Volume 13 - Corrosion Part 4 pptx

cessation of the crack growth. It is assumed that if it were not for the intervention of the corrosion product wedging the
curve would proceed to an arrest.

Fig. 47 Schematic of the variable effects of corrosion product wedging on SCC growth curves in a K-dec
reasing
test. Solid lines: measured curve. Dashed lines: estimated true curve excluding the effect of corrosion product
wedging. Asterisks indicate temporary crack arrests.
The threshold stress intensities determined by this method can be useful for ranking materials, but usually cannot be
considered valid. Therefore, they cannot be used in design calculations based on fracture mechanics. Displays of complete
V-K curves provide convenient comparisons of various materials, as shown in Fig. 48. Problems with the control of the
testing procedure and of correlations with service conditions have impeded the standardization of this test method (Ref
24, 34).

Fig. 48 SCC propagation rates for various aluminum alloy 7050 products. Double-beam specimens (S-
L; see
Fig. 28) bolt-loaded to pop-in and wetted three times daily with 3.5% NaCl. Plateau velocity a
veraged over 15
days. The right-hand end of the band for each product indicates the pop-in starting stress intensity (K
Io
) for the
tests of that material. Data for alloys 7075-T651 and 7079-T651 are from Ref 35. Source: Ref 82
Dead-weight loading, or a simulated dead-weight loading system used in conjunction with automatic data logging
equipment (Fig. 30(b), has proved to be a rigorous method for evaluating threshold stress intensities by SCC initiation
(Ref 33, 34). Because crack growth results in increasing stress intensity and an increasing crack opening, corrosion
product wedging is minimal, and each test usually has a definite end point (fracture). In these tests, fatigue precracked
compact or modified compact specimens (Fig. 25(b)) are loaded to various initial stress intensities K
Io
and exposed until
fracture or until completion of a designated time period (Fig. 29). The designated cut-off period should be long enough
for extended initiation times and yet not long enough to allow corrosion product wedging to exert a dominant influence.

The test results shown in Fig. 49 indicate that near-threshold values were reached within 1200 h, as judged by the
flattening tendency of the curves. The slight downward slope of some of the curves after 1200 h may be the result of
wedging by corrosion products, but this was not determined. The effect of such wedging would be to give lower estimates
of the threshold stress intensity.

Fig. 49 Initial stress intensity versus time to fracture for S-L (see Fig. 28
) compact specimens of various
aluminum alloys exposed to an aqueous solution containing 0.06 M sodium chloride, 0.02 M
sodium dichromate,
0.07 M sodium acetate, and acetic acid to pH 4. Asteri
sk indicates metallographic examination showed that SCC
had started. Source: Ref 33
The testing of longitudinal (L-T, L-S in Fig. 28) and long-transverse (T-L, T-S in Fig. 28) specimens presents special
problems with materials having typical directional grain structures. Stress-corrosion cracking growth is small and tends to
be in the L-T plane, which is perpendicular to the plane of the precrack (Ref 36, 83). Such out-of-plane crack growth
invalidates calculations of the plane-strain threshold stress intensity K
ISCC
. On the other hand, the testing of materials
having an equiaxed grain structure also presents problems with stress intensity calculations because of gross crack
branching; this would be applicable to specimens of any orientation.
The most widely used corrodent for testing precracked specimens is 3.5% sodium chloride solution applied dropwise to
the precrack two or (usually) three times daily (Ref 34, 35, 36, 37). This intermittent wetting technique accelerates SCC
growth but it also causes troublesome corrosion of the mechanical precrack. Less corrosive corrodents that have been
used include substitute ocean water (ASTM D 1141) and an inhibited salt solution containing 0.06 M sodium chloride,
0.02 M sodium dichromate, 0.07 M sodium acetate, and acetic acid to pH 4 (Ref 36, 37, 81). Some investigators have
tested 7000-series alloys in distilled water (Ref 78) and in water vapor at 40 °C (104 °F) (Ref 84). Typical test durations
that have been used range from 200 to 2500 h.
With low-resistance alloys, both of the first two corrodents listed in the preceding paragraph ranked alloys similarly and
in agreement with exposure to a seacoast and an inland industrial atmosphere. Plateau velocities in the laboratory tests
were about five to ten times faster than in the seacoast atmosphere and ten times faster than in the industrial atmosphere.

In these K-decreasing laboratory tests, corrosion product wedging effects dominated after exposure periods of about 200
to 800 h. The length of exposure time before the intervention of corrosion product wedging varies with several factors,
including the magnitude of K
Io
and the inherent resistance to crevice corrosion of the test material in the corrosive
environment (Ref 36, 41).
Slow Strain Rate Testing
Slow strain rate testing is not governed by any standards. Various aqueous solutions have been used in addition to 3.5%
sodium chloride. Because the 3.5% sodium chloride solution did not appear aggressive enough for slow strain rate testing,
more corrosive test mediums have been used, including oxidant additions to the sodium chloride solution or more acidic
solutions, such as aluminum chloride (Ref 52, 85).
In a round-robin testing program using several aluminum alloy types and several corrodents, a solution containing 3%
sodium chloride plus 0.3% hydrogen peroxide was considered the most promising candidate for possible standardization
(Fig. 33). Additional study is needed to determine the optimum composition of these constituents. Another promising
candidate was a solution of 2% sodium chloride plus 0.5% sodium chromate having a pH of 3.

Testing of Copper Alloys (Smooth Specimens)
Testing in Mattsson's Solution. According to ASTM G 37 (Ref 8), a stressed test specimen must be completely and
continuously immersed in an aqueous solution containing 0.05 g-atom/L of Cu
2+
and 1 g-mol/L of ammonium ion
( ) with a pH of 7.2. The copper is added as hydrated copper sulfate, and the is added as a mixture of
ammonium hydroxide and ammonium sulfate. The ratio of the latter two compounds is adjusted to achieve the desired
pH.
Mattsson's pH 7.2 solution is recommended only for brasses (copper-zinc base alloys). This test environment may give
erroneous results for other copper alloys and is not recommended. This is particularly true for alloys containing aluminum
or nickel.
This test environment is believed to provide an accelerated ranking of the relative or absolute degree of susceptibility to
SCC for different brasses. The test environment correlates well with the corresponding service ranking in environments
that cause SCC, which may be due to the combined presence of traces of moisture and ammonia vapor. The extent to

which the accelerated ranking correlates with the ranking obtained after long-term exposure to environments containing
corrodents other than ammonia is not known. Such environments may be severe marine atmospheres (chloride), severe
industrial atmospheres (predominantly sulfur dioxide), or superheated ammonia-free steam.
It is currently not possible to specify a time to failure in Mattsson's pH 7.2 solution that corresponds to a distinction
between acceptable and unacceptable SCC behavior in brass alloys. Such correlations must be determined on an
individual basis.
Mattsson's pH 7.2 solution may also cause some stress-independent general and intergranular corrosion of brasses.
Therefore, SCC failure may possibly be confused with mechanical failure induced by corrosion-reduced net cross section.
This is most likely with small cross-sectional specimens, high applied stress levels, long exposure times, and SCC-
resistant alloys. Careful metallographic examination is recommended for accurate determination of the cause of failure.
Alternatively, unstressed control specimens can be exposed to corrosive environments in order to determine the extent to
which stress-independent corrosion degrades mechanical properties.
Other Testing Media. The most widely used SCC agent for copper and copper alloys is ammonia (NH
3
) (Ref 86). The
ion does not appear to cause cracking in a stable salt, such as ammonium sulfate. Cracking will occur in a salt that
dissociates (such as ammonium carbonate) to form ammonia.
The ion (x is usually 4 to 5) is thought to be necessary to induce SCC in copper metals (Ref 87). Amine
groups also cause cracking, or are easily converted to ammonia. Amines and sulfamic acid also cause cracking. Dry
ammonia does not cause SCC of brass, as demonstrated by the successful use of brass valves and gages on tanks of
anhydrous ammonia.
Stress-corrosion cracking of copper metals in ammonia will not occur in the absence of oxygen or an oxidizing agent.
Carbon dioxide is also a requisite (Ref 88). Therefore, air rather than pure oxygen is necessary, and as a practical matter,
moisture is essential. When other factors are favorable, a very small amount of NH
3
is sufficient to cause cracking. The
controlling factor may therefore be moisture, because cracking may appear to be caused by the presence of a condensed
moisture film.
Other than ammonia, the most effective agents for causing cracking are the fumes from nitric acid or moist nitrogen
dioxide. Sulfur dioxide will also crack brass; but both maximum and minimum concentration limits exist, and the reaction

is slow (Ref 86). Alloy development studies have been conducted with a moist ammoniacal test atmosphere containing
80% air, 16% NH
3
, and 4% water vapor at 35 °C (95 °F). However, none of these corrodents has received the attention
that ammonia has garnered (Ref 87).
Historically, immersion of a copper alloy product in a mercurous nitrate solution has been used to test for residual stresses
(Ref 89, 90). Because these residual stresses are possible sources of failure by SCC in other environments, some have
regarded this test as a stress-corrosion test. However, it is only an indirect method of identifying SCC tendencies and does
not correlate to the presence of SCC as well as test methods based on specific attack by ammonia (Ref 86). It does
indicate, however, that mercury and other low-melting liquid metals can cause embrittlement and failure due to cracking.

Testing of Carbon and Low-Alloy Steels
Generally, steels with lower strengths are susceptible to SCC only upon exposure to a small number of specific
environments, such as the hot caustic solutions encountered in steam boilers, hot nitrate solutions, anhydrous ammonia,
and hot carbonate-bicarbonate solutions (Ref 91, 92).
Boiler Water Embrittlement Detector Testing. Caustic cracking failures frequently originate in welded structures
in the vicinity of faying surfaces, where small leaks cause soluble salts to accumulate in high local concentrations of
caustic soda and silica. As a general rule, crevices or splash areas on hot metal surfaces where the concentration of
dissolved soluble salts can occur are likely sites for SCC. This type of intergranular cracking failure has been produced
with concentrations of sodium hydroxide as low as 5%, but a concentration of 15 to 30% is usually required at 200 to 250
°C (390 to 480 °F) to produce this phenomenon. The apparatus and procedures used to determine the embrittling or
nonembrittling characteristics of the water in an operating boiler are detailed in ASTM D 807 (Ref 8).
Other Testing Media. Caustic cracking occurs in digester vessels used in the chemical-processing industries, and
laboratory studies have been conducted using sodium hydroxide concentrations of about 30 to 35% (Ref 93). Tests in
boiling nitrate solutions have frequently been used to study the effects of composition and metallurgical variables (Ref
92). In studies of low-carbon steel in boiling nitrate solutions having different cations, solutions containing the more
acidic cations in greater concentrations were found to be the most potent. This tendency is illustrated by the apparent
threshold stresses for failure of a 0.05% C steel in nitrate solutions with a range of concentrations, as shown in Table 5.
Table 5 Apparent threshold stress values for 0.05% C steel in nitrate solutions of varying concentrations

Apparent threshold stress values at a solution concentration of:

8 N 4 N 2.5 N 1 N
Nitrate solution
MPa

ksi

MPa

ksi

MPa

ksi MPa

ksi

Ammonium nitrate

14 2 21 3 48 7 83 12
Calcium nitrate
34 5 48 7 83 12 159 23
Lithium nitrate
34 5 55 8 131 19 at 2 N 159 23
Potassium nitrate
41 6 62 9 97 14 165 24
Sodium nitrate
55 8 131 19 152 22 179 26
Source: Ref 92

Cracking can be accelerated by the addition of small amounts of acid or oxidizing agents, such as potassium
permanganate, manganese sulfate, sodium nitrite, and potassium dichromate, but hydroxides and other salts, particularly
those forming insoluble iron products, such as sodium carbonate or sodium hydrogen phosphate, retard or prevent failure.
Sodium nitrite is also a known inhibitor if the nitrite concentration is equal to that of the nitrate ion. A standard test
environment has not been established, and conditions should be tailored to individual testing requirements.
The ranking of a given series of alloys may vary with exposure conditions (Ref 58). Consequently, selection of a
particular alloy for use in an environment that varies from that used in laboratory ranking tests may result in unexpected
service failure. This tendency is illustrated by the effects of alloying additions in ferritic steels on cracking in two
different environments (Fig. 50). Figure 50(a) illustrates that each of the alloying additions is beneficial in the carbonate-
bicarbonate solutions, with molybdenum having the greatest effect. However, the molybdenum addition has an adverse
effect in the 35% sodium hydroxide solution, although the beneficial effects of nickel and chromium additions remain the
same (Fig. 50b). Although nickel additions are beneficial in the above example, a similar addition of nickel to a carbon-
manganese steel produced susceptibility to SCC in boiling magnesium chloride; this did not occur in the steel without the
addition of nickel (Ref 95).

Fig. 50 Effect of various alloying elements on the SCC behavior of a low-
alloy ferritic steel in two different
corrosive environments. Behavior indicated by time to failure ratios i
n a slow strain rate test. (a) Immersed in 1
N sodium carbonate plus 1 N sodium bicarbonate at 75 °C (165 °F).
(b) Immersed in boiling 35% sodium
hydroxide. Source: Ref 94
The use of laboratory testing media that duplicate service conditions is equally important when accelerated tests are used
for quality control through the acceptance or rejection of production lots of a particular alloys. Reference 96 discusses
tests of prestressing steels intended for use as concrete reinforcing bars (rebars) in which an ammonium thiocyanate
solution was used to discriminate between heats of steel.
Use of the carbonate-bicarbonate solutions for testing pipeline steels by the slow strain rate method revealed that the
susceptibility to SCC was dependent on the electrochemical potential of the specimen surface in the test environment, as
shown in Fig. 50(a). A critical range in which SCC occurred was established. The critical range varies with the test
environment and alloy composition. Several tests at various carbonate-bicarbonate concentrations, temperatures, pH

levels, and corrosion potentials indicated that test conditions using an impressed potential of -650 mV versus the saturated
calomel electrode (SCE) and a temperature of 75 °C (165 °F) were optimal (Fig. 37).

Testing of High-Strength Steels (Ref 4, 97)
For steels with yield strengths greater than about 690 MPa (100 ksi) such as low-alloy and alloy steels, hot-work die
steels, maraging steels, and martensitic and precipitation-hardenable stainless steels the environments that cause SCC are
not specific. In many alloy systems, the phenomena of SCC and hydrogen embrittlement cracking are indistinguishable
(Fig. 1). This is particularly the case in environments that contain sulfides or other promoters of hydrogen entry.
Environments of major concern are natural waters for example, rainwater, seawater, and atmosphere moisture. Any of
these environments may become contaminated, which significantly increases the likelihood of SCC. Contamination with
hydrogen sulfide is particularly serious; consequently, the presence of hydrogen sulfide in high concentrations in salt
water associated with certain deep oil wells (termed sour wells; see the article "Corrosion in Petroleum Production
Operations" in this Volume) places an upper limit of approximately 620 MPa (90 ksi) on the yield strength that can be
tolerated in stressed steel in such environments without cracking.
Sulfide Stress Cracking. Determination of sulfide stress cracking is covered in NACE TM-01-77 (Ref 98). Stressed
specimens are immersed in acidified 5% sodium chloride solution saturated with hydrogen sulfide at ambient pressure
and temperature. The solution is acidified with the addition of 0.5% acetic acid, yielding an initial pH of approximately 3.
Applied stress at convenient increments of the yield strength is used to obtain cracking data that are plotted as shown in
Fig. 51. A 30-day test period is considered sufficient to reveal failure of susceptible material in most cases.

Fig. 51 Method of plotting results of sulfide
stress cracking tests. Open symbols indicate failure; closed symbols
indicate runouts. Source: Ref 98
The purpose of this test standard is to facilitate conformity in testing. Evaluation of data requires individual judgment on
several points based on the specific requirements of the end use. Consequently, the test should not be used as a single
criterion for evaluating an alloy for use in environments containing hydrogen sulfide or other hydrogen charging
elements. Attention should be paid to other factors that may affect SCC, such as pH, temperature, hydrogen sulfide
concentration, corrosion potential, and stress level, when determining the suitability of a metal for use.
The NACE test method recommends the use of smooth, small-diameter tension specimens stressed with constant-load or
sustained-load devices (Ref 98). However, different types of beam and fracture mechanics specimens may be included in

the testing standard in the future.
Another test method, known as the Shell Bent Beam Test, has been used for over 25 years in the petrochemical industry
to rank various materials for use in sour environments (Ref 99). However, acceptance has not been sufficient to generate
the interest for standardization.
Testing in sodium chloride solution constitutes a worst-case determination for high-strength steels; as such, it is
generally considered unrealistically aggressive for the useful ranking of steels in service environments that do not contain
hydrogen sulfide or other conditions favoring entry of hydrogen. Tests are usually performed in water containing about
3.5% sodium chloride, artificial seawater, natural seawater (rarely), or a marine atmosphere (Ref 4), unless specific
environmental conditions are under study. ASTM G 44 (Ref 8) is used where applicable.
In salt water and freshwater, a true threshold K
ISCC
exists for high-strength steels that is useful for characterizing
resistance to SCC. Ideally, K
ISCC
defines the combination of applied stress and defect size below which SCC will not
occur under static loading conditions in a given alloy and environment system. However, the reported value of K
ISCC
for a
given system often reflects the initial K
I
level and the exposure time associated with the testing. Table 6 illustrates the risk
of overestimating K
ISCC
by terminating the exposure test too soon when using the SCC initiation method (Ref 23, 24). A
similar risk exists in tests conducted with the arrest method. Table 7 shows that K
ISCC
values determined by the initiation
and arrest methods may be the same when testing times are sufficiently long and when compatible criteria are used for
establishing the threshold (Ref 24).
Table 6 Influence of cutoff time on apparent K

ISCC
using the SCC initiation method
Apparent K
ISCC
Exposure time, h

MPa ksi
100
187 170
1000
127 110
10,000
28 25
Note: The initiation method was used on a constant-load cantilever bend specimen (K-
increasing) of alloy steel with a yield strength
of 1240 MPa (180 ksi). Test environment was synthetic seawater at room temperature.
Source: Ref 24
Table 7 Comparison of K
ISCC
values determined by initiation and arrest methods
K
ISCC
, MPa (ksi )

Steel alloy
Initiation Arrest
10Ni, normal purity

24 (22) 26 (24)
10Ni, high purity

59 (54) 57 (52)
18Ni, normal purity

22-33 (20-30) 28 (25)
18Ni, high purity
<33 (<30) <33 (<30)
Note: Based on a crack growth rate of 2.5 × 10
-4
mm/h (10
-5
in./h). Modified compact specimens: constant load for initiation and
wedge-loaded with a bolt for arrest. Test environment: salt water at room temperature.
Source: Ref 24
Figure 52 illustrates a method used to compare various high-strength steels (Ref 4, 100). Data were obtained in salt water
or seawater, and K
ISCC
values are plotted versus yield strength. Envelopes are used to enclose all known valid data for the
various steels. The crosshatched envelopes or individual data points represent the featured steels, which allows
comparison with characteristics of the other steels. The straight lines in Fig. 52 illustrate how K
ISCC
values relate to the
maximum depth of long surface flaws that can be tolerated without stress-corrosion crack growth.

Fig. 52 Comparison of SCC behavior of several high-strength steels based on threshold stress intensity (K
ISCC
)
values in salt water. Source: Ref 100
Electrochemical Polarization. Although the mechanism of cracking in hydrogen sulfide environments is
predominantly one of hydrogen embrittlement, the mechanism of environmentally induced failures in environments not
containing sulfides or other promoters of hydrogen entry is not clearly agreed upon (Ref 97). Time to failure in a sodium

chloride solution depends on the corrosion potential (Ref 4, 101), which determines whether failure results from active
path corrosion or hydrogen embrittlement cracking. Electrochemical studies have shown that embrittlement of high-
strength steels by corrosion product hydrogen occurs when, for a given environment, the electrochemical potential of the
metal is equal to or more anodic than the reversible hydrogen potential, that is, for thermodynamic conditions that favor
the deposition of hydrogen on the surface of the steel.
Figure 53 compares the various types of cracking behavior that can be expected from electrochemical polarization (Ref
102). All of the curves except curve G were obtained experimentally. Curve A represents the case in which only hydrogen
embrittlement is obtained; curve B shows only active path corrosion. Both processes are shown in curves C and D.

Fig. 53
Use of electrochemical polarization to distinguish between SCC and hydrogen embrittlement
mechanisms in a high-strength steel immersed in sodium
chloride solution. See text for explanation of curves A
through H. Source: Ref 102
When both anodic and cathodic polarization shorten the cracking time, as in curve E, it is not possible to determine which
mechanism prevails without applied current. Curves F and G can be expected in acid solutions when the corrosion
potential is anodic to the reversible hydrogen potential.
In curve H, neither anodic nor cathodic polarization has any effect on cracking time. Therefore, it is possible that a
hydrogen embrittlement mechanism is involved. However, the mechanism by which hydrogen enters the steels is not
electrochemical. To perform realistic accelerated tests, the end use of the material and the environmental conditions
involved should be considered so that the test procedure involves the appropriate cracking mechanism. It should be noted
that hydrogen embrittlement cracking can also occur as a result of galvanic action between the test specimen and
components of the stressing system. In all SCC testing, therefore, all electrical contact between the specimen and
ancillary fixtures must be avoided, except when galvanic effects are desired.

Testing of Nonheat-Treatable Stainless Steels
The environments causing SCC that are encountered in the chemical industry are specific and are limited primarily to
chloride and caustic solutions at elevated temperatures and sulfide environments at ambient temperatures. In seawater at
or near room temperature, austenitic (iron-chromium-nickel) and ferritic (iron-chromium) steels do not experience SCC.
Fully ferritic stainless steels are highly resistant to SCC in chloride and caustic environments that cause austenitic

stainless steels to crack. However, laboratory studies have shown small additions of nickel or copper to ferritic steels may
render them susceptible to SCC in severe environments (Ref 4).
Testing in Boiling Magnesium Chloride Solution. ASTM G 36 (Ref 8) is applicable to wrought, cast, and welded
austenitic stainless steels and related nickel-base alloys. This method determines the effects of composition, heat
treatment, surface finish, microstructure, and stress on the susceptibility of these materials to chloride SCC. Although this
test can be performed with various concentrations of magnesium chloride, ASTM G 36 specifies a test solution
maintained at a constant boiling temperature of 155.0 ± 1.0 °C (311.0 ± 1.8 °F), that is, approximately 45% magnesium
chloride. Also described is a test apparatus capable of maintaining solution concentration and temperature within the
recommended limits for extended periods of time. Typical exposure times are up to 1000 h. However, historically, most
of the SCC data on austenitic stainless steels were obtained by using a boiling 42% MgCl
2
solution (boiling point: 154 °C,
or 309 °F). For this reason, much current testing is still done at the lower concentration.
Most chloride cracking testing has been carried out in accelerated test media such as boiling magnesium chloride (Ref 4,
103, 104). All austenitic stainless steels are susceptible to chloride cracking as shown in Fig. 54. It is noteworthy,
however, that the higher-nickel types 310 and 314 were appreciably more resistant than the others (Fig. 55). Although this
solution causes rapid cracking, it does not necessarily simulate the cracking observed in field applications.

Fig. 54 Relative SCC behavior of austenitic stainless steels in boiling magnesium chloride. Source: Ref 105


Fig. 55
Effect of nickel additions to a 17 to 24% Cr steel on resistance to SCC in boiling 42% magnesium
chloride. 1.5-mm (0.06-in.) diam wire specimens dead-weight loaded to 228 or 310 MPa (33 or 45 ksi).
Source:
Ref 106
Other ions in addition to chloride can cause cracking. Of all halogen ions, chlorides cause the most cases of SCC in
austenitic stainless steels. Known cases of fluoride and bromide SCC are few, and iodide is not known to produce SCC. In
addition, cations, such as Li
+

, Ca
2+
, Zn
2+
, , Ni
2+
, and Na
+
, affect test results to varying degrees (Ref 107). Although
chloride SCC occurs primarily at temperatures above about 90 °C (190 °F), acidified chloride solutions can produce SCC
at low temperatures (Ref 107, 108, 109). Therefore, in diagnosing service failures, it is necessary to establish which ions
(and other environmental and stress conditions as well) have caused the failure. In this manner, an appropriate test
procedure can be designed for the evaluation of alternative materials.
Reference 110 discusses laboratory reproduction of an environment that caused SCC at the top of a distillation tower in a
crude oil refinery. The service environment consisted of a very dilute hydrochloric acid solution (36 ppm chloride) with a
pH of 3 saturated with hydrogen sulfide gas at 80 °C (175 °F). In this test environment, austenitic stainless steels, such as
type 304 or 316 failed, but the ferritic types 430 and type 434 did not.
Testing in Polythionic Acids. Petrochemical refinery equipment is subject to polythionic acid cracking, which may
occur after shutdown. Polythionic acid forms by the decomposition of sulfides on metal walls in the presence of oxygen
and water. ASTM G 35 (Ref 8) describes procedures for preparing and conducting exposures to polythionic acids
(H
2
S
n
O
6
, where n is usually 2 to 5) at room temperature to determine the relative susceptibility of sensitized stainless
steels or related materials (high nickel-chromium-iron alloys) to intergranular SCC.
This test method can be used to evaluate stainless steels or other materials in the as-received condition or after high-
temperature service (480 to 815 °C, or 900 to 1500 °F) for prolonged periods of time. Wrought products, castings, and

weldments of stainless steels or other related materials used in environments containing sulfur or sulfides can also be
evaluated. Other materials that are capable of being sensitized can also be tested.
A variety of smooth SCC test specimens, surface finishes, and methods of applying stress can be used. Stressed
specimens are immersed in the polythionic acid solution, which can be prepared by passing a slow current of hydrogen
sulfide gas for 1 h through a fritted glass tube into a flask containing chilled (0 °C, or 32 °F) 6% sulfurous acid, after
which the liquid is kept in a stoppered flask for 48 h at room temperature. Solutions can also be prepared by passing a
slow current of sulfur dioxide gas through a fritted glass bubbler submerged in a container of distilled water at room
temperature. This is continued until the solution becomes saturated. The hydrogen sulfide gas is then slowly bubbled into
the sulfurous acid solution.
Prior to use, the polythionic acid solution should be filtered to remove elemental sulfur and then tested for acid content.
This can be done by analytical tests or by using a control test specimen of sensitized type 302 stainless steel. The control
should fail by cracking in less than 1 h.
The wick test can be used to evaluate the chloride cracking characteristics of thermal insulation for applications in the
chemical process industry. ASTM C 692 (Ref 8) covers the methodology and apparatus used to conduct this procedure.
When a dilute aqueous solution is transmitted to a metal surface by capillary action through an absorbent fibrous material,
the process is called wicking. Cracking occurs at much lower temperatures when alternate wetting and drying is used than
when the specimens are kept wet continuously.
Other Testing Media. Hot concentrated caustic solutions are another type of environment encountered in chemical
industries that causes SCC of stainless steels. However, the conditions leading to caustic cracking are more restrictive
than those leading to chloride cracking, and caustic environments have not received the attention that chlorides have.
There is little difference in the susceptibilities among types 304, 304L, 316, 316L, 347, and USS 18-18-2 austenitic steels.
All of these alloys crack rapidly in solutions of 10 to 50% sodium hydroxide at 150 to 370 °C (300 to 700 °F) (Ref 4, 104,
111).
Certain strong acid solutions containing chlorides, such as 5 N sulfuric acid plus 0.5 N sodium chloride, 3 N perchloric
acid plus 0.5 N sodium chloride, and 0.5 N to 1.0 N hydrochloric acid, are capable of causing SCC in austenitic stainless
steels at room temperature (Ref 4). Cracking in these environments is similar to the type of cracking that occurs in hot
chloride environments.
Electrochemical Polarization. Stress-corrosion cracking in austenitic and ferritic stainless steels can be delayed or
prevented by the application of cathodic current: however, if ferritic steels are overprotected by relatively large cathodic
current, they are apt to blister or crack due to the hydrogen discharged by the cathodic protection action. Anodic

polarization significantly accelerates the initiation of SCC, but appears to have a smaller accelerating effect on crack
propagation (Ref 112).
Testing of Magnesium Alloys
There is no standard accelerated test environment recommended for assessing the susceptibility of magnesium-base alloys
to SCC. Exposure of stressed specimens to the atmosphere has generally been used to determine the SCC susceptibility of
specific products.
The chloride-containing solutions typically used in accelerated tests for aluminum alloys are unsatisfactory for SCC tests
of magnesium alloys because of excessive general corrosion. In one investigation, a chromate-inhibited chloride solution
(35 g/L sodium chloride plus 20 g/L potassium chromate; pH 8) was found to be suitable for testing magnesium alloys
(Ref 113). Good correlation was observed between the SCC behavior of magnesium-aluminum-zinc alloys exposed by
total immersion in this solution and the behavior of the same alloys exposed to an industrial atmosphere. Cracking of
highly stressed susceptible alloys occurs within a few hours, but exposures can be continued up to 1000 h without
incurring excessive pitting. Laboratory tests also have been conducted using potassium hydrogen fluoride and a dilute
solution of sodium chloride plus sodium bicarbonate as the test medium (Ref 114).
Testing of Nickel Alloys
Nickel-base alloys are highly resistant to the chloride SCC that affects stainless steels. Iron-chromium-nickel alloys with
nickel contents greater than 50% are immune to cracking in boiling 42% magnesium chloride (Fig. 55). However, SCC of
nickel and high-nickel alloys has been experienced in high-temperature caustic soda and caustic potash solutions and in
molten caustic.
Cracking of some nickel-base alloys has also occurred under special conditions in fluosilicic acid, hydrofluoric acid,
mercuric salt solutions, and high-temperature water and steam that are contaminated with trace amounts of oxygen, lead,
fluorides, or chlorides (Ref 103, 106, 115). Sensitized alloys are susceptible to SCC in sulfur compounds such as sodium
sulfite, sodium thiosulfate, and polythionic acids.
The standard test environments that are most frequently used for high-nickel alloys are the same as those employed for
stainless steels. In a study of sulfur-induced SCC of sensitized Inconel alloy 600 steam generator tubing in water
contaminated by air and sodium thiosulfate at temperatures from 22 to 95 °C (72 to 203 °F), a solution of 0.1 M sodium
tetrathionate with a pH of 3.5 to 4.0 at 22 °C (72 °F) appeared to be an excellent test medium for sensitization in nickel
alloys and stainless steels. Slow strain rate testing was also found to be more effective than tests with statically loaded U-
bend specimens (Ref 116).
Slow strain rate testing was also effective for evaluating several nickel- and cobalt-base alloys in hot chloride and hot

caustic solutions. The average length of secondary stress-corrosion cracks, as determined by metallographic examination,
appeared to be a more appropriate parameter for quantifying the severity of SCC behavior than loss in ductility or loss in
fracture strength parameters; this is illustrated in Table 8 for Hastelloy alloy C-276. However, when using slow strain rate
testing methods, care must be taken not to confuse stress-assisted localized corrosion with SCC (Ref 117).
Table 8 Results of slow strain rate tests on Hastelloy alloy C-276
Ultimate
tensile
strength
Alloy condition

Strain
rate,
s
-1

Environment Reduction

in area,
%
MPa

ksi
Time
to
failure,

h
Average length of secondary
stress-corrosion cracks
3.4 × 10

-6


Air 71 745 108

60 0
Mill annealed
3.4 × 10
-6


50% sodium hydroxide, 147 °C
(297 °F)
61 593 86 52 13 × 10
-5
m (5 mils)
3.4 × 10
-6


Air 49 1524

221

17 0
3.4 × 10
-6


50% sodium hydroxide, 147 °C

(297 °F)
51 1503

218

18 <1 × 10
-5
m (<0.1 mil); no
obvious SCC
9 × 10
-7
Air 53 1558

226

29 0
9 × 10
-7
50% sodium hydroxide, 147 °C
(297 °F)
47 1524

221

30 2.5 × 10 m (1 mil)
5.3 × 10
-7


Air 51 1593

231

51 0
50% cold
swaged
5.3 × 10
-7


50% sodium hydroxide, 147 °C
(297 °F)
47 1565

227

60 4.8 × 10
-5
m (1.9 mils)
Source: Ref 116

Testing of Titanium Alloys
Although titanium alloys are not susceptible to SCC in either boiling 42% magnesium chloride or boiling 10% sodium
hydroxide solutions, which are commonly used to study SCC in stainless steels, the susceptibility of titanium and its
alloys to SCC has been demonstrated in several environments. This information is given in Table 9.
Table 9 Environments and temperatures conducive to SCC of titanium alloys

Environment Temperature, °C (°F)
Hot dry chloride salts
260-480 °C (500-900 °F)

Seawater, distilled water, and aqueous solutions

Ambient
Nitric acid, red fuming
Ambient
Nitrogen tetroxide
Ambient to 75 °C (165 °F)

Methanol, ethanol
Ambient
Chlorine
Elevated
Hydrogen chloride
Elevated
Hydrochloric, acid, 10%
Ambient to 40 °C (105 °F)

Trichloroethylene
Elevated
Trichlorofluoroethane
Elevated
Chlorinated diphenyl
Elevated

Testing in a Hot Salt Environment. The hot salt test consists of exposing a stressed salt-coated test specimen to an
elevated temperature for various predetermined lengths of time. The exposure periods are determined by the alloy, stress
level, temperature, and selected damage criterion (that is, embrittlement, cracking, or rupture, or a combination of these
phenomena). Exposures are typically carried out in laboratory ovens or furnaces equipped with loading equipment for
stressing specimens. Environmental conditions, the degree of control required, and the means for obtaining control are
described in ASTM G 41 (Ref 8).

This test method can be used to test all metals if service conditions warrant. The test limits maximum operating
temperatures and stress levels, or it categorizes different alloys according to their susceptibility if hot salt damage has
been found to accelerate failure by creep, fatigue, or rupture. Although limited evidence relates this phenomenon to actual
service failures, cracking under stress in a hot salt environment is a potential design-controlling factor.
The hot salt test should not be construed as being related to the SCC of materials in other environments. It should be used
only in an environment that may be encountered in service.
Hot salt testing can be used for alloy screening to determine the relative susceptibility of metals to embrittlement and
cracking and to determine the time-temperature-stress threshold levels for the onset of embrittlement and cracking.
However, certain types of specimens are more suitable for each of these types of characterizations. Precracked specimens
are unsuitable for testing of titanium alloys, because cracking reinitiates at salt/metal/air interfaces and results in many
small cracks that extend independently. Therefore, smooth specimens are recommended.
Testing in Water and Aqueous Solutions. Water, seawater, and almost any neutral aqueous solution (except
atmospheric water vapor) can cause SCC in many titanium alloys in the presence of preexisting cracklike flaws, although
susceptibility in these environments cannot be detected by smooth specimens. Therefore, fracture mechanics type
characterizations are necessary. For titanium alloys, the extremely rapid growth of stress-corrosion cracks in salt water
and the dependency on specimen geometry preclude the possibility of using crack growth rate data for design purposes.
Therefore, ranking of materials must be based on K
ISCC
values, and a true threshold stress intensity for SCC apparently
does exist (Ref 118). Titanium alloys do not exhibit stage I type crack growth kinetics (Fig. 3) in neutral aqueous
solutions. Tests have been performed for sufficient periods of time to allow detection of crack growth rates of 10
-9
m/s
(1.4 × 10
-4
in./h), but SCC has not been observed. The slowest crack velocity that has been detected is 10
-8
m/s (1.4 × 10
-3

in./h). Therefore, in neutral aqueous solutions, a threshold K
ISCC
exists at which SCC will not propagate (Ref 4, 118). The
above rates, however, are not as slow as those observed in high-susceptibility aluminum alloys (Fig. 46). Tests are
commonly performed in water containing about 3.5% sodium chloride, artificial seawater, or natural seawater unless
specific environments are being tested.
Electrochemical Polarization. The halide ions (chloride, bromide, and iodide) are SCC agents unique for titanium
alloys in aqueous solutions at room temperature. The crack initiation load and velocity are controlled by the applied
potential, as illustrated for the crack initiation load in Fig. 56. At potentials more negative than about -700 to -1400 mV,
depending on the solution, specimens were cathodically protected. Sodium fluoride solution and solutions of the other
anions that do not produce SCC (hydroxide, sulfide, sulfate, nitrite, nitrate, perchlorate, cyanide, and thiocyanate) yielded
results at all potentials in the same scatterband as the air values.

Fig. 56 Variation of crack initiation load with potential in 0.6 M halide solutions for Ti-8Al-1Mo-
1V. Specimen:
single-edge cracked sheet that was tension loaded by constant displacement. Source: Ref 118, 119
At potentials more positive than the above values, susceptibility in varying degrees occurred in the chloride, bromide, and
iodide solutions. The width of the critical potential range and the potential for maximum susceptibility varies with the
anion. A region of anodic protection occurred in the chloride and bromide solutions, but not in the iodide solution.
Crack propagation can be halted by switching the potential to either the anodic or cathodic protection zone. The corrosion
potential of titanium alloys in 3.5% sodium chloride and seawater about -800 mV versus SCE is similar (slightly more
negative) to the potential at which SCC susceptibility reaches a maximum (Ref 118).
Testing in Organic Fluids. A wide variety of organic fluids can cause SCC in some titanium alloys under specific test
conditions (Table 9). Most of these fluids attack the passive surface film that is characteristic of titanium alloy products.
Consequently, precracked specimens do not have to be used to accelerate the SCC initiation. A standard environment
does not exist; test conditions must be selected with appropriate consideration given to the type of environmental service
required.
Sustained-Load Cracking in Inert Environments. High-strength titanium alloys for use in highly stressed
components for military aircraft and other similar applications may be susceptible to sustained-load cracking in inert
environments (including dry air). Sustained-load cracking is similar to SCC except that it is much slower and occurs in

the total absence of a reactive environment. Sustained-load cracking is caused by, or is greatly aggravated by, hydrogen
dissolved in the titanium during processing. Vacuum annealing can reduce the hydrogen level to less than 10 ppm, at
which concentration the tendency toward sustained-load cracking is greatly reduced (Ref 4, 120).
Figure 57 illustrates an example of sustained-load cracking in mill-annealed plate of Ti-8Al-1Mo-1V containing 48 ppm
hydrogen. As shown in Fig. 57, the threshold stress intensity factor for sustained-load cracking in dry air is designated K
IH

because it is attributed to hydrogen in the metal. When the hydrogen concentration was reduced to 2 ppm by vacuum
annealing, the K
IH
value was increased to equal the inherent plane-strain fracture toughness, K
Ic
. However, the K
ISCC
value
was not affected (Ref 121). Therefore, in addition to the practical importance of sustained-load cracking, its potential
contribution to cracking should be taken into account when evaluating environmental effects, particularly in mechanistic
studies.

Fig. 57 Effect of sustained-load cracking compared to SCC in Ti-8Al-1Mo-1V mill-
annealed sheet. Hydrogen
concentration, 48 ppm; yield strength, 850 MPa (123 ksi); cantilever bend specimen (T-S); B
= 6.35 mm (0.25
in.). See Fig. 28 for an explanation of specimen orientation and fracture plane identification. Source: Ref 121

Special Considerations for Testing of Weldments
ASTM G 58 (Ref 8) covers test specimens in which stresses are developed by the welding process only (that is, residual
stress, Fig. 23), an externally applied load in addition to the stresses due to welding (Fig. 7e), and an externally applied
load only, with residual welding stresses removed by annealing.
The National Materials Advisory Board Committee on Environmentally Assisted Cracking Test Methods for High-

Strength Weldments recently published the following guidelines on SCC testing of weldments (Ref 24). Fracture
mechanics of cracked bodies was found to be a valid and useful approach for designing against environmentally assisted
cracking, although several limitations and difficulties must be taken into consideration. For static loading, K
ISCC
and da/dt
versus K
I
are useful parameters. They are specified to a material, temperature, and metal/environment system and are
functions of local chemical composition, microstructure, and so on.
Superimposed minor load fluctuations and infrequent changes in load can alter environmental cracking response. This
effect, which cannot be predicted from K
ISCC
and da/dt values, may be significant and detrimental. Reexamination of
static loading as a design premise may be required. Existing test methodology or environmentally assisted cracking
tendency is applicable to the evaluation of weldments. As in other structural components, residual stress must be treated
in a quantitative and realistic manner.
The National Materials Advisory Board report supports current design emphasis based on the presumption of preexisting
cracklike flaws in the structure and covers testing with precracked (fracture mechanics) specimens only. It contains a
critical assessment of the problems associated with environmentally assisted cracking in high-strength alloys and of state-
of-the-art design and test methodology.
Surface Preparation of Smooth Specimens
The pronounced effect of surface conditions on the time required to initiate SCC in test specimens is well known (Ref 5).
Unless the as-fabricated surface is being studied, the final surface preparation generally preferred is a mechanical process
followed by degreasing. However, chemical etches or electrochemical polishes can be used to remove heat-treating films
or thin layers of surface metal that may have become distorted during machining.
Care should be exercised to select an etchant that will not selectively attack constituents or phases in the metal and that
will not deposit undesirable residues on the surface. Etching or pickling should not be used with alloys that are
susceptible to hydrogen embrittlement.
Precautions should be taken when machining specimens to avoid overheating, plastic deformation, or the development of
residual stress in the metal surface. Machining should be performed in stages so that the final cut leaves the principal

surface with a clean finish of 0.7 m (30 in.) rms or smoother. The required machining sequences, types of tools, and
feed rate depend on the alloy and metallurgical condition of the testpiece. Lapping, mechanical polishing, and similar
operations that produce flow of the metal should be avoided.
Interpretation of Test Results
This is the most fallible part of SCC testing and evaluation; it includes the analysis that leads to the conclusions and
recommendations. Stress-corrosion test data are at best imprecise and test dependent, and they must be qualified with the
testing conditions. It is important to verify the mode of environmental cracking (Fig. 1) and then to review the data to
exclude all extraneous results, as discussed previously with the individual test methods. Following are some comments on
the nature of the test dependency of the most commonly used criteria of SCC behavior.
Criteria of SCC Behavior
Specimen Life (Time to Failure). Stress-corrosion testing frequently involves determining the lives of specimens
under specific test conditions. This includes the initiation (or incubation) of a stress-corrosion crack and its propagation to
the point of fracture (Fig. 2). Such a determination is easily accomplished when only a single crack forms and the
specimen fractures within the chosen test period. However, it often happens that SCC occurs but the specimen does not
fracture. This is especially likely when testing relatively low-strength materials by constant strain loading (Fig. 5b) and
when testing at applied stress or stress intensity levels only slightly above the threshold (Fig. 29a). Cracks may initiate at
multiple sites in constant-strain loaded smooth specimens with relatively low applied stress, and a difficult problem arises
in deciding when to consider a specimen failed if it does not fracture visibly.
It is often found that the majority of specimens in a set of replicates in a test fail rapidly; this leaves a few specimens that
fail at much longer times or do not fail at all before the test is discontinued. Such behavior presents difficulties, both
theoretical and practical, in deciding when to terminate a test, choosing a satisfactory representative value, and comparing
such values.
The arithmetic mean specimen life is widely used for smooth specimens because it can be manipulated algebraically and
can be used in many standard statistical tests of significance. It should be remembered, however, that extremely large or
extremely small values may cause the mean to be atypical of the true distribution. Moreover, in using the arithmetic
mean, it is assumed that the population is normally or very nearly normally distributed. The median, on the other hand,
has the advantages that it is influenced less by extreme values, requires no assumption about the population distribution,
and can be obtained much faster than arithmetic mean values because only about half the number of replicates exposed
need to be tested to failure. The median is used in a German specification (Ref 19) that also provides for the use of the
geometric mean if the replication is small.

References 1 and 122 contain examples for highly susceptible steel and aluminum alloys, respectively; these examples
demonstrate the normal distributions for the logarithms of the specimen lives. With such distributions, a geometric mean
would be the best representative value of the specimen life. It has also been shown that a Weibull distribution can be
appropriate for the non-normally distributed test data for a relatively resistant aluminum alloy (Ref 123). Thus, it should
not be assumed that any one distribution is applicable for all testing situations.
Comparisons of alloys with differing strength and fracture toughness based on time to failure can be completely
misleading. For example, SCC growth curves are illustrated schematically in Fig. 58 for alloys with different fracture
toughnesses. Curves A, B, and C represent materials with decreasing toughness, with curve C showing fast fracture
initiated by corrosion pits or fissures with no SCC.

Fig. 58
Various processes in SCC as influenced by the fracture toughness of the metal. Kinetics for pitting (or,
in material D, nonpitting), SCC (ma
terials A and B only), and fast fracture. Line at top illustrates how time to
failure data can be misleading. Source: Ref 124
The behavior of a material that does not develop localized pitting or intergranular attack is represented by a line
coincident with the abscissa, designated D in Fig. 58. The time-to-failure ranking above the graph indicates D as best and
A, B, and C as poorest. Actually, the SCC responses of C and D were not measured, and the true SCC ranking of A and B
(indicated by depth of SCC at the time of fracture) exhibits a trend opposite to that inferred from the time to failure data
above.
Further difficulties may arise, because the total time to fracture is also influenced by non-SCC factors. such as specimen
type and size, method of loading, initial stress level, and initiation behavior of the alloy. Consequently, the SCC ranking
of materials may vary among investigators using different testing techniques. Nevertheless, comparisons of specimen
lives derived from smooth specimens can be useful in certain mechanistic studies and in tests for comparing
environmental variations if the mechanical aspects of the investigation are held constant.
Threshold Stress (Stress-Time Curve). More information about the resistance to SCC of a material can be
obtained when testing smooth specimens by using a range of applied stresses. Such data are usually presented graphically
with the applied gross section stress plotted against specimen life (Fig. 54). The primary interest is generally in the long-
life portion of the curve to obtain an estimate of the threshold stress.
A common method of estimating threshold stress involves the experimental determination of the lowest stress at which

cracking occurs in at least one specimen and the highest stress at which cracking does not occur in several specimens (for
example, three or more, depending on variability). An average of the lowest failure and highest no-failure stresses is
usually taken as the threshold stress. Such determinations of critical stresses are carried out at specified test times that are
known through experience or preliminary tests to be sufficient to produce SCC in the alloy-environment system of
interest. Statistical methods are available for determining threshold stresses more precisely (for example, the Probit
method or staircase method) and are commonly used for the determination of fatigue limits. However, the additional
testing involved can be quite extensive.
Apparent threshold stresses determined in laboratory tests of coupon specimens are useful for ranking the SCC
susceptibility of various materials, but, because such data are dependent on test conditions, they are not realistic for the
purpose of engineering design (Ref 4). Also, when using such data as an aid in selecting the material for a specific
structure, caution should be exercised in trying to relate the laboratory test conditions to the anticipated service
conditions. Not only the environmental condition but also the geometry and size of the test specimens and the method of
stressing should be compared (Fig. 6, 17). Further, threshold stresses obtained with statically loaded smooth specimens
are likely to be nonconservative; it has been shown in tests on carbon-manganese steel that the threshold stress obtained
by static loads is reduced by applied constant slow strain rates and that it can be reduced even further with cyclic loading
(Ref 49). With appropriate frequency, load change, and temperature, average creep rates can be sustained over extended
periods, but with static loading, the creep rate may fall below the level needed to promote SCC. In general,
experimentally determined threshold stresses for materials with only limited susceptibility to SCC are more sensitive to
variations in testing conditions than in the case of highly susceptible materials.
Percent Survival (Curve). This method of analyzing stress corrosion test results is especially useful when some of
the specimens in a group survive the duration of the test. Examples of various comparisons by this technique are shown in
Fig. 17, 42, and 43. Although the curves can be drawn on regular coordinate graph paper, the percent survival values will
often lie along a straight line when plotted on normal probability paper, as illustrated in Fig. 59. The linearity of this plot
indicates that the statistical distribution of the test results is logarithmic-normal. The vertical positions of the lines indicate
the cracking ability of the environment, which can be represented by median cracking times. The slopes of the lines
correspond to the variance, which can be used to calculate confidence limits.

Fig. 59 Distribution of SCC test results for a stainless steel. Source: Ref 1, 125

Threshold Stress Intensity (K

ISCC
, K
th
). Linear elastic fracture mechanics is well established as a basis for materials
characterization, including environmental cracking (Ref 24, 126, 127). In practice, it is most practical to define K
ISCC
as
the K
I
level associated with some generally acceptable and definably low rate of crack growth that is commensurate with
the design service life. When K
ISCC
values are reported, the criterion for their assessment and the exposure time in the
environment must accompany the threshold values. A rational approach to the development of useful data for design is to
establish an operational definition of K
ISCC
that is appropriate for the structure under consideration.
Such characterization requires that linear elastic fracture mechanics and plane-strain conditions be satisfied. However, for
certain low-strength steels and aluminum alloys, existing data show that SCC can occur under conditions that deviate
substantially from plane-strain, and that SCC is by no means limited to or is most severe under plane-strain loading
conditions (Ref 24, 36, 128, 129). In these cases, the application of linear elastic fracture mechanics is no longer valid,
and the parameter K
ISCC
is no longer meaningful. Similarly, when testing materials with a high resistance to SCC, loading
to high percentages of K
Ic
may cause a relaxation of stress due to creep. In this case also, the apparent K
ISCC
values are
meaningless. Constant-load tests, therefore, are preferred for lower strength materials (Ref 130).

The symbol K
th
has been used to identify threshold stress intensity factors developed under test conditions that do not
satisfy all the requirements for plane-strain stress. Design calculations using such values should not be employed unless it
is clear that the laboratory tests exhibit the same stress state as that for the intended application. Nevertheless, properly
determined K
th
values can be useful for ranking materials.
In principle, experimentally determined K
ISCC
values should be the same whether they are determined by the initiation or
the arrest test method (Fig. 29). In both tests, there are dimensional requirements for ensuring that the test results are
independent of geometrical effects (see the section "Preparation of Precracked Specimens" in this article). However,
precautions must be exercised during testing to avoid the potential problems involved with the environmental exposure
(incubation, corrosion product) wedging, crack branching, crack tip blunting, and so on). Comparisons of K
ISCC
values
determined for selected steels by both methods, along with examples of overestimated values resulting from insufficient
length of exposure, are shown in Tables 6 and 7. It is advisable, when practicable, to use a test that matches the type of
loading encountered in the anticipated service.
From the parameter K
ISCC
, a value of a
cr
can be calculated using the relation a
cr
= 0.2 (K
ISCC
/TYS)
2

(see the section "Static
Loading of Precracked (Fracture Mechanics) Specimens" in this article). This is the shallowest crack (surface length is
long compared to its depth) that will propagate as a stress-corrosion crack at a yield strength level of gross stress under
the given environmental conditions. This can be a very useful parameter for comparing materials, especially when the
measured a
cr
values can be related to the capability of the flaw inspection system used for a given engineering structure
(Fig. 52). Straight lines representing assumed values of a
cr
in Fig. 52 illustrate how K
ISCC
values for the various steel
alloys relate to the maximum depth of long surface flaws that can be tolerated without growth of SCC.
Such a plot can be used as follows. If the inspection system to be used can detect all long surface flaws deeper than 0.25
mm (0.01 in.), then the materials engineer would select an alloy with a K
ISCC
above the 0.25-mm (0.01-in.) line.
Conversely, if the K
ISCC
and tensile yield strength of a material are known, the equation can be used to estimate the
maximum tolerable flaw size. Substitution of an anticipated design stress in terms of percentage of tensile yield strength
in the formula for a
cr
will generate a new series of a
cr
lines of lower slope.
Alternatively, the following method, which is specific for a given loading method, can be used, inasmuch as the flaw
depth and applied stress are uniquely related for a specific loading situation, a family of curves for constant K
th
values can

be developed within these parameters. Figure 60 shows such curves for long, shallow flaws; the curves were generated
according to the following equations (Ref 132, 133, 134);


(Eq 16)


(Eq 17)


(Eq 18)
where K
I
is stress intensity, is applied stress, a is flaw depth, Q is a shape parameter, and t is thickness. When K
th
is
substituted into Eq 16 and a flaw shape is assumed (Q = 0.8), this parameter is related to the flaw size and the applied
stress. In this representation, SCC growth will not occur below the curve for the appropriate K
th
. For example, for a steel
with K
th
of 60.5 MPa (55 ksi ), a very deep flaw (6 mm. or 0.25 in.) would be required to cause crack
propagation in hydrogen gas of a steel component stressed to 360 MPa (52 ksi) in bending. Such representations are
useful for relating test data and K
th
in design and in the development of crack inspection requirements, as well as for
ranking alloys.

Fig. 60

Relationship of applied stress and flaw depth to crack propagation in hydrogen gas. Dashed lines show
an example of the use of such a chart for a steel with K
th
of 60.5 MPa (55 ksi
) at an operating stress
of 359 MPa (52 ksi). Source: Ref 131
The requirements for K
ISCC
tests, as well as for other fracture mechanics tests, include very explicit criteria regarding the
minimum crack length for ensuring that the test results can be analyzed properly using existing linear fracture mechanics
concepts. Therefore, a typical K
ISCC
test uses a relatively large starting crack of the order of 25 mm (1 in.) long in a 25-
mm (1-in.) thick specimen. In many types of service, however, initial defects of this size are rare. For example, damage-
tolerant design criteria for military aircraft specify a flaw size of the order of 1.27 mm (0.05 in.) as the initial worst-case
damage assumption upon introduction of a new part into service (Ref 126). Experimental work on high-strength steels
exposed to hydrogen sulfide gas indicates that for a given combination of materials and applied stress there may indeed be
a defect size below which the direct applicability of linear elastic fracture mechanics is questionable (Ref 135). Because
of their susceptibility to SCC, however, high-strength steels should not be contemplated for service in the presence of
hydrogen sulfide.
For example, Fig. 61 represents a concept of combining SCC thresholds based on smooth specimen and linear elastic
fracture mechanics tests of aluminum alloy plate to give a conservative estimate of materials for design. As shown in Fig.
61, the threshold stress intensity analysis breaks down in the small flaw region (ABE) when the smooth specimen
threshold stress is exceeded. Therefore, the definition of a safe zone requires results from both types of tests; the exclusive
use of either one of the test methods can yield nonconservative conclusions. It is anticipated that application and further
development of elastic-plastic fracture mechanics theory will lead to improved estimates of critical stress/flaw size
combinations for the onset of SCC and tensile fracture, as proposed in Fig. 62.

Fig. 61
Concept for combining SCC thresholds obtained on smooth and linear elastic fracture mechanics

specimens to yield a conservative assessment of materials. (1) Minimum
stress at which small tensile
specimens fail by SCC when stressed in environment of interest. (2) Minimum stress intensity at which
significant stress-corrosion crack growth occurs in environment of interest. Source: Ref 136