Stress corrosion cracking of candidate waste container materials; Final report

Six alloys have been selected as candidate container materials for the storage of high-level nuclear waste at the proposed Yucca mountain site in Nevada. These materials are Type 304L stainless steel (SS). Type 316L SS, Incoloy 825, phosphorus-deoxidized Cu, Cu-30%Ni, and Cu-7%Al. The present program has been initiated to determine whether any of these materials can survive for 300 years in the site environment without developing through-wall stress corrosion cracks. and to assess the relative resistance of these materials to stress corrosion cracking (SCC)- A series of slow-strain-rate tests (SSRTs) and fracture-mechanics crack-growth-rate (CGR) tests was performed at 93{degree}C and 1 atm of pressure in simulated J-13 well water. This water is representative, prior to the widespread availability of unsaturated-zone water, of the groundwater present at the Yucca Mountain site. Slow-strain-rate tests were conducted on 6.35-mm-diameter cylindrical specimens at strain rates of 10-{sup {minus}7} and 10{sup {minus}8} s{sup {minus}1} under crevice and noncrevice conditions. All tests were interrupted after nominal elongation strain of 1--4%. Scanning electron microscopy revealed some crack initiation in virtually all the materials, as well as weldments made from these materials. A stress- or strain-ratio cracking index ranks these materials, in order of increasing resistance to SCC,more » as follows: Type 304 SS < Type 316L SS < Incoloy 825 < Cu-30%Ni < Cu and Cu-7%Al. Fracture-mechanics CGR tests were conducted on 25.4-mm-thick compact tension specimens of Types 304L and 316L stainless steel (SS) and Incoloy 825. Crack-growth rates were measured under various load conditions: load ratios M of 0.5--1.0, frequencies of 10{sup {minus}3}-1 Hz, rise nines of 1--1000s, and peak stress intensities of 25--40 MPa{center_dot}m {sup l/2}.« less


Abstract
Six alloys have been selected as candidate container materials for the storage of highlevel nuclear waste at the proposed Yucca Mountain site in Nevada. These materials are Type 304L stainless steel (SS), Type 316L SS, Incoloy 825, phosphorus-deoxidhT.ed Cu, Cu-30%Ni, and Cu-7%Al. The present program has been initiated to determine whether any of these materials can survive for 300 years in the site environment without developing through-wall stress corrosion cracks, and to assess the relative resistance of these materials to stress corrosion cracking (SCC). A series of slow-straln-rate tests (SSRTs) and fracture--mechanics crack-growth-rate (CGR) tests was performed at 93°C and 1 atm of pressure in simulated J-13 well water. This water is representative, prior to the widespread availability of unsaturated-zone water', of the groundwater present at the Yucca Mountain site.
Slow-strain-rate tests were conducted on 6.35-mm-dlameter cylindrical specimens at strain rates of 10-7 and 10-8 s-l under crevice and noncrevice conditions. All tests were interrupted after nominal elongation strain of 1-4%. Scanning electron microscopy revealed some crack initiation in virtually all the materials, as well as weldments made from these materials.
Fracture-mechanics CGR tests were conducted on 25.4--mm-thick compact tension specimens of Types 304L and 316L stainless steel (SS) and Incoloy 825. Crack-growth rates were measured under various load conditions: load ratios (]R)of 0.5-1.0, frequencies of 10-3-1 Hz, rise times of 1-1000 s, and peak stress intensities of 25-40 MPa_ml/2. The :: measured CGRs are bounded by the predicted rates from the current ASME Section XI correlation for fatigue CGRs of austenitic stainless steel in air. Environmentally accelerated crack gro;',-th was not evident in any of the three materials under the te3t conditions investigated.

Objectives
The objectives of the present program are to (1) determine whether any of the six candidate alloys currently under consideration as waste container materials can survive for 300 years in the Yucca Mountain repository without developing through-w_ll stress corrosion cracks, and (2) rank the candidate materials in terms of their re,_istance to stress corrosion cracking (SCC) in the repository environment.
In the overall test program, approximately 33 SSRTs were performed on six candidate materials, together with eGR tests on six CT specimens from three candidate materials in simulated waste repository groundwater. This is the final report for the program.

Background
In December 1987, Congress amended the Nuclear Waste Policy Act of 1982 and designated the Yu_._a Mountain site in southern Nevada as the nation's first high-level nuclear waste repository, if further investigations confiral that it is suitable. Yucca Mountain is located on the southwestern boundary of the Nevada Test Site, and the geologic horizon at the proposed location for the repository is compacted volcanic ash, or tuff, consisting of quartz, chrlstobalite, alkali feldspar, and other minor phases I situated above the permanent water table. Thus, interaction between groundwater and the waste packages should be minimal.
The high-level waste packages consist of three major components: the metallic containment barriers (the containers or canisters), the waste itself, and other materials, including packing materials, emplacement-hole liners, etc. The metallic containers are intended to provide essentially complete containment of the nuclear waste for 300 to 1,000 years after emplacement.
During the waste isolation or postcontainment period that extends for thousands of years after the metallic containment barriers are breached, the waste itself is expected to control the pate of release of radioactive nuclides into the immediate repository environment.
The temperature of the waste package will be sufficiently high for much of the time so that the limited groundwater in the vicinity of the package will be present only in the form of steam.
Nonetheless, the proposed tests will focus on aqueous environments at _.93°C, because these environments are expected to provide deleterious conditions (which may provide the "worst case") for SCC susceptibility. The elemental composition of Well J-13 groundwater, which is representative of the groundwater present at the Yucca Mountain site, is summarized In Six candidate materials are being tested, namely Type 304L stainless steel (SS), Type 316L SS, Incoloy 825, P-deoxidlzed Cu (CDA-122), Cu-30%Ni (CDA-715), and Cu-7%Al (Al bronze, or CDA-614). 3 These materials were chosen as candiaates because of their good corrosion properties and their extensive use in the marine, nuclear, and process industrles. 4 The alloys were obtained from Lawrence Livermore National Laboratory (LLNL) in the annealed condition. The following weldments, fabricated from these base materials by LLNL, are also being tested (the second alloy indicated in each pair is the weld filler metal): Type 304L SS/308L SS, Type 316L SS/316L SS, Incoloy 825/Incoloy 65, Incoloy 825/Inconel 625, Cu-30Ni/Monel 67, and Cu-7AI/AI bronze A2. The weldments were made by the gas-metalarc welding (GMAW) process.
In addition, LLNL supplied Cu and Cu-7AI weldments produced by the electron-beam welding (EBW) process. These latter weldments were produced without filler metal, and the fusion zones are narrow compared with the GMAW weldments.

Elemental compositions
of the base and weld filler metals are given in Tables 2-7,  and selected mechanical properties provided by vendors are given in Table 8. The yield strengths of the base materials and weldments were determined independently by Argonne National Laboratory at room temperature and at 93°C. and these values are presented in Table 9.

3,1 ExperimentalProcedures
Fabrication drawings for the SSRT specimens are shown in Figs. 1-3, and the locations of welds in the GMAW and EBW specimens are shown in Figs. 2 and 3, respectively.
Similar specimens were used in our earlier studies on SCC of austenitic stainless steels in simulated boiling w_ter environments. 5,6 The small-diameter {=0.8-0.9 mm) through-holes localize the strain and cracking, thereby facilitating observation of SCC | by scanning electron microscopy (SEM). When pins of matching materials are inserted in the holes, crevices are produced.
The SSRTs were conducted in simulated J-13 water at 93°C at strain rates of 10 -7 and 10-8 s-I for the base specimens and were interrupted after elongation (plastic) strains between 1 and 4%. The tests were conducted in a 1.270--L Ni container.
A worm-gear Jactuator, gear reducer, and a variable-speed control system were used to apply the strain. The temperature was maintained by clamp-on heaters with associated instrumentation. A once-through water system with a flow rate of =3 mL/min was used. The SSRT system, including the feedwater system and the electrochemical-potential (ECP) monitoring system, is shov, n in Fig. 4.
For tests W-1 through W-23_ the flow of simulated J-13 water was maintained by an overpressure in the feedwater tank and the flow rate was controlled by a capillary tube connecting the feedwater tank to the test chamber.
After a number of tests, the capillary tube was found to be clogged by the insoluble silicon compounds and bicarbonates. Therefore in tests W-23/W-33, a mechanical peristaltic pump was installed in the feedwater line to ensure a more continuous flow through the environmental chamber. Also, the water chemistry for the early tests was varied from test to test as various simulations of J-13 water were examined, to avoid excessive formation of precipitates at the test temperature. Details of the simulated J-13 water chemistry, its characterization, and the procedures for measuring ECP are described in a subsequent section of this report.
For tests W-13/W-33, a dial indicator was connected to the Jactuator to measure the displacement of the pull rod. This measurement, along with measurements of the system compliance, can be used to estimate the strains in the specimen during the test. Measurements of the total specimen length before and after the tests give even more accurate values of the actual elongation (plastic strain). After each test, the specimen was sectioned by electrical-discharge machining (EDM), as shown in Fig. 5, so that the entire hole region could be examined by SEM at magnifications up to 2000x to detect stress corrosion cracks. 6 In tests W-1 through W-7, pins of a matching material were inserted in the two through-holes to produce crevices. The tests were conducted at a strain rate of 10-7 s 1 and were interrupted after elongations of 3-3.7%.
During the tests, the top crevice was exposed alternately to water and vapor, and the bottom crevice was kept immersed in water. Siphon action in the inverted drain tube shown in Fig. 4 exposes the top crevice to alternate liquid and vapor. The bottom crevice is always immersed in liquid. 'When the hole at the top of the drain tube is opened to the atmosphere, both crevices can be exposed to liquid only.
Tests were also performed on the base materials at a strain rate of 10-8 s -l to investigate the effect of strain rate on SCC. In these tests, one of the holes was left open, but a pin was inserted in the other to produce a crevice. The entire specimen was continuously exposed to the liquid. These changes were made to determine whether a crevice is necessary for cracking of the materials. The tests were interrupted after plastic elongations or strains of _ 1.0--1.6%, which are somewhat lower than the strains at which the tests were internJpted at the higher strain rate, Welded spectmcns were strained at a i i       al = two holes, two pins: one hole exposed continuously to liquid; the other alternately to liquid and vapor. 2 = two holes, two pins: both holes exposed continuously to liquid. 3 = three holes, three pins: all holes exposed continuously to liquid. 4 = two holes, one pin: both holes exposed continuously to liquid.   In the tests on weld specimens, all of the crevices were continuously immersed in water.
lt should be pointed out that in the series of tests conducted at the two strain rates, namely, 10 -7 and 10-8 s-I, different exposure conditions were used. In the tests conducted at the faster strain rate (tests W-1/W-7), pins were inserted in both holes of the specimen (two c-evlces). Only one of the holes was exposed continuously to liquid; the other was exposed alternately to liquid and vapor phases.
In addition, the water chemistry for the first test (W-l) was vastly different from the rest of the tests. In the tests conducted at the slower strain rate of 10-8 s -l, the water chemistry is not comparable to that used for the tests conducted at the faster strain rates. Also, in the slower strain rate tests, only one of the holes was left open (noncrevice), and a pin was inserted in the other to produce a crevice. Both holes were continuously exposed to simulated J-13 water. Because of these differences in test conditions, the strain-rate effects observed in this work should be regarded as very preliminary.
The varying experimental conditions for the SSRT tests are summarized in Table I0. These differences should be borne in mind when reviewing the test results.

Observations of SCC by SEM
The SSRT results for the initial base metal tests are summarized in Tables 11 and 12 for the weldment tests in Table 13. The SEM observations for all the tests conducted at a strain rate of 10-7 s -I are summ.artT.ed in Tables 14 and 15. The results of the initial base metal and weldment tests show some cracking in virtually all the specimens, including those tested at the slower strain rate of 10-8 s-l. However, in most cases, crack observations were made at magnifications as high as 1000-2000x, because it was difficult to observe well-defined cracks at magnifications lower than lOOx. Hence, the actual crack depths inferred from the size of the crack opening are _mall (in the micron range) even for Type 304L SS. These findings are consistent with our unsuccessful efforts thus far to observe cracks >10--20 _tm in depth in the transverse metallographic specimens of Type 304L SS. The cracks in Type 304L SS are more open (and hence are probably deeper) than those in Incoloy 825, as shown in Fig. 6. This is consistent with the expected increased resistance of Incoloy 825 to SCC because of its higher Ni content. However. Incoloy 825 shows a larger number of cracks than Type 304 SS.
The cracks in Incoloy 825, which appear to have initiated at machining marks that are favorably oriented for cracking, appear as thin straight lines in Figs. 6 and 7. lt is not clear whether the larger number of cracks in the Incoloy 825 is due to the machining marks.
Such straight marks are also frequently seen in specimens of other' materials, although they were not observed in our earlier studles on SCC of austenitic stainless steels, s,e These machining marks are attributed to the use oi'a slightly modified drilling procedure that was adopted 'after completing 25 tests. However, this change in procedure was not found to affect the stress/strain response of the materials in the simulated J-13 water environment.
For the Ivcoloy 825 weldment, where the base metal is a different heat, poor surface finish made it difficult to determine whether cracks were present in the base metal, as shown in Figs°7a and 7c. The filler material, however, showed pronounced cracking (Fig.  7b).
Scanning electron microscopy of the Cu and Cu-base alloy spcclmens reveals that these materials also showed some signs of cracking in simulated J-13 water. As shown in Fig. 8, there is no discernible difference in cracking among the three Cu-based materials. Cracking in Cu (Fig. 8a) and in Cu-301,_i ( Fig. 8c) occurred both at the mac:,ining marks (which are nearly straight) and elsewhere (cracks with wavy appearance).
On the basis of these observations, it appears that the differences in SCC susceptibility of the three Cubased materials are small compared to those, for example, between Type 304L SS and Incoloy 825. Further quantification oi' SCC in terms of crack depth and growth rate is required to better delineate differences in cracking behavior.
Cracking has also been observed in all the weldments tested. For example, the Type 316L SS weldment with matching filler metal shows cracking both in the weld and base material (Fig. 9). The Cu-3ONi weldment specimen also exhibited cracking, and there was no clear--cut difference in the appearance of cracks in the base, weld, and heat-_,d'fected zone (Fig. I0).
The cracking behavior of the crevices continuously exposed to liquid was not much different from that of crevices alternately exposed to liquid and vapor during the tests. This is illustrated in Fig. 11 for an Incoloy 825 specimen, lt also appears that a crevice is not required for SCC to occur in these materials.
As shown in Fig. 12. the cracking at holes with and without pins appears similar for Incoloy 825 and Cu-3ONi. On the basis of SEM observations, it is also difficult to determine whether the change in strain rate had any effect on cr"acking, because the observed differences are small and the tests performed aThe local strain is computed as the product of a strain concentration factor I_, estimated from Neuber's rule, and the nominal strain in the unreduced gauge section, which is determined from the stress-strain curve and the measured load on the specimen.
Kr depends on the material and strain level.
dToo low to estimate accurately.   b The specimen has two holes with pins of the same material inserted into them to produce tight crevices. The specimen was exposed to J-13 water irl such a way that the top hole with the pin was alternately exposed to water in liquid and vapor phases, but the bottom hole was always immersed in J-13 water.
c This specimen was pulled to failure in liquid nitrogen following the interrupted test to determine the feasibility of determining the crack length, but it was not possible to examine the extent of cracking on the fracture surface.
at the slower strain rate 10-8 s -I were interrupted at lower strains than those performed at a strain rate of 10-7 s -I (see Tables II and 12).

Estimation of Local Plastic Strains
To _late the observed crack initiation and growth to plastic strain, it is necessary to estimate the local plastic strain at the hole.
The local strain is computed as the product of a strain concentration factor Ke, estimated from Neuber's rule, and the nominal strain in the unreduced gage section, which is determined from the stress/strain curve and the measured load on the specimen. b Except for the electron-beam-welded specimens, each specimen had three holes located as shown in Fig. 2. Each hole had a pin inserted into it. The electron-beam-welded specimen has two holes with pins. The specimens were completely immersed in simulated J-13 water duilng testing. c Electron-beam-welded.
Strvss/strain curves were determined for ali the materials from conventional tensile tests in air at a strain rate of-4.7 x 10-4 s-l. These stress/strain data were analyzed by the Ramberg and Osgood equation and Neuber's rule to determine the strain concentration at the hole as a function of plastic strain. As can be seen from Figs. 13 and 14, the results for Cu and Type 316L SS show that the strain concentration factor decreases with plastic strain, as expected.
Strain concentration Ke can also be computed by applying a simple power law relation between stress and plastic strain, namely, E 1 E where n = _ and C;o= _t-_• Locally, near the hole, Equations 8 and 9 imply that Ko--Kt (10) Substituting Eq, I0 in Neuber's rule, Eq. 3, and solving forK_, The two approachesforestimating ICegivealmostidentical results even forrelatively low plastic strains, as illustrated irt Figs.13 and 14.

Relative SCC Susceptibility of Waste Container Materials
To obtain additional information on the relative cracking susceptibility of the various materials, the load-deformation behavior of the specimen during the SSRT test was analyzed, and a cracking index parameter or stress ratio (SR) was formulated by comparing the stress in the environment with that in air at the same plastic strain.
For the present geometry, the SR, which characterizes the effect of the environment on the capability of material to sustain load in the plastic range, can be defined as 1131 where Sen is the nominal stress in the environment, aair is the nominal stress in air evaluated at the same plastic strain, and Cy is the yield stress of the material (see Table 9).
To minimize geometry effects, Oen and _air are determined for the same specimen geometry. Local plastic yielding is assumed to occur when the nominal stress reaches ay/3, because the stress concentration factor for the present specimen geometry is =3. The use of stress instead of load eliminates any possible differences resulting from variations in specimen diameter.
The SR focuses on the changes in stress after plastic yielding, because SCC is not expected without plastic deformation.
By eliminating the elastic contribution to stress, the SR appears to provide a sensitive indicator of SCC and is useful for screening materials, at least qualitatively.
As currently computed, we assume that the differences in strain rate between the In-air and SSRT values do not significantly affect the stress/strain response at the temperature of interest. With no cracking, SR = I, and if the material is susceptible to cracking, the value is < I. However, a measured value of 1 may indicate only that the degree of cracking is so small that the change in loads cannot be detected.
Although values of SR < 1 indicate that cracking has occurred, tests at different strain rates are needed to determine whether the cracking is environmentally assisted.
To determine the SR, the strains in the SSRT must be known. By measuring the pull-rod displacement during the test, we can determine the strains in "real time" by finding the compliance of the SSRT machine and then using the measured compliance to compute the strain from the measured loads and displacements, as is done in a conventional tensile test. Because the pull-rod displacement measurements were not made for the first 13 tests, "post-test" estimates can instead be obtained from the measured final plastic strain and load at the time of interruption of the tests. The total strain at the end of the test efinal is then £final where the stress is computed from the final load and the initial cross section of the specimen.
The total strain of the specimen as a function of time, e(t), is computed by assuming a constant strain rate _.
The plastic strain as a function of time, ep(t), is then Calibration tests show that the assumption of a constant strain rate is quite good, once yielding has occurred. Thus, plastic strains greater than -0.5% may be accurately determined. Figure 15 shows the variation of SR with plastic strain at a strain rate of 10--7 s -I for the six materials in J-13 water at 93°C. The ranking of materials in order of increasing resistance to cracking is '1_'pe 304L SS < Type 316L SS < Incoloy 825 and Cu-30Ni < Cu and Cu-7Al. Thus, P-deoxid 'ized Cu and Al bronze appear to be more resistant to cracking than the other alloys. The higher cracking resistance of Incoloy 825 relative to Type 304L SS can be attributed to the higher Ni content of Incoloy 825 and is qualitatively consistent with SEM observations of the cracks. The differences in cracking behavior of Incoloy 825. Cu-30Ni, Cu, and Cu-7Al are small, a finding that is also qualitatively consistent with the SEM observations. Also, the SR appears to be fairly sensitive to cracking even when the cracks are too small to be readily observed metaUographically.
lt is clear that a greater change in stress/strain response in the environment than in air is produced in the more SCC-susceptlble Type 304L SS than in the more resistant Incoloy 825 or copper and its alloys. A difference is also observed in the stress/strain behavior of the smooth specimen and the specimen with two holes in air; this is qualitatively consistent with the behavior expected from a reduction in cross-sectional area. For example, an 0.8-0.9 rnm through-hole in the gauge section causes a reduction in area of-20%.
But the difference between stress/strain response of the specimen with holes and that with holes and cracks cannot be explained on the basis of a reduction in area because the enlargement of the holes due to observed cracking is small. Therefore it is possible that part of the shift in stress/straln curve of the specimen in the environment relative to that in air can be attributed to the difference in strain rates used for tests in air and tests in tile environment, at least for Type 304L, which is known to exhibit anelastlc effects at low temperature.°A strain ratio analogou, to the SR can be defined: .Ep(air)l ,

£p(en)j_
where the strain in air and in the environment are evaluated at the same nominal stress.
A.,_can be seen from Fig. 16, the SCC susceptibility rankings determined from the two ra_tG_ jre similar. However, the strain ratio Indicates a greater degree of susceptibility for Type 304L SS, relative to Type 316L SS, than is suggested by the SR. Values of the strain ratio at the evd of the tests are included In Tables 11 and 12

SCC Susceptibility of Weldment Specimens
The results of SSRTs on weldment specimens of several materials are summarized in Table 12. The variation of SR with plastic strain or elongation for Types 304L SS and 316L SS,Incoloy 825,Cu, suggests that welding has a more deleterious effect on the cracking of Cu and Cu-30Ni than on that of the austenitic alloys. Because the heats of the base materials used in the fabrication of the Incoloy 825, Cu, and Cu-30Ni weldments were different from those used for the unwelded specimens, it is necessary to clarify the role of heat-to-heat differences in these tests compared with differences associated with welding. The SEM observations of the base metal used in the unwelded specimen (Fig. 1la) and the heat of the base metal used in the weldment (Fig.  7a) show different cracking behavior, possibly associated with heat-to-heat variation.
As is shown in Fig. 22, the SR suggests that Incoloy 825 with the filler material Inconel 625 is more resistant to cracking than is a similar weldment with the filler material Incoloy 65. The composition of the latter is similar to that of Incoloy 825. Anomalous behavior has been observed for Al bronze weldments produced by both GMAW and EBW. In these cases, the observed SR ts greater than I (Fig. 23), and further work is necessary to explain this unexpected behavior.   I  I  I  '  I  I  I  ' I i I f '    I x 10-7 s -I)

Influence of Strain Rate on SCC
Figures 24-29 sh_:v the effect of strain rate on the SR for Types 304L and 316L SS, Incoloy 825, Cu, Cu-TAI, and Cu-30Ni at different plastic strains.
The strain rate appears to influence the SR of Types 304L SS and 316L SS, Incoloy 825, and Cu, but it has virtually no effect on the SR of Cu-30Ni and Cu-7AI. The effect is more pronounced for Type 304L SS than for Type 316L SS or the other materials.
The lower values of the SR at a lower strain rate suggest an increase in susceptibility to cracking typical of SCC. Similarly, these results suggest that the cracking observed for Cu-3ONi and Cu-7AI at the higher strain rate may not be SCC. However, as pointed out earlier, both the exposure conditions and water chemist y are different for the tests periormed at the two strain rates, lt appears that the strain-_.,te effects on the SR are coupled with the environmental effects. More specfficaily, the simulated J-13 water used for the slower strain-rate experiments in tests on Types 304L and 316L SS and on Incoloy 825 are more oxidizing, which is reflected in the higher values of specimen ECP . A more oxidizing environment also tends to lower the SR. As will be discussed later, even at the same strain rates, the change in the SR appears to be correlated with the change in the specimen electrochemical potential (ECP). These results suggest that the ECP, pH, and other environmental variables must be controlled to more clearly delineate the straln-rate effects on the SCC susceptibility of diffel ent materials.

Measurement of Crack Depth in SSRT Specimens
Attempts have been made to measure the crack depths by transverse sectioning of the SSRT specimens.
In Type 304L SS (test W-4), multiple sections were examined by alternately grinding away _4-5 mils of material and metallographicaUy preparing the resulting new surface. Even after as much as 20 mils of material had been removed in this manner, only small cracks (_ 10 _m in depth) were observed. Even these cracks may be artifacts of the procedure used to produce the small-diameter holes in the gauge region. However, such cracks have not been observed in unstressed specimens with holes exposed to simulated J-13 water at 93°C for the same length of time. Consequently, it appears that we must modify our testing methods to produce cracks of greater depth. For exmnple, it may be desirable to consider tests strained to failure under identical environmental conditions so that the relative susceptibility of the materials can be quantified in terms of maximum crack depth and crack depth distribution or crack growth rate. This approach may provide another means of resolving the relatively small differences in the cracking behavior of the most resistant materials, and may also confirm some of the preliminary findings of the interrupted tests. Further attempts to determine the crack depth in specimens subjected to larger strains than in test W-4 will be discussed later.

influence of Water Chemistry on SCC
Tests W-1 and W-2 were performed on Incoloy 825 with different concentrations of ionic species in the water, but at the same strain rate of 1 x 10-.7 s -I. The cracking behavior in terms of SR (Fig. 30) appears to be very similar, The appearance of the cracks in the two specimens as observed by SEM is also very similar. Although the water chemistry is different for these two tests, the approximately identical stress ratio observed (see Table 11) is consistent with the ECP of the specimen (233--253 mV standard hydrogen electrode [SHE]) measured in the two environments (Fig. 30).
Tests W-5 and W-12 on Cu are similar, except that in test W-12 a small amount of H202 was added to the feedwater. Although this addltlon did not significantly change the ECP of the material, the lower SR (Fig, 31) suggests that H202 may be detrimental. However, it was difficult to observe differences in cracking by SEM.

SSRT Results (Tests W-23/W-33)
Because of the controlled water chemistry and water flow through the the environmental chamber as a result of introducing a mechanical pump discussed earlier, several tests were repeated for the five candidate alloys, namely, Types 304L SS and 316L SS, Incoloy 825, Cu-30Ni, and Cu-7Al. As shown in Table 16, all but two of the tests were performed at a strain rate of 1 x 10 -7 s -I The tests for "l_ype304L and 316L SS involved large plastic elongations, and these specimens were metallographically examined to determine the deepest crack or crack distribution.
Such observations would permit a better understanding of the relative SCC susceptibility of these materials.
In addition, irl general, the tests (W-23/W-33) were generally Interrupted at higher strains (ep = 5-18%) than were tests W-2/W-7 (ep = 3-4%) and these tests permit screening of the five alloys under better-defined environmental and slightly lower ECP conditions than the early tests. Furthermore.
for Type 304L SS, Incoloy 825. and Cu--30Ni, duplicate experimental results are obtained to check the reproducibility of stress/strain data used to obtain SR, because the rankings of different materials used irl the program are solely based on stress or strain ratio, For Cu-30Ni and Cu-7Al alloys. experiments were performed at the slightly lower strain rate of 5 x 10-8 s -l to determine any observable effect of strain rate on the SR. This strain-rate was chosen so that these tests could be interrupted at higher plastic strains near those of tests run at faster strain rates within a reasonable length of time.
For previous tests (W-2/W-7). we have computed strains from the measured plastic strain at the time of interruption and interpolated back by means of strain rate and time (strain-rate method, Eq. i6). For tests W-23 and W-33, we have computed strains by using both the strain-rate method and the displacement data from the dlal indicator on the Jactuator.
The extension data were recorded manually during the course of experiments, and these data, whlch are considered more accurate than those estimated from the strain-rate method, provide a comparison of SR values obtained by the two different methods.
Additionally, for the last two tests (W-32 and W-33), an L_rDT extensometer was attached to the specimen, and the elongation strains estimated from the displacement data provide another independent strain measurement.
The data indicate that evaluation of materials using stress and strain ratios is not affected by the manner in which the strains are computed in the SSRTs.

Crack Depth Distribution in SSRT Specimens
The SEM observations showed cracking in all specimens, and it was difficult to quantify these observations to determine whether cracking is more severe in specimens subjected to larger plastic strains and which material is most resistant to cracking.      Fig. 32. The distribution of maximum depths observed in four to five transverse sectlons is shown for both Types 316L and 304L SS specimens in Fig. 33. These results show that even in the materials most susceptible to cracking in simulated J-13 water, the maximum cracking observed is still small even after large plastic elongations. Therefore, the depth distribution results for the materials may not be very useful in relating to the SR, at least for the materials and environment under consideration.

Relative SCC Susceptibility oi Waste Container Materials
For tests W-23/W-33, we have examined the effect of different strain computation methods on calculated SR and reproducibility of the SR in duplicate tests performed on three materials, namely, Type 304L SS, Incoloy 825, and Cu-30Ni.
Once the most accurate method of computing the SR and its reproducibility is established under controlled environments, it may then be possible to screen two materials for resistance to SCC even though the difference in the SCC susceptibility is qualitatively small.
For tests W-23/W-33, the stress or strain ratio can be computed by using three measurements: strains estimated from the strain-rate method, pull-rod displacements from the dial indicator located on the Jactuator in the SSRT system, and displacements in the SSRT specimen determined with the LVDT extensometer. For displacements measured with the dial indicator, assuming that all of the plastic displacement ff_p) occurs in the SSRT specimen, 5p is given by where 8t Is the total displacement of the specimen and the loading system, P is the load, and Kt is the stiffness of the specimen and the machine.
The dt must be corrected for any backlash in the system. For the two SSRT systems, this backlash has been estimated 9 as 0.0025 and 0.0038 cm, depending on the machine used. Kt can be calculated from 8t and 8p with the values at the time of test interruption.
Here, 8t is obtained from the dial indicator and _ (Ep = 8p/L, where L is the gauge length of the SSRT specimen) is the plastic elongation determined by measuring the specimen length before and after the test. With the value of Kt thus determined, plastic strain can then be determined as a function of nominal stress from the load-extension data for the tests under consideration. The stress or strain ratio Is then estimated from the stress/straln data In a manner similar to that discussed belbre.
For tests W-32 and W-33, the LvITr extensometer attached to the specimen was used to obtain the displacement in the specimen as a function of P. For this case, _p is given by where Ks is the specimen stiffness equal to 0.93EA/L, E is the Young's modulus, and A is the cross-sectional area. The factor of 0.93 is introduced to account tbr the specimen shoulder deformation. 1o The specimen stiffness values thus determined for Incoloy 825 [test W-32) and Cu-30Nt [test W-33) are 1.56 x 107 and 1.23 x 107 kg.m" 1, respectively. The plastic strain is computed by dividing the plastic displacement by L. From the plastic strain determined as a function of nominal stress, the SR or the strain ratio is computed as before. As can be seen, the method of strain computation has virtually no effect for Incoloy 825 and only a small effect for Cu-30Ni at strains <6-7%. For both materials, the strain computation method using the dial indicator and extensometer gives very good agreement. We will later examine the ranking of different alloys in terms of increasing resistance to SCC on the basis of these data.  Fig. 44), "l_pe 316L SS (W-6 and W-24, Fig. 45), Incoloy 825 (W-2 and W-32, Fig. 46), and Cu-30Ni (W-3 and W--33, Fig. 47). However, [.here is close agreement between the results of two tests (W-7 and W-29, Fig. 481 for Cu-7Al, which also happens to be the most crack-resistant material. Similar comparison data for Cu are not avallable.

The disagreement
in results for the four alloys cannot be reconciled in terms of the method of computing strains (because all the methods discussed in thls report essentially lead to approximately identical results), but instead appears to be associated with the change in environmental and exposure condltions. In fact, the disagreement in results is better correlated with specimen ECP, at least for Type 304L (Fig. 441, Type 316L SS (Flg. 45), and Incoloy 825 (Fig. 46). In particular, the higher specimen ECP tends to give a lower SR, i.e., higher specimen ECP consistently results in g_Jater susceptibility. For tests conducted under approximately the same ECP conditions, almost identical results for the SR are obtalned, as shown in Figs. 30,39,41,and 43. lt is important to note that despite the vast difference in water chemistry for tests W-1 and W-2 (Fig. 301,  approximately identical values for the SR are obtained. This is consistent with the fact that the same ECP values were measured in these two tests. Although it is easy to comprehend why a slightly more oxidizing environment produces a greater susceptibility to cr,_cking with a resultant decrease in the SR, it is more difficult to obtain a fundamental understanding of how a shift in ECP value in the range of 250--380 mV(SHEI can affect the cracking susceptibility of austenitic stainless steels in simulated J-13 water at 93°C. Such a shift in ECP has not been found to significantly affect the SCC susceptibility of .Types 304 and 316L SS under boiling-water-reactor operating conditions.        Plastic 8train (%)

Fig. 48. Comparison of Stress Versus Plastic Strain for Two Tests on Cu-7Al in Simulated J-13 Water at 93°C _ = I x10 -Ts -I)
We can examine the ranking of the five alloys in terms of stress and strain ratios estimated from the dial-indicator and extensometer measurements of displacements in the specimen from tests W-23/W-33. Figure 49 is a plot of stress ratio versus plastic strain for the five candidate materials based upon the dial indicator readings. In this plot, we have included the duplicate test results for Type 304L SS, Incoloy 825, and Cu-30Ni (represented by the same symbol for each mater]al).
On the basis of this plot, the ranking of materials in order of increasing resistance to cracking is Type 304L SS < Type 316L SS < Incoloy 825 < Cu-30Ni < Cu-7AI. These results are similar to the ranking for tests W-2/W-7 except that Incoloy 825 appears to be slightly inferior to Cu-30Ni in cracking resistance.
A similar ranking is observed when the strain ratio is used (Fig. 50). The slightly superior SCC resistance of Cu-30Ni compared to that of Incoloy 825 can also be seen from Tests W-32 and W-33, where specimen displacement measurements were made with the LVDT extensometer.
Both of the tests for this case were interrupted at -12.0% elongation, strains. The slightly superior SCC resistance of Cu-30Ni, compared to that of Incoloy, in terms of both stress and strain ratios is shown in Fig. 51. lt is important to note that the above ranking for the five materials is again not affected by the method of computing strain, as is evidenced from Figs. 52 and 53, where stress and strain ratios are estimated from the measured plastic strain and strain rate. "[_e significance of the slight differences in resistance to SCC observed in this study for the two competing materials (Incoloy 825 and Cu-30 Ni) requires a better understanding of the stress and strain ratios, -" as weU as conltrmatlon by omer types of tests such as ira_;wrv-m_-c,am_ _._ _c_L=. Stress ratio is based on estimating strains n the SSRT specimens bN means of a dlal indicator (_ = I x 10 -7 s-l).  Stress ratio is based on estimating _ rate method [Eqs,(15) and (16) Stress ratio is based on estimating strains in the SSRT specimens by means of the straln-rate method [Eqs. (15) and (16)I.

Influence of Strain Rate on SCC
Two tests (W-27 and W-28; see Table 15) have been performed at the slower strain rate of 5.0 x 10--8 s-I, one on Cu-30Ni and another on Cu-TAl. Comparison of the results wlth those obtained at the slightly higher strain rate of 10-7 s -I showed no significant effects of strain rate on stress ratio. However, these tests show the potential for studying the effects of strain rate on cracking with the use of stress or strain ratios and hence for establishing whether the cracking ts environmentally assisted.

Preparation and Analyses of Test Solutions
The effluenbwater chemistries for the completed SSR:I" tests are reported in Tables 17-20. In test W-1, the concentrations of most of the species were higher by a factor of =20 relative to J-13 water; however, the values for Ca and Mg were very low (<0.04 ppm) to prevent precipitation of slightly soluble compounds such as CaC03, CaF2, MgF:_, CaSIO3, and MgSiO3. The HCO_ concentration was obtained by bubbling a 12% C02-20% 02--68% N2 gas mixture through the feedwater and maintaining a 5 psig overpressure in the feedwater tank. This gas mixture, also set the dissolved-oxygen concentration of the --feedwater at ,,7-8 ppm. The I4CO_ concentration in the effluent water was higher by a o factor of only .,6 than that in J-13 water because some of the CO2 escapes into the j atmosphere in the test vessel at 93°C. The relatively high concentration of sodium silicate -i and low concentration of HCO_ are reflected in the rather high pH value of =9.2.
Precipitation of silica compounds on the Incoloy 825 SSRT specimen and the test vessel ==' also occurred with this nominal 20x water chemistry.
-! 11 !  a Cation and anion concentrations determined by ICP spectroscopy and ion chromatography analyses, respectively, with the exception of  which was obtained by titration.   In tests W-2/W-7, the efffluent-water chemistry more closely approximated that of J-13 water; however, the concentrations of some of the anions (F-,CI-, and NO_), we_ higher by factors of --2-20 and the concentrations of HCO_ were lower by factors of -2-4.
The 30-ppm Si (_80 ppm (--80ppm SiO 2-) and 78-19-ppm HCO_ in the feedwater were obtained by adding Na2Si03 and bubbling with a 12% CO2-20% 02-68% N2 gas mixture, as in test W-1. The PH25oc of the effluent water ranged from 6.9 to 7.8 in these tests. Nickel from corrosion of the test vessels was also detected in the effluent water (typically <2 ppm). To increase the HCO_ concentration to the level in J-13 water (i.e, to --143 ppm), this species was added as NaHC03 and the gas mixture in the feedwater tank was changed to 20% 02-80% N2 in Tests W-8 and W-9. Approximately I ppm of H202 was also added to the feedwater after sparging with the 02-N2 gas mixture.
The measured H202 concentration in the effluent water was _0.2 ppm.
Beginning with test W-10, Na2Si03 was no longer added to the feedwater. "I_e reason for this change was that although silica solubility increases as temperature increases, the amount of material in solution depends on the phase (crystalline or amorphous) in equilibrium with the water, as well as on the nature and concentration of other species that are present.t1 The soluble form of silica is the monomeric species Si(OH)4, i.e., orthosillcic acid. Polysiliclc acid with molecular weights, as SiO2, up to 100,000 ainu and particle sizes <50 }k in diameter, as well as colloidal silica with a range of pm_icle sizes, can also be present in the water.
But impurities in the water can have a significant effect on the solubility of silica. For example, A1203 or AICI3.(6H20) can adsorb on silica and cause it to precipitate (i.e., decrease the SiO2 concentrations to <6 ppm}. This readily soluble hydrated salt was used to add Al3+ to simulated J-13 water in our experiments. Also, magnesium ion converts soluble silica (amorphous) to a magnesium silicate and also decreases the solubility. Organic impurities, although very low in our deionized water, can either increase or decrease silica solubility. Because silicates are commonly used as corrosion inhibitors in low-temperature systems, 12,13 it is unlikely that they would contribute to SCC of the candidate materials in our tests.

Although the effluent water chemistries
[n Tests W--8 and W-9 more closely approximate J-13 water (with the exception of Si), the HCO_ concentration was still somewhat lower (=50 versus 143 ppm} and the pH25oc was considerably higher (8.9 to 9.1 versus 6.9). Because of the high pH, CaC03 precipitation occurred, as evidenced by the high Ca content of the scale deposit and the low Ca values in the efflue_it water of Tests W-8 and W-9 (Table 17) relative to the feedwater (3-5 versus I I ppm).
In Tests W-10/W--23, the ionic species were of the same formulation as that in the two previous tests, but a 12% CO2-20°/o 02-68% N2 cover gas was bubbled through the feedwater to achieve the desired dissolved oxygen concentration and to produce an "equilibrium" CO2 overpressure to stabilize the HCO_ concentration and pH of the feedwater.
As is evident in Tables 17-20, the HCO_ and pH25oc values in the effluent water range from 125 to 141 ppm and 7.8 to 8.5, respectively. The effluent Ca and Mg concentrations were virtually the same as in the feedwater, and very little deposition occurred on the SSRT specimens or the vessel wall in these tests. With the exception of the Si concentration (<0.1 ppm), this water chemistry is a good simulation of the ionic and dissolved-gaseous species in J-13 water, although the pH25oc values are still somewhat higher than the reported near-neutral value for J-13 water. This, in part, can be attributed to some loss of CO2 to the gas phase above the water (i.e., air entrapped by a snug metal lid that fits on top of the test vessel) at ambient pressure.
The effluent water analyses from tests W-10/W-23 indicate that the feedwater can be made up in a reproducible manner.
Beginning with test W-10, a pH electrode was incorporated into the test vessel to obtain in-situ values at the test temperature of 93°C. The measured pH values at 25 and 93°C are given in Table 21. Average values of the measured effluent pH25oc and the pH93oc in the vessel for tests with the same feedwater chemistry (W-10/W-23) were 8.2 and 7.0, respectively, or approximately one pH unit on the basic side of neutral at the two temperatures.
The pH25oc values in tests W-l/W-9 encompass the range of =7.3 to 9.2, or approximately one pH unit on either side of the 8.2 value for the water chemistry that was used in the majority of the SSRTs.

Measurement of Electrochemical Potential
The ECP of the insulated SSRT specimens was measured against an external 2 x 10--4 M KCI/AgCI/Ag reference electrode at regular intervals during the test. Beginning with test W-19, a small rod of the container alloy and a Pt wire were also inserted into the test vessel to obtain an independent measure of the ECP of the material and the Pt, as well as of the SSRT specimen and the Ni vessel. The values for each electrode were plotted versus time, and the steady-state ECP values obtained from the plots are given in Table 22. Typical plots of the measured ECP of the SSRT specimens versus time are shown in Fig. 54 for Incoloy-825 and the Cu-6% Al alloy in tests W-13 and W-15.
The measured values were also converted to the SHE scale. The SHE conversion potential for an external AgCI/Ag reference electrode at 93°C as a function of KCI concentration is shown in Fig. 55 and was obtained from relations for the thermocell and liquid Junction potentials.
The calculated value for an electrolyte concentration of 2 x 10-4 M (obtained from the figure) is +365 mV. An experimental SHE correction potential of +378 mV for this electrode was obtained against a commonly used AgCl/Ag reference electrode with KCI electrolyte concentration of 0. I M. The latter value was used to convert the steady-state ECP values in Table 22 to the SHE scale. The influence of temperature on the SHE correction potential for the 2 x 10-4 M KCI/AgCI/Ag reference electrode was also calculated, and the result is shown in Fig. 56, The results in Table 22 indicate that the ECP values for the SSRT specimens range from =230 to 310 mV(SHE) for Cu and Cu-base alloys and -_320 to 390 mV(SHE) for the = austenitic stainless steels and Incoloy-825 in nominally the same water chemistry (tests W-10/W-23).
A comparison of these values with results for the con'esponding alloys in tests W-l/W-9 indicates that the ECP values are not strongly dependent on the water chemistry of these experiments.
The ECP values of electrodes fabricated from the same materials as the SSRT specimens are also shown in Table 22 fox"tests W-19/W-23.
The " ECP values of the unstrained austenitic steel electrodes are lower by _ 100 mV than those i of the SSRT specimens (strained electrodes).     HH2125F). The materials were tested in the as-recelved millannealed condition without additional heat treatment.
Elemental composition and mechanical properties of the materials are shown in Tables 2-9.
Side grooves having a semicircular cross section were cut into both sides of the specimen to a depth of" 1.27 mm to restrict crack growth to a single plane. The design of the CT specimens is in accordance with the ASTM specification E399, 14 except fo: the side grooves and six small threaded holes on the front face for instrumentation, as shown in Fig. 57. The stress intensity factors were calculated according to ASTM specification E-399. The direction of crack extension was perpendicular to the short transverse thickness direction of the plates. Rolling direction of the plates was not provided by LLNL or the vendors.
Metallographlc examination of the materials did not reveal any clues ibr rolling direction.
The specimens were fatigue-precracked irl air at room temperature for a length of 1.91 mm to introduce a sharp starter crack The initial machine notch measured from the load line was 17.78 mm. An isosceles triangular loading waveform at a frequency of I-2 Hz, a load ratio R (a ratio of minimum to maximum load) of 0.25, an initial maximum stress intensity of 16.1 MPa.m !/2 and a final maximum stress intensity of 17.5 MPa.m I/2 was used for precracking.
This final maximum stress intensity value is 70% of initial peak stress intensity ibr the subsequent crack-growth-tests in a simulated J-13 well water.
The initial peak stress intensity value for the tests in the simulated J-13 well water was chosen to be about 25 MPa.ml/2.
Welding residual stresses are the most important driving force for cracks.
Under these loads, the largest marginally detectable defects that would be created during the fabrication of a waste container would have associated stress intensity values approximately half the value chosen for these tests, which provides some conservatism.
The CGR tests were performed in a 5-L nickel vessel with a once-through flow system at a flow rate of 3 mL/min, under 1 atm pressure at 92-94°C. The vessel was not hermetically sealed. The simulated J-13 well water was prepared from deionized high-pu_'Ity water (resistivity >16 Mi'l.cm) and reagent-grade-purlty salts of CaS04, Ca(NO3)2, CaCI2, FeCl2, Li2SO4, MgS04, MnSO4, AICI3, Na2C03, NaHCO3, KHCO3, Na2SiO3, and HF. H_gh-purlty mixed gas containing 20% 02, 12%CO2, and 68% N2 was used as a cover gas at 3--.5 psig to maintain the desired dissolved O2 and HCO_ concentrations. Table 23 shows a typical composl.tion of J-13 well water I and analysis results for the simulated test solution used for the CGR t¢'_ts. The test solution is a good simulation of the ionic and dissolved gaseous species in the J-13 well water, with the exception of the Si concentration.
Si was not_ _dderl tn the te_t solution to avoid precipitation of silicate.
_ilicates are commonly used as corrosion inhibitors for low-.temperature appllcatlons,12.13 and it is considered unlikely that sillcates would contribute to envirorcnentally assisted crack growth of the test materials.
The higher PH25oc values for the effluent test solution are due to some loss of in Table 24. Three fatigue-precracked specimens were loaded simultaneously in a dalsy-chain serial loading arrangement.
The crack length in each specimen was continually monitored by means of a electric potential drop method.15 Potential drop across the specimens was about 40--50 laV at a measuring current of 2 A, and thermal electromotive force was on the order of 0.7-14 _LV. The resolution of the crack-length measurement was 50 _tm. A schematic representation of the experimental setup Js shown in Fig. 58.

High-Load-Ratio Tests
Specimens of all three alloys were initially loaded under a constant (R = I) stress intensity of 25 MPa.ml/2 in the simulated J-13 well water environment, but no crack growth was observed after 315 h. The load ratio was than decreased to R = 0.95, thus imposing a slight "wiggle" on the loading waveform, Out no crack growth was observed after an additional 334 h under these conditions. Following this, R was further decreased to 0.9 and the tests were continued for an additional 18,300 h {more than two yr), and again no crack growth was observed. Figures 59-431 show crack length as a function of tlme for the test times from 9300 to 19,000 h for the three alloys under R = 0.9 loading.

Experimental Setup and Test Specimen
The _naximum average crack growth rate in these tests can be estimated from the resolution of, the crack-growth-measurement technique (50 Ilm) and the time over which no crack extension was detected. Dividing the overall resolution of 50 I_m by the test duration of ]9_000 h results in an estimated maximum crack growth rate of ==8x t0 "13 m/s under the test conditions.
This compares with a maximum allowable crack growth rate of •, 1 x 10-12 m/s for the waste package canister, based upon a wall thickness of 1 cm and a target life of 300 years.

_ow-Load-Ratlo Tests
Crack-growth-rate tests in the same environment, conducted under more severe conditions of greater stress intensities and lower R ratios, resulted in observable crack extension.
In Fig. 62, Type 304L SS under conditions of R = 0.7, K = 36 MPa.m I/2, and frequency of 0.5 Hz, shows a CGR of 4.0 x 10 -9 m.s -I. The results oi' cyclic load tests on Types 304L and 316L SS and Incoloy 825 at R = 0.5 and 0.7 and frequencies of 10-3 to I Hz are summarized in Table 25. For comparison, fatigue CGRs of austenitic stainless steel in air at 93°C computed with the current ASME Section XI correlation, which is based on the work of james and Jones, le are also included. Figures 63 and 64 show a comparison between the CGRs observed for the test materials in simulated j-13 well water and the predicted growth rates of austenitic :_| stainless steel in air, based on the ASME Section Xl correlation. In tl_e lIgures, growth I= rates on the solid line (with slope of unity) would be equal to those predicted by the ASME _| -_ll -I --m correlation.
The observed CGRs for Types 304L and 316L SS and Incoloy 825 are generally lower than the predicted rates.
Because the code correlation represents a 95% confidence limit upper bound for the observed data base, it is expected to be conservative for most heats of material in the absence of environmental effects.
Crack growth rate under cyclic loads in a corrosive environment (da/dt) may be expressed as a sum of contributions by (I} stress corrosion cracking, (da/dt_scc; (2) corrosion fatigue, (da/dt}cF, representing the additional crack growth under cyclic loading due to the environment; and (31 mechanical fatigue, (da/dt)alr, representing the fatigue growth in air: (da/dt) = (da/dt)scc + (da/dt)cF + (da/dt)air.
The first two terms on the right side of the equation are envlronment-sensltive. They depend on loading history variables, such as rise-, unload-and hold-time, as well as on frequency.
In oxygenated-water environments, the environment-sensitive terms can contribute significantly to crack growth rates of austenitic stainless steels. 17-19 Under low-R and high-frequency loading, mechanical fatigue dominates. Environmental contributions would be expected to become more significant as the frequency decreases. Crack-growth rates as a function of cyclic frequency and CGRs per cycle versus rise time for the cun'c_nt tests are shown in Figs. 65-68. Figures 65 and 66 show that the timebased growth rates are proportional to frequency over the entire range of the frequencies used in the tests. This indicates that no erwironmental acceleration of crack growth is present for the test conditions considered. Figures 67 and 68 show that the growth rate per cycle is independent of rise time. This too indicates that no environmentally assisted crack growth occurred, and that the crack-growth mechanism ts pure mechanical fatigue°F or comparison, the environmentally accelerated behavior observed irl hlgh-temperature oxygenated environments 18,19 is shown schematically by the curved lines in Figs. 65 and 66.

Summary and Conclusions
A series of slow-strain-rate tensile tests on six candidate nuclear waste container materials was conducted under both crevice and noncrevlce conditions in simulated well J-13 water at 93°C at strain rates of 10 -7 and 10-8 s -I. The tests were performed under well-characterized environmental conditions. Similar tests were also performed on weldment specimens of Types 304L and 316L SS, Incoloy 825, Cu, Cu-7%Al, and Cu-30%Ni at a strain rate of 10-7 s-l The specimens contained small-dlameter through-holes (with or without pins of a matching material) to facilitate observation of cracks by scanning electron microscopy (SEM). The SEM observations showed cracking in virtually all of the materials under the severe testing conditions employed.
A stress rat lo was formulated on the basis of the ratio of the increase in stress following the initiation of local yielding for the material in water and the corresponding stress difference for an identical test in air at the same elongation.
A ratio of plastic strain in air to the plastic strain in the environment (strain ratio), both evaluated at the same stress, was also formulated to describe the cracking susceptibility.
Higher values of stress or strain ratio _ imply greater cracking or _¢_C susceptibtllty. Tile Stress or straln ratio appears to be uselul in screening the materials for SCC, even though the crack depths are small (<I00 _tm). On ii -! the basis of this stress ratio or strain ratio, the ranking of the materials in order of increasing resistance to cracking is Type 304L SS < Type 316L SS < Incoloy 825 < Cu-30%Ni < Cu and Cu-7%A1. Weldment specimens of Cu-30%Ni and Cu exhibited a somewhat higher susceptibility to cracking in terms of the index compared to the base metal specimens. The cracking index also suggests that a lower strain rate has a deleterious effect on cracking of types 304L and 316L FS, Incoloy 825, and Cu, but has virtually no effect on Cu-30°/oNi and Cu--7%AI. The relative SCC susceptibility of these materials from the preliminary SSRT tests is bei:lg confirmed by tests on fracturemechanics-type specimens.
Fracture-mechanics crack-growth tests were conducted on 25.4--mm-thick compact tension specimens of Types 304L and 316L SS and Incoloy 825 at 93°C and 1 atm of pressure in simulated J-13 well water in both high-and low-load-ratio tests. In the high-load-ratio tests, Kmax was 25 MPa.m 1/2 and R ranged from 0.9 to 1. In the lowload-ratio tests, Kmax ranged from 26.8 to 39.8 MPa.m 1/2 and R was 0.5 or 0.7. The results of these experiments lead to tile following conclusions: (1) No crack growth was detected in any of the alloys tested at K = 25 MPa.ml/2 and R = 0.9-1 for test times of 19,000 h. Based on the resolution of the crack-length-measuring technique, this indicates that the maximum average crack growth rate under these conditions was =,8 x 10 -13 m/s. (2} No cyclic or rise-time-dependent environmental acceleration of crack growth was observable in the simulated J-13 well water at 93°C under the test conditions. (3) Growth rates for R = 0.5-0.7 and maximum stress intensities of 26-40 MPa.ml/2 agree with the current ASME Section XI correlation for austenitic stainless steel within a factor of 0.9-3.8.