International Journal of Fatigue 30 (2008) 1756–1765

Contents lists available at ScienceDirect

International Journal of Fatigue

journal homepage: www.elsevier .com/locate / i j fa t igue

Corrosion fatigue behavior of extruded magnesium alloy AZ61 underthree different corrosive environments

Md. Shahnewaz Bhuiyan a, Yoshiharu Mutoh b,*, Tsutomu Murai c, Shinpei Iwakami c

a Department of Mechanical Engineering, Nagaoka University of Technology, 1603-1 Kamitomioka, Nagaoka-shi, Niigata 940-2188, Japanb Department of System Safety, Nagaoka University of Technology, 1603-1 Kamitomioka, Nagaoka-shi, Niigata 940-2188, Japanc Sankyo Aluminum Industry Co., Ltd., 13-3 Nagonoe, Shinminato, Toyama 934-8577, Japan

a r t i c l e i n f o

Article history:Received 1 November 2007Received in revised form 16 February 2008Accepted 24 February 2008Available online 7 March 2008

Keywords:Corrosion fatigueCorrosion pitHumidityNaClCaCl2

Magnesium alloys

0142-1123/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.ijfatigue.2008.02.012

* Corresponding author. Tel.: +81 258 47 9735; faxE-mail address: mutoh@mech.nagaokaut.ac.jp (Y.

a b s t r a c t

Corrosion fatigue process of extruded AZ61 magnesium alloy has been investigated under three differentcorrosive environments: (a) high humidity environment (80% relative humidity), (b) 5 wt% NaCl solutionenvironment, and (c) 5 wt% CaCl2 solution environment. The fatigue strength drastically reduced underthe three different corrosive environments: the reduction rates of fatigue limit under high humidity, NaCland CaCl2 environments were 0.22, 0.85 and 0.77, respectively. The drastic reduction in fatigue limitunder corrosive environments resulted from pit formation and growth to the critical size for fatigue cracknucleation. It is suggested that the NaCl environment enhances pit formation and growth more than theCaCl2 environment, due to the high Cl� concentration and low pH value. It is also found that the corrosionpit grows during fatigue cycles and the fatigue crack starts when the pit reaches a critical size for cracknucleation. The critical size is attained when the stress intensity factor range reaches the threshold value.The ratio between pit growth life to fatigue crack nucleation and total fatigue life was about 30%.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Magnesium alloys are now increasingly used for casings of elec-tronic products, vibrating plates of vibration test machines, wheels,etc., wherever the weight is the major concern. This is attributed totheir inherent superior properties such as very low density, highspecific strength, excellent castability and machinability, gooddamping capacity and so on. The demand for use of magnesium al-loys as structural materials in automobile industry has increased inrecent years. The use of magnesium alloys in automotive reducesthe total weight of the component up to 20%. At the same time thiswill substantially reduce the fuel consumption and the CO2 emis-sions. In order to apply the magnesium alloy to high strength struc-tural components in automobile, aerospace and other industries,the characterization of fatigue properties is highly required.

Corrosion fatigue may be defined as the combined action of anaggressive environment and a cyclic stress leading to prematurefailure of metals by cracking [1,2]. Several surveys of corrosion fail-ures in different industries have shown that 20–40% of the failuresexperienced have been due to corrosion fatigue [3,4]. Since manymechanically loaded parts are often subjected to prolonged cyclic

ll rights reserved.

: +81 258 47 9770.Mutoh).

stresses in an active medium, it is of significant importance tostudy the corrosion fatigue characteristics of magnesium alloys.

Wet argon has been found to decrease the fatigue limit of castmagnesium alloy AZ91D and AM60B [5,6]. It should be noted thatmost of these studies were performed on die cast materials, whichinclude casting defects such as pores. It is known that the poressignificantly affect the fatigue strength as stress concentrators,and also that their size, number and shape depend on the fabrica-tion process and condition. Therefore, it is important to investigatethe fatigue behavior of pore-free materials such as extruded mate-rials to understand basic fatigue behavior of magnesium alloys.Hilpert and Wagner [7] examined the fatigue behavior of extrudedhigh strength AZ80 magnesium alloy under a spray of aqueousNaCl solution. The fatigue life was considerably reduced at stressesbelow the fatigue limit under ambient air, while no significant ef-fect of NaCl solution on fatigue life was found at high stress ampli-tudes. Unigovski et al. [8] carried out the corrosion fatigue tests ofextruded AZ31 and AZ91D magnesium alloys under aggressiveenvironment as well as die cast AM60 alloy and reported that ex-truded alloys showed a significantly longer fatigue life in both airand NaCl solution in comparison with die-cast alloys. Sajuri et al.[9] studied the fatigue behavior of extruded AZ61 magnesium alloyunder tension–compression loading and reported that the fatiguebehavior of AZ61 magnesium alloy was highly sensitive to thehumidity under the ambient environment. However, number of

M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765 1757

research works on corrosion fatigue of magnesium alloys has beenstill limited. Detailed corrosion fatigue processes and effects of fre-quency, stress ratio, etc. have not yet been fully understood. Sys-tematic and basic study on corrosion fatigue of magnesium alloyshas been highly requested.

As corrosive environments, both NaCl and CaCl2 environmentsare used, where NaCl environment can simulate the seashoreatmosphere and CaCl2 is a deicing agent for traffic roads. So, it isimportant to know the corrosion fatigue behavior of CaCl2 environ-ment as well as NaCl environment. Therefore, in the present study,as the first step of the systematic study, for understanding basiccorrosion fatigue processes of magnesium alloys, fatigue tests ofextruded AZ61 magnesium alloy were carried out under three dif-

Table 1Chemical composition of the AZ61 magnesium alloy used (mass%)

Al Zn Mn Fe Si Cu Ni Mg

5.95 0.64 0.26 0.005 0.009 0.0008 0.0007 Bal.

Table 2Mechanical properties of the AZ61 magnesium alloy used

Yield stress,r0.2 (MPa)

Tensilestrength,rB (MPa)

Elongation(%)

Young’smodulus,E (GPa)

Vickershardness(HV)

AZ61 185 318 18 43 58.2

Fig. 1. Microstructure of as-received AZ61 magnesium alloy.

Fig. 2. Specimen with 6 mm gauge d

ferent corrosive environments: (a) high humidity (80% relativehumidity (RH)), (b) sprayed 5 wt% NaCl solution environment,and (c) sprayed 5 wt% CaCl2 solution environment. The extrudedmaterial was selected for avoiding the influence of casting defecton fatigue behavior.

2. Experimental procedure

2.1. The material

The material used in the present study was an extruded AZ61magnesium alloy. The chemical composition and the tensile prop-erties of the AZ61 magnesium alloys are shown in Tables 1 and 2,respectively. The microstructure of the as-received extruded AZ61magnesium alloy bar (16 mm in diameter) was observed under adigital microscope, as shown in Fig. 1. The grain structures wereuniform and equi-axed. Sajuri et al. [10] observed the microstruc-ture of extruded AZ61 magnesium alloys in three different planes:longitudinal, transverse and 45� to extrusion direction and con-cluded that no difference in grain structures among the three dif-ferent planes was observed. The specimen in 45� direction fromthe extruded direction showed lower tensile strength comparedto those in the longitudinal and transverse directions, where asthe fatigue strength of the specimen in the longitudinal directionwas higher compared to those for the specimens in the transverseand 45� directions. Moreover, they found that texture introducedby extrusion does not significantly influence on fatigue crackgrowth behavior.

2.2. The choice of the specimen

Preliminary fatigue tests were performed on the specimenswith 6 mm gauge diameter and 40 mm gauge length as per ASTME 466 under high humidity (HH) condition (relative humidity wasabout 80%), which resulted in failure of the threaded end region ofspecimen, as shown in Fig. 2. Also evidence of corrosion productswas found at the failure location. Such failure has not been ob-served in other structural materials like steels, aluminum alloysand so on. In the present experiment, the magnesium alloy speci-men was coupled with stainless steel grips. Since magnesium isranked the lowest among the metals in the electromotive force(emf) and galvanic series, galvanic corrosion due to potential dif-ferences between magnesium alloy and stainless steel could be in-duced in the magnesium alloy specimens. The combination of lowcorrosion resistance and high notch sensitivity of magnesium alloymight have promoted the observed failure not in the gauge partbut in the threaded region of specimen. In order to avoid the fail-ure in the threaded region, systematic modification of specimen

iameter failed from the thread.

1758 M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765

geometry (in terms of decreasing gauge diameter and correspond-ing gauge length) was carried out. It was finally found that thespecimen with 3 mm gauge diameter and 6 mm gauge length pro-vided fatigue failure in the gauge part. Therefore, this specimengeometry showed in Fig. 3 was adopted for all the fatigue testsin the present study.

The round bar fatigue specimens with threaded ends were ma-chined from the as-received round bar. After machining, the spec-imens were polished in the loading direction with 280–800 gritemery papers in laboratory air and with 1000–1500 grit emery pa-pers under kerosene oil to prevent corrosion of the specimen sur-face during polishing process. After all polishing processes werecompleted, the specimens were cleaned by ethanol in an ultrasoniccleaner.

2.3. Fatigue test

Tension–compression fatigue tests were conducted on a servo-hydraulic fatigue testing machine using a sinusoidal wave formwith a frequency of 20 Hz and a stress ratio of �1 under three dif-ferent corrosive environments; (a) high humidity environment(80% RH), (b) sprayed 5 wt% NaCl solution environment with apH value of 6.59 (noted as NaCl environment) and (c) sprayed

4325 25

93

εφ r

M12×1.256R40

M12×1.25

Fig. 3. Shape and sizes of the fatigue specimen.

103 104 10510

20

30

4050

60

70

80

90

100

110

120

130

140

150

160

170

180

Number of cycles

Stre

ss a

mpl

itude

, σa

[MPa

]

R = -1, f = 20 Hz Low Humidity (16°C-20°C, 3 High Humdity (50°C, 80-85%5 wt% NaCl (pH=6.59)

5 wt% CaCl2 (pH=8.07)

P

Fig. 4. S–N curves for AZ61 magnesium alloy un

5 wt% CaCl2 solution environment with a pH value of 8.07 (notedas CaCl2 environment). The fatigue tests for obtaining the basicfatigue properties of AZ61 magnesium alloy with no influence ofcorrosion were carried out under low humidity environment (at16–20 �C with 35–40% RH). In order to examine the influence ofhumidity, fatigue tests were conducted under high humidity(80% RH) in a specially designed chamber which can control therelative humidity level ranging from 30% to 80% RH. Fatigue testsunder the NaCl and CaCl2 environments were conducted in a31.5 � 26.5 � 17 cm3 chamber by spraying the solution with a flowvolume of 1.6 ml/h at 0.1 MPa air pressure, where the temperatureof solution was 25.1 �C and the temperature inside the chamberwas 25.4 �C. The fatigue test was continued up to complete failureof the specimen, while the test was stopped when the specimendid not fail up to 107 cycles.

Specimen surfaces and fracture surfaces of the specimens wereobserved in detail by using a scanning electron microscope (SEM)to investigate the pitting process and the fracture surface morphol-ogy under three different corrosive environments.

3. Results and discussion

3.1. S–N curve

The relationships between stress amplitude and number of cy-cles to failure under three different corrosive environments areshown in Fig. 4. The result under low humidity condition (35–40% RH), where no corrosion fatigue took place, is also indicatedin the figure. The extruded AZ61 magnesium alloy indicated a fati-gue limit of 135 MPa under the low humidity condition. The fati-gue limit was reduced to 105 MPa in the high humidityenvironment. However, no significant difference in fatigue life be-tween low and high humidity conditions was observed at stressamplitudes higher than the fatigue limit under the low humidity.A drastic reduction in fatigue limit was observed in the NaCl andCaCl2 environments. In the NaCl environment, the fatigue limitwas 20 MPa, while it was 30 MPa in the CaCl2 environment. TheNaCl environment is more detrimental for the fatigue limit of theAZ61 magnesium alloy compared to the CaCl2 environment, while

106 107 108

to failure, Nf [Cycles]

5-40% RH) RH)

P

P

der three different corrosive environments.

M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765 1759

the difference of fatigue limit was not much significant. The effectof corrosive environment on fatigue strength was evaluated byusing the reduction rate of fatigue limit (RRFL), as shown in Table3, where RRFL is given as

RRFL ¼ rLH � rCE

rLHð1Þ

where rLH is the fatigue limit under low humidity condition and rCE

is that under corrosive environment.It can be seen from the table that the NaCl environment reduces

the fatigue limit four times and the CaCl2 environment reduces it3.5 times more than the high humidity (80% RH) environment. Ithas been reported that Chloride ions promote rapid attack of mag-nesium in neutral, aqueous solutions. The corrosion rate increasesrapidly with increasing Cl� concentration [11]. Extruded magne-sium alloys with 3–8% Al and 0.5–0.8% Zn are susceptible to fili-form corrosion and pitting corrosion in aqueous chloridesolutions depending on chloride concentration [12]. In the presentstudy, in 5% NaCl solution the Cl� concentration is 0.85 mEq/L (mil-liequivalents per liter), where as in CaCl2 solution it is 0.45 mEq/L,indicating that the Cl� concentration in NaCl is almost double thatcompared to the Cl� in CaCl2. This high Cl� concentration in NaClsolution makes the NaCl environment more detrimental than theCaCl2 environment. On the other hand, it is also reported that cor-rosion fatigue life of magnesium alloys depends highly on the pHvalues of the solution [13]. The pH value of the medium has animportant impact on the corrosion morphology of magnesiumalloys. Magnesium is very resistant to corrosion by alkalis of thepH exceeds 10.5 which corresponds to the pH of a saturatedMg(OH)2 [14]. The corrosion rate of AZ91 ingot and die cast is highin acidic solutions (pH 1–2) as compared to that in neutral andhighly alkaline solutions (pH 4.5–12) [15]. Based on these studywe may speculate that since the pH (6.59) of NaCl is lower thanthe pH (8.07) of CaCl2, the NaCl is more corrosive than the CaCl2

environment. Therefore, on the basis of the above discussion, we

Table 3Effect of corrosive environment on the fatigue limit of the AZ61 magnesium alloy

RRFL (using Eq. (1))

High humidity 5 wt% NaCl 5 wt% CaCl2

Extruded AZ61 0.22 0.85 0.77

rLH is the fatigue limit under low humidity; rCE is the fatigue limit under corrosiveenvironment.

Fig. 5. SEM fractographs showing: (a) overview of the fracture surface and (b) high magcondition.

can say that due to the combine effect of the Cl� concentrationand the pH value, NaCl environment is more detrimental thanthe CaCl2 environment.

By using Eq. (1), similar results have been obtained on AZ80 inNaCl solution (RRFL = 73%) [7], extruded AZ31 and AM50 magne-sium alloys in NaCl solution (RRFL = 34% and 27%, respectively)[8], extruded AZ61 under high humidity (RRFL = 27%) [9], castAZ91E in 3.5% NaCl aqueous environment (RRFL = 83%) [16], elec-tro polished AZ31 and AZ80 in NaCl environment (RRFL = 27%and 81%, respectively) [17]. From these comparisons, it seemedthat, fatigue strength under NaCl environment of high strengthmagnesium alloys like AZ91E, AZ80 degraded more than those oflow strength magnesium alloys like AZ31, while the fatigue datafor magnesium alloys specially under NaCl environment fell in awider scatter band (RRFL = 27–83%).

3.2. Fractographic observations

SEM fractographs for the specimen tested at stress level of165 MPa under low humidity condition (36–40% RH) are shownin Fig. 5. The crack nucleation region is shown by a rectangularmark in Fig. 5a and is shown at high magnification in Fig. 5b. Thecrack nucleation region (Fig. 5b) shows no existence of foreign par-ticles such as inclusions and also no defect on the specimen sur-face. The crack nucleation region was relatively flat withtransgranular fracture, which was commonly observed in smoothspecimens of other structural metals. Fig. 6a and b shows SEM frac-tographs for the specimen tested under high humidity condition(80% RH) at stress amplitude of 155 MPa (higher than the fatiguelimit under low humidity condition). The crack nucleation regionrevealed the similar features as found under low humidity condi-tion. However, under high humidity environment, as the stressamplitudes become lower than 135 MPa (the fatigue limit underlow humidity condition), the failure of the specimen was attrib-uted due to the formation of corrosion pit. Fig. 6c and d showsthe presence of corrosion pit in the crack nucleation region onthe specimen surface tested under high humidity condition (80%RH) at stress amplitude of 115 MPa. Fig. 7 shows fracture surfacesof the specimen tested under the NaCl environment at stress levelsof 130 MPa (Fig. 7a and b) and 50 MPa (Fig. 7c and d). High magni-fication of the crack nucleation site revealed the existence of corro-sion pit. Moreover, the specimen surfaces and the fracture surfacestested under the NaCl environment were covered with corrosionproducts. Fig. 8 shows fracture surfaces of the specimens tested

nification of the crack nucleation region tested at 165 MPa under the low humidity

Fig. 6. SEM fractographs showing: (a) overview of the fracture surface and (b) high magnification of the crack nucleation region tested at 155 MPa; (c) overview of thefracture surface and (d) corrosion pit at the crack nucleation point on the specimen surface tested at 115 MPa under the high humidity condition.

Fig. 7. SEM fractographs showing: (a) overview of the fracture surface and (b) existence of corrosion pit at the crack nucleation site tested at 130 MPa; (c) overview of fracturesurface and (d) existence of corrosion pit at the crack nucleation site tested at 50 MPa under the NaCl environment.

1760 M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765

Fig. 8. SEM fractographs showing: (a) overview of the fracture surface and (b) existence of corrosion pit at the crack nucleation site tested at 110 MPa; (c) overview of fracturesurface and (d) existence of corrosion pit at the crack nucleation site tested at 50 MPa under the CaCl2 environment.

Fig. 9. Pit growth observations of the specimen tested up to (a) 10,000 cycles and (b) 30,000 cycles under NaCl environment at 120 MPa.

M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765 1761

at 110 MPa (Fig. 8a and b) and 50 MPa (Fig. 8c and d) under theCaCl2 environment. As can be seen from the figure, corrosion pitswere also found at the crack nucleation site.

3.3. Pit growth behavior

Since NaCl environment is more detrimental than CaCl2 envi-ronment (as shown in Table 3), the pit growth behavior was ob-served under this environment by performing interrupted fatiguetest. At a high stress amplitude of 120 MPa, the isolated singlepit was found to form and grow, as shown in Fig. 9. The size of

pit found on the specimen surface was about 7.5 lm at 10,000 cy-cles (28% of total life), which increased to 24 lm at 30,000 cycles(83% of total life). Therefore, the isolated pit is formed and growsto a critical size for starting a fatigue crack at high stress level. Sim-ilar pit growth behavior under high humidity environment was re-ported on AZ61 alloy [9], where the stress level was rather high. Itis known that extruded magnesium alloy indicates metallographicanisotropy. However, clear effect of anisotropy on pit geometrywas not observed in the present experiment. In order to investigatethe interactive action of cyclic loading and corrosive environment,one of the specimen was placed without loading under NaCl

1762 M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765

environment up to the time equivalent to 432,510 cycles. It wasfound that without loading under NaCl environment, few corrosionpits nucleated and grew on the specimen surface, the average sizeof which was about 28 lm. On the other hand, under simultaneousaction of cyclic loading and NaCl environment, at 80 MPa one ofspecimens failed after 305,540 cycles which corresponds to theclosest life duration of that without cyclic loading and the pit sizeat the fracture origin was about 70 lm, as shown in Fig. 10. Theseobservations suggested that formation and growth of corrosionpits under corrosive environment were significantly enhanced bycyclic loading. In order to compare the pit growth behavior atlow stress amplitudes, a similar interrupted fatigue test under NaClenvironment at a low stress amplitudes of 50 MPa was conducted.Fig. 11 shows the surface observations of the specimens inter-rupted at various loading cycles. As can be seen from the figurethat, after 50,000 cycles (2.8% of total life), corrosion pits werefound on the specimen surface (Fig. 11a). The number of corrosionpits was increased with increasing number of cycles. However, theformation sites of corrosion pits were not uniform but localized insome area, as seen from Fig. 11b. Coalescence of corrosion pits inthe localized area was found to form a large pit as further increaseof cycles, as found in Fig. 11c and d. Therefore, pit formation pro-cess under NaCl environment at low stress amplitudes is consid-ered as follows: multiple corrosion pits nucleated in somelocalized area on the specimen surface, number of pits increasedwith increasing number of cycles and finally localized dense pitscoalesced to form a large corrosion pit. This large coalesced pitcould become a fatigue crack starter at very low stress amplitude.

3.4. Fatigue crack nucleation life

Fatigue crack nucleation life after crack nucleation from a pitcan be predicted based on the Paris equation modified by thethreshold stress intensity factor range, as follows:

dadN

� �crack¼ C DKm � DKm

th

� �ð2Þ

DK ¼ rDrffiffiffiffiffiffipap

ð3Þ

Np ¼Z af

ai

1C aDr

ffiffiffipp

ð Þmam2 � CDKm

th

da ð4Þ

104 1050

100

200

300

400

500

No of cycles to

Cri

tical

pit

size

(μm

)

Pit size with fatigue loading un

Pit size without cyclic loading u

Fig. 10. Relationship between stress amplitude and critical pit size and number of c

From Ref. [18] C and m has been taken as 2.0 � 10�10 m cycle�1 and3.0.

Therefore, the crack nucleation life (Ni) can be obtained asfollows:

Ni ¼ Nf � Np ð5Þ

By using Eq. (4), the crack propagation life was evaluated as211,137 cycles at stress amplitude of 80 MPa under NaCl environ-ment. Therefore, the crack nucleation life at this stress amplitudewas 94,403 cycles (using Eq. (5)). Therefore, the crack nucleationlife is expected to be almost 30% of total fatigue life (Nf), whichis rather low compared to the reported crack nucleation life (about50% of the total fatigue) [9,18].

3.5. Relationship between stress amplitude and critical pit size

It can be considered that the pit size observed at the fractureorigin is the critical pit size for fatigue crack nucleation. Mechani-cal stress is a major factor for the pit formation and growth undercorrosive environment as seen from Fig. 10. Therefore, once a fati-gue crack starts from the pit, stress around the pit is released andfurther pit growth is not significant. The pit size was defined as theradius of semi-circular crack [19], which had the equivalent area tothe pit observed at the crack nucleation site. From the detailedfractographic observations, the relationships between stressamplitude and critical pit size under the NaCl and the CaCl2 envi-ronments are shown in Fig. 12. As can be seen from the figure,the relationship under the NaCl environment is almost a straightline for all stress amplitudes tested except for stress amplitudeshigher than 130 MPa, which indicates the so-called Kitagawa–Takahashi diagram for explaining small crack behavior. Thestraight line gives the threshold stress intensity factor range as0.8 MPa

pm. Therefore, this linear relationship suggests that a

fatigue crack starts from a pit when the stress intensity factorrange reaches the threshold value. A similar hypothesis has beenproposed by Hoeppner [20]. The relationship under the CaCl2 envi-ronment is also almost a straight line in the range of stress ampli-tudes lower than 100 MPa, which also gives the threshold stressintensity factor range of 0.8 MPa

pm. However, at stress ampli-

tudes higher than 100 MPa, the relationship shows a deviation

106 107

failure, Nf [Cycles]

der 5% NaCl

nder 5% NaCl

ycles under NaCl environments with cyclic loading and without cyclic loading.

10-1

100

101

102

103

102

Critical pit size (μm)

Stre

ss a

mpl

itude

, σa

[MPa

]

5% NaCl (pH=6.59)5% CaCl2 (pH=8.07)High humdity (HH)

ΔKTH = 0.8MPam1/2

Fatigue strength without influence of corrosion pitting

NaCl

CaCl2

HH

Fig. 12. Relationship between stress amplitude and critical pit size under corrosive environments.

Fig. 11. Specimen surface observations at the same location of the specimens tested under the NaCl environment at 50 MPa up to (a) 50,000 cycles, (b) 100,000 cycles, (c)600,000 cycles and (d) 800,000 cycles.

M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765 1763

from the linear line, which also indicates the so-called Kitagawa–Takahashi diagram similar to that under the NaCl environment.The similar tendency is observed under high humidity condition,as indicated in Fig. 12.

In the small crack region, applying the El Haddad equation [21–23] which is given as

DKth ¼ aDrth

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffipðaþ a0Þ

pð6Þ

1764 M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765

where a0 is a constant known as ‘‘an effective crack length” and thevalue of a0 can be obtained by

a0 ¼1p

DKth

are

� �2

ð7Þ

In the small crack problem without influence of corrosion, re hasbeen taken as the fatigue limit rw, as found in the El Haddad equa-tion. This can be interpreted in the following way: the rw gives thelowest stress, below which fatigue failure (crack nucleation andpropagation) cannot occur without pre-existence of small crack(or defect). In the present study, re is considered as the loweststress, above which the corrosion pit is not the crack nucleation siteand has no influence on fatigue crack nucleation. If the fatiguestrength without influence of pit re is assumed as 140 MPa underhigh humidity environment, 200 MPa under NaCl environment

10 3 10 4 10 50

200

400

600

Number of cycles t

Stre

ss a

mpl

itude

, σa

[MPa

]

103 104 1050

0.2

0.4

0.6

0.8

1

Number of cycle

σ a/σ B

a

b

Fig. 13. S–N curves for steel, aluminum alloy and magnesium alloy (a) under ambient amagnesium alloy.

and 160 MPa under the CaCl2 environment and a as 0.67 for semi-circular crack, the relationship between stress amplitude and criti-cal pit size in the small crack region can be obtained by the ElHaddad equation, as shown in Fig. 12. It can be found from the fig-ure that the data points under the three different corrosive environ-ments follow the El Haddad equation. It is also suggested that theNaCl environment enhances pit formation and growth more thanthe CaCl2 environment, since the fatigue strength without influenceof corrosion pitting is higher under the NaCl environment, as seenfrom Fig. 12.

From the foregoing discussion, it is found that the corrosion pitgrows during fatigue cycles and the fatigue crack starts when thepit reaches a critical size for crack nucleation. The critical size is at-tained when the stress intensity factor range reaches the thresholdvalue not only for long crack region but also for small crack region,where the El Haddad correction has to be taken.

10 6 107 10 8

Steel, air Steel, 3% NaCl Al alloy, air Al alloy, 3.5% NaCl Mg alloy, air Mg alloy, 5% NaCl

o failure, Nf [cycles]

106 107 108

s to failure, Nf [Cycles]

Steel (σB = 1012 MPa), airSteel (σB = 1012 MPa), 3% NaCl

Al alloy (σB = 493 MPa), air Al alloy (σB = 493 MPa), 3.5% NaCl Mg alloy (σB = 318 MPa), air Mg alloy (σB =318 MPa), 5% NaCl

ir and NaCl environment. (b) Normalized S–N curves for steel, aluminum alloy and

M.S. Bhuiyan et al. / International Journal of Fatigue 30 (2008) 1756–1765 1765

3.6. Comparison of corrosion fatigue strength

Corrosion fatigue strengths of structural materials other thanmagnesium alloy are available because of importance of corrosionfatigue problems. For example, Lin et al. [24] studied the fatiguebehavior of 7050-T73 high strength aluminum alloys in both labo-ratory air and 3.5% NaCl solution environment with a frequency of20 Hz and reported that the fatigue strength of the alloy in NaClenvironment was reduced to 67.4% relative to the fatigue strengthin the low humidity environment. Tokaji et al. [25] also studied thefatigue behavior of low alloy steel (Cr–Mo) under both laboratoryair and 3% NaCl environment with a frequency of 19 Hz and re-ported a large reduction of fatigue strength (50–80%) under NaClenvironment. The present work on magnesium alloy also shows asimilar trend of reduction of fatigue strength under corrosive envi-ronment, as shown in Fig. 13a. In all the cases, the reduction in fa-tigue strength under NaCl environment was attributed to theformation of corrosion pits. Since the materials plotted inFig. 13a has different ultimate tensile strengths (rB), the stressamplitude was normalized by the ultimate tensile strength of eachmaterial. The resultant normalized plot is shown in Fig. 13b. It isfound from the figure that under NaCl environment, S–N curvesfor all three materials fall in the same data band. This means thatthe difference in the corrosion fatigue strength shown in Fig. 13aprimarily results from the difference of ultimate strength, despitethe fact that the pit formation and growth behavior and thresholdstress intensity factor of each material are different.

4. Conclusions

Fatigue tests were carried out under three different corrosiveenvironments and also in air to understand the basic corrosion fa-tigue processes of AZ61 magnesium alloy. From the results ob-tained, the following conclusions are summarized:

1. The fatigue limit under low humidity (36–40% relative humid-ity) with no effect of corrosion was obtained at 135 MPa andthe ratio of fatigue limit to tensile strength was about 0.42.Sajuri et al. [9] also found that the ratio of fatigue limit to tensilestrength was about 0.45 under low humidity environment.

2. The fatigue limit was reduced to 105 MPa under high humidity(80% RH). The reduction rate of fatigue limit (RRFL) due to highhumidity was 0.22, which indicates that the fatigue strength ofAZ61 magnesium alloy is highly sensitive to the humidity levelof ambient environment.

3. The fatigue limits under the NaCl and CaCl2 environments were20 and 30 MPa, respectively. The RRFL values under the NaCland CaCl2 environments were 0.85 and 0.78, respectively.

4. The drastic reduction in fatigue limit under corrosive environ-ments resulted from pit formation and growth to the criticalsize for fatigue crack nucleation. Corrosion pit formation andgrowth were attributed to the interactive action of mechanicalloading and corrosive environment. It was found that at low

stress amplitudes under corrosive environments, multiple cor-rosion pits nucleated on the specimen surface, grew and thencoalesced to form a large corrosion pit. On the other hand, anisolated pit was formed and grew to the critical size at higherstress amplitudes. The ratio between pit growth life to fatiguecrack nucleation and total fatigue life was about 30%, which issimilar to that reported in Refs. [9,18].

References

[1] Fontana MG. Corrosion engineering. third ed. New York: McGraw-Hill; 1987.[2] Schweitzer PA. Corrosion and corrosion protection handbook. New

York: Marcel Dekker; 1989.[3] Uhlig HH, Mears RR. Corrosion handbook. New York: John Wiley; 1948. p. 21–

35.[4] Brown BF. Corrosion fatigue in naval structure. In: International meeting on

corrosion fatigue; 1971.[5] Ishikawa K, Kobayashi Y, Kido K. Endurance limit and fatigue crack

propagation of magnesium alloys. In: Proceedings of the internationalconference on fatigue in the very high cycle regime; 2001. p. 99–205.

[6] Mayer HR, Stanzl-Tschegg S. Environmental influences at very high frequency.In: Proceedings of the international conference on fatigue in the very highcycle regime; 2001. p. 267–74.

[7] Hilpert M, Wagner L. Corrosion fatigue behavior of the high-strengthmagnesium alloy AZ80. J Mater Eng Perform 2000;9(4):402–7.

[8] Unigovski Y, Eliezer A, Abramov E, Snir Y, Gutman EM. Corrosion fatigue ofextruded magnesium alloys. Mater Sci Eng A 2003;360:132–9.

[9] Sajuri ZB, Miyashita Y, Mutoh Y. Effect of humidity and temperature on thefatigue behavior of an extruded AZ61 magnesium alloy. Fatigue Fract EngMater Struct 2005;28:373–9.

[10] Sajuri ZB, Miyashita Y, Hosokai Y, Mutoh Y. Effect of texture on fatigueproperties of an extruded AZ61 magnesium alloy plate. Key Eng 2004;274–276:193–8.

[11] Song G, Atrens A. Understanding of magnesium corrosion. Adv Eng Mater2003;5(12):837–58.

[12] Lubbert K, Kopp J, Wendler-Kalsch E. Corrosion behaviour of laser beamwelded aluminium and magnesium alloys in the automotive industry. MaterCorros 1999;50:65–72.

[13] Eliezer A, Gutman EM, Abramov E, Unigovski Y. Corrosion fatigue of die-castand extruded magnesium alloys. J Light Met 2001;1(3):179–86.

[14] Makar GL, Kruger J. Corrosion of magnesium. Int Mater Rev1993;38(3):138–53.

[15] Ambat R, Aung NN, Zhou W. Evaluation of microstructural effects on corrosionbehavior of AZ91D magnesium alloy. Corros Sci 2000;42:1433–55.

[16] Stephens RI, Schrader CD, Lease KB. Corrosion fatigue of AZ91E-T6 castmagnesium alloy in a 3.5 percent NaCl aqueous environment. ASTM J EngMater Technol 1995;117:293–8.

[17] Hilpert M, Wagner L. Environmental effects on the HCF behavior of themagnesium alloys AZ31 and AZ80. Miner Met Mater Soc 2000:375–81.

[18] Sajuri ZB, Miyashita Y, Mutoh Y. Fatigue-life prediction of magnesium alloysfor structural applications. Adv Eng Mater 2003;5(12):910–6.

[19] Dolley EJ, Lee B, Wei RP. The effect of pitting corrosion on fatigue life. FatigueFract Eng Mater Struct 2000;23:555–60.

[20] Hoeppner DW. Model for prediction of fatigue lives based upon a pittingcorrosion fatigue process. ASTM special technical publication (675),symposium on fatigue mech; 1978. p. 841–70.

[21] Elhaddad MH, Topper TH, Topper TN. Fatigue life predictions of smooth andnotched specimens based on fracture mechanics. ASTM J Eng Mater Technol1981;109:91–6.

[22] Elhaddad MH, Topper TH, Topper TN. Fatigue crack propagation of shortcracks. ASTM J Eng Mater Technol 1979;101:42–6.

[23] Elhaddad MH, Topper TH, Topper TN. J-integral applications for short crack atnotches. Int J Fract 1980;16:15–30.

[24] Lin Chih-Kuang, Yang Sheng-Tseng. Corrosion fatigue behavior of 7050aluminum alloys under different tempers. Eng Fract Mech 1998;59:779–95.

[25] Tokaji K, Takahashi S. Corrosion fatigue behavior and fracture mechanisms innitrided low alloy steel. Fatigue Fract Eng Mater Struct 2002;26:215–21.