British Corrosion Journal

ISSN: 0007-0599 (Print) (Online) Journal homepage: http://www.tandfonline.com/loi/ycst19

Accelerated Atmospheric Corrosion Of Copper And

Copper Alloys

A. STERLING, A. ATRENS & I. O. SMITH

To cite this article: A. STERLING, A. ATRENS & I. O. SMITH (1990) Accelerated Atmospheric
Corrosion Of Copper And Copper Alloys, British Corrosion Journal, 25:4, 271-278, DOI:
10.1179/000705990799156391

To link to this article: http://dx.doi.org/10.1179/000705990799156391

Published online: 18 Jul 2013.

Submit your article to this journal

Article views: 8

View related articles

Citing articles: 6 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=ycst19

Download by: [UNSW Library] Date: 05 November 2016, At: 02:30

Accelerated atmospheric corrosion of copper and
copper alloys
A. STERLING The corrosion behaviour of a range of readily available commercial copper based alloys
A.ATRENS was examined in three accelerated atmospheric corrosion tests: acidified salt spray,
neutral salt spray at 80°C, and the GM (salt spray with thermal cycling) test. The
I. O. SMITH samples exposed to the G M test evinced very little corrosion. During exposure, the
weight of all the samples fluctuated in a random fashion, indicating that the corrosion
products were only partially adherent and became detached from the surface at random
times. The sequence of corrosion product formation on the specimen surfaces during
exposure to the neutral salt spray test was very similar to that observed on coppers
exposed to the atmosphere, indicating that this test can be used to give accelerated
indications concerning atmospheric corrosion in humid marine regions. Four 24 h
cycles of salt spray exposure are approximately equivalent to 1year's exposure to the
atmosphere. Alloying copper with phosphorus, tin, zinc, or silicon does not improve
atmospheric corrosion resistance. The alloy containing 30%Zn exhibited dezincijica-
tion. The nickel silvers exbibited corrosion rates in the neutral salt spray at 800C that
were lower by a factor of at least 13 than those of the high purity coppers. For six of the
alloys studied, the as received surface condition correlated with a slightly, but statisti-
cally significantly, higher corrosion rate, attributed to residual contaminants from
rolling.
Manuscript received 27 February 1987; in final form 25 October 1988. At the time the
work was carried out, the authors were in the Department of Mining and Metallurgical
Engineering, University of Queensland, St Lucia, Qld 4067, Australia. Dr Sterling is
now with Wilson Walton International, Mansfield, Qld and Dr Smith is with CRA
Advanced Technical Development, Connington, Western Australia.

INTRODUCTION EXPERIMENTAL PROCEDURE

The research presented in the present paper forms part of Test coupons
a wider investigation into the external corrosion of copper/ Details of the range of commercial alloys tested are given
brass automobile radiators. A standard radiator consists of in Table 1. Coupons, 100 x 75 mm in size, cut from rolled
a core connected to top and bottom header tanks. A sheet and drilled with a suspension hole, were stamped
typical core consists of about 50 parallel brass (often 70-30 with the appropriate alloy designation. Before testing,
brass) water channels, between which are cooling fins they were degreased in alcohol and washed in water, both
made from a finstock alloy that is essentially high purity processes being accompanied by rubbing with a rubber
copper containing 0·15 Sn. The service life of such radia-
%
stopper to remove any non-adherent material on the
tors can be limited in some geographical areas, for exam- surface.
ple in the south eastern regions of the USA, in Japan, and
in Australia, where they are exposed to an environment
Test methods
characterised by high temperatures, high humidity, and
Acidified salt spray
the presence of sea salt in the atmosphere. Failure modes
The testing conditions were as set out in ASTM
include rapid exterior corrosion of the cooling fins and
B287-1980, 'Standard method of acetic acid - salt spray
leaks in the water channels due to dezincification of the
(fog) testing', but with the temperature maintained at
brass. 50 ± 2°C. For this test, as well as for the other two tests,
The performance, corrosion mechanisms, and corrosion
the compressed air supply to the atomising nozzles was
products of a range of readily available commercial alloys
humidified in a saturation tower maintained 5 K above the
in three accelerated atmospheric corrosion tests has been
chamber temperature. This was done to prevent cooling of
studied. The tests used were:
the chamber and evaporation of water from the salt
(i) acetic acid salt spray
solution, thereby increasing its concentration. The pH of
(ii) neutral salt spray
the 50g1-1 NaCl spray solution was adjusted to 3·2±0·1
(iii) the GM test, which is designed to model the service
by the addition of glacial acetic acid. Analytical grade
environment of a radiator.
chemicals and distilled water were used. Specimens were
Direct prediction of service life is not expected from such
suspended vertically, by means of glass hooks; from cross-
accelerated tests, but they should provide the possibility of
members in the salt spray cabinet. The testing cycle lasted
quickly examining trends and ranking alloys.
24 h with the specimen exposed to the solution for 16 h
The scientific literature relating to accelerated atmos-
followed by washing and exposure to laboratory air for
pheric corrosion tests includes little in the way of analysis
8h.
of the corrosion products formed on copper alloys and
how these might influence the resulting corrosion rates.
The present study was carried out with this in mind. Neutral salt spray at BO°C
Consideration was also given to the influence of surface The testing conditions were as set out in ASTM
condition on corrosion rates, since it has been suggested1 Bl17-1979, 'Standard method of salt spray (fog) testing',
that the corrosion rates of as received copper alloys could but with the temperature maintained at 80 ± 5°C. The
be greater than those of carefully cleaned samples due to 50 g 1-1NaCl spray solution was maintained at pH 6·5-7·2.
the presence of residual surface contaminants from rolling. Analytical grade salt and distilled water were used. The

Dr. Corros. J., 1990,Vol. 25, No.4 271

272 Sterling et al. Accelerated atmospheric corrosion of copper

Table 1 Alloy designations, specifications, and compositions

Related
Alloy Designation specifications Composition, wt-%

Tough pitch copper 101 AS 1566-110A; >99·9Cu-0·030

BS 2870, 2875-C101;
ASTM B152, B11
Phosphorus 102 AS 1566-122A; >99·9Cu-0·025P
deoxidised copper BS 2870, 2875-C106;
ASTM B152, B11
Cu-Sn finstock alloy 144 AS 1566-144D 99·8Cu-0·15Sn-0·02P
90-10 brass 203 AS 1566-220A; 90Cu-10Zn
BS 2870-CZ101;
ASTM B36
70-30 brass 207 AS 1566-260A; 70Cu-30Zn
BS 2870-CZ106;
ASTM B36
Phosphor bronze 523 AS 1566-518B; 95Cu-5Sn, trace P
BS 2870, 2875-PB102;
ASTM B103-C51000
Coinage bronze 548 AS 1566-404D 97Cu- 2·5Zn-0· 5Sn-0·03P
Silicon bronze 801 AS 1566-655A; 96Cu-3Si-1Mn
BS 2870, 2875-CS101;
ASTM B96-C65000
18% nickel silver 905 AS 1566-770B; Zn-55Cu-18Ni-0·2Mn
BS 2870-NS107;
ASTM B122-C77000
75-25 cupronickel 912 AS1566-713B 74·7Cu-25Ni-0·3Mn
12% nickel silver 926 AS 1566-757B; Zn-63Cu-12Ni-0·2Mn
BS 2870-NS104;
ASTM B122-C75700

specimen suspension and test cycle were identical to those procedure was used for all samples, it is considered that
used in the acidified salt spray test. the comparisons made are valid.
After the completion of this first series of corrosion
GM test tests, part of the corrosion products was removed from
This test requires controlled cycling between 38 and 85°C, each specimen and an X-ray powder diffraction pattern
which was accomplished by mounting the specimens (at an was obtained using Cu Ka radiation. From the diffraction
angle of 60° to the horizontal) on a radiator through which patterns, the identities of the corrosion products were
warm or hot water was pumped as required. The test determined and the relative concentration of each was
requires the following 20 min cycle to be run continuously: calculated assuming a linear relationship between concen-
(i) 10 s salt spray (fog) while specimen is at 38°C tration and X-ray peak intensity. The percentage error
(ii) hold for 1 min at 38°C consequent on this, the simplest possible assumption, is
(iii) increase temperature to 85°C and hold for 10 min unlikely to be greater than 20 %
•

(iv) cool to 38°C and hold at this temperature to end of

20 min period. Second series of tests
The second series of corrosion tests was designed to
First series of tests measure corrosion penetration over 10 cycles of exposure
The first series of tests was designed to monitor corrosion to 80°C neutral salt spray. The specimens were either
behaviour, and to determine how it changed with time, in cleaned as described previously (designated as received)
each of the accelerated corrosion tests. The coupons were or acid cleaned for 5 min in 5%HzS04 followed by rubbing
examined, cleaned, dried, and weighed after each 24 h with a rubber stopper under flowing water to remove the
cycle of exposure. The cleaning procedure consisted of surface oxide (designated acid cleaned). The weight in this
washing in water while rubbing with a rubber stopper, so condition (before exposure to the environment) is desig-
removing loose deposits and corrosion products, but leav- nated Wbefore' After exposure to 10 cycles of neutral salt
ing any adherent corrosion products. This procedure was spray at 80°C, the specimens were washed, acid cleaned,
considered to have minimal effect on subsequent corrosion washed, dried, and weighed, to give Wafter' Separate
behaviour. The specimens were photographed, in colour, experiments were also carried out to measure the mass of
periodically to record the pattern of corrosion products material removed by acid cleaning a previously cleaned
and how this pattern changed with number of exposure specimen ~ w. The weight loss due to exposure to the
cycles. Subsequently, the specimens were sectioned and environment studied ~ W is given by
examined microscopically to determine the degree of non- ~ W = Wbefore - Wafter - ~ W (1)
uniform corrosion in terms of intercrystalline corrosion,
dezincification, and pitting. Sectioning was done at
random and a section of the order of 20-30 mm in length RESULTS AND DISCUSSION
was examined. The maximum depth of pitting attack was First series of tests
measured on these sections as the distance between the pit The weight change results for the specimens exposed to
bottom and the specimen surface. Such measurements do acidified salt spray, neutral salt spray, and the GM test are
not in~lude any contribution from general corrosion of the given in Tables 2-4 respectively. The results are presented
sample. This procedure does not allow for the possible in detail to draw attention to the considerable spread in
existence of extremely deep pits elsewhere on the surface, the weight changes. In fact, the results for all the speci-
but· since pitting is a statistical process and the same mens in all three tests display no discernible pattern, but

Br. Corros. J., 1990, Vol. 25, NO.4

 Sterling et al. Accelerated atmospheric corrosion of copper 273

Table 2 Weight change per cycle during exposure to acidified salt spray, mg

Alloy

Cycle 101 102 144 203 207 523 548 801 905 912 926

1 -61·0 -70·4 -4·6 -17·9 +1·3 -12·1 -40·1 -6·5 -12·0 -16·7 -29·1
2 -159·6 -98·1 -7·4 -1·8 +4·6 -13·0 -10·9 -1'8 -2·8 -28·8 -13·4
3 -23·2 -64·4 -18·6 0·0 +5·4 -24·4 -24·4 +4·4 -2·4 -14·6 -7·7
4 -7·3 -24·7 -23·0 -3·5 +4'1 -6·9 -7·9 -2·9 -2·4 -16·6 -8·1
5 -1·7 -8·2 -5·9 -2·4 +6·1 -11·3 -0·6 +5·7 +0·8 -28·2 -6·4
6 -6·3 -42·0 -8·6 +20·7 +10'8 -33·2 -24·6 +23·9 -3·0 -57·2 -7·2
7 -11·8 -40·2 -16·1 +5·1 +3·8 -27·1 -16·2 -25·6 -8·7 -27·5 -13·7
8 +7·2 -11·6 -1·5 +4·9 '+2·4 -21·5 -5·6 +7·7 -4·6 -21·4 -7·2
9 +6·4 +4·7 +0·9 +12·5 +4·5 -17·4 -13·0 +30·6 -5·4 -35·6 -5·0
10 -32·2 -6·1 -39·5 +9·2 +1·8 -6·0 -13·9 -10·2 -4·4 -52·7 -8·6
11 -3·0 -3·9 -3·5 +6·3 +3·7 -45·1 -8·2 -30·2 -4·0 -27·6 -10·0
12 -17·9 -25·5 -13·5 -5·9 +5·6 -9·6 -14·7 -46·7 -8·4 -37·0 -9·9
13 -2·1 -23·0 -3·8 +6·6 +2·1 -18·3 -22·6 -46·5 -5·4 -31·1 -12·5
14 -0·3 -22·3 -47·4 -3·0 -0·9 -28·8 -26·3 -23·7 -7·4 -26·2 -12·7
15 +0·4 -16·7 -4·6 +7·5 +4·3 -0·2 -17·4 +4·6 -0·7 -36·7 -9·0
16 +0·8 -13·1 -0·4 +6·4 +5·0 -15·0 -17·5 -8·4 -6·7 -22·3 -14·7
Total -311·6 -465·5 -197·5 +51·7 +64·6 -289·9 -241·3 -125·6 -77·5 -480·2 -175·2

appear to be random - even in respect of gains or losses in behaviour of the alloys in the two environments. Phosphor
weight. The weight losses during each GM test (Table 4) bronze (523), coinage bronze (548), and 75-25 cupro-
were much less than those measured during exposure to nickel (912) showed the least pitting attack with maximum
the other two environments. This is simply a reflection of penetration in the range 0-8 ~m. Medium penetration in
the fact that the exposure conditions during the GM test the range 11-18 ~m was observed for 12% nickel silver
are milder than those in the other two tests. (926), 18% nickel silver (905), phosphorus deoxidised
The weight change results reflect two different pro- copper (102), tough pitch copper (101), and eu-Sn fin-
cesses: weight increases as corrosion products are formed stock alloy (144), although both nickel silvers showed no
at the specimen surface, but decreases as non-adherent measurable pitting penetration after exposure to the
corrosion products are detached from the surface. The fact neutral salt spray. The alloys with the highest pitting
that the weight change per cycle is random with no penetration, in the range 21-80 ~m, were 90-10 brass
discernible pattern indicates that the corrosion products (203), silicon bronze (801), and}0-30 brass (207). The 70-
are only partially adherent and that variable amounts of 30 brass was the only alloy to exhibit dezincification.
corrosion products are detached from the surface at There is no apparent pattern of pitting susceptibility with
random. This implies that the corrosion products are at alloy composition. The corrosion products and the appear-
best only partially protective, and moreover implies that ances of the specimens were quite different for the three
weight change data cannot be used to predict corrosion accelerated atmospheric corrosion tests, as detailed below.
rate.
The maximum depths of pitting following exposure to Acidified salt spray test
acidified salt spray and to neutral salt spray, recorded in The corrosion products and their concentrations produced
Table 5, show a fairly close correlation between the during exposure to acidified salt spray are given in Table 6,

Table 3 Weight change per cycle during exposure to neutral salt spray at 80°C, mg

Alloy

Cycle 101 102 144 203 207 523 548 801 905 912 926

1 -32·3 -72·8 -1·3 +33·5 +35·7 -63·3 +24·6 +18·3 +20·0 +14·1 -+16·2
2 +8·4 -4·6 +21·4 +19·2 +25·8 -51·9 -23·5 +27·1 +4·9 +10·1 +11·5 .
3 +7'8 +5·7 +0·8 +10·6 +4·8 -28·5 -2·9 . +5·4 +4·4 +8·9 -1·0
4 +7·6 +6·9 +6·3 +3·6 +13·0 -7·2 +3·5 +8·1 +2·2 -2·2 +4·8
5 -9·2 -10·1 +11·8 +8·6 +9·7 -94·6 +13·3 +33·3 +10·2 +3·4 +7·5
6 -21·8 -150·8 +16·8 +41·9 +14·9 -27·4 +5·0 +17·9 +0·3 +2·0 +2·8
7 -3·9 -6·2 -143·4 -0·2 -5·2 -18·8 +1·5 +4·3 +1·6 +0·6 +1·7
8 +1·9 -0·5 -15·3 +8·2 +4·1 -87·7 +4·8 +9·5 +0·7 +1·4 -1·0
9 +8·1 -0·8 -4·4 +12·8 +8·9 -27·6 +6·8 +7·5 +4·1 +2·4 -0·3
10 +3·3 +7·5 -0·1 +0·3 +7·9 -57·2 +7·1 +9·4 +2·0 +2·2 -0·2
11 +4·7 -9·0 +1·2 +16·5 +3·7 -37·2 +2·7 +9·2 -0·7 -0·9 -1·3
12 +6·6 +12·9 +6·9 +2·0 +11·1 -52·0 +3·5 +12·6 +1·3 +1·4 +5·7
13 +2·5 -0·3 +7·2 -7·9 -0·3 -71·7 +3·3 +5·2 -2·9 -1·3 -8·9
14 +14~4 +9·7 +10·0 +6·9 +4·7 -26'3 +8·7 +17·3 +1·5 +1·3 +4·0
15 +13·5 +5·3 +7·2 +8·2 +6·8 -21·4 +15·8 +6·6 +0·7 +1·9 +7·1
16 -4·7 -18·1 -4·7 -3·4 +12·0 -63·2 -14·8 +9·8 +0·4 +0·5 -0·5
17 +10·9 -6·9 -7·2 +17·8 -0'8 -12·9 +5·5 +2·6 +0·8 -0·3 -0·4
18 +17·0 +15·5 +19·5 +13·6 +15·0 -53·7 +4·5 +7·8 +1·6 +2·1 +1·4
19 -6·7 +8·7 -15·2 +5·9 -6·7 -32·6 -42·1 +7·7 -1·7 -1·4 -1·6
Total +27·1 -207·9 -82·5 +218·1 +165·1 -835·2 +27·3 +219·4 +51·2 +46·2 +47·5

Br. Corros. J., 1990, Vol. 25, NO.4

274 Sterling et al. Accelerated atmospheric corrosion of copper

Table 4 Weight change per cycle during exposure to GM test, mg

Alloy

Cycle 101 102 144 203 207 523 548 801 905 912 926

1 +2·0 +1·5 +0·6 -0·6 +3·4 +2·9 -4·8 +0·6 +1·7 -3·5 +2·0
2 -0·8 -0·5 -0·9 +3·3 -0·7 -3·2 +3·6 -0·9 -5·2 +1·9 -6·3
3 +1·3 +1·4 +1·1 -1·6 +1·6 +5·0 +4·0 +0·6 +1·7 -2·2 +6·9
4 -0·1 -0·4 +0·2 +2·4 -0·1 +0·0 -10·1 +0·1 +1·8 +1·1 -3·7
5 +1·4 +1·0 +2·3 +3·2 +2·1 -1·0 +13·8 +2·2 -0·3 +0·7 +4·8
6 -0·6 -1·0 +0·0 -3·1 -0·6 +2·9 -4·1 -1·2 +0·3 -1·4 -5·9
7 +0·6 +0·2 +1·0 +3·5 +1·3 +1·0 +5·5 +3·9 +0·6 +0·6 -0·4
8 -0·4 -0·6 +0·6 -0·9 +0·5 +3·7 +0·2 +1·1 -0·5 -0·7 +4·6
9 +1·0 +1·3 +1·5 +5·7 +2·3 -1·6 +3·5 +2·3 -6·8 +2·7 -6·3
10 -1·7 -1·6 -1·7 -1·2 -0·9 +2·8 -5·8 +0·3 +4·6 -3·3 +5·3
11 +0·4 +0·4 +0·9 +2·2 +1·0 +2·5 -1·7 +1·2 +1·6 +0·2 -0·3
12 +1·9 +3·2 +3·3 -0·6 +2·0 +2·6 -1·0 +1·5 -2·3 -1·8 -4·3
13 -0·6 +0·4 +0·4 +0·0 +0·3 -2·0 -2·6 +1·1 -1·5 +0·5 +7·7
14 +1·0 +0·0 +0·2 +7·2 +0·4 +6·5 +8·7 +1·0 +9·3 +3·5 +4·0
15 -0·4 +0·1 +0·0 +0·6 +0·3 +0·8 +0·5 +0·6 +0·0 -0·3 -0·2
16 -0·4 -0·2 -0·5 +0·7 -0·1 +0·0 +0·5 +1·0 -0·1 -0·3 +0·0
17 +0·6 -0·4 +0·8 +1·0 +1·5 +1·9 +0·2 +0·3 +0·2 +0·4 +0·2
18 -0·4 -0·1 -0·3 +0·5 +0·3 +1·0 +0·6 +1·1 -0·1 -0·1 +0·5
19 -0·1 +0·1 -0·1 +0·4 +0·2 +1·0 +0·1 -0·2 +0·1 +0·0 -0·3
20 -0·9 -0·3 +0·0 +0·1 +0·2 +1·0 +0·6 +1·4 -0·6 -0·5 -0·3
21 +0·5 +0·9 +0·5 +0·7 +0·9 +1·2 +0·6 +0·8 +0·1 +0·4 -0·1
22 -0·7 -0·6 -0·5 -0·1 -0·5 +1·2 -0·2 -0·3 -0·3 -0·4 -0·5
23 +0·5 +0·4 +0·7 +0·7 +1·0 +0·8 +0·8 +1·2 +0·2 +0·2 +0·5
24 +0·1 +0·5 +0·3 +0·7 +1·0 +1·1 +0·8 +1·6 +0·3 +0·2 +0·4
25 +0·4 +0·5 +0·4 +0·0 +0·1 +1·4 +0·1 +0·1 -0·3 +0·0 -0·6
26 +0·5 +0·4 +0·5 +0·6 +0·5 +1·7 +0·9 +0·9 +0·1 +0·3 +0·2
27 -0·1 +0·5 +0·2 +0·1 +0·6 +1·5 +0·6 +0·6 -0·3 -0·1 -0·3
28 -0·2 +0·1 +0·3 +0·7 +0·2 +0·0 +0·5 +0·6 +0·5 +0·3 +0·4
29 +0·2 +0·4 +0·3 +0·3 +1·1 +1·4 +0·5 +0·7 -0·2 -0·2 +0·0
30 +0·5 +0·6 +0·7 +0·4 +0·6 +1·5 +1·1 +1·0 +0·3 +0·4 +0·3
31 +0·0 +0·1 +0·1 -0·1 +0·2 +0·4 +0·4 +0·4 -0·3 -0·5 -0·6
32 +0·4 +0·3 +0·4 +0·7 +0·8 +1·1 +0·9 +1·5 +0·3 +0·3 +0·4
33 +0·5 +0·5 +0·6 +0·0 +0·7 +1·2 +1·0 +0·8 -0·2 +0·1 +0·0
34 +0·2 -0·1 -0·4 +0·3 +0·2 +0·3 +0·4 +1·2 +0·0 -0·2 -0·1
35 -0·2 +0·3 +0·1 +0·5 +0·8 +1·1 +0·1 +0·8 -0·1 +0·4 +0·1
Total -6·3 +10·1 +13·7 +28·4 +23·2 +43·7 +30·2 +29·9 +4·6 -2·3 -8·1

which also lists the major alloy components: copper, products contaInIng cuprous oxide (CuzO) and basic
nickel, and zinc. Due to insufficient corrosion product on copper chlorides. The majority of the basic copper chlor-
some of the samples, grouping of corrosion product was ide was of the form of CuClz.3(Cu(OH)z), but there were
carried out for very similar alloys where it would be also other forms which included CuiOH)6Clz.3HzO,
expected that the corrosion product would be very similar. CU4(OH)6Ch.3HzO, and CU7CI4(OH)z.2HzO.These are all
Thus, the corrosion product of alloy 102 was combined very similar, differing only in their basicity and the amount
with that of alloy 101, and that of 926 with that of 905. of water of hydration which they contain. Consequently,
Alloys 101 and 102 are very similar high purity coppers, .all the basic copper chlorides have been grouped together
and alloys 905 and 926 are both nickel silvers, 905 having a as one constituent in the corrosion products and included
somewhat higher nickel content. under the heading of the major constituent. With some
The alloys in Table 6 have been listed in order of exceptions, within this group there is a correlation
decreasing copper content. Alloys 101, 144,203,523,548, between decreasing copper content in the alloy and
and 801 can be grouped together as having corrosion increasing amounts of basic copper chlorides and a
decreasing amount of cuprous oxide in the corrosion
Table 5 Maximum depth of pitting corrosion observed products.
following exposure to acidified salt spray and to Alloy 912 (75-25 cupronickel) produced a corrosion
neutral salt spray at 80°C, ,..m product containing 94°10 cuprous oxide and 6°1o basic
copper chlorides. This composition of corrosion product
Alloy Acidified salt spray Neutral salt spray could fit in the above sequence if nickel were considered to
act as a substitute for copper in these alloys corroding in
523 <3 0 acidified salt spray. This suggestion is supported by the
548 6 5 fact that in some applications a copper equivalent for
912 8 5 nickel with 1·2%Ni being equivalent to 1·0%Cu has been
926 11 0 proposed.2
905 15 0 Alloy 207 (70-30 brass) gave only 5°/0 cuprous oxide in
102 15 11
the corrosion product, the remainder being basic zinc
101 15 17
144 18 14 chloride formed as a result of dezincification.
203 33 21 The nickel silver (alloy 905) had a corrosion product
801 62 38 containing no cuprous oxide, 61°/0 basic copper chlorides,
207 64 60 and 39°/0 nickel chlorides. This make up of the corrosion
products is different from that observed on the related

Dr. Corros. J., 1990, Vol. 25, No.4

 Sterling et al. Accelerated atmospheric corrosion of copper 275

Table 6 Corrosion products produced during exposure to acidified salt spray

and their concentrations

Alloy compo-
sition, wt-% Corrosion product composition, wt-%

Alloy Cu Ni Zn CU20 CuC12.3(Cu(OHh) ZnC12·4(Zn(OH)2 NiC12

101 99·9 70 30
144 99·8 81 19
548 97 2·5 81 19
801 96 61 39
523 95 78 22
203 90 10 40 60
912 74·7 25 94 6
207 70 30 5 95
905 55 18 25 61 39

alloys 75-25 cupronickel (912) and 70-30 brass (207). sequence produced corrosion products containing decreas-
Thus the presence of both nickel and zinc in the alloy has ing concentrations of cuprous oxide, no cupric oxide, and
changed the corrosion process to produce nickel chloride increasing concentrations of basic copper chloride.
and basic copper chloride.
GM test
The GM test samples evinced very little corrosion. After
Neutral salt spray 35 cycles (each of 24 h duration) there was no sign of any
There are some interesting trends in the X-ray data from black oxide or patina; the only change was the loss of
the corrosion products (Table 7). The compounds and metallic lustre as the samples acquired a very thin cover-
their relative. concentrations were very similar on the three age of corrosion product. There was insufficient corrosion
high purity alloys: tough pitch copper (101), phosphorus product to carry out an X-ray identification of the
deoxidised copper (102), and Cu-Sn finstock alloy (144). compounds present.
The addition of extra tin up to 5% in phosphor bronze
(523) does not change the pattern. In all these alloys (101, Second series of tests
102, 144, and 523) the corrosion product was predomin- The weight loss results for specimens after 10 exposure
antly (83%) cuprous oxide, with small amounts of cupric cycles (each of 24 h duration) to neutral salt spray at 80°C
oxide (60/0) and basic copper chloride (9%) and traces of are given in Table 8. For each alloy, the weight loss due to
NaCI and CuClz. cleaning in acid for 5 min ~ wand the weight loss ~ W due
In alloys 548, 203, and 207, containing 2·5, 10, and to exposure to the salt spray are listed for both the acid
300/oZn respectively, increasing zinc content correlates cleaned and as received conditions. ~ W has also been
with corrosion products containing decreasing concentra- converted to an equivalent penetration based on the
tions of cuprous oxide, increasing concentrations of cupric assumption of uniform corrosion.
oxide, and a greater amount of chloride compounds. With the sole exception of alloy 905, ~ w is small
Coinage bronze (548) had in its corrosion products 14% of compared with d W, implying a small relative error in ~ W
the basic copper chloride, and amounts of NaCI and CuClz (see equation (1». For alloy 905 the weight loss due to
similar to those for alloy 101. The 90-10 brass (203) exposure of the acid cleaned samples to the neutral salt
showed slightly more basic copper chloride (16 %
and
), spray d Wacid is very low (0·021 g) and as a consequence
substantially more NaCI (9%) and CuClz than alloy 101. only slightly larger than ~w; in this case the fractional
Th~ corrosion product on 70-30 brass (207) contained no errors inherent in the measurement are likely to be large
cuprous oxide, but an increased amount of cupric oxide because the corrosion rate is low.
(16%
) and chlorides as basic zinc chloride; there was also a The results for alloys 101, 102, 144, 523, 548, and 801
large amount of ZnO. showed a greater weight loss for the as received condition
The alloying sequence 801, 905, and 912 can be com- than for the acid cleaned condition. There are two experi-
pareo with 101 in terms of decreasing total copper content mental considerations which, taken together, establish
in the alloy, the alloying elements being respectively that this difference in weight loss for the two conditions is
silicon and manganese, nickel and zinc, and nickel. This in fact statistically significant. First, ~w for the specimens

Table 7 Corrosion products produced during exposure to neutral salt spray at 80°C and their
concentrations

Corrosion product composition, wt-O/o

Alloying
Alloy addition, wt-% CU20 CuO CuCI2.3(Cu( 0 H)2) NaCI CuCI2 ZnO ZnCI2.4(Zn(OH)2)

101 0·030 83 6 9 1 1
102 0·03P 83 6 8 2 1
144 0·2Sn 82 7 9 1 1
523 5Sn 82 6 9 2 1
548 2·5Zn 78 7 14 1 1
203 10Zn 65 6 16 9 5
207 30Zn 16 73 11
801 3Si-1Mn 67 33
905 18Ni-25Zn 5 95
912 25Ni 100

Br. Corros. J., 1990, Vol. 25, NO.4

276 Sterling et at. Accelerated atmospheric corrosion of copper

Table 8 Weight loss* and penetration for as received and acid cleaned specimens exposed to
neutral salt spray at 80°C

Acid cleaned As received

Alloy ~w, g 6.W,g Penetrationt, Ilm 6.W,g Penetrationt, Ilm ~ Was received - 6. Wacid, g

101 0·016 0·611 4·6 0·690 5·2 0·079

102 0·012 0·436 3·3 0·597 4·5 0·161
144 0·015 0·625 4·7 0·780 5·8 0·155
523 0·010 0·631 4·7 0·779 5·8 0·148
548 0·011 0·778 5·8 0·865 6·4 0·087
801 0·012 0·546 4·1 0·563 4·2 0·017
203 0·014 0·657 4·9 0·646 4·8
207 0·037 0·940 7·0 0·876 6·6
912 0·002 0·272 2·0 0·147 1·1
905 0·010 0·021 0·16 0·043 0·32
926 0·004 0·025 0·19 0·029 0·22

* 6. w = weight loss during acid cleaning; ~ W = weight loss during test.

t Calculated assuming uniform corrosion.

in the as received condition was observed to be the same as those of pure copper (in particular 70-30 brass, which
that for the specimens already acid cleaned. This indicates exhibited dezincification) and there do exist alloys with
that no significant corrosion product was present on the corrosion rates lower than those of pure copper. One
surface of the as received specimens. Second, the differ- interesting development from the present work is the very
ences between the weight losses in the two conditions are, good performance of nickel silvers (alloys 905 and· 926),
in almost all cases1 much larger than those due to acid the corrosion rates for which were smaller by a very
cleaning. significant factor than those of high purity coppers.
These observations that the corrosion rate of some as Nickel silvers have also been observed by Herman and
received samples was greater than that of the correspond- Castillo,7 in a study lasting 4 years, to corrode in marine
ing acid cleaned samples support the suggestion of atmospheres at a rate substantially (by a factor of 7) lower
Moreton! that the corrosion rate can be increased by than that of coppers, but to undergo dezincification, with
residual contaminants from rolling, and indicate that this is an accompanying corrosion rate higher. than that of
an area where more systematic work is required. coppers, during exposure to industrial atmospheres. Holm
Alloys 101, 102, 144, 203, 523, 548, and 801 showed and Mattsson3 also observed that nickel silvers showed
essentially the same corrosion rate; the average weight loss corrosion rates somewhat lower than those of pure
for these alloys was 0·61 ± 0·10 g, which corresponds to a coppers during 16 years' exposure to .marine atmospheres
penetration of 4·6 (.tm, in the acid cleaned condition and and somewhat higher corrosion ~rates\vhenexpose9, ~o..an
o·70 ± 0·11 g, which corresponds to a penetration of industrial atmosphere; in this study no dezincification was
5·2 (.tm,for the as received condition. observed in either atmosphere. By contrast, in their 20
The 70-30 brass (207) showed a somewhat higher corro- year exposure tests, Scholes and Jacob8 found that nickel
sion rate due to dezincification. The corrosion rate of 75- silvers showed higher corrosion rates than coppers in both
25 cupronickel was less than half that shown by the alloy marine and industrial atmospheres and also observed:
group listed above, whereas the nickel silvers (905 and dezincification in both environments. .
926) have corrosion rates whi.ch are slower by a factor of The conclusions of this comparison between the present
30 and also lower by at least a factor of 13 than those of the study and the literature3,7,8 suggests that nickel silvers
high purity coppers. have corrosion rates lower than those of high purity
As noted above, alloys 101, 102, 144, 203, 523, 548, and coppers for 'short term' exposure (less than 16 years) to
801 have similar corrosion rates, which correlate, for all marine environments and that dezincification leads to
except 801, with similar corrosion products (Table 7), higher corrosion rates for nickel silvers exposed to indus-
composed of mainly cuprous oxide, some cupric oxide, trial environments, indicating scope for the development
and some basic copper chloride. However, the corrosion of an inhibited form of nickel silver. It should also be
product on alloy 801 is made up of cuprous oxide and basic noted that the work of Castillo and Popplewe1l9 has indi-
copper chloride only. cated that nickel silvers are susceptible to transgranular
Dezincification occurs" with 70-30 brass, producing stress corrosion cracking during exposure to marine
different corrosion products and a higher corrosion rate. atmospheres.
The alloys with lower corrosion rates, 912, 905, and 926, The observations that 70-30 brass exhibited a corrosion
have corrosion products containing very high concentra- rate higher than that of pure coppers is in disagreement
tions of basic copper chlorides, indicating that these may with the observations of Herman and Castillo.? In their
provide better protection, at least in the present short term tests the corrosion rafeof 7(f-30 brass was lower than
tests. The lowest corrosion rates were observed for the those of pure coppers during exposure to marine atmos-
nickel silvers, which had corrosion products containing pheres for 4 years, although dezincification was observed,
95% basic copper chloride and 50/0 cuprous oxide - a in agreement with the present study. In contrast, Holm
mixture that proved to be particularly protective in these and Mattsson3 observed that 70-30 brass showed a corro-
tests. sion rate higher than those of pure coppers during 16 year
exposure tests in a marine atmosphere; dezincification was
RELATION TO ATMOSPHERIC CORROSION also reported. Scholes and Jacob8 have observed dezincifi-
The corrosion rate data, specifically Tables 5 and 8, reveal cation in 70-30 brass and corrosion rates comparable to
a picture of copper corrosion consistent with the those of high purity. coppers during 20 year exposure tests
literature2-6 related to atmospheric corrosion; in general, in marine and industrial environments.
alloying does not decrease the corrosion rate of copper Thus, on balance, the present observations of dezincifi-
alloys; there are alloys with corrosion rates higher than cation in 70-30 brass, together with corrosion rates higher

Br. Corros. J., 1990, Vol. 25, No.4

 Sterling et al. Accelerated atmospheric corrosion of copper 277

than those of high purity coppers, would be expected corrosion and accelerated tests is an important practical
during long term atmospheric exposure on the basis of the one. The tests in the present study use a variety of means
literature evidence. For such applications therefore an to provide an acceleration of the corrosion mechanism
inhibited alloy is to be preferred. compared with atmospheric corrosion. All three tests
Similarities also exist between the products of atmos- included NaCI as the atmospheric contaminant, as would
pheric corrosion and the sequence in which corrosion be expected for atmospheric corrosion in humid marine
products formed and covered the surfaces of the speci- regions. Both the acidified salt spray test and the neutral
mens in the present accelerated atmospheric corrosion salt spray test contained NaCI concentrations much higher
tests. For alloys 101, 102, 144, 203, 523, 548, and 801 the than would be expected in atmospheric corrosion. The
following sequence was observed in the neutral salt spray D
neutral salt spray test had a temperature of 80 C, which is
tests: also greater than that during ambient atmospheric expo-
(i) the specimen loses its metallic lustre as the first sure. However, when operating radiators are considered,
corrosion products cover the surface 80 C is not a particularly elevated temperature, and the
D

(ii) an increasing fraction of the surface becomes radiator operating temperature cycle is the chief feature of
reddish brown the GM test. The results of the present study imply a
(iii) areas of the surface become black relationship between atmospheric corrosion and the
(iv) areas become covered with a green corrosion neutral salt spray test. This relationship is not so clear for
product patina. the acidified salt spray test. Moreover, the low pH values
The observations in the acidified salt spray tests are consis- used in this test probably occur seldom in atmospheric
tent with a similar sequence. Although the above sequence corrosion, leading to greater doubts about the validity of
was followed, differences in detail were found between this test with respect to atmospheric corrosion. The GM
individual alloys; for example, the time of observation of test proved to be very mild, so that no comment can be
the first black colour or patina. In particular, for alloys made concerning its relationship with atmospheric
101, 102, 144, and 523 the relative area covered with corrosion.
reddish brown or black corrosion product was significantly The literature reveals that the corrosion rate of coppers
different at the end of 10 cycles of the neutral salt spray exposed to humid marine environments is both time and
test. Since the analysis of the corrosion products for these locality dependent. Typical reported values from long
alloys under this condition showed a constant compo- term tests are 0·6-1·2 (Ref. 8),1·1-2·1 (Ref. 10), 2·2-3'2
sition, it must be concluded that both the reddish brown (Ref. 9), 2·1-2·6 (Ref. 7), and 0·5-0·9 (Ref. 2) I-tm/year.
and the black areas correspond to cuprous oxide. This is These values give an average corrosion rate of 1·7 Ilm/
substantiated by the fact that no reddish brown or black year, which is to be compared with an average penetration
areas were observed on alloy 207 and no cuprous oxide rate of 0·44 I-tm/(24 h cycle) measured for exposure of
was detected by the X-ray analysis. The different appear- alloys 101 and 102 to the neutral salt spray at 80 C. ThisD

ances and colours of the same compound may have various provides a calibration relating four 24 h salt spray cycles to
causes: differences in optical properties due to differences 1 year's atmospheric corrosion. This calibration must be
in oxide thickness is the most obvious. used with caution, as it is only expected to provide an
The sequence of corrosion products observed in the indication of the relative severity of the neutral salt spray
present tests corresponds to the sequence on coppers test and should not be used as a numerical predictor of
exposed to the atmosphere. 3--5,7.8.10 In atmospheric corro- lifetime for specific components. Moreover, the corrosion
sion of essentially pure copper, cuprous oxide first forms. rates taken from the literature exhibit a range of values
This is subsequently converted to a green patina, the differing by over a factor of five, and consequently a
composition of which depends on the composition of the similar factor must be inherent in the relationship between
atmosphere; in marine environments the patina consists of the accelerated test results and atmospheric corrosion.
basic copper chlorides.3--5,10The process. of atmospheric On the basis of this comparison, the maximum pit
patina formation on copper takes a considerable time, up depths recorded are not unreasonable compared with the
to 70 years, and during this time the corrosion rate values observed by Scholes and Jacob8 during 20 year
decreases as the patina becomes more protective. The exposure tests in a marine environment, when it is taken
present observations reveal that the same pattern of corro- into account that pit growth rate decreases with time.
sion product formation occurs on alloys 101, 102, 144,203,
523, 548, and 801, except that patina formation begins in CONCLUSIONS
the present tests much earlier than during atmospheric 1. Of the three atmospheric corrosion tests applied to
corrosion. Two deductions can be made: first, the neutral copper and some copper alloys, the GM test involving
salt spray test is useful in giving accelerated indications neutral salt spray with thermal cycling was less aggressive
concerning atmospheric corrosion; second, since alloys than the acetic acid salt spray (ASTM B289-1980) or
101, 102, 144, 203, 523, 548, and 801 had essentially the neutral salt spray at 80 C (ASTM Bl17"'::1979)tests.
D

same corrosion rate in the neutral salt spray, it is likely 2. The sequence of corrosion product formation on
that these alloys will also have similar atmospheric corro- specimen surfaces during exposure to the neutral salt spray
sion rates in humid marine environments. This indicates test was very similar to that on specimens exposed to
that there is no significant beneficial effect with respect to natural atmospheres, thus indicating that this test can be
atmospheric corrosion in alloying copper with phosphorus, used to give accelerated indications of corrosion to be
tin, zinc, or silicon. expected in humid marine regions. It is suggested that four
It is interesting to note that the alloys with the lowest 24 h salt spray cycles are approximately equivalent to 1
corrosion rates, 905, 912, and 926, have corrosion prod- year's exposure to such an atmosphere.
ucts containing very high concentrations of basic copper 3. The corrosion products formed under all three test
chlorides, similar to the naturally occurring patina on conditions were only partially protective~ but basic copper
coppers. The corrosion products on these would thus be chlorides were more protective than copper oxides.
expected to offer reasonable protection against further 4. Tough pitch copper, phosphorus deoxidised copper,
corrosion, as was in fact observed to be the case. This Cu-Sn alloy (as used in radiator fins), phosphor bronze,
implies that patina formation on these alloys can be rapid coinage brass, 90-10 brass, and silicon bronze all showed
compared with the times required for higher purity essentially the same corrosion rates in the neutral salt
coppers. spray test, indicating that these alloys will have similar
The question of the relationship between atmospheric atmospheric corrosion rates and that no improvement in

Br. Corros. J., 1990, Vol. 25, NO.4

278 Sterling et al. Accelerated atmospheric corrosion of copper

atmospheric corrosion resistance is achieved by alloying testing were supplied by Austral Bronze Crane Copper
with phosphorus, tin, zinc, or silicon. Ltd.
~5. The 70-30 brass exhibited dezincification in both the
acidified salt spray and neutral salt spray tests, giving REFERENCES
corrosion products that were quite different from those on 1. B. MORETON: International Copper Research Association
the other alloys. Inc., Potters Bar, personal communication, 1987.
6. The nickel silvers (12 and 18%Ni) exhibited very low 2. E. G. WEST: 'Copper and its alloys'; 1982, Chichester, Ellis
corrosion rates - compared with the other alloys - and this Horwood.
behaviour is proposed to be associated with the rapid 3. R. HOLM and E. MATTSSON: in 'Atmospheric corrosion in

formation of a patina containing a high proportion of basic metals', STP767, 83; 1982, Philadelphia, PA, ASTM.
4. w. H. J. VERNON: J. Inst. Met., 1932,49, 153.
copper chloride. 5. w. H. J. VERNON and L. WHITBY: J. Inst. Met., 1929,42, 181;
7. Six of the 11 coppers and copper alloys had a slightly, 1930,44, 389.
but statistically significantly, higher corrosion rate for 6. L. L. SHREIR (ed.): 'Corrosion', 2nd edn, Vol. 1,1:150-1:167;
specimens tested in the as received condition, this effect 1976, London, Butterworth.
being attributed to residual contaminants from rolling 7. R. S. HERMAN and A. P. CASTILLO: in 'Corrosion in natural
treatments. environments', STP558, 82; 1974, Philadelphia, PA, ASTM.
8. I. R. SCHOLES and w. R. JACOB: Proc. Conf. on 'Copper and its
alloys', 330; 1970, London, The Institute of Metals.
ACKNOWLEDGMENTS 9. A. P. CASTILLO and J. M. POPPLEWELL: in 'Atmospheric corro-
This work was supported by the International Copper sion in metals', STP767, 60; 1982, Philadelphia, PA, ASTM.
Research Association Inc., as part of project 391, 'Exter- 10. L. P. COSTAS: in 'Atmospheric corrosion in metals', STP767,
nal corrosion of copper/brass radiators'. Materials for 106; 1982, Philadelphia, PA, ASTM.

Advances in Physical Metallurgy

A collection of invited papers to mark the 70th birthday year of Professor
Sir Alan Cottrell
Edited by JA Charles and G C Smith

The papers presented are both nostalgic in their surveys of the contributions by Sir Alan in numerous
fields, and progressive in indicating how it had been possible to develop from the sure foundations that
he had laid.
Representing a broad survey of various areas of physical metallurgy, eleven papers, written by experts
in their respective fields, cover the following topics:

• Dislocations and Cracks • ' The Metallic State - Revisited '

• Theory of the Yield Point and Strain Aging in • Microstructural Characterisation of Defects and
Steel Mechanisms
• Ductile and Brittle Failure in Steel/Pressure (including field-ion observation)
Vessel Assessments • Creep and Creep Fracture
• Dislocation Reactions • Fibre Reinforcement at the Metallurgy
• The Nature of Grain Boundaries in Ordered Alloys Laboratory at Cambridge 1958-67
• Radiation Damage and Point Defects • Strengths of Grain Boundaries in Pure Metals
ISBN 0 901462 85 3 240x160mm 218pp (H) October 1990 £25.00 US$50.00

Orders with remittance to: The Institute of Metals, Sales & Marketing Dept., 1 Carlton House
Terrace, London SW1Y 5DB. Tel. 071-976 1338 Fax. 071-839 2078 Telex 8814813
Orders originating in Canada and the United States should be sent direct to: The Institute of
Metals North American Publications Center, Old Post Road, Brookfield, VT 05036, USA.
Tel. (802) 276 3162 Fax. (802) 276 3837
(Members of The Institute of Metals receive a 20% discount).

Hr. Corros. J., 1990, Vol. 25, No.4