Corrosion fatigue behaviour

of brass in aqueous solutions

V. K. G o u d a , A. A. R a m a d a n a n d G. I. Y o u s s e f

The corrosion fatigue behaviour of (x-brass was investigated in NaCI, NaNO 2, (NH4)2S04 and
(N H4)2S04 + CuS04 solutions by applying the reverse bending technique at 60 cycles/min.
With the exception of NaCI, all electrolytes reduced the fatigue life compared with that in air
or water. Anodic polarization caused further reduction in the fatigue life in all solutions. In
NaCI the failure was mainly due to electrochemical dissolution of the brass, while in ammoniacal
solution both thinning and corrosion fatigue occurred. In NaNO 2 solution corrosion was
inhibited but severe corrosion fatigue occurred. The fracture mode changed from transgranular
in air to mixed mode, with different percentages of the intergranular type depending on the
electrolyte.

Key words: corrosion fatigue; fatigue life; electrolytes; electrolytic corrosion; fractography;
(x-brass

Corrosion fatigue (CF) involves the combined interaction
of the corrosion environment and the applied cyclic stresses.
Despite the extensive published work on corrosion fatigue,
very little work has been reported on Cu-base alloys.
Birley and Tromans 1 studied the corrosion fatigue of
copper and a-brass in (NH4)2SO4 + CuSO4 and NaC1 solu-
tions and found that there was no specific dependence of
fatigue strength or crack path on the environment. In
contrast, Misawa2 reported that the fatigue life and cracking
path of (x-brass was dependent on the pH of an ammoniacal
solution.
Hahn and Duquette 3 reported that the fracture mode
of Cu-Ni alloy in air was transgranular and shifted either to
mixed transgranular and intergranular or to intergranular in
aqueous solutions according to the pH of the solution.
Hahn 4 found that, for a Cu-AI alloy, the relative amount
of intergranular cracking increased with increasing corrosion.
The aim of the present study was to determine the
effects of aqueous salt solutions of varying corrosivity and
of anodic polarization on the fatigue behaviour of (x-brass.
The choice of salt solutions was based on results obtained
from a parallel investigation on the stress corrosion cracking
(SCC) of the same alloy in different electrolytes. 5-8
It has been reported that increasing the corrosion
rate of the alloy by changing the environment or by anodic
dissolution tends to decrease the fatigue life.9-12 It has
also been mentioned that a critical corrosion rate 13-15
must be exceeded for corrosion fatigue to be possible.
However, the results of the present investigation show that
this correlation does not exist for (x-brass.
Experimental
The test specimens were machined from commercial-grade
annealed 72/28 brass sheet of 2.7 mm thickness and of
chemical composition (in weight %) 71.7 Cu, 28.02 Zn,
0.006 Pb and 0.01 Fe. The mechanical properties of the
material were: ultimate tensile strength (UTS) 283.44 MN/m 2,
yield strength (YS) 216.52 MN/m 2, % elongation 80% and
hardness (VHN) 600.36 MN/m 2. All specimens were ground
with grade 120 to grade 1000 silicon carbide papers and
were washed with water and then degreased with acetone.
The fatigue testing devices were constant deflection
machines. Two specimens can be tested simultaneously at
cyclic frequencies o f / 0 - 1 0 0 cycles/rain and at stresses
corresponding to specimen deflections of 0 - 2 5 . 4 ram. The
specimens were mounted in the test cells and loaded for
the desired bending stress at the minimum width, calculated
from the following equation: 16
6PL
S=~ kg/mm 2
where P = applied load in kg, S -- maximum bending stress
in kg/mm 2 for the minimum width, b = minimum width,
t = average thickness, and L = the distance in mm from the
loading point to the minimum width. The applied load ,P,
for a certain specimen deflection was measured using a
proving ring consisting of a steel ring fitted with a cali-
brated dial gauge, which was linear with the applied load.
A correction for the stress value was carried out when the
fracture did not occur at the minimum width. All tests
were operated at 60 cycles/rain and the number of cycles
to failure was registered on a mechanical counter which
tripped on failure.
The shape and dimensions of the fatigue test speci-
mens and the cell used are given elsewhere, le Tests were
carried out at room temperature in naturally-aerated
solutions. The solutions were prepared from reagent-
grade chemicals and double-distilled water. The strengths
of the solutions used were as follows: 1 M NaCI;
1 M (NH4)2SO4; 1 M (NH4)2SO4 + 0.05 M CuSO4; and
1 M NaNO2.
The fatigue tests were carried out either under the
free corrosion condition or at a constant applied anodic
potential of 300 mVH, which was controlled using a
'Wenking' potentiostat. The anodic overpotent/al during
the fatigue tests were 225 mV, 120 mV and 280 mV for
NaNO2, (NH4)2SO4 + CuSO4 and NaC1, respectively.
The fracture surface was studied directly by scanning
electron microscopy (SEM), while the exposed surface
was studied by optical microscopy. For fractographic
examinations, small pieces were cut about 10 mm from
the fracture surface of the upper part of the specimens,
which were automatically removed from the solution after

0142-1123/83/040207--09 $3.00 © 1983 Butterworth & Co (Publishers) Ltd

Int J Fatigue Vol 5 No 4 October 1983 207
failure. The specimens were cleaned in an ultrasonic cleaning
250
bath using acetone. F o r optical metallography, the cut
pieces were mounted in epoxy, polished with emery papers
down to No 600 and finally polished with fine alumina. 210
The specimens were etched with FeC13 solution.
The coupons used for corrosion rate measurements 190
under open circuit conditions had the dimensions 20 ×
50 x 3 mm, while the dimensions of the specimens used in 'E 170
z
the polarization experiments were 15 x 15 x 3 mm. The
surface preparation was the same as that applied for CF 150
specimens.
150
x N xx~
Results
I10 [] (NH4)2SO4 ~Cu$O4 %
S/N curves 0 NaNO 2

Under free corrosion condition 90-

To characterize the response of brass to cyclic bending 7N I J I I I I i~J I I I I I h J i

stresses in the different chemical environments the stress 04 105 I06

amplitude, S, versus number of cycles to failure, N, curves Number of cycles

were determined. Fig. 1 shows the S/N curves for the F i g 2 Stress v$ number o f cycles to failure at 300 m V H
specimens fatigued in the four salt solutions under free
corrosion conditions. The S/N curves of specimens fatigued
Based on the general behaviour shown in Fig. 1, the
in laboratory air and distilled water are also included in
S/N curves obtained can be divided into two groups. The
Fig. 1 for comparison. All tests were conducted at stress
first consists of the curves obtained in laboratory air, water
values in the range 155 MN/m 2 to 215 MN/m 2 ( 7 1 - 1 0 0 %
and NaC1 solution. At relatively high stresses, the fatigue
VS). resistance in water is better than that obtained in laboratory
air or in NaC1 solution; however, at low stresses the fatigue
life is greater in NaCI than in air or water. The second group
230 of S/N curves shows a considerable reduction in fatigue
life compared with those of the first group. The curves are
210 X ÷ almost parallel to each other, with the NaNO2 and
(NH4)2SO4 + CuSO4 curves very close to each other and
190- lower than the (NH4)2SO4 curve. The percentage reduction
in fatigue life is calculated according to the equation:
E I'/'0 ~- [(Nfa - N~s)/N~a] × 100
Z
O Air where Nfa and Nfs are the fatigue life in air and solution,
150~- [] Water
respectively. Table 1 gives the values of the reduction in
+ NoCI
fatigue life and a description of the surface appearance of
X (NH4)2 SO4
the specimens in all environments tested.
• NaNO z
IlO - Under applied anodic potential
• (NH4) 2S04÷CuSO 4

90- The S/N curves obtained for brass when anodically polarized
at 300 mVH in NaC1, NaNO2 and (NH4)2SO 4 + CuSO4 are
, I ,lJlJl I I i I I I I i shown in Fig. 2. The experiments were carried out at stress
7004 i05 106 values in the range 90 MN/m 2 to 205 MN/m 2 ( 4 1 - 9 4 % YS).
Number of cycles It is apparent from Fig. 2 that the curve obtained in NaC1
Fig. 1 Stress vs number of cycles to failure under free corrosion lies above the other two curves and is widely separated
conditions from them, especially at low stress values. Thus specimens

Table 1. Reduction in life at 190 M N / m 2 and surface appearance of brass during fatigue tests in different environments

Free corrosion conditions Applied anodic potential

Reduction Reduction
in life in life
Environment (%) Exposed surface (%) Exposed surface

Air O0
Water --8
NaCI --45 Homogeneous grey/brownish film 25 Rough, metallic lustre, reddish surface
NaNO 2 80 Homogeneous black film 90 Homogeneous dark grey colour
(NH4)2SO 4 60 Thin dark grey film
(NH4)2SO 4 + CuS04 80 Partly covered by blackish film 82 Non-continuous greyish film

208 Int J Fatigue October 1983

fatigued in NaC1 show a higher fatigue resistance than
230
those fatigued in NaNO 2 or (NH4)2SO4 + CuSO4 at all
stress levels. The fatigue resistance of brass in NaNO 2 is
lower than that in ammoniacal solution at high applied
stresses. This situation is reversed at lower applied stress
levels. 190
A comparison of the fatigue behaviour under free
corrosion conditions and under anodic polarization for ~7 170
each electrolyte is made in Figs 3 - 5 . It can be seen from
Fig. 3 that the curves are widely separated for NaC1 solu- ~ 150
tion, but they run almost parallel to each other at all
cyclic stresses. In NaNO2, the curves are almost parallel 130
and close to each other (Fig. 4). However, with
(NH4)2SO4 + CuSO4 solution, the curves are close to IlO
each other at high stresses but diverge rapidly as the
applied stress decreases (Fig. 5). 90- o Free corrosio
Thus, it is clear that in all solutions the fatigue life • A~ 500 mV H
generally decreases when the alloy is anodically polarized. 70104
I I I I I I I II
i0-~
I I I I I I II
I0 e
A remarkable reduction in thickness was noticed for
Number of cycles
specimens fatigued in NaCI and (NH4)2SO4 + CuSO4
solutions when anodically polarized. Fig. 5 Stress vs n u m b e r o f cycles t o f a i l u r e in ( N H 4 ) 2 S O 4 + C u S O 4

Table 2. Corrosion rate and current density at 300 mY H in

tested salt solutions
230
Steady state
current density
210
Corrosion rate at 300 mV H
Solution (mg/dm2/day) (mA/cm 2)
190
NaNO 2 0.0007 0.003
E 170 NaCI 0.0072 9
Z
(N H4) 2 SO4 0.0485 -
150 (NH4)2SO 4 + CuSO4 0.0598 12

130

Determination of the corrosion rate

IlO
O Free corrosion condit"ion The corrosion rate under open circuit conditions was
90
• At 3 0 0 mV H determined for unstressed brass coupons when immersed
in all the electrolytes used for the fatigue tests. The loss
70 I I I I I I Ill I l I I I I I in weight was recorded as a function of time. The experi-
IO4 IO s I06 ment lasted 21 days, and the results are summarized in
Number of cycles Table 2. It is clear that the order of aggressiveness of
Fig. 3 Stress v$ n u m b e r o f cycles t o f a i l u r e in NaCI these solutions is as follows: (NH4)2SO4 + CuSO4 >
(NI-I4)2SO4 > NaCl > NaNO2.
230
Anodic polarization experiments
210 Potentiostatically controlled anodic polarization experi-
ments at 300 mVH were conducted for unstressed speci-
190 mens in NaC1, NaNO2 arid (NH4)2SO4 + CuSO4 solutions.
The results are shown in Fig. 6. It is clear from this figure
'E 170 that the behaviour is almost the same in (NH4)2SO4 +
Z
CuSO4 and in NaC1 solutions: there is an initial decrease
150 - in current density during the first 20 min followed by
a slight increase. The steady state was obtained after
130 about 160 min at 8.6 mA/cm 2 and 12 mA/cm 2 for NaC1
and ammoniacal solutions, respectively. On the other
I10 hand, in NaNO 2 solution, the current density decreased
o Free corrosion condition slowly during the initial 60 rain, followed by a sharp
• A t 3 0 0 mV H decrease; the steady state was obtained after about
90-
200 min at 2.8 × 10-3mA/cm 2.
I i i i ll,I I I I i i i ii
7004 iO5 I0 e MetaUographic examination
Number o f cycles The mode of fatigue cracking of brass was examined by
Fig. 4 Stress vs n u m b e r o f cycles t o f a i l u r e in N a N O 2 micrography and fractography. The latter technique alone

Int J Fatigue October 1983 209

 conditions. This changed to entirely transgranular cracking
when the specimen was anodicaUy polarized. Examination

,°'l
of fatigued specimens in NaCl at high applied stresses
indicated the presence of secondary cracks parallel to the
X ×-- main crack spread over a distance of more than 10 mm.
IOI O ,,, O Striations were observed under free corrosion conditions
(Fig. 9h), as well as when anodically polarized (Fig. 9c).
Fracture in NaNO2 solution

Specimens fatigued in NaNO2 solution showed a mixed

A

~E IO°- mode of cracking, but the intergranular mode dominated

o
(Fig. 10). However, the percentage of intergranular cracking
decreased as the cyclic stress increased and when anodic
potential was applied. The specimens showed many secon-
dary cracks parallel to the main crack at a distance of
iO-I 2 - 3 mm from the main fracture surface, whether fatigued

IO-Z-- 0 NaCI
a NaNO z

i0o3 I I
0 I00 200
Time (rain)
Fig. 6 Current/time curves at constant applied potential of 300 mV H

does not clearly indicate the type of fatigue crack; exami-

nation with both techniques gives more accurate pictures
of the characteristics of the fatigue fracture. The fracture
surface was divided into nine zones to identify the charac-
teristic features of crack initiation near the exposed surface
and in the middle of the specimens.
Fracture in air

The mode of fracture was transgranular in nature for speci-

mens in air at low and high applied cyclic stresses, as shown
in Fig. 7a. Fatigue striations were defined and occupied a
very large area (Fig. 7h). Secondary cracks originating from
the main crack were observed at the fracture surface.
Fracture in water

Specimens fatigued in water showed a mixed mode of

fracture at relatively low stresses, as illustrated in Fig. 8a.
(Intergranular cracking can be seen in the upper right-hand
corner of Fig. 8a; transgranular cracking occurs in the
lower left-hand comer.) At relatively high cyclic stresses,
a mixed but predominantly transgranular mode of fracture
was observed (Fig. 8c). Striations were also seen (Fig. 8b),
and secondary cracks branched from the main one either
at the exposed surface (Fig. 8c) or at the fracture surface
itself. However, at low stresses, many microcracks at the
exposed surface parallel to the main crack were observed
within a distance of 2 - 3 mm from the main fracture
surface. The microcracks were especially observed through
or near the grains that contain slip bands, as shown in
Fig. 8c.
Fracture in NaCl solution
Fig. 7 Fatiguefrracture in air (198 MN/m2): (a) the transgranular
In NaC1 solution, the mode of cracking was mixed but mode (b) well defined striations and (c) an example of mixed-mode
predominantly transgranular (Fig. 9) under free corrosion crack initiation

210 I n t J Fatigue O c t o b e r 1983

 covered a great area of the fracture surface, especially under
applied anodic potential, as shown in Fig. 12b. Table 3
summarizes the mode of failure and the type of secondary
cracks observed in all environments tested.

Discussion
It is commonly accepted that C F is strongly influenced by
the aggresivenessof the environments, ie the more
aggressive the environment, the greater it damages the
fatigue properties of a metal. However, the resultsobtained
in the present study show that this generalization is not
justifiedfor 0~-brass.The objective of this discussion is to
assess the influence of different electrolytesof varying
corrosivity on the fatigue behaviour of brass under free
corrosion conditions and under applied anodic potential.

Fig. 8 Fatigue fracture in water: (a} mixed mode (169 MN/m 2)

(b) striations (169 MN/m 2) and (c) mixed mode with predomi-
nantly transgranular type and secondary cracks branched from
the main one (187 MN/m 2)

under free corrosion conditions or under applied anodic

potential, and under both low and high applied cyclic
stresses.This is similar to the behaviour observed for
water and NaCl.
Fracture in (NH4)2S04 + CuSO4 solution
Under open circuitconditions, a mixed mode of fracture
was observed with the intergranularpercentage increasing
as the applied stressdecreased (see Fig. 1 I). The same
mode was also observed when anodic potential was applied
under high stress.As the cyclic stressdecreased, the fracture
mode became entirelyintergranular (see Fig. 12). Striations
were also observed, but secondary cracks parallelto the
Fig. 9 Fractographs of fatigue fracture in NaCh (a) & (b) a mixed
main one, which were present with NaCl and NaNO2, were mode with predominantly transcrystalline type and striations
not noticed in ammoniacal solutions. Corrosion products (194 MN/m 2) (c) striations under applied potential conditions

Int J Fatigue October 1983 211

 tance in these solutions is found to be: N a N O 2 > NaCI >
(NIJ,4)2S04 > (NH4)2SO 4 + CuSO4, and thus the C F
resistance is not proportional to the corrosion resistance
of the electrolyte. It m a y be concluded that the degree of
the aggressiveness is not always a measure of C F severity.
The increase of fatigue life in an aggressive solution (such
as ammoniacal solution) compared with that in an inhibitive
solution (such as NAN02) may be attributed to enhanced
dissolution of surface microcracks in the former as soon as
they are formed, which m a y delay crack initiation.
Fatigue under applied anodic potential
The response of c~-brassanodically polarized at 300 m V H
and fatigued in the different electrolytes helps explain
the mechanism of CF. The steady state anodic current
density (anodic dissolution) is almost the same for NaCl
and (NH4)2SO 4 + CuS04 solutions and is about 300
times that obtained in N A N 0 2 solution, ie the anodic
dissolution of brass in N A N 0 2 is inhibited. However, the
fatigue lifein (NH4)2SO 4 + CuS04 solution is 3.3 times
and 0.9 times that obtained in NaCI and N A N 0 2 solutions,
respectively. If the C F of brass depended primarily on its
dissolution rate (as many investigators believe), one
would expect that the fatigue lifein NaCl and (NH4)2SO 4 +
CuS04 solutions would be at least of the same order of
magnitude, and much greater than that in N A N 0 2 solution.
Thus, the fatigue behaviour of brass when anodicaUy
polarized suggests that corrosion resistance is not the primary
factor in determining the severity of CF. It has been con-
cluded by some investigators9-L2 that the corrosivity of
the solution has most influence on the fatigue strength of
Cu-base alloys, and that a criticalcorrosion rate must be

Fig. 10 Fatigue fracture in NaNO2: (a) mixed mode with pre-

dominantly transgranular type (open circuit, 154 MN/m2), (b) mixed
mode with predominantly transgranular type (open circuit,
176 MN/m2), and (c) mixed mode with predominantly inter-
crystalline type (applied potential, 172 MN/m 2)

CF behaviour under free corrosion conditions

Exposure of ~-brass to water or to NaCI solution during
testing does not reduce the fatigue life compared with that
obtained in laboratory air; indeed, the fatigue life increases.
A similar increase in fatigue life has been also reported by
some investigators for Cu-base alloys when exposed to
either fresh or salt water, t° These results indicate that the
electrolytes can be arranged in the following order of
Fig. 11 Fracture in (NH4)2SO4+ C u S O 4 (open circuit condition,
fatigue resistance: NaC1 > (NH4)2SO4 > (NH4)2S04 + 172 MN/m2): (a) micrograph, and (b) fractograph showing a mixed
CuSO4 ~ NAN02. However, the order of corrosion rests- mode with predominantly intercrystalline cracking

212 Int J Fatigue October 1983

Table 3. Mode of fracture at different applied stresses in the tested environments

Free corrosion conditions Applied anodic potential

Stress Type* Stress Type*

Mode of of Mode of of
Environment MN/m 2 % YSt fracture** crack MN/m 2 % YSt fractu re** crack

Air 198 90 T II
Water 169 77 M I, II & I II
187 85 MPT I & II
NaCI 161 73 MPT
III
194 88 MPT 202 92 T III
NaNO 2 154 70 MPT 141 64 MPI
III III
176 80 M PT 180 82 M
(NH4)2SO 4 183 83 MPI
I, II
202 92 M
(NH4)2SO 4 163 74 MPI 95 43 I
+ I&lt I, II
CuSO4 189 86 MPI 150 68 MPI

* See definitions in the t e x t , for type o f secondary crack

* * T ---- transgranular; I = intergranular; M = mixed (T + I); and P = predominantly
t YS = yield strength

exceeded before CF takes place, z3,14 These conclusions were

Fig. 12 Fractographs in (N H4)2SO" + CuSO 4 (applied potential,
95 MN/m2): (a) near the centre of the specimen showing inter-
crystalline fracture (b) near the edge of the specimen showing
intercrystalline fracture covered by corrosion products
drawn from the data obtained in one electrolyte, NaC1.
However, our results suggest that NaC1 alone can not
explain the corrosion fatigue behaviour of brass in salt
solutions. Further, it can give misleading conclusions if
considered alone. The present study indicates that the
reduction in the fatigue life of brass in NaC1 when anodi-
cally polarized is essentially due to electrochemical dissolu-
tion of brass which causes severe thinning and consequently
mechanical failure. Such failure is better described as
'failure due to electrochemical dissolution', terminology
similar to that used in the field of SCC. Thinning is also
observed in ammoniacal solution, and is accompanied by
an enormous reduction in fatigue life compared with that
obtained in NaC1 solution; the failure is essentially the
result of the conjoint action of corrosion and mechanical
factors. By contrast, in NaNO2 solution, no thinning takes
place when brass is anodically polarized or under open
circuit conditions, yet considerable reduction in fatigue
life and consequently CF takes place. The occurrence of
severe CF in inhibitive solutions has not yet been reported
in the literature. It is of interest to note that the overall
CF behaviour under open circuit conditions in NaNO2
and ammoniacal solutions is almost the same, while at
300 mVH it is widely different. This behaviour may be
attributed to different mechanisms of CF operating in
these solutions. In NaNO2 solution, where an inhibitive
film is formed on the brass surface, the behaviour is
explained by rupture of the film, which may be followed
by adsorption of nitrites on to the bare metallic parts, z7
On the other hand in ammoniacal solution, where no film
formation takes place, a mechanochemical mechanism is
most likely operating. Accordingly, electrochemical
attack takes place at plastically-deformed areas of metal
and the non-deformed metal acts as cathode.

Int J Fatigue October 1983 213

 In non-corrosive solutions, whether enhancing CF
(NaNO2) or not (water and NaC1), the initiation stage of
cracking is similar, as apparent from the formation of many
microcracks near the fracture surface. However, the NaNO2
produces a propagation period which is not only faster
than the other non-corrosive environments, but also faster
than the corrosive environment. This behaviour supports
the idea that the mechanism of cracking in NaNO2 is
different from that in the other solutions, and may be
correlated with recent results on the SCC of the same alloy
in different salt solutions. 5-8 Using the slow rate technique,
the most severe SCC was obtained in nitrite solution com-
pared with sulphate, chlorate and chloride. These results
indicate that nitrite has a specific influence in initiating
and propagating SCC and CF, suggesting that there is an
interaction between SCC and CF in this solution.
Fracture characteristics
The S/N curves show only the response of brass to cyclic
stress but do not determine whether a change in fatigue life
is associated with a change in the fracture mode and type of
cracks or not. To clarify the fracture characteristics, the
different types of secondary cracks observed can be classified.
Secondary transgranular cracks are observed at slip
lines as illust,'ated in Fig. 13a. The intergranular cracks
occur at triple points due to grain boundary shear as shown
in Fig. 13b. Whatever the mode of failure, these secondary
cracks can be classified into three types. In types I and II,
the cracksare branched from the main one but are observed
on the exposed surface (Fig. 14a) and on the fracture
surface (Fig. 14c), respectively. Type III appears on the
exposed surface parallel to the main crack and within a

Fig. 14 SEM fractographs showing types of secondary microcracks

distance not more than 15 mm from the fracture surface,

Fig. 13 Fractographs showing examples of the beginning of micro-
cracks in e-brass: (a) transgranular at slip bands (b) intercrystalline
at triple point due to shear of grain boundary
as illustrated in Fig. 14b. Formation of many secondary
cracks parallel to the main one (type III) in some salt solu-
tions (NaC1 and NaNO2), especially under applied anodic
polarity, indicates that the mechanism of crack formation
and propagation has changed from a single crack, as in air,
to numerous parallel cracks. Subsequent crack growth
continues by joining these microcracks together. Although
there were many microcracks of type III in the specimen
fatigued in NaC1 solution under open circuit conditions, the
fatigue life was not reduced. This may be attributed to
bluntness of the secondary cracks and/or that NaC1 does
not have a specific effect on CF or SCC, as NaNO 2 does.
In ammoniacal solutions where the failure is caused by a
single main crack, uniform dissolution of the surface takes
place; even if secondary cracks were formed they would be

214 Int J Fatigue October 1983

removed and their main effect would occur within the
propagation period. According to Tomashov, ;8 this pheno-
menon is characteristic where corrosion factors predominate,
while for NaNO 2 the mechanical factor predominates.
The mode of fracture in the initiation stage of cracking
always showed some intergranular character. Thus, even in
the specimens fatigued in air where the fracture mode was
transgranular, a mixed mode was observed at the crack
initiation zone as shown in Fig. 7c. This special initiation
character was also found by Misawa 2 for brass specimens
fatigued in silicon oil. The transgranular mode was also
observed for pure copper and copper-base alloys fatigued in
air. 4 The fatigue crack in aqueous environments ranged from
mixed to mainly intergranular. It was observed that applying
anodic potential and lowering the applied stress resulted in
increasing the percentage of intergranular cracks,and it
became totally intergranular in ammonlacal solution. The
fact that low stresses increase the percentage of intergranular
cracking may be due to the increase in fatigue life (ie dura-
tion of experiment), and thus, the increase in the effect of
the environment. The result obtained in ammoniacal solution
(pH = 5.6) is in agreement with that of Misawa;2 he reported
that the fracture surface of (x-brass grov~n in a similar
solution (pH4.3-7.0) showed a predominately intergranular
mode of cracking and at pH = 7.5 it showed a mixed mode.
A similar shift in the mode of failure was also reported by
Gouda et al5 for SCC of brass.
The fact that intergranular-type cracking predominates
under CF conditions suggests that enhanced deformation
associated with grain boundaries causes enhanced prefer-
ential dissolution of the grain boundaries compared with
that at slip hands. Misawa 2 and Hahn 4 concluded that
shifting from transgranular in air to intergranular under
CF conditions may result from localized dissolutions due
to high dislocation densities and geometrical constraints at
grain boundaries. Further reduction in life does not
necessarily indicate an increase in intergranular cracking.
Conclusions
1) The corrosivity of the solution is not always the
primary factor in determining the severity of CF of
a-brass. Thus, the order of tendency towards corro-
sion is: sodium nitrite < sodium chloride <
ammonium sulphate < arnmoniacal copper sulphate,
while the order of tendency towards CF is: sodium
chloride < ammonium sulphate < ammoniacal
copper sulphate ~< sodium nitrite.
2) No reduction in life takes place in water or sodium
chloride, while in ammoniacal and nitrite solutions
considerable reduction in life takes place under free
corrosion conditions. With anodic polarization,
further reductions in life in all solutions take place.
However, in chloride, the enhanced failure is due to
electrochemical thinning.
3) No single mechanism could explain the behaviour
obtained in all solutions tested. The suggested mech-
anisms operating in nitrite solution are film rupture
and adsorption, while in ammoniacal solution the
mechanochemical model is most likely operating.
Metallographic examinations reveal that the corrosion
factor is predominant in ammoniacal solution, while
the mechanical factor is predominant in nitrite
solution.
4) The mode of failure in air is transgranular, while in
nitrite and ammoniacal solutions it is a mixed mode
I n t J Fatigue O c t o b e r 1983
with an increasing percentage of intergranular type
as the cyclic stress is lowered or anodi¢ potential is
applied.
Acknowledgment
The authors are grateful to the National Science Foundation,
USA, which supported most of this work under contract
no INT 76 17358.
References
1. Birely, S. S. and Tromans, D. 'Corrosion fatigue of copper
and ~-brass' J Electrochern Soc 119 (1972) p 1278
2. Mimwa, T. 'The corrosion fatigue and stress corrosion
cracking of (~-brass in ammoniacal solution' Corros Sci 18
(1978) p 199
3. Hahn, H. N. and Duquette, D. J. 'The effect of heat treat-
ment on the fatigue and corrosion fatigue behaviour of
CuNiCr alloy' Met Trans 10A (1979) p 1453
4. Hahn, H. N. 'Corrosion fatigue behaviour of copper and
copper base alloys' Ph D dissertation (Rensselaer Polytechnic
Institute, USA, 1977)
5. Gouda, V. K., El-,Sayed, H. A. and Sayed, S. M. 'Stress corro-
sion cracking of 72/28 brass in ammonium sulfate solution'
Proc 2nd ME-NACE Corrosion Conference, Bahrain, 1981
p 301
6. Gouda, V. K., EI-Sayad, H. A. and Sayed, S. M. 'Stress corro-
sion cracking of 72/28 brass in sodium chloride solutions'
Proc 8th Int Congress on Metallic Corrosion, Mainz, FRG,
1981 p 479
7. Gouda, V. K., Sayed, S. M. and EI-Sayed, H. A. 'Stress corro-
sion cracking of (~-brass in sodium nitrite solutions' (to be
published)
8. Gouda, V. K., EI-SayKI, H. A. and Sayed, S. M. 'Stress corro-
sion cracking of 72/28 brass in sodium chlorate solution'
Corrosion 38 (1982) p 609
9. McAdam, D. J. 'Corrosion fatigue of metal as affected by
chemical composition, heat treatment and cold working'
J TransAm Soc, Stee/Treat 11 (1927) p 355
10. Uhlig, H. H. Corrosion and Corrosion Control (John Wiley
and Sons, New York, 1971) p 150
11. Gould, A. J. 'Influence of temperature on the severity of
corrosion fatigue' Engineering 141 (1936) p 495
12. Cornet, I. and Golan, S. "Influence of temperature on
corrosion fatigue' Corrosion 15 (1959) p 262 t
1 3. Duquette, D. J. and Uhlig, H. H. 'Effect of dissolved oxygen
and sodium chloride on corros=on fatigue of 0.18% carbon
steel' Trans ASM 61 (1968) p 449
14. Masuda, H. and Duquette, O. J. 'Effect of surface dissolution
on fatigue crack nucleation in polycrystalline copper' Met
Trans 6A (1975) p 87
15. Uhlig, H. H. op cit p 156
16. Gouda, V. K. and Staehle, R. W. 'Effect of metallurgical
variables on the corrosion fatigue behaviour of 304 stainless
steel in sulphuric/chloride media' Br Corros J 15 (1980)
p111
17. Gouda, V. K., Ramadan, A. A. and Youssef, G. I. 'The
effect of sodium nitrite on the fatigue behaviour of brass'
Br Corros J 18 (1983)
18. Tomoshov, N. D. Theory of Corrosion and Protection of
Metals (The Macmillan Company, London, 1966) p 283
Authors
The authors are with the Corrosion Laboratory, Physical
Chemistry Department, National Research Centre, Dokki,
Cairo, Egypt. Inquiries should be addressed to Professor
Gouda.
215