A literature review and inventory of the effects of environment on the fatigue behavior of metals

A LITERATURE REVIEW AND INVENTORY OF THE EFFECTS OF ENVIRONMENT ON THE FATIGUE BEHAVIOR OF METALS C. MICHAEL HUDSON Mail Stop 378,NASA Langley Research Center, Hampton, VA 23665,U.S.A.

SUE K. SEWARD Mail Stop 185,NASA Langley Research Center, Hampton, VA 23665,U.S.A. Abstract-This paper reviews the current state of knowledge of the effects of gas environments (at atmospheric pressure and below) on the fatigue behavior of metals. Specifically, this report reviews (a) the mechanisms proposed to explain the differences observed in the fatigue behavior of vacuum- and air-tested specimens, (b) the effects of environment on the surface topo~aphy of fatigue cycled specimens, (c) the effect of environment on the various phases of the fatigue phenomenon, (d) the effect of prolonged exposure to vacuum on fatigue life, (e) the variation of fatigue life with decreasing gas pressure, and (f) gas evolution during fatigue cycling. Analysis of the findings of this review indicates (a) hydrogen embrittlement is primarily responsible for decreased fatigue resistance in humid environments,(b) dislocations move more easily during tests in vacuum than during tests in air, (c) fatigue cracks generally initiated more rapidly in air then in vacuum, (d) fatigue cracks always propagated more rapidly in air than in vacuum, (e) proIong~ exposure to vacuum does not adversely affect fatigue resistance, and (ff the variation of fatigue lie with decreasing gas pressure is sometimes stepped and sometimes continuous. NOTATION N, NA Nd Nh N* N,, NW &

fatigue life in laboratory air, cycles fatigue life in pure argon, cycles fatigue life in dry air, cycles fatigue life in humid air, cycles fatigue life in pure hydrogen, cycles number of cycles required to initiate a fatigue crack in laboratory air, cycles number of cycles required to initiate a fatigue crack in vacuum, cycles fatigue life in pure nitrogen, cycles fatigue life in pure oxygen, cycles (da ,dN;: rate of fatigue-crack propagation in laboratory air, nmicycle (in./cy~le) (daidNld rate of fatigue-crack propagation in dry air, nmicycle (in/cycle) (da/dNb, rate of fatigue-crack propagation in humid air, nmlcycle (in/cycle) (da/dN)” rate of fatigue-crack propagation in pure hydrogen, nmlcycle (in/cycle) (da/dNJo rate of fatigue-crack propagation in pure oxygen, nmlcycle (in/cycle) (da/dN), rate of fatigue-crack propagation in vacuum, nmlcycle (in/cycle) R ratio of the minimum stress to the maximum stress INTRODUCTION

A PREVIOUSreview of the literature[l] showed the fatigue resistance of metals is generally lower in atmospheric than in vacuumenvironments. Gold alone was unaffected by the vacuum environment[2]. The researchers cited in [l] proposed several types of mechanisms to explain this decrease in fatigue resistance. However, the data available at the time [I] was pubtished were insufficient to verify any of the proposed mechanisms. Since that time, researchers have conducted some very perceptive research into possible mechanisms. The primary objective of this literature review is to collate the findings of the various researchers, and to indicate the most probable mechanism for explaining the decreased fatigue resistance in the atmosphere. Identification of this mechanism could lead to the development of alloys which are not susceptible to this mechanism and which consequen~y would have higher fatigue resistance. EXPERIMENTAL

RESULTS FROM INVESTIGATIONS

REPORTED

IN THE LITERATURE

Effect of environment on the surface topography of fatigue cycled specimens Table 1 presents a summary of the effects of vacuum and air environments on surface

deformation due to fatigue cycling. Post-test inspection of most specimens’ surfacesl4-81 315

316

C. M. HUDSON

and S. K. SEWARD

Table 1. Effects of vacuum and air environments on surface deformation due to fatigue cycling Findings For a given strain level, specimens

Greeqwood Wilkow and A$piervhite Kr;lmer and Podlnseck Wadsworth and Hutchings

6

99.999% pure aluminum single crystals aluminum single crystals OFHC Copp*r

7

8

deformed

more in:

VXUUlll

Vacuom

VXUUIll

Superpure

Neither.

Same deformation air

in

vacuum and Air

indicated that, at a given stress level, the surfaces of vacuum-tested specimens rumpled considerably more than the surfaces of air-tested specimens. In contrast, Grosskreutz[9] found the surfaces of his specimens rumpled more in air than in vacuum. The more rumpled surfaces observed on most vacuum-tested specimens probably indicates greater dislocation mobility in the vacuum environment. Table 2 presents a summary of the effects of vacuum and air environments on the number of fatigue cracks which initiate during fatigue testing. In two investigations [4,7], numerous fatigue cracks developed in vacuum-tested single and polycrystalline aluminum specimens whereas relatively few developed in air-tested ones. In another investigation@], the same number of cracks developed in air- and vacuum-tested copper and polycrystalline aluminum specimens. In still another investigation[5], more cracks developed in air-tested lead specimens than in vacuum-tested ones. Table 2. Effects of vacuum and air environments on the density of fatigue cracks

Investigator(s)

Reference

TMaterial(s)

Findings

1In

Ham

4

llOO-H14

In tests in ~xwm.

tests in air,

/ I a single

fatigue crack initiated and propagated failure.

aluminum

a fatigue crack initiated, propagated a short distance and stopped. A second crack then initiated and repeated the propagation and stopping process. Eventually, either one crack propagated to failure, or the network of partially propagated cracks became so extensive the specimens were too flexible to test. specimens developed many, uniformly-distributed c%cks.

to

xramer and Podlaseck

*

Wadsworth and Rl;tchings

OFHC copper

specimens

Superpure polycry+ talline alurr,inum

specimens developed many cracks.

1

many

developed

cracks.

t

specimens cracks.

also developed

many

specimens cracks.

also developed many

3i7

The effects of environment on the fatigue behavior of metals Table 2. (Coa~d) Investigator(s)

Reference

Findings

Material(s) Tested

In tests in vacuum,

In tests in air, 5

Snowden and Greenwood

specimens

High-

/ purity

developed

i manycracks.

I

specimens developed few cracks.

lead

Tabie 3 presents a summary of the effects of vacuum and air environments on fracture surface topography. The vacuum-tested specimens generally exhibited fracture surfaces indicative of ductile failure whereas the air-tested ones exhibited surfaces indicative of more brittle fracture [ 10,l I]. The more ductile failure mode in the vacuum-tested specimens also indicates greater dislocation mobility in the vacuum environment. Table 3. Effects of vacuum and air environments on fracture surface topography Investigator(s)

Reference

Findings

Material(s) Tested For specimens in air,

Sumsion

10

xeyn

11

Magnesium, HM2lA magnesiumthhorbza aflay, and La 141A magnesiumlithium alloy 2024-T3 aluminum alloy

tested

For specimens tested in vacuum,

many flat, platelet-like areas, indicative of a more brittle fracture mode, were evident. These areas somewhat resembled crystallographic cleavage.

flow Ilnes, indicative of a more ductile fracture mode, were clearly evident on fracture surfaces.

small cleavagelike facets (indicative of somewhat brittle fracture) appeared on the fractmw surfaces. Fatigue striations developed on fracture surfaces.

rough and ragged areas (indicative of somewhat ductile fracture) appeared on the fracture surfaces. No fatigue st-iations developed on fracture surfaces.

J

Table 4 presents a summary of the effects of vacuum and air environments on sub-surface dislocation behavior. Generally, vacuum-tested specimens exhibited lower didocation densities near slip lines than air-tested ones [6,9]. Table 4. Effects of vacuum and air environments on subsurface dislocation behavior Investigator(s)

T

Reference

Material(s) Tested

Findings

Transmission electron microscopy showed that in the layer of material very near the specimen surface the fallowing conditions existed: 6

pure aluminum single 99.999%

CZ-p+lS

Grosskreutz

9

i

EFM VoL 8, No. 2-B

99.99% pure aluminum single crystals

a) Near slip lines, fewer dislocations were present in vacuum-lested spcimens that in airtested ones. b) In the regions between slip lines, the dislocation densities were essentially the same in vacuum and air-tested specimens. a) Fine, closely-spaced clusrers of dislocations and dislocation dipoles occurred in vacuumtested specimens. The spacing between these clusters was identical to the spacing of the fine slip markings on the specimen surface. b) Dense, relatively widely-spaced clusters of dislocation dipoles occurred in air-tested specimens with especially dense clusters in the vicinity of well-developed slip lines.

318

C. M. HUDSON and S. K. SEWARD

In the Discussion section of this paper, the findings reported in this section are related to the mechanisms proposed to explain the longer fatigue lives in vacuum.

Table 5 summarizes the effect of various gas constituents on fatigue life and fatigue-crack propagation. The identification of constituents in the atmosphere which have a detrimental effect on fatigue life may be the greatest contribution of research in vacuum fatigue. Experimental evidence already indicates that the absence of deleterious constituents rather than low gas pressure is responsible for the increased fatigue life in vacuum[l]. The results of a number of investigations [Z,4, 12-231 indicate strongly that water vapor significantly reduces the fatigue resistance of a wide range of materials. In contrast, the fatigue resistance of these materials is relatively high in pure oxygen, hydrogen, nitrogen and argon. Gas evolution during fatigue cyciing

Hydfogen gas evolved during fatigue testing of aluminum alloys and carbon steels[l3]. This evolution occurred only in tests in moist air. Apparently, the moisture reacted with freshly exposed metal at slip steps, extrusions, and crack surfaces to form an oxide layer. This reaction released hydrogen ions which subsequently combined to form the hydrogen gas which was detected. Table 5. Effects of various gas constituents on fatigue life and fatigue-crack propagation: (a) Constituent-Water Vapor;(b) constituent-Oxygen: (c)Constituent-Nitrogen;(d) Constituents-Hy~ogeu and Argon.

.-. ----

. ..- Findings

1.

Shives

IlOO-HI4

4

Ben

12

and

Fatigue

Life

Investigations not

Nd/Nb

=

3

3

Nd/Nh

7

I.4

85

3

N$Nb

=

3

85

3

NdiNb

=

/ lot

aluminum

given

given

composition

85

22 brass

Bennett

AZGiA magnesium /

alloy Ti-4AI-4Mn

I’3

I

titenium

Gaugh

and

I4

copper

SODwith 2.

not

not

eiven

eiven

Fatigue-Crack-Propagation v

~~estig~t~ans

Nd/Nh

-

=

1

-

0.05

(da,‘dX:),/(daidN),

=

8

0.05

(da,‘dN),/(da,‘dN)d

IT 8

.1

(da,‘dN)‘,

=

3

(da/‘dN)b/(da/dNId

! alloy

Spitzig

2nd

2024-T3

I / 100

aluminum

;

alloy

!

0.45C-Ni-

! 1

Cr-&lo

Wei

100

/(daidN);

2

steel Shives

AZ&A

and

I85

Bennett

1 Tests

corrfurted

iii dry and humid

sxgon

environments.

m

3

319

The effects of environment on the fatigue behavior of metals Table 5 (Cod) (a) Relative investigator

Reference

Materials Tested

2.

Fatigue-Crack-Propagation 17

Feeney, McMillan, and Wei

Finding8

Humidity, ‘% Moist fJrY Investigations

- (Concluded)

90

10

(da/dN),/(da/dN),

?i 1.1

7075-T6 aluminum alloy

90

10

(d~~)~/(d~dN)d

= I. 1

?178-T6 aluminum

90

10

(da./dN),,/(da/dNjd

* 1.1

DTD 5070A aluminum alloy

not

not given

(d~dN)k/(d~dN)d

* 3

given

2024-T3

100

25

(da/dN)h/(da/dNJd

w 1

100

25

(d~dN)k/(d~dN)d

=

2024-T3 aluminum alloy

all0y Bradshaw and Wheeler

18

Sartman

15

et al.

aluminum allOg

7075-T6

15

aluminum alloy

Table 5 {Contd) fb) Investigator(s)

Reference

MateriaIs Tested

Findings

1. Fatigue-Life Broom and

I9

Nicholson

Investigations

Al-4 percent Cu aluminum alloy

N&N,

= 0

B.S. L65 aluminum alloy

No/“,

= 4

D. T.D. 683 aluminum

NdN,

= 2

soy

I

2.

Fati~e-wrack-Pr~~tion

~ye6ffga~o~

D. T. D. 5070A aluminum alloy

(da/dN)a/(da/dN),

= 3

Overaged Al-7 percent Zn -5 percent Mg -I percent Mn aluminum a&y

(da/dN&/(dajdNfo

x 2

Peak aged Al-7 percent Zn -5 percent Mg -1 percent Mn

(da/dN)a/(da/dNIo

= 8

aluminum

alloy

I

320

C. M. HUDSON and S. K. SEWARD Table5 (Contd) cc) I 1Investigator(s)

Reference

Materials

Findings

Tested

Broom and ,Nlchalson

i9

j Al-4 percent cu aluminum alloy B.S.

L65 aluminum

NN/N,

= 6

NN/N,

= 10

Table 5 (Concluded) (d) Investigator(s)

Broom and Nicholson

Reference

19

Materials Tested

Findings

Al-4 percent cu aluminum alloy

NH/N,

= 3

B.S. L65 aluminum alloy

NH/N,

^I 3

D. T. D. 683 aluminum alloy

NH/N,

= 1

Bradshaw

(da/dN),/(da/dN)H=

5

and Whel?lf?r Wadsworth

2

OFHC copper

NA/N,

= 8

Nelson and

20

pure magnesium

NA/N,

2 8

t Williams

L

Variation of fatigue life with decreasing gas pressure The variation of fatigue life with gas pressure, in the range 101 kN/m’ to 1.3 pN/m* (760 to 1 x 10m8torr), has not been clearly established. In some instances, a continuous variation occurred between fatigue life and gas pressure, in other instances, a stepped variation occurred. Figure 1 (replotted from [2]) shows a continuous variation of fatigue life with gas pressure for pure copper, pure aluminum, and an aluminum alloy. Tests on pure aluminum[24], pure aluminum crystals[7], and 2014-T6 and 7075T6 aluminum alloys [25,26] demonstrated similar continuous variations of fatigue life with decreasing gas pressure. Figure 2 (replotted from [27]) shows a stepped variation of fatigue life with gas pressure for 1100-H14 aluminum. In this figure, fatigue life was nearly constant from 101 kN/m’ (760 torr) to approx. 13 N/m2 (1 x 10-l torr), increased steadily from 13 N/m* to 13 mN/m’ (1 X 10-‘-l x 10m4 torr), and was again nearly constant for lower pressures. Tests on pure lead [24], pure magnesium[9, 191, 1lOO-H14 aluminum [4,28], and pure aluminum [291 demonstrated a similar stepped variation of fatigue life with decreasing gas pressure. In all of these latter tests, fatigue life increased stepwise with decreasing gas pressure over the approximate pressure range 13 N/m2 to 13 mN/m’ (1 x 10-‘-l x 10m4torr).

321

The effects of environment on the fatigue behavior of metals

Gas

pressure, torr

Fatigue life, cycles

Fig. 1. Continuous variation between fatigue life and gas pressure (from Wadsworth[2]),

R = - 1.

Gas pressu e, g N/m

I10-11

0

1

2

3

4

5x105

Fatigue life, CyCkS

Fig. 2. Stepped variation between fatigue life and gas pressure (from Hordon and Wright[27]), R = - 1.

Efect

of environment on the various phases of the fatigue phenomenon

Table 6 summarizes the effect of environment of the various phases of the fatigue phenomenon. Researchers generally divide the fatigue phenomenon into three sequential phases: fatigue-crack initiation, fatigue-crack propagation and fracture. The fatigue-crack initiation phase begins with the application of the first load cycle and is generally thought to continue until a macroscopic crack develops in the test specimen. The fatigue-crack propagation phase then begins and continues up to the final load cycle (i.e. the one causing failure) in the fatigue test. The fracture phase includes only the loading cycle which causes final failure of the test specimen. At least 12 investigations considered the effects of vacuum on the fatigue-crack propagation phase [2,4,11, 18-20,22,26,27,30-321. In all 12 investigations, fatigue cracks grew slower in vacuum than in air. However, in 4 of the investigations [ 11,18,22,26] the fatigue-crack growth rates in vacuum approached the rates in air at high fatigue-crack growth rates, i.e. near the end of the tests. Six investigations considered the effects of vacuum on the fatigue-crack initiation phase [2,4,5,18,19,261. In four of these six [4,5,19,26], fatigue cracks initiated more rapidly in air than in vacuum. In the remaining two[2,18], fatigue cracks initiated in the same number of cycles in both vacuum and air. One investigation considered the effects of vacuum on fracture toughness[26]. The fracture toughness of 7075-T6 aluminum alloy was the same in vacuum and air.

322

C. M. HUDSON

and S. K. SEWARD

Table6. Effectsof environmenton the variousphasesof the fatiguephenomenon

Notconsidered 1100-814 aluminum

Not considered

Pure aluminum

Not considered

2024-T3 altminum For

I.

oxN8

Not considered

tom

high dajdN values:

For low da/dN values: (da/dN)a/(da/dN)~~250 For high da/dN valuea: (~dN)a/(da/d~)~ -i 1

DTD 50TOA

aluminum alloy

4.0

uN/m2

3.0x10-8 ton

NiJNia = 1

For low da/dN

values:

(da/dN)a/(da/dN),

c 6

For high da/dN values: (da~dN)a/(d~d~)~ rJ 1 Al-2.2 percent Gil

-1.6 percent E&g a&minum alloy

low da/dN valuea: ~d~~N)a~(da/d~)~ = 5

For

For high da/dN values: (da/dN)a/(da/dN),

=1

(da/dN)a/(da/dN),

>1

For low da/dN

v&W.:

Far high da/dS values:

d?@ectof prolonged exposure to vacuum on fatigue life In one of the earlier vacuum-fat~ue studies, the inyesti~ato~[25] found that the fatigue lives of ~uminu~-alloy specimens increased by a factor of about two after sb~~-tome exposure to vacuum environment, However, the longer the vacuum-exhume time prier to testing (up to

The effects of environment on the fatigue behavior of metals

323

161hr), the smaller was the increase in fatigue life in vacuum. This investigator suggested that if the specimens were subjected to the vacuum environment for a sufficiently long time, the vacuum-fatigue life would approach the fatigue life in air. This suggestion prompted several other investigations into the effects of long-time exposure to vacuum prior to testing. Experiments on aluminum and magnesium alloys[lO, II, 181 showed however, that prolonged exposure to vacuum prior to testing (up to 282 hr) had no deleterious effect on the vacuum-fatigue resistance of these alloys. Further, the fatigue lives of two aluminum alloys actually increased with increasing vacuum-exposure time [333. MECHANISTS

Three types of mechanisms were proposed to explain the increased fatigue resistance of metals in vacuum; coldwelding, oxidation, and hydrogen embrittlement. Detailed descriptions of the various mechanisms follow.

When two uncontaminated metal surfaces, e.g. the fracture surfaces of a specimen broken in high vacuum, contact each other under compressive stress, these surfaces may coldweld back together [34]. Wadsworth and Hutchingsl81 proposed that at low gas pressures, the new fatigue-crack surface that developed under tension loading is only partially contaminated because of the limited number of con~minating gas molecules available. During the compression portion of a loading cycle, the uncontaminated portions of the surface contact each other and coldweld. The net effect is a delay in fatigue-crack propagation, and an extended fatigue life at lower gas pressures. Ishii and Weertman [30] indicated that they also believe coldwelding contributes to increased fatigue life in vacuum.

Researchers have proposed two types of oxidation mechanisms to explain the increase in fatigue life in vacuum. One type involves dislocation pileups while the other involves slip reversal. Descriptions of these mechanisms follow. Dislocation pileups. Dislocation theory[35,361 shows that when dislocations in a relatively low modulus material closely approach an interface between this low-modulus material and a higher modulus material, a force repels the disIocations from the interface. This force can pile up these dislocations near the interface. These piled up dislocations can initiate fatigue cracks[37] or inhibit blunting of existing cracks by restricting plastic deformation around the crack tip. Several investigators [6,10,29] hypothesized that dislocation pileups occur more readily in air than in vacuum, and that this occurrence causes shorter fatigue lives in air. These investigators reasoned as follows. Fatigue cycles produce slip bands and fatigue cracks in metal components. These slip bands and fatigue cracks expose uncontaminated metal to the environment surrounding the components. In the atmosphere, this newly exposed metal oxidizes almost instantaneously. The modulus of this oxide was assumed to be higher than the modulus of the base metal since the moduli of most bulk oxides are considerably higher than the moduli of the corresponding metals (see [38] for example). Researchers propose this higher modulus oxide piles up dislocations and consequently initiates and/or accelerates the growth of fatigue cracks in the base metal. Thus fatigue life is shortened. In high vacuum, formation of even a monolayer of oxygen on a freshly exposed slip step can require hours. Consequently, dislocations do not pile up at an interface to initiate and/or accelerate the growth of fatigue cracks. Thus fatigue life is longer. St@ reversal. Slip develops when metallic grains (and some non-metalIic ones) are stressed beyond their yield strength[39]. This slip involves the relative sliding of parallel planes of atoms where equivalent lattice sites are again occupied by each atom after slip. This relative motion involves only a change in next-nearest-neighbor atoms and not the fracture of atomic bonds. Slip can be eliminated in some metals if the metallic grains are subsequently stressed in the direction opposite to that which caused the initial slip. This slip elimination process is termed slip reversal. Several investigators hypothesized that the oxide film which forms on slip steps during

324

C. M. HUDSON

and S. K. SEWARD

Table 7. Proposed oxidation mechanisms explaining the increased fatigue life in vacuum INVESTIGATOR(S)

REFERENCE

Wllkov

6

and Applewhite

Sumison

-It

MECHANISMS Wilkov and Applewhite proposed

that, during fat&w

tests in vacuum, dislocations escape uniformly across newly-formed slip lines since there is no oxide This uniform escape film to retard their escape. represents a more even distribution of damage across the slip line which, WIIkov and Applewhite suggest, does not readily initiate microcracks. Consequently, During fatigue tests in air, fatigue life is longer. the oxide film forms readily, some dislocations escape in highly localized regions, and microcracks initiate readily. Consequently, fatigue life is shorter.

10

!

Sumison suggested that in vacuum the lack of available oxygen molecules retards formation of an oxide The absence filnz on newly created crack surfaces. of this film permits dislocations to pass through the crack surfaces readily. Consequently, the metal at the crack tip deforms easily and blunts the crack tip.

This blunting slows fatigue crack growth and In air, the oxide film forms, increases fatigue lives. disl%ations are impeded, less blunting occurs, fatigue cracks grow faster, and fatigue lives are shorter.

L-

Shen, Podlaseck, and Kramer

29

Kramer and Demei- 1471 hypothesized that dislocations pile up in a region near the surface of.a This pile up occurs crystal during tensile testing. because the dislocations are unable to push through (The region the oxide film at the crystal’s surface. containing this pile up was subsequently called a “debris” layer !48iJ Shen, Podlaseck, and Kramer proposed that a “debris” layer also forms m the surface regions near a fatigue crack. The dislocations in this layer interact to form voids and cavity dislocations which facilitate the grovth of the fatigue crack. In vacuum, dislocations can escape readily through the unoxidized crack surface, Consequently. the “debris” layer with its voids and cavity dislocations de~eiops slowiy under fatigue cycling, and fatigue crack growth is ~O~respoRdin~ly slow. This slower crack growth produces longer fatigue lives. In air, the oxide film and consequently the “debris” layer form readily, the voids and cavity dis!ccations develop quickly, and fatigue crack growth is accelerat?d.

Grosskreutz

49

l-

Grosskreuts and Bowles proposed that in vacuum the slip steps produced by fatigue cycling do not immediately oxidize when formed. Consequently, reverse slip can occur when the fatigue loading Under reverse changes from tension to compression. slip conditions, the rates of slip step formation, and Grosskreutz dislocation dipole formation are lower. and Bowles suggest that these lower rates constitute a lower rate of fatigue damage accumulation and In air, consequently lead to longer fatigue lives. the slip steps oxidize readily. This oxidation inhibits reverse slip and consequently accelerates slip step and dislocation dipole formation, and fatigue damage accumulation.

and Bowles

4

L

Ram proposed that in the initial stage of fatiguedamage accumulation in llOO-Xl4 aluminum fprior to crack initiation) iocalized plastic deformation occurs in some surface crystals as a result of He assumed that sane of this fatigue loading. deformation occurs in the form of reversible slip, which initially exposes clean-surface material to

325

The effects of environment on the fatigue behavior of metals Table 7 (Concluded) 1

UWESTIGATOR(S)

MECHANISMS

REFERENCE

the gas molecules of the surrounding environment. The buildup of aluminum oxide on this clean-surface is time-dependent but tends towards an e,quilibrium thickness at a given pressure; the higher ttw pwssure is, the thicker the oxide film. At a higher pressure, a sufficient oxide thickness forms the structure characteristic of A1203. Ham proposed that only this structure can be bound tightly enough to the base metal to be carried into the metal when reverse slip occurs. Rebounding of the surface is thus In vacuum, if there inhibited and a crack nucleated. is an insufficient

thickness

of oxide to form A1203,

the attached oxygen can be displaced upon reversal the slip and the rebonding of the surfaces can be Fatigue lie is, therefore, longer accomplished. since no crack has been nucleated.

of

L

fatigue cycling in the atmosphere inhibits slip reversal thereby fostering slip step formation and fatigue damage accumulation. In vacuum, no oxide film forms; consequently, slip reversal can occur and fatigue damage accumulates more slowly. Detailed discussions of the various proposed oxidation mechanisms are given in Table 7. Hydrogen embrittiement

Broom and Nicholson[l9], Interrante[40], and Bradshaw and Wheeler[l8] suggested that hydrogen embrittlement was responsible for the decreased fatigue resistance in the atmosphere. They proposed that the reaction between atmospheric water vapor and clean metal surfaces (e.g. slip steps or cracks) generates hydrogen ions which embrittle the metal. Three possible mechanisms were proposed by which these ions could affect overall fatigue behavior. In the first mechanism, the hydrogen ions diffuse into the metal and accumulate at voids which exist in the metal. The constant diffusion of these ions builds up internal pressure in the voids which facilitates crack initiation. In the second mechanism, the diffusion of hydrogen ions alters the surface energy of cracks and, thus, modifies the rate at which the cracks propagate. In the third suggested mechanism, the rate of diffusion of the hydrogen ions governs the generation of voids by controlling the clustering of vacancies formed by moving dislocations. DISCUSSION

This literature review clearly shows the fatigue resistance of a wide variety of materials is significantly higher in vacuum than in air. This increase in resistance appears to result from exclusion of atmospheric water vapor from the environment surrounding the test specimen. Three types of mechanisms have been proposed to explain this difference in fatigue resistance: (1) coldwelding; (2) oxidation, and (3) hydrogen embrittlement. No definitive conclusions can be drawn as to whether any or all 3 types of mechanisms actually influence fatigue resistance. However, some comparisons between the proposed mechanisms and the experimental results are warranted. Coldwelding

Fatigue tests in vacuum certainly satisfy the cleanliness conditions for coldwelding. In such tests, cracks produce two uncontaminated metal surfaces which contact each other during the unloading portion of the fatigue cycle. Elber [41] showed that, regardless of the stress ratio used, newly created crack surfaces contact each other on unloading. Subsequent analysis by Newman[42] showed that these surfaces experience compressive stresses approaching the yield strength of the material as a result of this contact. In vacuum, stresses of this magnitude partially coldwelded the fracture surfaces of some metals 1341.Such coldwelding would, of course, reduce fatigue-crack-propagation rates in vacuum as is reported in Table 6. Coldwelding may have indirectly caused the increased dislocation mobility observed in

326

C. M. HUDSON andS.K. SEWARD

vacuum-tested specimens (discussion of Tables 1, 3 and 4). Coldwelding per se probably would not increase dislocation mobility in a vacuum environment. However, by increasing fatigue life, coldwelding would permit more time for dislocation generation and movement prior to specimen failure. Coldwelding may indeed contribute to some extent in increasing fatigue life in vacuum. Wowever, some non-coldwelding materials exhibit longer fatigue lives in vacuum than in air. For example, ~artin[43] found the fatigue lives of 1018 and 1113 carbon steels a factor of approximately two higher in vacuum than in air. Both of these steels showed very low coldwelding capabilities however. The resutts of fatigue tests in oxygen {Table 5) further indicate ~~ldwelding is not the only mechanism responsible for increased fatigue life in vacuum. Fatigue lives in dry oxygen were considerably longer than fatigue lives in humid environments. (In some instances, fatigue lives in dry oxygen were as long as fatigue lives in vacuum[l9].) Yet most metals could coldweld in neither dry oxygen nor humid air environments since freshly exposed metal would readily oxidize in both environments~~]. Obviously, some mechanism other than coldwelding contributes to the longer fatigue lives in dry oxygen and probably also contributes to the longer lives in vacuum.

There are indications that oxidation reduces fatigue resistance in air environments- For example, gold (which does not stably oxidize) has the same fatigue resistance in air and vacuum environments [2]. Such similar resistances would be expected if oxidation is primarily responsible for decreased fatigue lives in air. Experimental evidence indicates that dislocations move less freely in air than in vacuum (discussion of Tables 1, 3 and 4). Observation of this lower mobility contributed to the development of the oxidation mechanisms proposed to explain decreased fatigue resistance in air. However, 3radhurst and Leach(451, and ~ross~entz~9] conducted elegant investigations into the elastic properties of oxide films on the surfaces of aluminum specimens. These investigators found that, in contrast to the modulus of the bulk oxide, the elastic modulus of this oxide film at atmospheric pressure is actually lower than than the modulus of the base metal. This lower-modulus oxide him should increase dislocation mobility by attracting dislocations 135,361 to the metal surface. Thus for aluminum, it appears that some mechanism other than oxide-film formation reduces dislocation mobility (and consequently fatigue resistance) in air environments. The oxidation mechanism may also be questioned in light of the finding that the fatigue lives of some materials is considerably longer in dry oxygen than in laboratory air (Table 5). Since oxidation occurs as readily in dry oxygen as in humid air, there should be no difference in these fatigue lives if oxidation significantly affected fatigue behavior, Again, some additional mechanism must con~ibute to longer fatigue lives in dry oxygen. Finally, oxidation does not appear responsible for fatigue cracks propagating faster in air than in vacuum (Table 6) since fatigue cracks also propagate faster in air than in pure oxygen (Table 5). If oxidation was the primly mechanism affecting growth, the rates in air and oxygen should be comparable.

The hydrogen embrittlement mechanism could quite possibly explain the increased fatigue life in vacuum. Consider aluminum (which has a fee lattice) for example. The reaction between water vapor (which is primarily responsible for the decreased fatigue life in the atmosphere) and clean metal yields the atomic hydrogen necessary for embrittlement[l3]. Bradshaw and Wheeler [ 181found that this reaction generates sufficient energy to drive the atomic hydrogen into the metal, and that hydrogen dif?uses through the metal fast enough to embrittle the metal ahead of a slowly propagating fatigue crack. These findings indicate that hydrogen embri~lement resulting from the water vapor-clean metal reaction is a viable explanation for ~~minum’s decreased fatigue resistance in humid en~onm~nts. The sus~eptib~ity of bee and hcp metals to hydrogen embrittlement is well documented [46]. Hydrogen embrittlement could also cause the reduced ductility observed in fatigue tests in air (discussion of Tables 1,3 and 4). In proposing a mechanism to explain hydrogen embrittlement in

The effects of environment on the fatigue behavior of metals

321

metals, Cotterill[46] suggested that hydrogen-induced dislocation barriers cause dislocation pileups which, when stressed, form embrittling microcracks. A similar immobilization of dislocations appears to occur in fatigue tests in air. For example, slip-steps, which result from dislocation movement, develop slower in air as evidenced by the generally observed decreased surface roughening in fatigue tests in air. Further, the fatigue-crack surfaces of air-tested specimens show evidence of more brittle fracture than do the fracture surfaces of vacuum-tested ones. This more brittle fracture mode is another manifestation of restricted dislocation movement. This embrittlement could be responsible for accelerating fatigue crack propagation in humid environments (Table 4). Bradshaw and Wheeler[lS] also pointed out that the solubility of gaseous hydrogen in aluminum is very low at atmospheric pressure. This low solubility indicates hydrogen would permeate and embrittle aluminum very slowly in such an environment. Such limited embrittlement could produce the longer fatigue lives and lower fatigue-crack-propagation rates reported in Table 5 for tests in gaseous hydrogen environments. CONCLUDING REMARKS The literature has been surveyed for reports on the subject of fatigue of metals in gaseous environments. The following observations are made as a result of this survey. 1. Water vapor significantly reduces fatigue resistance. 2. Three types of mechanisms have been proposed to explain the observed decreased fatigue resistance in humid environments: (1) coldwelding, (2) oxidation, and (3) hydrogen embrittlement. All three mechanisms may contribute to this decrease, but a hydrogen embrittlement mechanism appears to be predominantly responsible for the decrease in a number of metals. 3. Dislocations move more easily during tests in vacuum than during tests in air. 4. Water vapor reduces fatigue resistance more than does oxygen, hydrogen, nitrogen and argon gases. 5. Fatigue cracks always propagate slower in vacuum than in air. 6. The variation of fatigue life with decreasing gas pressure is sometimes stepped and sometimes continuous. 7. Prolonged exposure to vacuum does not adversely affect fatigue resistance. Additional research is required to definitely establish whether hydrogen embrittlement is responsible for decreased fatigue lives in the atmosphere. Recently developed procedures such as the nuclear microprobe and muon tracking techniques for determining hydrogen diffusibility and concentrations in metals should greatly aid this research and may, in time, precisely define the role of hydrogen in the fatigue phenomenon. Positive identification of the mechanism(s) responsible for decreased fatigue resistance in the atmosphere could lead to the development of alloys which are not susceptible to this mechanism(s) and which consequently would have higher fatigue resistance. The use of such non-susceptible alloys in fatigue critical structures could save initial cost, weight, and material. In addition, structures made with these alloys would be more efficient to operate, and probably would require less frequent inspection and testing. Even if intrinsically non-susceptible alloys cannot be developed, knowledge of the mechanism(s) could lead to the development of coatings which will protect metals from the deleterious components of the atmosphere. REFERENCES [l] C. Michael Hudson, Problems of fatigue of metals in a vacuum enviroment. NASA TN D-2563 (1965). [2] N. J. Wadsworth, The E#ect of Environment on Metal Fatigue. Internal Stresses and Fatigue in Metals (Eds. Gerald M. Rassweiler and William L. Grube), pp. 382-396. Elsevier, Amsterdam (1959). [3] Anon: Metric practice guide. ASTM E 380-72.(June 1972). [4] John L. Ham, The influence of surface phenomena on the mechanical properties of structural materials (Fatigue of aluminumin vacuum). NRC Proj. No. 42-l-0105(Contract No. AF 49 (638)-lOOS),Natl. Res. Corp. (Cambridge, Mass.) (1%3). [5] K. U. Snowden and J. Neil1Greenwood, Surface deformation differences between lead fatigued in air and in partial vacuum. Trans. Met. Sot. AIME 212, 626-627 (1958). [6] M. A. Wilkov and B. Applewhite, The effect of vacuum on surface damage in fatigue. EARL-RM 1039,Engineering Mechanics Research Laboratory, The University of Texas (1%7).

328

C. M. HUDSON and S. K. SEWARD

171I. R. Kramer and S. E. Podlaseck, Effect of vacuum environment on the mechanical behavior of materials, PM-102 (Contract AF 49 (638)-946), Martin-Marietta Corp. (1961). I81 N. J. Wadsworth and J. Hutchings, The effect of atmospheric corrosion on metal fatigue. Phill. Msg., Ser. 8, 3(34), I I%-1166(1958). 191J. C. Grosskreutz. The effect of oxide films on dislocation-su~ace interactions in aluminum. Surface Science 8. 173-190 (1967). [lo] H. T. Sumsion. Vacuum effects on fatigue properties of magnesium and two magnesium alloys. J. Spacecraft j(6), 70&704 (1968). [ll] D. A. Meyn. The nature of fatigue-crack propagation in air and vacuum for 2024 aluminum. Trans. ASM 61, 52-61 (1968). ]12l T. R. Shives and J. A. Bennett, The effect of environment on the fatigue properties of selected engineering alloys. J. M~~er~a~s 3(3), 695-715( 1968). 1131John A. Bennett, Fatigue-An ~a#e~disc~~~iaary Ap~~oac~ (Eds. John J. Burke, Norman L. Reed and Volker Weiss), pp. 209-227. Syracuse University Press (1964). [14] H. J. Gough and D. G. Sopwith. Atmospheric action as a factor in fatigue of metals. Engineering CXXXIV(3491). 694-969 (1932). [ 151A. Hartman, F. A. Jacobs, A. Nederveen and P. De Rijk, Some tests on the effect of the environment on the propagation of fatigue cracks in aluminum alloys. NLR-TN M.2182,National1Lucht-Enruimtevaartlaboratorium, The Netherlands (1967). [ 161W. A. Spitzig and R. P. Wei, A fractographic investigation of the effect of environment on fatigue-crack propagation in an ultrahigh-strength steel. Trans. ASM 60, 279-288 (1967). 1171J. A. Feeney, J. C. McMillanand R. P. Wei, Environmental fatigue crack propagation of aluminum alloys at low stress intensity levels. Boeing Company Report D6-60114(1969). [18] F. J. Bradshaw and C. Wheeler, The effect of environment on fatigue crack growth in aluminum and some aluminum alloys. App. Mat/s. RPS. .5(Z). 112-120(1966). [t9] Trevor Broom and Anthony Nicholson, Atmospheric corrosion-fatigue of age-hardened aluminum alloys. J. hsf. ~~e~u~s. 89, 183-190(t%l). [2OJHoward G. Nelson and Dell P. Williams, The effect of vacuum on various mechanical properties of magnesium. The effects of the space environment on materials. SAMPE Pm. 11, 291-297 (1967). [21] M. R. Achter, G. J. Danek, Jr. and H. H. Smith, Effect on fatigue of gaseous environments under varying temperatures and pressure. Trans. Met. Sot. AIME 227(6), 12961301 (1963). (221J. C. Mabberley, The effect of environment on fatigue crack propagation in age hardened aluminum alloy AI-7%Zn-5% Mg-1% Mn. RAE. Tech. Memo CPM 53 (1966). [23] N. Thompson. El. Wadsworth and N. Louat, The origin of fatigue fracture in copper. Phil. Msg.. Ser. 8 l(2), I13-126 (1956). [24] K. U. Snowden, Effect of air pressure on the fatigue of lead and alumina. Nature 189(47SB),53-54 (1961). 1251R. H. Christensen, Fatigue cracking of metals accelerated by prolonged exposure to high vacuum. Engng Paper No. 1636,Missile Space Systems Div., Douglas Aircraft Co. (1963). 1261C. Michael Hudson, Investigation of the effect of vacuum environment on the fatigue and fracture behavior of 7075-T6. Proc. of the Vacuum Metallurgy Conf. pp. 1424-1427(1972). [27] M. J. Hordon and M. A. Wright. Mechanism of the Atmospheric Interaction with the Fatigue of Metals. NASA CR-l 16.5 (1968). [28] Harold Shen, Effect of vacuum environment on the mechanical behavior of materials. MCR 67-423, Martin Co., Denver, CO. 1291H. Shen, S. E. Podlaseck and I. R. Kramer, Effect of vacuum on the fatigue life of aluminum. Acta Metallurgica 14(3), 341-346(1966). 1301H. Ishii and J. Weertman, The effect of air pressure on the rate of fatigue crack growth. Scripta Metallurgica 3,?29-232

(1969). 1311 Roger N. Wright and A. S. Argon, Fatigue crack growth in Si-Fe. Met. Trans. l(il). 3065-3073(1970). f32] J. C. Grosskreutz, Research in Mechanisms of Fatigue. Qua~erly P~ag~ess report No. 5a, M.R.I. Project No. 2279-P (Contract No. AF 33 (616)-6383),Midwest Research Institute (Kansas City, MO) (1960). 1331D. W. Hoeppner and W. S. Hyler, The effect of vacuum outgassing time on the fatigue behavior of two structural aluminum alloys. Materials Research and Standards 6(1I), 599-601(1966). [34] John L. Ham. Investigation of adhesion and cohesion of metals in ultrahigh vacuum. NRCProj. No. 42-l-0121(Contract No. NASr-48) Natl. Res. Corp., Cambridge, Mass. (1962). ]3S] A. K. Head, The interaction of dislocations and boundaries. Phil. Mug. Ser. 7 44(348),92-94 (1953). [36] G. H. Canners, The interaction of a dislocation with a coated plane boundary. Inf. J. Engng Sci. 5(l), 25-38 (1967). 1371A. N. Strok, The formation of cracks as a result of plastic flow. PTOC.Roy. Sot. London, Ser. A 223(1154),404-414 (1954). 1381J. F. Lynch, C. C. Ruderer and W. H. Duckworth, Engineering properties of ceramics: Databook to guide materials selection for structural applications. AFML-TR-66-52 (1966). [39] A. D. Merriman, A Dictionary of Metallurgy. MacDonald and Evans, London (1958). 1401C. G. Interrante, Interpretive report on effect of hydrogen in pressure-vessel steels. Section I-Basic and research aspects. WRC Bulletin 145 (1969). 1411Wolf Elber, The signi~~anceof fatigue crack closure. Damage tolerance in aircraft structures. ASTM STP 486230-242 (1970). [42] J. C. Newman, Jr., A finite-element analysis of fatigue crack closure. 8th ASTM Nat. Symp. on hcture k&h. Providence, Rhode Island (August 1974). [43] D. E. Martin, Plastic strain fatigue in air and vacuum. Trans. ASME. J. Basic Engng 87D(4),850-856 (1%5). (441 F. Jona. Preparation and properties of clean surfaces of aluminum..f. Phys. Chem. Solids 28(1l), 2155-2160(1967). [45] D. H. Bradhurst and J, S. Llewdyn Leach, The mechanical properties of anodic films on aluminum. Trans. British Ceramic Sac. 62(9), 793-806 (1963). 146)P. Cotterill, Progress in ~ateriuis Science (Ed. Bruce Chalmers), Vol. 9, pp. 201-301. Pergamon Press, Oxford (1961).

The effects of environment on the fatigue behavior of metals

329

[47] I. R. Kramer and L. .I. Demer, The effect of surface removal on the plastic behavior of aluminum single crystals. Trans. Met. Sot. AIME 221, 780-786 (1961). [48] I. R. Kramer, The effect of surface removal on the plastic flow characteristics of metals-II. Trans. Met. Sot. AIME 227, 1003-1015(1963). [49] J. C. Grosskreutz and C. Q. Bowles,, Environment-Sensitive Mechanical Behavior (Eds. A. R. C. Westwood and S. N. Stoloff), Vol. 35, pp. 67-105. Metallurgical Sot. Conf. (1965). (ReceivedJanuary

1975)

APPENDIX Conversion of SI units to U.S. customary units

The International System of Units (SI) was adopted by the Eleventh General Conference on Weights and Measures held in Paris in 1%0[3]. Conversion factors required for units used herein are given in the following table:

Physical quantity Pressure

SI Unit (a) newtons per m*(N/m*)

Conversion U.S. Customary Unit factor (b) torr 0.7500x 1om2

‘Prefixes and symbols to indicate multiples of units are as follows: Symbol Prefix Multiple n 1om9 nano 10m6 micro F m lO_’ milli k kilo 10’ bMultiply value given in SI Unit by conversion factor to obtain equivalent in U.S. Customary Unit.