endurance limit and threshold stress intensity of die...

13
Endurance limit and threshold stress intensity of die cast magnesium and aluminium alloys at elevated temperatures H. Mayer a, * , M. Papakyriacou b , B. Zettl a , S. Vacic a a Institute of Physics and Materials Science, BOKU, Peter-Jordan-Str. 82, A 1190 Vienna, Austria b ARC Leichtmetallkompetenzzentrum Ranshofen GmbH, Ranshofen A 5282, Austria Received 23 August 2004; received in revised form 25 January 2005; accepted 10 February 2005 Available online 25 April 2005 Abstract High cycle fatigue properties of the high-pressure die-cast magnesium alloys AZ91 hp, AS21 hp and AE42 hp and of the aluminium alloy AlSi9Cu3 are investigated at elevated temperatures. Fatigue tests are performed at ultrasonic cyclic frequency and load ratio RZK1. Compared with ambient air environment, the S–N curves determined in warm air of 125 8C (magnesium alloys) and 150 8C (aluminium alloy) are shifted towards lower cyclic stresses. The mean endurance limits at 10 9 cycles of 41 MPa (AZ91 hp), 27 MPa (AS21 hp), 35 MPa (AE42 hp) and 61 MPa (AlSi9Cu3) are 70–90% of the respective stresses found in ambient air, and fracture mechanics tests delivered threshold stress intensities at elevated temperature of about 80% of the respective values at 20 8C. Fatigue cracks initiate, nearly, exclusively at casting porosity and fracture surfaces appear similar at low and elevated temperatures. Fatigue cracks may form at voids at stresses below the endurance limit, and non-propagating fatigue cracks of considerable length (greater 1 mm) are found in runnout specimens. Considering porosity as an initial crack, critical stress intensity, K crit is evaluated, which quantifies the sensitivity of a material to defects. Materials can sustain cyclic stresses without fatigue failure, if the cyclic stress intensity determined for their most damaging porosity is lower than K crit . q 2005 Elsevier Ltd. All rights reserved. Keywords: Cast magnesium; Cast aluminium; Porosity; High-cycle fatigue; Endurance limit; Threshold stress intensity; Ultrasonic fatigue 1. Introduction Magnesium alloys have received increased interest as possible construction materials in the automotive industry due to their low mass density, the possible benefit of weight reduction and consequently fuel saving. The rather high specific strength, the good castability and damping property, the improved corrosion resistance of high purity alloys and the possibility of recycling are additional advantages of these materials [1–3]. Today, cast magnesium alloys are used in several automotive applications like covers and door structures and heavier stressed components like wheels, housings and frames, for example. Effective weight reduction is possible using cast magnesium alloys progress- ively more often in load bearing components. Vehicles components are subjected to vibrations and repeated mechanical straining, and the number of load cycles may be very high, such as several 10 8 cycles in car wheels. When magnesium alloys are considered for powertrain components, like crank case or gearbox casing, fatigue properties at elevated temperatures are of great interest and have to be compared to the cyclic properties of well- established cast aluminium. Fatigue stress amplitudes in the range of 80 MPa at 10 8 cycles and room temperature are reported for several cast magnesium alloys [4], which is closely comparable to cast aluminium alloys. These high fatigue strengths, however, are obtained with separately cast specimens, whereas the cyclic strength of material extracted from actually cast components may be significantly lower. Material defects act as starting places for fatigue cracks and deteriorate the fatigue behaviour of the cast. Early fatigue investigations already showed that porosity acts as main crack initiation place in cast magnesium [5,6]. Studying fatigue properties of AM50 alloy samples obtained from different places of prototype control arms and wheels showed that fatigue performance was significantly affected International Journal of Fatigue 27 (2005) 1076–1088 www.elsevier.com/locate/ijfatigue 0142-1123/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijfatigue.2005.02.002 * Corresponding author. Tel.: C43 1 47654 5161; fax: C43 1 47654 5159. E-mail address: [email protected] (H. Mayer).

Upload: duongbao

Post on 17-Sep-2018

237 views

Category:

Documents


1 download

TRANSCRIPT

Endurance limit and threshold stress intensity of die cast magnesium

and aluminium alloys at elevated temperatures

H. Mayera,*, M. Papakyriacoub, B. Zettla, S. Vacica

aInstitute of Physics and Materials Science, BOKU, Peter-Jordan-Str. 82, A 1190 Vienna, AustriabARC Leichtmetallkompetenzzentrum Ranshofen GmbH, Ranshofen A 5282, Austria

Received 23 August 2004; received in revised form 25 January 2005; accepted 10 February 2005

Available online 25 April 2005

Abstract

High cycle fatigue properties of the high-pressure die-cast magnesium alloys AZ91 hp, AS21 hp and AE42 hp and of the aluminium alloy

AlSi9Cu3 are investigated at elevated temperatures. Fatigue tests are performed at ultrasonic cyclic frequency and load ratio RZK1.

Compared with ambient air environment, the S–N curves determined in warm air of 125 8C (magnesium alloys) and 150 8C (aluminium

alloy) are shifted towards lower cyclic stresses. The mean endurance limits at 109 cycles of 41 MPa (AZ91 hp), 27 MPa (AS21 hp), 35 MPa

(AE42 hp) and 61 MPa (AlSi9Cu3) are 70–90% of the respective stresses found in ambient air, and fracture mechanics tests delivered

threshold stress intensities at elevated temperature of about 80% of the respective values at 20 8C. Fatigue cracks initiate, nearly, exclusively

at casting porosity and fracture surfaces appear similar at low and elevated temperatures. Fatigue cracks may form at voids at stresses below

the endurance limit, and non-propagating fatigue cracks of considerable length (greater 1 mm) are found in runnout specimens. Considering

porosity as an initial crack, critical stress intensity, Kcrit is evaluated, which quantifies the sensitivity of a material to defects. Materials can

sustain cyclic stresses without fatigue failure, if the cyclic stress intensity determined for their most damaging porosity is lower than Kcrit.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: Cast magnesium; Cast aluminium; Porosity; High-cycle fatigue; Endurance limit; Threshold stress intensity; Ultrasonic fatigue

1. Introduction

Magnesium alloys have received increased interest as

possible construction materials in the automotive industry

due to their low mass density, the possible benefit of weight

reduction and consequently fuel saving. The rather high

specific strength, the good castability and damping property,

the improved corrosion resistance of high purity alloys

and the possibility of recycling are additional advantages of

these materials [1–3]. Today, cast magnesium alloys are

used in several automotive applications like covers and door

structures and heavier stressed components like wheels,

housings and frames, for example. Effective weight

reduction is possible using cast magnesium alloys progress-

ively more often in load bearing components. Vehicles

0142-1123/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ijfatigue.2005.02.002

* Corresponding author. Tel.: C43 1 47654 5161; fax: C43 1 47654

5159.

E-mail address: [email protected] (H. Mayer).

components are subjected to vibrations and repeated

mechanical straining, and the number of load cycles may

be very high, such as several 108 cycles in car wheels.

When magnesium alloys are considered for powertrain

components, like crank case or gearbox casing, fatigue

properties at elevated temperatures are of great interest

and have to be compared to the cyclic properties of well-

established cast aluminium.

Fatigue stress amplitudes in the range of 80 MPa at

108 cycles and room temperature are reported for several

cast magnesium alloys [4], which is closely comparable to

cast aluminium alloys. These high fatigue strengths,

however, are obtained with separately cast specimens,

whereas the cyclic strength of material extracted from

actually cast components may be significantly lower.

Material defects act as starting places for fatigue cracks

and deteriorate the fatigue behaviour of the cast. Early

fatigue investigations already showed that porosity acts as

main crack initiation place in cast magnesium [5,6].

Studying fatigue properties of AM50 alloy samples obtained

from different places of prototype control arms and wheels

showed that fatigue performance was significantly affected

International Journal of Fatigue 27 (2005) 1076–1088

www.elsevier.com/locate/ijfatigue

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1077

by the shrink and gas porosity at the extraction place [7].

Porosity as the source of fatigue cracks has been found in

several investigations of AZ91, AM50 and AM60 [8–16].

Besides porosity, oxides [17] and Manganese-particle-

enriched regions [18] may act as preferential sources of

fatigue cracks in cast magnesium. Fatigue lifetimes are

influenced by the size of the material defect, i.e. the larger

the defect the shorter the fatigue lifetime [7,8,15,17]. The

lower the cyclic load, the greater the influence of defects on

fatigue properties. Cycling AZ91 T4 at strain amplitude

10K3, for example, led to lifetimes between 103 and

5!105 cycles, depending on the defect size in the sample

[17]. Thus, the fatigue properties of magnesium alloys are

strongly influenced by the quality of the cast. Meaningful

comparison of the cyclic properties of cast magnesium and

aluminium requires testing both materials in conditions

comparable to the intended technical application.

In a previous work [19], the fatigue properties at room

temperature of high-pressure die-cast magnesium alloys

AZ91 hp, AS21 hp, AE42 hp and AM60 hp have been

compared to the aluminium alloy AlSi9Cu3, which is most

frequently used in high-pressure die-casting in Germany.

The materials were produced under conditions comparable

to the casting of actual components. Porosity, which is

inevitable in high-pressure die-casting, was most important

for crack initiation and fatigue lifetime, and cracks, nearly,

exclusively started at these material defects. All alloys

showed an endurance limit in ambient air with maximum

cycles to failure of about 2!107. Fracture mechanics

principles were applied considering porosity as initial

cracks. Failed specimens and runnouts were evaluated to

determine the so-called ‘critical stress intensity factor’, Kcrit,

which determines the maximum stress amplitude a material

can sustain without fatigue failure in presence of porosity.

This parameter was used to link the fracture probability with

the distribution of defects in the cast materials.

Considering porosity as initial cracks is a well estab-

lished method to model the fatigue behaviour of cast

aluminium alloys, and fatigue lifetime may be considered as

numbers of cycles necessary to propagate the crack to

fracture [20–23]. Investigations on 319 T7 aluminium

verified [24,25] that the crack initiation period at porosity

is negligible and cracks may initiate even at stresses below

the endurance limit. The larger the casting defect the shorter

the crack extension period. Thus the observed fatigue

lifetimes are inversely correlated to the size of the defect in

constant amplitude [20–23] as well as in variable amplitude

Table 1

Chemical compositions (in wt%)

Al Zn Mn Si

AlSi9Cu3 Rem 1.2 0.1–0.4 8–11

AZ91 hp 8.90 0.79 0.21 0.01

AS21 hp 2.20 !0.01 0.16 0.98

AE42 hp 3.90 0.004 0.37 0.007

tests [26,27]. Recently, fracture mechanics concept has been

similarly applied to predict lifetimes of cast magnesium

AZ91 by integrating growth rates of cracks initiating at

porosity [15].

In the present investigation, the fatigue properties of

high-pressure die-cast magnesium alloys AZ91 hp, AS21 hp

and AE42 hp are studied at elevated temperatures, and the

aluminium alloy AlSi9Cu3 serves for comparison. Fatigue

lifetime and crack growth experiments are performed and

the obtained results are compared to the fatigue perform-

ance at room temperature. The aluminium alloy is tested at

150 8C. The magnesium alloys are cycled at the lower

temperature of 125 8C due to their inferior creep resistance.

However, this temperature is meaningful for several

practical applications, in gearbox housings for example,

where one of the alloys (AZ91) is actually used [28]. Mainly

high cycle fatigue properties and near threshold fatigue

crack growth behaviour are investigated. The endurance

limit is determined at 109 cycles, and maximum fatigue

crack growth rates of 10K12 m/cycle are used to characterise

threshold stress intensity. The experiments are performed

using the ultrasonic fatigue testing method working at cyclic

frequency 20 kHz. Previous experiments, though at room

temperature, have shown that similar lifetimes are measured

cycling at 20 and 50 Hz [19] and no pronounced effect of

cyclic frequency on fatigue performance of these alloys may

be expected therefore. Fracture surface studies and fracture

mechanics considerations serve to evaluate the influence of

porosity on fatigue properties.

2. Materials and procedure

2.1. Materials

Fatigue properties of the aluminium alloy AlSi9Cu3 and

the magnesium alloys AZ91 hp, AS21 hp and AE42 hp are

investigated. The materials are produced by high-pressure

die-casting at a solidification pressure of 60 MPa (AZ91

hp), 53 MPa (AE42 hp, AlSi9Cu3) and 30 MPa (AS21 hp),

respectively. Table 1 shows the chemical compositions of

the investigated alloys, and Table 2 summarises their

static strength properties at ambient and elevated

temperatures [29].

The materials are cast in bars with stepwise varying

thickness. Specimens used in constant amplitude fatigue

lifetime investigations (S–N tests) are shown in Fig. 1(a).

Ce Fe Cu Ni Mg

1.0 2.0–3.5 0.1–0.5

0.003 0.001 0.001 Rem

!0.01 !0.002 !0.001 Rem

0.94 0.0003 0.0003 0.001 Rem

Table 2

Ultimate tensile strength (Rm), yield stress (Rp0.2) and fracture strain (A5) at ambient and elevated temperatures

Rm (MPa) Rp0.2 (MPa) A5 (%)

AlSi9Cu3 216 (20 8C), 198 (150 8C) 134 (20 8C), 123 (150 8C) 1.0 (20 8C), 1.1 (150 8C)

AZ91 hp 190 (20 8C), 165 (125 8C) 118 (20 8C), 93 (125 8C) 3.6 (20 8C), 12.7 (125 8C)

AS21 hp 131 (20 8C), 103 (125 8C) 84 (20 8C), 65 (125 8C) 2.1 (20 8C), 4.2 (125 8C)

AE42 hp 184 (20 8C), 146 (125 8C) 101 (20 8C), 82 (125 8C) 5.0 (20 8C), 18.8 (125 8C)

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881078

These specimens are obtained from the bar at a thickness of

7.5 mm and 1.25 mm was removed from both sides to obtain

the final specimen thickness of 5 mm. Edges in the centres

of the specimens were rounded (radius 0.5 mm) and the

surfaces were polished parallel to the specimen axis prior to

the measurements with abrasive paper of grade 1000.

Fatigue crack growth tests were performed using single

edge notched specimens as shown in Fig. 1(b). These

specimens were obtained from the bar at a thickness of

15 mm. A notch was introduced by spark erosion and served

to obtain a defined place for crack initiation.

2.2. Fatigue testing method

The ultrasonic fatigue testing method has been used to

perform fatigue lifetime and fracture mechanics

investigations at load ratio RZK1. Specimens are excited

to resonance vibrations at a frequency of approximately

20 kHz, which leads to maximum vibration amplitudes at

both specimen’s ends and maximum strain amplitudes in

54

14

108

R 60

7

5

43

15

86

15

R 60

3

Notch1

(a)

(b)

Fig. 1. Specimen shapes used in (a) S–N tes

the specimen’s centre, i.e. at the place of the vibration node.

Strain gauges are used to measure the cyclic strain and to

calibrate the experiments. At one specimen’s end a vibration

gauge serves to measure the vibration (displacement)

amplitude, which is used to control the experiment. The

displacement amplitude of the specimen is kept constant at

the pre-selected magnitude with high accuracy (typically

within G1%). Specimens are not cycled continuously but

in a sequence of pulses of 25 ms each (approximately

500 cycles) and pauses of adequate length (25–250 ms),

since continuous cycling would increase specimen

temperature due to internal friction. Stress amplitudes are

calculated using the measured strain amplitudes and the

Youngs’ moduli of the materials, i.e. 45 GPa at 20 8C and

41 GPa at 125 8C for the magnesium alloys [29] and 77 GPa

at 20 8C and 71 GPa at 150 8C for AlSi9Cu3.

S–N experiments at elevated temperatures are per-

formed, in principle, in a similar manner to the

experiments in ambient air described previously [19].

The main experimental difference is that heated air

14

15

15

ts and (b) fatigue crack growth tests.

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1079

provided by a hot air fan flows against the specimens.

Specimen surface temperatures are controlled either with

thermocouples or with temperature-dependent current

sources and are kept constant at 150 8C (aluminium

alloy) and 125 8C (magnesium alloys), respectively.

Previously obtained S–N data of the alloys measured in

ambient air of 18–22 8C and 40–60% relative humidity

are shown for comparison.

Specimens were loaded until failure or at least

109 cycles, if they did not fail (runnouts). At high cyclic

stresses, where all tested specimens failed, power law

functions of cyclic stress, s and cycles to failure, N are used

to approximate data (Eq. (1)):

skN Z const (1)

A log normal distribution of cycles to failure is assumed

to calculate 50% probability for failure. To evaluate the

endurance limit at 109 cycles, a procedure recommended by

Maenning [30] is used. If i is the number of specimens failed

below 109 cycles and n is the number of specimens tested at

this stress level, the probability for failure, PF(s) is

calculated using Eq. (2):

PFðsÞ Zð3i K1Þ

3n C1

�������� (2)

The Probit function serves to approximate the probabil-

ities for failure. The mean endurance limits at 109 cycles

(50% fracture probability below 109 cycles) and their

standard deviations are evaluated. Twenty-five specimens

or more are used in each testing series to obtain statistically

meaningful data.

Fatigue crack growth at elevated temperatures is studied

by introducing the single edge notched specimens in warm

airflow. Crack growth rates, Da/DN at certain cyclic stress

intensity amplitude, Kmax are measured by decreasing and

increasing the stress intensity in 5–7% steps in successive

experiments. Crack growth rates are determined as mean

growth rates over crack length increments of 0.15–0.20 mm.

Video equipment and magnification lenses serve to

determine crack length on the specimen’s surface with a

resolution of approximately 7 mm. If no crack growth

is detected, cycling the specimen with at least 2!107 cycles serves to determine threshold stress intensity.

With the optical resolution, the maximum growth rate of

3.5!10K13 m/cycle is used to characterise threshold.

Previously performed fatigue crack growth experiments in

ambient air (18–22 8C, 40–60% relative humidity) and in a

vacuum (pressure ranges from 10K3 to 5!10K3 Pa) are

included for comparison [31]. In ambient air, cycling the

specimen for 2!107 cycles served to determine threshold

stress intensity, similar to the experiments in warm air. In

vacuum, the threshold is determined loading the specimens

with 108 cycles to account for the lower optical resolution

using a vacuum chamber [31]. Details of ultrasonic fatigue

and fatigue crack growth experiments, determination of

cyclic stress intensity and the evaluation of data are

described elsewhere [32].

2.3. Evaluation of porosity

In the investigated alloys, nearly all fatigue cracks

initiate at porosity (voids and shrinkage) [19]. To determine

the size of material affected by porosity at the crack

initiating place, fracture surfaces were examined using

scanning electron microscopy (SEM). The projected area of

the defect in specimen’s length direction (i.e. the direction

of principal stress), termed ‘defect area’ is evaluated. Defect

areas are evaluated for two purposes. It may be assumed that

fatigue cracks initiate at one of the largest material defects

in the stressed volumes of the specimen. Thus, the

distribution of defect areas reflects the distribution of the

largest material defects of the alloy, which characterises

the quality of the cast and is most important for the fatigue

properties of the material.

Considering porosity as an initial crack, fracture surface

studies are used for fracture mechanics considerations.

Murakami et al. [33] have shown that the stress intensity

factor of a flaw is influenced by its shape by less than 10%

and may be calculated considering solely the applied stress,

the projected area of the flaw and its location. Assuming the

area affected by porosity (Defect Area) as an initial crack,

the stress intensity amplitude, Kmax is calculated for RZK1

loading with Eq. (3)

Kmax Z smaxa

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffip

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiDefect Area

pq

(3)

where smax is the cyclic stress amplitude, aZ0.65 for

surface defects (and for defects close to the surface [19]) and

aZ0.50, if the crack originates from internal porosity. If

specimens failed, the stress intensity was sufficient to

prolong the fatigue crack from the initial defect to rupture,

whereas runnouts either contained too small porosity or

were stressed at too low stress amplitudes. To include

runnouts in the analysis, these specimens were fractured at

higher stress amplitudes after low amplitude cycling.

Fatigue data produced with runnouts at the high stress are

not included in the diagrams or in the evaluations of the

endurance limits due to possible coaxing effects, and high

stress cycling of runnouts solely served to produce the

fracture surfaces.

Calculating the stress intensity factors for failed speci-

mens and runnouts serves to quantify the lowest stress

intensity factor necessary for fracture and the highest stress

intensity where no failure occurred. The range between

these two values is termed critical stress intensity amplitude,

Kcrit [19] and characterises the range where specimens may

fail whereas others do not. The surfaces of some AlSi9Cu3

and AZ91 hp specimens which were cycled in warm air and

did not fail within 109 cycles were examined in the SEM

using a rotating stage. This served to investigate non-

propagating cracks in runnout specimens.

104 105 106 107 108 10940

50

60

80

100

Ambient Air, 20 ˚CWarm Air, 150 ˚C

Number of Cycles

Stre

ss A

mpl

itude

(M

Pa)

150(a)

120

104 105 106 107 108 109

20

30

40

50

60

80

100

Ambient Air, 20 ˚CWarm Air, 125 ˚C

Number of Cycles

Stre

ss A

mpl

itude

(M

Pa)

(b)

104 105 106 107 108 109

20

30

40

50

60

80

100

Ambient Air, 20 ˚CWarm Air, 125 ˚C

Number of Cycles

Stre

ss A

mpl

itude

(M

Pa)

(c) (d)

104 105 106 107 108 109

20

30

40

50

60

80

100

Ambient Air, 20 ˚CWarm Air, 125 ˚C

Number of Cycles

Stre

ss A

mpl

itude

(M

Pa)

Fig. 2. S–N data of (a) AlSi9Cu3, (b) AZ91 hp, (c) AS21 hp and (d) AE42 hp at ambient (open circles, dashed lines) and elevated temperatures (closed circles,

solid lines).

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881080

3. Results

Table 3

Mean endurance limits (50% fracture probability at 109 cycles) and

standard deviation in warm air (150 8C for AlSi9Cu3, 125 8C for AZ91 hp,

AS21 hp and AE42 hp) and in ambient air (20 8C)

Mean endurance limitG

standard deviation, warm

air (MPa)

Mean endurance limitG

standard deviation, ambient

air (MPa)

AlSi9Cu3 61G12 75G14

AZ91 hp 41G6 45G7

AS21 hp 27G5 38G8

AE42 hp 35G5 42G9

3.1. S–N data

The results of constant amplitude lifetime investigations

are presented in Fig. 2. Fatigue data measured at elevated

temperatures are shown with closed circles, and previously

measured data at ambient temperature [19] are included with

open circles. Specimens which did not fail within minimum

109 cycles (runnouts) are marked with arrows. Two straight

lines in the double logarithmic diagrams are used to

characterise 50% fracture probability at elevated tempera-

tures (solid lines) and at ambient temperature (dashed lines).

Below 106–107 cycles, lines with constant slope indicate

increasing mean lifetimes with decreasing cyclic stresses.

The exponents k (Eq. (1)) are in the range from 6 to 8 for all

four materials at both temperatures. Above approximately

2!107 cycles, failures are rare, and data are approximated

with lines parallel to the abscissa. No AlSi9Cu3 specimen

(Fig. 2(a)), one AZ91 hp specimen cycled in ambient

air (Fig. 2(b)), two AS21 hp specimens cycled in warm air

(Fig. 2(c)) and one AE42 hp specimen cycled in warm

air (Fig. 2(d)) failed at numbers of cycles greater than

2!107.

Table 3 summarises the mean endurance limits and their

standard deviation in warm and ambient air. Mean

endurance limits decrease as the testing temperature

increases. In AlSi9Cu3, the mean endurance limit at

150 8C is approximately 80% of the cyclic stress measured

at 20 8C. The mean endurance limit of the magnesium alloys

at 125 8C is approximately 90% (AZ91 hp), 70% (AS21 hp)

and 85% (AE42 hp) of the respective stresses found in

ambient air. Standard deviations of the mean endurance

80

100

AlSi9Cu3

AZ91 hp

)

(a)

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1081

limits of all materials are about 15–20% of the respective

mean values in both environments. Specimens may fail in

ambient air at stresses where others survived in warm air.

The pronounced scatter in the cyclic properties is caused by

material defects.

0

20

40

60

< 0.5 0.5-1.0 1.0-1.5 1.5-2.0 2.0-2.5

AS21 hp

AE42 hp

Rel

ativ

e Fr

eque

ncy

(%

Defect Area (mm2)

Equivalent Diameter (mm)

< 0.8 0.8-1.1 1.1-1.4 1.4-1.6 1.6-1.8

AlSi9Cu3AZ91 hpAS21 hpAE42 hp

Prob

abili

ty o

f O

ccur

renc

e

0.01

0.05

0.20

0.50

0.80

0.95

0.99

(b)

3.2. Fatigue crack initiation

Porosity plays a most important role for crack initiation

and fatigue performance in the investigated alloys. Fatigue

cracks in 105 of the 107 specimens fractured in warm air

initiated at this kind of material defect. Large areas with

porosity may lead to drastically reduced lifetimes. In

AlSi9Cu3, one specimen cycled at 85 MPa failed after

1.0!104 cycles, whereas the other specimens stressed at

this amplitude survived at least 5.7!104 cycles (Fig. 2(a)).

Fatigue crack initiation area of this specimen is shown in

Fig. 3. Close to the surface, voids created by gas porosity

surrounded by dendrites and shrinkage porosity are visible.

The defect area of this specimen is approximately 2.5 mm2,

which is the largest defect area found on fracture surfaces of

all specimens. The damaging influence of this defect on the

fatigue behaviour is caused by its large size and the location

close to the specimen surface.

Defect areas at the crack initiation places of all specimens

have been evaluated, and the size distributions are shown in

Fig. 4. In Fig. 4(a), defect areas are classified in a histogram

and Fig. 4(b) shows a probability diagram. The defect areas

found on fracture surfaces of specimens ruptured in ambient

air [19] are included to enlarge the statistical basis. The

arithmetical mean and the maximum defect areas and

the equivalent diameters, i.e. the diameter of a circle with the

same area, are shown in Table 4. Defect areas range from

0.017 to 2.5 mm2, and defects greater than 1 mm2 could be

detected in each material. AE42 hp contains, on average, the

smallest defects. Defect areas in AS21 hp are, on average,

approximately a factor of 1.7 larger than in AE42 hp.

Fig. 3. Porosity visible on the fracture surface of AlSi9Cu3 specimen

which was cycled at 85 MPa in warm air of 150 8C and failed after

1.0!104 cycles.

Defect Area (mm2)

0.050.02 0.50.20.1 1 2

Equivalent Diameter (mm)

0.250.16 0.80.50.35 1.1 1.6

Fig. 4. Size distribution of defect areas at the crack initiation location and

equivalent diameters presented (a) in a histogram and (b) in a probability

plot. In (b), the abscissa indicates the size of defect areas and the ordinate

shows the probability that specimens contain porosity larger than this size.

The sizes of large defects in AZ91 hp are comparable to

large defects in AS21 hp, whereas the mean size is smaller.

Mean and maximum sizes of defect areas in AlSi9Cu3 are

larger than in the magnesium alloys.

Surface studies of runnout specimens in the SEM

served to find fatigue cracks, which did not grow until

fracture within 109 cycles (non-propagating cracks). The

upper parts of Figs. 5 and 6 show non-propagating cracks

Table 4

Arithmetical mean and maximum size of defect areas and respective

equivalent diameters at the crack initiation sites

Mean

defect

area

(mm2)

Equivalent

diameter of the

mean defect

area (mm)

Largest

defect

area

(mm2)

Equivalent

diameter of the

largest defect

area (mm)

AlSi9Cu3 0.63 0.89 2.50 1.78

AZ91 hp 0.30 0.61 1.43 1.35

AS21 hp 0.42 0.74 1.47 1.37

AE42 hp 0.25 0.56 1.42 1.34

Fig. 6. Non-propagating fatigue crack in AZ91 hp: the fatigue crack (upper

figure) was visible on the specimen’s surface after cycling at 38 MPa for

1.51!109 cycles. Then the specimen was ruptured at 64 MPa in 1.30!

105 cycles to produce the fracture surface (lower figure). Fatigue crack and

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881082

found on the surface of an AlSi9Cu3 specimen (Fig. 5)

and on the surface of an AZ91 hp specimen (Fig. 6),

respectively. The AlSi9Cu3 specimen was loaded with

1.05!109 cycles at 47.5 MPa, which is 78% of the mean

endurance limit of this material in warm air. The AZ91 hp

specimen was cycled at 38 MPa (90% of the mean

endurance limit) for 1.51!109 cycles. Several SEM

pictures of the specimens’ surfaces at different magnifi-

cations were taken to exactly locate the indicated crack

tips. Afterwards, the specimens were removed from the

SEM and fractured at higher cyclic stresses. This served

to produce the fracture surfaces, which are shown at the

same magnification in the lower part of Figs. 5 and 6. Gas

porosity is the origin of both fatigue cracks. Crack length

on the surfaces are 1.0 mm (AlSi9Cu3) and 1.6 mm

Fig. 5. Non-propagating fatigue crack in AlSi9Cu3: the fatigue crack (upper

figure) was visible on the specimen’s surface after cycling at 47.5 MPa for

1.05!109 cycles. Then the specimen was ruptured at 85 MPa in 1.15!105 cycles to produce the fracture surface (lower figure). Fatigue crack and

fracture surface are at the same magnification, indicated on the upper figure.

fracture surface are at the same magnification, indicated on the upper figure.

(AZ91 hp). These crack lengths are considerably larger

than the respective sizes of the crack initiating porosity.

The longest cracks observed on surfaces of runnout

AlSi9Cu3 specimens were 1.2 mm and were found on two

specimens cycled at 47.5 and 57 MPa, respectively. The

longest surface cracks of runnout AZ91 hp specimens

were 1.6 mm (Fig. 6) and 1.2 mm after cycling at 38 MPa.

3.3. Critical stress intensity amplitude

Assuming the area affected by porosity as the initial crack

and using Eq. (3), the stress intensity amplitudes of all

specimens were evaluated. Runnout specimens were frac-

tured at high stresses to locate defects, which were too small

to cause failure at the low loads. Stress intensity amplitudes

of fractured specimens and runnouts are shown in Fig. 7 with

closed and open circles, respectively. Shaded areas indicate

the ranges of the critical stress intensity factors, i.e. stress

intensity amplitudes, which caused failure of some speci-

mens and no failure of others. Table 5 summarizes the

critical stress intensity factors at elevated temperature.

Additionally, previously determined critical stress intensity

factors in ambient air are shown. Mean values of the critical

stress intensity factors in warm air are approximately 90%

(AlSi9Cu3, AZ91 hp) or 80% (AS21 hp, AE42 hp) of the

respective values found in ambient air. The ranges of

0.5 0.7 1 2 3 4 5

Kmax (MPam1/2)

No Failure

AlSi9Cu3

Failure

0.5 0.7 1 2 3

No Failure

AZ91 hp

Failure

Kmax (MPam1/2)

0.3 0.5 0.7 1 2 3

No Failure

AS21 hp

Failure

Kmax (MPam1/2)

0.3

0.3

0.5 0.7 1 2 3

AE42 hp

No Failure Failure

Kmax (MPam1/2)

Fig. 7. Stress intensities of broken specimens (solid circles) and runnouts (open circles) assuming the defect area as crack. The shaded region indicates the

critical stress intensity range where failures as well as runnouts occurred.

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1083

the critical stress intensity amplitudes in warm air are

typically twice as large as in ambient air.

The fraction of fatigue cracks initiating at material

defects at the surface is also shown in Table 5. In AlSi9Cu3,

about 1/3 of the fatigue cracks initiated at surface porosity

and 2/3 at internal defects in warm as well as ambient air.

Crack initiation in the magnesium alloys in ambient air is

preferentially at the surface, and 10% or less initiate from

the interior. In warm air, this preference for surface crack

initiation is less pronounced, and fatal fatigue cracks in

approximately 25% (AZ91 hp, AE42 hp) or more (AS21 hp)

of the fractured specimens initiated at internal porosity.

3.4. Fatigue crack growth and threshold data

In Fig. 8, the results of fatigue crack growth tests in warm

air of 150 8C (AlSi9Cu3, Fig. 8(a)) and 125 8C (AZ91 hp,

Fig. 8(b) and AS21 hp, Fig. 8(c)) are shown. For

comparison, previously measured data in ambient air

(open circles) and in a vacuum (triangles) are included

[31]. Arrows indicate that no crack growth was observed

within the optical resolution of crack length measurement

cycling the specimens for 2!107 cycles (warm air and

ambient air) or 108 cycles (vacuum). These measurements

indicate threshold stress intensities with limiting growth

Table 5

Critical stress intensity amplitudes and frequency of crack initiation at the surfac

Critical stress intensity

amplitude in warm air,

Kcrit (MPam1/2)

Fraction of cracks

starting at the surfa

warm air

AlSi9Cu3 1.4–1.8 0.37

AZ91 hp 0.85–1.05 0.72

AS21 hp 0.55–0.75 0.56

AE42 hp 0.7–0.9 0.76

rates of 3.5!10K13 m/cycle. Table 6 summarises the

threshold stress intensity amplitudes at load ratio RZK1

in the three environments.

Increasing testing temperature leads to increased fatigue

crack propagation rates and reduced threshold stress

intensities in all three materials. Compared with ambient

air, crack growth data measured in warm air are shifted

towards lower stress intensity amplitudes. Threshold stress

intensities at elevated temperatures are approximately 80%

of the respective values found in ambient air. Minimum

growth rates of propagating cracks in warm and ambient air

are about 10K11 m/cycle (AlSi9Cu3), 10K10 m/cycle (AZ91

hp) and 3!10K11 m/cycle (AS21 hp) and lower growth

rates result from experiments, in which a crack propagated

first and stopped afterwards, although stress intensity

amplitude was kept constant. In vacuum, threshold

behaviour in the magnesium alloys is not so distinct as in

ambient or warm air, and cycling close to the threshold

could produce mean fatigue crack propagation rates in the

regime of 10K12–10K11 m/cycle.

Fatigue fracture surface of AS21 hp after near threshold

cycling at 20 and 125 8C is shown in Fig. 9. The specimen

was initially cycled in ambient air at constant stress intensity

amplitude of 1.6 MPam1/2. Then the experiment continued

in warm air at the same stress intensity. Fig. 9 shows the area

e in warm air and in ambient air, respectively

ce,

Critical stress intensity

amplitude in ambient air,

Kcrit (MPam1/2)

Fraction of cracks

starting at the surface,

ambient air

1.75–1.95 0.33

1.0–1.1 0.92

0.7–0.8 0.96

0.9–1.1 0.90

10-13

10-12

10-11

10-10

10-9

10-8

10-7

1 2 3 4 5 6 7 8

Fatig

ue C

rack

Gro

wth

Rat

e (m

/ C

ycle

)

Kmax (MPam1/2)

10-13

10-12

10-11

10-10

10-9

10-8

10-7

0.8 1 2 3 4 5 6

Fatig

ue C

rack

Gro

wth

Rat

e (m

/ C

ycle

)

Kmax (MPam1/2)

10-13

10-12

10-11

10-10

10-9

10-8

10-7

0.8 1 2 3 4 5 6

Fatig

ue C

rack

Gro

wth

Rat

e (m

/ C

ycle

)

Kmax (MPam1/2)

(a) (b)

(c)

Fig. 8. Fatigue crack growth data of (a) AlSi9Cu3, (b) AZ91 hp and (c) AS21 hp in warm air (closed circles), ambient air (open circles) and in vacuum

(triangles).

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881084

on the fracture surface, where the environment has been

changed from low temperature (left side of the figure) to

elevated temperature (right side of the figure). Crack growth

is transcrystalline, relatively flat and shows no apparent

differences after cycling in both environments. The fracture

surfaces of the other testing materials were also similar after

cycling in ambient and warm air. Shrinkage porosity and

dendrites are visible in the upper section of Fig. 9. Porosity

is visible on several areas of the fracture surfaces and is

probably the main reason for the pronounced scatter of

fatigue crack growth data.

Table 6

Threshold stress intensities at RZK1 determined in warm air (150 8C for AlSi9C

Threshold stress intensity amplitude

in warm air, Kmax,th (MPam1/2)

Threshold s

in ambient a

AlSi9Cu3 1.95–2.2 2.45–2.7

AZ91 hp 1.05–1.2 1.30–1.55

AS21 hp 0.95–1.15 1.25–1.45

4. Discussion

Increasing the testing temperature deteriorated the

fatigue properties of high-pressure die-cast magnesium

alloys. S–N fatigue tests in warm airflow (125 8C) delivered

endurance limits at 109 cycles, which were 70–90% of

the respective stresses found in ambient air (20 8C), and

threshold stress intensities decreased to approximately 80%

of the values found at the low temperature. Loss of cyclic

strengths is within ranges comparable to the decrease of

static strengths, since tensile strengths at elevated

u3, 125 8C for AZ91 hp and AS21 hp), ambient air (20 8C) and in vacuum

tress intensity amplitude

ir, Kmax,th (MPam1/2)

Threshold stress intensity amplitude

in vacuum, Kmax,th (MPam1/2)

3.3–3.4

2.1–2.3

2.3–2.7

Fig. 9. Fatigue fracture surface of AS21 hp produced in ambient and warm

air. The picture shows the area where the temperature was increased from

20 8C (left side of the figure) to 125 8C (right side of the figure). Crack

growth direction is from left to right.

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1085

temperature are 80–85% and yield strengths are about 80%

of the respective stresses found at ambient temperatures.

Within the investigated magnesium alloys, AZ91 hp showed

better cyclic properties in warm air than AE42 hp, and AS21

hp exhibited the lowest cyclic strength, which is the same

ranking as in ambient air environment.

AlSi9Cu3 is a typical aluminium alloy used in high-

pressure die-casting, and cyclic properties of the magnesium

alloys may be compared with this material. Testing the

aluminium alloy at 150 8C, fatigue limit and threshold stress

intensity were about 80% of the respective values

determined in ambient air. Deterioration of fatigue proper-

ties of the magnesium alloys by increasing the temperature

to 125 8C is comparable to the loss of cyclic strength of the

aluminium alloy by increasing the temperature to 150 8C.

The mean fatigue limit of AZ91 hp at 125 8C is about 65%

of the stress amplitude determined for AlSi9Cu3 at 150 8C,

and threshold stress intensity is about 55% of the respective

value of the aluminium alloy. The fatigue limits of the

magnesium alloys at 20 and 125 8C are 40–45% of their

respective yield stresses and 25–30% of the tensile

strengths, whereas the fatigue limits of AlSi9Cu3 at low

and elevated temperatures are 50–55% of its yield strength

and 30–35% of the tensile strength.

Increase of temperature leads to greater cyclic plasticity

involved with fatigue cycling [11], greater accumulated

fatigue damage and thus lower cyclic strength of

magnesium alloys [34,35]. Several aspects of the fatigue

process at elevated and room temperature are, however,

comparable. Fatigue cracks start, nearly, exclusively (more

than 98%) at porosity in both environments. No obvious

temperature influence on crack path and fracture surface

appearance can be observed. Initial crack growth at porosity

and crack propagation at near threshold stress intensities is

mainly inside the (a-Mg) grains and tends to avoid

interdendritic regions [19,31], which is consistent with

the observations of Horstemeyer et al. [13] who found such

crack paths cycling AM60-B and AZ91-T4 [17] at low

stress intensity amplitudes. In contrast, at high cyclic

stresses [36], at high temperatures of 250 8C [34] or stress

intensities distinctly above threshold [13,17] cracks mainly

initiate and propagate in the interdendritic region. The

shapes of the S–N curves measured in warm air resemble the

respective curves determined in ambient air environment,

and increasing the testing temperature shifts the S–N curves

towards lower cyclic stresses. In both environments, an

endurance limit is found and cycles to failure above

2!107 cycles are rare.

Fatigue cracks were found on the surfaces of some

specimens, which were cycled with minimum 109 cycles

and did not fail. These cracks may be assumed as

non-propagating cracks, since possible mean growth

rates to reach 1 mm within 109 cycles would correspond

to 10K12 m/cycle, which is below technical significance, at

least in automotive applications. Comparable maximum

lengths of non-propagating cracks are observed by Caton

[24] who found crack length greater than 1 mm in 319 T7

aluminium specimens, which did not fail within 108 cycles.

In the magnesium alloy AZ91D, decreasing growth rates

with increasing crack length and non-propagating fatigue

crack are documented for cycling notched specimens at low

stress amplitudes [37]. Stop of crack growth of fatigue

cracks initiating at porosity may be explained by increasing

crack opening stress intensity with increasing crack length

[38], which eventually causes too low effective stress

intensity ranges to further propagate the crack.

The endurance limit behaviour of the investigated

materials is caused by porosity and the phenomenon of

non-propagating cracks and is not an inherent property of

the alloys. Testing AZ91 hp produced by gravity die-casting

[8] and Vacural die-casting [19] and evaluating solely

defect-free samples, considerably higher cyclic strength,

monotonic increase of lifetimes with decreasing cyclic

stresses and no endurance limit below 109 cycles was found.

Similarly, cast aluminium alloys frequently do not show a

fatigue limit [4]. Thus, the endurance limit stress of the

investigated alloys may be ascribed to the distribution of

porosity and the capability of the microstructure to stop

crack propagation. At low stress intensities, particles in the

interdendritic region and grain boundaries may serve as

barriers and decelerate fatigue crack propagation rates [17].

The critical stress intensity amplitude, Kcrit may be used

as measure for the sensitivity of the microstructure to

defects, i.e. the greater Kcrit, the higher cyclic stresses are

necessary to propagate a crack initiating at porosity to

rupture. Specimens can sustain cyclic stresses without

fatigue failure if the cyclic stress intensity determined for its

most damaging porosity is lower than Kcrit. Considering

the distribution of porosity and the preference for

crack initiation at the surface or in the interior, the

probability for failure at certain stress amplitude may be

linked to the distribution of porosity in a material [19].

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881086

Comparing the magnesium alloys, mean value of Kcrit of

AZ91 hp in warm air is about a factor 1.45 higher than Kcrit

of AS21 hp. The mean endurance limit of AZ91 hp is a

factor 1.52 higher than the mean endurance limit of AS21

hp. Thus, the lower fatigue strength of AS21 hp is mainly

caused by its greater defect sensitivity and to a far lesser

extent by the greater mean defect size. On the other hand,

mean defect size in AE42 hp is smaller than in AZ91 hp.

Mean value of Kcrit of AZ91 hp is, however, about a factor

1.2 higher than Kcrit of AE42 hp, and consequently its mean

fatigue limit is a factor 1.17 higher. Within comparable

microstructures, the growth rates of small fatigue cracks

increase as the yield strengths decrease [23]. This may be

the reason for the correlation of Kcrit with the yield strength

of the magnesium alloys, i.e. the higher the yield stress, the

higher the critical stress intensity. Kcrit is lower than the

threshold stress intensity of long cracks since short cracks

grow at cyclic stress intensities below the long crack

threshold [39]. Effective threshold stress intensities of AZ91

are found at 0.7 MPam1/2 [40,41] and 0.55 MPam1/2 [15],

and Kcrit is larger than this value.

The greatest effect of testing temperature on the fatigue

performance was found in endurance tests of AS21 hp,

where the mean endurance limit at 125 8C is 71% of the

stress found at 20 8C. A more pronounced influence of

testing environment is observed comparing crack propa-

gation properties in ambient air and vacuum. Threshold

stress intensity of AZ91 hp in vacuum is a factor 1.55 higher

than in ambient air, and a factor 1.85 higher threshold value

in vacuum is found for AS21 hp. The deleterious influence

of ambient air on fatigue damage in magnesium alloys is

caused by air humidity, which accelerates crack growth and

reduces the plasticity at the crack tip [42–45]. Chemical

processes on the newly created surfaces are especially

pronounced for near threshold cycling and slowly growing

fatigue cracks [31]. Fatigue cracks form more rapidly and

short cracks propagate faster if cycled in humid environ-

ment instead of in a vacuum [16]. In aluminium alloys, air

humidity has a similar deleterious effect on fatigue crack

growth [46–48]. However, compared with other aluminium

alloys [49,50], the influence of atmospheric moisture on

near threshold fatigue crack growth in AlSi9Cu3 is

relatively small, and threshold stress intensity in vacuum

is only a factor 1.3 higher than in ambient air.

In the absence of environmental effects, crack initiation

location would be expected at the defect with the greatest

value of a(Defect Area)1/4, according to Eq. (3). However,

fatigue cracks which initiate at surface defects grow under

the influence of atmospheric moisture, whereas initial crack

growth of internal cracks is in vacuum. In AlSi9Cu3, the

deleterious influence of atmospheric moisture is relatively

weak. Fatigue cracks preferentially initiate inside the

specimen due to the greater probability that large defects

are located in the interior of the stressed volume than at

the surface. In the magnesium alloys, the stronger

environmental effect causes preferential crack initiation at

surface defects. In ambient air, crack initiation is almost

exclusively (90% or more) at surface defects and the range

of the critical stress intensity is narrow (Table 5). At

elevated temperature, fatigue loading involves greater

cyclic plasticity, which makes initiation of fatal cracks at

(large) internal defects more probable. Minimum stress

intensities necessary to propagate small cracks to fracture at

125 8C are probably different in inert environment and air.

The greater variation of the crack initiation location is

probably one reason for the wider range of the critical stress

intensity at 125 compared with 20 8C.

5. Conclusions

The fatigue properties of the magnesium alloys AZ91 hp,

AS21 hp and AE42 hp and of the aluminium alloy AlSi9Cu3

produced by high-pressure die-casting have been investi-

gated in warm air. Experiments in ambient air served for

comparison, and the following conclusions may be drawn.

1.

Increasing testing temperature to 125 8C decreases the

endurance limits of the magnesium alloys at 109 cycles

to 70–90% of the stress amplitudes determined at room

temperature, and the threshold stress intensities are about

80% of the respective values at 20 8C. Loss of fatigue

strength of the aluminium alloy is in comparable ranges

by increasing the temperature to 150 8C. Shapes of the

S–N curves are similar at low and elevated temperatures

and the curves are shifted towards lower cyclic stresses.

Fatigue crack growth curves are shifted towards lower

cyclic stress intensities.

2.

The magnesium alloy AZ91 hp reaches about 2/3 of the

endurance limit stress of AlSi9Cu3 and threshold stress

intensities at low and elevated temperature are about

55% of the respective values of the aluminium alloy.

AZ91 hp shows better cyclic properties than AE42 hp at

room and elevated temperatures, and the lowest cyclic

strength was found for AS21 hp, which corresponds to

the ranking of the static strength properties of these

alloys.

3.

Fatigue crack initiation is, nearly, exclusively (greater

98%) at porosity in warm and ambient air and fatigue

crack propagation is mainly transgranular. Fatigue

cracks initiate at porosity even at stresses below the

endurance limit. The longest cracks found on surfaces of

specimens which survived 109 cycles were 1.0–1.6 mm.

The endurance limit is related to a non-propagating

condition of cracks rather than a non-initiating condition.

4.

Considering the area affected by porosity as initial crack,

failed and runnout specimens are used to determine the

critical stress intensity amplitude, Kcrit. Materials can

sustain cyclic stresses without fatigue failure if the cyclic

stress intensity determined for their most damaging

porosity is lower than Kcrit. The critical stress intensity

amplitude is smaller than the threshold stress intensity of

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–1088 1087

long cracks and larger than the effective threshold stress

intensity and can be used to quantify the sensitivity of a

material to defects.

Acknowledgements

Investigations were financed by the AUDI AG, Ingol-

stadt, which is gratefully acknowledged.

References

[1] Aghion E, Eliezer D, editors. Proceedings of the first Israeli

international conference on magnesium science and technology,

Dead Sea, Israel. Israel: Magnesium Research Institute; 1997.

[2] Mordike BL, Kainer KU, editors. Magnesium alloys and their

applications. Wolfsburg, Germany: Werkstoff-Info GmbH Frankfurt;

1998.

[3] Kainer KU, editor. Magnesium alloys and their applications.

Weinheim, Germany: Wiley–VCH Verlag GmbH; 2004.

[4] Henry SD, Davidson GM, Lampman SR, Reidenbach F, Boring RL,

Scott WW, editors. Fatigue data book: light structural alloys. New

Jersy: ASM International; 1995.

[5] Bhambri AK, Kattamis TZ. Cast microstructure and fatigue behavior

of a grain-refined Mg–Zn–Zr alloy. Met Trans 1971;2:1869–74.

[6] Chaudhuri SK, Nair K. Influence of microshrinkage on mechanical

properties of a magnesium–zinc–zirkonium–rare earth alloy. Mater

Sci Eng 1979;37:159–64.

[7] Donlon WT, Paige C, Morris CJ, Allison JE. The effects of casting

defects and microstructure on the mechanical properties of die cast

AM50 magnesium and 356 aluminum. Proceedings of the aluminum

and magnesium for automotive applications, Cleveland, Ohio, USA.

USA: Minerals, Metals and Materials Society/AIME; 1996, p. 17–27.

[8] Mayer H, Lipowsky H, Papakyriacou M, Rosch R, Stich A, Stanzl-

Tschegg S. Application of ultrasound for fatigue testing of lightweight

alloys. Fatigue Fract Eng Mater Struct 1999;22:591–9.

[9] Rodrigo D, Murray M, Mao H, Brevick J, Mobley C, Chandrasekar V,

et al. Effects of section size and microstructural features on the

mechanical properties of die cast AZ91D and AM60B magnesium

alloy test bars. SAE Trans: J Mater Manufact 1999;108:785–9.

[10] Liu Z, Wang Z, Wang Y, Liu Z. Cyclic deformation behaviour of high

pressure die casting alloy AM50. J Mater Sci Lett 1999;18:1567–9.

[11] Eisenmeier G, Holzwarth B, Hoppel HW, Mugrabi H. Cyclic

deformation and fatigue behaviour of the magnesium alloy AZ91.

Mater Sci Eng 2001;A319–321:578–82.

[12] Potzies C. PhD Dissertation: Zum Ermudungsverhalten der Mag-

nesium-Gusslegierung G-MgAl9Zn1 (AZ91). Berlin, TU Berlin:

Verkehrs- und Maschinensysteme; 2001.

[13] Horstemeyer MF, Yang N, Gall K, McDowell D, Fan J, Gullett P.

High cycle fatigue mechanisms in a cast AM60B magnesium alloy.

Fatigue Fract Eng Mater Struct 2002;25:1045–56.

[14] Renner F, Zenner H. Fatigue strength of die-cast magnesium

components. Fatigue Fract Eng Mater Struct 2002;25:1157–68.

[15] Sajuri ZB, Umehara T, Miyashita Y, Mutoh Y. Fatigue-life prediction

of magnesium alloys for structural applications. Adv Eng Mater 2003;

5(12):910–6.

[16] Gall K, Biallas G, Maier HJ, Gullet P, Horstemeyer MF,

McDowell DL, et al. In-situ observations of high cycle fatigue

mechanisms in cast AM60B magnesium in vacuum and water vapor

environments. Int J Fatigue 2004;26(1):59–70.

[17] Horstemeyer MF, Yang N, Gall K, McDowell DL, Fan J, Gullett PM.

High cycle fatigue of a die cast AZ91-T4 magnesium alloy. Acta

Mater 2004;52(5):1327–36.

[18] Mayer H, Papakyriacou M, Stanzl-Tschegg S, Tschegg E, Zettl B,

Lipowsky H, et al. Korrosionsermudung verschiedener Aluminium-

und Magnesium-Gublegierungen. Materials and Corrosion 1999;50:

81–9.

[19] Mayer H, Papakyriacou M, Zettl B, Stanzl-Tschegg SE. Influence of

porosity on the fatigue limit of die cast magnesium and aluminium

alloys. Int J Fatigue 2003;25(3):245–56.

[20] Couper MJ, Neeson AE, Griffith JR. Casting defects and the fatigue

behavior of an aluminium casting alloy. Fatigue Fract Eng Mater

Struct 1990;13(3):213–27.

[21] Ting JC, Lawrence FV. Modeling the long-life fatigue behaviour of a

cast aluminium alloy. Fatigue Fract Eng Mater Struct 1993;16(6):

631–47.

[22] Skallerud B, Iveland T, Harkegard G. Fatigue life assessment of

aluminium alloys with casting defects. Eng Fract Mech 1993;44(6):

857–74.

[23] Caton MJ, Jones JW, Allison JE. Use of small fatigue crack growth

analysis in predicting the S–N response of cast aluminium alloys. In:

Newman JC, Piascik RS, editors. Fatigue crack growth thresholds,

endurance limits, and design, ASTM STP 1372. West Conshohocken,

PA: ASTM; 2000, p. 285–303.

[24] Caton MJ. PhD Dissertation: predicting fatigue properties of cast

aluminium by characterizing small-crack propagation behavior. Ann

Arbor, MI: University of Michigan; 2001.

[25] Caton MJ, Jones JW, Mayer H, Stanzl-Tschegg S, Allison JE.

Demonstration of an endurance limit in cast 319 aluminum. Metall

Mater Trans A 2003;34A:33–41.

[26] Stanzl-Tschegg SE, Mayer HR, Tschegg EK, Beste A. In-service

loading of AlSi11 aluminium cast alloy in the very high cycle regime.

Int J Fatigue 1993;15(4):311–6.

[27] Dabayeh AA, Xu RX, DU BP, Topper TH. Fatigue of cast aluminium

alloy under constant and variable-amplitude loading. Int J Fatigue

1996;18(2):95–104.

[28] Schumann S, Friedrich F. The use of magnesium in cars—today and in

future. Proceedings of the magnesium alloys and their applications.

Wolfsburg, Germany: Werkstoff-Info GmbH Frankfurt; 1998, p. 3–13.

[29] Schindelbacher G, Rosch R. Mechanical properties of magnesium die

casting alloys at elevated temperatures and microstructure depen-

dence of wall thickness. In: Mordike BL, Kainer KU, editors.

Magnesium alloys and their application. Frankfurt: Werkstoff-

Informationsgesellschaft; 1998, p. 247–52.

[30] Maenning WW. Planning and evaluation of fatigue tests. In:

Lampman SR, Davidson GM, Reidenbach F, et al, editors. ASM

Handbook Fatigue and Fracture, vol. 19. Materials Park, OH: ASM

International; 1997. p. 303–13.

[31] Papakyriacou M, Mayer H, Fuchs U, Stanzl-Tschegg SE, Wei RP.

Influence of atmospheric moisture on slow fatigue crack growth at

ultrasonic frequency in aluminium and magnesium alloys. Fatigue

Fract Eng Mater Struct 2002;25(8/9):795–804.

[32] Mayer H. Fatigue crack growth and threshold measurements at very

high frequencies. Int Mater Rev 1999;44(1):1–36.

[33] Murakami Y, Endo M. The area parameter model for small defects

and nonmetallic inclusions in fatigue strength: experimental evi-

dences and applications. Proceedings of the theoretical concepts and

numerical analysis of fatigue, Birmingham, UK. Cradley Heath,

Warley, UK: Engineering Materials Advisory Services Ltd; 1992.

[34] May MJ, Honeycombe RWK. The effect of temperature on the fatigue

behaviour of magnesium and some magnesium alloys. J Inst Metals

1963;92:41–9.

[35] Ogarevic VV, Stephens RI. Fatigue of magnesium alloys. Annu Rev

Mater Sci 1990;20:141–77.

[36] Wang X, Fan J. SEM online investigation of fatigue crack initiation

and propagation in cast aluminium alloy. J Mater Sci 2004;39:

2617–20.

[37] Venkateswaran P, Raman SGS, Pathak SD, Miyashita Y, Mutoh Y.

Fatigue crack growth behaviour of a die-cast magnesium alloy

AZ91D. Mater Lett 2004;58:2525–9.

H. Mayer et al. / International Journal of Fatigue 27 (2005) 1076–10881088

[38] Ishihara S, McEvily AJ. Analysis of short fatigue crack growth in

aluminum alloys. Int J Fatigue 2002;24:1169–74.

[39] Pearson S. Initiation of fatigue cracks in commercial aluminium alloys

and the subsequent propagation of very short cracks. Eng Fract Mech

1975;7:235–47.

[40] Goodenberger DL, Stephens RI. Fatigue of AZ91E-T6 cast mag-

nesium alloy. J Eng Mater Technol 1993;115:391–7.

[41] Jung HC, Yim CD, Park WW, Lee CS, Shin KS. Fatigue crack

propagation behavior of AZ91D magnesium alloy. Mater Sci Forum

2003;419–422:75–80.

[42] Grinberg NM. The effect of vacuum on fatigue crack growth. Int

J Fatigue 1982;4(2):83–95.

[43] Serdyuk VA, Grindberg NM. The plastic zone and growth of fatigue

cracks in magnesium MA12 alloy at room and low temperatures. Int

J Fatigue 1983;5(2):79–85.

[44] Kobayashi Y, Shibusawa T, Ishikawa K. Environmental effect of

fatigue crack propagation of magnesium alloy. Mater Sci Eng 1997;

A234–236:220–2.

[45] Kobayashi Y, Takano T, Ishikawa K. Effect of oxygen and hydrogen

on fatigue crack propagation of AZ91 and AM60 alloys. Proceedings

of the seventh international fatigue conference, Bejing. Cradley

Heath, UK: EMAS; 1999.

[46] Bradshaw FJ, Wheeler C. The influence of gaseous environment and

fatigue frequency on the growth of fatigue cracks in some aluminium

alloys. Int J Fract Mech 1969;5(4):255–68.

[47] Wei RP. Some aspects of environment-enhanced fatigue-crack

growth. Eng Fract Mech 1970;1:633–51.

[48] Davidson DL, Lankford J. The effect of water vapor on fatigue crack

tip mechanics in 7075-T651 aluminium alloy. Fatigue Fract Eng

Mater Struct 1983;6(3):241–56.

[49] Holper B, Mayer H, Vasudevan AK, Stanzl-Tschegg SE. Near

threshold fatigue crack growth in aluminium alloys at low and

ultrasonic frequency: influences of specimen thickness, strain rate,

slip behaviour and air humidity. Int J Fatigue 2003;25(5):

397–411.

[50] Holper B, Mayer H, Vasudevan AK, Stanzl-Tschegg SE. Near

threshold fatigue crack growth at positive load ratio in aluminium

alloys at low and ultrasonic frequency: influences of strain

rate, slip behaviour and air humidity. Int J Fatigue 2004;26(1):

27–38.