Corrosion Science 47 (2005) 3336–3367

www.elsevier.com/locate/corsci

The corrosion of nickel–aluminium bronze

in seawater
J.A. Wharton a,*, R.C. Barik a, G. Kear b,
R.J.K. Wood a, K.R. Stokes a,c, F.C. Walsh a

a
Surface and Electrochemical Engineering Groups, University of Southampton,
School of Engineering Sciences, Highfield, Southampton SO17 1BJ, United Kingdom
b
BRANZ Ltd., Science and Engineering Services, Private Bag 50 908,
Porirua City, Wellington 6220, New Zealand
c
Physical Sciences Department, Dstl, Porton Down, Salisbury,
Wiltshire SP4 0JQ, United Kingdom

Available online 26 September 2005

Abstract

Nickel–aluminium bronze (NAB) alloys show good corrosion resistance under marine con-
ditions. The corrosion behaviour of cast and wrought NAB alloys is illustrated in this work
through a range of electrochemical techniques including open-circuit potentiometry with time,
oxygen reduction voltammetry, NAB dissolution voltammetry, potential step (or flow step)
current transients and linear polarisation resistance. The galvanic coupling of NAB to stainless
steel or copper is examined by zero resistance ammetery. The importance of using controlled
flow working electrodes is illustrated by the use of a rotating disc electrode, a rotating cylinder
electrode and a bimetallic (NAB/copper–nickel) rotating cylinder electrode. In addition to
controlling the hydrodynamics, such electrodes allow charge transfer data to separate from
those of mass transport control under mixed kinetic control. Longer term seawater immersion
trials on planar coupons coupled to titanium or cupronickel are also reported. The relative
contributions of erosion and corrosion attack are considered using a wall-jet electrode and
the corrosion characteristics of NAB are compared to those of copper and copper–nickel in
chloride media.

2005 Elsevier Ltd. All rights reserved.

*
Corresponding author. Tel.: +44 23 8059 2890; fax: +44 23 8059 3016.
E-mail address: jaw6@soton.ac.uk (J.A. Wharton).

0010-938X/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2005.05.053
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3337

Keywords: NAB; RDE; RCE; Erosion; Wall-jet electrode; Galvanic coupling; ZRA; LPR; Charge tran-
sfer kinetics; Mass transport

1. Introduction

Aluminium bronzes are copper-based alloys in which aluminium (<14 wt.%) is the
main alloying element. In addition, some of the alloys contain iron, nickel, manga-
nese or silicon. Both cast and wrought aluminium bronzes offer a good combination
of mechanical properties and corrosion resistance. Consequently, aluminium bronzes
have been widely used for decades in a variety of marine applications, including
valves and fittings, ship propellers, pump castings, pump shafts, valve stems and heat
exchanger waterboxes [1].
Nickel–aluminium bronze (NAB) alloys containing 9–12 wt.% aluminium with
additions of up to 6 wt.% each of iron and nickel represent one of the most impor-
tant groups of commercial aluminium bronzes. Increasing aluminium content results
in higher strength, which is attributable to a hard face-centred cubic (fcc) phase
which enhances the properties of castings as well as hot working in wrought alloys
[2]. The other alloying elements also improve properties and alter microstructure.
Nickel improves corrosion resistance, while iron acts as a grain refiner and increases
tensile strength. Nickel also improves yield strength, and both nickel and manganese
act as microstructure stabilisers [2]. Table 1 shows the composition of the main com-
mercial alloys: CuAl10Fe5Ni5, CuAl10Ni5Fe4, CuAl11Ni6Fe6 and CuA19Ni5-
Fe4Mn. NAB alloys are metallurgically complex alloys with several intermetallic
phases in which small variations in composition can result in the development of
markedly different microstructures [2–4], which can result in wide variations in sea-
water corrosion resistance. The microstructures that result in optimum corrosion
resistance can be obtained by controlling the composition and heat treatment [5].
For example, castings of CuA19Ni5Fe4Mn used for naval applications are given
an annealing treatment at 675 C for 2 to 6 h [2].
The microstructure of a sand cast NAB (Fig. 1a) consists of light etched areas of
a-phase, which is a fcc copper-rich solid solution and dark etched martensitic regions
(Ôb-phase or retained bÕ-a high temperature phase), surrounded by lamellar eutectoid
phases and a series of intermetallic j-phases. The jI-phase is globular or rosette
shaped and is reported to be iron-rich (based on Fe3Al). The jII-phase also takes
the form of dendritic rosettes which are unevenly distributed at the a/b boundary
and are smaller than the jI rosette. The jIII-phase can appear in either a lamellar
or sometimes a coagulated or globular (degraded lamellar) form. It grows normal
to the a/b boundary, as well as forming at the boundary of the large jI-phase,
and is described as being nickel-rich (NiAl). The jIV-phase is a fine precipitate within
the a-phase and is considered to be iron-rich. The microstructure perpendicular to
the extrusion direction for the wrought NAB bar is shown in Fig. 1b). The micro-
structure has been influenced by the extrusion with little evidence of the b and
jIII-phases. The heat treatment specified for the NES747 Part 2 British Naval
3338 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

Nomenclature

B Stern–Geary constant, V
co bulk concentration of reactant, mol m
3
Cv volume fraction of sand
d mean diameter of sand, m
Ee equilibrium potential, V
EL potential corresponding to limiting current conditions, V
Ek kinetic energy for erosion, J
Ecorr corrosion potential, V
Ecrit critical potential, V
F Faraday constant, C mol
1
jo exchange current density, A m
2
jL limiting current density, A m
2
ko standard rate constant
ks constant in Eq. (5)
k1 constant in Eq. (6)
k2 constant in Eq. (8)
l kinetic energy exponent in Eq. (10)
Dm mass loss, kg
merod mean erodent mass, kg
p peripheral velocity exponent in Eq. (6)
q peripheral velocity exponent in Eq. (8)
p O2 partial pressure of oxygen
R molar gas constant, J K
1 mol
1
Rp polarisation resistance, X cm
2
Qv slurry volume flow rate, m3 s
1
t time, s
T temperature, K
U peripheral velocity of electrode, m s
1
m mean erodent velocity, m s
1
DVu volume loss per impact, m3
z number of electrons in electrode process

Greek symbols
ac cathodic transfer coefficient
ba anodic Tafel slope, V
bc cathodic Tafel slope, V
q density of sand, kg m
3
x rotation speed of electrode, rad s
1

Abbreviations
BRCE bimetallic rotating cylinder electrode
EPS extracellular polymeric substances
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3339

LPR linear polarisation resistance

LSV linear sweep voltammetry
OCP open-circuit potential
PSCT potential step current transient
RDE rotating disc electrode
RCE rotating cylinder electrode
WJDE wall-jet disc electrode
ZRA zero resistance ammetry

standard results in the minimising or elimination of the more corrodible b-phase and
an increased density of fine j-phase precipitates in the a-phase, see Fig. 1c.
The corrosion resistance of NAB has been attributed to a protective layer, per-
haps 900 to 1000 nm thick, containing both aluminium and copper oxides [6,7].
The oxide layer is aluminium-rich adjacent to the base metal and richer in copper
in the outer regions. There are also oxides of nickel and iron, together with trace
amounts of copper salts and copper hydroxychlorides, e.g., Cu2(OH)3Cl and
Cu(OH)Cl, which form after longer exposure times to seawater. The oxide layer ad-
heres firmly to the base metal and consequently provides corrosion protection reduc-
ing the corrosion rate by a factor of 20–30 [7]. This protection has been attributed to
both a decrease of the anodic dissolution reaction which hampers the ionic transport
across the oxide layer, as well as a decrease in the rate of the cathodic reaction on the
oxide layer [6]. In quiet, tidal or flowing seawater, provided the flow velocity does not
exceed a certain limit, the protective film corrosion resistance continues to improve
until it achieves a long-term steady-state of 0.015–0.05 mm y
1 (0.6–2.0 lA cm
2)
[1,2]. NAB is the most resistant of the readily available copper-based alloys to
flow-induced corrosion. However, under conditions of service involving exposure
to seawater flowing at high speed, or with a high degree of turbulence, damage
can occur to the protective oxide layer, locally exposing the unprotected bare metal.
NAB is vulnerable to such attack in unpolluted seawater at flow speeds in excess of
4.3 m s
1 and the degree of attack is reported to vary logarithmically with velocity:
from 0.5 mm y
1 (20 lA cm
2) at 7.6 m s
1 to 0.76 mm y
1 (31 lA cm
2) at
30.5 m s
1, and even at 7.6 m s
1 corrosion rates could rise locally to 2 mm y
1 [8].
In contrast to the many studies of the corrosion of copper [9] and its alloys [10] in
chloride media, there have been few studies devoted to electrochemical kinetics at
NAB surfaces. Schüssler and Exner [6,11] studied the dissolution of cast NAB in sea-
water but they considered a fixed velocity of electrolyte and performed a limited
kinetic analysis. Kear et al. have considered both cathodic [12] and anodic [13] polar-
isation during NAB corrosion. In aerated seawater, the anodic dissolution of NAB
shows many featured in common with the dissolution of copper, a simplified anodic
reaction being the formation of the dichlorocuprous anion:
Cu
e
þ 2Cl
! CuCl

2 ð1Þ
 3340
J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367
Table 1
Composition of nickel–aluminium bronzes used in this work
Designation Composition/wt.% Mechanical properties
ISO/CEN standards British ASTM Cu Al Ni Fe Mn Hardness Tensile
(number) specification equivalent Vickers strength/MPa
Cast
CuAl10Fe5Ni5–C BS1400 AB2 C95800 Bal. 8.5–10.5 4.0–6.0 4.5–5.5 3.0 max 150–160 600–650
CuAl11Ni6Fe6–C – – Bal. 10.0–12.0 4.0–7.5 4.0–7.0 2.5 max 180–195 680–750
CuAl9Ni5Fe4Mn NES 747 Part 2a – Bal. 8.8–9.5 4.5–5.5 4.0–5.0 0.75–1.30 170 620 min
Wrought
CuAl10Ni5Fe4 (CW307G) CA104 C63200 Bal. 8.5 –11.0 4.0–6.0 3.0–5.0 – 180–220 590 min
CuAl11Fe6Ni6 (CW308G) – – Bal. 10.5–12.5 5.0–7.0 5.0–7.0 – 200–245 750–830
a
British Naval specification—castings annealed at 675 C for 2 to 6 h and cooled in air—for superior corrosion resistance.
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3341

Fig. 1. Optical micrographs of nickel–aluminium bronze: (a) sand cast NAB, (b) wrought bar NAB (as
used in RDE and RCE studies), (c) British Naval specification, cast then annealed (NES 747 Part 2, as
used in erosion–corrosion—WJDE and pontoon immersion studies).

The predominant cathodic reaction is the four electron reduction of dissolved

oxygen:
3342 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

O2 + 2H2 O + 4e
! 4OH
ð2Þ

At the mixed potential, the anodic kinetics are influenced by charge transfer and
mass transport contributions. Consequently, the rate and mechanism of corrosion
(which are dependent on film formation) are very flow sensitive [12,13] as is the bime-
tallic corrosion of NAB when coupled to copper–nickel or copper [14].
The aim of this paper is to illustrate the corrosion characteristics of NAB in sea-
water using both electrochemical and traditional exposure techniques. The short
term potentiometric and voltammetric techniques are complemented by bimetallic
coupling, erosion–corrosion studies and the monitoring of corrosion rates over ex-
tended periods of up to twelve months. It is important to consider flow conditions
during corrosion studies of NAB as mass transport is an essential component of
the anodic kinetics; the protective films formed are also flow sensitive. Rotating disc
electrode (RDE), rotating cylinder electrode (RCE), wall-jet disc electrode (WJDE)
and bimetallic rotating cylinder electrode (BRCE) geometries have been used as
complementary, hydrodynamic working electrodes (Fig. 2). The smooth RDE and
RCE provide well-defined surfaces for laminar and turbulent flow studies, respec-
tively, while a twin electrode, bimetallic RCE allows galvanic studies under well-con-
trolled turbulent flow conditions. The WJDE facilitates a comparison of NAB

Fig. 2. Types of hydrodynamic working electrode used in this study: (a) rotating disc electrode (RDE), (b)
rotating cylinder electrode (RCE), (c) bimetallic rotating cylinder electrode (BRCE), (d) wall-jet disc
electrode (WJDE).
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3343

dissolution rates under corrosion, erosion and combined erosion–corrosion condi-

tions using the jet impingement geometry. The longer-term corrosion rates of
NAB (including galvanic effects) have also been considered using planar coupons un-
der tidal conditions. A range of experimental time scales, electrode geometry, flow
conditions and NAB alloys have been used to quantify the corrosion characteristics
of NAB over a broad range of operational conditions.

2. Experimental details

2.1. Electrochemical measurements at RDE and RCE electrodes

Rotating electrode studies were carried out using discs or cylinders machined
from wrought NAB, supplied by Stone Manganese Marine Ltd. in rod form to BS
2874: 1986: CA 104 and having the composition Cu balance, Fe 4.43, Si 0.05, Mn
0.14, Pb 0.02, Al 9.31, Ni 4.65, Mg 0.01 max, Zn 0.11 wt.%. Copper (99.9% wt./
wt.) and 90-10 Cu–Ni (BS 2874: 1986: CN 102) were also used for comparison pur-
poses. A Pine Instruments Company, model AFMSRX analytical rotator was used
to control the rotation speed to within 1%. The dimensions of the active RDE were
0.40 cm diameter and 0.13 cm2 area while those of the RCE were diameter 1.99 cm,
length 1.60 cm and area 10.05 cm2. High-density polyethylene rod was used as the
inert sheathing material. The range of rotation speed was 200–9500 rpm (29–
995 rad s
1) for the RDE and 100–4000 rpm (10–398 rad s
1) for the RCE. The
working electrode surfaces were mechanically polished on micropolishing cloth
and degreased in ethanol prior to each individual polarisation and corrosion poten-
tial measurement. The polishing procedure consisted of a double, 3 min polish with a
0.3 lm a-alumina slurry. Natural seawater was collected from a short-term holding
tank in Langstone Harbour, Hampshire, UK. The seawater was subsequently vac-
uum filtered down to a 0.2 lm pore size membrane to remove suspended solids,
microorganisms and the majority of biological spores. Artificial seawater to BS
3900: Part F4:1968 (supplied by BDH) was used as a standard solution for compar-
ison with the filtered seawater. Typical mean values of pH, conductivity, chloride ion
concentration, salinity and dissolved oxygen (DO) concentration were pH 8.0,
50.8 mS cm
1, 0.550 mol dm
3, 33.4& (g kg
1) and 7.0 ppm. The electrolyte was
aerated for 10 min prior to measurements. When necessary, de-oxygenation was
achieved by purging with 99.99% nitrogen (British Oxygen Company) for 10 min fol-
lowed by a continuous N2 blanket to prevent air ingress. Electrochemical measure-
ments were made at 25 ± 0.2 C with a PC-controlled, Eco Chemie Autolab
potentiostat (PGSTAT20) using the General Purpose, Electrochemical Software
(GPES) version 4.5. The electrochemical cell incorporated a thermostatically con-
trolled, glass water jacket, a platinum gauze counter electrode and a (Radiometer
Analytical A/S, Ref 401) saturated calomel electrode (SCE) used in conjunction with
a Luggin–Haber capillary. LPR measurements were made by polarising the NAB
10 mV each side of the steady corrosion potential and monitoring the current.
ZRA data was obtained using a proprietary instrument (ACM).
3344 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

2.2. Extended seawater immersion tests using planar coupons

NAB (NES 747 Part 2) supplied by Meighs Ltd. (chemical composition: Cu bal-
ance, Al 9.32, Fe 5.00, Ni 5.38, Mn 1.10, Si 0.05 wt.%), Cu–15Ni alloy (chemical
composition: Cu balance, Ni 15.4, Mn 4.36, Al 1.61, Fe 0.94 wt.%) and commercially
pure titanium (Ti 99 wt.%) were used for galvanic coupling tests. The specimens were
wet polished with 600-grade SiC paper, degreased in acetone, rinsed in distilled water
then air-dried. The galvanic couple specimens had dimensions of 5.0 · 4.0 · 0.7 cm
for NAB/Ti, 5.0 · 4.0 · 0.4 cm for NAB/Cu15Ni and 5.0 · 5.1 · 0.4 cm for NAB/
NAB. The anode to cathode area ratio was 1:1. The galvanic couples (·2 replicates)
were connected via electrical leads, with a separation of 5 cm and were mounted in a
planar orientation on two non-metallic boards. The boards were immersed in seawa-
ter (below the splash zone) from a pontoon located at the National Oceanographic
Centre, University of Southampton. Boards 1 and 2 were retrieved after 6-months
and 12-months, respectively. The galvanic current and couple potential were mea-
sured using a zero resistance ammeter (ACM Instruments Galvogill). The reference
electrode used was a Ag/AgCl with seawater electrolyte. The salinity of the seawater
was approximately 31& reflecting its brackish nature and the pH was 7.9, which is
within the range normally cited for seawater (i.e. 7.7–8.3 for surface waters). The sea-
water temperature and DO were monitored using a Hanna Instruments H19145
probe, the DO level ranged between 6 and 10 ppm, being dependent on seawater
temperature and salinity.

2.3. Jet impingement testing

The tests were carried out using a slurry jet impingement rig, which has been pre-
viously described [15,16], for kinetic energies ranging from 0.02 to 0.41 lJ (calculated
from experimental parameters) and a working electrode of 4 · 4 cm. Modifications
were made to accommodate an Ag/AgCl reference electrode and a platinum counter
electrode. The ejector assembly, used for sand particle intake, was located down-
stream to prevent erosion damage on the counter and reference electrodes. A valve
situated near the ejector allowed the sand intake to be completely isolated under flow
corrosion conditions. The flow corrosion test electrolyte consisted of a 3.5% NaCl
solution, while the slurry was a 3.5% NaCl solution with 3% wt./wt. silica sand. Pure
corrosion (Jetcorrosion) experiments were carried out in the absence of slurry, whilst
the erosion–corrosion (Jeterosion–corrosion) experiments were carried out with sand.
A Gamry Instruments PC4-750 potentiostat and ESA400 software were used to
monitor the potential. For pure erosion (Jeterosion, a
0.200 V cathodic protection
was applied, based on the potentials observed from erosion–corrosion conditions.
The test duration was 5 h. The NAB (NES 747 Part 2) and high strength structural
steel BS 4360 (chemical composition: Fe balance, Mn 1.6, Si 0.5, C 0.25 wt.%) spec-
imens used in the erosion and erosion–corrosion experiments were wet-polished (up
to grade 1200 SiC paper) and lapped (up to 3 lm diamond suspension) to a surface
finish of Ra < 1 lm. Before and after each test, specimens were washed in water and
degreased with acetone, dried in a jet of cold air then weighed by a Mettler AT201
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3345

precision balance (with a range of <205 g and an accuracy of ±0.02 mg) to obtain
mass loss results.

3. Results and discussion

3.1. Open-circuit potential (OCP) for corrosion

The open-circuit potential vs. time characteristics of nickel–aluminium bronze

were examined in both filtered and BS 3900 artificial seawaters as a function of elec-
trode rotation rate. Fig. 3a shows the typical behaviour of the alloy in artificial sea-
water over a rotation speed range of 200–9500 rpm. After initial immersion, the

-0.240
Open Circuit Potential vs. SCE / V

NAB - Artificial seawater

-0.260

-0.280

-0.300 200 rpm

-0.320
9500 rpm

-0.340
0 400 800 1200 1600 2000
(a) Time Post Initiation of Rotation / s

, 90-10 Cu-Ni - Filtered & artificial

, Cu - Filtered & artificial
, NAB - Filtered & artificial
Open Circuit Potential vs. SCE / V

-0.210

-0.240

-0.270

-0.300

-0.330

-0.360
0 200 400 600 800 1000 1200
(b) Angular Velocity / rad s-1

Fig. 3. Open-circuit potential vs. time behaviour of the copper-based RDEs. (a) Mean open-circuit
(corrosion) potential transients for wrought NAB in the aerated, filtered and conductivity adjusted
seawater as a function of RDE rotation rate. The open data points indicate the locus of inflection at each
rotation rate (at 95% of total change in potential). (b) Directly measured, stabilised open-circuit
(corrosion) potentials of copper, the 90-10 copper–nickel and the wrought NAB within aerated, filtered
and BS 3900 artificial seawaters as a function of RDE angular velocity.
3346 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

measured potential became more negative with time until a steady potential was
achieved. The time taken to reach a steady OCP always decreased with increasing
electrode rotation rate and became more negative with increasing electrode rotation
rate.
It is likely that the initial time taken to achieve the stabilised potential is related to
the dissolution and/or conversion of the thermal oxide films produced by the wet
polishing procedure. The change from more positive potentials exhibited by the oxi-
des to more negative potentials is due to changes in the surface films. This process is
partially dependent on the rate mass transport of either the chloride ion or the cu-
prous dichloride complex, both of which are involved in the dissolution of cuprous
oxide. The time taken to reach the OCP inflection decreases at faster rotation rates,
i.e. an increasing rate of reactant/product mass transport allows the dissolution of
cuprous oxide films at a higher rate.
Fig. 3b shows the stabilised open-circuit potentials for both filtered and artificial
seawaters plotted as a function of RDE angular velocity. For NAB, the stabilised
OCP in both seawaters changed from
0.297 to
0.333 V vs. SCE over 200 to
9500 rpm (21–994 rad s
1). These values are rather more negative than those of
the copper metal and 90-10 copper–nickel. From examination of the open circuit po-
tential of copper in 0.5 mol dm
3 NaCl solution at various RDE rotation rates, Faita
et al. [17] showed that the measured potential of copper generally became more neg-
ative with increasing rotation rate (e.g.,
0.263 V vs. SCE [600 rpm] to
0.271 V vs.
SCE [2000 rpm at pH 7.5]). This mass transport effect for polished copper has been
verified by a number of authors [18–24]. Power and Ritchie [25] have classified the
mechanisms of corrosion reactions according to the rate determining step and the
behaviour of the open-circuit potential/corrosion potential as a function of reac-
tant/product concentration and conditions of mass transport. The present data are
consistent with their Case D, i.e. the cathodic reaction is charge transfer controlled
while the anodic reaction is mass transport controlled.

3.2. Cathodic kinetics (oxygen reduction)

Steady-state, cathodic polarisation of RDEs has been considered for the NAB, to-
gether with copper and 90-10 copper–nickel. Both LSV and PSCT techniques have
been used, with associated hydrodynamic (rotation rate) steps [26]. For the latter
technique the potential was either stepped to the region of total mass transport con-
trol of the reaction rate (at the limiting current density, jL) or to regions of mixed
charge transfer and mass transport control. An LSV sweep rate of 0.5 mV s
1 was
chosen from double potential step experiments at the copper RDE (at zero rpm)
within the mixed controlled region. From these measurements it was determined that
steady-state conditions would prevail at this potential sweep rate during the mea-
surement of oxygen reduction.
The cathodic LSV in Fig. 4a shows a single wave for oxygen reduction for NAB,
which at a fixed rotation speed in filtered seawater, is similar to those found at pure
copper surfaces [27]. At potentials slightly more negative than the open-circuit value,
rotation speed has no effect and the oxygen reduction is under complete charge
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3347

0.1 0.8
Mean background

Current Density / mA cm-2

0.0 0.0

-0.1 200 rpm -0.8

Current / mA -0.2 -1.6

-0.3 -2.4
9500 rpm
EL
-0.4 -3.2
NAB
-0.5 -4.0
-1.5 -1.2 -0.9 -0.6 -0.3 0.0
(a) Potential vs. SCE / V

, Cu - Filtered & artificial

, NAB - Filtered & artificial
15
Rate Constant (x104)/ cm s-1

9500 rpm 200 rpm

10
Cu

5 NAB

0
-0.34 -0.32 -0.30 -0.28 -0.26
(b) Open Circuit Potential vs SCE/V

Fig. 4. Oxygen reduction at a wrought NAB RDE. (a) In filtered seawater measured at a potential scan
rate of 0.5 mV s
1. The straight solid line gives the potential at the limiting current and the straight broken
line gives the limits of the potential range for mass transport control. (b) Extrapolated values of the
potential dependent rate constant for pure oxygen reduction at the open-circuit potential shown for the
copper and the NAB. The measured potential becomes more negative as the rotation rate is stepped
incrementally from 200 to 9500 rpm.

transfer control. At more negative potentials, there is mixed (charge transfer and
mass transport control) of the reaction rate and rotation speed has an increasing ef-
fect as the potential is made more negative. Eventually, the potential is sufficiently
negative for mass transport control of the reaction rate and a limiting current region
is observed. At very negative potentials, the current again rises due to hydrogen evo-
lution as a side reaction [28].
LSV was performed as a function of electrode rotation rate using both the RDE
and RCE. Under saturated dissolved oxygen conditions the observed limiting cur-
rent density, jL, for oxygen reduction was dependent on mass transport predicted
by the Levich relationship for the RDE, i.e. jL is proportional to the square
root of angular velocity. The potential at the limiting current density, EL, was also
3348 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

dependent on the electrode rotation rate. For oxygen reduction at the NAB, EL val-
ues ranging from
0.810 to
1.010 V vs. SCE for rotation rates between 200 and
9500 rpm (21–994 rad s
1), see Fig. 4a.
The cathodic Tafel slopes for copper, the 90-10 copper–nickel and the NAB were
measured directly from the LSV derived data and the PSCT derived data. Compared
to the Koutecky–Levich derived PSCT Tafel data, where the mass transport control-
ling component is removed, the values of the directly measured slopes were much
higher and generally showed some degree of dependence with electrode velocity.
The NAB PSCT values were
124 mV decade
1 (RDE—filtered),
119 mV decade
1
(RDE—artificial) and
149 mV decade
1 (RCE-filtered). The equivalent LSV derived
values for the RDE in the artificial and the filtered seawaters were generally
150 and

160 mV decade
1. The differences in Tafel slope between filtered and artificial sea-
water are attributable to differences in the surface film composition which, in turn,
arise from the compositional differences between the seawater types, the filtered
electrolyte having a much more complex chemical make-up than the artificial one
[12,13].
Fig. 4b provides a useful indication of the relative charge transfer controlled oxy-
gen reduction reaction rates occurring at the corrosion potential of NAB relative to
copper. These values were produced by extrapolation of the current to OCP and
extraction of the cathodic rate constant at each rotation rate. At 800 rpm, the poten-
tial dependent rate constants in filtered seawater for copper and NAB were 4 · 10
6
and 2 · 10
6 m s
1, respectively. In the artificial seawater, the estimated rate of oxy-
gen reduction on the copper was higher at 7 · 10
6 m s
1. The rate constant mea-
sured at the NAB was comparable to that measured in the filtered seawater. In
contrast to both the 90-10 copper–nickel and copper, therefore, the rate and mech-
anism of oxygen reduction at the nickel–aluminium bronze was essentially indepen-
dent of electrolyte type.
At 25 C the equilibrium potential, Ee, for the four-electron oxygen reduction
reaction can be calculated from the standard equilibrium potential, Eoe , of
+1.228 V vs. SHE (+0.986 V vs. SCE) [29].
E ¼ 1:228
0:059 pH þ 0:0148 1og pO2 ð3Þ

The resulting equilibrium potentials (+0.502 and +0.508 V vs. SCE for the filtered
and artificial seawaters, respectively) were used as the extrapolation point for the
Koutecky–Levich equation derived Tafel slopes for oxygen reduction to determine
exchange current density values (jo) for copper and the NAB, see Table 2. In both
the filtered and artificial seawaters, the magnitude of jo for oxygen reduction varied
in the order copper > NAB. In comparison to the other analysis used, this method
would lead to a significant decrease in precision due to the large overpotentials range
over which the extrapolation to the equilibrium potential was performed. Using
measured (experimentally derived) B and ko, exchange current densities can also
be determined by
ðac zFEe Þ
jo ¼ zF ~
k o co exp ð4Þ
RT
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3349

Table 2
Estimated values of the equilibrium potential and exchange current density for oxygen reduction at copper
and wrought NAB in both the filtered and artificial seawaters
Parameter Filtered seawater BS 3900 artificial seawater
Cu NAB Cu NAB

2
2
3
1
Tafel extrapolation jo/lA cm 3 · 10 9 · 10 2 · 10 1 · 10
2
Eq. (4) jo/lA cm
2 4 · 10
2 1 · 10
2 1 6 · 10
3
ks/cm s
2 2 · 10
10 5 · 10
11 7 · 10
9 3 · 10
11
Ee vs. SCE/V +0.502 +0.508

In general, a reasonable correlation is observed between the two methods of jo

determination. In the case of the NAB in the filtered seawater, similar values were pro-
duced with Eq. (4) relative to the direct Tafel extrapolation measurement (Table 2).
Further analysis allowed the standard rate constant, ks for oxygen reduction to be
calculated using,

jo ¼ zFk s co ð5Þ

The resulting rates of ks, derived from jo values taken from Eq. (4), are also given
in Table 2. The Tafel slope, rate constant and exchange current density data confirm
that the rate of oxygen reduction at the copper is generally higher than that of the
nickel–aluminium bronze. For the copper, oxygen reduction was faster within the
artificial seawater. This was not the case for the NAB where all data indicate very
similar rates of oxygen reduction in both electrolytes.

3.3. Anodic kinetics (metal dissolution)

The anodic voltammetry of NAB in Fig. 5a [13] shows a rotation rate dependence.
Between rotation rates of 800 and 1800 rpm, a transition from the typical quasi-pas-
sivation behaviour (also commonly observed for copper [9] and 90-10 copper–nickel
[10]). For the NAB RDE above 1800 rpm, the rate of reaction is effectively inhibited
from the OCP until a critical potential is reached (Ecrit ffi +0.025 V vs. SCE). At
potentials more positive than Ecrit, the current density steps directly to a convec-
tive-diffusion controlled value of jL. It is clear from Fig. 5a that there is a definite
change in the mechanism of polarisation (and protective formation) above a critical
range of rotation rates. This change in mechanism can possibly be explained by con-
sidering the limiting mole fractions for aluminium oxide formation (N Al2 O3 ) [30], as
considered in detail elsewhere [13].
The passivation of Cu–Al alloys is based upon the common system of oxidation
resistant materials, where solute Al has a greater affinity for oxygen than solute Cu.
Under standard conditions, Al2O3 is almost eleven times more stable than Cu2O rel-
ative to their metals in the zero oxidation state. For Cu–10%Al alloys, thermal oxi-
dation is based on a rapid initial production of Cu2O from which N Al2 O3 is achieved
at the alloy/oxide interface due to the depletion of Cu. Alumina subsequently forms
3350 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

25 198
-1
NAB - 1.0 mV s

Current Density / mA cm-2

20 159

15 1800 to 9500 rpm 119

Current / mA
1200 rpm
10 200 to 800 rpm 79

5 40

0 0

-0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

(a) Potential vs. SCE / V

200 rpm, 300 rpm, 400 rpm, 600 rpm,

800 rpm, 1000 rpm, 1200 rpm, 1400 rpm
-2.1

-2.4 NAB 1400 rpm

log (Current / A)

-2.7

-3.0 200 rpm

-3.3

-3.6

-3.9
-0.270 -0.255 -0.240 -0.225 -0.210 -0.195
(b) Potential vs. SCE / V

Fig. 5. Anodic dissolution of wrought NAB. (a) Anodic linear sweep voltammetry at the nickel–
aluminium bronze RDE as a function of electrode rotation rate (excluding 4800 rpm) in aerated, filtered
seawater at a scan rate of 1.0 mV s
1. (b) Apparent Tafel slopes (corrected for oxygen reduction)
describing the anodic dissolution of nickel–aluminium bronze in the filtered seawater as a function of RCE
rotation rate.

as a protective oxide which is highly impermeable to the passage of cuprous cations

which can no longer enter what is the outermost layer of cuprous oxide. The higher
the aluminium content of the alloy, therefore, the greater the corrosion resistance
due to the protective Al2O3 film since the limiting mole fraction is achieved over a
shorter exposure time and is maintained at lower copper dissolution rates.
The anodic kinetics have also been examined at a smooth RCE surface. Within
the mixed control region (at polarisations of approximately +300 mV) for each of
the copper-based alloys, anodic currents were plotted as a series of ÔapparentÕ Tafel
slopes (Fig. 5b). These are similar to those similar to those measured at copper in
chloride media [14]. Each apparent Tafel slope was plotted at a specific electrode
rotation rate between 200 and 1400 rpm (21–146 cm s
1). In all cases, the magnitude
of current density measured at these curves increased in a predictable and reproduc-
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3351

ible manner. It is clear from the plots that the current has both charge transfer and
mass transport controlling component (mixed control). This behaviour is attribut-
able to a shift of the equilibrium potential of the anodic reaction to more negative
values as the electrode rotation rate is increased [13].

3.4. The mixed(open-circuit corrosion) potential

Using the Tafel extrapolation technique and assuming uniform corrosion, the
influence of mass transport conditions on the overall corrosion mechanism for these
metals can be related directly to the angular velocity of the electrode. It can be as-
sumed here that the polarisation behaviour close to the OCP is representative of that
occurring at the mixed potential and extrapolation of the cathodic and anodic Tafel
slopes to the point of equal current density (at the mixed potential) will yield the cor-
rosion current density. It can also be assumed that the dominant cathodic reaction
(oxygen reduction) is slow enough to be taken as reversible at the mixed potential.
The anodic reaction in these cases is generally assumed to be under mixed charge
transfer and mass transport control where the apparent Tafel slope dominates at
these potentials.
The pure charge transfer controlled current density produced from the Koutecky–
Levich equation for oxygen reduction was used to produce cathodic Tafel slopes
completely free of the influence of mass transport of oxygen. Tafel slopes were also
measured within a large potential window from which the extrapolation of currents
to the mixed potential will be more accurate. Evans diagrams, therefore, can be pro-
duced which describe the polarisation behaviour of all three copper-based alloys at
the mixed potential for both filtered and artificial seawater. A typical example is pre-
sented in Fig. 6a for the NAB RDE in filtered seawater. The extrapolated Tafel slope
for pure oxygen reduction is also shown as a solid line. The anodic half-cell reaction
is represented by the apparent Tafel slopes (not corrected for oxygen reduction) for
the electro-dissolution reaction. The anodic polarisation curves are shown as a func-
tion of electrode rotation rate. As the rate of rotation is increased, the effective inter-
cepts between the cathodic Tafel slope and the anodic slopes increases in terms of the
corrosion current and becomes more negative in terms of the mixed potential. This is
due to an increase in anodic current density with rotation rate.
Fig. 6a can be used to demonstrate the corrosion mechanism of the copper, NAB
and 90-10 copper–nickel. At constant oxygen concentration, temperature and chlo-
ride based electrolyte, it is the rate of the dichlorocuprous anion, which is corrosion
rate controlling for each metal. It is clear from the preceding discussion that the
mixed potential can be derived from either LSV derived data, PSCT data or a com-
bination of both techniques. In order to establish the optimum combination of
polarisation data, the mixed potential for copper in the filtered seawater was derived
from three possible combinations of polarisation data: (a) LSV derived cathodic and
anodic data, (b) PSCT derived cathodic and anodic data and (c) LSV derived catho-
dic data and PSCT derived anodic data. The best fit between the directly measured
and derived values was observed for case (b), where only pure PSCT data was used
to reproduce the mixed potential. The percentage deviation from the means of
3352 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

-3.5
9500 rpm
Extrapolated Tafel plot
-4.0
for oxygen reduction

log (Current / A)
-4.5
200 rpm
-5.0
Icorr

-5.5
Ecorr
-6.0 NAB

-6.5
-0.40 -0.36 -0.32 -0.28 -0.24 -0.20 -0.16
(a) Potential vs. SCE / V

, 90-10 - Derived & directly measured

, Cu - Derived & directly measured
, NAB - Derived & directly measured
Open Circuit Potential vs. SCE / V

-0.210
Filtered seawater
-0.240

-0.270

-0.300

-0.330

-0.360
0 200 400 600 800 1000 1200
(b) Angular Velocity / rad s-1

Fig. 6. Mixed potential plots at wrought NAB RCE. (a) The mixed potential for the nickel–aluminium
bronze in the filtered seawater derived via Tafel extrapolation of curves derived from steady-state,
potential step, and current transients with hydrodynamic steps. (b) Comparison of the derived (from PSCT
data) and the directly measured open-circuit potentials for copper, the 90-10 copper–nickel and NAB in
the filtered seawater.

the derived data from the directly measured ones over the whole range of angular
velocities, were averaged as 1.57 ± 0.01%. The deviations measured for cases (a)
and (c) were much greater at 5.04 ± 0.01% and 4.51 ± 0.01% for the LSVcath. and
LSVanod. and the LSVeath. and PSCTanod. curves, respectively. This trend was also
observed in both the filtered and artificial electrolytes for all of the metals examined
in this section. It was assumed from these results, therefore, that the most accurate
method for the replication of the mixed potential conditions was Tafel extrapolation
using only the PSCT polarisation data.
The directly measured and derived mixed potentials for each of the copper-based
metals are compared in Fig. 6b using the filtered seawater as an example. Generally,
the fit between the two derivations is excellent where the deviation of the means is
within the standard deviation of each technique. Diagnostic plots of logarithmic
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3353

(mixed potential) vs. logarithmic (angular velocity) were produced in order to deter-
mine the dependence of the mixed potential on fluid velocity. Analysis of the open-
circuit corrosion potential vs. rotation speed data may be used to give the velocity
dependence of the mixed potential via the relationship:
Ecorr ¼ k 1 xp ð6Þ
Overall, the results given here strongly indicate that the OCP of copper, the 90-10
copper–nickel and the NAB can be modelled using the assumption that at the mixed
potential: (a) the cathodic reaction is essentially pure oxygen reduction which is
under pure charge transfer control and (b) both charge transfer and the mass trans-
port of dichlorocuprous anion from the electrode surface to the bulk solution con-
trols the rate of the anodic reaction.

3.5. Linear polarisation resistance

The Stern–Geary relationship:

ba bc B
jcorr ¼ ¼ ð7Þ
2:3Rp ðba þ bc Þ Rp
can be used to directly calculate the corrosion current density as can Tafel data.
Tafel extrapolation derived corrosion current densities for copper, 90-10 copper–
nickel and NAB in the filtered seawater have been studied as a function of RCE
peripheral velocity. A clear dependence on fluid velocity was seen for all the cop-
per-based metals where the current densities increased in a very reproducible man-
ner. The general rates of corrosion of the metals can be given in the order
NAB > 90-10 Cu–Ni > Cu. Corrosion current densities measured at 1000 rpm for
the copper, the 90-10 copper–nickel, and the NAB were 3.7 ± 0.6, 5.6 ± 0.4 and
7.8 ± 0.5 lA cm
2, respectively.
The LPR derived relative corrosion rates in Fig. 7 show that the order of suscep-
tibility to corrosion is 90-10 Cu–Ni ffi NAB > Cu. All the copper-based metals exhi-
bited a rotation rate dependence on corrosion current density. The corrosion rate
increased with electrode peripheral velocity. At 4800 rpm, the corrosion rate mea-
sured for the 90-10 copper–nickel was 9 lA cm
2 while for the NAB it was
10 lA cm
2. The equivalent corrosion current density measured at the copper was
6 lA cm
2. Diagnostic plots of logarithmic (corrosion current density) vs. logarith-
mic (peripheral velocity) were used to determine the dependence of corrosion rate on
flow velocity. The linear regression data taken from these plots, where B is derived
from pure charge transfer controlled anodic Tafel slopes, are given in Table 3 for the
RCE. This table also gives B values derived from mixed controlled anodic current
(the apparent Tafel slopes). For both cases, the general relationship:
jcorr ¼ k 2 U q ð8Þ
may be used to give an estimate of how the corrosion current density changes with
peripheral velocity. From the largely similar values of q given in Table 3 for the
RCE, it is clear that the corrosion rates derived for the copper-based materials are
3354 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

16.0

Corrosion Current Density / µA cm-2

NAB, 90-10 Cu-Ni
Cu

12.0

8.0

4.0

0 40 80 120 160
Peripheral Velocity / cm s-1

Fig. 7. Linear polarisation resistance measurements of corrosion current density vs. RCE peripheral
velocity at wrought NAB and other copper-based RCEs in filtered seawater.

Table 3
Linear regression data describing plots of logarithmic corrosion current density vs. logarithmic RCE
peripheral velocity
Metal Slope (log[ mA cm
2]/log[cm s
1]) Intercept (log[mA cm
2])
(value of q) (value of k2)
Cu 0.276 ± 0.017
2.802 ± 0.030 (0.002 ± 0.001)
(0.304 ± 0.016) (
2.914 ± 0.029 (0.001 ± 0.001))
90-10 Cu–Ni 0.219 ± 0.005
2.437 ± 0.010 (0.004 ± 0.001)
(0.388 ± 0.044) (
2.933 ± 0.080 (0.001 ± 0.001))
Wrought NAB 0.228 ± 0.023
2.491 ± 0.042 (0.003 ± 0.001)
(0.257 ± 0.010) (
2.597 ± 0.018 (0.003 ± 0.001))
Corrosion current density data are shown derived from B values produced from pure charge transfer
controlled Tafel slopes and with mixed controlled anodic Tafel slopes (the latter being in parentheses).

similarly dependent on fluid velocity hence on the rate of mass transport. This was
also the case for the RDE and laminar flow.
Table 4 provides a summary of the B values found for NAB, Cu and Cu–Ni in
chloride media in this work and in other studies. Although a relatively wide range
of values is apparent, the wrought NAB gives broadly similar values to copper
and the 90-10 copper–nickel.

3.6. Zero resistance ammetry at a BRCE

BRCE measurements are useful in the study of galvanic coupling under well
defined turbulent flow conditions and have been used by a number of workers
[14,31–33]. Fig. 8a shows the mean free corrosion and bimetallic corrosion potential
transients (measured at the same electrode at different times) of the 90-10 Cu–Ni/
NAB couple [14]. The uncoupled metal potentials were recorded until 900 s, after
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3355

Table 4
Experimental and literature values of the proportionality constant, B, in the Stern–Geary equation
Metal Conditions B/V References
Cu Filtered seawater 0.018 [9]
(charge transfer controlled bc and ba values)
Artificial seawater 0.026
(charge transfer controlled bc and ba values)
Filtered seawater (mixed controlled ba values) 0.019 ± 0.001
Artificial seawater (mixed controlled ba values) 0.021 ± 0.001
Natural seawater 0.006 [55]
3% NaCl 0.031 [55]
Natural seawater 0.005 [56]
3–3.5% NaCl and natural seawater 0.019 ± 0.001 [57]
Natural and artificial seawater 0.014 and 0.020 [58]
90-10 Cu–Ni Filtered seawater 0.023 [10]
(charge transfer controlled bc and ba values)
Artificial seawater 0.028
(charge transfer controlled bc and ba values)
Filtered seawater (mixed controlled ba values) 0.021 ± 0.001
Artificial seawater (mixed controlled ba values) 0.019 ± 0.001
Natural seawater 0.049 [55]
Artificial seawater 0.018 [55]
Natural seawater 0.049 [59]
Natural seawater 0.018 [60]
Natural seawater 0.070 [61]
Natural seawater 0.040–0.107 [62]
Natural seawater 0.018 [18]
3–3.5% NaC1 and seawater 0.018 and 0.016 [57]
Natural seawater 0.021 [63]
Natural seawater 0.034 [64]
Natural and artificial seawaters 0.014 and 0.019 [58]
NAB Filtered seawater 0.023 [12,13]
(charge transfer controlled bc and ba values)
Artificial seawater 0.021
(charge transfer controlled bc and ba values)
Filtered seawater (mixed controlled ba values) 0.018 ± 0.001
Artificial seawater (mixed controlled ba values) 0.030 ± 0.001
3% NaCl 0.012 [55]
3–3.5% NaCl 0.018 [65]
Values incorporating mixed control Tafel slopes and mean values of B are given for RDE rotation rates
from 200 to 9500 rpm.

which the metals were galvanically coupled. The uncoupled potential of the NAB is
very similar to that shown in Fig. 3a. At 200 rpm, the steady potentials were

0.296 V vs. SCE (Cu),
0.263 V vs. SCE (90-10 Cu–Ni) and
0.301 V vs. SCE
(NAB).
The potential of the couple and the galvanic current were then measured while
rotation speed steps were applied to the BRCE. Following a 100 s settling period,
the individual potential of copper alloys becomes more negative as the rotation speed
3356 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

-0.210
90-10 Cu-Ni Point of
bimetallic
connection

Potential vs. SCE / V

-0.240

-0.270

-0.300
200 rpm 1400 rpm

NAB
-0.330
0 600 1200 1800
(a) Time Post Initiation of Rotation / s

-0.440 V, -0.478 V, -0.515 V

-0.553 V, -0.590 V
-3.0
Derived from RCE polarisation data
log (jprot / A cm-2)

-3.5 Cu/NAB

-4.0

-4.5
0.8 1.2 1.6 2.0 2.4
(b) log (Peripheral Velocity / cms-1)

Fig. 8. Galvanic coupling using a twin electrode, bimetallic RCE (BRCE). (a) Mean single metal and
bimetallic potential transients of the 90-10 copper–nickel and the wrought NAB in the aerated, filtered
seawater as a function of BRCE rotation rate [14]. (b) Logarithmic plots describing the dependence of
mean applied current density on the copper/wrought NAB BRCE peripheral velocity in the filtered
seawater as a function of applied potential.

increases due to mass transport affects on the anodic process. At 1400 rpm, for exam-
ple, the potentials become
0.315 V,
0.284 V and
0.321 V, respectively.
Over a wide range of rotation speeds, the uncoupled NAB is base with respect to
both the uncoupled copper and copper–nickel electrodes. When coupled, the mean
bimetallic potentials ranged from
0.294 to
0.303 V vs. SCE (copper/NAB) and

0.282 to
0.295 V vs. SCE (90-10 Cu–Ni/NAB) over the rotation speed range of
200–1400 rpm. The directly measured and derived (from RCE PSCT polarisation
data for the single metals) current densities resulting from the impressed current
cathodic protection of the Cu/NAB bimetallic couple are given in Fig. 8b and show
good correlation. This behaviour shown is consistent with a system showing mixed
control kinetics [14].
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3357

3.7. Galvanic effects in seawater over an extended time (pontoon studies)

3.7.1. NAB/NAB couple (NES 747 Part 2)

On immersion, the couple potential for the NAB/NAB couple, see Fig. 9a, was

0.260 V, similar to the reported corrosion potential of NAB in seawater and is con-
sidered to indicate the formation of a Cu2O film [34–36]. The couple potential in-
creased to between
0.070 and
0.080 V after 3 months which has been
attributed to the transformation of Cu2O to CuO [34,35]. The most significant
change in couple potential occurred at the start of the biofouling season when sea-
water temperatures reach about 10 C (from a minimum of approximately 6 C in
February). The galvanic current density for this same period was relatively low at
0.1 to 0.5 lA cm
2 (Fig. 9b). After approximately 5 months immersion, however,
the decoupled potentials (Fig. 9a) show that the couple had become polarised, with
a subsequent increase in galvanic current density (1.0– 2.5 lA cm
2). While initially

Fig. 9. Extended seawater exposure tests involving NES 747 Part 2 NAB (pontoon studies: February 2004
to January 2005). (a) Coupled and decoupled potential measurements for NAB/NAB. (b) Seawater
temperature and galvanic current density for the NAB/NAB.
3358 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

immune from biofouling due to the toxicity copper-ion release within the protective
oxide film, this immunity diminishes over time.
For copper-based alloys, it has been previously reported that after exposure to
natural seawater for several months, a multilayer structure of microorganisms and
extracellular polymeric substances (EPS) were found entrapped between layers of
different copper corrosion products [37]. In the presence of a biofilm, the subsequent
corrosion behaviour will vary according to the extent of the interactions between
corrosion products layers and the various components in the biofilm. Even when
the biofouling season comes to an end, the presence of a patchy biofilm can create
chemical conditions very different from those expected from the ambient environ-
ment [38]. The biofilm can act as a diffusion barrier as well as a source or sink for
chemical species that are important to corrosion processes. Factors such as pH, dis-
solved oxygen and peroxides can vary greatly at the metal/biofilm interface. Patchy
biofilms on copper–nickel have been reported to produce complex corrosion beha-
viours resulting in physical heterogeneities on the metal surface, thus leading to
the formation of differential aeration cells [39]. Another aspect related to biofilm-cor-
rosion product interactions is spalling or sloughing of corrosion products associated
with EPS in the biofilms and this has been observed for copper-based alloys exposed
to seawater [40].

3.7.2. NAB/Cu–15Ni couples

Coupled and decoupled potential measurements for the NAB/Cu–15Ni couple
can be seen in Fig. 10a. On immersion the couple potential was approximately

0.230 V. A sharp shift in the noble direction to about
0.070 V occurred in the fol-
lowing 2 months. During this time there was a low net current galvanic current den-
sity (0.1–0.5 lA cm
2), see Fig. 10b. Here, both couple components are undergoing
film formation and growth processes. After 3-months, with mature protective oxide
layers, the couple components become increasingly polarised as evident from the
decoupled potentials. The Cu–15Ni is depolarised by 0.040–0.050 V in the noble
direction while the NAB depolarises in the active direction by 0.020 V and suggests
Cu–15Ni is cathodic to NAB. Increased corrosion resistance and the cathodic nature
of the copper–nickel alloys have been attributed to enrichment in iron and nickel
within the protective oxide film [10]. Once the couple became polarised, the galvanic
current density increased significantly reaching a maximum of approximately
30 lA cm
2 between 3 and 6 months. This behaviour also coincided with the sea-
sonal increase in seawater temperature and its associated biofouling activity. When
there is an established biofilm, modification of the oxygen reduction kinetics can
occur simply by the presence of the bacteria and bacterial metabolites/enzymes,
which act as electrocatalysts [41–43]. This bacterial influence appears to be respon-
sible for the increased galvanic currents observed during this period. At the end of
the summer (after 10 months exposure) the biofouling activity diminishes and the
depolarisation of the couple components is seen to reduce with a corresponding
decrease in the galvanic current density.
It has been suggested that enzymes (such as catalase) entrapped in the EPS could
be responsible for a possible increase in cathodic currents observed here [44]. Cata-
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3359

Fig. 10. Galvanic effects on NES 747 Part 2 NAB, in seawater, over an extended time (pontoon studies:
January 2003 to December 2003). (a) Coupled and decoupled potential measurements for the NAB/Cu–
15Ni couple immersed. (b) Seawater temperature and galvanic current density for the NAB/Cu–15Ni and
NAB/Ti couples. (c) Coupled and decoupled potentials measurements for the NAB/Ti couple immersed.

lase is an integral component of bacterial cellsÕ response to oxidative stress and is be-
lieved to limit the accumulation of reactive oxygen species in the cell [45]. The mech-
anism requires the electrochemical production of hydrogen peroxide during the
oxygen reduction process, and involves electron recycling during the enzymatic
decomposition of the peroxide to water and oxygen by catalase. The oxygen gener-
ated by the enzyme can be electrochemically reduced, reaching the surface by diffu-
sion. Thus, higher cathodic currents can result. It has been established that oxides
forming the protective films on copper and copper-based alloys play a key role in
both the hydrogen peroxide and oxygen reduction reactions [9,10,43,46]. Hydrogen
peroxide reduction proceeds through the chemical oxidation of Cu2O, to yield CuO.
Cupric oxide is then electrochemically reduced to regenerate Cu2O [46]. Cu2O acts as
a redox mediator allowing fast electroreduction of H2O2. However, on a CuO sur-
face the reduction of peroxide is inhibited and H2O2 accumulates at the interface
to finally desorb into the solution. The catalytic reduction of oxygen and hydrogen
peroxide on copper–nickel alloys appears to emulate that which occurs on copper
3360 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

[47]. This has been verified in a comparison of oxygen reduction on 90-10 copper–
nickel, copper and nickel–aluminium bronze in artificial and filtered seawater [26].
It was found that surface corrosion products in each case influenced the reaction rate
where cuprous and cupric species acted to catalyse and slow the kinetics of oxygen
reduction, respectively. Similarly, differences in the reduction kinetics of oxygen and
hydrogen peroxide on 70-30 and 90-10 copper–nickels have been attributed to the
nickel content within the protective oxide film [43], since H2O2 reduction is inhibited
on nickel oxide.

3.7.3. NAB/Ti couples

The NAB/Ti couple-potential was approximately
0.230 V on immersion
(Fig. 10c) and reveals little polarisation of the NAB by the titanium, i.e. the titanium
is readily polarised. During the first month, the galvanic current density increased
from 0.1 to 10 lA cm
2 (Fig. 10b). This is possibly related to the Cu2O to CuO
transformation when the protective oxide film thickens, as indicated by the rapid
shift in couple potential to
0.050 V. As titanium does not display any toxicity to-
wards marine organisms, biofouling can occur quickly on exposed surfaces. Initially
the decoupled potential for the titanium was about +0.075 to +0.090 V more noble
than the potential of the couple, whereas the NAB was depolarised in the active
direction by between 0.020 and 0.030 V. Titanium has been reported to undergo slow
ennoblement in both fresh and brackish water [48]. Once the biofouling season
started, in the third month of exposure, a significant ennoblement of the titanium
occurred with the decoupled potentials increasing to approximately +0.150 to
+0.180 V. However, the decoupled NAB potentials remained close to that of the
couple potential with little further depolarisation occurring. The galvanic current
densities after 6 months are significantly lower than the NAB/Cu–15Ni couple at
approximately 7 lA cm
2. The influence of the biofouling season appeared to be less
dramatic for the NAB/Ti couples compared with the NAB/Cu–15Ni, probably due
to the poor catalytic activity of titanium for oxygen reduction even in the presence of
a biofilm.

3.8. Jet impingement and erosion–corrosion

Mass loss measurements for flow corrosion, pure erosion and erosion–corrosion
of NAB (NES 747 Part 2) and 4360 steel (as a reference material) are shown in Table
5. Thus, direct comparisons can be made between the different test conditions,
assuming that Jetcorrosion, Jeterosion, and Jeterosion–corrosion effects occurred uniformly
over the specimen surface. The flow corrosion rates for the NAB and steel are rela-
tively constant, whereas for the Jeterosion, and Jeterosion–corrosion test conditions there
is increasing material loss which can be attributed to the increase in erodent kinetic
energy (Ek). A higher Ek value will produce greater plastic deformation/cutting wear
[15] and for the Jeterosion–corrosion condition, the freshly exposed metal from impact
events will also result in higher corrosion rates although this will be dependent on
the repassivation characteristics.
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367
Table 5
Erosion and erosion–corrosion characteristics of NES 747 NAB in seawater
Jet velocity/m s
1 Ek/lJ Mass loss/mg Erosion Erosion–corrosion Ek exponents
Jeterosion–corrosion Jeterosion Jetcorrosion rate/lm3 rate/lm3 impact
1 for erosion
impact
1 (erosion–corrosion)
4360 steel 1.03 (0.93)
3.1 0.016 12.2 14.1 12.3 0.015 0.026
3.1 0.089 18.5 20.3 10.8 0.041 0.046
5.0 0.230 27.5 27.5 11.7 0.300 0.344
6.7 0.410 100.5 88.1 9.6 0.280 0.301
NAB 0.79 (0.87)
3.1 0.016 3.3 2.9 1.5 0.113 0.089
5.0 0.040 8.2 7.6 1.7 0.886 0.746
5.0 0.230 12.1 10.5 1.7 0.924 0.090
6.0 0.330 13.3 12.2 2.4 1.550 1.915

3361
3362 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

3.8.1. Flow corrosion

Fig. 11a shows the flow corrosion rate for cast NAB (NES 747 Part 2) and 4360
steel for the different jet velocities. The relatively high and uniform corrosion rate for
the steel is consistent with a non-passivating metal. The formation of loosely adher-
ent corrosion products will be immediately removed by the impinging jet, exposing
fresh surfaces and allowing corrosion to continue. In contrast to the behaviour in
these aggressive flow conditions, NAB is generally regarded as remaining passive
under static and a range of flow conditions [1]. The measured flow corrosion rate
of cast NAB (0.5 to 0.8 mm y
1) is similar to those reported by Ault for cast
NAB alloys: between 0.5 and 2.0 mm y
1 at 7.6 m s
1 [8]. In comparison, corrosion
rates of 0.015 mm y
1 are typically reported for NAB alloys under static conditions
[1]. An SEM micrograph of the cast NAB exposed to a jet impingement of 5 m s
1
shows evidence of corrosion specifically of the a-phase [49] (Cu-rich with an Al con-
tent of around 7–8%: the limit is 9.6% Al in the a-phase [50]). The a-phase is more
susceptible to flow corrosion than the j-phases [51]. Under the jet impingement con-
ditions, corrosion of the a-phase could occur before the formation of the protective
oxide film, particularly over the relatively short test duration. Also, it is considered
that the shear stresses generated at the metal surface could be sufficient to either pre-
vent initial formation or cause damage to the developing protective film [2]. Fig. 11a
shows that the corrosion rate of NAB increases above a jet velocity of 5 m s
1, and
indeed for naval service conditions a maximum flow rate of 5 m s
1 is often the rec-
ommended limit.

3.8.2. Erosion
In addition to gravimetric mass loss measurements, erosion rates have been ex-
pressed as volume loss per impact (DVu) to allow for sand particle size variations
at a given volume fraction:
Dmpd 3
DV u ¼ ð9Þ
6qQv tC v
The erosion rate (DVu) is related to the kinetic energy, Ek by
l
DV u aðEk Þ ð10Þ
The kinetic energy Ek for erosion experiments is given by
Ek ¼ 1=2 merod m2 ð11Þ
The mean erodent velocity within the slurry jet is assumed to be equivalent to the jet
velocity.
Table 5 and Fig. 11b show the relationship between kinetic energy, Ek, and ero-
sion rates under pure erosion (DVu) and erosion–corrosion (DVuT) conditions. The
erosion rates for cast NAB and steel generally increase with increasing kinetic energy
with the cast NAB giving superior erosion performance in this case. Kinetic energy
exponents in Table 5 have been obtained for cast NAB and 4360 steel. Ductile mate-
rials have energy exponent values close to unity while brittle materials have values
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3363

Fig. 11. Jet impingement studies on NES 747 Part 2 NAB. (a) Flow corrosion rates vs. jet velocity.
(b) Relationship between erosion and erosion–corrosion rates vs. erodent kinetic energy.

between 2 and 3 [52–54]. Ductile erosion mechanisms are kinetic energy related,
whereas brittle materials are influenced by the presence of flaws that accelerate dam-
age rates. The studies show that cast NAB has an energy exponent close to unity,
which is consistent with the behaviour expected from a ductile material.

3.8.3. Erosion–corrosion
The relationship between DVu and erodent kinetic energy under erosion–corro-
sion condition shows the material loss increased with kinetic energy, see Fig. 11b.
For the cast NAB the kinetic energy exponents under erosion–corrosion conditions
3364 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

(l = 0.93) are slightly lower than the pure erosion conditions (l = 1.03), see Table 5.
The lower kinetic energy exponent is associated with the ability to form an adherent
protective oxide film under corrosive conditions. This could result in a reduction in
the wear rate, especially when the oxide layer remains intact under particle impinge-
ment. Conversely, for the 4360 steel, which is a non-passivating material, the kinetic
energy exponent under erosion–corrosion (l = 0.87) is higher than pure erosion con-
ditions (l = 0.79). In this instance uniform corrosion, coupled with the possibility of
stress corrosion cracking of the work hardened plastic deformation lips contribute to
an increased wear rate under erosion–corrosion conditions.

4. Conclusions

(i) The corrosion characteristics of NAB alloys have been studied using short term
electrochemical techniques, impingement studies and longer term immersion
trials.
(ii) A range of electrochemical techniques have been used to examine the electro-
chemistry of freshly polished NAB. It was observed in this case that a potential
step current transient method gave superior quantitative reproducibility in rel-
ative to linear sweep voltammetry. Overall corrosion rates measured with both
Tafel extrapolation and linear polarisation resistance were in broad agreement.
For thinly filmed surfaces, the overall electrochemical mechanism and mea-
sured corrosion rates of NAB were similar to that measured for copper and
90-10 copper–nickel. In each case, a mass transfer influenced component of
the anodic reaction provided the fluid flow dependency commonly observed
in the field.
(iii) In longer term exposure tests, accelerated corrosion rates were measured for
both the NAB/Ti and NAB/Cu–15Ni couples compared to uncoupled NAB.
This was the case particularly during the biofouling season where there was
an increased likelihood of biofilm formation and activity. The mechanism of
the noble depolarisation in the presence of a biofilm probably involved a
change in the kinetics of the oxygen reduction reaction. For the NAB/Cu–
15Ni couple, significant increases in the cathodic current have previously been
linked to copper species within the protective oxide film acting to catalyse the
reduction reactions. The NAB/Ti couple was less affected by the seasonal bio-
fouling activity, probably due to the relatively poor catalytic activity of the
oxide layer for oxygen reduction even in the presence of a biofilm.
(iv) Jet impingement velocities between 3 and 5 m s
1 have demonstrated the supe-
rior corrosion performance of NAB in contrast with non-passivating carbon
steel. However, NAB corrosion rates were enhanced above a flow velocity of
5 m s
1. The erosion and erosion–corrosion performance of NAB was consis-
tent with a ductile metal undergoing plastic deformation processes during sand
impacts. The ability to form an adherent protective oxide film under corrosive
conditions resulted in a lower kinetic energy exponent when compared with the
pure erosion condition.
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367
Table 6
A comparison of the corrosion rates (expressed as a corrosion current density) of NAB in seawater under various conditions
NAB type Electrode Seawater Technique Corrosion current
geometry type density/lA cm
2
Wrought bar RDE 20.9 rad s
1 Artificial Tafel extrapolation 5.6
RDE 146 rad s
1 (Laboratory cell) 8.8
Wrought bar RCE 0.20 m s
1 Filtered LPR 7.5
RCE 1.50 m s
1 (Laboratory cell) 10.9
British Naval specification WJDE 3.1 m s
1 Natural Weight loss 20
(cast then annealed) 747 Part 2 WJDE 6.0 m s
1 (Laboratory cell) 32
British Naval specification Tidal immersion Natural ZRA After 6-months
(cast then annealed) 747 Part 2 (long-term) (Tidal condition) 0.7–2.5

3365
3366 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367

(v) Table 6 provides a summary of the corrosion rates experienced under the var-
ious conditions used in this paper. The corrosion rate is expected to be depen-
dent on the nature of surface films and hence on factors including (a) the
composition of the seawater electrolyte, (b) the metallurgical composition, his-
tory and surface roughness of the NAB, (c) the electrode/electrolyte geometry,
(d) the magnitude and type of electrolyte flow, (d) the presence of erodent and
(f) the time of immersion.

Acknowledgments

The authors are grateful to Dstl (Defence Science and Technology Laboratory)
UK and QinetiQ-Haslar, UK for financial contributions to the research programme.
The wrought NAB alloy for the electrochemical studies was supplied by Stone Man-
ganese Ltd. The remainder of the copper alloys were supplied by Dr. Clive Tuck
from Meighs Ltd. The contents of this paper include material subject to
Crown
Copyright 2004 Dstl. Some of the electrochemical data in this paper was obtained
during Gareth KearÕs PhD at the University of Portsmouth, UK.

References

[1] A.H. Tuthill, Mater. Performance 26 (9) (1987) 12.

[2] H.I. Meigh, Cast and Wrought Aluminium Bronzes—Properties, Processes and Structure, Institute of
Materials, London, 2000.
[3] P.R. Howell, On the Phases Microconstituents in Nickel–aluminium Bronzes, Copper Development
Association Inc., A1310-XX/OO, 2000.
[4] M. Cook, W.P. Fentiman, E. Davis, J. Inst. Met. 80 (1951) 419.
[5] H.S. Campbell, Aluminium Bronze Corrosion Resistance Guide, Publication 80, Copper Develop-
ment Association, UK, July 1981, pp. 1–27.
[6] A. Schüssler, H.E. Exner, Corros. Sci. 34 (1993) 1793.
[7] B.G. Ateya, E.A. Ashour, S.M. Sayed, Corrosion 50 (1994) 20.
[8] J.P. Ault, Erosion Corrosion of Nickel–aluminium Bronze in Flowing Seawater, Corrosion 95, Paper
No. 281, NACE International, Houston, Texas, 1995.
[9] G. Kear, B.D. Barker, F.C. Walsh, Corros. Sci. 46 (2004) 109.
[10] G. Kear, B.D. Barker, K.R. Stokes, F.C. Walsh, J. Appl. Electrochem. 34 (2004) 659.
[11] A. Schüssler, H.E. Exner, Corros. Sci. 34 (1993) 1803.
[12] G. Kear, B.D. Barker, K.R. Stokes, F.C. Walsh, J. Appl. Electrochem. 34 (2004) 1235.
[13] G. Kear, B.D. Barker, K.R. Stokes, F.C. Walsh, J. Appl. Electrochem. 34 (2004) 1241.
[14] G. Kear, B.D. Barker, K.R. Stokes, F.C. Walsh, Corros. Sci. 47 (2004) 1694.
[15] J.B. Zu, I.M. Hutchings, G.T. Burstein, Wear 140 (1990) 331.
[16] Y. Puget, K.R. Trethewey, R.J.K. Wood, Wear 233–235 (1999) 522.
[17] G. Faita, G. Fiori, D. Salvadore, Corros. Sci. 15 (1975) 383.
[18] S.R. de Sanchez, D.J. Schiffrin, Corros. Sci. 22 (1982) 585.
[19] G. Venkatachari, K. Kannan, Bull. Electrochem. 9 (1993) 400.
[20] G. Kar, T.W. Healy, D.W. Fuerstenau, Corros. Sci. 13 (1973) 375.
[21] C. Deslouis, B. Tribollet, G. Mengoli, M. Musiani, J. Appl. Electrochem. 18 (1988) 374.
[22] P.A. Lush, M.J. Carr, Corros. Sci. 19 (1979) 1079.
[23] R.J.K. Wood, S.P. Hutton, D.J. Schiffrin, Corros. Sci. 30 (1990) 1177.
 J.A. Wharton et al. / Corrosion Science 47 (2005) 3336–3367 3367

[24] L.E. Eiselstein, B.C. Syrett, S.S. Wing, R.D. Caligiuri, Corros. Sci. 23 (1983) 223.
[25] G.P. Power, I.M. Ritchie, Electrochim. Acta 26 (1981) 1073.
[26] G. Kear, Electrochemical Corrosion of Marine Alloys under Flowing Conditions, PhD thesis,
University of Portsmouth, UK, 2001.
[27] G. Kear, C. Ponce de Leon, F.C. Walsh, Chem. Eng. Educ. J. 39 (1) (2005) 14.
[28] F.C. Walsh, A First Course in Electrochemical Engineering, The Electrochemical Consultancy,
Romsey, 1993.
[29] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon Press, Oxford,
1966.
[30] J.C. Scully, The Fundamentals of Corrosion, Pergamon Press, Oxford, 1990.
[31] F.B. Mansfeld, Corros. Sci. 32 (1976) 390.
[32] F.B. Mansfeld, J.V. Kenkel, Corros. Sci. 33 (1977) 236.
[33] F.B. Mansfeld, J.V. Kenkel, Corros. Sci. 33 (1977) 370.
[34] S.A. Campbell, G.J.W. Radford, C.D.S. Tuck, B.D. Barker, Corros. Sci. 58 (2002) 57.
[35] G. Bianchi, P. Longhi, Corros. Sci. 13 (1973) 853.
[36] R. Barik, J.A. Wharton, R.J.K. Wood, K.R. Stokes, in: Corrosion 2004, Paper No. 04301, NACE
International, Houston, Texas, 2004.
[37] G. Blunn, in: S. Barry, D.R. Houghton, G.C. Llewellyn, C.E. OÕRear (Eds.), Biodeterioration 6,
CAB International, London, 1986, p. 567.
[38] S.C. Dexter, Biofouling 7 (1993) 97.
[39] H.A. Videla, W.G. Characklis, Int. Biodeterior. Biodegrad. 29 (1992) 195.
[40] H.A. Videla, M.F.L. de Mele, G.J. Brankevich, in: Corrosion/89, Paper No. 291, NACE
International, Houston, Texas, 1989.
[41] D.J. Schiffrin, S.R. de Sánchez, Corrosion 41 (1985) 31.
[42] J.P. Busalmen, M. Vázquez, S.R. Sánchez, Electrochim. Acta 47 (2002) 1857.
[43] S. Ceré, M.V. Vazquez, S.R. de Sánchez, D.J. Schiffrin, J. Electroanal. Chem. 470 (1999) 31.
[44] J.P. Busalmen, M. Váaquez, S.R. Sánchez, Electrochim. Acta 47 (2002) 1857.
[45] H. Schellhorn, FEMS Microbiol. Lett. 131 (1994) 113.
[46] M.V. Vazquez, S.R. de Sánchez, E.J. Calvo, D.J. Schiffrin, J. Electroanal. Chem. 374 (1994) 179.
[47] F. King, M.J. Quinn, C.D. Litke, J. Electroanal. Chem. 385 (1995) 45.
[48] S.C. Dexter, H.J. Zhang, in: Proc. 11th Int. Corr. Congr., 1990, p. 4333.
[49] R.C. Barik, J.A. Wharton, R.J.K. Wood, K.S. Tan, K.R. Stokes, Wear 259 (2005) 230.
[50] D.T. Hawkins, R. Hultgren, in: T. Lyman (Ed.), Metals Handbook, Metallography, Structures and
Phase Diagrams, eighth ed., American Society for Metals, Metals Park, OH, 1973, p. 259.
[51] F. Hasan, A. Jahanafrooz, G.W. Lorimer, N. Ridley, Metall. Trans. A 13A (1982) 133.
[52] A.J. Moore, R.J.K. Wood, Erosive Wear Mapping in Pipe Materials, Plastic Pipes VIII, E1/4,
Koningshof, Holland, September, 1992, pp. 1–10.
[53] G.L Sheldon, I. Finnie, J. Eng. Ind. 88 (1966) 393.
[54] K.S. Tan, R.J.K. Wood, K.R. Stokes, Wear 255 (2003) 195.
[55] L.M. Callow, J.A. Richardson, J.L. Dawson, Br. Corros. J. 11 (1976) 123.
[56] J.C. Rowlands, M.N. Bentley, Br. Corros. J. 7 (1972) 42.
[57] H.P. Dhar, R.E. White, G. Burnell, L.R. Cornwell, R.B. Griffin, R. Darby, Corrosion-NACE 41
(1985) 317.
[58] F.B. Mansfeld, G. Liu, H. Xiao, C.H. Tsai, B.J. Little, Corros. Sci. 36 (1994) 2063.
[59] F.P. Ijsseling, Corros. Sci. 14 (1974) 97.
[60] H. Grubitish, F. Hilbert, R. Sammer, Werkst. Korros. 17 (1966) 760.
[61] D.D. MacDonald, B.C. Syrett, S.S. Wing, Corrosion 35 (1979) 367.
[62] D.D. MacDonald, B.C. Syrett, S.S. Wing, Corrosion 34 (1978) 289.
[63] M.R. Reda, J.N. Alhajji, J. Univ. Kuait-Sci. 20 (1993) 171.
[64] J.N. Alhajji, M.R. Reda, Corrosion 49 (1993) 809.
[65] J. Mathiyarasu, N. Palaniswamy, V.S. Muralidharan, Proc. Indian Acad. Sci.—Chem. Sci. 111 (1999)
377.