Environmental fatigue in a 7000 series aluminium alloy

Corrosion Science, Vol. 38, No. 10, pp. 1815-1837, 1996 Copyright 0 1996 Elsevier Science Ltd

Printed in Great Britain. All rights reserved 0010-938X/96 $15.00+0.00

PII: S0010_938X(96)0009&0

ENVIRONMENTAL FATIGUE IN A 7000 SERIES ALUMINIUM ALLOY

J. RUIZ and M. ELICES*

Departamento de Ciencia de Materiales, Universidad Politecnica de Madrid, ETSI de Caminos, Canales, y Puertos. Ciudad Universitaria, s.n. 28040 Madrid, Spain

Abstract-The effect of water vapour pressure on fatigue crack growth in an aluminium alloy is analysed in this investigation. Fatigue tests were performed in gaseous atmospheres with different water vapour concentrations and the experimental results were correlated with the morphology of the fracture surfaces examined by SEM. An important effort has been made in thoroughly examining the characteristic features associated with the fatigue fracture for environments with different water vapour pressures. Crack growth response at the lower water vapour pressures (between 1 and 5 Pa) is satisfactorily explained in terms of a model based on the hypothesis of hydrogen embrittlement. Copyright 0 1996 Elsevier Science Ltd

Keywords: A. aluminium, B. SEM, C. corrosion fatigue, hydrogen embrittlement.

INTRODUCTION

Aluminium alloys have been used successfully in a variety of applications due to their good mechanical properties and low density. The greatest potential for commercial exploitation is offered by the medium strength Al-Zn-Mg alloys (7xxx series), which are stronger than other weldable alloys. Commercial exploitation of these alloys, nevertheless, remains limited owing to fears of their suffering environment-sensitive fracture in service, particularly under cyclic loading.le3 Most of the published data on corrosion-fatigue for these alloys involve cracking in aqueous saline environments.2’4-6 However, some apparently innocuous environments, such as air, are also responsible for enhanced crack growth; vacuum.

F;mid air can increase crack velocity one order of magnitude with respect to high Tests in gaseous environments are scarce, possibly due to experimental

difficulties in monitoring high purity environments in a high vacuum chamber properly fitted to the testing machine.‘29’3

Significant progress has been made in recent years in studying corrosion-fatigue of 7xxx aluminium alloys in gaseous environments. Nevertheless, a detailed mechanistic understanding to improve the performance of these alloys under adverse environments needs further efforts. Experimental results favour a mechanism based on hydrogen embrittlement (HE) instead of anodic dissolution: Speidel’ argued that an HE mechanism must control the corrosion fatigue of 7xxx alloys in moist gaseous atmospheres. His reasoning is based on the observation that at sufficiently low humidity levels, (below 30% relative humidity) condensation does not occur at the propagating crack tip’,14 and therefore the possibility of anodic dissolution is eliminated. Other arguments are based on

Manuscript received 22 March 1996.

1815

1816 J. Ruiz and M. Elites

the striations of the fracture surfaces of corrosion-fatigue cracks in moist environments. Such striations suggest that the fracture process is discontinuous.‘5,‘6 Thus it appears possible that bursts of fracture occur each time the reaction between water and the aluminium alloy has generated enough hydrogen and this hydrogen has penetrated into the alloy.

The purpose of the present work is to characterize the corrosion-fatigue crack growth rates between 2 x 10V8 and 2 x 10e6 m/cycle in a 7017-T651 (Al-Zn-Mg) aluminium alloy in humid, high-purity, gaseous environments. Tests were designed in the hope of providing new information about the role of hydrogen and water vapour in the fatigue process and the access of environment to the crack tip. Two main results are provided: first, data of da/dl\ras a function of AKand a wide range of water vapour pressures (from lo-’ Pa to 1300 Pa), and as a second result, a detailed microscopic survey of fracture surfaces. Different types of striations were found and their presence was correlated with water vapour pressure. It is expected that this information will serve to improve our understanding of this involved problem.

EXPERIMENTAL METHOD

Materials In this research, 70 17 aluminium alloy in the T65 1 temper (peak strength) was employed.

It was supplied as a 30mm thick rolling plate. It is a medium strength precipitation hardenable alloy, with a chemical composition Al 5.0 Zn 2.4 Mg (the detailed composition is given in Table 1). The fabrication process produces a fairly anisotropic microstructure, as can be seen in Fig. 1, with the grains elongated in the rolling direction and flattened in the other two directions.

Experimental method Fatigue tests were performed in a servohydraulic machine in load control with constant

load amplitude, following the recommendations of the ASTM E647 standard, Test Method for Measurement of Fatigue Crack Growth Rates. Compact tension samples (CTS) with W= 50 mm and a thickness B= 5 mm machined in the L-T orientation (with the notch plane perpendicular to the rolling direction) were used (see Fig. 1). The mechanical properties of the material in this orientation are shown in Table 2 (each value represents the average of three tests). The experimental procedure involved pre-cracking of the samples in air at a frequency of 10 Hz and stress ratio R = 0.1, with a sinusoidal waveform. A manual load shedding technique was used to reduce the stress intensity factor amplitude, AK, during pre- cracking from AKz9.5 MPa,/m, to the initial value for the fatigue tests (AKz6.5 MPaJm). The last two steps were conducted with the specimen in the testing environment to avoid transient effects in the fatigue crack growth rates. Fatigue tests were carried out with the same waveform and stress ratio as were used during pre-cracking at a frequency of 5 Hz.

Table 1. Chemical composition of 7017-T651 aluminium alloy

Zn Mg Fe Si Mn Cr cu Zr Ti Ni Al

5.01 2.44 0.23 0.11 0.29 0.17 0.12 0.13 0.05 0.01 bal

Environmental fatigue in a 7000 series Al alloy 1817

Fig. 1, Orientation of the samples relative to the rolling direction and representative microstructure of 7017-T651 aluminium alloy.

Table 2. Mechanical properties of 7017-T651 aluminium alloy (L-T orientation)

E c70.2 Kc

(@a) P&W (4% (2, (MPaJm)

65.8 415 465 13.7 34.6

1818 .I. Ruiz and M. Elites

Crack length was measured on one specimen side with a travelling microscope (accuracy of 0.01 mm.). After breaking the sample, the curvature of the crack front was examined and corrected when necessary according to the recommendations of the standard. Crack growth rates were calculated by the secant method. The stress intensity factor is computed from Saxena and Hudak.”

Testing environments In the experiments, the specimens were inside an ultra-high-vacuum chamber,

specifically designed for fatigue testing in precisely controlled environmental conditions, which has been described elsewhere.‘**i3 The experimental arrangement allows fulfilment of the most demanding requirements regarding the purity and the control of the environment surrounding the sample. Tests were carried out to ascertain the influence of water vapour pressure on fatigue crack growth rate in a reference inert environment: high vacuum (pzz 10B5 Pa), and in high purity water vapour at different pressures below atmospheric pressure. In addition, some tests were carried out in atmospheric air (50% relative humidity) to obtain another reference curve.

The composition of the atmosphere surrounding the sample is analysed with a quadrupolar mass spectrometer placed in a secondary small chamber joined on to the main vacuum chamber. This device provides identification and individual quantification of the gases in the testing environment. It can measure variations in partial pressure lower than 10e8 Pa. The operation of this spectrometer and the water vapour purifying techniques are fully described elsewhere.13 As an example of the strict environmental control achieved, a representative spectrum of the high purity water vapour atmospheres used in the experiments is shown in Fig. 2. The plot (Fig. 2) shows a very clean spectrum, with the two peaks associated with water: one of HzO+ (m/e = 18) and the other of OH+ (m/e = 17) and a very small peak of nitrogen (m/e = 28). There is also a fairly high peak of hydrogen, typical of vacuum systems evacuated with turbomolecular pumps, as is the case, due to their low compression ratio for the lightest gases.”

Fractography Fracture surface morphology was analysed with a scanning electron microscope using

secondary and backscattered electrons to form the image. Secondary electron images show the topographical aspect of the fracture surface, whereas the backscattered electron images allow assessment of the existence of phases with different chemical composition. Chemical composition of the different phases was qualitatively analysed with energy dispersive X-ray microanalysis (EDX).

EXPERIMENTAL RESULTS

Environmental-fatigue tests In Fig. 3, crack growth rate, da/dN, is represented as a function of the stress intensity

factor amplitude, AK, at a frequency of 5 Hz for different values of the water vapour pressure, including the data corresponding to the reference environments: atmospheric air (50% relative humidity) and high vacuum @z 10e5 Pa). The data are fitted by means of power law functions (Y = A l .J?) and plotted in a log-log scale. Water vapour pressure in humid air can be calculated from the product of the saturation pressure (approximately 2600 Pa at room temperature) and the relative humidity, which gives a value of 1300 Pa in

Environmental fatigue in a 7000 series Al alloy

f 3 0,4 2 P

= 0,2

z * 0

H,O+

OH*

H’ 2

N2*

L -

L I I I I I

0 20 40 60 60 m/e

Fig. 2. Spectrum of a water vapour atmosphere (1 Pa).

I

AA 7017-T651

loa- WATER VAPOUR

v=5Hz

Humid air (50% Rh) .

. 1300 Pa (50% Rh air)

High vacuum

0 10 Pa 0 5 Pa 0 1 Pa . 10” Pa (vacuum)

6 7 8 9 10 20 25

AK (MPadm) Fig. 3. da/dN vs AK data for 7017-T651 aluminium alloy tested in environments with different

water vapour pressure (v = 5 Hz).

1820 J. Ruiz and M. Ekes

this case. As regards the high vacuum atmosphere, the spectra recorded with the quadrupolar mass spectrometer show the characteristic composition of the residual gas in a system evacuated with a clean turbomolecular pump. Water vapour is the most abundant species, as corresponds to an unbaked vacuum system where the inner walls slowly release the water vapour they had adsorbed when the system was at atmospheric pressure.” Consequently the water vapour pressure can be estimated as approximately 10e5 Pa.

The results shown in Fig. 3 indicate that crack growth rates in high vacuum are very similar to those observed in water vapour at the lowest vapour pressures (1 Pa). As the water vapour pressure is increased, crack velocity also increases until a value of approximately 5 Pa where, as described later, a saturation in the embrittlement process seems to occur. The results at a water vapour pressure of 10 Pa are almost identical to those found at 5 Pa. For AK%7 MPa Jm, where the sharpest differences in da/dN are encountered, the crack propagation rate at 5 Pa is twice the value at 1 Pa, and the fractographic results to be described later do indicate a clear difference in the micromechanisms for crack growth in both cases. Crack propagation rates in atmospheric air (p = 1300 Pa) are the highest, irrespective of the value of AK. Particularly at AK=7 MPaJm, crack velocity in atmospheric air is five times higher than in high vacuum and almost three times higher than in water vapour at 5 Pa. These results are consistent with crack velocities measured by others in a high-strength Al-Zn-Mg-Cu alloy (7075-T651)” tested in similar experimental conditions.

Fractographic analyses Detailed examinations were made of the fracture surface morphology of the samples

tested in different environments, with the aim of characterizing the surface appearance as a function of water vapour pressure. The correspondence between the transitions observed in crack propagation rate (macroscopic parameter) and the fracture surface morphology gives additional clues in elucidating the embrittlement mechanism. The results of the fractographic analysis in different testing environments are shown. In all micrographs, the crack propagates from bottom to top. The microfractographic features shown in the figures were observed uniformly over the entire fracture surface. No significant differences were found between near surface and interior regions of the sample, which suggests that the same micromechanisms are involved irrespective of the location of the surface region.

High vacuum (p E IO-’ Pa). For the lower values of AK (AKx 7 MPa Jm) the fracture surface has a regular and plain appearance. Most of the surface is covered with features containing structures perpendicularly oriented to the crack direction, which are periodically repeated and superimposed with a puff pastry-like appearance, as can be seen in Fig. 4. These features are characteristic of fatigue in high vacuum in 7xxx aluminium alloys and have been called coarse striations by other authors.i6 As measured from Fig. 4, striation spacing for AK%7 MPaJm is around 2 urn, which is much higher than the crack growth rate for this value of AK (approximately 0.01 nm/cycle). It suggests that if these microscopic features are correlated with crack propagation-which seems to be reasonable-in this case crack advance does not proceed cycle by cycle, but is an intermittent process.‘5*‘6

Regions are sometimes observed with markings parallel to the overall crack direction, generally at the confluence between features with coarse striations, as shown by the arrows in Figs 4a and 5. These markings, which have not been found in the literature, could be related to consecutive positions of the crack front in grains unfavourably oriented with

Environmental fatigue in a 7000 series AI alloy 1821

Fig. 4. Microfractographs in high vacuum at 10e5 Pa (AK%7 MPaJm and v= 5 Hz): (a) typical appearance of the fracture surface, (b) higher magnification.

respect to the average crack advance direction, because the spacing between successive markings is very similar to the coarse striations spacing.

When the backscattered electrons are employed to form the image (BEI), a number of particles with a bright contrast are found in the fracture surfaces. Energy Dispersive X-ray (EDX) analyses showed that these particles are rich in iron, manganese, silicon and aluminium. They are sometimes isolated and aligned in the rolling direction with a needle- like form, although generally they form agglomerates with a typical diameter or 20pm inside some deep holes in the fracture surfaces. These impurities cannot be detected with the secondary electrons image (SEI), as shown in Fig. 6. Both micrographs correspond to the same region of the fracture surface when secondary electrons (Fig. 6a) and backscattered electrons (Fig. 6b) are used to form the image. As the impurities are generally associated


Fig. 5. Microfractograph in high vacuum at lo-‘Pa (AKz7MPa,/m and v=5 Hz) showing

coarse striations perpendicular to the direction of crack growth, and parallel markings in the

direction of growth.

with the deep holes (typical diameter of 20-30 pm) found in the fracture surfaces, it suggests that they act as stress concentrators ahead of the crack tip. When the crack front encounters one of these impurities, the plastic deformation is steadily concentrated around it until a microfracture takes place leaving a deep hole behind the crack wake.

As the value of AK is increased, the fracture surface morphology does not change substantially. The coarse striations perpendicular to the crack direction remain as the most outstanding microscopic feature, although their appearance becomes less clearly defined. However, an interesting feature arises: the regions where the coarse striations appear are almost completely covered with small dimples (approximately 0.2pm diameter for AK% 11.5 MPaJm), as shown in Fig. 7. The size of the dimples suggests they might be associated with the dispersoids in the aluminium alloy matrix.

For the highest values of AK, the fractography becomes very complex and irregular. The fracture surfaces are characterized by constant changes in the crack front plane, deep holes, regions subjected to a large amount of plastic deformation and a number of impurities of the same composition as the ones described above. All these features correspond to the predominance of static modes of fracture at values of AK close to the fracture toughness of the material. In these conditions, the plastic zone size is large and the amount of plastic deformation involved in the process may hide any other micrographical feature typical of the environment. Although the coarse striations have disappeared, the dimples found at AK% 11.5 MPaJm are still present. As can be seen in Fig. 8, their morphology has not changed, although they are elongated in the crack growth direction and their typical diameter has increased to approximately 1 pm as a consequence of the greater plastic deformation.

Water vapour (p = I Pa). The most remarkable characteristic of the fractography in high purity water vapour at a pressure of 1 Pa is the similarity to the results found in high vacuum (see Fig. 9). For AK%7 MPa,/m coarse striations, perpendicular to the crack growth direction are clearly visible over the entire fracture surface, as corresponds to the fatigue process in inert environments. However, in the steeply inclined regions of the surface appear some distinct features which were never observed in high vacuum. They consist of shallow


Fig. 6. Microfractographs in high vacuum at 10m5 Pa (AK%7 MPaJm and V= 5 Hz) showing the impurities in the fracture surface: (a) secondary electron image; (b) backscattered electron image.

dimples and round particles (around 1 pm diameter) easily recognisable by their bright contrast in SE1 as shown in Fig. 9b. EDX analyses showed that these regions are rich in oxygen, and also contain the habitual proportions of aluminium, zinc and magnesium found in other regions of the fracture surfaces. They might be corrosion products, coming from the fretting of the fracture surfaces upon unloading in a fretting-fatigue process. Although the water vapour content may be not enough to change the global appearance of the fracture surfaces and significantly accelerate crack propagation rate, there exists a difference of five orders of magnitude in water vapour pressure as compared with the high vacuum atmosphere employed before.

If backscattered electrons (BEI) are selected to form the image in the SEM, a number of particles with a bright contrast are found on the fracture surfaces. When analysed by EDX


Fig. 7. Microfractograph in high vacuum at lo-’ Pa (AK% 11.5 MPaJm and v= 5 Hz) showing small dimples in the fracture surface.

they show the same composition as those encountered in high vacuum, and the same arrangement: they are aligned in the rolling direction and frequently they form large agglomerates inside some deep holes in the fracture surfaces.

As the value of AK is increased, the fracture morphology becomes more irregular, although it maintains the typical features previously described. The coarse striations are less defined and are mostly covered with shallow dimples like those found in vacuum at AK% 11.5 MPa,/m (see Fig. 10 and compare with Fig. 7). Even for the highest values of AK (close to 20 MPa,/m) fine striations typical of the fatigue process in atmospheric air are scarcely observed.

The similarity between the fractographic results in water vapour pressure at 1 Pa and high vacuum correlates very well with the macroscopic results of crack propagation rate; they are almost identical in both environments, as can be seen in Fig. 3. Based on this evidence it is reasonable to think that the same cracking micromechanisms operate in both environments.

Fig. 8. Microfractograph in high vacuum at lo-’ Pa (AK= 15.5 MPaJm and v = 5 Hz) showing holes in the fracture surface.

Environmental fatigue in a 7000 series Al alloy 182.5

Fig. 9. Microfractographs in water vapour at 1 Pa (AK% 7 MPaJm and v= 5 Hz): (a) typical appearance of the fracture surface, (b) higher magnification.

Water vapour (p = 5 Pa). When the water vapour pressure is increased up to 5 Pa, the fracture surface morphology changes completely. The characteristic features of the fatigue surfaces in vacuum vanish, and instead a fracture with a microscopically brittle appearance emerges, as shown in Fig. 11. For AK% 7 MPa,/m, the coarse striations, a characteristic feature of the fracture surfaces in water vapour at a pressure of 1 Pa, are replaced by flat facets frequently covered with fine fatigue striations-termed brittle striations in the literature 19-perpendicularly oriented to the crack advance direction. As computed from Fig. 11 b, striation spacing is about 0.4,~m for the brittle striations, whereas the crack propagation rate measured at AKw 7 MPa,/m is around 0.04 ,um/cycle. Consequently, if it is postulated that every striation is associated with a discrete advance of the crack front, the crack propagation does not proceed cycle by cycle in this case either, but is an intermittent process. Several fatigue cycles are needed for the plastic deformation to be accumulated in the crack tip region before a brittle crack extension event takes place.r6 From the values


Fig. 10. Microfractograph in water vapour at 1 Pa (AK= 11.5 MPaJm and Y= 5 Hz) showing coarse striations and small dimples in the fracture surface.

computed for the striation spacing and the crack growth rate it can be estimated that for AK%7 MPaJm in high vacuum at least 200 cycles would be necessary for every discrete advance of the crack, whereas in water vapour at 5 Pa, this number would be reduced to 10.

In the fracture surfaces other features can be observed that resemble some aspects of the fractography in water vapour at 1 Pa. In the steeply inclined regions which separate adjacent flat facets, shallow dimples and round particles similar to those found in water vapour at 1 Pa are clearly visible (see Fig. 1 lb). When analysed with EDX, these regions show a high oxygen content, as happened previously, and they are attributed to the effect of water vapour in the fretting-fatigue process of the fracture surfaces on unloading.

If the backscattered electrons are selected to form the image, bright impurity particles rich in iron, manganese, silicon and aluminium are also found, although at a glance it seems that their surface fraction is lower than in high vacuum.

For higher values of AK, the fracture surfaces still retain the same basic appearance, with featureless flat facets close to others covered with brittle striations, as shown in Fig. 12. When AK reaches a value of 15 MPa,/m, in addition to the brittle striations which were previously seen, some relatively featureless flat facets when observed at high magnification show very fine ductile striations” (see Fig. 13). As computed from Fig. 13, striation spacing is approximately 0.3 pm, whereas the crack growth rate ranges between 0.4 and 0.5 pm/cycle for this value of AK. These results support a cycle by cycle process for the advance of the crack in this case.

Air (50% relative humidity). For the lowest values of AK(AKx 7 MPaJm), the fracture surface morphology is very uniform and regular. As can be seen in Fig. 14a it is composed of small facets with a brittle microscopic appearance periodically repeated every 5 pm approximately. These features resemble the periodic structures found by other authors in a 7075-T651 aluminium alloy in air at AK%6 MPa,/m.16 In general the periodic structures do not show remarkable features, although some times very fine fatigue striations can be guessed. At AKx 7 MPa,/m these striations are very difficult to resolve in the SEM for this aluminium alloy. The size of these facets is considerably smaller than that of the flat facets which appeared in water vapour at 5 Pa (see Fig. 1 la). Consequently, a fracture surface with


Fig. 11. Microfractographs in water vapour at 5 Pa (AK% 7 MPa,/m and v = 5 Hz): (a) typical appearance of the fracture surface, (b) higher magnification.

a fragmentary aspect is produced, which seems to indicate a complex and irregular propagation process.

In the steeply inclined regions (see Fig. 14b) there are traces of oxygen on the fracture surfaces in the EDX spectra. Dimples and round particles similar to those found previously in water vapour are clearly seen in these regions. The proposed explanation is the same as before: they are produced in the presence of water vapour by fretting-fatigue between the fracture surfaces on unloading. A number of impurity particles rich in iron, manganese, silicon and aluminium are also present on the fracture surfaces.

As AK is increased, the global appearance of the fracture surfaces does not change abruptly. However, due to the larger plastic zone size it becomes more irregular, with frequent changes in the local crack plane and deep holes within which impurity particles are found, sometimes fractured by the local tension. Nevertheless, the most remarkable characteristics are the fine ductile striations which cover a great portion of the fracture


Fig. 12. Microfractograph in water vapour at 5 Pa (AK% 11.5 MPaJm and Y= 5 Hz) showing brittle striations in the fracture surface.

surfaces (see Fig. 15) and become easily resolvable from AK% 11.5 MPaJm. From Fig. 15, striation spacing can be estimated at around 0.2-0.3 pm, which correlates very well with the measured crack propagation rate of 0.3 pm/cycle at the same value of AK. Consequently, it is reasonable to assume that crack advance proceeds cycle by cycle in this case.

DISCUSSION

The experimental results obtained in this investigation show a clear relation between the fatigue crack propagation rate (macroscopic parameter) and the fracture surface morphology (microscopic parameter). To facilitate discussion, the influence of water vapour pressure on fatigue growth rates and on fractography are dealt with separately and a model is proposed to explain the relationships between the microscopic and macroscopic results.

Fig. 13. Microfractograph in water vapour at 5 Pa (AK% 15.5 MPa,/m and V= 5 Hz) showing ductile striations in the fracture surface.


Fig. 14. Microfractographs in 50% relative humidity air (AK= 7 MPa,/m and v= 5 Hz): (a) typical appearance of the fracture surface, (b) higher magnification.

The data on fatigue behaviour exhibit a strong dependence of crack propagation rates on pressure at the lower water vapour pressures tested. There seems to exist a threshold value for the pressure under which no noticeable embrittlement is produced, as crack growth rates coincide with the results in high vacuum. In the 7017-T651 aluminium alloy, the threshold water vapour pressure for a frequency of 5 Hz is approximately 1 Pa. If the pressure is increased, crack propagation rates are steadily accelerated until a saturation value is reached. Using the results obtained in the present investigation, an estimate of 5 Pa for the water vapour saturation pressure can be made. In the saturation region (between 5 and 100 Pa) the crack velocity remains almost constant. At higher vapour pressures (above 100 Pa) additional increases in crack propagation rate exceeding the saturation value are found (see Fig. 3).20

The morphology of the fracture surfaces depend both on the environment and on the stress intensity factor amplitude. For the lower values of AKthe plastic zone size is small and the amount of plastic deformation accumulated ahead of the crack tip is not enough to hide


Fig. 15. Microfractograph in 50% relative humidity air (AK% 11.5 MPaJm and Y = 5 Hz) showing well defined ductile striations in the fracture surface.

the characteristic features associated with the environment. However, for the highest values of AK, the plastic zone size is larger and the plastic deformation modifies the fatigue fracture morphology by inducing static modes of fracture such as void growth and coalescence, deep holes and continuous changes in the crack propagation plane. As regards the influence of the environment, a careful fractographic analysis shows that for the threshold value of the water vapour pressure (1 Pa in this case) the fracture surface morphology is very similar to that in high vacuum. Both exhibit the characteristic features of the fatigue process in inert environments: the coarse striations perpendicular to the crack growth direction which are superimposed on each other (see Figs 4a and 9a). When the water vapour pressure is increased up to 5 Pa, the fracture surface changes completely. The features associated with the fatigue in inert environments tend to vanish, and they are replaced by some features characteristic of the fatigue process in aggressive environments. The surface is composed of flat facets with a microscopically brittle appearance which are often covered with fine fatigue

Fig. 16. Microfractograph in water vapour at 10 Pa (AK% 11.5 MPaJm and v = 5 Hz) showing both types of striations ductile and brittle.


striations of the two kinds described in the literature: ductile and brittle.” Both types of striations correspond to the same phenomenon, as can be seen in Fig. 16, where facets with brittle striations are close to other facets with ductile striations. For water vapour pressures above the saturation value (i.e. above 100 Pa), the fracture surfaces are very similar to those in atmospheric air (1300 Pa). The morphology is plain and regular, and the surface is sprinkled with small facets frequently covered with very fine fatigue striations (see Fig. 17).

From the experimental results found in the 7017-T65 1 aluminium alloy, it seems proven that the water vapour contained in the environment is responsible for the embrittlement of the material when it is subjected to cyclic loads. Some investigators agree on an explanation based on the chemical reaction between the water vapour and the fresh fracture surfaces produced by fatigue in the crack tip. They attribute the embrittlement caused by the environment in aluminium alloys”” and intermetallic aluminium compounds** to the following reaction (1):

2Al+ 3H20 =+ A1203 + 6H+ + 6e- (1)

According to this sequence, the water vapour molecules react with aluminium and as a consequence atomic hydrogen is produced. Subsequently, hydrogen enters the material and causes the embrittlement by some process still unknown.

Some models have been proposed to explain environmental fatigue. Most of them are based on the proposition that the rate of fatigue crack growth in a deleterious environment (da/dN), is composed of the sum of three terms:

(2)

where (da/dh’), is the fatigue crack growth rate in an inert reference environment and represents the contribution of pure fatigue, (du/dN& represents the contribution from sustained-load crack growth, and (da/diV),.is the corrosion fatigue component, associated with the synergistic interaction between the mechanical fatigue process and the environmental attack. The corrosion fatigue component is the most difficult to work out.

Fig. 17. Microfractographs in water vapour at 100 Pa (AKx 7 MPa,/m and V= 5 Hz) showing the typical appearance of the fracture surface.


For the structural materials the contribution of (da/dN)sCC is negligible” and it will not be taken into account in the following discussion.

From the models proposed in the literature, the one postulated by Wei and co- workers22*23 effectively predicts the dependence of (da/dAQon the testing frequency and the water vapour pressure in gaseous environments. This is why it has been employed to analyse the results obtained in the present investigation.

According to the model, in aluminium alloys the environmental fatigue process is controlled by the transport of the aggressive species to the crack tip region, as the surface reaction (1) is very fast. With this hypothesis (transport-controlled process), the model postulates a linear dependence of the corrosion fatigue component, (da/dN),f, on the quotient between the water vapour pressure p. and the testing frequency,f22,23:

Fatigue crack growth data at three different AK levels (7, 11.5 and 15.5 MPaJm) have been used for applying the model to the results obtained in this investigation. A comparison of the model predictions with crack propagation rates on the 7017-T65 1 aluminium alloy as a function of water vapour pressure is shown in Fig. 18 for a frequency of 5 Hz.~’ Theoretical curves are matched to the experimental results and show remarkable agreement.

109- = 15.5 MPdm

1 10 100 1000

p, (water vapour pressure) (Pa) Fig. 18. Influence of water vapour pressure on the fatigue crack growth rate in 7017-T651

aluminium alloy at 5 Hz (solid lines represent the predictions of the model).


The linear relationship between crack growth rate and water vapour pressure (predicted by the model) under the saturation pressure is confirmed, and it can be estimated around 5 Pa in this case.

According to equation (3), the parameter which governs crack propagation rates in gaseous environments is the environmental exposure, defined as the quotient between water vapour pressure and frequency. If this statement is correct, the fracture surface morphology must depend also on this parameter.

In order to verify this affirmation two experiments were performed where the water vapour pressure was maintained constant meanwhile the frequency was changed during the test. For a water vapour pressure of 10 Pa, changing the frequency between 1 and 10 Hz does not have any effect on the crack propagation rate, as can be seen in Fig. 19. The values of the environmental exposure, between 10 Pa sand 1 Pa s, lie in the saturation region (see Fig. IQ, and fatigue crack growth data are very similar to those obtained at 10 Pa and 5 Hz (exposure of 2 Pa s). When the fracture surfaces are analysed, no remarkable differences are found between the two testing conditions either. However, for a water vapour pressure of 1 Pa, testing frequency has a great influence on fatigue growth rates. As shown in Fig. 20, if the frequency is lowered from 10 Hz to 1 Hz-the environmental exposure changes from 0.1 Pa s (threshold region) to 1 Pa s (saturation region)-the crack velocity is increased up to the values corresponding to 5 Pa and 5 Hz (exposure of 1 Pa s). Furthermore, the effect is

I

AA 7017-T651

IO” ’ I

6 7 6 9 10 AK (MPadm)

20 25

Fig. 19. du/dN vs AKdata for 7017-T65 1 aluminium alloy tested in water vapour at 10 Pa changing the frequency between 1 and 10 Hz.


AA 7017-T651 WATER VAPOUR

H,O (5 Pa)

Hz0 (1 Pa)

5Hz

p=l Pa

. 10 Hz

0 1l-k

6 7 6 9 10 AK (MPadm)

20 25

Fig. 20. da/dN vs AKdata for 7017-T651 aluminium alloy tested in water vapour at 1 Pa changing

the frequency between 1 and 10 Hz.

totally reversible, since if the frequency is again increased to 10 Hz, the crack propagation rate falls to the previous values. Again the fractographic morphology correlates very well with the macroscopic results of crack growth rate. In Fig. 21 the fracture surface appearance in water vapour at a pressure of 1 Pa when the frequency is changed from 10 Hz to 1 Hz can be observed. At 1 Pa and 10 Hz (exposure of 0.1 Pa s which corresponds to the threshold region) the fracture surface morphology is very similar to that found in water vapour at 1 Pa and 5 Hz (exposure of 0.2 Pa s), with the characteristic features of the fatigue process in inert environments (see Fig. 21a). When the frequency is reduced to 1 Hz in the same environment, the fracture surface appearance changes completely. In place of the features associated with the fatigue in inert environments, flat facets with a microscopically brittle aspect appear on the fracture surface (see Fig. 2 1 b). Often these facets are covered with fine fatigue striations characteristic of the fatigue process in aggressive environments. If the frequency is again increased, the features associated with aggressive environments disappear and the fracture surfaces recover a vacuum-like aspect.

The remarkable correlation between the experimental fatigue crack growth data and the fracture surface appearance, in addition to the good agreement between the experimental results and the semi-empirical model based on the hypothesis of hydrogen embrittlement, lead to a reliance on the proposed explanation of the embrittlement process. Moreover, the fact that a liquid electrolyte cannot exist in the crack tip-very unlikely at the water vapour


Fig. 21. Microfractographs in water vapour at 1 Pa (AK% 11.5 MPaJm) showing the differences in fracture surface morphology: (a) inert environment appearance (v= 10 Hz), (b) saturation water

vapour pressure appearance (v = 1 Hz).

pressures employed, which give rise to relative humidities between 0.04% and 7.7%‘,14- leaves hydrogen embrittlement as the only convincing explanation of the environmental fatigue process in this aluminium alloy in gaseous environments.

These experimental results obtained in the 7017-T651 aluminium alloy are wholly consistent with the behaviour observed by other investigators’,” on two aluminium alloys of different composition (2219-T851 and 7075-T651) tested in similar experimental conditions. In spite of the differences between the three alloys, the results exhibit the same tendencies. The chemical composition of the 2219 and the 7075 aluminium alloys are different. However, they share an alloying element: copper (6.3% in AA2219 and 1.6% in AA707524), whereas in the 7017 the copper content is far less (0.12%). When the crack propagation data in these materials, which present important differences in their microstructures, are globally analysed, the similitude in behaviour is surprising. It leads to


the conclusion that the micromechanisms of fatigue crack growth must be very much alike in the three aluminium alloys compared, and they may be common to other materials with different chemical composition.25

CONCLUSIONS

From this research it may be concluded that:

1. The species responsible for the embrittlement of the 70 17-T651 aluminium alloy when it is subjected to fatigue in gaseous environments is the water vapour contained in the environment.

2. Fatigue crack propagation rates in 7017-T651 aluminium alloy increase as the water vapour pressure of the environment is increased.

3. Fractographic results correlate very well with the macroscopic fatigue crack growth rates: at a frequency of 5 Hz, crack propagation rate in high vacuum and in water vapour at a pressure of 1 Pa are almost identical, and the fracture surfaces in both gaseous atmospheres show a very similar appearance, which is characteristic of the fatigue process in inert environments. However, when water vapour pressure reaches 5 Pa (the saturation value), crack velocity is twice that in high vacuum, and the features associated with fatigue in inert environments vanish without trace in the fracture surfaces; features characteristic of the fatigue in aggressive environments appear instead.

4. The data on crack growth response at the lower water vapour pressures (between 1 and 5 Pa) are satisfactorily explained in terms of a corrosion fatigue model based on the hypothesis of hydrogen embrittlement.

5. The model used to analyse the experimental data predicts a linear dependence of crack propagation rate on the environmental exposure-the quotient between the water vapour pressure and the frequency. The preliminary results obtained seem to agree with this prediction, although more research is needed to fully corroborate this aspect.

6. The behaviour of the 70 17-T65 1 aluminium alloy when subjected to fatigue cycling in gaseous environments is consistent with the results obtained in other aluminium alloys with different chemical composition (2219-T851 and 7075-T651). This similitude suggests that the micromechanisms responsible for fatigue crack growth must be very much alike in these alloys, despite their microstructural differences.

7. The remarkable correlation between the macroscopic fatigue results and the fractographic appearance, in addition to the good agreement between the experimental data and the model, lead to a reliance on the proposed explanation. The evidence points to hydrogen embrittlement as the mechanism responsible for the environmental fatigue process in 7017-T651 aluminium alloy. Hydrogen is thought to arise from the surface reaction of water vapour with the fresh aluminium surfaces created by fatigue in the crack tip region according to the following sequence:

2Al+ 3H20 =+ A1203 + 6Hf + 6e-

Acknowledgements-The authors are grateful to Dra. C. Garcia-Cordovilla and Prof. E. Louis for providing the

aluminium alloy, to Prof. J. Llorca for useful discussion and to D. Jose Miguel Martinez for his help in drawing the figures. Support of this research by the Universidad Polidcnica de Madrid under Contract A-9530 is gratefully acknowledged.


REFERENCES

1. M. 0. Speidel, Hydrogen embrittlement and stress corrosion cracking, p. 271. ASM (I 984). 2. N. J. H. Holroyd, Environment-induced cracking of metals, p. 31 I. NACE, Houston (1990). 3. C. Garcia-Cordovilla. E. Louis, A. Pamies, L. Caballero and M. Elites. Mater. Sri. and Engineering A174.

I73 (I 994). 4. R. E. Stoltz and R. M. Pelloux, Metal/. Trans. 3, 2433 (1972). 5. N. J. H. Holroyd and D. Hardie. Corr. Sci. 23, 527 (1983). 6. N. J. H. Holroyd and D. Hardie, Environment sensitive fracture: evaluation and comparison of test methods, p.

534. ASTM STP 821 (1984). 7. F. J. Bradshaw and C. Wheeler, Int. Journal of Fract. Mech. 5, 255 (1969). 8. R. P. Wei, P. S. Pao, R. G. Hart, T. W. Weir and G. W. Simmons, Metall. Trans. llA, I51 (1980). 9. D. L. Dicus, Environment sensitive fracture: evaluation and comparison of test methods, p. 5 13. ASTM STP 82 1

(1984). IO. M. Gao, P. S. Pao and R. P. Wei, Merall. Trans. 19A, 1739 (1988). I I. M. 0. Speidel. Hydrogen in Metals, p. 249. ASM (1974). 12. J. Ruiz and M. Elites. Vacuum 45. 1069 (1994). 13. J. Ruiz and M. Elites, Jour. of Test. and Eva/. 23, 275 (1995). 14. S. M. Wiederhom. Jour. of the Am. Ceram. Sot. 50, 407 (1967). 15. D. L. Davidson and J. Lankford, Fatigue of Eng. Mat. and Strut. 6, 241 (1983). 16. J. Lankford and D. L. Davidson. Actu Metal/. 31, 1273 (1983). 17. A. Saxena and S. J. Hudak Jr. Inr. Jour. of Fracture 14, 453 (1978). 18. G. Osterstrom, Vacuum physics and technology, Academic Press Inc. Orlando (I 979). 19. N. J. Nix and H. M. Flower, Acta MeraN. 30, 1549 (1982). 20. J. Ruiz. Ph.D. Dissertation. Universidad Complutense de Madrid (1993). 21. C. T. Liu, Scripta Metall. et Mater. 27, 25 (1992). 22. T. W. Weir, G. W. Simmons, R. G. Hart and R. P. Wei, Scripta Metall. 14. 357 (1980). 23. R. P. Wei and G. W. Simmons, Inr. Jour. of Fracture 17, 235 (1981). 24. Metals Handbook, Vol. 2 Properties and selection: nonferrous alloys andpure metals, 9 Edn. American Society

for Metals, Ohio (I 979). 25. D. L. Davidson and J. Lankford, Int. Mater. Reviews 37. 45 (1992).

Documents

Environmental fatigue in a 7000 series aluminium alloy