Study of hydrogen embrittlement in aluminium alloy 2024 in the longitudinal direction

Study of hydrogen embrittlement in aluminium alloy 2024 in the longitudinal direction

Materials Science and Engineering, A 125 (1990) 21 - 30 21 Study of Hydrogen Embrittlement in Aluminium Alloy 2024 in the Longitudinal Direction E Z...

1MB Sizes 0 Downloads 21 Views

Materials Science and Engineering, A 125 (1990) 21 - 30

21

Study of Hydrogen Embrittlement in Aluminium Alloy 2024 in the Longitudinal Direction E ZEIDES and I. R O M A N

Graduate School of Applied Science and Technology, The Hebrew University of Jerusalem, Jerusalem 91904 (Israel) (Received March 15, 1989; in revised form July 5, 1989)

Abstract

A comparative study of the hydrogen effect on overall ductility in the rolling direction of aluminium alloy 2024 in the T351 temper condition and aluminium alloy 2024 aged at 190 °C for 7 h has been conducted. Hydrogen-pre-charged and hydrogen-free sheet tensile specimens have been tested at either a constant or an alternating strain rate. It has been found that internal hydrogen caused severe degradation in overall plastic elongation of aluminium alloy 2024-T351 and that in artificially aged alloy 2024 the embrittlement was observed only after the onset of necking. It has been observed that hydrogen-induced modification of the strain rate dependence of flow stress by reducing dynamic hardening in alloy 2024-T351 aged at 190°C for 7 h and enhancing dynamic softening in alloy 2024-T351. It is suggested that hydrogen embrittlement is caused by hydrogeninduced enhancement of plastic deformation localization that mainly affects non-uniform plastic deformation. 1. Introduction

Stress corrosion cracking (SCC) has been a long-standing problem in structural application of high strength aluminium alloys. For a long time, preferential anodic dissolution at the crack tip dominated the explanations of SCC in aluminium alloys [ 1]. In 1956, aluminium alloy 1100 was reported to disintegrate along grain boundaries above 200 °C in a water vapour atmosphere [2]. The severity of the attack was reduced by coupling aluminium to more cathodic metals. It was concluded that the attack was mechanical damage caused by hydrogen penetration along grain boundaries. This seems to be the first experi0921-5093/90/$3.50

mental evidence of hydrogen-induced loss of integrity in aluminium and its alloys. The detrimental effect of internal hydrogen on the ductility of high strength A I - Z n - M g aluminium alloys (7XXX series) was first investigated by Gest and Troiano [3]. Since then, the possible role of absorbed hydrogen as an operating mechanism of SCC has received a great deal of attention and persuasive evidence for hydrogen embrittlement (HE) in high strength aluminium alloys has accumulated [4-7]. The effect of hydrogen was observed in mechanical tests that employed either continuous straining or static loading. Upon continuous straining, hydrogen caused ductility loss, manifested in either a decrease in the total elongation to fracture [7-9] or in the reduction in fracture area (RA) [3, 10-16]. In samples loaded in the elastic region, HE was manifested by delayed fracture [17]. The determination of the RA loss in tensile specimens strained slowly to fracture has become an almost universal method for HE studies. The severity of the hydrogen effect has been related to external and internal parameters, such as hydrogen fugacity, loading mode, strain rate, temperature and microstructure [3-6, 12, 15, 16]. The hydrogen fugacity--an important variable in HE--is established by hydrogen overpotential set by the free corrosive reaction or controlled externally [18]. This has been studied qualitatively either by varying the humidity level and temperature in water vapour [4, 7] or by changing the applied electromechanical potential of the specimens in liquids [3, 4, 10, 16]. The general conclusion that a higher hydrogen fugacity level is related to a more severe hydrogen effect can be inferred from these investigations. The susceptibility of aluminium alloy 7075 in the T6 temper condition to environmentally induced embrittlement has been demonstrated to © Elsevier Sequoia/Printed in The Netherlands

22 depend on the loading mode [19]. The fact that SCC was still observed in mode III (pure torsion), although to a lesser extent than in mode I (pure tension), was taken as evidence that anodic dissolution is possible and is a dominant mechanism in mode II loading, However, the use of loading mode with no hydrostatic component as a diagnostic tool must be exercised with care since strong hydrogen-induced embrittlement has also been observed in in situ high voltage electron microscopy (HVEM) experiments on alloy 7075T6 [20] and also in thin sheets of high purity aluminium which is inherently soft material [21]. In both cases the hydrostatic stress component was small. HE depends on the applied strain rate and the test temperature. When hydrogen was supplied to the tensile specimens during testing, higher strain rates corresponded to reduced ductility losses [10, 16], whereas in hydrogen-pre-charged specimens the highest ductility losses were observed at the same intermediate strain rates [12, 22]. The reduced severity of HE in pre-charged specimens at the slower strain rates has been attributed to the depletion of atomic hydrogen from the matrix by leakage to inclusions and large precipitates [22]. The temperature dependence of HE in hydrogen-pre-charged specimens was found to be non-monotonic. The effect was absent at the lowest test temperature ( - 200 °C) and exhibited a maximum which was dependent on the microstructure [15]. The relationship between HE and microstructural features such as strengthening precipitates [5, 6, 9, 12-15] and grain boundary particles [9] has been studied. The role of strengthening precipitates in HE has been attributed to their effect on the slip mode control [5, 6, 9, 12-15], as was also suggested earlier for SCC [23]. In both HE and SCC, microstructures that gave rise to the more planar or coarse slip were reported to be more susceptible to embrittlement. However, the effect of microstructure on HE in alloy 7075 was small, if any, when mechanical tests were conducted at - 1 0 0 ° C [15]; at this temperature, the extent of HE was close to its peak value and almost equally high for microstructures with very different slip planarities. The explanation of HE microstructure relations in terms of slip characteristics has been based on the premise that moving dislocations transpoi-t hydrogen to the sites at which fracture initiates [5, 6, 9, 12-15, 24]. According to this, a higher slip concentration results in a higher

hydrogen content and therefore greater damage potential at the sites at which slip bands terminate. The observations made on alloy 7075T6--that the embrittled layer was deeper after hydrogen cathodic pre-charging with simultaneous slow straining compared with cathodic precharging alone--support the mechanism of hydrogen transport by dislocations [13]. When a similar charging procedure was employed for pure nickel, no enhancement of the depth of embrittlement occurred [25]. Direct measurements of hydrogen permeation during plastic deformation in polycrystalline nickel [26] and of hydrogen penetration profiles established during plastic deformation with the use of secondary-ion mass spectrometry [27] also did not reveal any enhancement of hydrogen diffusivity by plastic deformation. In the very special case where hydrogen trapping due to sessile dislocations was suppressed by the use of single crystals deformed in the easy-glide region, evidence for dislocation transport of hydrogen has been reported [28, 29]. A different approach to the interrelation between microstructure and HE has been made by Christodoulou and Flower [9]. They have demonstrated that grain boundary precipitates can serve as preferential sites for hydrogen bubble nucleation, provided that the precipitates have reached a critical size. Such bubbles collect atomic hydrogen from the metallic matrix, resulting in reduced sensitivity to HE. Yet another suggestion has been made by Holroyd and Hardie [16]. They have related the beneficial effect of overaging to a reduction in hydrogen permeation and their results indicate that overaged aluminium alloy 7049 could be as embrittled as the HE-sensitive alloy 7049-T651, given sufficient time for hydrogen absorption. The fracture morphology of the 7XXX series hydrogen-charged samples changes from a ductile transgranular appearance to a brittle intergranular topography, and the greater proportion of intergranular morphology has been related to a larger degree of embrittlement, or ductility drop [7, 10]. Although intergranular hydrogen-assisted fracture is most common in A1-Zn-Mg alloys, transgranular cleavage-like brittle fracture has also been reported [9, 10, 14, 16, 30]. However, a brittle fracture appearance, either intergranular or transgranular, is not a necessary feature of HE. Thompson and coworkers have

23 demonstrated that hydrogen was able to cause embrittlement without altering the transgranular ductile fracture mode, both in aluminium alloy 7075 [15] and in aluminium alloy 2124 [11], when the sample long axis coincided with the rolling direction. The dimple size on the fracture surfaces was analysed for alloy 7075. The hydrogen effect on fracture development was suggested to be specific to microstructure, giving rise to enhancement of either void growth or nucleation [15]. They also found that H E (measured by the RA loss) was less pronounced in alloy 2124 than in alloy 7075 and that, in addition to cathodic hydrogen charging, simultaneous slow straining was necessary to reveal H E in alloy 2124. This was the first reported evidence of H E in the 2XXX series of aluminium alloys. Prior to this observation, the lack of evidence for H E in A1-Cu alloys was suggested to support the pure anodic dissolution mechanism of SCC in this system [31]. More recently it has been shown that hydrogen embrittles not only aluminium-based alloys but also affects high purity aluminium [21]. The purpose of the present study is threefold: first to show that, in fact, alloy 2024 (essentially alloy 2124 with higher iron and silicon contents) is very sensitive to HE; second to demonstrate that macroscopic appearance of H E in alloy 2024 differs in naturally aged and artificially aged heat treatment conditions; finally, to show that the H E phenomena in aluminium and its alloys can be treated in the framework of hydrogen-induced enhancement of plastic deformation localization.

2. Experimentalprocedure An alloy 2024-T351 plate 25.4 mm thick (made by Reynolds) was used in this study. The T351 heat treatment condition means that this material has been quenched, stretched and then naturally aged at ambient temperature. The grain structure of the material is given in Fig. 1, and the grain size was 1.3 mm x 0.3 mm x 0.08 mm. Flat tensile specimens having 32 m m x 6.25 m m x 1 mm gauge dimensions were machined with the tensile axis parallel to the rolling direction and the specimen plane perpendicular to the rolling surface. One group of specimens was tested in the as-received condition and another group has been additionally aged at 190°C for 7 h to increase the strength level. Two techniques for hydrogen charging have been used. Specimens

Fig. 1. Typical micrograph showing grain morphology in alloy 2024-T351 plate in a section perpendicular to the rolling plane and containing the rolling direction. for hydrogen effect evaluation were either soaked in 0.1 N N a O H water solution stirred continuously for different periods of time up to 35 days or exposed to deionized water in an ultrasonic bath. A commercial ultrasonic cleaner was employed for this purpose. It was observed that the hydrogen pick-up was higher in the ultrasonic water bath 105 wt.ppm (after 1 week) compared with a maximum of 50 wt.ppm obtained by soaking in 0.1 N N a O H water solution. The improved effectiveness of employing an ultrasonic water bath for hydrogen charging was realized after most of the experiments with the T351 samples had been completed in 0.1 N N a O H water solution. Consequently, additional samples of the T351 condition were tested after charging in an ultrasonic water bath to evaluate whether the two charging techniques result in a similar hydrogen effect on tensile properties. The hydrogen charging was confirmed in several randomly chosen specimens analysed for hydrogen by the vacuum extraction technique. Both hydrogen-charged and uncharged tensile specimens were polished mechanically with 1200 grit paper followed by polishing on cloth soaked with 3 /~m diamond paste. Shortly after polishing, the specimens were strained in a screw-driven tensile machine (JJM30K) in laboratory air either continuously, at initial strain rates in the range 10-5_ 10-1 s-l, or with strain rate changes during the test. From two to four specimens were tested for each strain rate. The uncharged specimens were coated with silicone oil prior to straining to prevent contact with humidity in the atmosphere. Acoustic emission (AE) measurements were

24

employed to examine the possible role of internal hydrogen on the failure mechanism, Le. microvoid nucleation. A E characteristics were evaluated during tensile deformation utilizing a microcomputer-based system (model 5000, A E T ) consisting of a 175 kHz resonant transducer (MAC 175, AET), pre-amplifier (160, A E T ) with 125-250 kHz bandpass filter and an r.m.s, voltmeter (201, AET). The A E sensor was attached to the side surface of tensile specimens close to the gauge section.

strength and strain hardening, were similar for both hydrogen-free and hydrogen-pre-charged specimens. A small, if any, hydrogen effect on the loss of thickness reduction at fracture was noted. The results of the mechanical tests conducted at an initial strain rate of 10-4 s-1 are given in Table 1 and indicate that hydrogen caused significant deterioration in the overall ductility as measured by elongation to fracture. The severity of this effect increased continuously with increas-

15

3. Results

3.1. Alloy 2024-T351 3.1.1. Mechanical testing Jerky flow, or the Portevin-Le Chatelier (PLC) effect, characterized the load-deflection curves of the alloy 2024-T351 tensile specimens. PLC serrations were strong at 10 -5 s -1 and weak at higher strain rates. PLC bands could be seen on the gauge length of both hydrogen-charged and uncharged specimens. Fracture occurred predominantly by shear with little necking for both hydrogen-charged and uncharged specimens. Load-deflection and the corresponding r.m.s, of A E voltage curves for hydrogen-free and hydrogen-pre-charged specimens are given in Fig. 2. HE appears in the smaller extent of plastic deformation prior to failure. Except for that, the load-deflection characteristics, i.e. the yield

uncharged

_~

'~

"-.'.,

I0

2024 T351 ALOY

uJ

~

~

~

g//

50 z ata¢

/

20

/ /

8

I0

I00 HYDROGEN CHARGING TIME ( hrs )

ro

I00'

Fig. 3. Plastic elongation ( zx ) of hydrogen-pre-charged alloy 2024-T351 specimens tested at 10 -4 s -~ initial strain rate and hydrogen content (o) after different soaking times in 0.1 N NaOH solution. 18 16

o

o

li 3K ff

~\

sOeclmens

o

o ~

.

o

S iz N Io

~e

|

~6

H-charged O~ I /T'H-charged

,-----, '-----'

0

DEFLECTION

10_<.

i

10.5

i

10.2

i/

i0.1

INITIAL STRAIN RATE (sec -I )

Fig. 2. Load-deflection charts with corresponding r.m.s. curves of A E for alloy 2024-T351 tensile specimens. The leftmost curves are for the hydrogen-pre-charged (in 0.1 N NaOH solution) specimen. TABLE 1

i

110.5

Fig. 4. Total elongation of hydrogen-free (o) and soaked (for 35 days in 0.1 N NaOH solution (e)) alloy 2024-T351 specimens, tested at different strain rates. The elongation loss (zx) is greatest at about 10- 4 s 1initial strain rate. -

Results of mechanical tests for alloy 2 0 2 4 - T 3 5 1 tested at • = 1 0 - 4 s -

H charging time in 0.1 N NaOH water solution

True yield strength (MPa)

True tensile strength (MPa)

Strainhardening coefficient

True total elongation (%)

Elongation loss (%)

Uncharged 35 days (840 h)

398.0 ___11.3 404.1 ± 15.6

590.3 ± 13.5 534.7 ± 9.4

0.16 0.16

14.5 ± 0.8 7.8 ± 1.6

0 46.2

25

ing hydrogen content as shown in Fig. 3. The hydrogen content achieved saturation level well before lo3 h of charging. This trend is presumably also true for the hydrogen effect on elongation. In several cases, tensile specimens were pre-charged in ultrasonic water bath. This treatment resulted in higher hydrogen pick-up and higher extent of HE. The experimental results, summarized in Fig. 4, show that the hydrogen effect on the total elongation depends on the

strain rate. In the range of strain rates used in this study, the ductility drop was most severe when testing was conducted at an initial strain rate of 10-4 s-1. 3.1.2. Fractography As mentioned earlier, all specimens failed predominantly by shear fracture. Fractographs, showing a typical fracture surface topography of both charged and uncharged broken samples, are shown in Fig. 5. The shear-type fracture morphologies of both charged and uncharged specimens are similar and consist of dimpled areas and tear ridges. The only distinct difference between the two was the presence of a “mud crack’ pattern occasionally found on the fracture surfaces of the hydrogen-charged specimens, demonstrated in Fig. 6. This pattern, however, was found rarely and not on all specimens that exhibited hydrogen-induced ductility loss. 3.2. Alloy 2024- T351 aged at 190 “Cfor 7 h 3.2.1. Mechanical testing

Specimens for hydrogen effect evaluation were soaked in water in an ultrasonic bath for 1 week. This procedure resulted in hydrogen pick-up of about 105 wt.ppm. All the specimens were tested at an initial strain rate of 10e4 s-l. Unlike alloy 2024-T351, artificially aged specimens broke after remark-

Fig. 5. Typical fractographs showing the fracture morphology of ahoy 2024-T351 tensile specimens tested at an initial strain rate of lo-“ SC’ (the fracture surfaces are composed of shear dimples and tear ridges): (a) uncharged; (b) hydrogen charged for 35 days (840 h) in 0.1 N NaOH solution.

Fig. 6. Fractograph showing the “mud crack” pattern on the shear fracture surface of an alloy 2024-T351 hydrogencharged specimen.

26 TABLE 2

Results of tensile tests for alloy 2024-T 351 aged for 7 h at 190 °C

o0.2 (MPa)

OUTs (MPa)

EtotaI (%)

Eunif (%)

Eneck (%)

Eneck Reduction loss

(%)

in thickness

(%)

Reduction in thickness

toss (%)

WithoutH

486.6+4.8

545.7+2.5

5.1+0.5

3.7+0.5

1.4_+0.0

0

26.2+2.8

0

Hpre-charged

481.6+3.3

562.3±2.7

5.3+0.6

4.5±0.0

0.8-+0.0

42.9

18.1+6.0

30.9

APneck Pmax (%)

AP /)max loss

(%) 10.3+0.8 2.3-+1.0

0 77.7

00.2, yield strength measured at 0.2% of plastic deformation; OUTS, ultimate tensile strength; Etotal, total plastic deformation elongation at break; Eun~f,uniform plastic deformation elongation; E,eck, elongation during necking development; Reduction in thickness, reduction in thickness at the fracture surface, related to the initial specimen thickness; AP~eek,load drop during necking development, i.e. from the maximum load/)max to the load at fracture. All the stress and strain values are in engineering units.

0.I KN

0 _I

O~5mm i

,

x

DEFLECTION

Fig. 7. Load-deflection (top curves) with corresponding r.m.s, voltage of AE (bottom curves) charts of hydrogen-free (right-hand curves) and hydrogen-pre-charged (left-hand curves) specimens of alloy 2024-T351 specimens aged at 190 oC for 7 h. The smaller necking portion for hydrogencontaining specimens should be noted.

able neck development; nevertheless, shear fracture was still predominant. The results of the mechanical tests are compiled in Table 2. The loss of reduction in thickness of the fracture area was found to be a more convenient measure of H E than the loss of R A as commonly used for round tensile specimens. The projection on the plain normal to tensile axis rather than the true fracture area is usually measured as a good approximation. In our case, where thin sheet specimens were employed, the resulting shear fracture was inclined both to the specimen surface and to the tensile axis with variations in the extent of the inclination along the fracture path or from specimen to specimen. Typical load-deflection curves with corresponding A E charts axe given in Fig. 7, for both reference and hydrogen-pre-charged samples. Hydrogen precharging caused delay in the onset of necking as seen from the results of uniform elongation and ultimate strength that were higher for hydrogenpre-charged specimens. The H E was manifested only in the neck development stage, as measured by the elongation, the reduction in thickness at fracture area, and the relative load drop after

k

DEFLECTION

Fig. 8. Enlarged necking portion of load-deflection charts defined in Fig. 7. The charts have been superimposed so that the ultimate loads of the individual curves were coincided. It should be noted that fracture in hydrogen-pre-charged specimens (curves 1, 2) has occurred at a smaller load drop and after a shorter deflection after the onset of necking.

necking onset. Magnified necking parts of the load-deflection curves are given in Fig. 8. The total elongation to fracture was not affected by hydrogen. A E characteristics during tensile tests were similar for both hydrogen-charged and uncharged specimens. 3.2.2. Fractography Fracture surface topographies of hydrogencharged and uncharged specimens were similar to each other and were composed of ductile dimples and tear ridges as shown in Fig. 9. 3.3. Alternating strain rate tests

The strain rate dependence of the flow stress was measured for both heat treatment conditions. The cross-head velocity was changed by a factor of 10 either upwards or downwards during tension tests. Characteristic load-deflection curves for both hydrogen-charged and uncharged specimens are given in Fig. 10. Alloy 2024-T351 specimens exhibited a negative strain rate sensitivity, i.e. an increase in strain rate

27

O-3KN

l

~

v

~ t

2024

C~ 0 -J 7hrs at Igo"c

[ O.~KN 0.,?.mm

DEFLECTION

Fig. 10. Load-deflection charts of alloy 2024-T351 and

190 °C-aged tensile specimens tested with alternating strain rates. The upward-pointing arrows correspond to the switch to the faster straining and the downward-pointing arrows to the slower staining. TRUE STRESS (MFo) 0

450

'/,tu

u

490 ~

550

i

=

570

]

u

610

I

u

650

u

u

-I

£

-2 -5 -4 -5 -6

Fig. 11. Strain rate softening v s . true stress for hydrogenpre-charged (~, ~) and hydrogen-free (o, Q) alloy 2024-T 351 specimens (el = 5 x 10 -5 s-]; ez = 5 x 10 -4 s- l): slow-fast transitions; fast-slow transitions. Two specimens for each group were used. 0 4~

,

TRUE STRESS (MPo) 470 510 5t50 I

Fig. 9. Typical fractographs showing the fracture morphology of tensile specimens of alloy 2024 which had been aged at 190 °C tested at 10 4 s-J initial strain rate (the fracture surfaces are composed of dimples and tear ridges): (a) specimen exposed to an ultrasonic field in pure water for 7 days; (b) hydrogen-free specimen.

gave rise to a r e d u c t i o n in flow stress a n d vice versa. In artificially aged specimens the strain rate sensitivity was positive so that the strain rate and flow stress changes h a d a parallel trend. T h e stress changes A o c a u s e d b y rate changes were o b t a i n e d b y extrapolating the l o a d - d e f l e c tion curve after its stabilization b a c k to the point o f rate alteration. T w o separate l o a d - d e f l e c t i o n curves, o n e f o r e a c h strain rate, c o u l d be thus constructed. T h e results o f the strain rate d e p e n d e n c e of flow stress are s u m m a r i z e d in

-3

t

I

~.,.,..~

1

650

590 ,

I

I

I

~

-4

Fig. 12. Strain rate softening v s . true stress for alloy 2024T351 specimens tested by the alternating method between e I = 1 0 - 3 s - n and e2=5 x 1 0 - 4 s -l strain rates. One specimen was used for each condition. Figs. 1 1 - 1 3 . It is evident that h y d r o g e n m o d i f i e d this d e p e n d e n c e f o r b o t h heat t r e a t m e n t c o n d i tions, e n h a n c i n g d y n a m i c softening f o r the T 3 5 1 c o n d i t i o n and r e d u c i n g d y n a m i c h a r d e n i n g for the artificially aged condition.

4. Discussion T h r e e m a j o r observations m a d e in the present study are, first, the deleterious effect o f internal

28

2.5

1.5 b .=w ID

~c~ o.5 "

490

500

0

520

530

540

5

TRUE STRESS (MPa)

Fig. 13. Strain-rate-hardening results for s p e c i m e n s of alloy 2 0 2 4 - T 3 5 1 aged at 190 °C for 7 h, tested by the alternating m e t h o d between e ~= 1 0 - 3 s - J a n d e2 = 5 x 1 0 - 4 s - 1 O n e specimen was used for each group.

hydrogen on the ductility of aluminium alloy 2024, second, the distinction in the characteristic of HE in naturally and overaged T351 heat treatment conditions and, third, the hydrogen-induced modification of the strain rate dependence of flow stress. In naturally aged alloy 2024, HE was manifested in the reduction in total elongation to fracture with no effect on the reduction in thickness at fracture surface. In contrast, HE in overaged alloy T351 appeared only after the onset of necking and exhibited itself in a diminution in elongation during necking, in a decrease in the reduction in thickness at the fracture surface and in a reduction in the relative load drop from the maximum load to the break. In this case, the total elongation to fracture remained unaffected by hydrogen. The fracture mode was transgranular ductile for both hydrogen-free and hydrogencharged specimens. For both hydrogen-precharging techniques employed in this study the hydrogen effects on the tensile test were qualitatively similar. It seems that HE in aluminium alloy 2024 could be understood in terms of the interrelationship between the macroscopic plastic flow character and fracture. A ductile type of fracture results from internal void nucleation, growth and linking up, either by void impingement or by the formation of void sheets between grown voids [32]. When the plastic flow is macroscopically non-uniform but is localized in a shear band or neck, fracture initiates there, imposing a limit on overall deformation. The stability of a material to flow localization depends on the strain hardening, strain rate flow stress dependence and void population [33-35]. In the case of the alloy 2024-T351, the strain rate sensitivity was negative, so that the plastic flow became unstable shortly after general yielding. Shear bursts appeared and began to grow with

plastic flow. Finally, fracture initiated within a shear band. In the alloy 2024-T351 aged at 190 °C, the strain rate sensitivity was positive and plastic deformation localization was delayed until necking. One finding made in the present study-that hydrogen caused a modification in strain rate flow stress dependence for both heat treatment conditions--makes interpretation of HE straightforward. In both naturally and artificially aged conditions, HE was confined to non-uniform portions of plastic flow, i.e. PLC bursts in the T351 condition, and necking in the overaged T351 condition. This observation indicates that internal hydrogen causes intensification of plastic deformation localization. Any localization is accompanied by a local increase in strain rate. In the present case this transition induces smaller hardening in alloy 2024 aged at 190 °C and larger softening in alloy 2024-T351 hydrogen-precharged specimens than in hydrogen-free samples of the two heat treatment conditions. Consequently, the resistance of the alloy to plastic flow localization is expected to decrease owing to hydrogen, thus leading to premature fracture. A higher hydrogen content results in a larger extent of HE as can be judged from the mechanical test after the two hydrogen-pre-charging techniques employed in the present study. The preceding discussion is a new approach to HE in aluminium alloys. Traditionally, HE in aluminium alloys has been attributed to either earlier void nucleation or faster void growth [15]. In order to address the possible role of hydrogen in the void nucleation process, A E measurements were employed in the present study. A E during plastic deformation of aluminium alloys 2024-T351 and 2024-T6 has been exclusively related to fracture or to decoherence from the matrix of the intermetallic particles [36, 37], giving information about void nucleation. Since the A E of hydrogenfree and the A E of hydrogen-charged specimens were very similar, little or no effect of hydrogen on void nucleation can be suggested. In addition to A E r.m.s, voltage measurements, the distribution of individual A E events by peak amplitude was studied, employing a microcomputer-based system. Again, no difference between hydrogencharged and uncharged groups of specimens was found. The usefulness of A E in the study of HE in aluminium-based alloys has been demonstrated recently on AI-Li alloy 2091, where A E was strongly influenced by the presence of hydrogen [38].

29

The observation made in this study for transgranular fracture in alloy 2024--that HE operates only during non-uniform plastic flow--is consistent with experimental results reported in the literature for other metallic systems. It has been found that HE in porous iron influenced only neck development in smooth specimens but caused remarkable ductility loss when notched specimens were used and therefore plastic flow was non-uniform from the beginning [39]. Hydrogen enhancement of plastic instability in ironbased alloys has been extensively reviewed [35]. Also in all in situ H V E M experiments on a variety of metallic systems, the presence of a notch was essential to reveal hydrogen-induced flow modification and loss in ductility [20, 40, 41]. A film structure was found on the fracture surfaces of several hydrogen-embrittled specimens. In some cases this structure was layered and fragmented by a network of short cracks, the so-called "mud cracks" pattern. Generally, an area exhibiting this kind of structure also bore the appearance of a mechanically damaged surface. It is believed that the filmed areas resulted from mechanical damage and were formed during relative sliding of opposite fracture surfaces in the course of crack propagation. Shear mode deformation at the crack tip is thus presumed. The formation of the film fracture in hydrogencontaining alloys is assumed to result from the hydrogen-induced plastic deformation localization accompanying the fracture process. A more complete discussion of the phenomenon for pure aluminium and aluminium-based alloys has been given elsewhere [21]. Finally, it should be noted, that a new method of ductility evaluation, namely relative load drop during neck development, has been employed in the present study. This method seems to be very attractive since it can be obtained directly from the load-deflection chart in a very simple fashion.

degradation is sensitive to the elongation strain rate with the maximum effect at about 10-4 s- 1. (3) Hydrogen-pre-charged alloy 2024-T351 sheet specimens aged at 190 °C for 7 h exhibit ductility degradation only after necking onset as measured by a decrease in relative load drop (from the maximum load to the fracture load), in elongation during the necking stage and in reduction in thickness at the fracture surface. (4) Hydrogen-induced modification of the strain rate dependence of flow stress, either by reducing dynamic hardening in alloy 2024-T351 aged at 190 °C for 7 h or by enhancing dynamic softening in alloy 2024-'T351. (5) It has been suggested that HE in aluminium alloy 2024 is caused by hydrogeninduced enhancement of plastic deformation localization. Uniform plastic flow is thus suggested to be resistant to HE in this alloy. (6) The experimental results obtained indicate that hydrogen does not alter the fibrous transgranular fracture mode in longitudinal samples of aluminium alloy 2024. (7) In several cases, a film fracture morphology was found in hydrogen-containing alloy 2024-T351. Part of the film surfaces appeared cracked and layered, the so-called "mud crack" pattern. This feature is believed to be the result of mechanical damage occurring between opposite fracture surfaces due to the hydrogen-induced plastic shear localization accompanying fracture development. Acknowledgments Valuable discussions with Professor I. M. Bernstein, Illinois Institute of Technology, are greatly appreciated. One of the authors (F. Zeides) wishes to acknowledge the Center for Absorption Scientists of the Israeli Ministry of Absorption for financial support.

5. Conclusions References (1) Internal hydrogen caused remarkable ductility loss in aluminium alloy 2024. The HE characteristics during tensile deformation differ for the two heat treatments studied. (2) Hydrogen-containing aluminium-based alloy 2024-T351 suffers significant ductility loss which was manifested in a reduction in the extent of plastic elongation. The degree of ductility

1 R.B. Mears, Proc. Syrup. on Stress Corrosion Cracking of Metals, Philadelphia, PA, American Society for Metals, Metals Park, OH, 1945, p. 329. 2 J.E. Draley and W. E. Ruther, Corrosion, 12 (1956) 480t. 3 R. Gest and A. R. Troiano, Corrosion, 30 (1974) 274. 4 M. O. Speidel, in I. M. Bernstein and A. W. Thompson (eds.), Hydrogen in Metals, 1974, p. 249. 5 A. W. Thompson and I. M. Bernstein, Rev. Coatings Corros., 2 (1975) 2.

30 6 A. W. Thompson and I. M. Bernstein, Adv. Corros. Sci. Technol., 7(1980) 53. 7 G. M. Scamans, R. Alani and P. R. Swann, Corros. Sci., 16 (1976) 443. 8 T. E Klimowicz and R. M. Latanision, Metall. Trans. A, 9 (1978) 597. 9 L. Christodoulou and H. M. Flower, Acta Metall., 28 (1980) 481. 10 D. Hardie, N. J. H. Holroyd and R. N. Parkins, Met. Sci., 13 (1979) 603. I 1 D. A. Hardwick, M. Taheri, A. W. Thompson and I. M. Bernstein, Metall. Trans. A, 13 ( 1982) 235. 12 M. Taheri, J. Albrecht, I. M. Bernstein and A. W. Thompson, Scr. Metall., 13 (1979) 871. 13 J. Albrecht, I. M. Bernstein and A. W. Thompson, Metall. Trans. A, 13 (1982) 811. 14 D. A. Hardwick, A. W. Thompson and I. M. Bernstein, Metall. Trans. A, 14(1983)2517. 15 J. Albrecht, A. W. Thompson and I. M. Bernstein, Metall. Trans. A, 10 (1979) 1759. 16 N. J. H. Holroyd and D. Hardie, Corros. Sci., 21 (1981) 129. 17 G.H. Koch, Corrosion, 35 (1979) 73. 18 M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon, Oxford, 1966. 19 J. A. S. Green, H. W. Hayden and W. S. Montague, in A. W. Thompson and I. M. Bernstein (eds.), Effect of Hydrogen on Behavior of Materials, Metallurgical Society of AIME, New York, 1976, p. 200. 20 G. M. Bond, I. M. Robertson and H. K. Birnbaum, Acta Metall., 35 (1987) 2289. 21 F. Zeides, Ph.D. Thesis, University of Illinois, 1986. 22 M. Mueller, I. M. Bernstein and A. W. Thompson, Scr. Metall., 17(1983) 1039. 23 H.A. Holi, Corrosion, 23 (1967) 173. 24 J. K. Tien, A. W. Thompson, I. M. Bernstein and R. J.

Richards, Metall. Trans. A, 7(1976) 821. 25 A. Kimura and H. K. Birnbaum, Scr. Metall., 21 (1987) 219. 26 G. S. Frankel and R. M. Latanision, Metall. Trans. A, 17 (1986) 861. 27 B. Ladna and H. K. Birnbaum, cited by A. Kimura and H. K. Birnbaum, Scr. Metall., 21 (1987) 219. 28 G. S. Frankel and R. M. Latanision, Metall. Trans. A, 17 (1986) 869. 29 C. Hwang and I. M. Bernstein, Acta Metall., 34 (1986) 1001. 30 S.P. Lynch, Corros. Sci., 22 (1982) 925. 31 R. M. Latanision, H. Gastine and C. R. Compeau, in Z. A. Foroulis (ed.), Environment-Sensitive Fracture of Engineering Materials, Metallurgical Society of AIME, Warrendale, PA, 1979, p. 48. 32 G. T. Hahn and A. R. Rosenfeld, Metall. Trans. A, 6 (1975)653. 33 A.K. Ghosh, Metall. Trans. A, 5 (1974) 1607. 34 N. Chung etal., Acta Metall., 25 (1977) 377. 35 O. A. Onyewuenyi, in R. A. Oriani, J. P. Hirth and M. Smialawski (eds.), Hydrogen Degradation of Ferrous Alloys, Noyes, Park Ridge, NJ, 1985, p. 44. 36 Y. Fukuzawa et al., in M. Onoe, K. Yamaguchi and H. Takahashi (eds.), Progress in Acoustic Emission II, Japanese Society for Non-Destructive Inspection, Tokyo, 1984,p. 1. 37 S. Mck. Cousland and C. M. Scala, Mater. Sci. Eng., 57 (1983) 23. 38 E Zeides and I. Roman, J. Acoust. Emiss., 8 (1989) 109. 39 J. Kameda and W. A. Spitzig, Mater. Sci. Eng., 96 (1987) 311. 40 T. Tabata and H. K. Birnbaum, Scr. Metall., 18 (1984) 231. 41 I. M. Robertson and H. K. Birnbaum, Acta Metall., 34 (1986) 353.