Tensile and Fatigue Properties of 17-4 PH Stainless Steel at

High Temperatures
JUI-HUNG WU and CHIH-KUANG LIN

The tensile and high-cycle fatigue properties for 17-4 PH* stainless steels in three different conditions
were investigated at temperatures ranging from room temperature to 400 ⬚C. Results indicated that
the yield strength and fatigue strength for the three conditions at a given temperature took the following
order: condition H900 ⬎ condition A ⬎ condition H1150. The yield strength of each condition
decreased with increasing temperature except for condition A, which was tested at 400 ⬚C with longer
hold times, where a precipitation-hardening effect took place. The S–N curves showed that the fatigue
strengths of each condition in the short-life regime were decreased with an increase in temperature.
In the long-life regime, the fatigue strengths of condition A at 400 ⬚C were greater than those at
lower temperatures as a result of an in-situ precipitation-hardening effect. The fatigue strengths of
condition H900 in the long life regime at 300 ⬚C were superior to those at lower temperatures, due to
the mechanisms of surface oxidation and thermal activation of dislocations. Fractography observations
indicated that a shift of fatigue fracture from surface to internal crack initiation occurred at higher
temperatures (300 ⬚C and 400 ⬚C) with long fatigue lives.

I. INTRODUCTION was on the effect of embrittlement, caused by long-term

PRECIPITATION-HARDENING stainless steels aging at 300 ⬚C to 400 ⬚C, on the mechanical properties. As
have been widely used as structural components for various some applications of 17-4 PH stainless steels are subjected to
long-term cyclic loading at high temperatures, it is important
applications, such as nuclear, chemical, aircraft, and naval
to characterize the high-temperature fatigue behavior and
industries, due to their excellent mechanical properties, good
associated microstructural variation of this alloy in order to
fabrication characteristics, and excellent corrosion resist-
better predict the fatigue lifetime for components made from
ance. Of the former, 17-4 PH* stainless steel is currently
this material.
*17-4 PH is a trademark of the Armco Steel Corporation, Pittsburgh, PA. The objective of this study is, macroscopically and micro-
scopically, to characterize the effect of environmental tem-
one of the most commonly used alloys in this type.[1] The perature on the tensile and fatigue properties of variously
alloy 17-4 PH is a typical martensitic precipitation-hardening heat-treated 17-4 PH stainless steels by systematic experi-
stainless steel and is most often supplied from the mill in ments. Tensile tests with different hold times at temperature
the solution-annealed condition (condition A), in which the and high-cycle fatigue tests were conducted at room tempera-
microstructure is comprised of low-carbon equiaxed mar- ture, 200 ⬚C, 300 ⬚C, and 400 ⬚C for 17-4 PH steels in
tensite and up to several volume percent of ␦ -ferrite string- the solution-annealed (condition A), peak-aged (condition
ers.[1,2,3] By applying suitable heat treatments in the H900), and overaged (condition H1150) conditions, in order
temperature range of 482 ⬚C to 621 ⬚C (900 ⬚F to 1150 ⬚F), to study the variation of tensile and fatigue strength with
a wide range of mechanical properties can be obtained for microstructure.
this type of stainless steel.[1,3–7] The maximum strength and
hardness values are obtained after aging at 450 ⬚C to 510 ⬚C,
by which the precipitation of coherent copper-rich clusters II. EXPERIMENTAL PROCEDURES
occurs.[1,3,4] Aging at higher temperatures (above 540 ⬚C)
The commercially available 17-4 PH stainless steels used
would result in the precipitation of incoherent fcc copper-
in the current study were supplied by the vendor in the
rich precipitates, lower strength and hardness, and an
form of hot-rolled, solution-annealed bars. The chemical
enhancement in toughness.[1,3,4]
Most of the previous investigations on 17-4 PH were composition of this alloy is given in Table I. Three types of
heat treatments were applied to the specimens, i.e., the as-
concentrated on the analyses of the microstructural, mechan-
received condition A, peak-aged condition H900, and over-
ical, and fatigue properties at room temperature;[2,3,8–13] little
aged condition H1150. Details of the heat-treatment proce-
work has been on the high-temperature mechanical and
dures are given in Table II. The mechanical properties at
fatigue properties of this alloy. Although a few studies on
the high-temperature characteristics of the 17-4 PH alloy room temperature for each condition are listed in Table III.
have been reported,[8,14] the emphasis in those investigations Scanning electron microscopy (SEM) micrographs of the
original microstructure in each condition are shown in Figure
1. The matrix structure for each condition mainly consisted
of equiaxial martensite laths within the prior-austenite
JUI-HUNG WU, Graduate Research Assistant, and CHIH-KUANG LIN, grains, and the grain size was about 20 to 30 ␮m.
Professor, are with the Department of Mechanical Engineering, National
Central University, Chung-Li 32054, Taiwan R.O.C. Contact e-mail:
Both tensile and high-cycle fatigue tests were carried out
t330014@[Link] on a commercial closed-loop servohydraulic test machine
Manuscript submitted August 7, 2001. at room temperature, 200 ⬚C, 300 ⬚C, and 400 ⬚C in ambient

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JUNE 2002—1715

air. The axial smooth-surface specimens used in both tests ⬚C) for 17-4 PH steel in three different conditions. It can be
had a uniform cylindrical gage section of 6 mm in diameter seen in Figure 2 that the yield strength of the three conditions
and 18 mm in length. A commercial compact two-zone, at a given temperature generally takes the following
SiC-heated furnace was used to heat the specimens in both
tensile and fatigue tests. Tensile tests were conducted under
stroke control, with a stroke rate of 0.5 mm/min. A commer-
cial direct-contact, high-temperature extensometer with a 12
mm gage length was used to measure the tensile strains. In
high-temperature tensile tests, specimens from each condi-
tion were heated to the given temperatures and held at tem-
perature for different periods of time (0.25, 1, 4, 16, and 32
hours) before the start of each test. The high-cycle axial
fatigue tests were performed under a sinusoidal-loading
waveform with a load ratio of R ⫽ 0.1 and a frequency of
f ⫽ 20 Hz until failure or 2 ⫻ 106 cycles, at which point
the specimen was considered to be a runout. Each fatigue
specimen was held in the furnace at a given temperature for
15 minutes to reach thermal equilibrium prior to the start
of fatigue loading.
The SEM was used for characterization of the fracture- (a)
surface morphology and high-temperature oxidation phe-
nomenon. Microstructures of the tensile and fatigue speci-
mens were analyzed with transmission electron microscopy
(TEM) to observe the precipitates and substructure of dislo-
cations. The thin foils for TEM analysis were prepared by
cutting out 3-mm-diameter disks from some of the tensile
and fatigue specimens, followed by standard grinding and
twin-jet electropolishing procedures. In addition, differential
scanning calorimetry (DSC) was applied to detect whether
secondary age hardening would occur in the condition H900
and H1150 specimens when tested at the given high tem-
peratures.

III. RESULTS AND DISCUSSION

A. Effects of Temperature and Hold Time at Temperature (b)
on Tensile Strength
Figure 2 shows plots of the yield strength as a function
of hold time at three temperatures (200 ⬚C, 300 ⬚C, and 400

Table I. Nominal Chemical Composition of 17-4 PH

Stainless Steels Tested (Weight Percent)
Cr Ni Cu Mn Si Nb ⫹ Ti Mo S C P Fe
15.18 4.47 3.47 0.65 0.38 0.2 0.15 0.03 0.02 0.016 bal

Table II. Heat-Treatment Conditions of 17-4 PH Stainless

Steels Tested
Condition Treatment
Condition A 1038 ⬚C ⫻ 0.5 h → air cool (c)
H900 Condition A → 482 ⬚C ⫻ 1 h → air cool
H1150 Condition A → 621 ⬚C ⫻ 4 h → air cool Fig. 1—SEM micrographs of the original microstructure of 17-4 PH stain-
less steel in three conditions: (a) condition A, (b) H900, and (c) H1150.

Table III. Room-Temperature Mechanical Properties of 17-4 PH Stainless Steel in Different Conditions
Ultimate Tensile Yield Strength Elastic Modulus Elongation Hardness V-Notch Impact
Condition Strength (MPa) (MPa) (GPa) (in 25 mm) (Pct) (HRc) Toughness (J)
Condition A 1018 992 199 13.4 32 67
H900 1414 1387 223 12.5 44.5 21
H1150 966 880 196 18.4 31.5 75

1716—VOLUME 33A, JUNE 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 with the hold time to a value equivalent to that of condition
H900, as a result of a pronounced precipitation-hardening
effect. This explains why the yield strengths of condition
A with longer hold times at 400 ⬚C were superior to the
corresponding ones at lower temperatures. However, except
for condition A tested at 400 ⬚C, the yield strength of each
condition at a given temperature was independent of the
hold time and almost stayed at a constant value. This is due
to the lack of significant change in the microstructure at a
given high temperature when the hold time was increased
from 0.25 to 32 hours. Previous studies[9,10] have shown that
condition A needed a certain long period of time to complete
the precipitation process when age treated at a temperature
below 400 ⬚C. Therefore, the invariance of the yield strength
with the hold time for condition A tested at 200 ⬚C and 300
(a) ⬚C can be attributed to the incompletion of the precipitation
process during the given hold times. For the two aged condi-
tions, neither condition H900 nor H1150 showed any exo-
thermic peak when analyzed with DSC from room
temperature to 600 ⬚C. This implied that secondary precipita-
tion did not occur for the two aged conditions. Note that
conditions H900 and H1150 have been initially age treated
at 482 ⬚C and 621 ⬚C, respectively, prior to high-temperature
tensile testing. Hence, the yield strengths of conditions H900
and H1150 at each given high temperature remained nearly
unchanged as the hold time changed from 0.25 to 32 hours.
Microstructural analyses of the failed tensile specimens pro-
vided further support for the previous arguments.
The TEM micrographs of the original microstructure in
each condition prior to any tensile testing are shown in
Figure 3. Figure 3(a) shows the typical nearly parallel mar-
tensite laths with a very high density of dislocations in
(b)
condition A. For condition H900, it can be seen in Figure
3(b) that the dislocation density was reduced and regular
dislocation arrays within the martensite lath could be
detected. This suggested that the recovery of dislocations
occurred in condition H900 during the initial age-hardening
treatment at 482 ⬚C. In addition, the fine, uniformly distrib-
uted copper-rich precipitates were not detected by TEM in
condition H900, as the precipitates were very small and
coherent with the martensitic matrix structure. Previous
investigations[2,9,10] have also shown that the copper-rich
phase formed at the initial stage of the precipitation process
was a very fine structure and coherent with the matrix, such
that it was difficult to be detected by TEM. However, as
shown in Figure 3(c), the matrix of condition H1150 con-
sisted of large rod-like copper-rich precipitates and a very
low dislocation density. Zang[10] similarly indicated that for
(c) the 17-4 PH alloy aged at 650 ⬚C, the matrix would com-
pletely recrystallize and copper-rich phases became coarse,
Fig. 2—Variation of yield strength with hold time for 17-4 PH stainless rod-like shapes which were incoherent with the matrix.
steels tested at different temperatures: (a) 200 ⬚C, (b) 300 ⬚C, and (c) 400 ⬚C.
The TEM analysis indicated that at a given high tempera-
ture, the microstructures in conditions H900 and H1150 were
sequence: condition H900 ⬎ condition A ⬎ condition nearly unchanged as the hold time was increased from 0.25
H1150. It can also be inferred from Figure 2 that the yield to 32 hours. This might explain why the yield strength in
strength of each condition was reduced with increasing tem- conditions H900 or H1150 at a given high temperature was
perature except for condition A, tested at 400 ⬚C with longer almost constant regardless of the length of hold time. For
hold times (⬎16 hours), where strengths were greater than condition A tested at 400 ⬚C with a 32 hour hold time, the
those obtained at lower temperatures. The variation of ulti- TEM micrograph shown in Figure 4 exhibited a microstruc-
mate tensile strength with testing temperature for each condi- tural morphology similar to that of the original condition
tion exhibited a similar trend to that of yield strength. The H900 (Figure 3(b)), in which the fine, uniformly distributed
yield strength of condition A tested at 400 ⬚C was increased precipitates could not be detected by TEM. However, the

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JUNE 2002—1717

 (a) (b)

(c)
Fig. 3—TEM micrographs of the original microstructure of 17-4 PH stainless steel in three conditions: (a) condition A, (b) H900, and (c) H1150.

reason why condition A, tested at 400 ⬚C with longer hold Apparently, at fatigue lives of ⬍2 ⫻ 105 cycles, an increase
times, had a tensile strength equivalent to that of condition in temperature generated a similar effect on the fatigue
H900 was believed to be caused by the formation in the strength as it did on the tensile strength, such that the fatigue
matrix of coherent precipitates, which were too small to be resistance in this regime was related to the yield strength.
detected by TEM (Figure 4), like those in condition H900 In other words, the fatigue strengths at fatigue lives of ⬍2 ⫻
(Figure 3(b)). 105 cycles for each condition were decreased with increasing
temperature due to the reduction of yield strength. At fatigue
lives of ⬎2 ⫻ 105 cycles, the differences of fatigue strength
B. Influence of Environmental Temperature on the in various temperatures for each condition were decreased,
Fatigue Behavior and the fatigue strengths of condition A at 400 ⬚C and
Figure 5 shows the S–N curves of each condition tested condition H900 at 300 ⬚C were greater than the correspond-
at various temperatures. It can be seen in Figure 5 that the ing ones at lower temperatures. For condition A tested at
fatigue strengths of a given condition were decreased with 400 ⬚C, the greater fatigue strengths at fatigue lives of ⬎2
increasing temperature at fatigue lives of ⬍2 ⫻ 105 cycles. ⫻ 105 cycles, in comparison with other lower temperatures,

1718—VOLUME 33A, JUNE 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 (a)

Fig. 4—TEM micrograph of the microstructure in condition A of 17-4 PH

stainless steel tested at 400 ⬚C with a 32-h hold time.

can be attributed to an in-situ precipitation-hardening effect.

As described in the previous section, the tensile strength of
condition A tested at 400 ⬚C with a longer hold time at
temperature was increased due to a precipitation-hardening
effect. This precipitation-hardening effect may also enhance
the fatigue resistance of condition A tested at 400 ⬚C to
extend the fatigue lifetime over that produced at lower tem-
peratures. However, it was considered that, at fatigue lives
of ⬎2 ⫻ 105 cycles, the slight differences in fatigue strength (b)
for each condition tested at various temperatures were related
to the variation of fractography with the environmental tem-
perature. A similar relationship may also be applied to
explain the greater fatigue strengths observed for condition
H900 specimens tested at 300 ⬚C over the other temperatures
at fatigue lives of ⬎2 ⫻ 105 cycles. These fatigue strengths
can be supported by the analyses of fractography, surface
oxidation, and microstructure, as described subsequently.
The representative fracture surfaces of condition H900
fatigue specimens tested at room temperature and 200 ⬚C
in all life regimes and at 300 ⬚C and 400 ⬚C in the short-
life regime are shown in Figure 6. It can be seen in Figure
6 that in each of the four specimens, crack initiation occurred
at the surface followed by a flat river-pattern crack-growth
region. A similar morphology of the fracture surface was
also observed for the condition A and H1150 fatigue speci-
mens tested at lower temperatures (room temperature and (c)
200 ⬚C) in all life regimes and at higher temperatures (300 Fig. 5—S–N curves at various temperatures for 17-4 PH stainless steel in
⬚C and 400 ⬚C) in the short-life regime, where fatigue cracks three conditions: (a) condition A, (b) H900, and (c) H1150. (Arrows desig-
were initiated at the surface. On the contrary, Figure 7 shows nate runout tests.)
the fracture surfaces of condition H900 fatigue specimens
tested at 300 ⬚C and 400 ⬚C with fatigue lives ⬎2 ⫻ 105
cycles. In both specimens, crack initiation occurred within surface crack initiation. Therefore, the small differences in
the specimen body. In fact, all the fatigue specimens of the fatigue strength among the given temperatures at fatigue
three conditions tested at 300 ⬚C and 400 ⬚C in the long- lives of ⬎2 ⫻ 105 cycles for each condition might be associ-
life regime showed internal crack initiation. In other words, ated with this type of internal crack-initiation behavior at
the fatigue specimens of each condition tested at higher high temperatures. Similarly, the change from surface crack
temperatures (300 ⬚C and 400 ⬚C) in the long-life regime initiation to internal crack initiation might be responsible
showed internal crack initiation, while the others showed for the greater fatigue strengths at fatigue lives of ⬎2 ⫻

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JUNE 2002—1719

 (a) (b)

(c) (d )
Fig. 6—SEM fractographs of condition H900 failed at (a) room temperature, Sa ⫽ 518 MPa, and Nf ⫽ 284,180; (b) 200 ⬚C, Sa ⫽ 540 MPa, and Nf ⫽
111,267; (c) 300 ⬚C, Sa ⫽ 540 MPa, and Nf ⫽ 23,014; and (d ) 400 ⬚C, Sa ⫽ 501 MPa, and Nf ⫽ 70,342 (i: crack initiation site).

105 cycles in condition H900 specimens tested at 300 ⬚C, along the loading direction. Figure 9 showed an oxide layer
as compared with those tested at lower temperatures. An of about 1 ␮m in thickness and some protrusions on the
earlier study by Isomoto and Stoloff[15] also indicated that oxide-layer surface. These protrusions were believed to be
the fatigue strengths in the long-life regime for a powder the side view of the oxide products shown in Figure 8.
metallurgy Astroloy tested at 650 ⬚C ambient air were greater However, no such surface oxidation products were observed
than those at room temperature due to an internal crack- for specimens tested at a temperature equal to or below 200
initiation mechanism. This change in crack-initiation site ⬚C. Botella et al.[16] observed that a 17Cr-2Ni austenitic
was related to the pronounced surface oxidation observed stainless steel heated at 700 ⬚C developed a homogeneous
for high-temperature specimens, as described subsequently. Cr-rich oxide layer accompanied by many Fe-rich nodules,
Figures 8(a) and (b) show the surface oxidation products on which Cr- and Mn-rich octahedral crystals were formed.
of condition A heated and held at 300 ⬚C for 8 hours and Apparently, similar surface oxidation processes and products
at 400 ⬚C for 4 hours, respectively. It can be seen in Figure may also occur in the given 17Cr-4Ni alloy at 300 ⬚C and
8(a) that the surface of the condition A specimen held at 400 ⬚C in the current work, as evidenced by Figures 8 and
300 ⬚C consisted of a homogeneous oxide layer, on which 9. Previous studies[15,17,18] have shown that internal fatigue-
sporadic oxide nodules were also observed. On the surface crack initiation in high-temperature ambient air was attrib-
of the condition A specimen held at 400 ⬚C, in addition to uted to a reduction in the surface-slip intensity, as a thin
the formation of a uniform oxide layer and more oxide oxide layer on slip steps would lead to greater surface work
nodules, acicular crystals were also observed to have formed hardening and prevent surface fatigue-crack initiation. More-
on top of the nodules, as shown in Figure 8(b). Similar over, in the long-life regime, the oxide layer would not be
surface-oxide layers were also directly observed on the severely damaged by a smaller deformation due to a lower
fatigue specimens of each condition tested at 300 ⬚C and fatigue load, such that it could effectively enhance the resist-
400 ⬚C, as evidenced in Figure 9, where an SEM micrograph ance to surface crack initiation.[15] A comparison of the
of a condition H900 fatigue specimen tested at 400 ⬚C was cyclic stress level with the yield strength may also support
presented as an example. Note that the SEM micrograph in this argument. Table IV listed the ratios of the transition
Figure 9 was an axially sectioned view of the gage section stress (defined as the maximum cyclic stress corresponding
in the fractured specimen, which was longitudinally cut to the transition fatigue life of 2 ⫻ 105 cycles in Figure 4)

1720—VOLUME 33A, JUNE 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 (a) (a)

(b)
(b)
Fig. 7—SEM fractographs of condition H900 failed at (a) 300 ⬚C, Sa ⫽
518 MPa, and Nf ⫽ 1,376,047; and (b) 400 ⬚C, Sa ⫽ 484 MPa, and Nf ⫽ Fig. 8—SEM micrographs of surface oxidation products for condition A
1,443,109 (i: crack initiation site). at (a) 300 ⬚C ⫻ 8 h and (b) 400 ⬚C ⫻ 4 h.

over the yield strength for the given three conditions tested
at 300 ⬚C and 400 ⬚C. It can be seen in Table IV that the
ratios are very close to 1. This implied that at a fatigue life
of ⬎2 ⫻ 105 cycles, the cyclic stress level was lower than
the yield strength and, thus, the oxide layer was barely
damaged by the smaller deformation. Apparently, in the
current study, it was considered that at high temperatures
and fatigue lives of ⬎2 ⫻ 105 cycles, the protective oxide
layer strengthened the surface resistance to fatigue-crack
initiation, such that crack initiation occurred inside the speci-
men and then increased the fatigue strength to a level close
to or greater than the low-temperature one.
The TEM observations on the substructure of dislocations
for condition A fatigue specimens tested at 300 ⬚C and 400
⬚C are shown in Figures 10(a) and (b), respectively. It can Fig. 9—SEM micrograph of the axially sectioned view in a condition H900
be seen in Figure 10 that the microstructures at both tempera- fatigue specimen tested at 400 ⬚C with a fatigue life of 1,443,109 cycles.
tures consist of nearly parallel lath martensites with disloca-
tion arrays which can be clearly resolved, unlike those in
the original condition A, as shown in Figure 3(a). The forma- generated at the tip of stress concentrations such as micro-
tion of regular arrays of dislocations is characteristic of a pores or defects. Therefore, the formation of intense slip
recovered structure. This inferred that thermally activated bands was retarded, the crack-initiation resistance was
recovery processes, such as dislocation climb and/or cross enhanced, and stage I crack initiation and propagation were
slip, occurred at 300 ⬚C and 400 ⬚C in the present study, suppressed.[15,17,18] Consequently, in the present work, the
resulting in a reduction in dislocation density and the forma- recovery by thermal activation of dislocations was consid-
tion of regular arrays of dislocations. Previous studies[15,17,18] ered to be partially responsible for the small differences in
indicated that thermal activation allowed dislocations to fatigue strength among all given temperatures in the long-
cross slip and climb out of the planar bands which were life regime for each condition, as well as for the greater

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JUNE 2002—1721

 Table IV. Comparison of Transition Stress and Yield
Strength of 17-4 PH Stainless Steel at High Temperatures
Transition Yield
Stress* Strength** Transition Stress/
Condition (MPa) (MPa) Yield Strength
Condition A, 300 ⬚C 924 895 1.03
H900, 300 ⬚C 1184 1175 1.01
H900, 400 ⬚C 1100 1060 1.04
H1150, 300 ⬚C 845 825 1.02
H1150, 400 ⬚C 807 750 1.08
*Transition stress: the maximum cyclic stress corresponding to
the fatigue life of 2 ⫻ 105 cycles for each condition in Fig. 4.
**The yield strength with 0.25-h hold time at temperature.

fatigue strengths of the H900 condition at 300 ⬚C, over those

at lower temperatures in the long-life regime. It has also
been recognized by Isomoto and Stoloff[15] that in the long-
life regime, the fatigue strengths of a powder metallurgy
Astroloy tested at 650 ⬚C ambient air were greater than (a)
the corresponding ones at room temperature, due to the
formation of a thin surface-oxide film and the thermal activa-
tion of dislocations.

C. Effect of Heat Treatment on the Fatigue Behavior

Comparisons of the S–N curves of variously heat-treated
17-4 PH steels at different temperatures are shown in Figure
11. Figure 11 indicated that at a given temperature, the peak-
aged H900 condition exhibited the highest fatigue strength
among the given three conditions, while the overaged H1150
condition showed the lowest one. In general, fatigue failure
can be divided into two steps: crack initiation and crack
propagation.[19] For steels, the fatigue life of smooth surfaces
was usually predominated by the crack initiation stage,
which took up most of the lifetime, in particular at low stress
levels.[19] It is generally found that the fatigue strength of
steel is increased with tensile strength.[19] For an alloy with
a higher yield strength, dislocations in the matrix are more
difficult to move and, thus, the resistance to crack initiation
is enhanced and the fatigue life will be increased.[4,15] As
shown in Figure 11, the fatigue strength in the given three
conditions appeared to be increased with tensile strength (b)
for a given testing temperature. In other words, at a given Fig. 10—TEM micrographs of the microstructure in condition A tested at
temperature, both the tensile strength and fatigue strength (a) 300 ⬚C, Sa ⫽ 411 MPa, and runout; and (b) 400 ⬚C, Sa ⫽ 482 MPa,
for the given three conditions mainly take the following and Nf ⫽ 923,841.
order: condition H900 ⬎ condition A ⬎ condition H1150.
This indicated that the initial, coherent precipitates in the
matrix of the peak-aged condition H900 provided more
resistance to crack initiation in 17-4 PH steel as compared not be increased to the same level as the condition H900
with the unaged condition A and overaged condition H1150. specimen.
Although the tensile strength of condition A tested at 400 In addition, Figure 11 indicated that the S–N curves for
⬚C with a longer hold time at temperature was equivalent each condition revealed less-steep slopes at 300 ⬚C and 400
to that of condition H900, the precipitation-hardening effect ⬚C. It can also been seen that the high-temperature data
under cyclic loading at 400 ⬚C did not increase the fatigue points (300 ⬚C and 400 ⬚C) are mostly located at two distinct
strength of condition A to the same level as the condition regions, where the fatigue life is either less than about 105
H900 specimen. Apparently, before the in-situ precipitation- cycles or longer than 106 cycles. This was likely related to
hardening effect gained enough time to become completely the effectiveness of the two aforementioned mechanisms:
effective at 400 ⬚C for condition A, the cyclic loading had surface oxidation and thermal activation of dislocations. An
already generated some damage on the condition A speci- earlier study by Isomoto and Stoloff[15] indicated that at a
mens, such that the fatigue strength of condition A could high stress level, the brittle oxide film on the specimen

1722—VOLUME 33A, JUNE 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A

 (a) (b)

(d )
(c)
Fig. 11—S–N curves for variously heat-treated 17-4 PH stainless steels tested at different temperatures: (a) room temperature, (b) 200 ⬚C, (c) 300 ⬚C, and
(d ) 400 ⬚C (arrows designate runout tests).

surface was easier to fracture as a result of a greater deforma- for each condition at higher temperatures exhibited less-
tion, and a notch was then introduced. The crack was readily steep slopes and, thus the data points were readily located
initiated from this notch and propagated more rapidly due either in the very-short-life regime or the very-long-life
to the more severe attack by oxygen.[15] Moreover, Leverant regime.
and Gell[17] indicated that the planar bands became unstable
at high temperatures due to the thermal activation of disloca- IV. CONCLUSIONS
tions. When the rate of dislocation recovery out of the slip
band exceeded that of crack development along the slip band, 1. For variously heat-treated 17-4 PH stainless steels tested
stage I crack initiation and propagation were suppressed.[17] at a given temperature, the yield and fatigue strength
Hence, in the present study, it was possible that at high generally took the following sequence: condition H900
temperatures and greater fatigue loads, the rate of formation ⬎ condition A ⬎ condition H1150.
of a crack along the slip band was greater than that of 2. With the exception of condition A tested at 400 ⬚C, the
dislocation recovery out of the slip band. Therefore, the yield strength of each condition at a given temperature
fatigue lives of specimens tested at high temperatures and was independent of the hold time at temperature. The
high stress amplitudes could not be extended by the mecha- yield strength of condition A tested at 400 ⬚C was
nisms of surface oxidation and thermal activation of disloca- increased with the hold time to a value equivalent to that
tions. Once the stress amplitude was reduced to a certain of condition H900, due to a precipitation-hardening
level, where the protective surface-oxide film became stable effect.
and the dislocation recovery rate outranged the crack forma- 3. The yield strength of each condition was generally
tion rate, the fatigue life could be significantly increased to decreased with an increase in testing temperature, except
a great extent. Consequently, at lower stress levels, both for condition A tested at 400 ⬚C with longer hold times.
mechanisms of surface oxidation and thermal activation of 4. In the short-life regime, the fatigue strengths of each
dislocations effectively extended the fatigue lives of speci- condition were decreased with an increase in temperature
mens at high temperatures. In this regard, the S–N curves due to the reduction of yield strength.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 33A, JUNE 2002—1723

5. In the long-life regime, the fatigue strengths of condition 2. U.K. Viswanathan, S. Banerjee, and R. Krishnan: Mater. Sci. Eng.,
A tested a 400 ⬚C were greater than those at lower temper- 1988, vol. A104, pp. 181-89.
3. H.J. Rack and D. Kalish: Metall. Trans., 1974, vol. 5, pp. 1595-1605.
atures because of an in-situ precipitation-hardening 4. C.-K. Lin and W.-J. Tsai: Fatigue Fract. Eng. Mater. Struct., 2000,
effect. However, the small differences in fatigue strength vol. 23, pp. 489-97.
among the given temperatures in the long-life regime for 5. F.B. Pickering: Int. Met. Rev., 1976, vol. 21, pp. 227-68.
each condition, as well as the greater fatigue strengths 6. R.D.K. Misra, G.Y. Prasad, T.V. Balasubramanian, and P.R. Rao:
of condition H900 in the long-life regime at 300 ⬚C Scripta Metall., 1986, vol. 20, pp. 713-16.
7. R.D.K. Misra, G.Y. Prasad, T.V. Balasubramanian, and P.R. Rao:
over those at lower temperatures, can be attributed to the Scripta Metall., 1987, vol. 21, pp. 1067-70.
surface oxidation and thermal activation of dislocations. 8. K.C. Antony: J. Met., 1963, vol. 15, pp. 922-27.
6. For the given three conditions of 17-4 PH alloy, fatigue 9. U.K. Viswanathan, P.K.K. Nayar, and R. Krishnan: Mater. Sci. Tech-
cracks were initiated at the surface in all life regimes at nol., 1989, vol. 5, pp. 346-49.
lower temperatures (room temperature and 200 ⬚C) and 10. X. Zang: Trans. Met. Heat Treatment, 1991, vol. 12, pp. 23-29 (in
Chinese).
in short life regime at higher temperatures (300 ⬚C and 11. G. Huang: Iron and Steel, 1998, vol. 33, pp. 44-46 (in Chinese).
400 ⬚C). On the contrary, in long life regime at higher 12. F.J. Heymann, V.P. Swaminathan, and J.W. Cunningham: Report No.
temperatures (300 ⬚C and 400 ⬚C), crack initiation CS-1967, Electrical Power Research Institute, Palo Alto, CA, 1981.
occurred inside the specimen. 13. B.C. Syrett, R. Viswanathan, S.S. Wing, and J.E. Wittig: Corrosion,
1982, vol. 38, pp. 273-82.
14. M. Murayama, Y. Katayama, and K. Hono: Metall. Mater. Trans. A,
ACKNOWLEDGMENTS 1999, vol. 30A, pp. 345-53.
This work was funded by the National Science Council 15. T. Isomoto and N.S. Stoloff : Mater. Sci. Eng., 1990, vol. A124, pp.
171-81.
of the Republic of China (Taiwan) under Contract No. NSC- 16. J. Botella, C. Merino, and E. Otero: Oxid. Met., 1998, vol. 49, pp.
89-2216-E-008-018 297-324.
17. G.R. Leverant and M. Gell: Metall. Trans. A, 1975, vol. 6A, pp. 367-71.
REFERENCES 18. D.J. Duquette and M. Gell: Metall. Trans., 1972, vol. 3, pp. 1899-1905.
19. J.A. Bannantine, J.J. Comer, and J.L. Handrock: Fundamentals of
1. W.F. Smith: Structure and Properties of Engineering Alloys, 2nd ed., Metal Fatigue Analysis, Prentice-Hall, Inc., Englewood Cliffs, NJ,
McGraw-Hill, Inc., New York, NY, 1993. 1990.

1724—VOLUME 33A, JUNE 2002 METALLURGICAL AND MATERIALS TRANSACTIONS A