Czichos H., Saito T., Smith L.E. (Eds.) Handbook of Metrology and Testing

Подождите немного. Документ загружается.

708 Part D Materials Performance Testing

In the case of cracked specimens the limiting condi-

tions are deﬁned by the critical threshold stress intensity

factor K

iscc

Tests with Constant Strain

The various types of bending tests that are in use of-

ten belong to this category. These can simulate stresses

from production which are often the cause for failures

in service conditions.

The advantage of bending tests is the use of simple

and hence often cheap specimens and clamping devices.

Problems with this kind of tests usually result from the

low reproducibility of the stress level – if indeed this

can be measured at all. Tests to improve this situation

have led to reﬁned types of bending tests, e.g. four-point

loading instead of three-point loading. But the limita-

tions of the simple bending theory which is usually used

to calculate the stress level can lead to errors, especially

if straining beyond the elastic limit is necessary. The ap-

plication of strain gauges to measure the stresses at the

surface can be useful. The preparation of bend-shaped

parts for U-formed bending specimens introduces sig-

niﬁcant plastic deformation which could affect crack

initiation.

Tensile tests with constant strain are sometimes

preferred to bending tests as the application and calcu-

lation of the stress is simpliﬁed. However, more massive

stressing frames are required than for bending speci-

mens of similar cross section.

Apart from affecting the value of the maximum ini-

tial load the stiffness of the frame can also affect the

time to failure of the specimen. In most of the tests with

constant strain, especially when testing ductile mater-

ials, the initial elastic elongation is partly converted into

plastic deformation as the crack formation proceeds.

The magnitude of any loading relaxation can dif-

fer from specimen to specimen and can affect the time

to failure of the specimen depending on the number of

pits or cracks that are present. In the case of specimens

with many cracks or pits considerable relaxation can

be observed while in the presence of only a few cracks

little relaxation can be observed. Thus if only one sin-

gle crack occurs it does not necessarily have to grow

to a considerable size for sudden failure (fracture) of

the specimen to occur as the applied load will remain

high. On the other hand the signiﬁcant relaxation of the

loading that takes place in the presence of many stress

corrosion cracks means that these cracks have to pro-

ceed further to reach a size that will be sufﬁcient to

generate stress conditions for a sudden fracture to occur

at relatively low loading.

Tests with Constant Load

These tests are more suitable for simulating failures

due to stress corrosion cracking brought about by loads

or stresses occurring in practice. As the effective cross

section of a specimen is reduced by propagating crack

formation tests with constant load include an intensi-

ﬁed stress situation. Hence, it is more probable that such

tests will lead to an earlier failure or fracture compared

to tests with constant total elongation.

Tests with constant load lead to an increasing stress

situation as the initiated cracks continue to grow and

therefore it is more unlikely that cracks, once started,

will be arrested in these tests than those in a constant

strain test below the critical stress threshold. As a conse-

quence, in any particular system the value for the critical

stress threshold is probably lower when determined un-

der constant load than under constant strain.

Tests with Slow Strain Rate

The application of dynamic slow straining, which orig-

inally was only employed as a quick selection test, has

now become more important for simulating conditions

in practice.

This method includes a relatively slow strain or

deformation rate (e.g. 10

−6

−1

) of a specimen under

deﬁned environmental conditions, which is continued

until fracture occurs.

For stress corrosion cracking, crack propagation

rates are generally in the range of 10

−3

to 10

−6

mm s

−1

which means that in laboratory tests under constant to-

tal strain or constant loading failure of specimens with

usual dimensions will take place within a few days.

This has also been observed in practice in those systems

where stress corrosion cracking can easily be initiated.

On the other hand it is common experience that speci-

mens do not fail even after very long testing times; for

such situations the test will be stopped after an arbitrary

time period.

The ﬁrst applications of this test procedure were

aimed at getting data to compare the effects of such vari-

ables as material composition and microstructure or the

addition of inhibitors to stress corrosion cracking initi-

ating environments. It was also used to initiate stress

corrosion cracking for material/environment combi-

nations which did not show crack formation under

constant load or constant total strain in the laboratory.

This means that this test is relatively severe as it often

initiates failure of the specimen due to stress corrosion

cracking where other types of loading of smooth spec-

imens did not lead to crack formation. In view of that

it must be classiﬁed in the same category as tests with

Part D 12.6

Corrosion 12.6 Corrosion with Mechanical Loading 709

precracked specimens where the failure of the specimen

is also inherent to the test.

Composition of Test Solution

Although it is unavoidable that the medium is a very

important factor for SCC tests, some of the solutions

are often used in combination with special alloys. Two

examples are boiling MgCl solution for stainless steels

and boiling nitrate solutions for carbon steels. Such so-

lutions have been criticized for various reasons; most

importantly because they do not represent practical ser-

vice conditions. This can be very signiﬁcant, because

relative susceptibilities of a series of materials are not

always the same in different media.

The effect of a change in the pH value is well known

in the case of general corrosion in electrolytes, and

as far as corresponding experiments exist, such effects

are no less obvious with SCC. Changes of media pH

value during a test can be as important as the initial pH

value. Any change of pH value during a test will depend

on solution volume, surface area of exposed specimen

and time. Use of a relatively large solution volume

combined with a small surface area of the metal or re-

plenishment or exchange of solution during the test is

likely to lead to a smaller pH change and therefore a dif-

ferent time to failure of the test specimen than would be

the case with a small solution volume and large metal

surface.

Small changes of O concentration can have signiﬁ-

cant inﬂuence where oxygen plays an important role in

corrosion reactions causing crack initiation.

Electrochemical Aspects

Free corrosion experiments, i. e. at the free corrosion

potential, can be used for reference and serial exper-

iments. They are always problematic if the corrosion

medium does not contain a deﬁned redox system with

enough redox capacity since electrochemical reactions

can change with time. Controlled electrochemical ex-

periments can be used to determine the dependence of

SCC on potential and to obtain information on the effect

of system parameters on the potential. These parameters

include, for instance, the redox systems of aggres-

sive environments and, under special circumstances, the

alloying components of the metal. Controlled electro-

chemical experiments have been used to achieve a re-

duction in the time to failure of the specimen by setting

the test electrochemical conditions to give maximum

susceptibility to SCC. In such cases care has to be taken

that the corrosion type and mechanism do not change.

This can be examined by metallographic analysis.

Discussion and Evaluation of Results

In tests under constant load, the SCC sensitivity of dif-

ferent materials can only be assessed by determination

of the following values.

•

Fracture: yes/no;

•

Cracks: yes/no;

•

Depth and amount of cracks.

An assessment of relative resistance to SCC of different

materials based only on the time to failure is not possi-

ble, for the reasons given above. Tests with slow strain

rate provide the following evaluation possibilities for

determining SCC susceptibility. For experiments con-

tinued until fracture

(a) depth of secondary cracks;

(b) elongation after fracture;

(d) breaking load.

For experiments ﬁnished before fracture

(e) experimental duration and measured crack depth.

Evaluation regarding (a) and (e) is usually more sensi-

tive than (b–d). Information regarding the SCC behavior

of a system can more easily be obtained if the mechan-

ical values (b–d) are related to values measured in an

inert environment (air, oil).

In general, a metallographic evaluation of the speci-

men following the experiment is necessary to determine

reliably the resistance to SCC of metallic materials using

the slow-strain-rate test method. For other types of tests

metallography is also in many cases the only method for

the reliable detection of SCC and for distinguishing dif-

ferent types of SCC. The average or maximum depth of

secondary cracks and resulting crack growth rates can be

used as the basis for comparison tests.

12.6.2 Corrosion Fatigue

Corrosion fatigue is a process which involves conjoint

corrosion and alternating straining of the metal, often

leading to cracking. It may occur when a metal is sub-

jected to cyclic straining in a corrosive environment.

Corrosive or otherwise chemically active environments

can promote the initiation of fatigue cracks in metals

and alloys and increase the rate of fatigue crack propa-

gation. Corrosion fatigue processes are not limited to

speciﬁc metal/environment systems and reliable esti-

mates of fatigue life for all combinations of loading and

environment cannot be made without data from labora-

tory tests.

Part D 12.6

710 Part D Materials Performance Testing

Regarding the parameters involved in the process

there are some similarities between stress corrosion

cracking and fatigue but especially the mechanical load-

ing is different. For testing a metal’s resistance to

corrosion fatigue in general two types of test have been

developed: cycles to failure testing (ISO 11782-1) and

crack propagation testing (ISO 11782-2).

Cycles to Failure Testing

In the presence of an aggressive environment the fa-

tigue strength of a metal or alloy is reduced to an extent

which depends on the nature of the environment and the

test conditions. For example, the well-deﬁned fatigue

strength limit observed for steels in air may no longer

be evident as illustrated in Fig. 12.42. Interpretation of

results is then based on the assumption of an acceptable

life of the component.

The test involves subjecting a series of specimens

to the number of stress cycles required for a fatigue

crack to initiate and grow large enough to cause failure

during exposure to a corrosive or otherwise chemically

active environment at progressively smaller alternating

stresses in order to deﬁne either the fatigue strength at N

Cyclic stress amplitude S

(MPa)

400

200

600

Number of cycles to failure N

Fig. 12.42 Schematic comparison of S–N behavior during

fatigue and corrosion fatigue for steel. Key: 1 fatigue in

air, 2 fatigue limit in air, 3 corrosion fatigue (no fatigue

limit)

cycles, S

, from an S–N diagram or the fatigue strength

limit as the fatigue life becomes very large.

The test is used to determine the effect of environ-

ment, material, geometry, surface condition, stress, etc.,

on the corrosion fatigue resistance of metals or alloys

subjected to applied stress for relatively large numbers of

cycles. The test may also be used as a guide to the selec-

tion of materials for service under conditions of repeated

applied stress under known environmental conditions.

Specimens

The design and type of specimen used depends on the

fatigue testing machine used, the objective of the fatigue

study and the form of the material from which the spec-

imen is to be made. Fatigue test specimens are designed

according to the mode of loading, which can include

axial stressing, plane bending, rotating beam, alternate

torsion or combined stress.

Specimens may have circular, square, rectangular,

annular or, in special cases, other cross sections.

The gripped ends may be of any shape to suit the

holders of the test machine. Problems may arise unless

the gripped portion of the specimen is isolated from the

corrosive test environment.

The test section of the specimen shall be reduced

in cross section to prevent failure in the grip ends and

should be of such a size as to use the middle to upper

ranges of the load rating of the fatigue machine to op-

timize the sensitivity and response of the system. The

transition from the gauge section to the gripped ends of

the specimen shall be designed to minimize any stress

concentration. It is recommended that the radius of the

blending ﬁllet should be at least eight times the speci-

men test section diameter or width. The cross-sectional

area of the gripped ends should, where possible, be at

least four times that of the test section area. The test

section length should be greater than three times the test

section diameter or width.

For tests run in compression, the length of the test

section shall be less than four times the test section

diameter or width in order to minimize buckling. For

the purposes of calculating the load to be applied to ob-

tain the required stress, the dimensions from which the

area is calculated shall be measured to within 0.02 mm.

Specimens should be identiﬁed by an indelible marking

method, such as stamping, on surface areas, preferably

on the plain ends, without having an inﬂuence on the

test results. Specimens should be stored after appropri-

ate cleaning under desiccated conditions prior to testing

in order to avoid corrosion which may inﬂuence the test

results.

Part D 12.6

Corrosion 12.6 Corrosion with Mechanical Loading 711

Cylindrical Specimens

Two types of specimens with circular cross section are

frequently used for corrosion fatigue tests.

1. Specimens with tangentially blending ﬁllets be-

tween the test section and the grip ends; these are

suitable where axial loading is employed;

2. Specimens with a continuous radius between the

grip ends with the minimum diameter at the center;

these are suitable for rotating bend tests.

A minimum cross-sectional diameter of 5 mm is

preferred.

Flat Sheet or Plate Specimens

Flat specimens for fatigue tests are reduced in width in

the test section and may have thickness reductions.

If the specimen thickness is less than 2.5 mm, and

the tests are performed in compression, provisions for

lateral support should be made to prevent buckling with-

out affecting the applied load by more than 5%.

The most commonly used types include

1. specimens with tangentially blending ﬁllets be-

tween the test section and the grip ends;

2. specimens with a continuous radius between the

grip ends.

Notched Specimens

The effect of machined notches on corrosion fatigue

strength can be determined by comparing the S–N

curves of notched and unnotched specimens.

The data for notched specimens are usually plotted

in terms of nominal stress based on the net cross section

of the specimen.

The effectiveness of the notch in decreasing the fa-

tigue limit is expressed by the fatigue notch factor K

(the ratio of the fatigue limit of unnotched specimens to

the fatigue limit of notched specimens).

The notch sensitivity of a material in fatigue is ex-

pressed by the notch sensitivity factor q

q =

−1

, (12.78)

where K

is the stress concentration factor; q = 0for

a material that experiences no reduction in fatigue limit

due to a notch; q = 1 for a material where the notch

exerts its full theoretical effect.

If the standard size cannot be met other specimen

conﬁgurations can be used with appropriate caution.

Specimen Size Effects, Surface Condition

and Environmental Considerations

Size effects can be important in fatigue for several rea-

sons, including residual stress distribution, variations in

the stress gradient across the diameter (plain or notched

specimens in bending or tension and notched specimens

in axial tension–compression loading), variations in

surface area, variations in hydrogen concentration gra-

dient (in appropriate environmental conditions). There

is a tendency for the fatigue strength to decrease as the

specimen size increases but this may not always be the

case. The size effect means that it can be difﬁcult to

predict the fatigue performance of large components

directly from the results of laboratory tests on small

specimens.

Corrosion fatigue properties can be very sensitive

to surface condition since fatigue cracks usually initiate

at the surface. In general, fatigue life increases as the

magnitude of the surface roughness decreases. There-

fore, attention must be paid to the surface preparation of

corrosion fatigue test specimens. Unless it is required to

investigate the behavior of an as-manufactured surface,

it is recommended that a metallographically polished

surface free of machining grooves and scratches should

be used. The direction of polishing can be important

and in axial loading the ﬁnal surface preparation should

involve grinding, or polishing in the longitudinal direc-

tion, i. e. in the same direction as the applied stress.

Other important parameters which must be consid-

ered in the test design and evaluation include surface

and residual stress and microstructural surface effects.

More detailed information on these topics is given in the

relevant standards. For environmental factors all con-

siderations presented in Sect. 12.5.1 on stress corrosion

cracking apply.

Stressing Considerations

Cyclic Frequency. Cyclic frequency is of far greater im-

portance when cycles to failure tests are conducted in

aggressive environments rather than in air, where cyclic

frequency usually has little, if any, effect. This sensi-

tivity to frequency is due to time-dependent processes

associated with the material–environment interaction.

In a narrow context this may simply reﬂect the timescale

of overall testing for signiﬁcant pit development but

more broadly the cycle period affects the extent of reac-

tion or transport during a load cycle and consequently

the extent of crack advance.

In the presence of an aggressive environment the fa-

tigue strength of a metal or alloy generally decreases as

the cyclic frequency is reduced. It is important, there-

Part D 12.6

712 Part D Materials Performance Testing

log (da/dN)

log ΔK

fore, that a cyclic frequency relevant to the service

application be used during testing.

Waveform

In some cases, corrosion fatigue strength is strongly af-

fected by the waveform of the loading cycle. This is par-

ticularly so where the cycle incorporates hold times dur-

ing which time-dependent corrosion or stress corrosion

processes may inﬂuence crack initiation and growth.

The rate of loading and unloading may also inﬂu-

ence environmental effects if these involve, for exam-

ple, diffusion or repassivation processes. It can be im-

portant, therefore, to employ a waveform representative

of that encountered during service. Sinusoidal, triangu-

lar, sawtooth and square waveforms are often employed

to simulate service loading conditions and, where ap-

propriate, hold times can be imposed during the cycle.

Variable-Amplitude Loading Patterns

Some practical applications involve exposure to random

loading cycles or to well-deﬁned periodic changes in

the cyclic loading conditions. While some insight into

Reference plane

Fig. 12.44 Three or four-point single-

edge notch bend specimen (SENB3)

Fig. 12.43 Corrosion fatigue crack propagation rate as

a function of the range of the stress intensity factor 

the inﬂuence of these ﬂuctuations may be gained by the

summation of the effects observed during a series of

tests under different loading conditions, it is preferable

to simulate the service conditions by computer control

using block or random loading programs.

Test Report

The test report should include the following informa-

tion.

1. Specimen design, dimension, machining processes

and surface condition;

2. For notched specimens, details of the notch and its

stress concentration factor;

3. Description of the test machine, including the

method of veriﬁcation of dynamic load monitoring;

4. Test material characterization in terms of, for exam-

ple, chemical composition, melting and fabrication

process, heat treatment, microstructure, grain size,

nonmetallic inclusion content and mechanical prop-

erties; product size and form shall also be identiﬁed;

the method of stress relief, if applicable;

5. Specimen orientation and its location with respect

to the parent product from which it was removed;

6. Test loading variables, including stress amplitude

and stress ratio, fatigue life or cycles to end

of test, cyclic frequency and waveform for each

specimen;

7. The initial solution composition, pH, degree of aer-

ation (or concentration of other relevant gases),

ﬂow conditions, temperature and electrode poten-

tial; speciﬁcation of ﬂow rate should be in terms

of approximate linear rate past the specimen if de-

termined by the recirculation rate; the reference

electrode used shall be indicated; the potential shall

be reported and referred to an appropriate standard

electrode (for example, the standard hydrogen elec-

trode or saturated calomel electrode at 25

◦

C);

Part D 12.6

Corrosion 12.6 Corrosion with Mechanical Loading 713

8. The starting procedure for the test, for example any

change in initial electrode potential;

9. Transients in the environment or in the loading (in-

cluding test interruptions) during testing, noting the

nature and duration;

10. Description of the environmental chamber and all

equipment used for environmental monitoring or

control;

11. Failure criterion;

12. An S–N diagram plotting the maximum stress,

minimum stress, stress range or alternating stress

against the number of cycles to failure; it is

conventional to plot fatigue life, N, in cycles log-

arithmically on the abscissa while stress is plotted

arithmetically or logarithmically on the ordinate; all

data should be plotted in the S–N diagram along

with a best-ﬁt regression analysis line; this proce-

dure develops the S–N diagram for 50% probability

of survival when the logarithms of the lives are de-

scribed by normal distribution.

Crack Propagation Testing

Using Precracked Specimens

This part describes the fracture mechanics method of

determining the crack growth rates of preexisting cracks

under cyclic loading in a controlled environment and

the measurement of the threshold stress intensity factor

range for crack growth below which the rate of crack

advance falls below some deﬁned limit agreed between

parties.

Principle of Corrosion Fatigue Crack

Propagation Testing

A fatigue precrack is induced in a notched specimen

by cyclic loading. As the crack grows the loading con-

ditions are adjusted until the values of AK and R are

appropriate for the subsequent determination of AK

Reference plane Notch

Fig. 12.46 Center-cracked tension

specimen (CCT)

Reference plane

Fig. 12.45 Compact tension (CT) specimen

crack growth rates and the crack is of sufﬁcient length

for the inﬂuence of the notch to be negligible.

Corrosion fatigue crack propagation tests are then

conducted using cyclic loading under environmental

and stressing conditions relevant to the particular ap-

plication. During the test, crack length is monitored as

a function of elapsed cycles. These data are subjected

to numerical analysis so that the rate of crack growth,

da/dN, can be expressed as a function of the stress-

intensity factor range AK.

Crack growth rates presented in terms of AK are

generally independent of the geometry of the specimen

used. The principle of similitude allows the comparison

of data obtained from a variety of specimen types and

allows da/dN versus AK data to be used in the design

and evaluation of engineering structures provided that

appropriate mechanical, chemical and electrochemical

test conditions are employed. An important deviation

from the principle of similitude can occur in relation

Part D 12.6

714 Part D Materials Performance Testing

Reference plane

Nominal 60°

Root radius

0.1 max.

Root radius

0.1 max.

0.25 max.

Fig. 12.47 Permitted notch geometry

to short cracks because of crack-tip chemistry differ-

ences, microstructurally sensitive growth and crack-tip

shielding considerations.

The threshold stress-intensity factor range for cor-

rosion fatigue, AK

may be higher or lower than

the threshold in air depending on the particular

metal/environment conditions. It may be determined by

a controlled reduction in load range until the rate of

growth becomes insigniﬁcant for the speciﬁc applica-

tion. Practically, from a measurement perspective it is

necessary to assign a value to this.

Results of corrosion fatigue crack growth-rate tests

for many metals have shown that the relationship be-

tween da/ dN and AK can differ signiﬁcantly from

the three-stage relationship usually observed for tests

in air, as shown in Fig. 12.43. The shape of the curve

depends on the material/environment system and for

some cases time-dependent (as distinct from cycle-de-

pendent) cracking modes can ensue which can enhance

crack growth producing frequency-dependent growth-

rate plateaux as shown in Fig. 12.43.

Specimen Design

A wide range of standard specimen geometries of the

type used in fracture toughness testing may be used

(see ISO 11782-2). The particular type of specimen se-

lected will be dependent upon the form of the material

to be tested and the conditions of test. Pin-loaded spec-

imens such as compact tension (CT) specimens are not

suitable for tests with R values of zero or less than

zero because of backlash effects. For such purposes

four-point single-edge notch bend (SENB4) or center-

cracked tension (CCT) specimens loaded by friction

grips are suitable.

A basic requirement is that the dimensions of the

specimens be sufﬁcient to maintain predominantly triax-

ial (plane strain) conditions in which plastic deformation

is limited in the vicinity of the crack tip. Experience with

fracture toughness testing has shown that for a valid K

measurement a, B and (W −a) should not be less than

2.5





, (12.79)

where σ

is the yield strength.

It is recommended that a similar criterion be used

to ensure adequate constraint during corrosion fatigue

crack growth testing where K

max

is substituted for K

in the above expression.

Specimen geometries which are frequently used for

corrosion fatigue crack growth rate testing include the

following.

1. Three-point single-edge notch bend (SENB3)

2. Four-point single-edge notch bend (SENB4)

3. Compact tension (CT)

4. Center-cracked tension (CCT).

Details of standard specimen designs for each of

these types of specimen are given in Figs. 12.44–12.46

and permitted notch geometries are given in Fig.12.47.

ISO 11782-2 is recommended for speciﬁc and de-

tailed information regarding test procedure, determina-

tion of corrosion fatigue crack propagation rates and

stress intensity factor range where an overview on meth-

ods for measuring crack lengths is also

available.

12.7 Hydrogen-Induced Stress Corrosion Cracking

Hydrogen-induced stress corrosion cracking or hydro-

gen embrittlement is a process where the uptake of

hydrogen into a material leads to brittle cracking of the

material under nominally uncritical loads. Its display

Part D 12.7

Corrosion 12.7 Hydrogen-Induced Stress Corrosion Cracking 715

has many similarities in comparison to stress corrosion

cracking but there are also certain differences, namely

that it most frequently in high-strength materials and

the brittle fracture must be traced back to a different

mechanism.

12.7.1 Electrochemical Processes

In the case of hydrogen-induced stress corrosion crack-

ing or hydrogen embrittlement the adsorbed hydrogen

at the steel surface is responsible for the crack growth.

As hydrogen is mostly generated by the cathodic par-

tial reaction of the corrosion process this kind of

corrosion mechanism is also called cathodic stress

corrosion cracking. A prerequisite for the absorption

of hydrogen into the metal is that the hydrogen is

present in a dissociated form. The practical importance

of hydrogen-induced stress corrosion cracking lies in

the fact that, even under otherwise harmless humidity

conditions (condensation) on the steel surface, minor

corrosion reactions could provide sufﬁcient hydrogen

to initiate crack growth. Experimentally determined

crack growth of stressed high-strength steels in low-

aggressive media such as humid air or distilled water

have shown that even small amounts of hydrogen are

sufﬁcient to embrittle these steels [12.72].

For high-strength steels this type of corrosion does

not require a speciﬁc medium. Only a sufﬁcient amount

of atomic hydrogen capable of adsorption is neces-

sary and this can be formed by the Volmer reaction

in acid and neutral media or by water dissociation in

alkaline media. In both cases the hydrogen evolution

is quantitatively coupled to the anodic partial reac-

tion (iron dissolution) of the corrosion process even

though a considerable metal loss does not necessar-

ily take place. Often pitting corrosion due to chlorides

and subsequent acidiﬁcation of the electrolyte within

the corrosion pits is the cause for hydrogen evolution

(pitting-induced stress corrosion cracking). In this case

there is an overlapping of local acid corrosion with

local cathodic hydrogen evolution and anodic metal dis-

solution. This has led to the idea of anodic hydrogen

embrittlement. During anodic crack growth hydrogen

evolution takes place in the electrolyte at the crack tip

and atomic hydrogen could then enter the metal lattice

leading to embrittling processes [12.73,74].

In fracture mechanics investigations too it has been

observed that hydrogen-induced fracture mechanisms

exist not only with cathodic polarization but also with

anodic polarization [12.75]. Even with passive steels

in aqueous alkaline solutions hydrogen evolution has

been proved experimentally to be the result of the ca-

thodic partial reaction of the dissolution of passive

iron [12.76]. On the other hand, the free corrosion

potential is strongly dependent on the aeration condi-

tions. Measurements of the free corrosion potential of

prestressing steel in saturated Ca(OH)

solution gave

values between −0.79 V normal hydrogen electrode

(NHE) with nitrogen saturation and −0.03 V(NHE)

with aeration. In oxygen-free solutions the range of

free corrosion potentials and the range of cathodic hy-

drogen evolution potentials overlap and hydrogen can

consequently also be generated at prestressing steels in

alkaline solutions. Even small amounts of cathodically

evolved hydrogen have resulted in strong changes of the

fracture properties determined in constant-extension-

rate tests (CERT) [12.77].

The hydrogen reduction reaction can be split into

different steps. In the ﬁrst step protons are discharged at

the phase boundary

+ e =H

. (12.80)

H atoms adsorbed at the metal surface recombine in

a second step according to the Tafel equation

(12.81)

or the Heyrovsky reaction

+ e =H

. (12.82)

molecules formed by this reaction can escape to the

atmosphere. With kinetic inhibition of the recombina-

tion adsorbed hydrogen atoms build up near the surface

and can lead to high H activities, resulting in the pene-

tration of hydrogen into the metal

→H

. (12.83)

An equilibrium exists between the concentration of ad-

sorbed hydrogen at the surface and dissolved H atoms

in the metal matrix. With an increasing degree of cover-

age by adsorbed atomic hydrogen the probability of H

absorption increases.

For hydrogen-induced corrosion the activity of the

adsorbed hydrogen and not the evolution rate of hy-

drogen is the main determining factor. Hydrogen only

leads to a chemical reaction with the material, or to

a metal–physical hydrogen absorption if additional crit-

ical conditions exist with respect to the state and type of

material, environment, electrochemical conditions and

mechanical loading.

Part D 12.7

716 Part D Materials Performance Testing

12.7.2 Theories of H-Induced Stress

Corrosion Cracking

From the metal physics point of view there are four

main theories of H-induced corrosion [12.78]. The basic

principles of these different mechanisms will be brieﬂy

described below.

Pressure Theory

The ﬁrst theory to explain the mechanisms of H-induced

corrosion was the pressure theory [12.79]. It is based on

the recombination of atomic hydrogen at internal sur-

faces of the metal especially at sharp-edged inclusions.

This results in high pressure, which can initiate pore for-

mation or microcracks. The pressure theory is able to

explain the formation of blisters and H-induced internal

cracks.

Adsorption Model

The basic idea of the adsorption model is that, as a re-

sult of the adsorption of certain species, e.g. atomic

hydrogen, at the crack tip the surface energy is re-

duced [12.80]. The reduction of the surface energy

lowers the critical stress which is necessary for crack

propagation. An improvement of this model that is

mainly based on thermodynamical considerations was

achieved using an atomistic point of view. According to

this, the binding energy of the atoms forming the crack

tip is reduced by adsorption of certain species; brittle

fracture is favored when the cohesive strength of the lat-

tice drops below the critical shear stress at the crack tip.

But the applicability of this model to explain H-induced

material fracture is limited as H-induced crack nuclei

are not formed at the surface of the crack tip itself but

within the material in the vicinity of the crack tip due to

the concentration of stresses there.

Dislocation Theory

Today it is well accepted that atomic hydrogen is en-

riched in areas of high dislocation densities and can be

transported together with the dislocations through the

metal lattice when tensile loading of the metallic ma-

terial is present [12.81]. A possible cause of H-induced

material damage arising from the interaction of hydro-

gen and dislocations could be the diffusion of hydrogen

into the dilatation zone of the dislocation and, by form-

ing a Cottrell cloud, restraining the dislocation mobility,

changing the gliding mechanisms and handicapping the

deformation behavior, especially the deformation abil-

ity in front of the crack tip due to strengthening and thus

favoring brittle fracture.

The interaction of dissolved hydrogen and dislo-

cations is often taken either as the only cause or in

combination with other mechanisms of H-induced cor-

rosion to explain H-induced material fractures. The

increased hydrogen mobility due to the transport via

dislocations is especially important for face-centered-

cubic metals due to the low diffusion rate of the

interstitially dissolved hydrogen atoms in these lattice

structures.

As a result of tests with high-strength martensitic

steels which indicated an increased dislocation mobility

after hydrogen uptake the slip softening model was de-

veloped. According to this model the preferential crack

growth of high-strength steels is explained by the in-

creased dislocation mobility resulting from hydrogen

uptake, which facilitates the deformation ability.

Decohesion Theory

The decohesion theory proposed by Troiano [12.82]and

further improved by Oriani [12.83] differs from the

adsorption model in that it considers the formation of

fracture nuclei in the vicinity of the crack tip within the

bulk metal. The basic idea of this model is that the bind-

ing force of the metal atoms in the metal lattice can be

reduced as a result of interactions with atomic hydrogen

and then, in combination with high mechanical stresses,

result in pure elastic rupture of the material. Hydrogen

absorbed by the metallic material diffuses to areas of

high stresses which are present at notches and therefore

also in front of the crack tip because of the multiaxial

state of stress. The area with the highest stresses is the

boundary between the plastically deformed skin layer

and the elastically deformed center. With sufﬁcient hy-

drogen concentration the cohesion forces between the

single atomic layers in the lattice are reduced so that

crack initiation and crack growth are possible. Crack

initiation occurs near the surface but contrary to anodic

stress corrosion cracking not at the surface in contact

with the medium. Critical hydrogen contents are most

likely to occur in areas of high lattice defect density, i. e.

at dislocation pile-ups, and at phase or especially grain

boundaries vertically oriented to the acting stress. The

crack propagation develops stepwise. For each prop-

agation step enough hydrogen must be accumulated

through diffusion into the region of the highest triaxial

state of stress that occurs at the transition from the plas-

tically deformed zone at the crack tip to the elastically

stressed center. This produces an additional crack which

then grows and coalesces with the initial crack. The

decohesion theory enables material ruptures to be ex-

plained even with low hydrogen absorption. Among the

Part D 12.7

Corrosion 12.7 Hydrogen-Induced Stress Corrosion Cracking 717

different hypotheses put forward to explain hydrogen-

induced embrittlement, the decohesion theory seems to

be the most reasonable for high-strength prestressing

steels.

12.7.3 Environment

and Material Parameters

Since hydrogen is adsorbed in the atomic and not

the molecular form its uptake will be substantially in-

creased by the presence of agents (promoters) which

hinder the recombination of hydrogen atoms to hydro-

gen molecules by reaction [12.80]. Effective promoting

agents are certain compounds of the ﬁfth and sixth main

groups of the periodic system, e.g. As, S, Se and Te

compounds. Sulﬁdes and thiocyanates have been iden-

tiﬁed as technically relevant substances in connection

with prestressing steels. Even small amounts of contam-

inants or other active substances in building materials

can substantially accelerate stress corrosion cracking of

susceptible steels [12.84].

Because of its low atomic diameter hydrogen forms

an interstitial solid solution with iron, the hydrogen

atom releasing its electron to the electron gas of the

metal and so remaining present as a proton. The low

diameter and the interstitial position of the hydrogen

atoms as well as the high mobility of protons lead to

a high diffusibility of hydrogen within the iron lat-

tice. The dissolved hydrogen can interact with itself,

with dislocations and with other lattice defects and

thus inﬂuence the mobility of the dislocations in the

metal which, in turn, can change the deformability of

the material. For the damage mechanism of hydrogen-

induced stress corrosion cracking the actual source of

the hydrogen is of minor importance. The damage prob-

ably proceeds via a two-step process. In the ﬁrst step

hydrogen can be picked up from the material in the

absence of mechanical loading, and then, in a second

step, failure occurs after stress is applied. The extent

of the corrosion damage is determined by the hydrogen

activity.

Hydrogen solubility depends on lattice structure,

amount of internal surfaces and temperature. Internal

surfaces are interfaces of cracks, pores and other voids

as well as phase boundaries [12.85]. More hydrogen

can be dissolved in face-centered-cubic lattices (by

a factor of about three) than in body-centered-cubic lat-

tices. This results in a lower hydrogen susceptibility for

austenitic steels. The diffusion coefﬁcient is also depen-

dent on the lattice structure. At a certain temperature in

a body-centered-cubic lattice it is from 3–6 orders of

magnitude higher than in a face-centered-cubic lattice.

Therefore under critical conditions there is a higher risk

of hydrogen-induced stress corrosion cracking for fer-

ritic and martensitic steels. Those structural anomalies

of a perfect iron lattice with which hydrogen can react

are often called traps. These are disturbances of the lat-

tice which can trap hydrogen and hold it for a certain

period of time.

As well as the parameters mentioned above hydro-

gen-induced stress corrosion cracking and propagation

also depend on hydrogen concentration and tensile

stresses. With high tensile stresses susceptible steels re-

quire only very low hydrogen activity, to initiate crack

growth [12.86].

Hydrogen-induced stress corrosion cracking of

high-strength steels of the quenched and tempered type

is characterized by cracks with intergranular corrosion

along the former austenite grain boundaries.

12.7.4 Fractographic

and Mechanical Effects of HISCC

The process of hydrogen-induced brittle fracture can

be described as follows. Initially, the change of ten-

sile stress immediately in the crack initiation zone is so

slow that hydrogen can diffuse to the crack tip and act

decohesively. The consequence is to produce a stable

intergranular fracture along the grain boundaries which

act as weak spots of the microstructure. Ductile regions

and micropits can be observed at the grain surfaces.

With increasing crack propagation rate hydrogen can-

not diffuse sufﬁciently to the crack tip and an area with

a dimple fracture appearance is formed. As the fracture

develops unstable crack propagation takes place with

the formation of a radial structure (quasi-cleavage frac-

ture area) and ﬁnally, when the monoaxial state of stress

is reached, the formation of a shear stress lip, which is

characterized by the appearance of a fracture with ﬁne

dimples.

The effect of hydrogen is mainly to reduce the bind-

ing energy of the metallic atoms thus favoring crack

initiation and propagation. When the crack growth has

reached a level where the diffusion of hydrogen to the

crack tip is no longer possible the further crack progress

is similar to a nonembrittled sample. In some cases in

the area of the radial structure hydrogen can also inﬂu-

ence the structure of the fracture surface, i. e. sometimes

the cleavage fracture will show clear signs of intergran-

ular components.

When assessing hydrogen embrittlement of pre-

stressing steels the type of mechanical loading has

Part D 12.7