgL, there was no
effect on copper corrosion at room temperature under any flow condition.
At 60"C, there was a marked increase in general corrosion rates in both
waters, an increase in the susceptibility to Type I1 pitting under slow flows
in the soft water, and an increased susceptibility to erosion corrosion at
high flow velocities. The effect of chlorine on Type I1 pitting is discussed
in more detail in Chapter 5.
Bond eta]. reported the results from tests lasting up to 12 months using
the same disinfection regimes, flow rates of 0.1 and 2 m/s, and ambient and
60°C temperature^.^^ After 12 months, the corrosion rates of all the tubes
were in the range 1-3 pm/y-a very low rate. The corrosion rates after 4
months were somewhat higher, but all were less than 0.01 mm/y-still a
low corrosion rate. Type I1 pitting was not seen in the hot, soft water with
continuous chlorine dosing because the soft water used was different to that
used in the 30-day tests and did not support Type I1 pitting.
There are other ways of controlling fouling in addition to chlorine.
Campbell and Searle compared the fouling of various metals and non-
metals under slow flow conditions in seawater.47One tank was untreated,
as a control; another was dosed to give a chlorine residual of approximately
0.5 mgL; and a third had electrically stimulated copper anodes to provide
approximately 60 ppb copper ions in solution. The fourth tank had elec-
trically stimulated anodes of copper and aluminum as this was claimed by
some to be superior to copper alone, with levels of approximately 60 ppb
copper and approximately 30 ppb aluminum. Fouling and corrosion of a
range of copper alloys was greatest in the control tank and least in the
tank dosed with chlorine. There was no significant difference between the
results from the tanks dosed with copper and with copper and aluminum.
The degree of fouling and corrosion was intermediate between that of the
control tank and the tank dosed with chlorine.
Where heat exchanger tubes become fouled with deposits, it is pos-
sible to restore the heat transfer efficiency by passing sponge balls through
the heat exchanger. These are injected into the inlet water box and are
collected from the discharge water box. Excessive use of sponge balls can
cause accelerated corrosion of copper alloy condenser tubes, and Sat0 and
colleagues described the optimum cleaning regime to maintain heat trans-
fer without accelerating corrosion of copper alloy tubes.48-”’ They found
that fewer than six balls per tube every 1-2 weeks was enough to give
satisfactory heat exchanger performance.
Carrera and Gabetta reviewed the biocide options available to the
offshore oil and gas i n d ~ s t r y . They
~ ’ considered a wide range of equipment
and the best way to control bacteria in each one, as biocides are not always
the best solution. They considered both oxidizing biocides, such as chlorine,
and a range of organic biocides, listing both the advantages and disadvan-
tages of each. Many, but not all, of MIC problems are due to SRB, and one
advantage of some biocides is their toxicity to a wide range of bacteria.
When considering sulfide attack by SRB of 90/10 copper-nickel in
seawater, Little et al. observed no beneficial effects on corrosion of iron or
ferrous sulfate additions.” However, additions of sodium sulfite (NazS03)
reduced the attack, although they did not prevent it. The ineffectiveness
of ferrous sulfate additions in preventing sulfide attack of copper alloys
has been noted by many researchers. Sat0 et al. found that ferrous sulfate
did work against sulfides if a protective film was first formed in clean
seawater.52
In Chapter 5, the attack of copper tubes by Type I11 pitting was shown
to have a microbiological contribution. Fischer et al. described the measures
adopted at a German hospital to prevent failures due to Type I11 pitting.53
In the hot water pipes, increasing the water temperature above 55°C and
eliminating dead water regions prevented attack. In the cold water pipes, it
was necessary to use ultraviolet radiation to reduce the active bacteria in
the water. In the long term, a plant was built to increase the bicarbonate
alkalinity of the supply water, as described in Chapter 5.
References
I . K.D. Efird, MP 15,4 (1976): p. 16.
2. K.D. Efird, D.B. Anderson, M P 14, 1 (1975): p. 3.
3. S. Campbell, R. Fletcher, C. Powell, “Long Term Exposure Trials Evaluating the
Biofouling and Corrosion Resistance of Copper-Nickel Alloy Sheathing
Materials,” 12th International Congress on Marine Corrosion and Fouling
(Southampton, U.K.: 2004).
4. C.A. Powell, “The Biofouling Resistance of Copper-Nickel-Properties and
Experience,” CDA Report, July 1997.
5. C. Powell, H. Michels, “Review of Splash Zone Corrosion and Biofouling of
C70600 Sheathed Steel During 20 Years Exposure,” EUROCORR 2006
(Maastricht, Holland: European Federation of Corrosion, 2006).
6. E.K. Albaugh, “Copper-Nickel Piping Reduces Costs, Biofouling/Corrosion,”
World Oil, November 1984.
7. J.W. Costerton, G.G. Geesey, P.A. Jones, M P 27,4 (1988): p. 49.
8. T.R. Bott, .!?fluent Water Treat. J. 19 (1979): p. 453.
9. A.H.L. Chamberlain, B.J. Garner, Int. Biodeteriorution 24 (1988): p. 213.
10. J.E. Castle, D.C. Epler, “The Development of Inorganic and Biological Fouling
Layers on Copper Based Alloys,” Progress in the Prevention of Fouling in
Industrial Plants (Nottingham, U.K.: Institute of Corrosion, 198 I ) .
11. T.H. Rogers, J. lnst. Met. 75 (1948): p. 19.
12. J.G. Jolley, G.G. Geesey, M.R. Hankins, R.B. Wright, P.L. Wichlacz, Surf:
lnteqace Sci. 11 (1988): p. 371.
13. J.G. Jolley, G.G. Geesey, M.R. Hankins, R.B. Wright, P.L. Wichlacz, Appl.
Spectrosc. 43 (1989): p. 1,062.
14. J.G. Jolley, G.G. Geesey, M.R. Hankins, R.B. Wright, P.L. Wichlacz, App/. Surf
Sci. 37 (1989): p. 469.
15. G.G. Geesey, P.J. Bremer, MTSJ. 24 (1988): p. 36.
16. G.G. Geesey, L. Lang, J.G. Jolley, M.R. Hankins, T. Iwaoka, P.R. Griffiths, Whfer
Sci. Technol. 20 (1988): p. 161.
17. W. Fischer, I. Hanssel, H.H. Paradies, Microbial Corrosion, vol. I (New York, N Y
Elsevier, 1988): p. 300.
18. M. Critchley, R. Taylor, R. O’Halloran, MP 44,6 (2005): p. 56.
19. S.C. Davis, R.M. Mitchell, F.J. Ansuini, International Copper Research
Association Project Final Report, 316, November, 198I.
20. J.E. Castle, M.S. Parvizi, A.H.L. Chamberlain, Microbial Corrosion (Book 303:
Metallurgical Society, 1983): p. 36.
21. B. Little, P. Wagner, J. Jacobus, MP 27, 8 (1988): p. 57.
22. A. Tuthill, “Base Metal Resistance of Alloys to Microbially Influenced
Corrosion,” unpublished report prepared for NiDI, June 1992.
23. G.J.E. Nicklin, “Living with the Threat of Microbially Influenced Corrosion in
Submarine Systems: The Royal Navy’s Perspective MOD.” Ninth International
Naval Engineering Conference and Exhibition (Hamburg, Germany: British
Crown Copyright 2008).
24. C.M. Rangel, M.L. Quinta, M.E. Simas, Microbial Corrosion-I, Conference held
in Sintra, Portugal (New York: Elsevier, 1988): p. 346.
25. V.K. Gouda, I.M. Banut, W.T. Riad, S. Mansour, Corrosion 49 (1993): p. 63.
26. B.J. Little, P.A. Wagner, R.I. May, M.B. McNiell, Marine Tech. Soc. J. 24, 3
(1991): p. 10.
27. J.R. Davis, ed., Metals Handbook, 2nd ed. (Materials Park, OH: ASM
28. J.S. Lee, K.L. Lowe, R.I. Ray, B.J. Little, CORROSION 2003, paper no. 214
29. S.C. Dexter, J.P. LaFontaine, Corrosion 54 (1998): p. 85 I.
30. R. Francis, Galvanic Corrosion-A Practical Guide,forEngineers (Houston, TX:
3 1. F.E. Wells, International Copper Research Association Project Report, 348, March
1984.
32. F.E. Wells, International Copper Research Association Project Report, 348,
October 1989.
33. A.A. West, J. Rogers, J.V. Lee, C.W. Keevil, J.S. Colbourne, P.J. Dennis, K. Smith,
International Copper Research Association Project Report, 401 A, July 1989.
34. J.O. Noyce, H. Michels, C.W. Keevil, J. Hospital Infect. 62 (2006): p. 289.
35. H. Michels, S.A. Wilks, J.O. Noyce, C.W. Keevil, “Copper Alloys for Human
Infectious Disease Control,” Materials Science and Technology Conference
(Pittsburgh, PA: 2005).
36. D.C. Mangum, B.P. Shepherd, J.C. Williams, 3rd International Congress on
Marine Corrosion and Fouling, NBS Special Publication, October 1972.
37. T.J. Lamb, 3rd International Congress on Marine Corrosion and Fouling, NBS
Special Publication, October 1972.
38. F. Inui, Bull. MESJ 1 (1 973): p. 2 I .
39. P.D. Goodman, Brit. Corros. J. 22 (1987): p. 56.
40. R. Francis, M P 21, 8 ( 1982): p. 44.
41. E.B. Shone, G.C. Grimm, Trans. Znst. Marine Eng. 98 (1985): paper I 1.
42. R. Cigna, K. De Ranter, 0.Fumei, L. Giuliani, Brit. Corros. J. 23 (1988): p. 190.
43. D.B. Anderson, B.R. Richards, J. Eng. Power July (1966).
44. A.H. Tuthill, R.E. Avery, S. Lamb, G. Kobrin, CORROSION/98, paper no. 708
January 1991.
46. S. Bond, R. Francis, H. Saxton, BNF Metals Technology Centre Research Report,
R658/12, August 1992.
47. I. Campbell, N.K. Searle, “Comparative Effects of Fouling and Antifouling
Measures on Corrosion,” Fouling and Corrosion of Metals in Seawater
(Dunstaffnage, Scotland: SMBA, April, 1982).
of Copper Alloys and Titanium,” Corrosion in the Power Industry (Montreal,
Canada: NACE International, 1981).
49. S. Sato, K. Nagata, Sumitomo Light M e t Tech. Rep. 19 (1978): p. 83.
SO. S. Sato, K. Nagata, S. Yamauchi, CORROSION/8 I , paper no. 195 (Houston, TX:
5 1. P. Carrera, G. Gabetta, European Federation of Corrosion Publication, EFC 19,
IOM 1996, p. 44.
52. S. Sato, T. Nosetani, Y. Yamaguchi, K. Onda, Sumitomo Light Met. Tech. Rep. I6
(1975): p. 23.
53. W.R. Fischer, D. Wagner, H. Siedlarek, M P 34, 10 (1995): p. SO.
CHAPTER 14
Galvanic Corrosion
In the past, galvanic corrosion has caused numerous failures of equipment in

service, some of them very costly. This chapter describes the causes of gal-
vanic corrosion, the main factors affecting its severity, and some preventa-
tive measures that can be adopted. For further information, the reader is
referred to the work of Francis.'
When a metal is immersed in a conducting liquid, it takes up an elec-
trode potential (also known as the corrosion potential). This is determined
by the equilibrium between the anodic and cathodic reactions occurring on
the surface and is usually measured with reference to a standard electrode,
such as the saturated calomel electrode (SCE).
Galvanic corrosion occurs when two metals with different potentials
are in electrical contact while immersed in an electrically conducting, cor-
rosive liquid. Because the metals have different natural potentials in the
liquid, a current will flow from the anode (more electronegative) to the
cathode (more electropositive) to equalize the potentials. This will increase
the corrosion on the anode (Figure 14.1).This additional corrosion is gal-
vanic corrosion, which is also known as bimetallic corrosion, dissimilar
metal corrosion, or contact corrosion. These terms are used interchangeably
throughout this chapter.
In general, the reactions that occur under bimetallic corrosion are
similar to those that would occur on a single, uncoupled metal, but the rate of
attack is increased, sometimes dramatically. With some metal combinations,
the change in the electrode potential in the couple compared with the
uncoupled potential can induce corrosion that would not have occurred in
the uncoupled state, for example, pitting. In some environments, the change
in potential of the cathode in the couple, compared to the uncoupled state,
can also introduce problems, for example, hydrogen embrittlement (HE).
267
Electrolyte
Material I Material II
FIGURE 14.1 Schematic of a galvanic corrosion cell.
The effect of coupling two metals together increases the corrosion

rate of the anode and reduces, or even suppresses, corrosion of the cathode.
Hence, coupling a component to a sacrificial anode can prevent corrosion;
this is the principle of cathodic protection (CP).
When a metal is corroding, two processes are occurring. One is the
dissolution of metal at the anode, for example, iron:
Fe + Fe2+ 2e + (14.1)
This must be balanced by a cathodic reaction. Most practical cases of
galvanic corrosion occur in solutions containing dissolved oxygen (DO),
and the primary cathodic reaction is the reduction of DO:
0 2 + 2H20 + 4e + 4 0 H (1 4.2)
Under a few circumstances, the cathodic reaction can be the reduction of
hydrogen ions to hydrogen gas:
2H+ + 2e + H2 (14.3)
Under uncoupled corrosion conditions, the anodic and cathodic reactions
occur at small, local areas on the metal surface. In a bimetallic couple, the
cathodic reaction is mostly, or totally, on the electropositive member of the
couple, and the anodic reaction is mostly, or totally, on the electronegative
member of the couple.
14.1 Conditions Necessary for Galvanic Corrosion
The basic requirements necessary to cause bimetallic corrosion are as

follows:
Galvanic Corrosion 269
1. An electrolyte bridging the two metals, which may not be aggressive

to the individual metals when they are not coupled, and which may be
in the form of bulk volume of solution, a condensed film, or a damp
solid such as soil, salt deposits, or corrosion products
2. Electrical connection between the two metals, usually arising from
direct physical contact but also arising where electrical continuity
is established between two metals, for example, by an insulation-
coated conductor, by structural metal work, or by electrical earthing
systems; it is not necessary for the metal junction to be immersed in
the electrolyte
3. A sufficient difference in potential between the two metals to provide
a significant galvanic current
4. A sustained cathodic reaction on the more noble (electropositive) of
the two metals by one of the mechanisms described previously-in
most practical situations, the consumption of DO
14.2 Factors Affecting Galvanic Corrosion

Many factors affect galvanic corrosion, including electrode potential, ca-
thodic efficiency, area ratio, and the nature of the electrolyte.
14.2.1 Electrode Potential

The potential, as defined at the beginning of this chapter, is determined by
many factors, but the only environment for which this is really well docu-
mented is seawater. Unfortunately, the potential, even in a single liquid, is
not a fixed value; rather, it can vary with changes in temperature, liquid flow
rate, the level of aeration, and so on. However, the relative ranking of alloys
remains largely unaffected by such changes, even though the potential may
change by several hundred millivolts for some alloys. Figure 14.2 shows the
electrochemical series in seawater at 10°C (50"F), a typical temperature in
the North Sea. The difference between the most electropositive and the most
electronegative metals is nearly 2 V, and a couple between these materials
generates high currents and, hence, high corrosion rates on the anode.
Three metals are not shown in Figure 14.2: platinum, gold, and sil-
ver. These are used, for example, in electrical contacts, where bimetallic
corrosion can sometimes be a problem. Platinum and gold are similar in
potential to titanium, that is, they are very electropositive, whereas silver is
similar in potential to nickel.
Ni.
Le
I= ICarbon Steel
€3ast Iron
AI-2.7Mg
Zinc
D -1,000 -5
Potential (mV SCE)
FIGURE 14.2 Galvanic series in seawater at 10°C. (Data courtesy of SINTEF and the
author.)
However, the magnitude of the potential difference alone will not nec-
essarily indicate the amount of bimetallic corrosion. For instance, metals
with a potential difference of only 50 mV have shown severe bimetallic
corrosion problems, while metals with a potential difference of 800 mV
have been successfully coupled together. This is because the potential
difference gives no information about the kinetics of bimetallic corro-
sion, which depend on the current flowing between the two metals in the
couple.
Figure 14.2 shows the electrochemical series for seawater at 10°C
but the relative position of the metals is very similar for many near-neutral,
aerated solutions. Hence, this series has applications outside its initial,
anticipated area. It can be seen that copper and its alloys are midway in
the galvanic series and so can be anodes when connected to more noble
(electropositive) metals and cathodes when connected to less noble (elec-
tronegative) metals.
d10y 6251 C-276 €

Uper Stainless Steel 3
Titanium I
316L
Nickel
m 41 Bronze
n9( 1 Copper-Nickel
13Copper
r"l Lead
I Tin
D C .ban Steel
17
Cast ,n
AI-2.7Mg
1 7 luminium
-1 500 -1 000 -500 0 500

Potential (mV SCE)
FIGURE 14.3 Galvanic series in seawater at 40°C. (Data courtesy of SINTEF and the
author.)
14.2.2 Potential Variability

Changes in the corrosion potential of individual metals can result in polarity
changes for some couples. A change in potential can be caused by a change
in the kinetics of the anodic andlor cathodic reaction or a change in the
nature of the reaction, for example, a change from the reduction of DO to
the reduction of hydrogen ions as the cathodic reaction. Potential changes
can be caused by complexing ions; changes in pH, temperature, or velocity;2
and also by intense aeration when it enhances film formation on the hitherto
anodic metal.
Changing the temperature of the seawater can have a significant effect
on the potential of some alloys. Figure 14.3 shows the galvanic series at
40°C. Some metals, such as aluminum, nickel-chrome-molybdenum alloys,
high-alloy stainless steels (SS), and titanium, show significant changes in
potential with temperature, whereas metals like zinc and 90/10 copper-
nickel (UNS C70600) show only small changes. The reasons for this
are mostly connected with the nature of the surface film. For the most
electropositive alloys, the potential decrease with temperature increase is

due to the absence of a biofilm at 40°C which forms in natural seawater at
ambient temperature.
Temperature has little effect on the potential of copper alloys in sea-
water and similar conducting liquids. Small additions of oxidizers, such
as hypochlorite (C103-), produce small increases in the potential of most
copper alloys. Similar to temperature, high water velocities cause only
small changes in potential of copper alloys.2 If the water is deaerated, or
very low in oxygen, then the potential of most copper alloys will move
electronegatively some 200 to 300 mV.
Under some conditions, it is possible to get galvanic corrosion with
a single metal. This occurs when part of a system is replaced with new
metal, while some of the old metal remains. If there is a potential difference
between the old, filmed metal and the new, bare metal, significant galvanic
corrosion may occur before the new metal forms a protective film. One
solution is to clean the old metal to remove the film so that the entire system
starts at the same potential. Corbett reported such a problem for copper pipes
in a fresh water system with a high carbon dioxide (CO2) content.3
14.2.3 Electrode Efficiency

When a metal forms part of a galvanic couple, the rate of the anodic or
cathodic reaction is not the same for all metals. This is because the kinetics
of the reaction vary from metal to metal. The rate of the reaction at the
metal surface determines its efficiency as an anode or cathode. In galvanic
corrosion, a change of efficiency in the cathodic reaction usually has a more
significant effect than a change in efficiency in the anodic reaction.
The efficiency can also be explained in electrochemical terms. Current
flow between the two metals in contact in a couple is accompanied by a
shift in the potential of the anodic member to a more electropositive value
and in the potential of the cathodic member to a more electronegative value.
These changes, calledpolarization, result in the two metals approaching the
same potential, any potential difference between them being equal to the
product of the current and the resistance through the electrolyte. The extent
of polarization depends on both the metal and the environment. In neutral
electrolytes, the cathode is almost always polarized much more than the
anode, which accounts for the fact that a small area of sacrificial anode will
effectively provide protection to a relatively large cathodic area. The extent
of polarization will determine how effective any particular metal may be as
a cathode to drive the corrosion of the anode.
800
600
I
400
c
0
200
v)
>
E
- o
m
z
-200
0
n
400
-600
-800
0.001 0.01 0.1 1 10 100
Current Density (pA/cm2)
FIGURE 14.4 Cathodic polarization curves for high-alloy SS in three types of

~eawater.~
Some metals, such as titanium, are not very efficient at reducing

DO compared with, say, copper alloys, so it is possible for a metal, such
as carbon steel (CS), to corrode more coupled to a copper alloy than to
titanium, despite titanium being much more electropositive than copper.
All copper alloys are efficient cathodes, including the nickel-copper alloys,
Alloys 400 and K-500.
The cathodic efficiency can also change under different conditions.
An example is high-alloyed SS in seawater. Figure 14.4 shows the cathodic
polarization curves for a 6% Mo SS under three condition^.^ Curve A
shows the behavior in natural seawater after the biofilm has formed. The
rest potential is about +300 mV SCE, but under cathodic polarization,
very large currents are produced for relatively small changes in poten-
tial. Hence, the SS with the biofilm is an efficient cathode, and corrosion
rates of the anode in the couple can be very high. Both titanium and the
nickel-chromium-molybdenum alloys behave in a similar manner in this
environment.
Curve B shows the behavior in chlorinated seawater (>0.3 mg/L
chlorine). The cathodic reaction is now mainly the reduction of hypochlo-
rite ions to chloride, instead of the reduction of DO. SS in chlorinated
seawater is not such an efficient cathode as it is with the biofilm. and
FIGURE 14.5 Galvanic corrosion of a NAB tube plate with titanium tubes. (See color
section.)
hence, lower currents are produced at any given potential, compared with
natural (i.e., unchlorinated) seawater, despite the more electropositive rest
potential.
Curve C shows the behavior in seawater when no biofilm forms.
The cathodic reaction is the reduction of DO, but without the biofilm, the
reaction on the SS is very slow, and hence, even lower currents are produced
at any given potential than occur in chlorinated seawater.
An example that demonstrates this is the use of 6% Mo SS or ti-
tanium tubes in heat exchangers with nickel aluminum bronze (NAB)
tube plates. Sometimes galvanic corrosion of the tube plates has been
severe, as shown in Figure 14.5. However, there have been other cases
in which there has been little acceleration of corrosion of the NAB. The
author believes that this is due to chlorination of the seawater. At plants
where little or no chlorination has been carried out, the cathodic reac-
tion would not be polarized, and severe corrosion could occur. At sta-
tions with low-level continuous chlorination, the cathodic efficiency would
be much reduced, as shown in Figure 14.4, and corrosion would be less
likely.
14.2.4 Electrolyte
Electrolyte factors that have a major influence on bimetallic corrosion are
composition, pH, and, in particular, electrical conductivity, which affects
both the intensity and the distribution of corrosion.
The severity of corrosion often increases with increasing electrical
conductivity of the electrolyte because, in practice, high conductivity is
often caused by the presence of aggressive ions, such as chloride, or by
acid or alkali.
The electrical conductivity of electrolytes can vary widely; typical
values are as follows:
Fresh distilled water 0.5-2 pS/cm
Stored distilled water 2-4 pS/cm
Supply water 50-1,500 vS/cm
Seawater 40,000 pS/cm
Saturated sodium chloride (NaC1) solution 250,000 pS/cm
Sulfuric acid (H2S04) up to 800,000 pS/cm
Molten salts 20,000- 1o7pS/cm
Except for very critical items of equipment, such as electronic com-
ponents, and where it is essential to suppress metal dissolution, bimetallic
corrosion is seldom a problem when the conductivity is very low such as in
pure water.
In a solution of low conductivity, or when only a thin film of a highly
conducting electrolyte covers the metals, corrosion is confined to an area
near the junction between the two metals. Although the total amount of
corrosion may be low, it is likely to be highly localized and, therefore,
may be intense. In a highly conducting solution, corrosion will still be
concentrated at the metal junction, but it will be more widespread than
in a solution of low conductivity. Intense corrosion is likely to arise even
with nominally pure water under condensing conditions if the atmosphere
is polluted with acid gases, such as sulfur dioxide (SO2),or with salt, for
example, at industrial or sea coast locations.
In common with local cell corrosion, bimetallic corrosion is sensi-
tive to the presence of constituents in the electrolyte that affect stability
of the corroded metal ions. In some cases, insoluble corrosion products
from the anodic metal can deposit on, and induce crevice corrosion of, the
cathodic member of a couple. For instance, corrosion products from an
electronegative alloy, such as steel or aluminum, could deposit on a copper
cathode and induce crevice corrosion due to the establishment of a copper
ion concentration cell between the metal under the deposit and the freely
exposed metal (see Chapter 6). It is also possible to modify the composi-
tion of the environment by adding inhibitors to control bimetallic corrosion.
This practice has proved effective in the treatment of waters in engines and
recirculating industrial cooling plants. Bicarbonate, cyanide, and tartrate
ions form soluble complexes with copper and zinc, and thus, their presence
increases the rate of corrosion. Other metals and alloys can also have their
corrosion rates accelerated by the presence of suitable organic or inorganic
complexants.
As mentioned previously, in some instances, the cathodic reaction
can become hydrogen evolution. In seawater, this becomes significant at
potentials of -850 mV SCE and more electronegative. This means that
alloys that are susceptible to HE may crack when connected to anodes
that produce these low potentials. Materials like Alloys 400 and K-500
can embrittle when coupled to some aluminum alloys, zinc, or magnesium.
As an example, Efird reported the failure of Alloy K-500 bolts by HE
when used subsea and connected to zinc anodes, which were installed to
protect adjacent C S 5 Other copper alloys are resistant to HE (see Chap-
ter 12).
14.2.5 Area Ratio

The ratio of the exposed areas of the anode and cathode materials in a
couple is very important in the consideration of the likelihood of galvanic
corrosion. The larger the cathode compared with the anode, the more oxygen
reduction, or other cathodic reaction, can occur, and, hence, the greater the
galvanic current. This results in more corrosion on the anode.
Adverse area ratios are likely to occur with fasteners and at joints.
Weld, braze, or rivet metal should be of the same potential as, or, better still,
slightly cathodic to, the base or parent metal. For example, the use of mild
steel bolts or rivets to fasten nickel, copper, or SS alloy plates should be
avoided. Figure 14.6 shows the effect of area ratio on a joint with various
liquid conductivities and potentials.
The effect of area ratio is not confined to general corrosion, but
may also affect other types of corrosion. In Chapter 7, the effect on
the depth of erosion corrosion of 90/10 copper-nickel at high velocity
in seawater by coupling it to other copper alloys was described. A fail-
ure from a North Sea platform also demonstrates this effect. The failure
occurred at the junction of a 6% Mo SS pipe and one made of 90/10
copper-nickel. In chlorinated seawater, this would not normally result in
FIGURE 14.6 Galvanic corrosion in a riveted joint for three electrolytic cases: (A) large
anode area, small cathode area, and high-conductivity electrolyte; (B) large anode area,
small cathode area, and low-conductivity electrolyte; and (C) small anode area and large
cathode area.
any significant acceleration of the corrosion of the 90/10 alloy.' However,

there was turbulence at the junction due to a protruding gasket, and this,
combined with the coupling to SS, induced erosion corrosion. This was
over a small area, and hence, the cathode-anode area ratio was very high
and resulted in rapid penetration of the copper-nickel.6
If the electrical conductance of the electrolyte bridging the bimetallic
contact is low, either because the bulk conductivity is low or because the
electrolyte is present only as a thin film, the effective areas taking part
in bimetallic cell reactions are small, and the total amount of corrosion
is relatively small, although it may be severe immediately adjacent to the
metal junction.
Under immersed conditions in many supply waters, which generally
have a relatively low electrical conductivity, adverse effects are uncommon
if the contacting metals are of similar area. Thus, galvanized steel pipes can
be used with brass or gunmetal connectors, but serious corrosion to the pipe
end is likely to result if the galvanized steel is connected directly to a large
area of copper such as a tank or cylinder. Similarly, SS and copper tubes
FIGURE 14.7 Severe corrosion of a copper pipe entering an SS hot water cylinder. (See
color section.)
can usually be joined without problems resulting, but accelerated corrosion

of the copper tube is likely to occur if it is attached to an SS tank, as shown
in Figure 14.7.
Under immersed conditions in a highly conducting electrolyte, such
as seawater, effective areas will be greater, and severe corrosion may be
encountered on small anodic areas of many metals.
Extremely small anodic areas exist at discontinuities, such as cracks
or pinholes, in cathodic coatings such as copper plating on steel. In seawater,
where electrolyte conductivity is high, it is necessary to apply protection or,
in the case of metal coatings, to specify a thickness of adequate integrity.
Similar considerations apply to pores or defects in a paint coating if the
metal is in contact with a more electropositive metal, that is, the cathode-
anode area ratio is dramatically increased compared with unpainted metal.
This should be avoided if at all possible, as shown in Figure 14.8, which
shows the corrosion on a steel shaft connected to a copper alloy propeller.
The lower line shows that the depth of attack on the bare shaft decreases
rapidly to the uncoupled corrosion rate only 1 m from the propeller. When
the shaft is coated, but there are defects in the coating, the depth of attack
- 2.5
-
E
E
Y
0 2
.d
.
d
2
0 1.5
5Q
D 1
0.5
Distance (m)
IABare Shaft OCoated Shaft I
FIGURE 14.8 Depth of attack of a steel shaft coupled to a copper alloy propeller after
43 days in seawater.
is much greater and remains at a high value over 10 m from the propeller,
as shown in the upper curve.
By correct selection of area ratio, dissimilar metals have frequently
been used successfully together. This is essentially done by making sure
that the less noble anode is large and the more noble cathode is small. Note
that it is not the absolute area of the components that is important, but the
wetted area.
14.2.6 Aeration and Flow Rate

The majority of practical situations involving galvanic corrosion arise in
aqueous solutions under conditions in which the cathodic reaction is the
reduction of DO. As with single-metal corrosion, bimetallic corrosion is,
therefore, partly dependent on the rate at which oxygen can diffuse to the
cathodic surface from the bulk electrolyte.
The galvanic corrosion rate of many copper and nickel-based alloys,
and of SS in seawater, depends on the flow rate of the water as well as the
area ratio. Copper and copper-nickel alloys tend to become more noble,
that is, more electropositive, and corrode less as the flow rate increases. In
280 THE CORROSION OF COPPER A N D ITS ALLOYS
well-aerated, flowing solutions, nickel-based alloys and SS are also likely

to become passive and corrode less than under stagnant conditions.
In neutral electrolytes, complete deaeration will, in many instances,
suppress single-metal and bimetallic corrosion. However, under such anaer-
obic conditions, cathodic depolarization and corrosion can occur if sulfate-
reducing bacteria are present, that is, an alternative reaction to the reduction
of DO is available. In closed, recirculating systems, deaeration is extremely
practical, and very low DO concentrations can be maintained with chem-
ical oxygen scavengers. Theoretically, the corrosion of iron and steel in
a closed, recirculating system should consume the DO fairly quickly. In
practice, there are often places where small quantities of air can be drawn
into the system and so corrosion inhibitors and oxygen scavengers become
necessary.
A car’s cooling system often contains metals such as aluminum, steel,
cast iron, copper, brass, and so on, and bimetallic corrosion does not occur
because commercial antifreeze solutions contain corrosion inhibitors and/or
oxygen scavengers. In once-through cooling systems, it is much more dif-
ficult to control oxygen levels, and 100% deaeration is rarely achieved. In
land-based installations, such as multistage flash (MSF) desalination plants,
deaeration to t 2 0 ppb is routinely achieved. Under these conditions, the
corrosion rate of CS is very low. However, on offshore oil platforms, where
space and weight are at a premium, deaeration of injection water rarely gets
below 50 ppb DO, and levels of 100 or 200 ppb are not uncommon. Under
these conditions, corrosion rates are significant for some metals, such as
CS, and bimetallic corrosion is also a possibility. Figure 14.9 shows severe
corrosion of a CS water box that was coupled to copper alloy heat exchanger
tubes in an MSF desalination plant. The deaeration was not well controlled
and the corrosion of the water box was so severe that it had to be scrapped.
In this case, the copper alloy acted as the cathode.
14.2.7 Metallurgical Condition and Composition

In some cases, differences in corrosion potential can exist between metals
or alloys of nominally the same composition. Subjection to cold working
often tends to make an alloy more anodic. In some alloys, heat treatment
can produce galvanic differences, although this is a rarity with single-phase
copper alloys.
In multiphase alloys, it is common for one phase to be anodic to an-
other. Normally, these differences are small and do not give rise to problems,
FIGURE 14.9 Corrosion of a CS water box coupled to copper alloy heat exchanger tubes
and a tube plate. (See color section.)
but in some instances, severe bimetallic corrosion can occur. A good ex-
ample is NAB (C95800), in which the main alpha phase is anodic to the
kappa I11 phase in seawater. The kappa I11 phase is present in long, thin
stringers (see Figure 2.1), and the relatively large area of the alpha (anodic)
phase means that there is no significant galvanic corrosion. However, when
the pH drops from 8 to 3, the potentials are reversed, and there is now a large
cathode-anode area ratio. Severe corrosion then occurs along the kappa I11
phase, and corrosion rates of >I m d y have been observed in crevices and
shielded areas.7
The Royal Navy has experienced numerous failures of NAB, many
associated with welds. However, commercial vessels have not experienced
this problem. The reason is that naval vessels usually spend a long time
being fitted out, with stagnant and intermittent flow in the seawater system.
Commercial vessels are put into operation as soon as possible and rarely
spend more than the minimum time fitting out. Rowlands showed that deep
pitting along the kappa I11 occurred easily in NAB under intermittent or
very slow flow conditions; however, with a normal flow rate of 3 m/s, a
protective film quickly formed, and there was no ~ i t t i n g . ~
Local changes in composition can also arise at joints made by welding,
both in the weld bead and in the heat-affected zone of the parent metal.
Problems are usually avoided by selecting appropriate filler metals and

welding techniques (see Chapter 18). Because the weld bead has a very
small area compared with the parent metal, it is important that it has the
same potential as the parent metal or, preferably, is slightly electropositive.
Even a reduction of 50 mV in the potential of the weld metal compared to
the parent may cause severe galvanic corrosion in some fluids.
Bimetallic corrosion can occur between alloys of similar type, but
somewhat different composition. Thus, 90/10 copper-nickel can be anodic
to 70/30 copper-nickel under some circumstances.
14.2.8 Bimetallic Corrosion without Physical Contact

Localized corrosion of a metal may give rise to soluble corrosion prod-
ucts, which deposit by a displacement reaction onto a less noble metal
exposed in the same solution and form local, intense bimetallic cells. This
is particularly evident with the more electropositive metals, such as copper,
and problems have been encountered in water heating systems as a result
of copper dissolving from pipe work and depositing onto steel radiators,
aluminum alloy pump impellers, and galvanized steel tanks. Insulation of
joints between dissimilar metal components is not necessarily a safeguard
if cuprosolvent water flows or diffuses from copper to components in sus-
ceptible metals. An example is the filling of aluminum saucepans from the
hot water tap in an area where the water is slightly cuprosolvent (i.e., a low
pH and/or a high free carbon dioxide content). If the copper forms deposits
on the aluminum, it can result in localized corrosion that may eventually
lead to perforation.
Copper can cause corrosion of steel and galvanized steel, particularly
in warm or hot water. Hence, it is usual to specify that copper and its
alloys may be used downstream of galvanized steel, but not upstream, in
supply waters. An interesting example of this occurred in a block of flats
in Denmark with a common heating system. The main feed pipes for the
hot water were in galvanized steel, with copper risers to the individual flats.
This is as recommended, that is, copper after galvanized steel. However, as
the hot water system was not in use all the time, it was possible for copper
corrosion products to descend under gravity into the larger galvanized pipe
during periods of nonuse. The galvanized steel pitted near the takeoff points
due to the copper corrosion products. The solution was to move the takeoff
points to the underside of the galvanized steel feeder pipe so that the copper
corrosion products would no longer enter it.
In plants and equipment that involve recirculated electrolyte, such
as vehicle cooling systems, central heating systems, and industrial heat
exchangers, this type of problem can sometimes be avoided by removing

oxygen from the system or by adding inhibitors, as described previously.
14.2.9 Stifling Effects

In general, the rate of corrosion of coupled and noncoupled metals decreases
with time of exposure. This is partly due to the diminishing rate of diffusion
of oxygen (or whatever the cathodic reactant) through the electrolyte and
through films of corrosion product to cathodic regions, and partly due
to the protection afforded to anodic regions by the corrosion product. To
some extent, therefore, the bimetallic corrosion rate is affected by the
permeability of the corrosion product. In seawater and, to a lesser extent,
in hard supply waters, the alkaline conditions produced at the cathode
result in the formation of calcareous deposits of calcium and magnesium
carbonates, and magnesium hydroxide, which may decrease the rate of
bimetallic corrosion. With steel, for example, corrosion is decreased by
one order of magnitude after less than a year’s exposure. Where corrosion
products are water-permeable or hygroscopic, however, their presence can
sometimes increase the rate of bimetallic corrosion, either because they are
more aggressive than the bulk environment or because they maintain wet
conditions at the metal surface.
14.3 Alloy Groups
When looking at the galvanic compatibility of materials in seawater, it is

found that they can be divided into four major groups, according to their
corrosion behavior, numbered from 1 to 4. Table 14.1 shows the alloy
groups and the common alloys found in each class. Group 1 includes all
the passive alloys, which, by and large, do not usually suffer corrosion in
seawater at around ambient temperature. Some definitions are required to
identify alloy classes. The nickel-chrome-molybdenum alloy class covers
all the nickel alloys with more than 7 wt% molybdenum, such as Alloy
625 and Alloy C-276. Titanium covers not only grade 2 (commercially
pure), but also the common alloys, such as Ti-6A1-4V and Ti-0.15% Pd.
Superaustenitic SS cover all those alloys with a molybdenum content equal
to 6 wt% or greater and having a PREN >40 (for which PREN = %Cr
+ +
3.3x%Mo 16x%N). The superduplex SS cover all duplex alloys with
25% or more chromium and a P E N >40.
In group 2 are the passive alloys that are not totally immune to lo-
calized corrosion (for example, crevice corrosion) in seawater. Such alloys
TABLE 14.1 Alloy groupings for seawater at ambient temperature

Category Type Alloys
1 Noble; passive Nickel-chrome-molybdenum alloys
(Mo > 7%)
6% Mo austenitic SS
Superduplex SS
Titanium and its alloys
2 Passive; not truly Alloy 400/K-500
corrosion-resistant 904L
22% Cr duplex
Alloy 825
Alloy 20
316L
3 Moderate corrosion Copper alloys, Austenitic cast iron
resistance
4 Poor corrosion resistance CS, Cast iron
Aluminum alloys
are mostly SS and iron-nickel-chromium-molybdenumalloys with PREN

values below 40 (for example, Type 316L SS [UNS S316031 and alloy
825). Another type of alloy that falls into this group are the nickel-copper
alloys, the best known of which are Alloys 400 and K-500. These can suffer
crevice corrosion in The common group 2 alloys are shown in
Table 14.1.
Group 3 covers alloys of moderate corrosion resistance that do not
show true passivity. The most common ones are copper and its alloys and
the austenitic cast irons. These cast irons are more noble than conventional
cast irons because of their high nickel content.
Group 4 covers alloys with poor corrosion resistance and has been
subdivided into an iron and steel group and an aluminum alloy group. The
reason for this is that there can be a substantial potential difference between
aluminum alloys and iron and steel in seawater, and so these need to be
treated separately, as shown in Table 14.1.
The principal reason for grouping the alloys as shown in Table 14.1
is that there is ample evidence that bimetallic couples between any of the
alloys in a particular group will not generally give rise to any significant
bimetallic corrosion. This does not necessarily mean that no corrosion will
occur, but no significant additional corrosion will occur. For example, naval
brass will dezincify in seawater, but coupling it to 90/10 copper-nickel will
not significantly increase the rate of penetration. There are a few exceptions
to this rule, but they are not common, and even then, with a suitable area
ratio, bimetallic corrosion need not be a problem. Hence, when there is any
doubt about materials compatibility, the selection of an alloy from the same
group will prevent problems. There are obviously cases in which alloys
from different groups can be connected together safely, and some of these
are discussed in the following sections, but selecting alloys from the same
group is a good rule of thumb when in doubt.
14.4 Galvanic Corrosion in Seawater
A wide range of copper alloys are used in seawater, but the main groups are
the copper-nickels, brasses, tin bronzes, aluminum bronzes, and gunmetals.
The latter category is only available as castings. All these have been used
extensively, and the data for some of the alloys are reported later. However,
there is not much difference between any of the alloys unless selective
phase attack is occurring, for example, alpha-beta brasses and two-phase
aluminum bronzes.
Walleen and Anderson examined couples of 6% Mo austenitic SS
and 90/10 copper-nickel exposed in natural seawater for 1 year.' Although
the corrosion rate increases with area of SS, the corrosion only becomes
significant at area ratios greater than 1:1. However, in chlorinated seawater
systems, the biofilm does not form, and the SS is a less efficient cathode. This
means that area ratios of 10:1 or more can be tolerated without a significant
increase in the corrosion of the copper-nickel.6 This was demonstrated
on a Norwegian offshore platform, where a 90/10 copper-nickel column
pipe from an NAB pump was joined to a 6% Mo superaustenitic SS ring
main. The seawater had a chlorine residual of -0.7 mg/L, and after several
years' operation, there was no significant acceleration of corrosion of the
copper-nickel at the junction."
Two-phase brasses, such as muntz metal and rolled naval brass, are
often used for heat exchanger tube plates with copper alloy tubes. The
dominant form of corrosion is dezincification, and this is countered by
making the tube plates thick (35 mm or more). Coupling to other copper
alloys does not significantly change the rate of dezincification. However,
coupling to 6% Mo superaustenitic SS or titanium showed increases in the
depth of dezincification of 10 to 30 times."-'3 Clearly an alternative tube
plate material or CP is required in such cases.
A copper alloy commonly used in seawater because of its good resis-
tance to corrosion is aluminum bronze. A large number of these alloys are
used around the world, but the most common are discussed in Chapter 2.
Examples are Cu-7% Al, a single-phase alloy; Alloy D, a U.S. alloy used
mostly for condenser tube plates; and NAB, the strongest alloy of this type
and widely used in both cast and wrought forms.
Gehring and colleagues carried out studies on 6% Mo SS and ti-
tanium coupled to aluminum bronze, Alloy D, in seawater, with an area
ratio of SS to bronze of 400:1, as a function of salinity, temperature,
and pollution The results showed that the corrosion rate of alu-
minum bronze (Alloy D) is much less than that of muntz metal, particularly
at low water temperatures. However, at temperatures of about 20°C or
more, the rate of attack was unacceptable. As with muntz metal, the so-
lution is to make the tube plate very thick or to protect it cathodically,
either by impressed current or by the attachment of soft iron or mild steel
anodes.
When aluminum bronze is coupled to titanium or SS, it shows severe
attack in natural seawater and erratic behavior in power stations when used
as a tube plate material in steam condenser^.'^ This may be due to the
presence or absence of a chlorine residual from the antifouling treatment.
As previously discussed, chlorine prevents the formation of the biofilm,
which results in the cathode being far less efficient.
The author investigated this aspect when NAB valves were being
considered for a superduplex SS piping system.IS Panels of superduplex
and cast NAB were bolted together and exposed in natural seawater or
chlorinated seawater (0.7 mg/L chlorine) at about 23°C. Two area ratios
were investigated, that is, SS to NAB of 1: 1 and 10:1. The main problem
was localized attack. In natural seawater, this took the form of both crevice
corrosion and pitting on open surfaces. The pitting was along the kappa I11
phase and hence was deep, up to -0.5 mm after 60 days. In chlorinated
seawater, there was no open surface pitting, and crevice corrosion only
occurred at an area ratio of 10:1.
These results suggest that NAB valves would corrode rapidly in
an unchlorinated SS seawater cooling system, and even in a chlorinated
system, some crevice corrosion might occur. The service experience is
that severe corrosion of NAB valves in superduplex piping systems oc-
curred on both the Woodside Goodwyn A and BP Andrew platforms.
The seawater on Goodwyn A was chlorinated, but the seawater tem-
perature varied from 20 to 40°C that is, higher than in the author's
tests. The seawater on BP Andrew was unchlorinated and about lO"C,
and severe corrosion of NAB was seen in service, as predicted by the
tests.
14.5 Galvanic Corrosion in Fresh Water

There are two main differences between fresh water and seawater, apart
from the much lower total dissolved solids content: conductivity and hard-
ness. The conductivity of a typical fresh water is about two orders of
magnitude less than that of seawater. This means that bimetallic corrosion
tends to be confined to the point of contact of the two metals. It also means
that galvanic corrosion is often less severe than in seawater, unless the area
ratios are very different or the potential difference is very large. The exam-
ple of copper and SS was discussed previously, where copper fittings with
SS pipe have been very successful, but a copper pipe entering an SS tank
was badly corroded.
Seawater is very hard because of the high magnesium and sulfate
contents. Fresh waters are usually subdivided into soft, moderately hard,
and hard, for which the measure of hardness is the total hardness, expressed
as calcium carbonate (CaC03). A soft water is usually from 0 to 50 mg/L
total hardness, a moderately hard water is from 50 to 150 mg/L, and a hard
water has a total hardness > 150 mg/L. In fresh waters, much of the hardness
is often present as calcium bicarbonate (Ca[HC03]2), and many hard and
moderately hard waters tend to deposit calcium carbonate scale with time.
The scale inhibits the access of water to the metal surface, slowing down
corrosion. This is the reason that CS, galvanized steel, and cast iron can
last for many years in hard waters.
Soft waters do not usually deposit scale, but they also generally have
lower dissolved solids contents, including chlorides, and so corrosion rates
in soft waters are not necessarily much higher than in hard waters. One
factor that is important in soft waters is pH. Even if the pH is quite high (for
example, pH S S ) , there is little buffering action because the bicarbonate
content is low and the creation of hydrogen ions, for example, at a corrosion
site, can result in a large local decrease of pH.
Linder and Mattsson conducted extensive galvanic corrosion tests
both at room temperature (25°C [77"F]) and at 75°C ( I67"F).l6 Most of the
tests were in a water of intermediate hardness (100 mg/L as CaC03), but
additional tests were also carried out in a soft water (hardness = 2.5 mgL)
and a hard water (hardness = 245 mg/L). During these tests, they investi-
gated both coupled and uncoupled potentials as well as galvanic currents
for a range of materials commonly used in fresh water systems.
From these data, they produced galvanic series for both hot and cold
moderately hard water, which are shown in Figures 14.10 and 14.1 1. The
Stairiless Steel
Copper
90/10 Copper-Nickel
Al-Brass
DZR Brass
m Carboi steel
Cast Irc
Aluminium
-
Zinc
-800 -600
- -400
Potential (mV SCE)
-200 0 200
FIGURE 14.10 Electrochemical series for some common metals in a moderately hard
fresh water at 25°C.’6
heading “stainless steel” covers the four types they tested, that is, 13% Cr
and 18% Cr ferritic and Types 304 and 3 16 austenitic SS (UNS S30400
and S3 1600). There were no significant differences between these al-
loys, but there were no tight crevices on the samples and the maximum
chloride content was only 75 mg/L, with most of the data generated at
11 mg/L chloride. Higher chloride levels ( A 0 0 mg/L) occur in some fresh
waters, and these might cause crevice corrosion with these alloys, particu-
larly at higher temperatures. The silver solder tested was a copper-silver-
phosphorus type, while the soft solder was a tin-silver alloy recommended
for potable waters (see Chapter 18). The names of the other alloys are
self-explanatory.
Both series show similar potentials for the SS and most of the copper
alloys, with 60/40 brass and the soft solder being slightly more electroneg-
ative. The order of the other alloys is identical to that in seawater at 25”C,
although the potential values are different. However, at 75”C, the relative
positions of zinc and aluminum were reversed.
Sta nless Steel
Copper
90110 Copper-Nickel
Al-Brass
DZR Brass
Gunmetal
Silver Solder
60140 B a s s
Soft Solder
0Carbon Sl
1st Iron
-800 -600 -600 -200 0 260
Potential (mV SCE)
FIGURE 14.11 Electrochemical series for some common metals in a moderately hard
fresh water at 75"C.16
The corrosion currents that were measured did not always reflect
the position of the alloys in the two series. From the corrosion data, the
metals have been arranged into groups in a similar manner to that done for
seawater in Section 14.4. The different groups are shown in Table 14.2 for
the moderately hard water at 25°C.
Although Figure 14.10 shows similar potentials for copper alloys and
the SS, it is well known that copper can suffer severe galvanic corrosion
when coupled to large areas of SS, as described previously. Hence, the
copper alloys have been put into a separate group, while acknowledging
that under the right circumstances, they can be coupled to S S . Linder and
Mattsson observed some galvanic currents for the soft solder coupled to
copper, although service experience shows few corrosion problems with
solders at copper joints. Hence, soft solder has been put into a subsection
of group 2, meaning that problems may arise in some circumstances.
TABLE 14.2 Grouping of compatible materials in

moderately hard fresh water at 25°C
Group Type Alloys
1 Passive Austenitic SS
(304, 3 16)
Ferritic SS
(13% Cr, 18% Cr)
2 Good corrosion Copper and its alloys
resistance Silver solder
Soft solder (Sn-Ag)
3 Moderate Cast iron
corrosion cs
resistance
4 Low corrosion Aluminum
resistance Zinc
Cast iron and CS have been grouped together because they show sim-
ilar potentials. When coupled together at 25"C, a little corrosion occurred
on the cast iron.
Zinc and aluminum have been grouped together because they behaved
similarly when coupled to other metals. The potential of aluminum was
more electropositive than the potential of zinc, but when the alloys were
coupled together, it was the aluminum that became the anode, although the
current was small. The reason for this is believed to be that in the couple,
the presence of Zn2+ ions modifies the potential of the aluminum, moving
it electronegatively.
We can now see what effect increasing the temperature to 75°C has on
Table 14.2. There are two effects, one of which is that soft solder becomes
more anodic to copper than at 25°C and cast iron becomes cathodic to
CS. The higher than expected galvanic current when CS and cast iron
were connected together was probably due to the graphite in the cast iron
increasing its cathodic efficiency. The other effect was that zinc became
more electropositive than aluminum, with the result that there was more
corrosion of the aluminum when it was coupled to zinc.
In soft water at 25"C, there were no significant differences in corrosion
behavior compared with the moderately hard water. At 75"C, the grouping
of alloys did not change, but the currents between incompatible metals
were higher than for the moderately hard water, and therefore corrosion
rates were higher. This may have been because the pH was not so well
buffered and decreased more in the soft water during the exposure tests.
Alternatively, it may have been that some scale was being deposited on the
cathodes in the moderately hard water, which reduced the corrosion rate.
Soft waters generally show no real scale-forming tendency.
In hard water at 25"C, the corrosion between incompatible metals
(i.e., from different groups) was much less than in the moderately hard
water. Aluminum was less anodic to copper and SS and was cathodic to
zinc. Zinc was less anodic to CS, which was itself less anodic to SS.
Only aluminum coupled to CS gave large corrosion currents. The reduc-
tion in corrosion was probably due to the formation of calcium carbon-
ate scale on the cathode and the buffering of the pH by the bicarbonate
ions.
When the temperature of the hard water was increased to 75"C, alu-
minum became anodic to zinc and corroded substantially when coupled
to any other alloy. Couples between the other alloys were similar in be-
havior to that at 25°C but generally, there was less corrosion, presumably
because of the increased scale-forming tendency of the hard water at high
temperatures.
14.6 Galvanic Corrosion in the Atmosphere

The most important factor in atmospheric corrosion is water. Corrosion in
direct contact with air at ambient temperature is generally not significant
from an engineering point of view. Therefore, water must condense on
the exposed metal surfaces or precipitate as rain. This will be followed
by evaporation as the components dry out, and thus, a most important
parameter for atmospheric corrosion is the time of wetness. Condensation is
controlled by the relative humidity, and it is generally accepted that a relative
humidity of 80% or more and a temperature greater than 0°C are required
for significant condensation. l 7 Rainfall is dependent on the local weather
and can vary greatly, even in a small country like the United Kingdom.
Corrosion in the atmosphere is generally not so great as corrosion
fully immersed in water. However, galvanic corrosion rates can be as high
as those under immersed conditions.
As the water evaporates, anything dissolved in it will concentrate.
Hence, when these species are aggressive to a metal, the corrosion rate will
increase as the water evaporates. The water films formed in the atmosphere
are generally thin, and rapid diffusion of oxygen can occur. Hence, there
is a ready supply of oxygen that can be reduced as the cathodic reaction.
Ammonia (NH3) is common in rural areas and is readily soluble, causing

increases in pH. This is beneficial for most alloys, although stress corrosion
cracking of some copper alloys is possible. Sulfur and nitrogen oxides are
common in industrial and urban areas as effluent gases, for example, fossil-
fuelled power stations. All these gases dissolve readily, forming strong
acids, which are the cause of acid rain.
In addition to these gases, two other factors affect atmospheric cor-
rosion: temperature and chlorides. Temperature is self-explanatory: the
warmer the climate, the greater the rate of chemical reactions, includ-
ing corrosion. Chloride is found adjacent to the sea, and the closer a site
is to the sea, the higher the chloride concentration will be in condensed
water.
Using all these factors, sites are generally divided into four types
by atmosphere: rural, urban, industrial, and marine. Rural sites have low
chloride and sulfur dioxide and may have ammonia. The urban atmosphere
has some sulfur dioxide, whereas industrial areas have high sulfur dioxide
and NO, content. Finally, the marine atmosphere is characterized by high
chloride content.
The cathodic reaction in atmospheric corrosion is usually the reduc-
tion of DO, and as described previously, the reaction rates can be high,
especially with thin water films. If the dissolved solids and gases content of
the water is low, the conductivity will be low, reducing the rate of corrosion,
but as the films evaporate, the conductivity will increase.
Galvanic corrosion in the atmosphere tends to be confined to the dis-
similar metal joint where the condensation has occurred. Hence, area ratios
are usually around 1:1. From the preceding discussion, it can be deduced
that atmospheric galvanic corrosion is similar to galvanic corrosion when
immersed in a low-conductivity electrolyte. The corrosion rate obviously
varies as the species dissolved in the water vary, and as the water films
evaporate and condense again. This is not readily simulated in the labora-
tory, but there is a body of data from long-term exposure trials of galvanic
couples that is very useful.
Kucera and Mattsson published bimetallic corrosion data for a variety
of atmospheres. In addition, they made an attempt to categorize the cor-
rosion in terms of engineering risk, and the data are summarized below."
Table 14.3 shows the atmospheric corrosion rates for 1 1 commonly used al-
loys. Weathering steel is a low-alloy CS with -0.5% copper. This produces
an adherent rust layer that can lower atmospheric corrosion rates compared
to CS. In addition, aluminum was tested in the anodized condition, as this is
commonly used to reduce aluminum corrosion. Three types of atmosphere
TABLE 14.3 Corrosion of some common metals in

various atmospheric environmentsL8
Corrosion rate (prn/y)
Alloy Rural Urban Marine
Magnesium 6.5 11 14
Aluminum 0.1 0.5 0.6
Anodized aluminum 0 0.2 0.2
Zinc 0.5 2.5 1.3
Carbon steel 22 53 36
Weathering steel 22 48 32
Tin 1.4 2.7 10
Lead 0.3 0 0
Copper 1.2 1.4 3.2
Nickel 0.4 2.7 1 .o
Stainless steel (Type 304) 0.03 0.04 0.04
are reported: rural, urban, and marine. Some metals suffered little corrosion
in any of the environments, for example, lead and SS. However, while the
corrosion rate of some metals is low in rural atmospheres, it is much greater
in urban and/or marine atmospheres. Zinc is clearly more susceptible to
acid sulfur gases than chlorides, that is, it is more susceptible in urban,
rather than marine, atmospheres, whereas tin and copper corroded most in
marine atmospheres.
Kucera and Mattsson tested many, but not all, of these metals in com-
bination, and the results for marine atmospheres are summarized in Table
14.4.'* The data from urban and rural areas are similar to those in Table
14.4 and may be found in the work of Francis.' The aim has been to show
couples with no significant change in corrosion rate, a slight increase in cor-
rosion rate, and a large increase in corrosion rate. This has to be compared
with the uncoupled corrosion rate in Table 14.3, as an increase in corrosion
rate may still not represent much loss of metal and may be acceptable
for a particular application. There are no absolutes here, and some metal
combinations may be acceptable in one case and not in another, depending
on the criticality of the application. Chromium is included in Table 14.4
because of its widespread use as electroplating in atmospheric applications.
With many metals used uncoated in the atmosphere (steel, zinc, and
aluminum), copper is the cathode. Being an efficient cathode, the corrosion
rate of the anode is accelerated, but not often dramatically, as the effective
area ratio is usually 1:1.
It might be thought that coupling copper, or one of its alloys, to SS
may result in accelerated corrosion of the copper. This is not so because
TABLE 14.4 Risk of galvanic corrosion in a marine

atmosphere for some common metals'x
(See color section.)
Anodised aluminium
Carbon steel
Weathering steel
,
Stainless steel
No risk of galvanic corrosion Large increase in corrosion
Some increase in corrosion 0 No data
the effective area ratio is usually 1:1, and SS is not a very efficient cathode.
This was why SS was used to replace corroded steel internal supports of
the copper skin on the Statue of Liberty.'"''
The following sections review the main methods of preventing and combat-
ing galvanic corrosion. To do this properly, it is necessary to know as much
as possible about the environmental conditions. With more information,
more confidence can be placed in the decisions made to mitigate galvanic
corrosion.
14.7.1 Materials
Where the design is being completed before any materials are purchased,
there is the maximum opportunity to select galvanically compatible metals
for the specific application. The following guidelines offer some helpful
suggestions.
Use compatible metals wherever possible. Try to select alloys from
the same group, as shown in Tables 14.1 and 14.2. Avoid metals that are
far apart in the electrochemical series, unless it is well documented that
they can be coupled successfully, for example, 90/10 copper-nickel and
high-alloy SS in chlorinated seawater.
When welding, pay particular attention to the condition of the weld
after fabrication. It is desirable that it be cathodic to the parent metal.
When carrying out autogenous welding, choose welding parameters that
will achieve this. When using filler metal, select a filler that produces weld
metal that has a similar potential or is electropositive to the parent metal.
With such a high cathode-anode area ratio as occurs in a weld, a weld metal
potential only 50 mV negative to that of the parent metal can result in severe
corrosion in some fluids.
Where compatible materials cannot be used, for whatever reason, one
of the following methods of prevention of galvanic corrosion should be
considered.
14.7.2 Insulation and Separation

If the anode and cathode can be prevented from coming into electrical
contact, then no galvanic corrosion will occur. In some situations, this can
be done quite conveniently, while in others, it is impractical.
Several methods for achieving this when bolting two flat surfaces
together were discussed by Parkins and Chandler, depending on whether it
is the bolts or one of the plates that is the anode.'q2'
Another method of preventing galvanic corrosion between two adja-
cent metals could involve the use of a thick coat of nonconducting, insoluble
mastic at a joint. The mastic may additionally contain chemicals to inhibit
corrosion of the metals as they are leached out.
One method for preventing galvanic corrosion of piping at dissimilar
metal joints is the isolating flange. This uses a nonconducting insert between
the two metal flanges and nonconducting sleeves and washers for the bolts.
This means that the two metals are no longer in electrical contact. However,
there are two major disadvantages to this technique. First, the sleeves and
washers have to be undamaged and correctly assembled for the isolation
flange to work properly. On many industrial sites, sufficiently skilled labor
is not available, and this is difficult to ensure. Second, there is frequently
a requirement by electrical engineers to earth all major items of metal
equipment for safety reasons. This has the effect of shorting across the
isolation flange and, unfortunately, is a common industrial practice.22Pipe
hangers or supports can also act as inadvertent electrical shorting between
adjacent pipe runs. Where isolating flanges are used it is important that the
resistance between the two components be measured prior to start-up, to
ensure that isolation has been achieved.
An alternative method of separating the anode and cathode when one
is a pipe and the other is a pipe, vessel, valve, etc., is to use an insulated spool
piece. This is a length of pipe between the two items that is either internally
coated or is made from a nonmetallic, for example, fiber-reinforced plastic
(FRP/glass-reinforced plastic). Note that the coating must extend up onto
the flange faces as well as along the pipe bore. The metal pipe being coated
must be made of the same metal as that forming the cathode so that defects in
the coating do not give rise to rapid localized corrosion. If a coating is being
used, it should have a life consistent with the planned maintenance intervals
for the plant. A coating that needs replacing at more frequent intervals is
often not cost-effective, and an alternative solution to the galvanic corrosion
problem may be required.
The length of the spool piece depends on the electrochemical prop-
erties of the two metals and the particular fluid under consideration. An
example for seawater serves to demonstrate this principle. When joining a
copper alloy, such as 90/10 copper-nickel, to a group 1 alloy, it is common
to use a spool piece approximately six pipe diameters long, provided that
there is no undue turbulence to cause localized erosion corrosion of the
copper alloy. BP Tankers had an oil tanker with 90/10 copper-nickel piping
and a titanium-plate heat exchanger. Galvanic corrosion occurred on the
copper-nickel adjacent to the heat exchanger, which was solved by fitting a
coated titanium spool piece five diameters long. The length of the spool was
limited by the available space in the ship, but it prevented further corrosion
of the copper-nickel pipe.
Bardal et al. showed that with high-alloy SS and 90/10 copper-nickel
in chlorinated seawater, a spool piece is unnecessary.' However, if the
chlorination were likely to be turned off at all, a spool piece would be a
wise precaution to prevent corrosion at the dissimilar metal joint. In natural
seawater, the corrosion rate of 90/10 copper-nickel was -1.2 mm/y at the
junction with an SS pipe.
14.7.3 Paints and Coatings

When trying to prevent a bimetallic corrosion problem, it is sometimes
possible to use paints or coatings on the affected area. Preferably, both the
anode and cathode should be coated, but it is often sufficient just to coat
the cathode. A small defect in the coating will still mean that only a very
tiny cathode area is exposed to the seawater. On no account should only the
anode be coated. The cathode-anode area ratio will be very high at defects
in the coating, and the rate of corrosion at the anode will usually increase
dramatically.
It is important to select coatings that have a life similar to that of the
components they are protecting, especially if access is limited or difficult.
Otherwise, regular maintenance of the coating to prevent galvanic corrosion
will be necessary as the coating breaks down.
An alternative method to organic coatings is metallic coatings. The
most commonly applied metals are aluminum and zinc, although many
other metals can be used as coatings. The most common, and cheapest,
method is thermal spraying. However, this produces porous coatings, and for
maximum life, polymeric sealers are required after coating. An alternative
is to use high-velocity oxy fuel (HVOF) spraying, which produces very
dense coatings with no need for sealing, although this process is a little
more expensive.
Note that where a metallic coating is porous and no sealer is used, it is
essential that the coating be anodic to the base metal; otherwise, accelerated
corrosion will occur at the defects in the coating due to the high cathode-
anode area ratio. An example of a poor choice of coating is chromium
plating on steel, which is unsuitable in seawater.
For the case of atmospheric corrosion only, some additional coating
measures can be considered. These all rely on the galvanic corrosion being
confined to the joint area with an anode-cathode area ratio of approximately
1:l. One solution is to use a nonconducting mastic at the junction, which
will, effectively, increase the resistance between the anode and cathode. A
similar effect can be obtained with polymeric tapes, which may contain
inhibitors that can be leached out. The tape is usually applied up to about
25 mm to either side of the bimetallic joint. Rubber coating has also been
used successfully to prevent atmospheric galvanic corrosion.
14.7.4 Cathodic Protection

This method of protection relies on changing the coupled potential of two
dissimilar metals. It is not always necessary to suppress corrosion totally,
but only to prevent or reduce the corrosion due to the galvanic couple.
Two methods are used: impressed current and sacrificial anodes. With
the former, a current sufficient to change the potential to the desired value
is applied. With a sacrificial anode, the final potential is usually close to the
open circuit potential of the anode, and so the metal chosen for the anodes
is important. Not only must the potential be in the desired range, but the
anode must also be of a size and have a life and a cost that are practical.
Some examples will show this. In tube and shell heat exchangers,
using seawater for a coolant, muntz metal or rolled naval brass is com-
monly used as a tube plate material with 90/10 or 70/30 copper-nickel
tubes. The brass tube plates dezincify, but the tube plates are usually thick
(235 mm), and so this may be ignored. In more critical applications, sac-
rificial anodes have been attached to the tube plates. Zinc and aluminum-
indium are common alloys that are readily available. However, the open
circuit potential of these alloys is about -1 V SCE. If a large number of
anodes are fitted, overprotection results, which can give rise to unexpected
problems. A heat exchanger at a Spanish power station was protected in
such a manner, and the low potential caused the precipitation of a heavy cal-
careous scale in the tube inlets. This increased inlet end turbulence, which
caused some erosion corrosion of the tubes, which was exacerbated by the
presence of sulfides.
The solution is to use soft iron or mild steel (i.e., low carbon) anodes.
The potential is now too electropositive for significant calcareous scale to
deposit, whereas it is sufficiently low to reduce dezincification of the tube
plates. An added benefit is that the supply of iron corrosion products from
the anodes aids in the formation of protective films on the copper alloy
tubes.
A similar principle can be applied in piping systems by fitting a
sacrificial spool piece, usually of CS, five to six pipe diameters long between
the two components of the bimetallic couple. In seawater, the spool piece
requires replacement every 1 or 2 years, but if access is easy, this can be
a cost-effective solution. Remember that iron corrosion products from the
spool piece will be carried downstream, and this must be acceptable in
the rest of the system. The corrosion products usually form soft deposits
of hydrated iron oxides up to -1 mm thick, which may not always be
desirable.
14.7.5 Inhibitors
Inhibitors are rarely used in once-through seawater systems, although they
may be used in recirculating closed systems or in seawater injection sys-
tems in the oil and gas industry. Note that ferrous sulfate (FeS04.7H20),
sometimes added to improve the corrosion resistance of copper alloy heat
exchanger tubes, is not an inhibitor. The iron hydroxide (FeOOH) deposit
that is produced acts as a barrier between the turbulent water and the pro-
tective film on the metal surface, as described in Chapter 3.23
Where inhibitors are used, it is important to select one that is com-
patible with all the metals in the system.
In Section 14.7.3, the use of mastics and tapes to prevent atmospheric
galvanic corrosion was discussed. In some cases, these can be obtained
with inhibitors to provide additional protection against galvanic corrosion.
As with fully immersed conditions, it is important that suitable inhibitors
be used that are compatible with both metals in the couple being protected.
A slightly different method of inhibition is deaeration, which can
usually only be applied to recirculating systems. Provided that hydrogen
evolution cannot be the cathodic reaction, the addition of oxygen scavengers
to prevent the cathodic reduction of DO is a viable method of preventing
corrosion, including galvanic corrosion. It is essential to prevent additional
entry of oxygen into the system or regular dosing with a scavenger will be
necessary.
14.7.6 Design
If it is addressed early enough in the design stage, galvanic corrosion can be
prevented by one of the methods described previously. Once construction
has started, changes and retrofits become much more difficult and expensive.
A range of good practices should be adopted to minimize the risk of
galvanic corrosion. For structures exposed to the atmosphere, one of these
is to prevent water accumulation and the entrapment of debris. Designs that
allow water to run off freely pose fewer problems than those that create
areas where water can collect (see Chapter 16).
Wherever possible, for fully immersed conditions, a design should be
chosen to minimize the area of the cathode and/or maximize the area of the
anode. If the cathode-anode area ratio is kept sufficiently small, additional
methods of restricting galvanic corrosion may not be necessary.
Finally, one further factor is essential to the implementation of a good
design: an understanding by the work force carrying out the construction,
fabrication, and so on. If the workers are told the importance of what they
are doing and why it must be done a particular way, then they will be more
likely to implement the design correctly. Any design to prevent galvanic
corrosion must include communication with the work force of the necessary
precautions required during assembly.
References
2. T.S. Lee, E.W. Thiele, J.H. Waldorf, M P 23, 11 (1984): p. 44.
3. R.A. Corbett, MP 36, 12 (1999): p. 63.
4. T. Rogne, U. Steinsmo, “Practical Consequences of the Biofilm in Natural Sea
Water and of Chlorination on the Corrosion Behaviour of Stainless Steels,” in Sen
Water Corrosion of Stainless Steels-Mechanisms and Experiences, EFC
Publication 19 (London, U.K.: Institute of Materials, Minerals, and Mining,
1996): p. 55.
5. K.D. Efird, M P 24,4 (1985): p. 37.
6. E. Bardal, R. Johnsen, P.O. Gartland, Corrosion 40 (1984): p. 12.
7. J. Rowlands, “Studies of the Preferential Phase Corrosion of Cast Nickel
Aluminium Bronze in Seawater,” Eighth International Congress on Marine
Corrosion and Fouling (Mainz, Germany: 1981).
8. B. Walleen, T Anderson, “Galvanic Corrosion of Copper Alloys in Contact with a
Highly Alloyed Stainless Steel in Sea Water,” Proc. of the 10th Scandinavian
Corrosion Congress (Stockholm, Sweden: I986), p. 149.
9. R. Francis, Weir Materials and Foundries Technical Report, TN1099, July 1994.
10. 0. Strandmyr, 0. Hagerup, CORROSION/98, paper no. 707 (Houston, TX: NACE
11. G.A. Gehring, J.R. Maurer, CORROSION/81, paper no. 202 (Houston, TX:
12. G.A. Gehring, R.J. Kyle, CORROSION/82, paper no. 60 (Houston, TX: NACE
13. G.A. Gehring, “Inlet End Corrosion Problems in Condensers,” Proc. of the
Seventh Annual Conference on Materials for Coal Conversion and Utilization
(Gaithersburg, MD: National Institute of Standards and Technology, 1982).
14. B. Case, J.O. Davies, W.E. Heaton, J.M. Sketchley, “Assessment of the Corrosion
of Tube Plate Materials in Titanium Tubed Condensers,” Proc. of the International
Colloquium on Choice of Materials for Condenser Tubes and Plates and Tube and
Tightness Testing (Avignon, France: SFEN, 1982), p. 17 1.

16. M. Linder, E. Mattsson, “Galvanic Corrosion of Combinations of Metals in Fresh
Water,” Proc. of the Seventh Scandinavian Corrosion Conference (Trondheim,
Norway: 1975),p. 19.
17. E. Mattsson, Basic Corrosion Technologyfor Scientists and Engineem, 2nd ed.
(London, U.K.: Institute of Materials, Minerals, and Mining, 1996): p. 75.
18. V. Kucera, E. Mattsson, “Atmospheric Corrosion of Bimetallic Structures,” in
Atmospheric Corrosion, W.H. Ailor, ed. (New York, NY: John Wiley, 1982): p.
561.
19. R. Baboain, E.B. Kliver, The Statue of Liberty Restoration (Houston, TX: NACE
20. R. Baboain, MP 35, 11 (1996): p. 5.
21. R. Parkins, M. Chandler, Corrosion Control in Engineering Design (London,
U.K.: DOVHMSO, 1978).
22. R. Johnsen, S. Olsen, CORROSION/92, paper no. 397 (Houston, TX: NACE
23. J.E. Castle, M. Parvisi, “Minor Alloying Elements and the Marine Corrosion of
Copper Alloys,” in Proc. of the International Colloquium on Choice of Materials
for Condenser Tubes and Plates and Tube and Tightness Testing (Avignon, France:
SFEN, 1982), p. 171.
CHAPTER 15
Underground Corrosion
Corrosion in soils is complex and depends on many variables. A wide range

of corrosivity is also possible in soils; failures of buried components have
occurred within a year, while archaeological samples have been preserved
for hundreds of years. Although a wide range of equipment may be buried,
the most common items are pipes (for water, sewage, gas, etc.), cables
(electricity and telecom), tanks, pilings, and foundations. These are often
required to last for periods of 50 to 100 years. Copper and copper alloy
artifacts that are thousands of years old have been recovered from archae-
ological sites. This shows that given the right soil conditions, copper can
resist corrosion for long periods of time.
Soil, essentially, has three components: particles, which may be in-
organic (clay, sand, etc.) or organic (peat, muck,(') etc.); moisture, which
includes the solids dissolved in the water; and gas. Both nitrogen and oxy-
gen can diffuse into soils from the air, but other gases may be produced
by local organic activity, for example, carbon dioxide (CO2) and hydrogen
sulfide (H2S). The properties of the soil that influence corrosion are as
follows:
1. Soil moisture includes the dissolved solids and varies seasonally

2. Oxygen, the quantity of which depends on the ability of the gas to
penetrate the soil
3. Redox potential, generally determined by the oxygen content of the
soil
4. pH, influenced by the dissolved solids and any chemicals being pro-
duced by local organic activity
(I) Muck is decaying vegetation, nor manure.
303
5. Conductivity, which affects corrosion in a similar way to liquids, that

is, the more conductive the soil, the greater the corrosion current
6. Microbial activity, by which a range of chemical by-products may be
produced that influence corrosion, for example, H2S
Soil conditions can vary considerably, even over a small distance, and this
can cause large changes in corrosivity. When corrosion in soils is severe,
it is similar to aqueous corrosion in seawater or fresh water; when corrosion
is mild, it is similar to atmospheric corrosion. In underground corrosion, it
is common for the anode and cathode of a corroding metal to be adjacent,
as in aqueous corrosion. However, under certain conditions, they may be
more widely separated, and in extreme cases, they may be a kilometer or
more apart.' In such cases, any corrosion products tend to form between
the anode and cathode, possibly causing the anode to pit and corrode at a
high rate.' This is also true of galvanic corrosion underground, although
the corrosion is more commonly found at the couple junction.
Although conductivity is used when assessing aqueous solutions, it is
more common to use resistivity when considering underground corrosion.
This is the inverse of conductivity, that is, 400 pS/cm = 2,500 Qcm. Typi-
cally, soils with a resistivity below 1,000 Qcm are regarded as aggressive,
whereas those with a resistivity >5,000 Qcm are mildly corrosive.
To assess the corrosivity of a soil, a number of methods have been
proposed. One of the best known is that used by the German gas industry
for assessing the risk of corrosion of buried pipelines.2 This involves an
index calculated from 10 parameters of the water and soil, and the more
negative the index, the more corrosive the soil.
Stray currents from electrical conductors are well known to affect
corrosion underground. This is a specialist subject and outside the scope of
this book. For more information, the reader is referred to specialist texts on
the subject, while Mattsson gives an introduction to the subject. '
Note that when equipment is buried and the soil is replaced, the con-
ditions around the buried item can be very different to those in the adjacent,
undisturbed soil3 The moisture and oxygen content around the buried item
can be much higher, leading to greater corrosion than would be expected
from general testing of the soil.
15.1 Long-Term Exposure Tests
Between the 1930s and 1950s, several long-term exposure tests were con-
ducted on copper and its alloys buried at a variety of sites in both the
Underground Corrosion 305
United States and United Kingdom. The U.S. National Bureau of Stan-
dards (NBS) data have been summarized by Romanoff, and the highlights,
relating to copper, are presented here.4 The exposures varied from 8 to 14
years in a variety of soils, some of whose properties are listed in Table 15.1,
while the corrosion data are summarized in Table 15.2. The NBS study
used multiple specimens of both phosphorus-deoxidized and tough pitch
coppers. There was no significant difference between the corrosion rates
for the two coppers, and so the average rate for all the samples has been
quoted.
The resistivity figures in Table 15.1 show a very wide range, as does
the drainage. There was more corrosion at poorly drained sites, whereas the
corrosion at well-drained sites was slight. The annual rainfall varied from
750 to 1,500mm at most sites and appeared to have little effect on corrosion
within this range. It should be remembered that at sites with heavy rainfall,
much of it may fall in a few days and mostly run off the surface, so that the
subsoil is comparatively dry for much of the year.
Any effects of temperature are masked by the more marked effects
of the other conditions. Very high sulfate (>300 ppm) and/or chloride
(> 100 ppm) content is associated with poor drainage and usually occurs
in soils containing considerable humus. The combination appears to be
responsible for most of the worst cases of corrosion (sites 43, 63, 60, and
64). The rate of corrosion, however, bears no direct relation to the sulfate or
chloride content. Soils with moderate sulfate and/or chloride content still
cause appreciable corrosion when the drainage is bad.
The most acidic soils (pH 5 4.0) all have considerable sulfate content,
and generally chloride, too, so the comments in the preceding paragraph
apply to these soils as well. Of the alkaline soils (pH >8.0), the most
corrosive was soil 67, with very poor drainage (category 4 in Table 15.1)
and a moderate sulfate content, possibly because this was associated with
a particularly aggressive sulfide. The pH figures appear to give only a very
rough guide to soil corrosivity, but it must be remembered that the NBS
report states that some of the pH values may be incorrect, as they were
liable to change on exposure to the air.4
Soils with low resistivity (< 100 Stcm) are always very corrosive, but
the worst corrosion occurred with soils 67 and 60, which had resistivities in
the range 200 to 500 Qcm and poor drainage. Other sites with resistivities
in this range had less corrosion but also had low rainfall and drainage,
varying from poor (category 3) to good (category 1). It should be noted that
the resistivities are quoted for saturated soils at 16°C and that these soils
would have higher resistivities for most of the year. A better correlation
TABLE 15.1 Summary of soil conditions during NBS exposure tests4
Soil Cornp. of water (pprn)
No. Type Mean temp. (“C) Ann. rainfall (rnm) DrainagecA) pH Resistivity (Qcrn) Na HC03 CI SO4
67 Cinders 8 762 4 8.0 455 8 6 0.8 29
43 Tidal marsh 11 1,092 4 3.1 60 45 1 0 433 370
63 Tidal marsh 19 1,143 4 2.9 84 336 0 127 366
60 Peat 9 940 4 2.6 218 29 0 0 567
33 Peat 8 762 4 6.8 800 15 - 23 21
58 Muck(B) 21 1,448 4 4.0 712 20 0 5 25
29 Muck 21 1,448 4 4.2 1,270 22 0 17 23
45 Alkali soil 8 38 1 3 7.4 263 82 2 2 120
64 Clay 14 406 4 8.3 62 28 1 9 288 3
56 Clay 21 1,245 3 7.1 406 . 31 8 16 30
61 Clay 21 1,448 3 5.9 943 7 7 1 9
27 Clay 19 1,422 3 6.6 570 5 20 1 15
28 Clay adobe 16 254 3 6.8 408 15 1 10 9
5 Clay adobe 13 584 3 7.0 1,346 9 7 0.3 2
3 Clay loam 16 1,219 1 5.2 30,000 - - - -
8 Clay loam 9 533 3 7.6 350 14 7 0.1 44
25 Clay loam 8 762 2 7.2 2,980 2 10 0.3 1
36 Sandy lodm 18 1,346 1 4.5 1 1,200 - - - -
10 Sandy loam 10 1,041 2 6.6 7,460 - - - -
12 Sandy loam 17 38 1 2 7.1 3,190 4 4 0 1
16 Sandy loam 19 1,549 2 4.4 8,290 - - - -
37 Fine sand 21 1,194 3 3.8 1 1,200 - - - -
31 Fine sand 21 1,194 1 4.1 20,500 - - - -
66 Gravelly loam 21 203 1 8.7 232 66 7 28 30
6 Gravelly loam 11 864 1 5.9 45,100 - - - -
4 Loam 12 1,016 2 5.6 6,670 - - - -
35 Loam 17 38 1 1 7.3 2,060 7 11 0.6 4
23 Silt loam 18 152 3 9.4 278 84 19 11 56
1 Silt loam 9 864 3 7.0 1,215 7 1 1 8
20 Silt loam 9 864 3 7.5 2,870 2 5 0 2
19 Silt loam 10 813 1 4.6 1,970 4 2 0.3 5
18 Silt loam 11 711 1 7.3 1,410 3 9 0 2
~
Drainage is indicated as follows: I =good; 2 =moderate; 3 =poor; 4= very poor.

(B1 Decaying vegetation.
TABLE 15.2 Summary of corrosion results for copper from NBS

exposure tests4
Soil
Exposure Mean corrosion Maximum pitting
No. Type time (years) rate (mmly) rate (mm/y)
67 Cinders 9.2 0.0400 0.150
43 Tidal marsh 8.0 0.02 10 (A)
63 Tidal marsh 9.6 0.0 160 0.020

60 Peat 9.2 0.0230 0.107
33 Peat 8.0 0.0040 0.025
58 Muck(B) 9.5 0.0070 0.033
29 Muck 8.0 0.0040 <0.020
45 Alkali soil 8.0 0.00 10 0.025
64 Clay 9.2 0.0150 0.043
56 Clay 9.4 0.0028 <0.018
61 Clay 9.5 0.00 10 0.020
27 Clay 13.6 0.0004 <0.010
28 Clay adobe 8.0 0.0028 <0.020
5 Clay adobe 13.4 0.00 10 0.0 15
3 Clay loam 8.0 0.0010 t0.020
8 Clay loam 8.0 0.0008 0.036
25 Clay loam 8.0 0.0004 <0.010
36 Sandy loam 13.6 0.0066 t0.010
10 Sandy loam 13.2 0.0030 0.015
12 Sandy loam 8.0 0.0102 0.038
16 Sandy loam 8.0 0.0020 0.066
37 Fine sand 8.0 0.0058 0.041
31 Fine sand 13.7 0.0003 t0.011
66 Gravelly loam 9.2 0.0020 t0.023
6 Gravelly loam 13.3 0.0004 <0.010
4 Loam 8.0 0.0008 0.033
35 Loam 8.0 0.0005 <0.0 18
23 Silt loam 8.0 0.0046 0.04 1
1 Silt loam 8.1 0.0020 0.028
20 Silt loam 8.1 0.00 13 0.036
19 Silt loam 8.0 0.00 13 0.023
18 Silt loam 8.0 0.0003 No Ditting
- I
(A) For no. 43, depth of pits is uncertain.

(B) Decaying vegetation.
with corrosivity would be expected if the resistivity values were related to

the average moisture conditions during the year.
Of the various soil types examined, cinders was the most aggressive,
although its sulfate and chloride contents were not high. The high corrosion
rate is probably due to the fact that cinders contain a high proportion of
TABLE 15.3 Summarv of soil conditions at BNFMRA test sites’

Soil Water extract composition (ppm)
Moisture loss
Site Type at 110°C (%) pH CI SO3 CaO SalineNH3
F Wet acid peat 41.2 4.2 4.7 372 262 40
C Moist acid clay 2.4 4.6 1.7 12 17 1.8
G Made-up ground(*) - - - - - -
E Dry acid sand 0.9 5.1 2.0 4.5 12 3
A Moist normal clay 5.8 7.2 1.2 6.2 24 0.4
D Chalk 2.6 8.4 5.1(B) 1.6 36 0.3
B Slightly acid sand 2.3 6.2 0.3 3.1 8 0.3
(*) Miscellaneous builder’s rubble.
(B) This is the mean of two widely differing values (1.2 and 9.0). Gilbert explained why two such
different values should be found for the same soil.’
carbon, which is both noble and an efficient reducer of dissolved oxygen

(DO) (the most common cathodic reaction). Hence, the attack can be re-
garded as galvanic corrosion, with a high corrosion rate because of the high
cathode-anode area ratio. Soils composed mainly of decaying organic mat-
ter are the next most aggressive, particularly soils 43 and 63 (tidal marsh)
and 60 (peat). All of these have high sulfate and/or chloride content and
a high moisture content. Soils with less sulfate and/or chloride but similar
moisture content (soils 33, 58, and 29) caused much less corrosion. Note
that muck in the United States means a soil largely composed of decaying
vegetation, and not “manure,” as it does in the United Kingdom.
The important fact from this study was that the corrosion rate of
copper is very low in most soils and is not particularly high even in the
most aggressive soils. Pitting is of more concern in the most aggressive
soils, where penetration of a copper pipe could occur in -10 years. In the
less aggressive soils, little or no pitting occurred, and perforation would not
occur in 50 years.
BNF Metals Research Association (BNFMRA) conducted two studies
of underground corrosion that included copper. The first examined samples
that had been exposed for up to 11 years at seven U.K. sites.5The conditions
at the sites are summarized in Table 15.3, whereas the corrosion data are
given in Table 15.4. The made-up ground (site G) was composed mainly of
builder’s rubble and included a significant quantity of cement; hence, the
water would not be acidic. This explains the low corrosion rates at this site.
As in the NBS tests, the peat soil (F) with high sulfate and moisture
content was very aggressive, although the corrosion rate was only about
one-third of that seen in the NBS tests. The BNFMRA tests also included
TABLE 15.4 Corrosion of buried copper in BNFMRA exposures4

Soil
Exposure Mean corrosion Maximum pitting
Site Type (years) rate (mm/y) rate (mm/y)
F Wet acid peat 1.25 0.0009 -
5.67 0.0030 <0.003

10.5 0.0066 0.0 18
C Moist acid clay 1.42 0.0066 -
6.0 0.005 1 t0.003

10.5 0.0053 0.038
G Made-up ground(*) 1 .O 0.0009 -
3.08 0.0005 0.064
9.83 0.0003 <0.002
E Dry acid sand - - -
5.92 0.0001 10.003

11.08 0.0002 <0.002
A Moist normal clay 1.25 0
5.75 0.000 1 <0.003
11.08 0.0002 <0.002
D Chalk 1.58 0
5.83 0 <0.003
10.92 0.000 1 t0.002
B Slightly acid sand 1.25 0 -
5.75 0.0001 0.003
10.83 0.000 1 t0.002
(A) Miscellaneous builder’s rubble.
a chalk soil, a type not included in the NBS study. The higher pH of the
water in this soil is believed to be the reason for the low corrosion rate.
The highest sulfate contents were at sites C and F, the two most
corrosive soils. The corrosion rates at the two sites after 10 years were
much the same, despite the sulfate being 30 times higher at site F compared
with site C.
The greatest corrosion occurred at the sites with the lowest pH values,
but only in conjunction with a high moisture content. This is demonstrated
by soils C and E, which had a similar pH, but soil E had a very low moisture
content.
Because the BNFMRA tests removed samples at different time inter-
vals, it is possible to examine the corrosion rate with time.5 Figure 15.1
shows the average loss of wall thickness for sites C and F, the most cor-
rosive. The average metal loss for all the other sites was much lower, at
less than 0.3 x lo-* mm after 10 years. The results show that corrosion at
site C (moist acid clay) was more or less constant with time. At site F (wet
0 2 4 6 8 10 12
Exposure Time (y)
1 - t Site c +Site F I
FIGURE 15.1 Effect of exposure time on copper thickness 1 0 ~ s . ~
acid peat), the corrosion rate was low initially, but gradually increased with
time. Hence, although corrosion at site F was much lower than at site C
after 1 or 2 years, there was little difference in the average metal loss after
10.5 years.
Gilbert and Porter presented the results of 5-year underground tests of
copper at various sites in the United Kingdom, one of which was the same
as one used in the earlier BNFMRA s t ~ d y . Their
~ - ~ results also showed low
corrosion rates for copper, but the corrosion at Rothhampstead, site C in
the earlier study, was much greater than in the earlier study.5 The average
metal loss over 5 years was 5 times that seen in the earlier study, and the
pitting was also deeper. These results show just how variable underground
corrosion can be, even in the same location. However, despite this variation,
the average metal loss of copper was still very low.
Gilbert reviewed the data from all three studies discussed previously
and compared the data, as shown in Table 15.5.7The results show a good
correlation between the mean corrosion rates and general soil corrosivity.
The correlation of the maximum pitting rate with soil corrosivity was not
so good, but pitting rates were highest in the most aggressive soils.
TABLE 15.5 Comparison of corrosion data for copper from NBS

and BNFMRA tests'
Mean Maximum
Soil Exposure corrosion pitting
type Source Reference (years) rate (rnm/y x I 04) rate (rnm/y)
Mildly BNFMRA Gilbert" 10 0.5-2.5 0
corrosive BNFMRA Gilbert and 5 5.0-25 0.040
Poster6
NBS Romanoff4 14 4.0-25 0.043
Moderately NBS Romanoff4 14 25- 130 0.033
corrosive BNFMRA Gilbert' 10 53-66 0.046
BNFMRA Gilbert and 5 66 0.32
Porter6
Very NBS Romanoff4 14 160-355 0.1 15
corrosive
Campbell and Sutton tested copper underground for exposure times

up to 15 years at a single site.' The soil was a clay marl overlying a bed of old
peat. The samples were buried in the clay, close to the peat-clay interface.
The corrosion rates from the 5-, lo-, and 15-year trenches were all low but
showed no consistent trends with time. These results show the variability
in corrosion that is possible in the same soil at slightly different locations.
The authors ascribed the variations in corrosion rate to the variations in
proximity of the samples to the peat-clay interface.'
In addition to these research exposure tests, the results of a large
number of service experiences with copper pipe underground have also
been reported.' Copper pipe has been used extensively for underground
transportation of both gas and water, particularly in the United States, the
United Kingdom, and Australia. The experiences have been very good,
with examinations after 10 to 20 years showing little significant corrosion
or ~ i t t i n gThese
. ~ service experiences confirm the results from the long-term
exposure tests.
15.2 Copper Alloys

Samples of various copper alloys were included in the NBS exposure tests
for periods of 8 to 13 years.4 The alloys included are shown in Table 15.6.
Unfortunately, none of the single-phase brasses included arsenic, and they
all suffered dezincification over large areas in most soils, as did the duplex
brasses. Thus, the corrosion rates, calculated from weight loss, do not reflect
TABLE 15.6 Copper alloys tested in NBS underground exposure

tests for 8-1 3 years4
Type Name Composition (wt%)
Brass Red brass CuIl5Zn
Admiralty brass CuI30Znll.3Sn
Leaded brass Cu13 IZnl0.8Pb
60140 brass Cul40Zn/0.4Pb
Muntz metal CuI40Zn12.5Pb
Bronze Tin bronze Cu/l.8Sn/l Si
Silicon bronze Cull .5Si
Silicon bronze CuI3.2Si
Aluminum bronze Cu19.4A113SFe
Nickel silver Leaded nickel silver Cu/4OZn/ I ONil2.5Pb
5/20 nickel silver CuI5Znl20Ni
this and give too favorable an impression of their performance. The most
serious dezincification occurred in soils 67 (cinders) and 64 (clay with very
poor drainage), as described in Table 15.1. Hence, uninhibited brasses are
unsuitable for use underground.
The aluminum bronze that was tested does not represent a modern
composition in that it contained no nickel, whereas nickel aluminum bronze
(C63200) contains 4% to 5 % nickel. The aluminum bronze in the NBS study
suffered dealloying, but a single-phase aluminum bronze, such as C6 1400,
would probably have performed much better.'
The two nickel silvers also suffered dezincification, as might be ex-
pected, but, as noted previously, this is not reflected in the general metal loss
rate data. The tin and silicon bronzes performed better and generally gave
corrosion rates similar to copper. There were no copper alloys included in
the tests that consistently performed better than copper.
15.3 Galvanic Corrosion
Copper is sometimes connected to other metals underground, and it can be

seen from the previous chapter that copper will usually be the cathode in
such couples. Waters underground are usually similar to fresh waters, and
in fresh waters, copper can safely be connected to stainless steel (SS). This
was confirmed by Escalante and Gerhold, who found a constant reversal of
current flow between coupled copper and SS in an underground exposure
test." This shows that the potentials were similar. Hence, copper and SS
may safely be coupled together underground in most cases.
In many practical cases, copper will be connected to steel, cast iron, or

galvanized steel. Any galvanic corrosion will tend to be close to the metal
junction, but in some cases, it may be located elsewhere. This is because
the resistance of the soil between two metal items is not as important as
might be thought. There is little difference in the circuit resistance through
the soil between equal areas 0.3 m apart and two similar areas 1 m apart.'
This explains why corrosion can take place some distance away from a
dissimilar metal junction and why applying a coating to the metals at the
junction does not always stop galvanic corrosion underground, although it
often reduces it.
There are four conditions under which copper might commonly be
connected to steel, cast iron, or galvanized steel underground; in all of these,
copper will be the cathode:
1. Copper to bare anode (for example, steel)
2. Copper to partially coated anode
3. Copper to fully coated anode
4. Copper to fully coated anode with coating defects
These possibilities were examined by Hough and Alleyne, who concluded
that in cases (1) and (2) the anode would be attacked, but not significantly in
excess of the uncoupled corrosion rate because of the relatively large area
of the anode.'' In case (3), the coating would prevent any corrosion from
taking place, but case (4) could cause severe corrosion of the steelkast iron
at the defects. Hough and Alleyne exposed a coated steel pipe connected
to a bare copper pipe and put defects in the steel coating that were 6 to
25 mm in diameter." After 9 months' exposure, perforation of the steel
pipe had occurred at some of the defects. The artificial defects were from
1.2 to 6.7 m from the bimetallic junction, but the severity of the attack was
not noticeably affected by the distance from the copper.
It has also been observed that corrosion can occur underground when
dissimilar metals are close but not connected. In two of the BNFMRA
studies, aluminum was placed parallel to the copper at separations from 75
to 300 mm.6*8Only at the closest spacing did accelerated corrosion of the
aluminum occur. This was probably due to diffusion of copper corrosion
products through the soil to the aluminum. Copper corrosion products are
well known to accelerate localized corrosion of aluminum and its alloys.I2
Concern has been expressed that iron corrosion products might accel-
erate underground corrosion of copper. Hough exposed copper and steel in
an accelerated test simulating underground c o r r o ~ i o n . lSome
~ - ~ ~tests were
run in soil alone and some in 75% soil plus 25% iron oxide. The results
TABLE 15.7 Relative corrosion of copper and steel in an

accelerated underground corrosion test in the
presence of iron oxide12
Condition Metal Relative weight loss
Soil Copper 1.2
Steel 5.8
Soil + rust Copper 3.9
Steel 218
Copper connected to steel 1 .o
are shown in Table 15.7, where the weight loss of copper coupled to steel
in soil plus oxide is taken as 1. The results showed that copper coupled
to steel in the presence of iron oxide corrodes slightly less than when it is
uncoupled in soil alone. Even when uncoupled in soil plus iron oxide, the
acceleration of corrosion of copper compared with soil alone was only a
factor of 3.3. As the corrosion of copper in soil is very low, this magnitude
of increase is not significant.
It is also possible that the soil itself may contain a material that can
act as one electrode in a galvanic couple. Cinders contain high levels of
carbon, which is well known to be noble to most metals and is also a very
efficient reducer of D0.12 In addition, the area of cinders will often be much
greater than that of the exposed metal, thus exacerbating attack. Such a case
was described in Section 15.1, as part of the NBS study of underground
corr~sion.~
The corrosion rate of copper underground is generally very low, and ser-
vice experience shows that long lives are possible for copper pipe used
underground.’ However, the propensity of brasses to dezincify when ex-
posed underground means that they should not be used unless they are in-
hibited against dezincification. The two alloys most commonly used for fit-
tings underground are LG2 gunmetal ((33600) and dezincification-resistant
(DZR) brass (British Standard CZ132). Both of these have performed sat-
isfactorily underground for many years.
In very aggressive soils, coextruded, plastic-coated copper pipe may
be necessary. This coating has a high integrity and can give a long, trouble-
free life in very aggressive soils.
Where copper is connected to steel or cast iron, then coating of the
copper and iron or steel adjacent to the joint will significantly reduce any
galvanic corrosion. If the steevcast iron pipe is fully coated, then coupling
to copper could produce rapid corrosion at defects in the coating. It will
then be necessary either to fully coat the copper to reduce the cathodic area
or to apply cathodic protection to the steelkast iron.
References
1. E. Mattsson, Basic Corrosion Technologyfor Scientists and Engineers, 2nd ed.
(London, U.K.: Institute of Materials, Minerals, and Mining, 1996), p. 75.
2. “Merkblatt fur die Beurteilung der Korrosiongefahrdung von Eisen und Stahl im
Erdboden,” Standard DVWG GW9 (Frankfurt, Germany: German Gas and Water
Works Engineers Association, 1971).
3. J.O. Harris, D. Eyre, “Soil in the Corrosion Process” in Corrosion, 3rd ed., L.L.
Shreir, R.A. Jarman, G.T. Burstein, eds. (Oxford, U.K.: Butterworth Heinemann,
1994), p. 2:80.
4. M. Romanoff, “Underground Corrosion,” NBS Circulation 579, 1957.
5. P.T. Gilbert, J. Inst. Met. 73 (1946): p. 139.
6. P.T. Gilbert, F.C. Porter, J. Iron Steel Inst. Spec. Rep. 45 (1952), p. 55.
7. P.T. Gilbert, “Copper and Copper Alloys,” in Corrosion, 3rd ed., L.L. Shreir, R.A.
Jarman, G.T. Burstein, eds. (Oxford, U.K.: Butterworth Heinemann, 1994),
p. 4:48.
8. H.S. Campbell, L.J. Sutton, BNF Metals Research Association Research Report,
A1385, March 1962.
9. Copper Development Association Publication (U.K.), “Copper Underground: Its
Resistance to Soil Corrosion,” 40 1967.
10. E. Escalante, W.F. Gerhold, MP 14, 10 (1975): p. 16.
1 1. F.A. Hough, A.B. Alleyne, Gas 12 (1936): pp. 18,90
13. F.A. Hough,Am. GasJ. 144, l(1936): p. 31.
14. F.A. Hough, Am. Gas J. 144,2 (1936): p. 17.
CHAPTER 16
Atmospheric Corrosion
Copper and its alloys are often used for architectural applications both
outdoors (Figure 16.1) and indoors (Figure 16.2). Copper also has nu-
merous applications in mechanical and electrical equipment exposed to
the atmosphere. Engineering equipment frequently has to operate in the
open, without protection from the weather, and resistance to corrosion is
important. While paints and coatings are often used to prevent corrosion,
uncoated metals are also frequently used. As with materials fully immersed
in water or some other conducting fluid, corrosion can occur in the atmo-
sphere, and under the most adverse conditions, it can lead to premature
failure. Atmospheric corrosion is somewhat different from full immersion
corrosion in liquids, such as water, and in the next section, the important
factors influencing atmospheric corrosion are discussed.
16.1 Factors Affecting Atmospheric Corrosion
The most important factor is water. Corrosion in direct contact with air at
ambient temperature is generally not significant from an engineering point
of view. Therefore, water must condense on the exposed metal surfaces or
precipitate as rain. This will be followed by evaporation as the components
dry out, and thus, a most important parameter for atmospheric corrosion
is the time of wetness. Condensation is controlled by the relative humidity
(RH), and it is generally accepted that an RH of 80% or more and a
'
temperature >O"C are required for significant condensation. Rainfall is
dependent on the local weather and can vary greatly, even in a small country
like the United Kingdom.
31 7
FIGURE 16.1 Copper-covered spire on St. Magnus Cathedral, Kirkwall, Scotland.

(Photograph courtesy of the Copper Development Association.) (See color section.)
Atmospheric Corrosion 319
FIGURE 16.2 Brass escalator safety balustrades in a U.K. shopping center. (Photograph
courtesy of the Copper DevelopmentAssociation.) (See color section.)
Corrosion in the atmosphere is generally not so great as corrosion

fully immersed in water. However, galvanic corrosion rates can be as high
as rates under immersed conditions.
As the water evaporates, anything dissolved in it will concentrate.
Hence, when these species are aggressive to a metal, the corrosion rate will
increase as the water evaporates. The common species found in condensa-
tion and rainwater are briefly discussed:.
1. Oxygen. Oxygen is always present in the atmosphere, and when the

water films are thin, rapid diffusion of oxygen can occur. Hence,
polarization of the cathodic reduction of dissolved oxygen (DO) does
not usually occur.
TABLE 16.1 Emission fluxes of atmospheric aerosol

Darticles from natural sources3
Source Flux (Tg/y)
Direct Soil dust 1,500
Sea salt 1,300
Biological debris 50
Volcanic dust 33
Indirect Sulfates from natural precursors 102
Organic matter from biological 55
volatile organic compounds
Nitrates from NO, 22
2. Carbon dioxide (C02). This gas is common in the atmosphere and

dissolves readily, causing the water to be mildly acidic. It was
not thought to be a major contributor to atmospheric corrosion,2
but new research suggests that it can play a synergistic role with
chloride.
3. Ammonia (NH3). Ammonia is common in rural areas and is readily
soluble, causing increases in pH. This is beneficial for most metals,
although stress corrosion cracking of some copper alloys is possible.
4. Sulfur and nitrogen oxides. These are common in industrial and ur-
ban areas as effluent gases, for example, from fossil-fuelled power
stations. All these gases dissolve readily, forming strong acids, which
are the cause of acid rain.
In addition to these gases, two other factors affect atmospheric cor-

rosion: temperature and particles.-? Temperature is self-explanatory: the
warmer the climate, the greater the rate of chemical reactions, including
corrosion.
Particles fall into two groups: fine and coarse.-?Fine particles range
in diameter from -0.1 to 2 pm and consist chiefly of low-volatility gases
condensed onto the surface of atmospheric water droplets. These particles
can react with a metal surface but will tend to be removed by rainfall. Coarse
particles cover the size range 2 to 100 pm and are generated by a variety
of mechanical processes such as wind-blown dust. These particles tend to
remain on metal surfaces and are not so readily removed by rain. Table 16.1
shows the emission fluxes of some common atmospheric aerosol particles
from natural ~ o u r c e sThe
. ~ table shows that soil dust and sea salt are by far
the biggest contributors to the particle count, with natural sulfate forming
just 3%.
The closer a site is to the sea, the higher the chloride concentration
in condensed water. This has been well demonstrated over the last 50 years
by exposures at the LaQue Laboratory in North Carolina, USA, where
corrosion rates on the test site 25 m from the sea are higher than those on
the test site 250 m from the sea.
Chen investigated the effects of sodium chloride (NaCl) particles on
the atmospheric corrosion of copper in the short term.4 He found that the
presence of carbon dioxide had a significant effect on the corrosion rate
in the presence of salt. At low salt particle densities, the corrosion rate
increased due to the formation of copper chloride. At higher salt particle
densities, the corrosion rate also increased initially, but the corrosion prod-
uct was copper oxide, which was more protective, and the corrosion rate
decreased once the oxide layer had formed.
Compared with sea salt, soil particles are relatively inert, but they
can create a shielded area if they settle on a metal surface, and crevice
corrosion could occur. This can lead to much higher localized metal loss
rates compared to open surface corrosion.
Using all these factors, sites are generally divided into four types
by atmosphere: rural, urban, industrial, and marine. Rural sites have low
chloride and sulfur dioxide and may have ammonia. The urban atmosphere
has some sulfur dioxide (SOz), whereas industrial areas have high sulfur
dioxide and NO, content. Finally, the marine atmosphere is characterized
by a high chloride content.
Mattsson has pointed out that with recent environmental concerns and
the cleanup of pollution, industrial atmospheres were more corrosive from
'
the mid- 1950s to about 1970 in Europe. By 1977, atmospheric corrosion
rates in many industrial areas had fallen to pre-Second World War levels.
This means that corrosion data from this era may be overly conservative
for industrial atmospheres.
The cathodic reaction in atmospheric corrosion is usually the reduc-
tion of DO, and as described earlier, the reaction rates can be high, especially
with thin water films. If the dissolved solid and gas content of the water is
low, the conductivity will be low, reducing the rate of corrosion, but as the
films evaporate, the conductivity will increase.
16.2 Atmospheric Aggressivity
At any given site, the factors affecting atmospheric aggressivity can vary
from month to month. In addition, there can be huge variations from site to
site, even within one atmospheric class, for example, industrial.
TABLE 16.2 Wetness classes from IS0 9226"

Wet time
Class (%) WY) Example
TI <o. 1 < 10 Indoor
T2 0.1-3 10-250 Indoor (unheated)
T3 3-30 250-2,600 Shed storage or cold climate
T4 30-60 2,600-5,200 Outdoors temperate climate
T5 >60 >5,200 Tropical outdoor or surf
In an effort to standardize atmospheric classifications, the IS0

CORRAG program tested a range of metals, including copper, at many
sites across Europe and North America for periods of up to 8 years. From
these data, a classification system was developed, as described in I S 0 9226,s
which uses the variables time of wetness ( T ) , sulfur dioxide content ( P ) ,
and chloride content (S) and ascribes values to them for different ranges,
as shown in Tables 16.2-16.4.
Five corrosion classes exist, with increasing corrosion rates as the
class number increases. The corrosion rate for different metals in each class
varies, as shown in Table 16.5 for some common metals. Table 16.6 shows
how the corrosion class for copper varies depending on the value of the
three variables S, T , and P . It is clear from Table 16.6 how important
the time of wetness is. With high values of T , the corrosion class is very
aggressive for copper, irrespective of the values of S and P . However, even
class 5 for copper corresponds to a corrosion rate of -3 pm/y, which is not
a high corrosion rate, although it may be critical in some applications.
One important feature of this classification system is that it allows a
wide range of times of wetness and includes not just open exposure out-
doors, but also exposures with some level of screening from the elements.
Indoor corrosion can be very important, particularly on critical compo-
nents such as electronic circuit boards. This is discussed in more detail in
Section 16.5.
TABLE 16.3 Classes for sulfur dioxide in I S 0 9266

SO2 deposition rate SO2 concentration
SO2 class (mg/m2/d) (mg/m3)
PO < 10 <I2
PI 10-35 12-40
P2 36-80 4 1-90
P3 8 1-200 91-250
TABLE 16.4 Classes for chloride

in IS0 9226
CI- deposition
Chloride- class rate (mg/m2/d)
so t3
S1 3-60
s2 6 1-300
s3 >300
While the I S 0 classification system works well in many cases, there

are still some factors that require incorporation into the system.' One of
these is the frequent difference in corrosion rate between the upper and
lower faces of flat standard specimens. Another is the role of pollutants
other than chloride and sulfur dioxide.
16.3 Corrosion Rates Outdoors
Kucera and Mattsson conducted atmospheric exposure tests on a range of

metals and alloys at various Swedish sites, and Table 16.7 summarizes their
data.6 These results show that the corrosion rate of copper is not very high
in any environment, but it is highest in a marine environment. Other metals,
such as zinc and nickel, had the highest corrosion rates, after carbon steel,
in the urbadindustrial environment.
There have been a number of studies of copper exposed for up to
20 years in a variety of atmospheric condition^.^-^ Table 16.8 summarizes
the data from these and other references. The data show reasonable agree-
ment for the different atmospheric types, with the Swedish 16-year data
giving the lowest corrosion rates.' The variability of the severity of in-
dustrial locations is also shown in that some sites show more attack than
experienced under marine conditions, whereas other sites were less severe
than marine sites.
TABLE 16.5 Corrosion classes in Is0 9226

Average corrosion rate for first 10 years (pm/y)
Corrosion
class Steel Copper Zinc Aluminum Weathering steel
I <0.5 <0.01 tO.O 1 10.01 <O. 1
2 0.5-2 0.01-0.1 0.01-0.5 0.01-0.025 0.1-1
3 2-10 0.1-2 0.5-2 0.025-0.2 1 4
4 10-35 2-3 2-4 0.2-1 .o 4-10
5 >35 >3 >4 >1.0 > 10
TABLE 16.6 Corrosion classes for different environmental classes for copper
and zinc from IS0 9226
Corrosion class
T2 T3 T4 T5
so-Sl s2 s3 so-s1 s2 s3 so-s1 s 2 s3 so-SI s 2 s3

PO-P 1 1 1-2 2 3 3 3 4 3 4 5 4 5 5
P2 1-2 2 3 3 3 4 4 3 4 4 5 5 5 5
P3 2 33-4 3 3-4 4 4 - 5 5 5 5 5 5
TABLE 16.7 Corrosion of some common metals in various

atmospheric environments6
Corrosion rate ( p d y )
Alloy Rural Urban Marine
Magnesium 6.5 11 14
Aluminum 0.1 0.5 0.6
Anodized aluminum 0 0.2 0.2
Zinc 0.5 2.5 I .3
Carbon steel 22 53 36
Weathering steel 22 48 32
Tin 1.4 2.1 10
Lead 0.3 0 0
Copper 1.2 1.4 3.2
Nickel 0.4 2.7 1 .o
Stainless steel (Type 304) 0.03 0.04 0.04
TABLE 16.8 Comparison of copper corrosion rates from

different sources
Country Time (years) Reference Rural Urban Industrial Marine
Sweden 5 5 I .2 1.4 - 3.2
United States 4 6 - 0.9 1.1 2. I
Sweden 16 7 0.3-0.5 0.9-1.3 - 0.5-0.9
United States 2 8 1.2 - I .9 1.8
20 8 0.7 - I .3 1.1
Typical values - 3 0.5 1-2 ~2.5 -I

TABLE 16.9 Comparison of corrosion rates of copper alloys in atmospheric

exposure tests
Country Time (years) Reference Rural Urban Industrial Marine
United States 4 7 - 0.5-1.0 0.9-1.4 0.4-2.7
Sweden 16 8 0.3-0.5 0.9-1.3 - 0.5-0.9
United States 20 9 0.5-1.1 - 1.2-2.3 0.5-1.7
All of the authors of References 6-8 comment that the films formed
slowly, with time, on the metal surface and were still forming after years of
exposure, although corrosion rates were decreasing with time as the films
f ~ r m e d . This
~ - ~is shown in Table 16.8 by the data of Costas after 2 and 20
years’ exposure.’
In addition to pure copper, there are also data on a range of copper
alloys. All the studies mentioned previously tested a range of brasses,
bronzes, copper-nickels, and nickel The results in Table 16.9
show the range of corrosion rates measured for the various alloys. There
were no significant differences between the different copper alloys. It can
be seen that these are very similar to the rates for pure copper in Table 16.8;
in fact, Holm and Mattsson do not separate them.8
All the authors comment that general corrosion is not the main prob-
lem with copper alloys and that localized corrosion caused more severe
a t t a ~ k . Dezincification
~-~ was seen with many of the uninhibited brasses
containing more than 15% zinc, although this was not deep in most cases.
Also, the results from the three sets of tests were somewhat variable. For
example, Castillo saw deep dezincification only on a manganese bronze
(C66900), whereas Holm and Mattsson saw attack on all alphaheta brasses
and Costas saw dezincification only on 70/30 brass. Dezincification of the
nickel silvers was slight or nonexistent. Holm and Mattsson reported de-
zincification depths varying from 50 to 250 pm, which is greater than any
metal loss due to general corrosion but may be of little concern from an
engineering point of view.8 Castillo and Popplewell reported pitting and/or
crevice corrosion with the phosphor bronzes containing from 4% to 8%
tin.7 The depth of attack varied from 76 to 127 pm, again, much deeper
than would occur by general corrosion.
Castillo and Popplewell also exposed U-bend samples of many cop-
per alloys in the cold-worked condition (hard or extra hard).7 Those that
suffered stress corrosion craclung (SCC) failed in a relatively short time. In
the industrial environment, failures occurred with cartridge brass (C26000),
leaded brass (C35300), and other duplex brasses. The nickel silvers (with the
exception of C75200), one aluminum brass sample, and 40% cold-worked
admiralty brass also failed by SCC in the industrial environment, although
the time to failure for aluminum brass was much greater than for the other
alloys. The 10% cold-worked admiralty brass did not suffer SCC at all. In
the marine environment, SCC failures occurred only with the nickel silvers
and a high beta-phase brass. Copper alloys for atmospheric exposure are
more often used in the annealed or half-hard condition. As no alloys in this
condition were tested, it is not possible to say how susceptible they might
be to SCC over long periods of time. The susceptibility of copper alloys to
SCC is discussed in more detail in Chapter 10.
The data presented so far are from northern temperate climates. In
tropical climates, conditions can be more aggressive, both because of the
higher ambient temperature and the longer time of wetness. This means
somewhat higher corrosion rates, but these will still not be very high for
copper, although it is important to guard against localized corrosion with
copper alloys.
16.4 Galvanic Corrosion
The subject of galvanic corrosion in the atmosphere has been reviewed by

Francis, and galvanic corrosion of copper alloys was discussed in Chap-
ter 14.'" Galvanic corrosion in the atmosphere tends to be confined to the
dissimilar metal joint where the condensation has occurred. Hence, area
ratios are usually around 1:1. From the preceding characteristics, it can be
deduced that atmospheric galvanic corrosion is similar to galvanic corrosion
when immersed in a low-conductivity electrolyte. The corrosion rate obvi-
ously varies as the species dissolved in the water vary and as the water films
evaporate and condense again. With most metal couples, copper is the cath-
ode, and it can be more efficient than well-known cathodic materials such as
graphite. Doyle and Wright investigated the galvanic corrosion of aluminum
'
coupled to a range of alloys at several sites in North America.' The results
show that copper was a more efficient cathode than graphite at three indus-
trial sites and was only a little less efficient than graphite at three marine
sites (Figure 16.3). It can also be seen that Alloy 400, with -35% copper,
is much less efficient as a cathode than both pure copper and graphite.
Where copper is connected to more noble metals, such as stainless
steels, nickel, and chromium, there is either no increase in the corrosion
rate of copper or only a small increase.6 This is essentially because these
noble metals are not very efficient cathodes.
Copper Graphite Alloy 400 Copper Graphite Alloy 400

Industrial Marine
I
I n s i t e I =site 2 ~ ~ i3 t e
FIGURE 16.3 Atmospheric galvanic corrosion loss of aluminum coupled to various

materials.”
Even in a tropical atmosphere, copper will not corrode very much

when coupled to a noble metal. Beavers et al. coupled a titanium alloy
(Ti/6A1/4V) to various metals in the tropical marine atmosphere of Panama,
in Central America.12 The corrosion rate of 70/30 brass was 1 p d y , no
different from the rate for uncoupled brass. This is because titanium and its
alloys are not very efficient cathodes in atmospheric galvanic couples.
16.5 Corrosion Indoors
Atmospheric corrosion indoors can also be very important. Copper and its
alloys are often used in large buildings, such as offices, hospitals, and hotels,
where the appearance of the metal is important, as shown in Figure 16.4.
Corrosion can also be important in museums, where further degradation
of ancient metal artifacts is undesirable. It should be recognized that the
response of an indoor atmosphere to changes in the concentration of external
pollutants can be very rapid, of the order of minute^.^ Hence, it is important
to keep the atmosphere under control to minimize corrosion. This involves
FIGURE 16.4 Copper used in a decorative internal curtain wall. (Photograph courtesy of
NDM.) (See color section.)
control of both temperature and humidity and, in severe circumstances, the

control of aerosols containing aggressive species such as sulfur-containing
gases or chlorides.
The Instrument Society of America (ISA) has classified atmospheres
indoors by the thickness of the corroded layer after a 1-month exposure.
The different levels and their descriptions are shown in Table 16.10. The
ISA has also categorized the gaseous constituents of an atmosphere for the
various corrosivity levels, as shown in Table 16.11. Note that if the gases
in a given environment correspond to different corrosivity levels, then the
overall level is determined by the gas concentration at the most severe level.
The data in Table 16.11 are valid provided that the RH is less than 50%.
Following the ISA classification, the corrosivity class should increase by
one level if the average RH increases by
Leygraf and Graedel reported some corrosion data for copper in
Swedish churches and muse urn^.^ Most of the buildings were at level G1
or G2, but one was in category G3. This shows how severe the environment
may be for critical components, even in buildings where one would not
expect it.
TABLE 16.10 Corrosivity classification according to the ISA

Corrosivity Corroded layer Description of
level thickness (nm)(A) severity level
G1 mild t30 Corrosion is not a factor in determining reliability.
C2 moderate 30 to < 100 Effects of corrosion are measurable and may be a
factor in determining reliability.
G3 harsh 100 to ~ 2 0 0 There is a high probability that corrosion will
occur. Environmental control or special design and
packaging is recommended.
CX severe >200 Only specially designed and protected equipment
is expected to operate. Use of equipment in this
class is a matter of negotiation between user and
supplier.
(A)
All thickness values after 1month exposure.
Atmospheric corrosion is most important for electronic components

in which copper is used as the conductor on printed circuit boards. Even a
small amount of corrosion can interfere with the correct working of a board,
even if it does not stop it working altogether. Corrosion products act as a
resistance, which causes a voltage drop. As much of modern electronics
operates at low voltages, any voltage drop can affect an instrument’s ability
to function correctly.
The copper used as a conductor in printed circuit boards is usually
coated to prevent corrosion, but contacts and connectors are not. These
are often made of copper, or a copper alloy, and are coated with gold or
tin. In addition, corrosive elements can penetrate defects in the coatings of
integrated circuits and can cause them to malfunction. A detailed description
of the corrosion of electronic equipment is contained in the work of Leygraf
and Graedel.3
TABLE 16.11 Maximum concentrations of gaseous corrodants

according to ISA corrosivity classification
Maximum concentration (ppbv)
Corrosivity level H2S SO2 Clz NO,

G1 t 3 < 10 <I <so
G2 <lo < 100 t 2 < 12s
G3 t50 <300 < 10 < 1,250
GX >so 3300 110 11,250
r-
Debris Debris
collects collects
Weld dressed
on outside not
on inside of tube
FIGURE 16.5 Design details likely to result in corrosion because of entrapment of

debris.13
The more severe the operating environment, the more rigorous will be
the protective measures that are required to ensure the successful operation
of electronic equipment. These are discussed in the following section.
When selecting a method of corrosion prevention or reduction, it is impor-

tant to know the severity of the environment and the desired performance
life of the component. However, there are some factors that exacerbate
atmospheric corrosion.
For components that are to be exposed to the atmosphere, it is im-
portant to use a design that removes or minimizes areas where water and
debris can accumulate. Figure 16.5 shows some poor designs that can lead
to c o r r ~ s i o n .Where
'~ mixed metals are to be joined and galvanic corrosion
could be a problem, use a design that insulates the two metals or apply a
suitable coating to the cathode in the couple." The prevention of galvanic
corrosion is discussed in more detail in Chapter 14 and also by Francis."
When selecting copper alloys for use in corrosive atmospheres, ensure
that an alloy is chosen that will not suffer SCC or be susceptible to pitting in
the particular environment. Using alloys in the annealed or stress-relieved
condition minimizes internal stresses and the subsequent risk of SCC. SCC
can lead to rapid failure, compared to general corrosion, as described in
Section 16.3, while pitting corrosion usually looks unsightly but does not
generally cause failure.
Where it is desired to preserve the metallic appearance of copper or a
copper alloy outdoors, then a suitable coating system must be chosen. One
developed by BNF Metals Research Association for the International Cop-
per Research Association consisted of a clear, cellulose lacquer containing
benzotriazole, a well-known inhibitor for copper and its alloys. Known
as INCRALAC, this coating has given a long life when properly applied.
It is still available using more modern lacquer formulations. Waxes have
also been used to preserve the appearance of copper, either alone or in
combination with a lacquer.
Where a decorative patina is required, the color of this and the rate
of formation depend very much on the composition of the specific atmo-
sphere in which it is exposed. To avoid problems, commercial chemical
treatments are available that will create an aesthetically pleasing patina
prior to installation.
One area where problems can occur is during transport, particularly
on items being transported by sea. Critical components may need to be
packed with a desiccant to keep the moisture content low. An alternative
is to use a vapor phase inhibitor, ensuring that it is compatible with all the
materials used in the component.
It is also important to recognize that under some storage conditions,
packing material may release chemicals that are aggressive, resulting in
severe corrosion during transport or storage. Such a case was described in
Chapter 5 , where a packing material used with copper tubes released formic
acid (HCOOH) during storage and caused severe formicary corrosion of
the tubes.
With indoor applications, a patina will form on copper and its alloys
even more slowly than outdoors. Where a shiny, metallic appearance is
required, an inhibited, clear lacquer may offer the best method for long-term
preservation of the appearance. Components that are handled a lot, such as
handrails, require no such treatment as handling alone will keep them clean.
Electronic components may require special protective measures, par-
ticularly if they are in a critical application. It is important to check the
corrosivity of the atmosphere, both before and after installation, and this
should be done at the most corrosive time of the year (usually the most
humid months). For a category G2 atmosphere, the selection of materials is
important. For categories G3 and GX, additional protective measures, such
as a protective housing and/or an organic coating, are required.
TABLE 16.12 Climatic and operating conditions for electronics to Swedish

pulp and paper industry standard
Maximum level Average
Compound (PPW over (h)
Gaseous pollutants so2 30 1
HZS 10 1
NOS 30 1
CE2 + HF 10 24
HF 10 24
NH3 500 1
0 1 5 1
Thickness of corroded copper layer <30 nm during first month
Overpressure 20 Pa
Temperature &2"C in the range 18"-25"C
Temperature variations <O. 1"C/min
Relative humidity 35 5% *
Relative humidity variations <O. 1%/min
Solar radiation Protection against direct solar radiation
Air velocity t0.2 m / s in rooms with human beings
Particulate pollutants tO.01 mg/m3, average over 5 minutes
t0.005 mg/m3, average over 24 hours
For electronic components, there are three types of protective

measures3:
1. Selection of materials and design of components

2. Protective measures during manufacture, transport, and storage
3. Protective measures during installation and service
Design of electronic packages can have an adverse effect on performance

by encouraging corrosion. Fans used for cooling can increase the deposition
rate of pollutants. Coatings and other organic materials that outgas corrosive
substances, such as hydrogen sulfide (H2S), must be avoided.
The manufacturing process can add corrosion problems to electronics
by means such as fingerprints or metallic or organic contaminants. Residues
from soldering or cleaning may also lead to corrosion. Hence, care is needed
in all cleaning operations.
The atmospheric conditions during transport and storage may be dif-
ferent from those of the service environment. It is important that protective
measures are chosen to ensure full protection until installation. This may
involve hermetically sealed transport containers, desiccants, or vapor phase
inhibitors. Inhibitors must be compatible with all the materials used in the
component being shipped. Because corrosion requires water, it is important

that all materials are dry during packing.
Critical components in service may require a complete indoor air
treatment system, which involves special design of the room. This usually
includes hermetically sealed doors or air locks, seals on the ceiling and
floor, and seals on all the service entry points. To remove corrosive pollu-
tants, the air may need to be filtered, using activated carbon or aluminum
oxide impregnated with potassium permanganate.3 Remember that the air
flow through the filters must be sufficiently slow to allow full removal of
the pollutants. Table 16.12 shows a clean air specification for electronics
adopted by the Swedish pulp and paper industry. Similar specifications have
been adopted in other countries.
References
1. E. Mattsson, Basic Corrosion Technologyfor Scientists and Engineers, 2nd ed.
(London, U.K.: Institute of Materials, Minerals, and Mining, 1996),
p. 80.
2. D. Fyfe, “The Atmosphere,” in Corrosion, 3rd ed., L.L. Shreir, R.A. Jarman, G.T.
Burstein, eds. (London, U.K.: Butterworth Heinemann, 1994), p. 2:33.
3. C. Leygraf, T.E. Graedel, Atmospheric Corrosion (New York, NY John Wiley,
2000).
4. Z.Y. Chen, “Initial Atmospheric Corrosion of Copper (Influence of Sodium
Chloride Particles, Humidity, Carbon Dioxide and Gaseous Pollutants),” Licentiate
Thesis (Ph.D.) (Stockholm, Sweden: Royal Institute of Technology, 2005).
5. IS0 9226, “Corrosion of Metals and Alloys4orrosivity of Atmospheres-
Determination of Corrosion Rate of Standard Specimens for the Evaluation of
Corrosivity” (Geneva, Switzerland: ISO).
6. V. Kucera, E. Mattsson, “Atmospheric Corrosion of Bimetallic Structures,” in
Atmospheric Corrosion, W.H. Ailor, ed. (New York, NY: John Wiley, 1982).
7. A.P. Castillo, J.M. Popplewell, ASTM International Special Publication, STP 767,
1982, p. 60.
8. R. Holm, E. Mattsson, ASTM International Special Publication, STP 767, 1982,
p. 85.
9. L.P. Costas, ASTM International Special Publication, STP 767, 1982, p. 106.
NACE International, 2001), p. 81.
11. D.P. Doyle, T.E. Wright, ASTM International Special Publication, STP 978, 1988,
p. 161.
12. J.A. Beavers, G.H. Kock, W.E. Berry, Corrosion of Metals in Marine
Environments (Columbus, OH: Battelle, Metals and Ceramics Information Centre,
1986).
13. R. Parkins, C. Chandler, Corrosion Control in Engineering Design (London, U.K.:
DOUHMSO, 1978).
CHAPTER 17
Corrosion in Waters
The previous two chapters dealt with the specific environments of the
atmosphere and underground. In this chapter, the use of copper and its alloys
in applications associated with waters of one type or another are reviewed.
Where specific types of corrosion may occur, the reader is referred to the
relevant chapters. The intention of this chapter is to highlight the widespread
applications for copper and its alloys in waters and some of the common
problems, not all of which are corrosion related.
17.1 Fresh Water
Copper and its alloys are widely used for pipes and fittings to handle fresh
and potable waters, for hot waters, and for central heating systems. Although
there are occasionally failures due to corrosion, thousands of kilometers of
copper and copper alloy piping systems are giving reliable service.
Dezincification of brass fittings, valves, etc., is rarely a problem these
days, with the wide availability of dezincification-resistant brass for both
forgings and die castings. In some countries, LG2 gunmetal ((283600) is
available at lower cost than brass, and so dezincification is virtually un-
known there (see Chapter 9).
One problem that still sometimes occurs with brass components in
fresh waters is stress corrosion cracking (SCC; see Chapter 10). This is
usually from the outside and is often associated with ammonia (NH3) from
phenolic insulation systems. It is important to select insulation systems that
cannot release ammonia or formic acid (HCOOH). The latter can lead to
formicary corrosion, as described in Chapter 5.
335
In an attempt to determine the most common causes of corrosion

failure at the time of writing, the author had discussions with Phillip Munn,
a U.K. consultant who regularly examines copper and copper alloy failures.
It appears that there are differences between domestic and commercial
properties.
In domestic properties, in the United Kingdom, the most common
corrosion problems are blue water and Type I pitting corrosion, sometimes
associated with flux residues. The causes of this and the methods to avoid
them are discussed in Chapters 4 and 5 . In commercial properties with
recirculating hot water systems, at least half of the failures are due to
erosion corrosion. The causes of this and suitable remedies are discussed
in Chapter 7. Such problems can be avoided by attention to water velocity
and potential turbulence raisers at the design stage. Microbially influenced
corrosion (MIC), or Type I11 pitting, and deposit attack account for another
30 to 40% of the failures in commercial recirculating hot water systems.
The causes of these types of corrosion and their avoidance are discussed in
Chapters 5 and 6.
In commercial cold water systems, the two most common causes
of failure in the United Kingdom are MIC/Type I11 pitting (-40%) and
Type I pitting (-40%), as discussed in Chapter 5 , with erosion corrosion
accounting for only 5 to 10% of failures. The remainder of the problems
are caused either by excessive chlorination or long periods of stagnation.
The ways in which these two factors can affect corrosion in fresh waters
are discussed in Chapters 4 and 5.
As mentioned previously (Chapters 4 and 5 ) , there is an increasing
tendency to find phosphate deposits associated with problems such as blue
water and Type I pitting. Research is needed in this area to determine the
role of phosphate in copper corrosion and what levels will not adversely
affect corrosion in various types of water.
BNF Metals Technology Centre published a guide on avoiding corro-
sion problems in domestic central heating systems.' Most of the problems
occur when air can get into the system. Normally, the dissolved oxygen
(DO) from the air is consumed by corrosion of the metal components in
the first day of operation, and thereafter, corrosion is negligible. However,
leaking joints near the pump suction and overpumping can both lead to the
introduction of fresh DO. This does not normally have a significant effect
on the corrosion of copper central heating pipes but may cause dezincifi-
cation of any brass fittings. In severe cases, this may cause blockage of the
system with meringue corrosion products (see Chapter 9).
Corrosion in Waters 337
Overpumping occurs if the pressure difference between the points

where the vent pipe and cold feed pipe are connected exceeds the head
between the highest point of the swan neck on the vent pipe and the water
level in the tank supplying the system. Overpumping leads to water being
driven out of the vent pipe into the header tank, and fresh, aerated water is
drawn into the system from the tank to make good the loss. Hence, the DO
content never decreases to low levels. Overpumping is prevented by correct
positioning of the vent pipe and cold water feed pipe into the system to
keep the pressure difference between them low. This is discussed in more
detail in various commercial design guides.
Another possible cause of failure is when central heating pipe work is
buried in cement or concrete screed with insufficient allowance for thermal
expansion and contraction, as the system heats up and cools down. If the pipe
is overly restrained, fatigue failures of the pipe can occur. The incorporation
of thermal expansion sections in long pipe runs prevents this problem.
External corrosion can be prevented on small-bore central heating
tubing (up to 10 mm outside diameter) by installing it with a coextruded
PVC covering. With larger-diameter pipe ( 212 mm outside diameter), it is
necessary to leave a duct in concrete when it is being laid if copper central
heating tubing is to be installed. This avoids restraining the pipe, such that
thermal fatigue might occur, as well as putting an air space around the pipes.
The pipes should be insulated using a moisture-resistant coating, avoiding
types that may release ammonia or formic acid (see Chapters 5 and 10).
17.2 Seawater
Copper alloys are widely used in seawater, particularly in seawater cooling

systems. They are used in both cast and wrought forms for pumps, pipe
work, valves, heat exchangers, filters, etc. Copper alloys also find extensive
use in firewater systems and for ships’ propellers and propulsers. The
most commonly used alloys are 90/10 copper-nickel (C70600) for wrought
products and nickel aluminum bronze (NAB; C95800) for cast products.
It is not the intention of this section to go over seawater applications
in detail; rather, this section intends to point the reader toward sources of
further information with regard to design and applications.
A good review of the behavior of 90/10 copper-nickel in seawater
was prepared by Parvizi et al., which includes historical data, proper-
ties and composition, and corrosion behavior in seawater, including the
effects of velocity, sulfides, temperature, etc.’ The properties of cast NAB,
including corrosion resistance in seawater, are compared with other al-

loys in a publication by Oldfield and master^.^ The corrosion resistance
of NAB in seawater is also described in Meigh’s book on the aluminum
bronze^.^
Most seawater applications involve surface or near-surface water,
where there are changes in flow rate and temperature due to tides and the
DO level is close to saturation. Reinhart published corrosion data on a wide
range of copper alloys exposed at depths from 760 to 1,800 m (2,500 to
6,000 ft) for periods from 4 months to 3 years.5 The study examined general
corrosion, pitting, and SCC. Many of the results were similar to those seen
in surface waters, but some of the alloys corroded differently due to the
lower temperature and DO content at great depth.
Firewater systems operate at high velocities when working, up to 10
m/s, and the use of copper alloys might seem problematical. However, fire-
water systems are only tested for about an hour per week, at most, and
spend the rest of the time stagnant. Hence, any erosion corrosion damage
sustained during testing will reform protective films when inactive. Al-
though there have been a few problems due to corrosion by sulfides, 90/10
copper-nickel piping and NAB or gunmetal sprinkler heads have given
good service in firewater systems all over the world. One advantage of
copper alloys over high-alloy stainless steels (SS) is their ability to func-
tion without accelerated corrosion at elevated temperatures. In hot climates,
firewater piping can be exposed to direct sunlight and surface temperatures
may reach 60 or 70°C in the summer. Such high temperatures have caused
corrosion of high-alloy SS systems, but the author knows of no failures of
copper alloy systems at elevated temperatures.
The author described the selection of materials for seawater cooling
systems, including the use of wrought and cast copper alloys.6 It is also im-
portant to stress the use of correct specifications and perform quality-
assurance and control procedures to avoid potential problems. Having se-
lected a suitable copper alloy, it is not just sufficient to specify it to be
supplied to a standard such as ASTM, EN, or BS. While some standards
do cover all the relevant properties, others do not. The problem of carbon
films was discussed in Chapter 5 ; however, at the time of writing, only
the old BS 2871, part 3,7 specifically requires heat exchanger tubes to be
free of films in the bore, and even then, a method of testing is not speci-
fied. The author’s suggestions for detecting carbon films were discussed in
Chapter 5.
Another problem can occur with 90/10 copper-nickel. The alloy gives
the best corrosion resistance when the iron content is in the range 1.5 to
2.0 wt%, as discussed in Chapter 7. However, many of the common stan-

dards for the alloy give the iron limit as 1.0 to 2.0 wt%. To ensure adequate
resistance to erosion corrosion, it is necessary to specify that the minimum
iron content be 1.5 wt%.
This level of iron exceeds the solubility limit for the alloy at room
temperature. However, if the iron is not in solution, the corrosion resistance
decreases quite markedly, although not the resistance to erosion corrosion
(see Chapter 7). To retain the iron in solution, it is necessary to cool the
-
alloy rapidly from an annealing temperature of 900°C. It has been stated
that this problem is well understood, and modern manufacturing methods
will always produce good-quality material. However, the author examined
failures of 90/10 copper-nickel pipes from the cooling water system of an oil
platform in the North Sea. They were suffering a very high rate of general
corrosion plus erosion corrosion after the bends, even though the water
velocity was only -3 d s . The clue to the cause of the problem was at the
welds: not only was the 70/30 copper-nickel weld metal unattacked; neither
was the parent metal in the heat-affected zone (HAZ). Figure 7.8 (Chapter 7)
shows a half-section of pipe, and the cut edge clearly shows the thicker metal
in the weld and HAZ. This was because the iron was not in solution in the
pipe, but the weld thermal cycles in the HAZ had re-solutionized the iron and
restored the corrosion resistance. The weld metal was not affected because
iron is more soluble in the 70/30 copper-nickel filler metal. Determination
of the magnetic permeability with a simple device, such as the Severnt
magnetic gauge, showed the parent metal to have a value of > 1.2, while
the weld metal and HAZ were < 1.05. It is suspected that the attack of the
parent metal started as accelerated general corrosion that was particularly
rapid because the alloy was being used close to its recommended limit. As
the weld and HAZ became increasingly proud of the rest of the surface, they
would increase local turbulence such that erosion corrosion immediately
downstream became possible. It is recommended that a simple magnetic
permeability check be carried out on the parent metal at the same intervals
as the other tests (e.g., for carbon films and residual stresses). When asked
about this problem, Yorkshire Imperial Metals recommended that material
with a magnetic permeability of >1.10 be rejected. It should be stressed
that this type of failure is not common but can be avoided by additional
tests during manufacture.
NAB is a widely used alloy in both cast and wrought forms. NAB
has sometimes been observed to suffer dealuminification. It has been found
t Trade name.
that keeping the nickel content 20.5% above the iron content will prevent
dealloying (Chapter 9). This is not covered by most specifications and
should be added as a requirement when specifying this alloy, in either cast
or wrought form.
At the high temperatures used in hot working or during welding,
NAB has a structure that is alpha and beta. During cooling, the beta phase
transforms to alpha and kappa. However, if the cooling rate is too fast, the
beta phase does not fully transform. Beta phase is highly anodic to the alpha
phase, and it is attacked rapidly in seawater (Chapters 3 and 9). The Royal
Navy experienced corrosion problems with weld-repaired castings in NAB,
and seam-welded pipe in a desalination plant corroded and unzipped along
the weld (see Chapter 18).
The author was involved in investigating corrosion at a large multi-
stage flash (MSF) desalination plant in the Middle East, where two NAB
tube plates out of 128 showed pitting due to the presence of beta phase. It
was thought that these two were the first in a batch to be rolled and cooled
too quickly off the rolls. As subsequent plates were piled up, the thermal
mass increased and cooling rates slowed. The tube plates could not be weld
repaired because that would have created more beta phase. The solution
was to grind out the corrosion and fill the space with a cold-setting ceramic
filler.
It is recommended that all NAB be given a heat treatment after the final
hot operation (welding, rolling, etc.). For castings, it is recommended that
the Royal Navy heat treatment of 675" f 25°C for 2 to 6 hours, followed
by air cooling, be specified. For wrought products (such as thick plate), a
heat treatment of 725" f 25°C for the same period, followed by air cooling,
is recommended. The author has specified this treatment for the tube plates
in another MSF plant, and there have been no problems after 10 years in
service.
Some nickel alloys often come in a number of different heat-treated
forms, each with somewhat different mechanical properties. It is important
to make sure that the design has been carried out using the mechanical
properties for the required heat treatment of the finished product.
One example is Alloy K-500, which is an age-hardening nickel-copper
alloy that is frequently used for marine fasteners. It is usually used in
the aged condition to obtain a 0.2% proof stress of 730 MPa or more.
However, it can be purchased in both annealed and aged conditions. After
manufacturing the component, the incorrect strength may be produced if
the material was purchased in the wrong heat treatment form. If the material
was already aged and is aged again, the strength will be higher than desired,
with a consequent loss of ductility. If the material was purchased annealed
and is not aged, the strength will be below the requirement for the project.
This demonstrates not only the need for detailed purchasing specifications,
but also the need for a Q N Q C system that will detect such errors.
A number of authors have detailed their experiences with copper
alloys in seawater. However, these tend to discuss failures and frequently
neglect to mention the thousands of tons of copper alloys that are giving
excellent service.
Birn and Skalski described experiences with copper alloys by the
Polish marine industry.8 Although there have been problems due to dezinc-
ification of brasses and erosion corrosion, the principal problem has been
accelerated corrosion due to sulfides in polluted waters.
The U.S. Navy also experienced failures of copper-nickel due to
polluted water, and these failures, and the extensive research program that
followed, are discussed in Chapter 7. This is because, although sulfides
cause corrosion that has the appearance of pitting, the rate of propagation is
very flow-rate-dependent, with high rates of attack at normal flow velocities
(2 to 3 m / s ) compared with stagnant conditions.
Johnsen described the experiences with copper alloys in the Norwe-
gian sector of the North Sea by six oil and gas companies.’ Their experi-
ences were very similar, with the main corrosion problems being erosion
corrosion; accelerated corrosion due to sulfides, produced under stagnant
conditions; and galvanic corrosion when coupled to higher-alloy materials.
These problems led some of the companies to switch to other materials, but
others devised strategies to make copper alloys work.’
The author has had discussions with oil and gas companies in the
U.K. sector of the North Sea, and they have also had some corrosion prob-
lems with copper alloys, mostly due to galvanic corrosion, but also due
to erosion corrosion. One factor that emerged was that the galvanic prob-
lems are not accelerated general corrosion; instead, the copper alloys were
suffering from one of several forms of localized corrosion, that was then
greatly accelerated by coupling to a higher-alloy material. This is discussed
in Chapter 14, with suggestions for alleviation. Despite some corrosion
failures, copper alloys are still used in the U.K. sector of the North Sea,
where they mostly perform well.
Johnsen pointed out that although the design and operating rules for
copper alloys are well known, for example, keeping to velocity limits and
flushing out stagnant legs of piping with fresh water, these are frequently
ignored, both at the design stage and in operation, at least in the Norwegian
sector of the North Sea.' The author's discussions with U.K. sector operators
suggest that the rules are implemented a little better, as shown by the greater
success of copper alloys in this sector.
The author has examined a number of failures of NAB pumps and
valves from seawater service over the last 30 years. The main cause of
failure has been erosion corrosion due to the presence of sulfides or over-
chlorination (see Chapter 7). Failures due to corrosion of beta phase (Chap-
ters 9 and 18) are rare, now that the importance of postweld heat treatment
is recognized.
Probably the most widely used copper alloy in seawater is 90/10
copper-nickel, and Schleich and Powell give details on how to avoid the
typical failures reported previously and also how to use copper alloys to
avoid biofouling problems. l o They stress the importance of getting the
conditions at start-up right so that protective films form that give significant
resistance to accelerated corrosion under upset conditions, for example,
sulfide ingress.
When corrosion problems do occur, or there is significant buildup of
scale, it may be necessary to use an acid-cleaning treatment to restore the
heat transfer coefficient in copper alloy heat exchanger tubes. The selection
of the correct acid and strength is important to avoid doing further damage
to the tubes. Le Guyader et al. described the effects of three commercial
cleaning solutions on 70/30 copper-nickel and the optimum way to clean
them to avoid subsequent problems."
17.3 Swimming Pools
Although copper alloys were once widely used for handrails, grates, and
other such features in swimming pools, they have largely been superseded
by SS. However, copper alloys, particularly copper-nickels, are still used
for heat exchangers, especially in pools utilizing brines or seawater. Brasses
are not used because of the risk of SCC from the ammonia (NH3) and/or
chloramines in the pool water. Largely, these units work well, but the author
has seen a few failures of 70/30 copper-nickel due to ammonia-induced hot
spot attack (see Chapter 9) from the pool water side.12 The problem can
be avoided by keeping the water velocity on the pool side high (>1.5 d s ) .
This means that pool water temperature control should be done on the hot
product side of the heat exchanger. It may be necessary to modify the heat
exchanger to a multipass configuration, rather than single pass, to obtain a
sufficiently high flow rate."
17.4 Desalination
With continuing industrial development in many parts of the world, the

demand for fresh water, particularly drinking water, is rising sharply. In
many areas, there are no further surface or underground sources of water,
and desalination, the removal of salt from brackish water or seawater,
must be considered. There are three main methods of desalination used
on a large, commercial scale: reverse osmosis (RO), MSF distillation, and
multiple effect distillation (MED).
Copper alloys have seen little use in RO plants, except that the main
seawater pumps are sometimes NAB. However, copper alloys are used
extensively in both MSF and MED plants.
17.4.1 Multistage Flash

In the basic MSF process, steam heats the recirculating brine in the heat
input section, and the steam condensate produced is returned as feed to the
boilers. Figure 17.1 shows a schematic of a typical MSF plant. The hot brine
(typically 1 10°C) then passes through a series of flash chambers in the heat
recovery section, each at a lower pressure than the preceding one. In each
chamber, some water vapor flashes off and is condensed on heat exchanger
tubes. In the last few heat rejection stages, cold, aerated seawater is pumped
through the condensers to lower the temperature for further condensation
to occur and to remove the heat from the heat input section and complete
the operating cycle.
Some of the raw seawater is discharged to waste and some is deaer-
ated and dosed with chemicals to control scaling and foaming, before being
added to the recirculating brine as feed makeup. The feed is necessary to
replace evaporative losses and to maintain the recirculating brine concen-
tration at acceptable levels. The brine is then heated in the heat recovery
condensers, before returning to the heat input section.
It is apparent from the preceding description that an MSF plant is
quite complex and presents a range of corrosive environments. However,
it can conveniently be divided into four main areas: the heat rejection
section, the heat recovery section, the brine heater, and the ejector section,
which handles the removal of incondensable gases. The flash chambers
are found in the first two areas and generally utilize the same materials of
construction. The heat recovery and brine heater sections operate with low
oxygen concentrations in the water ( - 4 0ppb), whereas the seawater in the
heat rejection side is fully aerated.
Vent
Heat
t *' I
DemiSter bistillste Flashing Condenser
trough brlne tube
Condensate return
HEAT
BRINE RECOVERY
HEATER SECTION
Decarbonator
. .
(when emDiovedl
SIMPLIFIED FLOW DIAGRAM
I LT-', Deaerator
Make up water chemical injection
of a multi-stage flash desalination plant (when employed)
BRINE HEATER i * HEAT RECOVERY SECTION HEAT REJECTION SECTION
Copper alloys are widely used for the heat exchangers in the heat
recovery and brine heater sections for tubes, tube plates, and sometimes as
linings for water boxes. Aluminum brass and 90/10, 70/30, and 66/30/2/2
copper-nickels are all used for tubing, while both NAB and rolled naval
brass are used for tube plates. The 90/10 copper-nickel is also sometimes
used as a lining for carbon steel (CS) water boxes.
Although all four tubing alloys have been used in MSF plants, 70/30
copper-nickel has had relatively little use, and aluminum brass tends to be
mostly used in the lower-temperature part of the heat recovery section. The
90/10 and 66/30/2/2 copper-nickel-iron-manganese are currently the most
widely used copper tubing alloys. The highest temperatures occur in the
brine heater (- 110°C), and 66/30/2/2 copper-nickel-iron-manganese has
been very successful as tubing in this application.
A survey of materials used in desalination plants was carried out under
A.D. Little in the 197Os.l3 More recently, in 2003, Oldfield conducted a
similar survey, and the results of these surveys are compared in Table 17.1.l 4
In Table 17.1, the materials used for heat exchanger tubes in the
brine heater, heat recovery, and heat rejection sections are compared. Note
that the 2003 survey divides the heat recovery section into two sections:
the high- and low-temperature sections. Table 17.1 averages the data from
TABLE 17.1 A comparison of the use of heat exchanger alloys in MSF plants
from the surveys of A.D. Little and Oldfield l 3 - l 4
197213 200314
Brine Heat Heat Brine Heat Heat
heater recovery rejection heater recovery rejection
Alloy (%) (“A) (Yo) (%) (%) (Yo)
Al-brass 17.9 68 23 0 25.7 5.6

90/10 Cu-Ni 49.0 24.5 34 27.7 52.8 18.8
70/30 Cu-Ni 1.5 1.0 1.5 7.9 11 4.2
66/30/2/2 Cu-Ni-Fe-Mn 3 1.3 6.0 39.5 51.8 5.2 26.5
Titanium 0.3 0.5 2.0 8.2 3.5 42.5
these two subdivisions. Table 17.2 shows the difference in materials selec-
tion for these two subdivisions from the 2003 survey. l4
The results in Table 17.1 show that the main material for the brine
heater tubes has changed from 90/10 copper-nickel to 66/30/2/2 Cu-Ni-
Fe-Mn. In the heat recovery section, aluminum brass has remained the
most popular alloy in the cooler regions, but 90/10 copper-nickel is the
predominantly used alloy in the hotter regions (Table 17.2). In the heat
rejection section, there has been a big change from copper alloys to titanium.
Although titanium is the most widely used single alloy, the copper alloys
together still account for a majority of the tubes. One feature of interest from
both surveys is the low use of 70/30 copper-nickel. The 66/30/2/2 Cu-Ni-
Fe-Mn alloy appears to be regarded as a significant and useful upgrade from
90/10 copper-nickel in some parts of the plant such as the brine heaters.
The A.D. Little survey also produced data on the failure rates of heat
exchanger tubes. Unfortunately, no such data were obtained by Oldfield.
Table 17.3 shows the A.D. Little survey results. These numbers need to be
compared with the usage data in Table 17.1. For example, the low failure
rate of 70/30 copper-nickel in the recovery and rejection sections is not
surprising, given its low use. The low failure rate of 66/30/2/2 Cu-Ni-Fe-Mn
TABLE 17.2 Materials used for the heat recovery section

heat exchanger tubest4
Usage (%)
Alloy Low temperature High temperature
Al-brass 46.7 4.7
90/10 Cu-Ni 47.1 58.4
70/30 Cu-Ni 0 21.0
66/30/2/2 Cu-Ni-Fe-Mn 0.4 10.0
Titanium 3.5 3.5
TABLE 17.3 Heat exchanger tube failure rates in MSF plants”

Failure rate (%)
Alloy Brine heater Heat recovery Heat rejection
Al-brass 15.7 1.1 6.8
90110 Cu-Ni 2.8 0.4 2.3
70/30 Cu-Ni 15.9 0 1.6
661301212 Cu-Ni-Fe-Mn 1.5 0.02 0.05
is significant, given the alloy’s high use. 90/10 copper-nickel is also shown
as a reliable alloy, as is aluminum brass in the heat recovery section.
Over the years, a wide range of failures have been described for
MSF plants. In the heat rejection section, copper alloy tubes have failed
from all the common causes seen in other seawater-cooled plants: erosion
corrosion due to lodged obstructions or excessive chlorination and pitting
due to the presence of pollutants such as sulfide or ammonia. The first two
can be controlled by improved filters and better chlorination control. The
effect of pollutants on copper alloys was described in Chapters 3,7, and 9.
The effects of ammonia can be mitigated by ferrous sulfate (FeS04.7H20)
dosing. When high levels of sulfide are present, ferrous sulfate dosing is
not beneficial, and a change of alloy may be required.
17.4.2 Multiple Effect Distillation

MED plants, like MSF plants, require a source of heat, which can be steam or
hot water (MED only) or electricity, and so are often run in conjunction with
a power station. In most MED plants, a thermocompressor or a mechanical
vapor compressor is added to improve efficiency. These are known as
thermocompression distillation (TCD) and mechanical vapor compression
(MVC) plants, respectively. The higher efficiency of TCD plants means
that they are the most common variant.
Figure 17.2 shows a schematic of a typical MED plant, in this case,
a TCD plant. A TCD plant consists of a number of evaporator stages (2 to
14), plus a final heat rejection section. Vapor at -75°C is introduced into
the first-stage evaporator tubes, where it is condensed by externally sprayed
raw water. In cooling the evaporator tubes, the raw water in stage 1 is heated,
and part of it vaporizes. The vapor enters the tubes of the second stage and
is condensed by raw water, as in stage 1, forming more condensate. This
process repeats itself in subsequent stages. Part of the vapor produced in an
intermediate stage is drawn up by the thermocompressor, which increases
its pressure and mixes the water vapor with high-pressure steam to feed
Operating Process
1 HPsteam
2 Air removal
3 Cooling water reject
4 Raw water feed

4@
5 Fresh water output
6 Brine blow-down
FIGURE 17.2 Schematic of the operation of a typical TCD plant. (See color section.)
the first stage. The remainder of the vapor passes to the following stages
and, finally, to a heat exchanger, where it condenses on the outside of tubes
cooled internally by raw water. Part of this raw water becomes the feed to
the evaporators, and the rest goes to waste. The product water is collected
and piped away, while the brine goes to blow-down. An MED plant operates
in the same way, but without the thermocompressor; in an MVC plant, the
thermocompressor is replaced with a mechanical vapor compressor.
A wide variety of materials are used in an MED plant, but copper
alloys are still the principal choice for the tubes in the evaporator section.
Titanium and high-alloy SS tend to be the first choice for the final heat
recovery heat exchanger, which operates under fully aerated conditions,
although 90/10 copper-nickel is sometimes used. In the evaporators, the
environment at the top is fully aerated, but oxygen levels decrease dramat-
ically in the lower sections. The highest temperature is currently -75°C
so aluminum brass and 90/10 copper-nickel perform well and are the most
common choice for the evaporator tubes. To increase efficiency, it is nec-
essary to use thin-walled tubes, so a -0.7-mm wall is common, and it is
important to provide adequate tube supports to avoid fatigue problems.
The tube plates in the evaporators may be copper-nickel or naval
brass, although Type 316L SS (UNS S31603) has also been used. Because
of the lower water temperatures, as compared with MSF plants, water boxes
tend to be rubber-lined steel.
Copper alloys have performed well in MED plants, although there
is an occasional problem with erosion corrosion in the top two rows of
tubes due to excessive velocity from the spray nozzles. This can be cured
by redesign of the seawater spray system, although titanium has also been
used in the top two rows to prevent this problem.
17.5 Oils and Oily Waters
Most oils are nonconductive and, as such, do not pose a corrosion problem
to copper and its alloys. For example, 90/10 copper-nickel is widely used
for hydraulic piping on offshore platforms because of its resistance to
the external marine atmosphere. In addition, 90/10 copper-nickel has been
used for hydraulic brake tubing on cars, in which its resistance to external
corrosion means that it will normally last the life of the car. Tungumt is
a 70/30 brass with 1% aluminum, and this alloy has also been used for
t Trade name.
hydraulic tubing because of its resistance to external corrosion. However,

Tungum can suffer SCC when ammonia or other aggressive species are
present (see Chapter 10).
Only where the oil can break down to release ionic compounds,
particularly acids, can internal corrosion occur. Provided that such oils are
used within their appropriate temperature range, where decomposition does
not occur, corrosion is not a problem.
When water is present with the oil, then two conditions can arise.
One is a condition in which the oil and water remain as separate phases,
with little mixing. The second case is one in which the oil is miscible with
the water or the oil and water are mixed in an emulsion, usually under
fast-flowing conditions.
In the first case, in which the oil and water remain separate, corrosion
can occur where there is sufficient water to wet the metal surface. This can
occur with very small water fra~tions'~--exactlyhow much depends on the
nature of the specific oil and the solubility of water in it; however, solubility
of water in most oils is low, and as little as 1% water can cause corrosion.'6
Even if the water and oil form an emulsion, the oil and water can separate
into their distinct phases if the flow rate decreases, and a series of stratified
layers can form.I7 Where a water layer contacts the metal surface, corrosion
can occur.
If the water is deaerated or has a very low DO content (t50ppb),
then corrosion will not be significant for alloys such as 90/10 copper-nickel,
gunmetals, and aluminum bronzes, provided that there are no other species
present that can provide an alternative cathodic reaction to the reduction of
DO, for example, carbon dioxide (CO2). If, however, the water is aerated,
then its composition and pH will determine the corrosivity. Any of the
forms of corrosion described in Chapters 4-14 may occur, depending on
composition, pH, flow rate, stress, and so on. Appropriate precautions to
prevent such corrosion should be taken, as suggested in these chapters. For
example, if Type I pitting is possible, then the tube bores must be free of
carbon films.
Where the oil-water mixture forms an emulsion at high flow rates,
the corrosion can be less than in the water a10ne.l~One reason is that the
conductivity of the emulsion is less than that of the water. Craig reported
corrosion of CS in some emulsions with as little as 1% water and no
corrosion in others with as much as 50% water." This is because the heavier
the oil, the more easily it forms an emulsion. l 7 Wetting of the metal by water
decreases and wetting by oil increases as the interfacial tension between the
oil and water reduces. This is why heavier oil emulsions with water are less
corrosive; that is, they can contain more emulsified water before separation
occurs.17
The lower the conductivity, the higher the oil content, and the less
easy it is for fresh reactants to reach the metal surface. However, emulsions
normally only form under turbulent conditions so that classic diffusion
control is not a rate-determining reaction. Following are the conductivities
for some different types of water, without oil, which might indicate the
types of corrosion problems that may arise:
Fresh water 50-1,000 pS

Brackish water 1,000-20,000 pS
Brines >20,000 ps
Seawater -40,000 pS
Although the effect of oil-water emulsions in reducing the corrosion rate of

CS has been studied, there are no published data for copper al10ys.l~While
the formation of an emulsion may reduce corrosion compared with the
water phase alone, it is advisable to carry out corrosion tests, particularly if
the water is aerated. Remember that it may be necessary to carry out tests
for more than one type of corrosion, depending on the water composition. In
addition, the oil and water will separate if the flow rate decreases sufficiently,
and corrosion will then occur wherever the water contacts the metal surface.
References
1 . BNF Metals Technology Centre Miscellaneous Publication, MP 586, August 1976.
2. M.S. Parvizi, A. Aladjem, J.E. Castle, Znr. Matel: Rev. 33 (1988): p. 169.
3. J.W. Oldfield, G.L. Masters, Copper Development Association (U.K.) Publication
115, June 1996.
4. H.J. Meigh, Cast and Wrought Aluminium Bronzes: Properties, Processes, and
Structures (London, U.K.: Institute of Materials, Minerals, and Mining, 2000).
5. EM. Reinhart, “Corrosion of Material in Hydrospace Part I V Copper and Copper
Alloys,” Naval Civil Engineering Laboratory Report, AD835,104, April 1968.
6. R. Francis, The Selection of Materials for Seawater Cooling Systems: A Practical
Guide f o r Engineers (Houston, TX: NACE International, 2006).
7. BS 287 1, “Specification for Copper and Copper Alloys Tubes: Tubes for Heat
Exchangers” (London, U.K.: BSI).
8. J. Birn, I. Skalski, “Corrosion Behaviour of Non-ferrous Alloys in Seawater in the
Polish Marine Industry,” in Corrosion Belzaviour and Protection of Copper and
Aluminium Alloys in Seawater, D. Feron, ed., EFC Publication No. 50

(Cambridge, U.K.: Woodhead and Maney, 2007): p. 3.
9. R. Johnsen, “Experience with the Use of Copper Alloys in Seawater Systems on
the Norwegian Continental Shelf,” in Corrosion Behaviour and Protection of
Copper and Aluminium Alloys in Seawater, D. Feron, ed., EFC Publication No. 50
10. W. Schleich, C. Powell, “ C u N 90/10: How to Avoid Typical Failures of Seawater
Tubing Systems and Marine Biofouling on Structures,” in Corrosion Behaviour
and Protection of Copper and Aluminium Alloys in Seawater; D. Feron, ed., EFC
Publication No. 50 (Cambridge, U.K.: Woodhead and Maney, 2007): p. 73.
11. H. Le Guyader, A.-M. Grolleau, E. Lumieux, K. Lucas, T. Wolejszu, “70/30
Copper-Nickel Piping Systems-Use of Descaling Agents and Their Effect on
Corrosion Properties,” in Corrosion Behaviour and Protection of Copper arid
Aluminium Alloys in Seawater, D. Feron, ed., EFC Publication No. 50
13. E.H. Newton, J.D. Barckett, J.M. Ketteringham, “Survey of Materials in Large
Desalting Plants Around the World,” A.D. Little Report to the U.S. Office of
Saline Water, March 1972.
14. J.W. Oldfield, “Novel Material Selection to Improve Corrosion Resistance,”
Report to the Middle East Desalination Research Centre, Oman, 97-AS-001,2003.
15. N.G. Galindez Ruiz, B.D. Craig, MP 34,4 (1995): p. 50.
16. B.D. Craig, Corrosion 54 (1998): p. 657.
17. C. de Waard, L.M. Smith, B.D. Craig, CORROSION 2003, paper no. 629
CHAPTER 18
Joining and the Corrosion of Joints
In many service applications, there is a need to join metallic components.

Where the joint is permanent, it is common with copper and its alloys to use
soldering, brazing, or welding. Each of these three processes is considered
in this chapter for copper alloys, and the potential corrosion risks of each
are described.
18.1 Soldering
For many applications where the stresses are not high, a joint made with a
low-melting temperature alloy will be satisfactory. Solders are traditionally
leaatin alloys with typical compositions around the eutectic (63/37 Sn/Pb)
to give a low melting temperature (190°C or less).
When joining copper, or its alloys, with solder, it is necessary to
use a flux to wet the surface. Traditionally, these have been based on
ammonium (NH4+) and/or zinc chloride (ZnC12). The flux residues after
soldering are not very soluble and are often acidic, which can result in deep
corrosion beneath the flux deposit; alternatively, crevice corrosion may
form at the edge of the residue. Fluxes used for electronics applications are
typically based on rosin or organic resins, the residues of which tend to be
relatively benign. However, if these are supplemented with more aggressive
ingredients, the residues may be active enough to cause electromigration
and subsequent electrical shorting in humid environments.
In some countries and in some industries, it is necessary to use a
soluble flux so that residues can be washed off after soldering. Some of these
still use a chloride activator and can be aggressive if not removed. Others
use an iodide activator, which is less aggressive than chloride.
353
A traditional composition for solders has been 60/40 Sn/Pb, with a

higher tin content where greater strength or resistance to fatigue is required.
Lead is toxic, and so it is no longer permitted in solders for joints in potable
water systems and in some areas in the electronics industry. However,
t i d e a d solders are still preferred for some electronics applications, for
example, in aerospace and military applications, where guaranteed long-
term reliability is critical to operations.
Where lead-free solders are mandatory, there are several popular op-
tions. For high-temperature applications, a tidcopper alloy containing -3%
copper may be used. This has a liquidus above 300°C and tends to be less
costly than the lower melting point alternatives that contain silver. Tin/
silver and related alloys melt in the range 210 to 240°C and have found use
in the electronics industry as tin/lead replacements. When using tin-based
solders on copper, it is important to avoid the formation of excessive copper-
tin intermetallic compound layers. Many of these are highly cathodic to
copper and may cause accelerated corrosion in aqueous solutions and are
also relatively brittle. Intermetallics are most likely to grow under pro-
longed heating such as can occur in rework or high-operating temperature
environments.
The high tin content of solders means that in most waters, a stable
film of tin oxide (SnO) will form over a wide range of pH values. This oxide
is very tenacious and corrosion-resistant. Erosion corrosion is sometimes
seen in copper piping systems carrying fresh water, immediately after sharp
bends or elbows. If the fitting has been soldered into place, sometimes a
protruding bead of solder increases the local turbulence. The erosion corro-
sion is always on the copper and never on the solder, although the potential
'
of solder is significantly more electronegative than that of copper. Francis
and Campbell believed that this was because tin oxide is more stable and
mechanically strong than copper oxide in fresh waters. In unpublished work
carried out at BNF Metals Technology Centre, they exposed samples of cop-
per coated with solder in the May jet test (see Section B.2 in Appendix B)
at 10 m/s, with 3% added air, in a hard borehole water containing 100 mg/L
chloride. No measurable erosion corrosion had occurred on the solder after
30 days at 15°C. These results demonstrated the high resistance of solders
to erosion corrosion in fresh water. Even in seawater, tin solders have good
resistance to both general corrosion and erosion corrosion.
Aluminum bronzes and copper-nickel alloys are difficult to solder be-
cause of their protective films. Where they must be soldered, it is necessary
either to coat the joint area in copper first, or to use a very aggressive flux
that can break down these films.
Joining and the Corrosion of Joints 355
18.2 Brazing
Brazing involves the joining of copper and its alloys with silver alloys
having melting temperatures in the range 600-800°C. Confusingly, the pro-
cess is sometimes called silver soldering, despite the fact that according to
conventional nomenclature, the temperature range clearly defines a brazing
process. Cadmium is no longer widely used in brazing alloys because of
its high toxicity, and so for practical purposes, the minimum melting tem-
perature of brazing alloys is about 640°C. A brazed joint requires a higher
temperature than a soldered joint, but it is also much stronger.
A range of copper-silver-phosphorus alloys is commonly used for
brazing coppers and brasses, and these alloys contain 0 to 15% silver and
4 to 7% phosphorus, depending on the melting point. The higher the silver
content is, the lower the melting temperature will be. These alloys are not
suitable for brazing other copper alloys, for which silver brazing alloys free
of phosphorus should be used.
As with soldering, it is necessary to use a flux, and the most common
contains borates and fluorides. The exception to this is brazing copper with
copper-phosphorus alloys; these are self-fluxing and will wet easily if the
surface is clean. Any residues when using borax flux are inert, and so the
corrosion problems associated with soldering fluxes are absent. However,
fluxes containing active ingredients, such as fluorides, must be removed.
Brazing alloys have good corrosion resistance in waters, including
seawater. The silver in the alloy generally means that the braze metal is
cathodic to the parent metal, and localized galvanic corrosion does not
occur.
Heavy oxide films formed during extensive brazing operations usually
consist of a thin layer of cupric oxide, with cuprous oxide underneath. Such
scales are highly cathodic to copper and can give rise to pitting corrosion at
the boundary between the scaled and unscaled parts of the metal surface.2
The severity of the attack will depend on the water composition, pH, and
temperature.
18.3 Welding
Welding involves joints made above the melting point of the parent metal.
Such joints involve the most heat but usually have the highest strength.
Sometimes the joints are autogenous; sometimes filler is used, often of a
matching composition to the parent alloy. For some of the copper alloys,
the selection of a filler metal is very important. Copper and its alloys can be
welded by all the common arc-welding techniques. In addition, autogenous

welding of thick (50 mm), oxygen-free copper cylinders has been achieved
by both electron beam welding and friction stir welding. In both cases, weld
deposits were produced that were not anodic to the parent metal. This is the
most important factor in the corrosion of welds. The weld bead has a much
smaller area than the parent metal, and so it is important that its potential
be the same as the parent metal or more positive. If this is not achieved,
rapid, localized attack of the weld metal can occur.’
The copper-nickel alloys are often used in seawater, and welding is
preferred for high-strength joints. The iron content of 90/10 copper-nickel
exceeds the solubility limit at room temperature (see Chapter 2), and so
iron may come out of solution if a matching composition filler is used and
the weld cools too slowly. This can reduce the corrosion resistance of the
weld metal, particularly the erosion corrosion resistance (see Chapter 7).
For this reason, it is common to weld 90/10 copper-nickel with 70/30
copper-nickel filler. The higher nickel content greatly increases the iron
solubility limit, and there is no risk of iron precipitation on cooling. Note
that slow cooling of thick section welds may cause iron to precipitate out of
solution in the heat-affected zone (HAZ). This can be avoided by suitable
joint design and the selection of welding parameters when qualifying the
welding procedure.
The 70130 copper-nickel alloys, such as C71500, and the single-
phase copper-nickel-chromium alloys are usually welded with a matching
composition filler. This produces welds that are not anodic to the parent
metal and so will not corrode preferentially. Alloy C7 1500 has been welded
with Alloy 400 filler, but this is generally unnecessary as it is more expensive
than matching filler and does not generally confer any significant corrosion
benefits.
Little et al. examined the corrosion of welded copper-nickel alloys in
seawater polluted by sulfide^.^ They concluded that any corrosion would
occur, not on the weld, but downstream of a girth weld, on the HAZ.
In an examination of the corrosion of welded Alloy 400, Lee et al.
carried out tests in seawater containing inorganic sulfide or sulfate-reducing
bacteria (SRB) culture^.^ The welds used a filler that was similar to the
parent metal, but with an increased silicon and manganese content plus -2%
titanium (UNS N04060). In the inorganic sulfide, there was, more or less,
uniform corrosion with no pitting anywhere, even after 20 weeks’ exposure.
Samples exposed to seawater containing SRBs had suffered pitting of the
parent metal after 12 weeks’ exposure. After 14 weeks’ exposure, there was
attack at the weld fusion line, but there was no attack of the weld metal after
20 weeks. This shows that the welds are no more susceptible to localized
attack in this environment than the parent metal.
The welding of the aluminum bronzes is somewhat more complex,
particularly for the multiphase alloys. The subject is discussed in detail by
Meigh and also by Spiller.s-6 Here a brief overview is presented, readers
wishing more information should consult the preceding references.
One problem with welding the aluminum bronzes is the ductility dip.
The alloys suffer a loss of ductility around 500°C, and the width of this dip
varies with alloy composition but can be extensive for some of the aluminum
bronzes.’ This means that cracking may occur on cooling, particularly
when the weld is under severe restraint. Meigh gave recommendations
for minimizing the risk of cracking, involving both the choice of welding
parameters and the selection of the filler metal composition.’
Spiller reported that there is a risk of HAZ cracking when welding
aluminum bronzes.6 However, this has not been seen for many years, and
it is believed that modern manufacturing methods produce alloys that are
no longer susceptible to HAZ cracking.’
When welding the single-phase aluminum bronzes, it is usual to use
matching composition filler. The alloy containing 8% aluminum and 3%
iron is sometimes preferred because iron increases the strength of the weld
metal and reduces the risk of ~ r a c k i n g Iron
. ~ also prevents the formation
of gamma 2 phase, which is anodic to the alpha phase and can corrode
preferentially.*
When welding nickel aluminum bronze (NAB), it is common to use
a matching filler. However, the weld bead cools very quickly, and not all
the beta phase present at high temperature may transform to alpha and
kappa phases. Beta phase is highly anodic to alpha phase and will cor-
rode rapidly in seawater (Chapter 9). Hence, a postweld heat treatment is
desirable to ensure that all the beta phase transforms. The most common
heat treatment is that specified by the Royal Navy for all welds to NAB.
This involves heating at 675°C for 6 hours, followed by air cooling. Where
postweld heat treatment is not possible, then the use of Cu/8A1/3Fe filler
is an alternative. Although no postweld heat treatment will be required,
the joint will not be as strong as one made with matching filler and may
not be as strong as the parent metal. The use of a single-phase filler does
not result in any preferential weld corrosion in seawater as the potentials
of both aluminum bronzes are similar. Figure 18.1 shows a seam-welded
NAB pipe that had not been postweld heat treated. After exposure to sea-
water, the beta phase corroded rapidly, with the result that the seam weld
unzipped.
FIGURE 18.1 Corrosion of the beta phase in an NAB seam-welded pipe that had not
been postweld heat treated. (Photograph courtesy of the Nickel Institute.) (See color
section.)
NAB has also been successfully joined by friction stir el ding.^ This
has the advantage of requiring no filler, but an assessment of the corrosion
resistance of such joints has not been made to date.
Provided that they have been heat treated, NAB welds do not generally
cause a corrosion problem in seawater. However, the Royal Navy has seen
corrosion of NAB, both welded and unwelded, when it was exposed to
slowly flowing seawater or alternate stagnant and flowing conditions for
extended periods. The work of Maselkowski and Francis, described in
Chapter 9, showed that attack was most pronounced in the HAZ of NAB
welds, with corrosion taking the form of selective attack of the kappa 111
phase.' As this phase is in long, thin stringers, the attack could be very deep.
However, if the NAB is put into flowing seawater such that a protective film
forms, then no selective phase attack occurs. This explains why NAB welds
give few problems in industrial service, where there is pressure to get
equipment into service rapidly.
References
1. R. Francis, Gulvanic Corrosion-A Practical Guide,for Eiigineers (Houston, TX:
2. H.S. Campbell, BNF Metals Research Association Miscellaneous Publication, MP

574, August 1972.
3. B.J. Little, P.A. Wagner, J. Jacobus, CORROSION/88, paper no. 81 (Houston, TX:
4. J.S. Lee, K.L. Lowe, R.I. Ray, B.J. Little, CORROSION103, paper no. 214
5. H.J. Meigh, Cast and Wrought Aluminium Bronzes-Properties, Processes, and
Structures (London, U.K.: Institute of Materials, Minerals, and Mining, 2000).
6. K.R. Spiller, “Practical Aspects of Welding Aluminium Bronze,” Welding and
Metal Fabrication, November 1971, p. 401.
7. M.F. Gittos, The Welding Institute, correspondence to author, 2006.
9. C.R. Maselkowski, R. Francis, BNF Metals Technology Centre Research Report,
R506 (Supplementary Report), July 1985.
FIGURE 1.1 Copper cladding on the towers of Barnberg Cathedral, Germany.
FIGURE 1.2 Copper wires being used in the construction of a modern, high-efficiency
electric motor. (Photograph courtesy of CDA.)
90 18
16
14
I-Cl -0 -Fe -Ni -CuI
FIGURE 3.2 Concentration of elements in the film on 90/10 copper-nickel vs. potential.
FIGURE 4.7 Appearance of films on copper tube exposed in a cuprosolvent water.

FIGURE 4.9 Appearance of a copper tube carrying so-called blue water. (Photograph
courtesy of P. Munn.)
FIGURE 5.1 Typical appearance of Type I pitting. (Photograph courtesy H.S. Campbell.)
FIGURE 5.2 Type I pitting at a solder flux run. (Photograph courtesy of P. Munn.)
FIGURE 5.3 Type I pitting in a water with high phosphate. (Photograph courtesy of
P. Munn.)
FIGURE 5.7 Typical appearance of Type II pitting. (Photograph courtesy of H.S. Camp-
bell.)
FIGURE 5.9 Typical appearance of Type 111 pitting. (Photograph courtesy of H.S. Camp-
bell.)
FIGURE 5.12 Typical appearance of Rosette corrosion. (Photograph courtesy of
P. Munn.)
(B)
FIGURE 6.1 Appearance of crevice corrosion on (A) 70/30copper-nickel and (B) Alloy
400, exposed to quiescent seawater. (Photographs courtesy of R.M. Kain.)
FIGURE 6.2 Appearance of 70/30copper-nickel tubes with O-ring crevices after one and
three cycles of 500 hours in flowing seawater followed by 500 hours of stagnant seawater.
(Photographs courtesy of R.M. Kain.)
FIGURE 7.1 Erosion corrosion of 15-mm copper pipe after a 90-degree elbow. (Photo-
graph courtesy of P.Munn.)
FIGURE 7.8 Selective corrosion of parent 90/10 copper-nickel pipe because most of the
iron was not in solution.
C70600 C71500 C72200 C71640
FIGURE 7.13 Maximum depth of attack around a blockage of various copper-nickel alloys
after 12 months in seawater.
FIGURE 7.23 Erosion corrosion at a scratch line on an aluminum brass tube, where
repassivation did not occur due to overchlorination.
FIGURE 7.29 Severe erosion of a gunmetal impeller due to a high level of suspended
solids.
FIGURE 8.1 Cavitation damage, showing typical jagged, honeycomb appearance.

FIGURE 9.1 Meringue dezincification blocking a duplex brass fitting. (Photograph cour-
tesy of Copper Development Association U.K.)
(B)
FIGURE 9.2 Dezincification of brass wire: (A) fractured wire; (B) microsection showing
extent of dezincification.
FIGURE 9.5 Appearance of NAB/superduplex stainless steel couples after 63 days’ ex-
posure: (A) natural seawater; (B)chlorinated seawater.
FIGURE 9.7 Half section of die-cast brass water meter body with fracture around base
due to dezincification.
TABLE 9.2 Relative susceptibilityto hot spot corrosion of some common

copper alloys
ALLOY UNS No. SUSCEPTIBILITY TO HOT SPOT ATTACK

Classical I Sulphide I Ammonia
Admiralty C44300 Erratic results I Erratic results I NT
Brass
Aluminium I C68700
Brass
I Intermediate I Intermediate I
NT = Not Tested
m
n
E
ln
ln
2 350
fj
300
250
L"" I
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time to Failure (h)
I + DGS1043 w CZ114 A CZ132 0 CZ121 A CZ122 + 70130 brass I
FIGURE 10.7 Applied stress vs. time to failure in a mercurous nitrate test for some stress-
relievedcopper alloys.
FIGURE 11.2 Typical appearance of a fatigue fracture, including beach marks.

FIGURE 13.1 The 5-year sequence of fouling on CS in seawater (exposure periods from left to right): (top) 3
months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.
FIGURE 13.2 The 5-year sequence of fouling on copper in seawater (exposure periods from left to right}: (top)
3 months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.
FIGURE 13.3 The 5-year sequence of fouling on 90/10 Cu-Ni in seawater (exposure periods from left to right):
(top) 3 months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.
FIGURE 13.4 The 5-year sequence of fouling on 70/30 Cu-Ni in seawater (exposure periods from left to right):
(top) 3 months, 9 months, 18 months; (bottom) 36 months, 48 months, 60 months.
.c. BULK SEAWATER
METAL
ANAEROBIC AEROBIC
BACTERIA BACTERIA
FIGURE 13.5 Schematic of biofilm on an immersed metal surface.
1 OOE-08
.
1 00E+02 -
0 10 20 30 40 50 60 70 80 90 100
Time (min)
1+1E7 +1E6 +1E5 +1E4 -1E3 +1E2 1

FIGURE 13.6 Effect of innoculum size of EMRSA-16 on the time for total kill when ex-
posed to copper.
1.00E+08
1.00E+06
T
E
CI
S
1.00Ec04
0
z
+.
0
m" 1.00E102
l.OOE+OO
0 50 100 150 200 250 300 350 400
Time (min)
[+Copper +Brass +Nickel Silver +304 SSI
FIGURE 13.7 MRSA viability on some metals at 20°C.

FIGURE 14.5 Galvanic corrosion of a NAB tube plate with titanium tubes.
. .
;*I/
.,
FIGURE 14.7 Severe corrosion of a copper pipe entering an SS hot water

cylinder.
FIGURE 14.9 Corrosion of a CS water box coupled to copper alloy heat exchanger tubes
and a tube plate.
TABLE 14.4 The risk of galvanic corrosion in a marine atmosphere for some common
metals’*
-
9) -
- 9 ) 9)
€ 3 gG5 S €
.-= = o w t n 8 - % . :
.5 2 E E .z .r: Q i ?
CI
E
E S i C j g a + j & 0 Q ) g
4 No risk of galvanic corrosion Large increase in corrosion
0 Some increase in corrosion 0 No data

P
FIGURE 16.1 Copper-covered spire on St. Magnus Cathedral, Kirkwall, Scotland.

(Photograph courtesy of the Copper Development Association.)
FIGURE 16.2 Brass escalator safety balustrades in a U.K. shopping center. (Photograph
courtesy of the Copper Development Association.)
FIGURE 16.4 Copper used in a decorative internal curtain wall. (Photograph courtesy of
NDM.)
BRINE RECOVERY
HEATER SECTION
SIMPLIFIED FLOW DIAGRAM Make up water chemical lnlectlon

of a multi-stage flash desalination plant (when employed)
1 -m I I HEAT RECOVERY SECTION I HEAT REJECTION SECTION I

FIGURE 17.1 Schematic of the operation of a typical MSF plant.
Op erating Process
l>® HPsteam
2 Air removal
.. ® 3 Cooling water reject
Raw water Ia ad
~0
5 Fresh water output
6 Brine blow-down
...
... ®
®
FIGURE 17.2 Schematic of the operation of a typical TCD plant.
FIGURE 18.1 Corrosion of the beta phase in an NAB seam-welded pipe that had not
been postweld heat treated. (Photograph courtesy of the Nickel Institute.)
FIGURE B.3 Campbell condenser tube rigs at BNF test site.

To all my colleagues at BNF
For your friendship, support, and the patience to pass on
so much of your expertise
APPENDIX A
Nominal Compositions of Some

Commonly Used Copper Alloys
The following tables give the nominal compositions of some copper alloys
commonly used in engineering. They are grouped by type, as described in
Chapter 2.
TABLE A.l Brasses

Nominal composition (wt%)
Form Cornrnonnarne UNSno. Cu Zn Pb As Sn Mn Other
Wrought Commercial bronze c22000 90 Bal(A) - - - - -
Red brass C23000 85 Bal - - - - -

Cartridge brass C26000 70 Bal - - - - -
Arsenical brass C26130 70 Bal - 0.04 - - -
Admiralty brass C44300 70 Bal - 0.04 1 - -
Aluminium brass C68700 70 Bal - 0.04 - - 2A1

Muntz metal C28000 60 Bal 0.5 - - - -
Naval brass C46400 61 Bal - - 1 - -
Hot stamping brass C38000 58 Bal 2 - - - -
DZR brass CZ 132(B) 62 Bal 2 0.06 - - -

Manganese bronze CZ114(” 58 Bal 1 - 0.7 1.5 0.8Fe
Cast DZR brass (diecast) BS EN 62 Bal 2 0.06 - - 0.5A1
1982‘’’
Manganesebronze C86550 58 Bal 0.5 - 0.5 2 1Fe
Beta brass C86200 60 27 - - - 3 3Fe
4A1
(A) Bal = balance.
(B) British Standard quoted as no UNS number.
361
362 APPENDIX A
TABLE A.2 Copper-nickel and nickel-copper alloys

Nominal composition (wtY0)
Form Commonname UNSno. Cu Ni Fe Mn Cr Other
Wrought 90/10 copper-nickel C70600 Bal(A) 10 1.5 1 -
70130 copper-nickel C71500 Bal 30 0.7 0.7 -
661301212 C71640 Bal 30 2 2 -
CuMilFeMn
Alloy 722 C72200 Bal 16 0.7 0.7 0.5
Alloy 7 19 C71900 Bal 30 0.3 0.7 2.5
Alloy 729 C72900 Bal 15 - - - 8Sn
Marine1 C72420 Bal 18 0.8 4 - 2A1
0.7Nb
Alloy 400 NO4400 Bal 65 2 2 - -
Alloy K-500 NO5500 Bal 65 1.5 1 - 3A1

0.4Ti
Cast IN768 (B) Bal 30 0.7 0.6 1.6 -
(*) Bal = balance.
(B)No UNS number.
TABLE A.3 Tin bronzes and gunmetals

Form Common name UNSno. Cu Sn Zn Pb P Other
Wrought Phosphor bronze C50900 Bal(A) 3 - - 0.2 -
Phosphor bronze C51000 Bal 5 - - 0.2 -
Phosphor bronze C52100 Bal 7 - - 0.2 -
AP bronze (B) Bal 8 - - - 1Al
0.1Si
Cast CT1 phosphorbronze C90700 Bal 10 - - 0.1 -
LB2leadedbronze C93700 Bal 10 - 10 0.05 -
Admiralty gunmetal C90600 Bal 10 2 1.5 - -
LG2 gunmetal C83600 Bal 5 5 5 - -
LG4 gunmetal C92410 Bal 7 2 3 - -

M bronze C92200 Bal 6 4 1.5 - -
(A) Bal = balance.
(B) No UNS number.
TABLE A.4 Silicon bronze

Nominal composition (wt‘70)
Form Common name U N S no. Cu Si Mn Zn Other
Wrought Siliconbronze C65500 Bal(A) 3 1 - -
(A) Bal = balance
Appendix A 363
TABLE A.5 Aluminum bronzes

Form Common name UNS no. Cu Al Fe Ni Other
Wrought Copper-aluminum C60800 Bal(A) 5.5 - - 0.3As
Alloy D C61400 Bal 7 2.5 - -
Alloy E C63300 Bal 10 3 5 -

NAB C63200 Bal 9 4 4.5 -
Cast AB 1 C95200 87 9 3 - -
NAB C95800 Bal 9 4 4.5 -
CMA 1 (B) Bal 8 3 2.5 12Mn

(A) Bal =balance.
(B)No UNS number.
TABLE A.6 Nickel silvers

Form Common name UNS no. cu Ni Zn Pb
Wrought 45110143 C77400 45 10 BdA) 1.5
40114/44 (B) 40 14 Bal 1.5
63110127 C74500 63 10 Bal -
63112/25 C75700 63 12 Bal -
63/15/22 C75400 63 15 Bal -
63118119 C75200 63 18 Bal -
5511 8/27 C77000 55 18 Bal -
63120117 (8) 63 20 Bal -
58/25/17 (B) 58 25 Bal -
(A) Bal =balance.
(B)No UNS number.
APPENDIX B
Laboratory Tests to Evaluate

the Erosion Corrosion Resistance
of Copper Alloys
When examining laboratory erosion corrosion data, it is often difficult to

compare results from different test methods, employing different velocities.
This appendix briefly summarizes the most commonly used test methods
and what can be gleaned from the data.
B.l Test Methods

The test most representative of service is a flow loop or model condenser,
operating at normal flow velocities or maximum design velocities. Although
this reproduces field conditions, exposures need to be long (6 months to
2 years) to generate useful data. Sometimes these tests are accelerated by
increasing the flow velocity to, say, twice design to produce data more
quickly. Erosion corrosion usually then occurs anywhere that local turbu-
lence is increased such as after sharp bends. This must be taken into account
when considering the data, for example, not rejecting an alloy because it
showed a little attack. At normal flow rates, this alloy could be perfectly
satisfactory.
Most erosion corrosion data have been generated by a method involv-
ing much higher flow velocities. The oldest of these is the May, or BNF
jet test, in which water is fed to a circular water box designed to avoid
cavitation.' Water is fed via flexible hose from the water box to multiple
jet assemblies (12 or 24), which are submerged in a tank of the water being
considered. These rigs can be used in either once-through or recirculating
mode. The jet assembly consists of a block holding a nozzle, with the test
sample fixed a few millimeters from the jet (Figure B. 1). The distance from
365
366 APPENDIX B
TEST SAMPLE
I JET NOZZLE
t
WATER
FIGURE B.l Schematic of the sample holder in the BNF jet test.'
the jet to the sample must be set at exactly the right distance to ensure that
the sample is in the most turbulent part of the jet.' Jet velocities are usually
in the range of 5-10 d s , with tests lasting for 30 days. Note that under
recirculating conditions, it is usual to inject 3% air into the water. This is
discussed in more detail in the next section.
The LaQue Laboratory, in the United States, used a variant of the BNF
jet test, which involved using a cylinder as a high-pressure water box. The
jet nozzles were fastened around the base of the cylinder, with the samples
held a fixed distance from the jet. The whole assembly was immersed in
a tank of seawater. The advantage of this test rig design was that it could
operate to even higher velocities than the BNF jet test, without cavitation
in the water box.
The jet test is used to simulate turbulent conditions that can occur at
condenser tube inlets or at partial blockages within tubes. Sat0 and Nagata
showed that flow velocities around a blockage could be up to five times the
nominal flow ~ e l o c i t y . ~
Thus,
- ~ for a nominal flow velocity of 2 m/s, the
velocity around a blockage could be up to 10 m/s.
The parallel flow rig was designed to study the shear forces that are
required to remove protective films and initiate erosion corrosion. Samples
are placed in the center of a long pipe, parallel to the axis, with nonconduct-
ing spacers to separate each sample. By using long, straight pipes, uniform
flow is produced so that the shear force can be calculated.
Campbell designed a test rig that includes not only an impinging jet,
but also crevices, slow flow, and heat t r a n ~ f e r . ~The
- ~ condenser tube test
rig uses a water box identical to that of the BNF jet test, but this feeds
10 nozzles that have short (200-mm) lengths of 25.4-mm-outside-diameter
Appendix B 367
I -
WARM WATER CREVICE
SLOW MOVING WARM WATER -
HEAT TRANSFER - mfl HEATER BLOCK
CONDENSER TUBE -
SLOW MOVING COLD WATER
COLD WATER CREVICE - IMPINGEMENT ZONE
INLET NOZZLE -
FIGURE 8.2 Schematic showing the various corrosion zones in the Campbell condenser
tube test rig.4.5
condenser tube over them, as shown schematically in Figure B.2. Fig-

ure B.3 shows several rigs in parallel at the BNF test site. In this test, the
heaters were not fitted. These rigs have mostly been used with jet velocities
in the range of 5 to 10 m/s in a once-through mode."' The inclined angle
of the jet, compared to the BNF jet test, does not seem to produce levels of
erosion corrosion very different to the levels seen in the jet test under the
same operating conditions.
When it is desired to specifically study the effects of a partial blockage
on the performance of heat exchanger tubes, Gaffoglio described a plastic
insert that can be fixed in a condenser tube to simulate this effect.8
368 APPENDIX B
FIGURE 8.3 Campbell condenser tube rigs at BNF test site. (See color section.)
B.2 Recirculating Vs. Once-Through Testing
Some test rigs use recirculating water and some use once-through water,
but it is important to recognize that erosion corrosion testing in synthetic
seawater (e.g., ASTM D1 1419)is not the same as testing in natural seawater.
This has been clearly demonstrated and is thought to be due to a combination
of the organic material and fine suspended solids (typically 50 mg/L) present
in natural seawater."-" In Chapter 3, the different films that form on some
alloys in natural and synthetic seawater are discussed.
Gilbert and LaQue carried out extensive comparative tests with the
BNF jet test under both once-through and recirculating conditions. They
found that once-through seawater was more aggressive to some copper
alloys compared with recirculating tests. This was thought to be due prin-
cipally to very fine abrasive particles from plankton-either their bodies or
their hard shells. Tests with once-through seawater gave similar results to
recirculating tests if the jet velocity in the once-through tests was reduced
to about one-third the jet velocity in the recirculating tests. The results from
the recirculating tests gave similar depths of attack to those sometimes seen
at the inlets of copper alloy condenser tubes. Although the conditions were
more erosive under once-through conditions at jet velocities of 5 m/s or
more, Gilbert and LaQue felt that these velocities were useful for studying
the more erosive conditions that can sometimes develop, for example, at a
partial blockage in a condenser tube.
Appendix B 369
It was also found that the addition of 3% air only had an effect when
no erosion corrosion occurred without it. The extra turbulence introduced
by the addition of air can then induce some erosion corrosion. As entrained
air is quite common in seawater cooling systems, it is useful to add it when
conducting jet tests in a recirculating mode. It would only be necessary to
add air under once-through conditions if it was being conducted at a very
low jet velocity.
References
1. R. May, R.W. DeVere Stacpoole, J. Inst. Met. 77 (1950): p. 33 I
of Copper Alloys and Titanium,” Conference on Corrosion in the Power Industry
(Montreal, Canada: NACE International, 1977).
3. S. Sato, K. Nagata, Sumitomo Light Met. Tech. Rep. 19 (1978): p. 83.
4. H.S. Campbell, BNF Metals Research Association Miscellaneous Publication,
MP577, February 1973.
5. H.S. Campbell, “The Campbell Condenser Tube Apparatus,” Third International
Congress on Marine Corrosion and Fouling (Gaithersburg, MD: 1972).
6. R. Francis, M P 21 (1982): p. 44.
7. R. Francis, Brit. Corms. J. 20 (1985): pp. 167, 175.
8. C.J. Gaffoglio, in Power Engineering, April 1982, p. 60.
9. ASTM D1141, “Standard Practice for the Preparation of Substitute Ocean Water”
(West Conshohocken, PA: ASTM International).
10. P.T. Gilbert, EL. LaQue, J. Electrochenz. Soc. 101 (1954): p. 448.
11. P. Gallagher, R.E. Malpas, E.B. Shone, Brit. Corros. J. 23 (1988): p. 229.
INDEX
Index Terms Links
acetic acid 44 50–52 80

82 91
admiralty brass
atmospheric exposure and 202 204
carbon films 69 70
composition of 361
condenser tubes and 52 206
cracking and 202
dezincification and 7
erosion corrosion and 112 114–115 117
ferrous sulfate and 23 134
fresh water and 112 114–115
hot spots and 190
jet tests and 117
pitting and 69
SCC and 326
seawater and 17 157 158t
temperature and 34 36t
tin and 7
underground tests 312
See also specific topics
This page has been reformatted by Knovel to provide easier navigation.

Index Terms Links
admiralty gunmetal 9 132 169

362
aeration 50 52 54
acetic acid and 50
aluminum bronze and 47
flow rate and 279–280
potential and 269 271
sulfuric acid and 45–46
alkalinity
alkalis and 50–52 53
bimetallic corrosion and 283
erosion and 113
pitting and 66 76 263
SCC and 201 216
soils and 305–306
See also pH value
alloy 400 13 257
acetic acid and 50–52
alkalis and 50–52
biofilms and 79
caustic soda and 50
cavitation and 171
crevice corrosion and 13 96f 99
102t 103t 106t
electropositive potential 130
erosion corrosion and 130
formicary corrosion and 50
galvanic corrosion 326

Index Terms Links
alloy 400 (Cont.)

HF and 48
hydrochloric acid and 48
isocorrosion curve for 51
jet tests and 117–118
MIC and 256–257
nickel-copper alloy and 13
nitric acid and 47
pitting and 78–80 90
seawater and 34 78–79 79f
95 238 256
284
SRB and 257 261 346
sulfides and 256
sulfuric acid and 46 47
temperature and 34 36f
alloy 722
composition of 362t
crevice corrosion and 99
erosion and 119 127 131
147 153 154f
158 160
heat exchangers and 12
hot spot corrosion and 190
alloy 729 12 362

Index Terms Links
alloy K-500
age-hardening and 340
alloy 400 and 13
cavitation and 169 170 171
mechanical properties of 6
seawater and 237 239f 276
aluminum brass 17–18 78
arsenic and 174 215f
atmospheric corrosion and 327f
bronze and 10–11
See aluminum bronze
chlorine and 25 140 141f
147
copper-nickel and 12 121
corrosion fatigue and 226 228f 231
erosion corrosion and 127 140 141f
142f 150–151 150f
152f
films and 17 135 136
137
hydrotalcite and 22
iron dosing and 136 136f 137f
pitting and 144
power stations and 149
redox potential 261
sand and 154
SCC and 213 215

Index Terms Links
aluminum brass (Cont.)

seawater and 228f 231 253
sulfides and 144 147 154
159 160 191
surface finish 70
tubes and 70 141f 160
zinc and 290
aluminum bronze 8 22 255
acids and 46– 47
brass and 179
cavitation and 169 171
citric acid and 50
composition of 363t
dealumnification and 179 189
ductility dip and 357
HAZ cracking 357
HF and 48
pH values and 201t
sand and 156
SCC and 201 214 221
seawater and 285 286
underground corrosion and 312
welding and 357
aluminum chloride 83
amines, SCC and 196
amino acids 77

Index Terms Links
ammonia 51
aeration and 51
airborne 195
alkalinity of 51
ammonium chloride 202
atmospheric corrosion and 320
condenser tubes and 51
corrosion fatigue and 227 231
ferrous sulfate and 24 138 346
fog tests 52
fog tests and 52t
fresh water and 335
galvanic corrosion and 83
hot spots and 24 138 186
188 191
MSFplants and 138 191
nitrates and 209–210
pH control and 51
rural areas and 342
SCC and 13 51 196
197 200 203t
205 209 217
221–222 231 320
349
seawater and 202 231 231f
232f

Index Terms Links
ammonia (Cont.)
spray tests and 52t
swimming pools and 342
arsenic
brass and 176 214–215 215f
dezincification and 7 8 174
176
pitting and 59
SCC and 7 199 213
215 215f
atacamites 16
atmospheric corrosion 201t 317–333
acid rain and 320
aerosol particles and 320t
aluminum and 326–327 327f
ammonia and 320
See also ammonia
bimetallic data 292
carbon dioxide and 320
cathodic reaction in 321
chlorides and 292
common metals and 293 293t
condensation and 319
corrosion classes 322 323 323t
324t 325
See also dezincification

Index Terms Links
atmospheric corrosion (Cont.)

exposure tests 201 201t 323
323
factors in 291 317–321
galvanic corrosion and 291–294 326
humidity and 291 328
indoors 327–329 328
industrial locations 321 323
ISA classification 328 329t
low-conductivity and 326
methods of prevention 330–333
nitrogen oxides and 320
oxygen and 319
printed circuit boards and 329
rainwater and 319
SCC and 325
site types 321
sodium chloride and 321
sulfur dioxide and 322
temperature and 292
wetness classes 322t
bacteria 42 53 139
anaerobic 143
biofilms and 77–78 251
See biofilms
biofouling. See biofouling
Index Terms Links
bacteria (Cont.)
blue water and 42
chlorine and 139 260
copper toxicity 257–258
E. coli 258
exopolymers and 76–77
hydrogen sulfide and 143
iron-oxidizing bacteria 253
L. pneumophila 258
rnacrofouling 243–251
microfouling and 243–251
MRSA and 259
pH and 258
pitting and 89–90 263
polybutylene and 259
seawater and 243 252–260
SRB. See sulfate-reducing bacteria
superbugs 259
toxicity and 257–260
ultraviolet radiation 89
beach marks 226 227f
benzotriazole 108
beryllium 127
bimetallic corrosion. See galvanic corrosion
biofilms 7
alloy 400 and 79
bacteria and. See bacteria
biofouling 251–252 251t 342

Index Terms Links
biofilms (Cont.)
fouling and 251–252
galvanic corrosion and 257 273–274 285
MIC and 251
pipe surfaces and 258
255
polysaccharides 77
253 273–274
stainless steel and 257
temperature and 272
biofouling 251–252 251t 342
See also biofilms
blue water 42 43 43f
53 255 336
BNFjet tests. See jet tests
brass 1 7–8 15–17
114 176
acetic acid and 50
admiralty brass and. See admiralty brass
alkaline solutions and 216
aluminum brass. See aluminum brass
arsenic and 7 174 176
361
beta phase 7 176 213

Index Terms Links
brass (Cont.)
brazed joints 173 355
cartridge cases 195 204 325
citric acid and 50
commercial bronxe 8
composition of 361t
corrosion fatigue and 226–227
cuprous oxide and 17
dealloying and 173 176
dezincification and 173 174
See DZR brass
DZR. See dezincification-resistant brass
ferrous sulfate and 23
fresh water and 112 335
hydrochloric acid and 46
lead and 8
mercury and 209t
naval brass 176 285 298
344 348
red brass and 114
SCC and 199 205 209
213 335
See also SCC
70/30brass 17 117 199
202 208 210
214 216 253
325

Index Terms Links
70/30brass (Cont.)
single-phase 199
sodium nitrate and 210
sulfur dioxide and 212 222
sulfuric acid and 45
tin and 7
tungum and 348
zinc and 7 176
See also specific types, topics
calcium carbonate 16 62 68
calcium hydroxide 51
Campbell tube tests 99 135 142f
367f
cannons. 9
carbon dioxide 84
atmospheric corrosion and 320–321
erosion corrosion and 113 113t
galvanic corrosion and 272 282
pitting and 63 89–90 349
soils and 303
well water and 113t
carbon films 69–70
carbon steel 54 237 245f
carboxylic acids 80
cast alloys 117 146
Index Terms Links
caustic soda 50 51f

cavitation 165–172
alloy K-500and 171
bubbles 168
cavitation tunnel 168
copper alloys and 168–171
cracking and 167
four stages of 165 166f
honeycomb appearance 165 166f
manganese aluminum bronze and 171
manganese bronze and 171
NAB and 169 171
nozzle and 168
orifice plate 171
pitting and 165–167
pressure and 165
pumps and 165 171
seawater and 169 170t
ships and 165
test methods 167–168
ultrasonic tests 169 169t
vapor cavities and 165
vibration and 165
charcoal 59

Index Terms Links
chlorine 25–26 79 138

aluminum brass and 25 140 141f
147
bacteria and 260
Campbell test rig and 142f
cathodic reactions 26 184 192
260
copper chloride and 26
copper-nickel and 147 148f 149
crevice corrosion and 100 101 103
Cu-Ni-Fe-Mn alloy 140
disinfection regimes 261
drinking water and 261
effects of 139–143
erosion corrosion and 148f 261
159 261
films and 25 140
fouling and 25
fresh water and 336
heat exchangers and 139 159
See also heat exchangers
in hot water 72
hypochlorite and 100 138 159
191 260
NAB and 143
73 101 261

Index Terms Links
chlorine (Cont.)
redox potential 261
safe levels 141
seawater and 23 100 138
143 160
SRB and 261
sulfide and 141 147 149
See also specific materials
chromium
atmospheric corrosion and 293 294t
alloy 722 and. See alloy 722
as-cast condition and 125
beryllium and 153
copper-nickel alloys and 12
molybdenum and 273 283 284
seawater and 297
welding and 356
citric acid 50–51 53
coatings 314 331
contact corrosion. See galvanic corrosion
copper alloys, compositions of 361–363
See also specific alloys
copper carbonate 16
copper chloride 26
copper hydroxycloride 244
copper-manganese-aluminum alloys 179

Index Terms Links
copper-nickel alloys 12
aging and 119
alkalis and 50
aluminum and 12 121 216
antifouling properties 249–251
area ratio and 50
beryllium and 127
bicarbonate and 39
cathodic polarization 28–29 28f 255
chlorine and 25 147 148f
149
chromium and 125 126f 127
common alloys 11–12 362t
corrosion fatigue and 227–232 231f 232f
crevice corrosion and 95 96f 98–88
100 100t 101t
105f 106
cuprous oxide on 18
denickelification and 184
erosion corrosion and 118 119 121
125 129 129t
132 148f 157–158
fatigue and 216–218 230
films and 18–22 20f 70
133–134 255
See also films

Index Terms Links
copper-nickel alloys (Cont.)

galvanic corrosion and 279 285 296
heat exchangers and 24 69 82
144 149 184
191 348
HF and 48
hydrogen embrittlement and 242–242
ironand 11 19 119
121 124 132
mercurous oxide tests 219 233
MIC and 256
molybdenum and 121 124
90/10 alloy. See copper-nickel (90/10) alloy
niobium and 12 121 124
124t 125f
oil platforms and 232 285
protective film on 255
reverse bend tests 229f
sand and 151 154
SCC and 53 202 203t
216–218 219
seawater and 19f 29 34
99–100 127f 129t
144 l53 229
231f 247f 252
338 341 356
See also seawater; specific topics

Index Terms Links
copper-nickel alloys (Cont.)

70/30 alloy. See copper-nickel (70/30) alloy
ships and 144 249 250
silicon and 124 124t 125f
single-phase 196 206 216
sodium sulfite and 262
soldering and 354
SRB and 255 256
strength of 12 202 216–218
147 149 154
256
temperature and 21 37 132
tin and 12
tubes and 24 51 138
145 191 348
See also tubes
Vibraphone tests and 229
welding and 121 250 356
zinc and 127
zirconium and 124
See also specific alloys, topics
copper-nickel (90/10) alloy
alloy 400 and 130
aluminum in 124
ammonia and 231
antifouling properties of 249–250

Index Terms Links
copper-nickel (90/10) alloy (Cont.)

area ratio and 266
bends and 134
bimetallic corrosion 282 285
biofilms and 244 255
cathodic process and 19 26 27
28t 282 298
cavitation and 171
chlorine and 141 147 160
296
chromium and 127
crevice corrosion and 98 100
dealloying and 284
desalination plants and 138 191 344
erosion corrosion and 118 119 120–121
128 130t 153
157 160
fatigue and 232
films on 29 70 133
138
202 216 338
345–346 348 360
iron in 119 121 132
144 145 356
niobium in 124
quenching and 11

Index Terms Links

sand and 151
SCC and 202 209
seawater and 19t 21 34
34t 36 128
129 134 170
231 251–252 295
337 342 360
SRB and 256 262
159 256 262
338
temperature and 38 38t 39
turbulence and 277
welds and 356
See also copper-nickel alloys; specific
topics
copper-nickel (70/30) alloy 18–22 362t
aluminum and 7 145
admiralty brass and 17
ammonia and 52 342
arsenic and 199
bimetallic corrosion and 282 298
biofilms and 257
cavitation tests 167
chromium and 125
citric acid and 50

Index Terms Links

crevice corrosion and 95 98t 100
101 103t 106
127
dealloying and 184 325
desalination and 103
erosion corrosion and 117 128 129t
130 157
fatigue and 230 231 232t
films and 20 21 133
157
fouling resistance 12 244 250
HF and 48
hot spots and 185 191
ironin 11 118 121
132 138 144
151
MIC and 256
niobium and 124 124t 125
reverse bend tests 229
RO conditions and 104
sand and 161
SCC and 231

Index Terms Links

seawater and 29 34 36t
96 138 158
229 253 254
356
silt and 151
strength of 124 171
sulfide and 144 159
temperature and 37 37t 38t
titanium and 346
tubes and 23 127 344
welds and 339 356
zirconium and 124
See also copper-nickel
alloys; specific topics
corrosion fatigue 225–236 231
aluminum brass and 226 228f 231
ammonia and 227
See also ammonia
anodic currents 234
beach marks 226 227f
brass and 226–227
closed-loop system and 235
copper-nickel and 227–232 231f 232f
rnultiphase alloys and 280
NAB and 233–234 235
offshore platforms and 232
organic coatings and 235

Index Terms Links
corrosion fatigue (Cont.)

peak stress and 225
prevention of 234–235
R value 225
S/N curves 225 226f
See also seawater
sulfides and 27
sulfuric acid and 47t
sunlight and 16
temperature and 36 37 38f
See also specific alloys, environments
coupling, of metals. See galvanic corrosion
crevice corrosion 95–109
alloy 400 and 13 96f 99
102t 104
avoidance of 108
benzotriazole and 108
Campbell condenser tube tests and 99
cathodic reaction 97
copper-nickel and 95 96f 100t
101t 106
copper-nickel-iron alloys 14 344
See also Cu-Ni-Fe-Mn alloy
crevice formers 101–102 101t 102t
depth of 98t 99 100t

Index Terms Links
crevice corrosion (Cont.)

flowing conditions 102 103t 106t
fresh water and 97
kappa III phase 106 107f 108
lay-up conditions 104
mechanism of 95–97
NAB and 106 108 182
O-rings and 102–105 105f
107f
polluted waters and 99
quiescent conditions and 98
river waters and 97
seawater and 97–108 98t
ships and 98
sodium chloride and 106 107f
stagnant conditions 106t
sulfide and 99
tubes and 103
Cu-Ni-Fe-Mn alloy 152 185
chlorine and 140
crevice corrosion 127
erosion corrosion 132t 142t 144
147 149 160
films and 70

Index Terms Links
Cu-Ni-Fe-Mn alloy (Cont.)

hot spots and 190 190t
MSF plants and 345 346t
NAB and 158
sand and 151
sulfides and 149
cuproammonium complexes 51
cuprosolvency 4043 41f 53
cuprous chloride 67 68
cuprous oxide 15 16 17
58 68
dealloying 173–193
See specific types
dealumnification 179–184
aluminum bronze and 179 189
NAB and 189
denickelitication 184
desalination plants 21 340 343
See specific types
dezincitication 173–179
anodic reaction and 173
arsenic and 174 176
avoidance of 188
of beta phase 176
brass fittings and 173
Index Terms Links
dezincitication (Cont.)
of brass wire 175f
brazing and 177
cathodic reaction and 173
copper-zinc equilibrium diagram 176
crevice corrosion and 173 174f
die-cast brass and 189f
DZR brass. See dezincification-resistant
brass
fresh water and 335 336
ground water and 177
gunmetals and 174
plugs and 173
resistance to 189
seawater and 177
softened waters and 177
tin and 176
water composition 178f
dezincitication-resistant (DZR) brass 176
arsenic and 8
cracking in 200
invention of 12
Mattsson’s solution 196 197f
mercury and 208
SCC and 197f 199
disinfection 44–45

Index Terms Links
dissimilar metal corrosion. See galvanic

corrosion
dissociation constant 82
drinking water 41 78 215
261 343
DZR brass. See dezincitication-resistant brass
electrochemical tests 84
electrolytes, conductivity of 275
electronic components 329 331 332
ennoblement 21
environmentally assisted cracking. See stress
corrosion cracking
erosion corrosion 11 23 111–164
aging and 119
aluminum and 127
aluminum brass and 141f 150f
annealing and 119
area ratio and 276–279 277
BNF jet test and 367
bronzes and 117
cast alloys and 117
cathodic protection 149
cavitation and 165
chlorine and 148f 261
chromium and 12 125
cold rolling and 119
Index Terms Links
erosion corrosion (Cont.)

corrosion products and 130
elbows and 112f 116
films and 130 149
flow velocities and 54
in flowing systems 157
fouling and 243
fresh water and 112–116
gunmetals and 117 150f
heat exchanger tube and 140
high-velocity tests 133
iron and 11 19 118
119 153 153f
jet tests and 131 132t 149
laboratory tests for 365–369
MED plants and 348
molybdenum and 124
NAB and 117 129t 155f
158
nickel-copper alloys and 118 119 121
122f 123f 125
129 129t 132
148f 158
See also copper-nickel alloys
nickel silvers and 127
niobium and 124

Index Terms Links

oil platforms and 341
once-through water 368
phases of attack 151
physical factors in 128
pitting and 111 261
See also pitting
recirculating water 368
resistance to 119 131 153
365–369
by sand 153f
sand and 151 155f 160
See also sand
seawater and 116–134 116t
See also seawater
shear stress and 131t
silt and 160
sulfide and 148f 155f
suspended solids and 149 150f
synergistic effects 149
temperature and 128 134
turbulence and 121 125 130
159 277 339
354 369
velocity and 115f

Index Terms Links

welding and 120
well water and 113t
See also specific materials, environments
Evans diagram 27
fatigue. See corrosion fatigue

ferrous sulfate 23–25 138
aluminum brass and 138
ammonia pollution 24
261
hypochlorite dosing and 138
iron and 134–139
MSF plant and 24 191
pretreating and 150f
films
air-formed 22
aluminum brass and 17 135 136
137
bacterial. See bacteria
biofilms. See biofilms
carbon films 83 338
cathodic polarization 26 255
chloride-rich 20
Index Terms Links
films (Cont.)
chlorine and 25 140
copper and 40
copper-nickel and 20f
corrosion resistance and 22
cuprosolvency and 41f
cuprous oxide 20
depolarization and 29
efficiency of 137
erosion and 130 149
film-water interface 16
formation of 15–31
heat exchanger tubes and 84
hydrotalcite and 253
hygroscopic 196
iron-rich 23 135
manganese dioxide and 137
microbiological 42
oxide layer 19
pH and 39
phosphates and 43
pitting and 57
polarization and 27
SCC and 196 210
seawater and 17 22
shear stress and 131 157
sulfide-oxide 29

Index Terms Links
films (Cont.)
temperature and 271
See also specific types
firewater systems 134
fluorescence 58 62
fog tests 51 52t
formaldehyde 82 90–91
formicary corrosion 50 80–42 90–91
fouling 243–265
antifouling paint 249
bacteria and 253
See bacteria
biofilms and 251
See biofilms
biofouling and 251t
chlorine and 25
coating 249
copper-nickel and 247f 248f 251t
macrofouling and 243
microbial. See bacteria
microfouling and 243
resistance to 244
seawater and 25 246f 247f
248f 251 260
ships and 249 250
SRB and 253
See also specific types, materials

Index Terms Links
fresh water 335

blue water 255
chlorination and 336
cold water systems 336
compatible materials in 289–290 290t
coupling in 290–291
electrochemical series for 288f
flow velocities 53
galvanic corrosion and 287–291
hardness of 287 290–291
heating systems and 336 337
overpumping and 337
pitting and 57
SCC and 335
seawater and 33–40 287
soft waters and 287
temperature and 290–291
velocities for 114t
galvanic corrosion 267–301 277f 326–327

aeration and 279–280 299
alloy groups 283–285
area ratio and 281
atmospheric corrosion and 291–294 326
cathodic protection and 268
Index Terms Links
galvanic corrosion (Cont.)

coatings and 297
conditions for 268–269
copper-nickel and 279
deaeration and 299
electrode efficiency 272–274
electrode potential and 269
electrolyte factors 275
fresh water and 287–291
hardness and 287
insulation and 295–297
ionization and 268
isolating flanges 295
low-conductivity and 326
mastics and 299
NAB and 274f 281
oil platforms and 276 341
paints and 297
polarization and 272
potential variability and 271–272
seawater and 285–286
separation and 295–297
soft waters and 287
spool pieces and 296

Index Terms Links
galvanic corrosion (Cont.)

temperature 272
titanium and 274f
underground corrosion and 312–314
without contact 282–283
gold, potential and 269
gunmetals 9 132
bronze and 9
composition of 362t
erosion and 150f
SCC and 198
seawater and 338
hard waters 44 62
heat-affected zone (HAZ) 339
heat exchangers 152f
carbon films and 70 338
cathodic protection and 298
chlorine and 139–140 159
cooling of 135 143
copper-nickel alloys and 24 149
corrosion and 23
Cu-Ni-Fe-Mn and 185
efficiency of 146
Index Terms Links
heat exchangers (Cont.)

films and 70 84 338
fouling in 243
hot spot corrosion and 184 191
impressed current and 298
inhibitors and 299
iron additions to 23
iron hydroxide deposit 299
macrofouling of 139
MIC and 256
microfouling of 139
MSF plants and 344 345
See also MSF plants
multipass 159
muntz metal and 176
naval brass and 176
pitting and 69
pollution and 143
power stations and 139 149 160
sacrificial anodes in 298
sand and 149 153–154
SCC and 51
seawater-cooled 135
ships and 143

Index Terms Links
heat exchangers (Cont.)

silt and 149 153
solution and 135
sponge balls 146 262
spool pieces and 298
spray tests 51 52t
thermal conductivity and 2
titanium and 345
tubes in. See tubes
turbulence and 125 158
water boxes 158–159
heat shrink tubing 209
heating systems 282 335–337
HE See hydrofluoric acid
hot spot corrosion 184–188
ammonia-induced 186–188 187f
classical 185 191
heat exchanger design and 191
pitting and 184
sulfide-induced 185–186 187f 188
191
temperature and 186
thermogalvanic effects and 185
three types of 184
water flow and 190

Index Terms Links
hot water systems 60 157 263

bimetallic corrosion and 282
erosion corrosion and 157 336
formic acid and 82
Legionnaires’ disease and 72 89
pitting in 71 72–75 77
82 89
SCC and 197
stainless steel and 278f
humic acid 65 77
hydraulic systems 348
hydrochloric acid 4647 48t
hydrofluoric acid 48
hydrogen embrittlement 237–241
seawater and 238
test method 237
hydrogen sulfide 255
bacteria and 143
polluted waters and 185
SCC and 217
seawater and 143
SRB and 261
hydrotalcite 17 18
films and 253

Index Terms Links
impingement attack. See erosion corrosion

INCRA. See International Copper Research
Association
Instrument Society of America (ISA) 328
International Copper Research Association
(INCRA) 195 331
International Organization for Standardization
(ISO) 220
CORRAG program 322
iron 23
aluminum brass and 136f 137f
ferrous sulfate. See ferrous sulfate
manganese and 121
solubility limit 11
in solution 120 339
See also specific alloys, topics
ISA. See Instrument Society of America
ISO. See International Organization for
Standardization
jet tests
Campbell test rig 365 366
erosion corrosion and 131 132t 367

Index Terms Links
jet tests (Cont.)

high-velocity tests 111
impingement tests 119 125
multivelocity 133
once-through conditions 368
recirculating tests 368
schematic of 366f
seawater and 368
water box method 366
joints, corrosion in 353–359
lacquers 331
Langelier index 44
lead 44 176
bacteria and 253
chlorine and 139
dissolution rates 44 54 61
74 83
DZR brass and 176
gunmetals and 9
phosphates and 61
soil and 321
soldering and 353–354
LECO-type test 84
Legionnaires’ disease 72 89
Index Terms Links
lime 54
Little, A.D. survey 344 345
Lucey model 66 74
agressivity ratings and 86
chloride and 85
chloride contents and 65
ionization and 86
pitting and 63 64f 65
66 67–69
pitting propensity rating and 87f
PPR and 86
sulfate and 85
magnetic permeability 120 121 339

malachite 16
manganese 10 127 138
aluminum alloys and 171 179
bronze and 8 9 10
See manganese bronze
cavitation and 171
Cu-Ni-Fe-Mn alloy. See Cu-Ni-Fe-Mn alloy
as deoxidant 11
iron and 121
seawater and 23

Index Terms Links
manganese bronze 8 169

cavitation and 171
propellers on marine vessels 130
SCC and 213 221
sulfides and 146
manganese dioxide 137
martensite 10
Mattsson’s solution 196 197f 200
May test. See BNF jet test
mechanical vapor compression (MVC) plants 346
MED plants. See multiple effect distillation
plants
Meigh, H. 10
mercury 206–208
brass and 209t
DZR brass and 208
mercurous nitrate test 206 207f 208f
SCC and 206 217
meringue corrosion 177 336
MIC. See microbially influenced corrosion
microbial action. See bacteria
microbially influenced corrosion (MIC) 53 243–265
See bacteria
molybdenum 121 124 217
MSF desalination. See multistage flash
desalination

Index Terms Links
multiple effect distillation (MED) plants 34 343

evaporation in 348
operation of 347f
titanium and 348
multistage flash (MSF) desalination plants 21 34 280
ammonia pollution and 138 191
ferrous sulfate dosing and 191
heat exchangers and 343–346 345t
NAB and 340
operation of 344f
scaling and 22
seawater and 343
temperatures in 36
muntz metal 156 176 286
museums, corrosion and 327
nickel aluminum bronze (NAB) 10 29 118

179 274
beta phase 340
castings 12
cavitation and 169 171
chlorine and 143
corrosion fatigue and 233–234 235
dealumnification and 189 339
Index Terms Links
nickel aluminum bronze (NAB) (Cont.)

desalination plants and 340
erosion and 117 129t 155f
158
galvanic corrosion and 274f 281
heat treated 179
jet impingement tests 128
kappa III phase and 180
microstructure of 128
MSF plants and 340
nickel content 180
nitrates and 209
phase corrosion 179
pitting and 182
potential-time curves 181f
sand and 154 155f
SCC and 209
337 339 340
selective attack 182
selective phase attack 180 184 190
stainless steel and 182t
sulfides and 146
welding and 340 357
nickel-chrome-molybdenum alloys 13 273

Index Terms Links
nickel silvers 13
composition of 363 363t
SCC and 326
niobium 217 229 237
copper-nickel alloys and 12 121 124–125
nitrates
DIN specifications 209
mercurous nitrate test and 210
SCC and 209–213
nitric acid 47
nitrogen oxides 320
noble materials 54
North-Pryor model 16
nozzles, cavitation and 168
oil platforms 120 184

biocides 262
biofouling 251
carbon steel legs 232
cladding 217 232 250
copper-nickel alloys and 216 250 285
corrosion fatigue and 232
corrosion-resistant fasteners 238
deaeration and 280
Index Terms Links
oil platforms (Cont.)

fouling and 250
hydrogen embrittlement and 276
NAB and 286
oily waters and 348–350
piping systems 286
SCC and 217
seawater and. See seawater
SRB and 262
See also specific materials, topics
organic acids 49–50
See specific types
oxide films 53 255
oxygen 5
atmospheric corrosion and. See atmospheric
corrosion
bacteria and 252 254
cathodic reaction and 99 114 254
267–268 271 276
308
concentration cells 95 185
deaeration and 54 280

Index Terms Links
oxygen (Cont.)
dissolved 19 27 33
36f 37 39
66 83 103
114 280 320–321
LECO tests and 84
oxide formation 29
oxygen-free alloys 5
pitting and 68 83
scavengers 280 299
seawater and 337–339 348–349 356
358
soils and 303
sulfides and 27
packing agents 222

PDO. See phosphorus-deoxidized copper
peroxide 77
pH values 355
alkalinity and See alkalinity
bacteria and 258
bicarbonates and 71 76 291
biofilms and 252
chlorides and 52 83 177
copper-nickel alloys and 37 216
Index Terms Links
pH values (Cont.)
cuprosolvency and 39 40 53
215
deaeration and 349
drinking water and 40 41 44
97
hydrotalcite and 18 22
73 74 89
plumbosolvency and 54
polarity reversal and 180 271 275
281
PPR values and 65-66
SCC and 201 202 205
210 214 219
220 292 320
soils and 303 305 309
solders and 354
phosphates 85
corrosion and 43
films and 43
lead and 61
pitting and 61 62f
phosphoric acid 49 49t
phosphorus-deoxidized (PDO) copper 59 196–197
photoelectric sensitivity 16

Index Terms Links
pickling 47 89
piping. See tubes
pitting 57–94
activated charcoal and 86
ammonia and 188 205
annealing and 89
arsenic and 59
bacteria and 263
bicarbonate-sulfate ratio and 71 72
biofilms and 77
calcium carbonate and 57 62 66
68
carbon films 60 83 89
cavitation and 165–167
101 261
copper carbonate and 57
107f
cuprous chloride and 60 67
cuprous oxide and 58
films and 57–58 77
flux residues 336
fresh water and 57
hardness and 66
horseshoe imprints 111

Index Terms Links
pitting (Cont.)
hot spot corrosion and 184
hot water 77
initiation of 72 76
ionization and 66 67 73
83
kappa III phase 107f
Lucey model and 63–69 86 87f
manganese dioxide 71
mechanism of 67-69
microbial action and 90
natural inhibitors 58
oil-water mixture 349
orthosilicates and 72
peroxide and 77
pH values and 62 74
phosphates and 61 62f
potential zones 60
PPR and 86 87f
propensity rating 86 87f
SCC and 205
schematic of 68f
soils and 308
solder and 61 61f
stagnation and 85

Index Terms Links
pitting (Cont.)
341
in surface-derived waters 58
Swedish studies 75–76
test cell 67
tubes and 60 60t
See also tubes
Type I 57–70 58f 60
67–69 83 85
336
Type II 70–74 71f 72f
89
Type III 74–78 75f 90
336
water composition and 62 66f 86
water temperature 89
See also sperific materials, topics
platinum, potential and 269
plugging 173
plumbosolvency 43-44 54
polybutylene 259
polysaccharide films 76
polytetrafluorethylene (PTFE) 98

Index Terms Links
potable water 41 43 335

See drinking water
potassium permanganate 333
potentiostatic method 73
Pourbaix diagram
SCC and 220
seawater and 39 39f
power stations 127 138 143
condensers in 159
exchangers and. See heat exchangers
ferrous sulfate 135
silt and 151
tubes in. See tubes
turbulence in 159 161
printed circuit boards 329
propellers 165
PTFE. See polytetrafluorethylene
pumps 165 168 171
rainwater 319
red brass 114
redox potential 54
reverse osmosis (RO) desalination plants 103 104t 343
river waters 97
See fresh water

Index Terms Links
rosette corrosion 43 82–83 84f

91
S/N curves 225

sand
aluminum alloys and 154 156
condenser tubes and 156 156t
erosion and 151 153f 155f
160
NAB and 154 155f
seawater and 154 156t
See also seawater
settling and 161
sulfide and 154
saturated calomel electrode (SCE) 267
scaling 22 44
scanning electron microscope 62
SCC. See stress corrosion cracking
SCE. See saturated calomel electrode
season cracking 195 197
seawater 85 179 337–342
alloy 400 and 79f 95
alloy groups 284 284t
See sperific alloys
alloy K-500 and 239f
aluminum brass and 228f 231 253
Index Terms Links
seawater (Cont.)
aluminum bronze and 285 286
ammonia and 202 231 231f
232f
anaerobic bacteria and 27
area ratio and 278
bacteria and 252
See also bacteria
biofilms and 252 252f 253
274
C70600 alloy 337–338
cathodic polarization 28f
cavitation and 169 170t
138 143 160
273 274 286
cooling systems and 338
copper-manganese-aluminum alloys 179
copper-nickel and 19f 127f 129t
229f 231f 247f
252 338 339
356
corrosion fatigue and 226 231 232f
corrosion-resistant fasteners 238
crevice corrosion and 97–108 98t 99
101
cuprous oxide 39
desalination plants. See specific types

Index Terms Links
seawater (Cont.)
electrochemical series for 270
erosion corrosion and 116–134 116t
fasteners and 340
films and 17 22
See also films
firewater systems 338
flow velocities 53
fouling and 25 246f 247f
248f 251 260
See also fouling
fresh water and 33-40 287
See also fresh water
galvanic corrosion and 8–11 270f 271f
285–286
general corrosion rates for 33 34t
gunmetal and 338
hardness of 287
hydrogen embrittlement and 238
hydrogen sulfide in 27 143
jet tests and 368
manganese and 23
muntz metal and 286

Index Terms Links
seawater (Cont.)
NAB and 143 180 286
337 339 340
342
pH and 39
See also pH values
pigments and 254
piping and 251
pitting and 7 78
Pourbaix diagram and 39 39f
pretreating and 150f
sand and 154 156t
See also sand
SCC and 203t
See also SCC
ships and 144
See ships
SRB and 27 262 356
sulfur and 254
surface finish and 70
synthetic 368
temperature and 21 34 128
128f 129t 271
272
welding and 356
See al.so specific alloys, topics

Index Terms Links
selective phase attack 173

settling 161
Severn magnetic gauge 121
shear stress 131 131t 132
157
ships
cavitation and 165
cladding and 244
copper-nickel and 144 249 250
fouling and 249 250
heat exchanger and 143
oil tankers 140 250
propellers and 165
seawater and. See seawater
silica gel 222
silicon 124 124t 125f
silicon bronze 8–10 201 205f
composition of 362t
dealloying and 184
SCC and 203 221
silt
erosion and 160
heat exchanger and 149 153
power stations and 151
settling and 161

Index Terms Links
silver
potential and 269
soldering and 355
slimes 139
sodium bicarbonate 16
sodium chloride 321
sodium dimethyldithiocarbainate 145
sodium hydroxide 50–51
sodium silicate 53
sodium sulfite 262
sodium zeolite 112
soft water
chlorine and 45 74
pitting and 45 262
soils
cinders 307
corrosion in. See underground corrosion
organic matter in 308
pH values 305 309
pitting and 308
properties of 303
resistivity of 305
sulfate content 305
types of 306t 307
soldering 353–354
spray tests 52 52t

Index Terms Links
SRB. See sulfate-reducing bacteria

stainless steel 13 293
biofilms and 257
cathodic reaction 257
copper and 312
isocorrosion curves 51
NAB and 182t
polarization curves for 273
seawater and 286
underground corrosion and 312 314t
stress corrosion cracking (SCC) 51 127 195–224
absorbants and 222
Aebi test 219
aluminum brass and 213 215
aluminum bronze and 201 214 221
amines and 196
ammonia and 196 197 200
205 320
arsenic and 7 199 213
215 215f
atmospheric conditions and 196
beta phase content and 221
C72420 alloy and 217
cathodic protection 222
chloride solutions and 213
citrate and 214
common tests for 218t

Index Terms Links
stress corrosion cracking (SCC) (Cont.)

copper-nickel alloys and 216–218 219
DIN specifications 209
DZR brass and 197f 199
Eichorn test 219
films and 196 210
fresh water and 335
gunmetals and 198
heating pipes 209
high stresses and 219
industrial environment 326
manganese bronze and 213 221
mercurous nitrate test 218–221
mercury and 206 217
molybdenum disulfide and 217
NAB and 209
nitrates and 209–213
passivity breakdown and 210
PDO copper and 197
pH and 220
phenylthiourea and 222
phosphorus levels 216
pitting and 205
Pourbaix diagram and 220
season cracker tests 197 203
seawater and 203t

Index Terms Links
stress corrosion cracking (SCC) (Cont.)

silicon bronze and 203 221
stress corrosion curves and 195
sulfur dioxide and 211–213 220
test methods 218–221
tin and 199 201
tungum and 349
zinc and 198f
See also specific materials topics
sulfate-reducing bacteria (SRB) 27 143 280
bacteria and 253
chlorine and 261
fouling and 253
MIC and 262
sulfides 27–29 127 143–149
alloy 400 and 256
aluminum brass and 147 154 160
191
copper-nickel and 145 147 149
erosion corrosion and 148f 155 155f
heat exchanger and 159 160
heat exchanger tubes and 154

Index Terms Links
sulfides (Cont.)
NAB and 146
pitting and 90 188
pollution and 138 185
redox potential 160
sand and 154
sponge ball cleaning and 146
sulfur dioxide 320
atmospheric corrosion and 275 322
brass and 222
ISO classes 322–323
rural areas and 292 321
SCC and 196 211–213 220
sulfuric acid 45–46
aeration and 45
brass and 45
corrosion and 47t
electrolytes and 275
MIC and 263
pitting and 68
SRB and 263
tin bronzes and 45
sunlight, corrosion and 16
surface-derived water 65 86
swimming pools 342

Index Terms Links
tap water 42f

tarnishing 13
tartrates 53
temperature
acids and 49
biofilm and 272
corrosion and 36 37 38f
cuprosolvency and 40
lead dissolution 44
MED desalination plants and 36
MSF desalination and 36
seawater and 21 34
thermocompression distillation (TCD) plants 346 347f
tin
dealloying of 184
SCC and 199
tin brass 146
tin bronze. See tin bronze

Index Terms Links
tin bronze 50 184

bronzes and 22–23
composition of 362t
HF and 48
oxides of 23
SCC and 201
titanium 151 156 286
galvanic corrosion and 273–275 274f
heat exchanger tubes and 156 345
MED plants and 348
tubes 51–52 59–60 158
aging and 119
ammonia and 197
bacteria and. See bacteria
blue water and 43
chlorine and 73 103
cleaning of 85 251 262
342
colloids in 138
condenser tubes 127 131
exchangers and. See heat exchangers
films on 20 69–70 77
84 89 156
fouling in 243 262

Index Terms Links
tubes (Cont.)
iron content 120
pass-fail criterion 85
77 85
plastic 258
potable water and 45 52 72
78 112 205
261
pretreatment of 145
SCC and 51
shear stress on 132
temperature and 185–186 191 213
turbulence and. See turbulence
water boxes and 158
welding and 132
tungum 348–349
turbulence
bends and 158 365
159 277 296
339 354 369
films and 16
heat exchanger tubes and 125 158 159

Index Terms Links
turbulence (Cont.)
soldering and 116 132 354
tubes and 131 132 134
water boxes and 158 159
Unified Numbering System (UNS) 4 6t 206

206t 362t 363t
See specific materials,topics
ultraviolet irradiation 89 263
underground corrosion 303–305
bimetallic junctions 313
brass and 314
cinders 314
copper alloys and 312 312t 313
galvanic corrosion and 312–314
iron oxide 314t
iron products 313
long-term exposure tests 304–311
methods of prevention 314–315
pitting and 310
soil properties and 303–304
steel and 314t

Index Terms Links
Vibraphone tests 229

vibration, cavitation and 165
water quenching 119

water treatment
charcoal and 59
organic material 59
welding 121 281 355–358
ductility dip and 357
HAZ and 120 339 356
357
NAB and 340 357
pipe and 132
seawater and 356
well water 65 112–113 113t
x-ray spectroscopy 21

Index Terms Links
zinc alloys 7 9 176

198f 290
zirconium 124

(EngineeringEBookspdf) Corrosion of Copper PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(EngineeringEBookspdf) Corrosion of Copper PDF

Uploaded by

Copyright:

Available Formats

The Corrosion of

Copper and I t s Alloys

A Practical Guide for Engineers

THE CORROSION SOCIETY

1440 South Creek Drive

THE CORROSION SOCIETY

Printed in the United States of America

Library of Congress Cataloging in Publication Data

Francis, Roger, Engineer.

Neither NACE International, its officers, directors, or members thereof

2 Copper and Its Alloys 5

3 Film Formation and Properties 15

7 Erosion Corrosion and Erosion 111

10 Stress Corrosion Cracking 195

11 Corrosion Fatigue 225

12 Hydrogen Embrittlement 237

13 Fouling and Microbially Influenced Corrosion 243

14 Galvanic Corrosion 267

14.2.5 Area Ratio 276

15 Underground Corrosion 303

17 Corrosion in Waters 335

17.4 Desalination 343

18 Joining and the Corrosion of Joints 353

Appendix A: Nominal Compositions of Some Commonly Used

Copper and Its Alloys

In this chapter, the composition and important properties of the common

to more than 320 MPa, with an elongation of about 4%. An alternative

Copper alloys containing zinc, commonly in concentrations from 5 % to

2.3.1 Tin Bronzes and Gunmetals

2.3.2 Silicon Bronze

cracking resistance of silicon bronze is discussed in more detail in Chap-

2.3.3 Aluminum Bronze

KN U ICIy K, KnI martensite

FIGURE 2.1 Schematic of the structure of nickel aluminum b r o n ~ e . ~

alloys, and it is more susceptible to corrosion. However, these alloys have

2.4 Copper-Nickel and Nickel-Copper Alloys

copper-nickel with a simple magnetic permeability gauge to ensure that the

Alloy 400 is a nickel-copper alloy with good resistance to corrosion

2.5 Nickel Silvers

4. J.E. Bowers, P.W. Oseland, G.C. Davies, “Development of a Hot-Stamping Brass

Film Formation and Properties

3.1 Copper and the Brasses

When copper is first placed in water, the initial corrosion rate is

3.2 Aluminum Brass

(Mg6A12C03[OH] 16 .4H20).This compound was identified by several lab-

3.3 Copper-Nickel Alloys

-300 -250 -200 -150 -100 -50 0

I--1-0 -Fe -Ni -CuI

The potential at which 90/10 copper-nickel is first exposed has a strong

hydrotalcite, as seen on aluminum brass (Section 3.2), was also detected on

3.4 Aluminum Bronze

The air-formed film on aluminum bronze is mainly alumina (Al2O3), and

3.5 Tin Bronzes

on a commercial 8% tin bronze alloy ((252100) exposed in seawater.” From

3.6 Ferrous Sulfate

Ferrous sulfate dosing has also solved corrosion problems caused

Current Density (pA/cmZ)

-2hl NO CI --- 60dl No CI - -6Od/ 4mglL CI I

Current Density (pA/cm2)

I-Aerated --- Deaerated I

0.01 0.1 1 10 100 1,000

Current Density (pA/crn2)

5. S.M. Wilhelm, Y. Tanizawa, C.-Y: Liu, N. Hackerman, Corros. Sci. 22 (1982):