SCIENCE

These were interpreted by the extraordinarily large specific surface REFERENCES

area that caused corrosion analogous to that in crevice and by the
capability of hydrolysis reaction among ferrous/ferric Tons and
1. C. Wood, Prog. Nucl. Energy 20(1987): p. 1.
water. 2. W.J. Marble ei al., CORROSION/85, paper no. 107 (Houston, TX: NACE, 1985).
3. K. Yamazaki et al., 4th Int. Conf: Water Chem. of Nucl. Reactor Systems (BNES)
It was confirmed that hydrogen monitoring of iron-suspended (Bournemouth, OK: Oct. 1986), p. 103.
solutions could be a very simple and useful way to investigate cor- 4. K. Otoha et al., 1 st. Int. Conf, Water Chem. in Nuci. Power Plants Operational
rosion in high-temperature water. Silica was proved to be an effec- Experience and New Tech. for Management (Tokyo, Japan: Apr. 1988), p. 315.
5. K. Oshima, ibid., p. 3.
tive corrosion inhibitor by this method, since it generated thin, pro-
6. G.P. Marsh et al., Corrosion Sci. 28(1988): p. 289.
tective layers through the coordination with iron surfaces and the 7. V.J. Linnenbom, J. Electrochem: Soc. 105 (1958): p. 322.
olation among hydroxyl ion and ferrous/ferric ion. The specific 8. W.E. Bornak, Corrosion 44(1988): p. 154.
amount of silica required for corrosion inhibition was estimated to 9. H. Leidheiser Jr. et al., Corrosion Sci. 28(1988): p. 631.
be 3.9 x 10 -10 mol Si/cm 2 Fe. 10. N. Fujita et al., Corrosion 45, 11(1989): p. 901.
11. R.K. (Ier, The Colloid Chemistry of Silica and Silicates (Ithaca, NY: Cornell Univ.
Silica also inhibited corrosion of iron under y-ray irradiations, Press, 1955).
12. E.C. Potter et al., Brit. Corrosion J. 1(1965): p. 26.
whose effectiveness showed a marked dependence on the dis- 13. L. Tomlinson, Corrosion 37(1981): p. 591.
solved oxygen concentrations. This dependence was postulated 14. K. Krauskopf, Introduction to Geochemistry (New York, NY: McGraw-Hill, 1967).
by considering the elemental reactions between e; q and oxygen. 15. F.H. Sweeton et al., J. Chem. Thermodyn. 2(1970): p. 479.
16. P.C. Searson et al., J. Electrochem. Soc. 135(1988): p. 1358.
17. K. Ishigure et al., Nucl. Tech. 50(1980): p. 169.
It was suggested that fine powder of metals could be used ef-
18. Y. Katsumura et al., tst. Int. Cont. Water Chem. in Nucl. Power Plants: Opera-
fectively to investigate the crevice chemistry, since it could offer as tional Experience and New Technologies for Management (Tokyo, Japan: April
large a surface-to-solution ratio as crevice. 1988), p. 511.

The Effect of Low-Temperature Isothermal

Heat Treatments on Intergranular Corrosion
of Type 316 Stai n l ess Steel
Simulated Weld Heat-Affected Zones
M.C. Juhas* and B.E. Wilde -

ABSTRACT

Type 316 (UNSO S31600) stainless steel (SS) is a candidate al-

) tion (DOS) was determined using two corrosion methods; the
loy for containment of high-level nuclear waste in deep under- double-loop electrochemical potentiokinetic reactivation (DL-EPR)
ground storage. The heat-affected zones (HAZs) of welded con- test and the ferric sulfate (Streicher) test. Optical metallography
tainers could become sensitized under the appropriate thermal was used to reveal the presente of grain boundary carbides and
conditions. Prior to the present study, it was suspected that car- analytical electron microscopy (AEM) was employed to determine
bides that nucleated during welding could grow during the con- the local composition of grain boundaries. The corrosion tests in-
tainment life of the nuclear waste. In this investigation, type 316 dicated that significant chromium depletion did not occur regard-
stainless steels containing variations in carbon and nitrogen have less of the heat treatment or carbon and nitrogen content. The
been exposed to thermal cycles simulating a single-pass weid Streicher tests showed that grain boundaries were selectively at-
HAZ, followed by long-term (2000 to 2800 hours) isothermal heat tacked after the HAZ simulations; however, the ensuing low-tem-
treatments in the range of 300 to 400°C. The degree of sensitiza- perature exposure did not increase the susceptibility to intergra-
nular attack (IGA). Discrete gram boundary carbides were
observed on 15 to 25% of the boundaries in the high-carbon al-
loys (0.08 wt% C). The AEM results showed that when carbides
"Submitted for publication May 1989; in revised form, May 1990. were present, the chromium level at the gram boundaries was
'Laboratoire de Métallurgie Structurale, Université de Paris-Sud, Batiment 413, slightly depressed with respect to the matrix level of 16.7 wt%;
91405 Orsay Cedex, France. however, the minimum level was never less than approximately
**Fontana Corrosion Center, Department of Materials Science and Engineering, 16.0 wt%.
The Ohio State University, 116 W. 19th Ave., Columbus, OH 43210.
t' 1 UNS numbers are listed in Metals and Alloys in the Unified Numbering System,
published by the Society of Automotive Engineers (SAE) and cosponsored by the KEY WORDS: AEM, EPR tests, low-temperature sensitization,
American Society for Testing and Materials (ASTM). sensitization, stainless steel

0010-9312/90/000233/$3.00/0
812 © 1990, National Association of Corrosion Engineers CORROSION -October 1990
 SCIENC E

INTRODUCTION The design of permanent containment for high-level nuclear

waste has been the subject of national interest for the past
decade. 49 Type 316 SS has been considered among many candi-
Austenitic stainless steels have been used extensively in high-tem- date materials for this purpose. After emplacement of the nuclear
perature applications such as turbines, steam generators, and eip- waste, the containers would be permanently sealed, most likely
ing in the nuclear industry. These materials are especially attrac- using a welding technique. Thermal radiation produced from the
tive because of their weldability and excellent resistance to many waste itself is predicted to maintain temperatures in the 150 to
forms of corrosion. Although stainless steels are highly immune to 250°C range for hundreds of years. The physical, mechanical, and
general corrosion, they are susceptible to intergranular attack compositional integrity of the meta) barrier should remain intact for
(IGA) under the proper conditions of microstructure and environ- the projectel containment life of 1,000 years.
ment. This form of localized corrosion, often occurring in or near a
In this study, the thermal conditions of a single-pass weid
weid joint, is particularly insidious because it can develop over a
HAZ have been simulated, followed by long-term isothermal heat-
period of several years.
ing at low temperatures. Four 316 SS alloys with systematic varia-
IGA occurs as a result of local breakdown of the passive film tions in carbon and nitrogen were examined. The purpose of this
along grain boundaries due to prolonged heating in the tempera- work was to determine whether LTS occurred in 316 SS as a re-
ture range 550 to 800'C. 1-3 This temperature excursion produces sult of the above thermal treatments. The simulation of weid resid-
precipitation of Cr-rich M 23 C 6 carbides along grain boundaries with ual stresses was intentionally omitted. Two corrosion tests were
the concurrent impoverishment of Cr in the adjacent matrix, as first used to detect IGA as related to chromium depletion that may
suggested by Bain and co-workers ° and Strauss and his have occurred during the thermal treatments. AEM was employed
co-workers. 5 Hall and Briant 6 have recently shown that Cr deple- to determine the local composition of the grain boundaries.
tion can also occur along boundaries between the carbides. A ma-
terial is said to be sensitized when the intergranular chromium
level drops below —12 wt% and has been shown by numerous
investigators' - " to be susceptible to IGA when exposed to oxidiz-
ing conditions. Materials and Experimental Technique
Sensitization is influenced by many factors, most of which are The materials examined in this study had a base composition
controllable in common industrial practice. Among the factors af- of type 316 SS with systematic variations in the levels of carbon
fecting sensitization are; time and temperature, 31418-24 composi- and nitrogen. Two 32-kg (70-Ib) heats were vacuum/argon induc-
tion (especially carbon and nitrogen) 2.14.24-3° cold work, 431-39 and tion melted from high-purity materials. The nitrogen content of the
grain boundary structure. 4° heats was 0.04 (alloy Nl) and 0.16 wt% (alloy N3). The split-heat
technique was used to produce two carbon levels of 0.02 (alloy
Cl) and 0.07 wt% (alloy C3) within each nitrogen heat. The alloy
Low-Temperature Sensitization compositions appear in Table 1. The material was supplied in the
Researchers 1641-48 have shown that if non-sensitizing car- form of 0.2-cm-thick plate that had been solution annealed at
bides are first precipitated on grain boundaries by a brief high- 1038°C for 30 minutes and water quenched. The final grain size
temperature exposure, sensitization can develop during subse- was approximately 50 to 60 µm. The ratio of the wt% of Cr/Ni was
quent thermal treatments at unexpectedly low temperatures balanced to yield a single-phase, austenitic microstructure. For this
(- 400°C) This behavior has been termed low-temperature sensiti- reason, the amount of nickel (art austenite stabilizer) is relatively
zation (LTS). The motivation for investigating LTS resulted from high compared to the AISI 12 f compositional specification and the
weid HAZ cracking in boiling water reactor (BWR) water transfer chromium level (a ferrite stabilizer) is relatively low.
lines. In most cases, the degree of sensitization (DOS) in the HAZ
that occurs upon welding is not extensive. In other words, the LTS heat treatments were performed in two stages; (1) HAZ
width of chromium-depleted area adjacent to the boundaries is simulation and (2) low-temperature, isothermal aging. These will
insufficient to produce detectable cracking. However, after years of henceforth be referred to as HAZ and LTS heat treatments, re-
service at operating temperatures of 288°C, the HAZ has been spectively. The purpose of the HAZ thermal cycle was to simulate
shown to become increasingly susceptible to IGA. The increase in the microstructure in the weid HAZ at a distance of 2 to 5 mm
DOS has not been associated with a wide or deep chromium de- from the fusion zone. This region has been shown to be the most
48
pletion profile. 43-°s In other words, the minimum chromium content susceptible to sensitization in welded BWR recirculating pipes.
at the grain boundary was greater than 12 wt%. The temperatures and cooling rates for the HAZ thermal cycle are
Carbide nucleation does not occur in this low-temperature given in Table 2. The HAZ simulation was performed using a
regime because of the rapid decrease in nucleation rate with de- Gleeblet 3) which, through resistance heating and water cooling, is
creasing temperature. Povich 41 and Povich and Rao 42 first capable of inducing rapid thermal gradients in materials. The ther-
explained that thermodynamically stable carbides that precipitate mal profile was monitored by a control thermocouple that was
at high temperatures, i.e., in the normai sensitization range, do not spot-welded to the center of the sample, as shown in Figure 1.
dissolve at lower temperatures. Rather, they continue to grow and The other five thermocouples were used only in initial studies to
establish a new interfacial chromium level and chromium depletion map the thermal profile across the specimen. The specimen thick-
profile. ness and distance between the jaws of the Gleeble were adjusted
Several LTS studies include estimations of the activation en- to obtain homogenous heating across the span section (S) of the
ergy for LTS. 41 • 42446,47 The results are inconclusive as to sample. The dashed circle in the center of the specimen repre-
whether LTS is controlled by diffusion of chromium through the sents the corrosion test sample that was machined out of the
austenite matrix or along grain boundaries. Gleeble blank after the 2-stage heat treatment. The specimen
grips were placed in a floating fixture that allowed free motion dur-
ing thermal expansion and contraction of the sample.
Background for the Present Investigation
In this study, type 316 SS has been evaluated for resistance
to LTS under specific environmental conditions. These conditions
(2) American Iron and Steel Institute (AISI), Washington, DC.
and their engineering application wilt be briefly described in the
following paragraphs. (3 ^Gleeble is a trade name of buffers Scientific Inc., Poestenkill, rn.

CORROSION—Vol. 46, No. 10 813

SCIENCE

TABLE 1 The LTS heat treatments were performed after the HAZ sim-
Chemical Composition (wt%) ulation according to the schedule shown in Table 3. Each of the
of Experimental Alloys four alloys was heat treated in triplicate at the time/temperature
combinations indicated in Table 3. In addition, alloys Cl N1 and
Element Cl N1 Cl N3 C3N1 C3N3 C3N1 were heat treated for 24 and 54 hours at 650C without
prior HAZ simulation in order to compare the effects of LTS heat
treatments with sensitization in the normai temperature range. The
C 0.029 0.021 0.082 0.081
heat treatments were performed in air.
N 0.049 0.171 0.45 0.176
Corrosion test specimens 2 cm in diameter and 0.32 cm thick
Mn 1.60 1.55 1.60 1.57
Si 0.40 0.43 0.41 0.44 were electro-discharge machined (EDM) from the center of the
Cr 16.6 16.6 16.7 16.5 Gleeble blanks. Two corrosion methods were employed to mea-
Ni 12.4 12.5 12.4 12.4 sure the DOS: (1) the DL-EPR test s° and (2) the ferric sulfate
Mo 2.7 2.7 2.7 2.7 weight-loss test or Streicher test. 51
P 0.003 0.003 0.003 0.003 The test parameters for the DL-EPR test were identical to
S 0.006 0.005 0.005 0.005 those given in Reference 50. The DOS was measured as the ratio
Fe bal bal bal bal of the maximum current density of the anodic scan (i e ) to the max-
imum current density of the reactivation scan (i r ).
Specimens were repolished after the DL-EPR test and re-
used for the Streicher test. For each alloy, triplicate specimens
TABLE 2
representing each heat treatment were placed in common flasks.
HAZ Thermal Simulation Using the Gleeble
The DOS was reported as corrosion rate in mpy as calculated
from the weight loss during the 120-hour test. The specimen sur-
1 Heat from 25° to 900°C in 10 s (88'C/s) faces were examined using optical microscopy.
2 Hold at 900°C for 5 s Thin foils were made from the corrosion specimens using the
3 Cool from 900° to 690°C at a rate of 7°C/s
twin-jet electropolishing technique. Analytical electron microscopy
4 Cool from 690° to 490°C at a rate of 5°C/s
5 Free cool from 490°C to 25°C was performed with a JEOL 41 2000X (200 kV) equipped with a
Tracor Northern' S) energy-dispersive spectrometer (EDS) for com-
positional microanalysis. Characteristic x-ray intensities were con-
verted to compositional weight percent using the Cliff-Lorimer 52
t=4.6 mm method.
(0.18")
RESULTS

Corrosion Tests
The DL-EPR test produced identical results for all of the al-
loys, regardless of the LTS heat treatment. A representative EPR
Thermocouple curve is shown in Figure 2. The stable passive behavior during the
positions reactivation scan indicates that the grain boundaries were not de-
pleted in chromium enough to cause local breakdown of the pas-
sive layer.
X ` 2.5 cm In order to verify the ability of the DL-EPR test to detect chro-
mium depletion in these materials, samples of alloys Cl Nl and
C3N1 were each tested after aging for 24 and 54 hours in the nor-
S=6.4 cm - ►{ mal sensitization temperature range without prior HAZ simulation
(2.5") (as indicated in Table 3). Figure 3 shows the DL-EPR curve for
alloy Cl Nl after aging 54 hours at 650°C. The heat treatment pro-
Gleeble jaws duced a profusion of carbide precipitation on the grain boundaries
and considerable chromium depletion as evidenced by the high
FIGURE 1. Schematic drawing of Gleeble specimen used for HAZ
current density of the reactivation scan in Figure 3. The i r /i a values
simulation. The dashed circ/e represents the corrosion specimen.
for the 650°C heat treatments scaled directly with the carbon con-
(X) marks the locations of the thermocouples that were used to map
tent and time at temperature as shown in Table 4.
the thermal profile of the span width.
The Streicher tests were evaluated by both the weight loss
and the surface appearance. The weight-toss results were plotted
TABLE 3 as the corrosion rate as a function of LTS heat-treatment time.
LTS Heat Treatments Figure 4 shows an example of the results for the four experimental
alloys heat treated at 350°C. The corrosion rate did not behave as
Temperature Time a monotonic function of time, but rather it exhibited various max-
ima and a minimum at 500 hours. The corrosion rates at the re-
'C ('F) (hours)
maining LTS temperatures of 300° and 400°C showed similar be-
havior. Alloy Cl N1 deviated from this trend in Figure 4; however,
300 ( 570) 50 500 1,000 2,300 the trend is obeyed at 300° and 400°C. Variations in the amount of
350 ( 660) 50 500 1,000 2,000 carbon and nitrogen showed no distinguishable effect on the rela-
400 ( 750) 50 500 1,000 2,800 tive corrosion rate of the alloys.
500 ( 930) 24
650 (1200) 1 "
(4 jJEOL USA Inc., Peabody, MA.
(A) C1N1 and C3N1 only. No prior HAZ simulation. (5) Tracor Northern Inc., Middieton, WI.

814 CORROSION -October 1990

 SCIENCE

0.2 CIN1
T=350C
-- --- C1N3 ----- C3N1 C3N3

50
0.1
lll
U j. 40

't 0 E
>
30
c
-0.1 0
0
0
t

20
<— As
C 0

N -0.2
0
0
10
1o1 102 10' 101
-0.3
Time of LTS Heet Treatment (hrs)
1 Oe-5 10e-4 10e-3 1 Oe-2 10e-1
FIGURE 4. Streicher test results. Corrosion rate as a function of time
Current Density, i (Alcm2) at LTS temperature of 350 ° C.
FIGURE 2. Representative DL-EPR curve for all LTS heat treat-
ments.

0.2 The weight-loss data was also plotted for the three LTS tem-
peratures at constant composition, as shown in Figure 5. Here, the
data for alloy Cl N3 obey the trend of a minimum corrosion rate at
500 hours, regardless of temperature. The corrosion rate in-
0.1 creased as the LTS temperature increased at 50 hours and 500
lll hours although this correlation was not observed for the remaining
0 three alloys. The data representing the as-received condition and
the 500°C/24-hour LTS heat treatment are also shown. It is impor-
N tant to note the relative scale of the corrosion rates. While distinc-
tive trends are evident, the absolute values of the corrosion rates
are quite low with respect to the material that was sensitized in the
-0.1 normal sensitization range. For example, the corrosion rates of
alloys Cl Nl and C3N1, aged for 24 and 54 hours at 650°C, were
s 245 and 3,160 mpy, respectively, two to three orders of magnitude
c greater than the data reported in Figures 4 and 5. Therefore, the
0
41 -0.2 corrosion rates produced by the LTS heat treatments are quite
0 mild relative to those in the normal sensitization range.
0

Microstructures
-0.3 The HAZ simulation produced carbide nucleation in the high-
carbon alloys, C3N1 and C3N3, while the low-carbon alloys were
1 Oe-5 1 Oe-4 10e-3 1 Oe-2 10e-1 essentially carbide-free. Figure 6 shows an example of the car-
Current Density, i (A/cm2) bides observed in the high-carbon alloys. Alloy C3N1 clearly con-
tained more carbides than alloy C3N3; however, it was difficult to
FIGURE 3. DL-EPR results of alloy C1N1 after 54 hours at 650°C. quantify the average carbide population because of compositional
banding of carbon that occurred during initial processing of the
alloys. The majority of the carbides appeared as discrete precipi-
tates of less than 0.5 µm in size, although semicontinuous net-
TABLE 4 works were observed along some grain boundaries in alloy C3N1
DL-EPR Resuits as seen in Figure 7. Here, the majority of the boundaries contain
for Sensitized 316 SS no carbides whereas semicontinuous carbide strings appear on
others. In order to estimate the relative amount of carbide precipi-
tation in the two alloys, several metallographic samples were sur-
Condition Cl N1 C3N1 veyed (at least 10 samples for each alloy). Carbide population was
reported as the percentage of grain boundaries containing at least
one carbide. In alloy C3N1, approximately 25% of the boundaries
650°C/24 h 0.061 0.292 contained carbides while 15% of the boundaries in alloy C3N3
650°C/54 h 0.266 0.331 contained carbides. The population of carbides in alloy C3N3 was
highly variable among the samples examined.

CORROSION —Vol. 46, No. 10 815

SCIENCE

AIIoy C1N3
7000 ----- 360C 400C

50

n 40
E

30
c SOOC
0
0
0
t
0
V 20
Aa Rec'd

10
101 10' 103 10'

Time of LTS Heat Treatment (hrs)

FIGURE 5. Streicher test results. Corrosion rate as a function of time FIGURE 7. Alloy C3N1 after HAZ simulation showing the variation in
for alloy C1N3 at 300, 350, and 400°C. the number of graln boundaries containing carbide precipitation.

In regions that contained carbides after LTS heat treatments,

the population and size remained essentially the same as before
the LTS heat treatments. However, alloy C3N1 also contained
very smal) carbides that were not obvious before the LTS heat
treatments. The small carbides were only observed in groupings
with larger carbides. The appearance of the small carbides was
inconsistent with respect to the duration and temperature of LTS
heat treatment. That is, within the same alloy, they appeared after
50 hours at a given LTS temperature but were not observed after
2,800 hours at the same temperature or vice versa.

The low-carbon alloys remained free of grain boundary car-

bides after the LTS heat treatments.

Surface Appearance After the Streicher Test

Figure 8 shows the surface of alloy C3N1 in the as-received
condition after the Streicher test. The surface had been polished
with 120 grit silicon carbide paper before the test. Some of the
grain boundaries are slightly etched; however, the majority are
unattacked. This was also the case for the remaining three alloys
FIGURE 6. Microstructure of alloy C3N1 after HAZ simulation in the as-received condition.
showing carbides <0.5 µm.

The surfaces of the four experimental alloys after HAZ and

LTS heat treatment of >2000 hours at 400°C are shown in Figure
9. In all of the alloys, the boundaries are more severely attacked
In order to ascertain that carbides did not exist prior to the than the as-received condition (Figure 8). However, there is also a
HAZ simulation, samples of C3N1 in the as-received condition range of the degree of attack within each alloy. In order to quantify
were re-solution annealed at 1038°C for 2 hours and thermally cy- the relative extent of boundary attack among the four alloys, the
cled on the Gleeble. Metallographic examination revealed that the widest and the narrowest boundaries were measured on several
HAZ simulation produced the same size and population density of samples. It can be seen from Table 5 that the widest boundaries
carbides as previously observed. were observed in the high-carbon, low-nitrogen alloy, C3N1,
The LTS heat treatments that followed the HAZ simulation whereas the least amount of attack occurred in the low-carbon,
did not produce further growth of existing carbides or significantly high-nitrogen alloy, Cl N3. A width of zero was assigned to bound-
change in the population density compared to the HAZ-only heat aries that were lightly etched as in the as-received condition. Ta-
treatment. Although measurement using metallographic samples is ble 5 also shows the results of specimens that received the LTS
admittedly imprecise, the carbide population density was too heat treatment without prior HAZ simulation. The surfaces of these
sparse for statistically significant measurement using TEM foils, samples resembied the solution-annealed condition, regardless of
which would have provided more accurate size measurements. the alloy or the LTS temperature.

816 CORROSION— October 1990

 SCIENCE

FIGURE 8. Surface of atloy C3N1 in the as-received condition after

the Streicher test.

FIGURE 9 continues on the next page.

CORROSION —Vol. 46, No. 10 817

SCIENCE

In contrast, the concentration profile perpendicular to a

boundary in alloy Cl Nl after the HAZ simulation is shown in Fig-
ure 12. Although carbides were not observed in the thin foil, the
boundary is slightly depleted in chromium compared to the matrix
on either side. In fact, the chromium concentration is slightly en-
riched at a distance of --100 nm on both sides of the boundary
and the actual boundary composition is that of the bulk. Further,
compositional fluctuations along the boundary were more
pronounced after the HAZ simulation than in the as-received con-
dition. However, the fluctuations were within ± 1.5 wt% of the ma-
trix value for chromium and ± 1 wt% for molybdenum.
Figures 13 and 14 show the concentration profiles along and
perpendicular to grain boundaries in alloy C3N1 after the HAZ
simulation. In this case, the boundaries contained carbides of size
—0.2 to 0.5 µm. The composition of the carbides is approximately
26 to 30 wt% chromium, 4 to 5 wt% molybdenum, 8 to 10 wt%
nickel, and 55 wt% iron. There is no depletion of chromium or
molybdenum along the boundary. The levels of chromium and mo-
lybdenum remain above their respective matrix concentrations in
the intercarbide region, which is a few hundred nm in length.
The composition profile perpendicular to a different boundary
(Figure 14) that contains a carbide has chromium minima of 16.7
wt% on either side of the boundary. The maximum chromium level
of the carbide is —18 wt%. The molybdenum concentration profile
is roughly a mirror image of the chromium profile. Thai is, the
FIGURE 9. Surface of the four experimental alloys after Streicher chromium maxima coincide with minima in the molybdenum profile
test. HAZ + LTSlh2,000 hours at 400°C: (a) alloy C1N1, (b) alloy and vice versa. This behavior was also observed in alloy Cl N1
CIN3, (c) a ll o y C3N1, (d) alloy C3N3. after heat treatment, both along and normal to boundaries. Asym-
metrical molybdenum maxima appear on both sides of the bound-
ary and the level drops to 1.3 wt% in the carbide.
TABLE 5
The composition profile shown in Figure 15 is normal to a
Width of Grain Boundary Attack
boundary in alloy C3N1 after HAZ and LTS heat treatment. Car-
in Streicher Test After HAZ Simulation and
bides were not observed on the boundary. There is a distinctive
LTS Heat Treatment of 2,000 hours at 400°C
drop in the chromium concentration at the boundary compared to
the matrix. However, the minimum of -15.9 wt% chromium is sub-
AIIoy Grain Boundary Width (µm) stantially greater than 12 wt% necessary to produce IGA in corro-
sion tests such as the EPR and Streicher tests. Again, the mini-
C1N1 0 to 10 mum chromium level is coincident with the maximum in
ClN3 O to 8 molybdenum concentration.
C3N1 0 to 25
C3N3 0 to 15
C3N1 Oto4
DISCUSSION
°'2,O00 hours/400°C without prior HAZ simulation.

Carbide Nucleation and Growth

Compositional Microanalysis A theoretical model describing the thermodynamic and kinetic
of the Grain Boundaries behavior of the experimental alloys in this study has been previ-
The AEM investigation was limited to alloys Cl N1 and C3N1. ously reported. 53 The temperature—time-to-sensitization (TTS)
The two alloys were each examined in three conditions; (1) as- curves produced from the model are shown in Figure 16 along
received, (2) after HAZ thermal simulation, and (3) after HAZ ther- with the locus of points representing the HAZ thermal simulation.
mal simulation and LTS heat treatment of >2000 hours at 400°C. The model is based on solving the Stawstrom—Hillert equation;'
Compositional analyses were obtained both along and perpendicu- therefore, the time-to-sensitization depends on diffusion of chro-
lar to boundaries. A minimum of two foils for each heat treatment mium in the austenite lattice. The family of curves in Figure 16
was examined. Composition profiles were obtained for four or live show that the propensity for sensitization increases as the carbon
boundaries within each foil. content increases and the nitrogen level decreases. Also, the tem-
The levels of chromium, molybdenum, phosphorus, and sili- perature at the "nose" of the TTS curve increases with increasing
con were fairly consistent in the as-received condition, showing susceptibility to sensitization. The effects of carbon and nitrogen
little deviation from their respective matrix compositions. Figures are incorporated in the model by modifying the activity coefficients
10 and 11 show compositional profiles along and perpendicular to of chromium with temperature-dependent constants for carbon and
a boundary, respectively for alloy C3N1 in the as-received condi- nitrogen.
tion. The compositional fluctuation along the boundary (Figure 10) Figure 16 shows that the four experimental alloys should be
is within --1 wi% for chromium and molybdenum. Figure 11 illus- immune to sensitization as defined by the Stawstrom—Hillert equa-
trates that the boundaries are compositionally similar to the auste- tion; the TTS curve indicates the time and temperature at which
nite matrix. That is, there is neither appreciable enrichment nor the chromium-depleted zone (wt% Cr < 12) extends 20 nm on
depletion of alloying elements at the grain boundaries. The com- either side of the grain boundary. The model does not take into
position profiles were similar for alloy Cl Nl. In both materials account the size of the growing carbide but rather the equilibrium
(Cl N1 and C3N1), the element that showed the most inhomoge- chromium content at the austenite/carbide interface, which is a
neity in the as-received condition was molybdenum. function of temperature and time.

818CORROSION — October 1990

 SCIENCE

—A— Si —•— Cr —v— Mo —•— P —A— Si —•— Cr —v— Mo —m— P

20 4 40 7

6
19 35
30. a
5
t- 0 ó
0 S S
18 0 30 4 co
2

17 25

1 3 :1
16 20

1s 0 15 0
0 100 200 300 400 0 100 200 300 400 500 600 700 800
Distante (nm) Distance (nm)

FIGURE 10. Concentration profile along graln boundary of alloy FIGURE 13. Concentration profile along boundary with carbides in
C3N1 in as-received condition. alloy C3N1 after HAZ simulation.

— A — Si —•— Cr —v— Mo —m— p — A — Si —•— Cr — v — Mo —•— P

20 4 20 4
boundary
carbide
19 1 19
30. 30.
0 0
0 S V S
18 18
y 0
2 2
á
17 17
m Ó
1 3 1 3
16 16

is . ! •i•^u - E • U • n • .0 15 0
0 100 200 300 400 500 0 100 200 300 400 500 600
Distance (nm) Distance (nm)

FIGURE 11. Concentration profile perpendicular to grain boundary FIGURE 14. Concentration profile perpendicular to boundary in
in alloy C1N1 in as-received condition. alloy C3N1 with carbide after HAZ simulation.

—A— Si —•— Cr —v— Mo —m— P — A — Si —•— Cr —V— Mo —•— P

20 4 20 4

19 19
30. 30.
ó 0
U S 2
18 0 18
N v)
2 2
t
»
17 17
m S
1 1 3
16 16

15 • • arm--n - n-- '0 1511 w. O —

0 100 200 300 400 500 600 0 100 200 300 400 500 600
Distance (nm) Distance (nm)
FIGURE 12. Concentration profile perpendicular to grain boundary FIGURE 15. Concentration profile perpendicular to boundary in
ae
in alloy C1N1 ft r HAZ simulation. alloy C3N1 after HAZ simulation and LTS heat treatm e nt.

CORROSION—Vol. 46, No. 10 819

SCIENCE

1000 rIi rmj Grain boundaries serve as rapid transport paths because of
their relatively open structure compared to the lattice. Grain
boundary dislocations further enhance migration of chromium st-
900 oms via pipe diffusion. In order to estimate the relative amount of
chromium diffusion along the boundaries and through the matrix
during the HAZ simulation, the characteristic diffusion distances,
r
;> 800 (Dt) 72 , were calculated. The choice of the diffusion coefficient of
^^ •-.-•-.-.-•- chromium through the lattice and along the boundaries dramati-
cally affects the resulting diffusion distances. An average of two
í`o
700 values obtained from the literature 55.5e was used. The calculation
yielded a total chromium diffusion distance during the 75-second
HAZ heat treatment of 6.8 µm along the boundaries and 0.03 µm
y 600 in the matrix. This estimation illustrates that the 0.5 µm-sized car-
Gleeble ------- C1 N1 bides observed after the HAZ simulation correspond to growth
heat ......••••• C1N3 from grain boundary diffusion of chromium. Similarly, the value of
500 (Dt)' 12 after 2,000 hours at 400C preclude lattice diffusion as the
treatment ----- C3N1
peak T=900 ° C C3N3 major contributor to low-temperature carbide growth, (D L t)" 2 =
0.003 µm and (D e t) 2 = 6.6 µm.
400
The total diffusion distance along grain boundaries during the
10 -1100 101 102 103
heat treatments is relatively small with respect to the average
Time (min) grain size. However, the sparse carbide population allows constant
replenishment of chromium to the regions immediately surrounding
FIGURE 16. TTS curve produced from theoretical model for the particles.
experimental alloys. The grain boundary diffusion distance can be used to esti-
mate the carbide size and spacing based on the conservation of
The alloy compositions and thermal treatments were mass. A simple geometrical model was employed 57 using the dif-
designed specifically to yield the resuits that appear in Figure 16. fusion distance during the HAZ simulation as one-half of the inter-
That is, the peak temperature and cooling curve for the HAZ ther- carbide spacing. The purpose of the calculation was to estimate
mal cycle fall to the left of the TTS curve. This represents the first whether the grain boundaries contained enough chromium to form
stage of the classical LTS heat treatment as first defined by Cr 23 C 6 carbides of approximately 0.5 µm in size while leaving at
Povich 41 and shown by others 42-ae,5e to produce a sensitized least 12 wt% chromium in solution to prevent IGA. The calculation
microstructure in austenitic stainless steels. It is noted in these yielded a carbide diameter of 0.55 µm with an intercarbide spac-
studies that sensitization was detected in type 304 (UNS S30400) ing of 12 µm. Thus it was estimated that the carbides observed
SS while the low-carbon grades of 304 SS and 316 SS (regard- after the HAZ simulation could have formed entirely Erom chro-
less of the carbon content) were generally immune. Ljungberg 47 mium supplied from the boundaries while the boundaries them-
found that significant LTS occurred only in materials containing selves remained immune to IGA. It should be mentioned that this
continuous carbide stringers on the grain boundaries. In other model slightly overestimates the carbide size because the compo-
words, only materials that were nearly sensitized by the nucleation sition is assumed to be Cr 23 C 6 . The AEM results revealed that the
heat treatment could be low-temperature sensitized. This observa- M portion of the M 23 C 6 is actually about 30 wt% chromium. The
tion was also reported by Nakagawa, 58 net effect on the carbide size is virtually negligible since the parti-
The TTS curves for the experimental alloys in Figure 16 con- cle radius is proportional to 1/n 3 where n is the density of chro-
verge after longer times at low-temperatures. Mozhi and mium atoms per carbide.
co-workers 53 used Fullman's LTS model 48 to illustrate that sensiti-
zation behavior after several hundreds of years is governed by the
diffusion of chromium through the austenite matrix (D L ) and along Direct Grain Boundary Observations
grain boundaries (D B ) whereas compositional variations are impor- The composition profiles obtained using AEM indicate that
tant only in the high-temperature/short-time regime. the HAZ simulation produced an obvious local fluctuation in the
The Stawstrom—Hillert model is in agreement with much of levels of chromium and molybdenum at the grain boundary. Al-
the experimental work on 304 and 316 stainless steels. 14.25,26,37,54 though the effects were not dramatic, the compositional variations
The agreement is greatest in the case of long-term isothermal an- could be related to the presente of carbides in the high-carbon
neals in the temperature range 600 to 850°C. These heat treat- alloy.
ments typically result in substantial carbide precipitation accompa- Including the effects of electron beam spreading, the ana-
nied by well-developed chromium depletion profiles. The range of lyzed volume is approximately 50 nm in diameter through the
carbide precipitation, as depicted by the TTS curve, is greater than thickness of the foil. Therefore, even in the limiting case of a per-
the range of intergranular attack associated with chromium deple- fectly aligned grain boundary, the spot size is larger than the vol-
tion. That is, carbide precipitation occurs before the boundaries ume of sensitized material as estimated in the previous section.
become significantly depleted in chromium. At temperatures above Further, any misalignment of the grain boundary with respect to
the nose of the curve, depletion is relatively rapid but chromium is the electron beam and x-ray detector would result in additional
easily replenished from the interior of the grafins and the bound- toss of resolution. The infrequency of carbide precipitation left few
aries (healing). Therefore, it is possible that carbides form without choices of grain boundaries to be analyzed and therefore optimum
detectable chromium depletion. spatial resolution was often unobtainable using the double-tilt
In the case of LTS, the time spent in the nucleation stage is stage. Consequently, even though the beam appeared to be cen-
insufficient for chromium to diffuse through the matrix to the grain tered on the boundary, significant contributions from outside the
boundaries. Since the extent of diffusion depends on the time at activation volume could have interfered with the signal. The result
temperature, the earliest stages of nucleation and growth arise of such a misalignment is the detection of a greater level of chro-
from the most available source of chromium, namely the grain mium than actually occurs at the boundary.
boundaries. Thus, carbide precipitation is unrelated to chromium Several of the concentration profiles displayed chromium min-
depletion in the matrix. Instead, the majority of the chromium is ima coincident with molybdenum maxima and vice versa. This
supplied to the growing carbide through the grain boundaries, trend occurred in profiles both along and also perpendicular to the
boundary dislocations and nearby lattice dislocations. grain boundaries. The calculated diffusion distances for molybde-

820 CORROSION—October 1990

 SC I ENC E

num in the austenitic matrix during the HAZ simulation and LTS various maxima and minima was observed in large-diameter pipe
heat treatment are 0.050 µm and 1 µm, respectively. These val- weid HAZs after LTS heat treatments. 44 The corresponding AEM
ues indicate that if the carbide contains a significant amount of results showed that chromium depletion did not occur.
molybdenum, its corresponding depletion layer could be wider The nonlinear relationship between time at LTS temperature
than the chromium depletion layer. However, the molybdenum and IGA is not understood. A possible explanation may be related
concentration gradient, which drives diffusion, is not as steep as to site competition between phosphorus, sulfur, and nitrogen st-
that for chromium. Further, not all of the carbides analyzed con- oms on the grain boundaries. The activity of these impurity ele-
tained molybdenum. ments could be influenced by the alloying elements silicon and
Compositional profiles obtained for the alloys heat treated in molybdenum. Conversely, phosphorus and sulfur may affect the
the normal sensitization range (Table 3) showed extensive chro- activities of silicon and molybdenum. The Jatter case could pro-
mium depletion with minima of - 10 wt°° chromium. Therefore it duce local susceptibility to IGA if silicon and molybdenum were
can be concluded that if significant depletion had resulted from the incorporated info the passive film.
HAZ simulation, it would have been detectable using the existing
AEM capabilities.
SUMMARY
Grain Boundary Attack in Corrosion Tests
The EPR results corroborated the AEM data, which indicate The HAZ thermal simulation produced small (0.1 to 0.5 µm)
that chromium depletion did not occur from the HAZ simulation. If isolated carbides on grain boundaries in the high-carbon alloys.
a very narrow chromium depletion layer had formed, it was unde- Carbides were observed on approximately 25% of the boundaries
in the high-carbon (0.08 wt°%), low-nitrogen 0.04 wt%) material
tectable using both methods. Recent findings 59 show that EPR
and on -15°° of the boundaries in the high-carbon, high-nitrogen
tests can detect narrow depletion profes if they are deep; that is,
(0.18 wt°°) alloys. The low-carbon (0.02 wt%) alloys were free of
when the minimum chromium level is less than -12 wt°°. In other
words, the EPR test detects a volume of chromium depletion at carbides.
the boundary. It can be concluded from the EPR test results in this
study that if a chromium depletion layer existed, it was neither The LTS heat treatment of 2,000 hours at 400°C did not, in
wide nor deep. This was demonstrated by direct measurement general. change the size and population of the carbides. In some
using AEM. samples, very small carbides (<0.1 µm) were observed in group-
Examination of the specimen surfaces after the Streicher ings with those produced from the HAZ simulation. These may
tests showed that the grain boundaries were moderately suscepti- have been too small to defect after the HAZ simulation, but grew
ble to IGA after the HAZ simulation and LTS heat treatments. This during the LTS heat treatment.
was in striking contrast to the specimens that received heat treat-
ments in the normal sensitization range where attack was severe The DL-EPR test results demonstrated that grain boundary
enough to cause entire grains to dislodge from the surface. The chromium depletion did not occur regardless of the composition or
HAZ and LTS heat treatments produced an increase in surface LTS heat treatment. The DL-EPR test detected extensive chro-
attack compared to the as-received condition. The maximum mium depletion in samples that were sensitized in the normal sen-
boundary width scaled directly with the amount of carbon and ni- sitization range (24 hours/650°C).
trogen: the widest boundaries corresponded to the high-carbon,
low-nitrogen alloy and the narrowest boundaries corresponded to
the low-carbon, high-nitrogen alloy. The maximum boundary width The HAZ simulation produced a substantial increase in grain
can certainly not be used as a single measure of the average re- boundary attack over the as-received (solution-annealed) condition
sistance to IGA as a function of carbon and nitrogen level. This is in the Streicher test. This was evident by visual inspection; how-
especially true since the weight-loss results were inconclusive ever, the weight-loss data do not reflect a change in corrosion be-
among the alloys. Although most of the boundaries in the high- havior from the as-received condition. The ensuing LTS heat treat-
carbon alloys (75 to 85%) were free of carbides, they were ments did not change the extent of local attack. Specimens that
attacked nevertheless. Moreover, the low-carbon alloys that con- received the LTS heat treatments without prior HAZ simulation
tained no carbides after heat treatment were more susceptible to were unchanged from the as-received condition. Thus it is con-
IGA than in the as-received condition. The susceptibility of the cluded that the high-temperature excursion during the HAZ simula-
grain boundaries to IGA in the Streicher test is unrelated to chro- tion was necessary to activate the boundaries. The extent of at-
mium depletion as explained in the preceding sections. tack was greater for the high-carbon alloys than for the low-carbon
Impurity segregation to grain boundaries has been shown to alloys. The addition of nitrogen increased the resistance to IGA for
increase the susceptibility to IGA in the absence of chromium both levels of carbon.
depletion."• 25 26 The levels of phosphorus and sulfur in these al-
.

loys were 0.003 wt% and 0.005 wt%, respectively. While these The weight-loss data from the Streicher test revealed that type
levels are quite low compared to commercial grade materials, they 316 SS is highly resistant to IGA. The effects of carbon and nitro-
cannot be excluded as grain boundary segregants. The analytical gen were not apparent from these data. The most striking result
methods used in this investigation were not sensitive enough to was the relatively high resistance to IGA after —500 hours at LTS
detect the presence of small amounts of sulfur and phosphorus temperature, regardless of the temperature or alloy composition.
accurately, especially on the grain boundaries. Although the AEM This may be due to the segregation of site-competitive impurity
analysis included sulfur, the number of counts per spectrum was elements to the grain boundaries, although this was not shown in
insufficient for statistical significance. Auger electron spectroscopy the present work.
(AES) would be the most effective method of detecting grain
boundary phosphorus or sulfur in these alloys.
► Estimation of the average chromium diffusion distance through
The trends that appear in the weight-loss data from the Strei- the austenite Iattice and along grain boundaries shows that it is
cher test have not been explained in this paper. With few excep- possible for carbides to form during the HAZ simulation. High diffu-
tions, the minimum corrosion rate occurred at 500 hours. At longer sivity paths such as grain boundaries and dislocation pipes can
LTS times, the corrosion rate increased, but never appreciably supply enough chromium to produce carbides of the size and den-
above that of the as-received condition. The same general trend of sity observed in this study.

CORROSION —Vol. 46, No. 10 821

SCIENCE

AEM results show mild chromium depletion at some grajn 23. S M. Bruemmer, CORROSION/89. paper no. 561 (Houston, TX: NACE, 1989).
boundaries, however, the minimum chromium content was never 24. H . Betrabet. Ph.D. dissertation (Columbus, OH: The Ohio State University,
1986).
less than -16 wt%. There is evidente that molybdenum and chro- 25. R.A. Mulford. E.L. Hall. and CL. Briant. Corrosion 39(1983): p: 132.
mium diffuse in opposite directions when the former is not tied up 26. C.L. Briant. R.A. Mulford. and E.L. Hall, Corrosion 38(1982): p. 468.
in a carbide. 27. T.A. Mozhi, H.S. Betrabet. V. Jagannathan, B.E. Wilde, and W.A.T. Clark, Scripta
Metall. 20(1986): p. 723.
28. J.J. Eckenrod and C.W. Kovack. ASTM-STP 679. ed. C.R. Brinkman and H.W.
ACKNOWLEDGMENTS Garvin (Philadephia. PA: ASTM. 1979). p. 17.
29. R.E. Smith, 'Basic Studies on the Variabilities of Fabrication Related Sensitiza-
tion Phenomena in Stainless Steel." EPRI Report No. RP 1072-1 SRD-79-017
The authors wish to thank the project sponsor, Lawrence- (Palo Atto. CA: Electric Power Research Institute. 1979).
30. C.L. Briant, °Effects of Nitrogen and Cold Work on the Sensitization of Austenitic
Livermore National Laboratory, and technical liaison, Dr. R.D. Mc- Stainless Steels. - G.E. Report No. NP-2457 (Schenectady, NY: General Electric
Cright. Company. 1982).
31. P.K. De, G.C. Palit, and K. Elayaperumal. Trans. Indian Inst. Metals 26(1973): p.
56.
REFERENCES 32. A. Randak and F.W. Trautes, Werks. und Korros. 21(1970): p. 97.
33. D.A. Vermilyea. Corrosion 29(1973): p. 442.
34. A. Honkasalo, Scand. J. Metall. 2(1973): p. 156.
1. C. Stawstrom and M. Hillert, J. Iron Steel Inst. 207(1969): p. 77.
35. C. Hahin. R.M. Stass, B.H. Nelson. and P.J. Reucroft, Corrosion 32(1976): p.
2. C.S. Tedmon, O.A. Vermilyea. and J.N. Rosolowski, J. Electrochem. Soc.
229.
118(1969): p. 162.
3. V. Cihal. Intergranular Corrosion of Cr-Ni Stainless Steel. presented at Unieux 36. C.L. Briant and A.M. Ritter. Scripta Met. 13(1979): p. 177.
Conf., on May 5. 1969. 37. C.L. Briant and A.M. Ritter, Metall, Trans. 11(1980): p. 2009.
4. E.C. Bain, R.H. Aborn, and J.J 0. Rutherford. Trans. Amer. Steel Treating Soc. 38. C.L. Briant. Corrosion 38(1982): p. 596.
21(1933): p. 481. 39. E.P. Butler and M.G. Burke, Acta Metall. 34(1986): p. 557.
5. B. Strauss, H. Schottky, and J. Hinnuber. Zeitsch. fur Anorgan. und Alleg. Che- 40. B.W. Bennett and H.W. Pickering, Acta Meiall, 36(1988): p. 539.
mie 188(1930): p. 309. 41. M.J. Povich. Corrosion 34(1978): p. 60.
6. E.L. Hall and C.L. Briant, Metall. Trans. 15A (May 1984): p 793. 42. M.J. Povich and P. Ran, Corrosion 34(1978): p. 260.
7. R.L. Cow and C.S. Tedmon. "Intergranular Corrosion in Fe-Ni-Cr Alloys." Ad- 43. W.G. Morris and A.M. Ritter, Proc. Seminar on Countermeasures for Pipe Crack-
vances in Corrosion Science. vol. 3. ed. M.G. Fontana and R.W. Staehle (New ing in BWRs, paper no. 25 (Febo Alto, CA: EPRI, 1980).
York, NY: Plenum Press. 1973), 44. C.G. Schmidt, R.D. Caliguiri, L.E. Eiselstein, S.S. Wing, and D. Cubiciotti, Metall,
8. V. Cihal et al., "Tests d'etude et d'evaluation de la sensibilite des aciers inoxyd- Trans. 18A (Aug. 1987): p. 1483.
ables a la corrosion intergranulaire." Communication au 15eme Congres Euro- 45. P.L. Andresen. H.D. Solomon, and D.F. Taylor, "Basic Studies on the Variability
peenne de la Corrosion. Par is, Sept., 1973 (Rueil-Malmaison, France: L'Impri- of Fabrication-Related Sensitization Phenomena in Stainless Steels," EPRI Final
merie Hiblot. 1973). Report NP-1823 (Palo Alto. CA: EPRI. May 1981).
9. W.L. Clark, R.L. Cowan, and W.L. Walker, "Comparative Methods for Measuring 46. S. Hatton, T. Yokosuka. Y. Urayama, and Y. Mori, International LTS Workshop,
Degree of Sensitization in Stainless Steel." Intergranular Corrosion in Stainless paper no. 6 (Palo Alto. CA: EPRI, Jan. 1982).
Alloys, ASTM STP 656, ed. R.F. Steigerweld (Philadelphia. PA: ASTM, 1978). 47. L. Ljungberg, International LTS Workshop, paper no. 5 (Palo Alto, CA: EPRI, Jan.
pp. 99-132. 1982).
10. C.L. Briant, Corrosion 36(1980): p. 497. 48. R.L. Fullman, Proc. Seminar on Countermeasures for Pipe Cracking in BWRs,
11. CL. Briant and P.L. Andresen, Meiall. Trans. 19A(March 1988): p. 495. paper no. 26 (Pa(o Alto, CA: EPRI. 1980).
12. CL. Briant, Metall. Trans. 18A (April 1987): p. 691. 49. Scientific tnvestigation Plan for Nevada Nuclear Waste Storage tnvestigation
13. A.P. Majidi and M.A. Streicher. Corrosion 4 0(1984): p. 393. Project. Metal Barrier Select os and Testing. Rev. O, June 15, 1987.
14. T.A. Mozhi. W.A.T. Clark, K. Nishimoto. W.B. Johnson, and D.D. Macdonald, 50. A.P. Majidi and M.A. Streicher, Corrosion 40(1984): p. 584.
Corrosion 41(1985): p. 555. 51. ASTM A262-77a, "Standard Recommended Practices for Detecting Susceptibility
15. S.M. Bruemmer. L.A. Charlot, and D.G. Atteridge, "Compositional Effects of the to lntergranular Attack in Stainless Steels" (Philadelphia, PA' ASTM, 1977).
Sensitization of Austenitic Stainless Steels," NUREG/CR-3918 (Washington, DC: 52. G. Cliff and G. Lorimer, Proc. sth European Congress on Electron Microscopy
U.S. Government Printing Office. December 1984). (Bristol and London. England: Institute of Physics, 1972), p. 141.
16. C.D. Lundin, C.H. Lee. R. Menon. and E.E. Stansbury, Weld. Res. Council Bulle- 53. T.A. Mozhi, M.C. Juhas, and B.E. Wilde, Scripta Met. 21(1987): p. 1547.
tin No. 319 (New York NY: Welding Research Council. Nov. 1986). 54. H.S. Betrabet, K. Nishimoto, B.E. Wilde, W.A.T. Clark, Corrosion 43(1987): p. 77.
17. C.L. Briant and E.L. Hall, Corrosion 43(1987): p. 525. 55. R.A. Perkins, R.A. Padgeft, and N.K. Tunali, Metall. Trans. 4(1973): p. 2535.
18. R.L. Fullman, Acta Metall. 30(1982): p. 1407. 56. A.F. Smith, Metal Sci. 9(1975): p. 375.
19. F. Kohier, Monatsh. Chem. 91(1960): p. 738. 57. M.C. Juhas. Ph.D. dissertation (Columbus, OH: The Ohio State University, 1989).
20. G.S. Was and R.M. Kruger, Acte Met. 33(1985): p. 841. 58. Y.G. Nakagawa, "LTS Study," Metallurgy Dept., Research Institute, Ishikawajima-
21. R.A. Carolan and R.G. Faulkner. Acts Met. 36(1988): p. 257. Harima Heavy Industries, Ltd., Report No, HTGE-045-I-011, 1978-1979.
22. B.W. Bennett and H.W. Pickering, Acta Met. 36(1988): p. 539. 59. S.M. Bruemmer, L.A. Charlot, and B.W. Arey, Corrosion 44(1988): p. 328.

822 CORROSION -October 1990