- Email: [email protected]
308L austenitic stainless steel fusion welds
Corrosion Science 77 (2013) 210–221
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Effect of low temperature on hydrogen-assisted crack propagation in 304L/308L austenitic stainless steel fusion welds H.F. Jackson ⇑, C. San Marchi, D.K. Balch, B.P. Somerday Sandia National Laboratories, Livermore, CA 94550, USA
a r t i c l e
i n f o
Article history: Received 5 February 2013 Accepted 5 August 2013 Available online 11 August 2013 Keywords: A. Stainless steel B. SEM C. Hydrogen embrittlement
a b s t r a c t Effects of low temperature on hydrogen-assisted cracking in 304L/308L austenitic stainless steel welds were investigated using elastic–plastic fracture mechanics methods. Thermally precharged hydrogen (140 wppm) decreased fracture toughness and altered fracture mechanisms at 293 and 223 K relative to hydrogen-free welds. At 293 K, hydrogen increased planar deformation in austenite, and microcracking of d-ferrite governed crack paths. At 223 K, low temperature enabled hydrogen to exacerbate localized deformation, and microvoid formation, at austenite deformation band intersections near phase boundaries, dominated damage initiation; microcracking of ferrite did not contribute to crack growth. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Austenitic stainless steels are used extensively in the nuclear power, chemical processing, and oil and gas industries, in applications where they are susceptible to stress corrosion cracking (SCC) [1–3]. SCC begins with an electrochemical process, such as local anodic dissolution, however synergistic processes of anodic dissolution and cathodic hydrogen damage are thought to contribute to SCC in stainless steels [4,5]. Hydrogen damage is typically characterized by a significant loss of tensile ductility. In various alloy/ solution systems, interactions between dislocations at a crack tip and corrosion products, such as hydrogen, can play a key role in SCC. Detailed fractographic and microstructural studies support the premise that adsorbed or absorbed hydrogen promotes crack growth by localized plastic flow [3,6–10]. This suggests that effects of a corrosive environment and sustained tensile stress are made worse by the presence of hydrogen. Understanding the contribution of hydrogen to deformation and fracture processes is therefore central to developing a mechanistic understanding of SCC. Austenitic stainless steels are highly resistant to embrittlement promoted either by hydrogen or low temperature, retaining significant ductility and fracture toughness in these environments [4,11–15]. Austenitic stainless steel welds can be more susceptible than base materials to fracture, corrosion, and other degradation. The compositions of austenitic welds are typically selected to promote primary solidification as d-ferrite and solid-state transformation to austenite to suppress solidification cracking [16]. The as⇑ Corresponding author. Present address: Structural Integrity Associates, Inc., San Jose, CA 95138, USA. Tel.: +1 408 833 7201. E-mail address: [email protected] (H.F. Jackson). 0010-938X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.corsci.2013.08.004
welded microstructure usually retains several volume percent of the bcc ferrite phase at room temperature, resulting in a duplex austenitic/ferritic microstructure. Studies of duplex stainless steels emphasize their attractive mechanical and corrosion properties, intermediate between those of austenite and ferrite. High hydrogen fugacities at the metal surface promote hydrogen pickup, especially in applications with cathodic protection or sour gas or oil containing significant H2S [1,2]. For hydrogen-exposed austenitic welds and other duplex (two-phase) austenite/ferrite microstructures, the detrimental influence of ferrite on room-temperature ductility and fracture toughness has been noted in several studies [17–23]. Hydrogen transport is complicated by the lower solubility but higher diffusivity and permeability of hydrogen in the ferrite phase [24]. Austenitic steels are not as readily embrittled by hydrogen as ferritic steels [2]. Low temperature can induce a ductile– brittle transition in ferrite and localizes deformation in austenite, yet few studies are reported for hydrogen-assisted fracture of austenitic stainless steel welds at low temperature, particularly studies that quantify crack growth resistance by fracture mechanics methods. Mills [25] reviewed cryogenic fracture toughness of AISI 304 and 316 base metals and their welds at temperatures as low as 4 K, however, studies at less severe sub-ambient temperatures are needed. The mechanism of environment-assisted crack propagation is known to differ depending on the specific combination of metallurgical, environmental, and mechanical variables [5,26]. For example, features of hydrogen damage in austenitic and duplex stainless steels differ when they result from exposure to high pressure hydrogen gas versus cathodic charging in an aqueous solution; or upon tensile straining after precharging with hydrogen versus in situ straining in a hydrogen environment [1,27–30]. Austenitic al-
211
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
loys’ susceptibility to embrittlement has been correlated with a propensity for localized deformation, which can arise due to metallurgical factors such as low stacking fault energy (SFE) [31–33] or environmental factors such as low temperature [33–37] or hydrogen exposure [4,38,39]. Hydrogen-assisted crack growth is sensitive to the combination of metallurgical, environmental, and mechanical variables, hence it is important to understand effects of relevant combinations of material and environmental variables (absorbed hydrogen, low temperature, ferrite/austenite phase distribution) on fracture of austenitic welds. In the present manuscript, we have isolated effects of a high-purity gaseous hydrogen environment. However, hydrogen derived from aqueous sources (e.g. corrosion reactions at a stress-corrosion crack tip) is thought to affect alloys in a similar manner [4,14,38]. The present study expands on the results of a previous investigation of room-temperature hydrogen-assisted crack propagation in gas-tungsten arc (GTA) welds having a 304L base metal and 308L filler [21]. The objectives of the present study are to characterize hydrogen-assisted crack propagation of these welds at low temperature. Fracture mechanics specimens were thermally precharged in hydrogen gas, and fracture initiation toughness and crack growth resistance curves were measured at 223 K (50 °C). The effects of low temperature on hydrogen-assisted crack initiation and propagation mechanisms were assessed via electron microscopy of fracture surfaces and fracture profiles. 2. Experimental The processes of weld fabrication, hydrogen precharging, and fracture testing of welds at room temperature were detailed by Jackson et al. [21] and are summarized here. 2.1. 304L/308L Welds The weld, referred to as 304L/308L throughout this work, was fabricated from a round bar (64 mm diameter) of annealed 304L stainless steel having a longitudinal U-shaped groove. The groove was filled with 1.1 mm diameter 308L filler wire using 9 gas-tungsten arc (GTA) weld passes. Bulk elemental compositions of the bar, wire, and weld fusion zone are reported in Table 1. The volume fraction of ferrite in the weld fusion zone was estimated by magnetic measurement (Feritscope, Helmut-Fischer GmbH, Sindelfingen, Germany) at several distances from the weld root along the centerline. The as-welded microstructure was imaged using optical and scanning electron microscopy (SEM, JSM-840, JEOL, Tokyo, Japan). The elemental compositions of the ferrite and austenite phases within the weld were quantified by electron probe microanalysis (EPMA, JXA-8200, JEOL, Tokyo, Japan). Variation in tensile properties as a function of depth in the 304L/308L weld was characterized by tensile tests of subsized specimens which were extracted from the root, mid-height, and top of the weld. Upon going from the root to the top of the weld, yield strength decreases
from 420 MPa to 350 MPa, while strain at failure increases from 40% to 50% [21]. 2.2. Preparation of fracture specimens Disk-shaped compact-tension (CT) specimens, illustrated schematically relative to the weld in Fig. 1, had a width W (distance from the load-line to the back face of the specimen) of 40.6 mm, thickness B of 6.4 mm, and net thickness Bn (between the sidegrooves) of 4.8 mm, consistent with ASTM E1820 [40] and ASTM E1737 [41]. Specimens tested at room temperature in the noncharged condition had the same W but with a B of 12.7 mm and Bn of 9.9 mm. In order to locate the fatigue precrack tip in a similar microstructure in all specimens, the precrack was grown to the same distance into the weld fusion zone, about 1.5 mm. To accommodate variation in the distance from the notch tip to the root of the weld in the machined specimens, the final crack length-to-width ratio varied between 0.63–0.68. Precracks were grown in air along the radial direction of the welded bar along the weld centerline at 10 Hz under a load ratio of 0.1 and final maximum stress intensity factor Kmax of 30 MPa m1/2. Residual stresses in precracked specimens are minimized by conducting precracking in accordance with standard procedures in ASTM E1820 [ref]. Specimen dimensions conformed to allowable dimensions (with respect to specimen geometry) of the starter notch, final precrack, and amount of fatigue crack growth. The Kmax applied during precracking was kept well below the material fracture toughness that is measured during subsequent fracture testing. During the precracking procedure, the displacement is kept constant, hence the Kmax and DK range decrease progressively as the crack grows. Thermal precharging of CT specimens was in 99.9999% hydrogen gas at 138 MPa and 573 K for a minimum of 39 days and up to 61 days. The charging time and temperature were selected to achieve at the specimen midthickness a minimum hydrogen concentration of 90% of the equilibrium hydrogen concentration at the surface. After hydrogen precharging but before mechanical testing, specimens were stored at below 250 K to minimize hydrogen outgassing. The specimen was equilibrated at the test temperature, either 293 or 223 K, for at least 30 min prior to mechanical testing. The hydrogen concentration in the fusion zone was 140 wppm (0.8 at.%), as measured by inert gas fusion (Wah Chang, Albany, OR). This concentration is consistent with that predicted based on the thermal precharging parameters and the hydrogen solubility of 300-series alloys [42]. 2.3. Fracture mechanics testing Elastic–plastic fracture mechanics tests of the fatigue-precracked disk CT specimens were conducted according to ASTM E1820 [40]. Three or four replicate specimens were tested for each
Table 1 Elemental compositions (wt%) of base metal, filler, and weld fusion zone (balance Fe).
a
304L 308La 304L/308L weld b,c 304L/308L weldb,d a b c d e
Cr
Ni
Mn
Si
Mo
C
S
P
N
Cu
Nb
Creqe
Nieqe
19.85 20.5 21.27 20.75
10.73 10.3 10.23 10.19
1.6 1.56 1.69 1.65
0.57 0.5 0.51 0.51
0.17 <0.01 0.05 0.04
0.030 0.028 0.04 0.02
0.001 0.012 0.02 0.011
0.005 0.006 0.006 0.005
0.02 0.055 0.047 0.048
– 0.018 0.02 0.02
– <0.005 0.01 –
19.85 20.51 – –
12.18 12.38 – –
Per manufacturer certification. By emission spectroscopy. Analysis at mid-height of weld. Analysis at root of weld Nieq = Ni + 35C + 20N + 0.25Cu, and Creq = Cr + Mo + 0.7Nb per Kotecki and Siewert [43].
212
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
tiation of crack extension following blunting was identified as the second slope change in the DCPD vs. COD record, which is consistent with observations from interrupted fracture mechanics tests of stainless steels [22]. The fracture initiation toughness was identified from the intersection of the 0.2 mm offset blunting line with the R curve. 2.4. Post-testing characterization Fracture surfaces and fracture profiles parallel to and normal to the crack growth direction were characterized by SEM. Elemental analysis of compositional partitioning by energy dispersive X-ray spectroscopy (EDS) was used to infer phases in the weld fusion zone. Fig. 1. Optical image of the 304L/308L weld in cross-section, with the orientation and location of disk-shaped CT specimens illustrated schematically.
3. Results 3.1. As-welded microstructure
condition. Room-temperature tests were conducted in the ambient laboratory air. Tests at 223 K were conducted within an environmental chamber mounted in the load train of a servo-hydraulic test frame (MTS Corporation, Eden Prairie, MN). Evaporated liquid nitrogen was used as a coolant, with a resistive heater and temperature controller maintaining the test temperature of 223 K. Testing was conducted at a constant (monotonic) displacement rate of 0.4 mm min1. The crack-opening displacement (COD) at the load-line was measured with a clip gauge, and crack extension was monitored continuously using the direct current potential drop (DCPD) method of ASTM E1737 [41]. The test was terminated manually after a greater than 10% load drop from peak load and an increase in DCPD by at least 15% from its initial value, or when load-line displacement exceeded the measurement capacity of the clip gauge (4–5 mm). Following the test, CT specimens were heat-tinted at 623 K for 60 min to mark the extent of crack growth during the test. The specimens were then broken apart in liquid nitrogen, and the final crack length and precrack length were measured from the fracture surfaces. The calculated crack length based on DCPD data was linearly scaled so that the final calculated and measured crack lengths were equal. Prior to correction, the discrepancy between the final measured and calculated crack lengths was less than 1.5%. The J-integral crack growth resistance (J–R) curve was constructed from the load vs. COD and DCPD vs. COD records. The initial J vs. crack growth increment (Da) relationship was assumed to be a linear crack blunting response following ASTM E1737. The ini-
The microstructure of the fusion zone consisted of primary dferrite in a matrix of austenite (c), as seen in Fig. 2. The ferrite number (FN) as measured at distances of 2, 6, 10, and 14 mm from the weld root along the centerline was 9.4, 10.1, 10.7, and 10.1, respectively. These values represent the mean of five measurements per location. The FN approximates the volume percent ferrite, and the measurements are consistent with values of 8–10 predicted from the WRC-1992 diagram [43], based on the calculated Creq and Nieq for the weld metal (Table 1). Table 2 shows the elemental compositions of the austenite and ferrite phases within the weld, as measured by EPMA spot analyses at two locations: at mid-height of the weld and near the weld root. Table 2 also compares the compositions of the individual phases to the average composition of the weld fusion zone, as measured by emission spectroscopy in the same locations. In the fusion zone, the average composition by emission spectroscopy closely matched that of the original 308L filler wire. Weld austenite had approximately the same wt% Ni and Cr as in the 304L and 308L starting materials, while the approximately 8–10 vol% of d-ferrite was depleted in Ni and enriched in Cr. 3.2. Fracture toughness Compared to the non-charged condition, precharged hydrogen markedly decreased the fracture initiation toughness of 304L/ 308L welds at both 293 and 223 K, as shown by crack growth resis-
Fig. 2. (a) Optical and (b) backscattered electron images of the fusion zone in the 304L/308L weld near the fusion zone/base metal interface at the weld centerline. The solidification direction is nominally from bottom to top of the image. The microstructure contains skeletal ferrite (d) in an austenite (c) matrix.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221 Table 2 Elemental compositions (wt%) of ferrite and austenite phases within the weld. Phase
a b c
Location
Cr
Ni
Weld ferrite
a
Weld mid-height Weld roota
28.17 27.86
4.41 4.38
Weld austenite
Weld mid-heighta Weld roota
20.02 19.73
10.24 10.40
Weld average
Weld mid-heightb Weld rootb 308L filler wirec
21.27 20.75 20.5
10.23 10.19 10.3
Base metal
Outside fusion zonea 304L annealed barc
18.44 19.85
10.21 10.73
EPMA, mean of three measurements for ferrite and nine for austenite. Emission spectroscopy. Per manufacturer certification.
tance (J–R) curves (Fig. 3) and fracture initiation toughness (Jmax or JIH) values (Table 3). Also reported in Table 3 are the corresponding stress-intensity factor KJ values, determined by:
KJ ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi EJ 1 m2
ð1Þ
where E is Young’s modulus, J is fracture initiation toughness, and m is Poisson’s ratio [40]. The J–R curve slope over the first 0.5 mm of crack extension immediately following blunting is reported as crack growth resistance dJ/dDa.
213
Welds not exposed to hydrogen and tested at both 293 and 223 K exhibited significant crack tip blunting, and tests were terminated manually before the onset of noticeable crack extension, as confirmed by post-test analysis. Crack extension in non-charged specimens was insufficient to permit calculation of JIC, and instead we report as Jmax the lesser of two values, either (1) the maximum J value recorded during the test or (2) the maximum J-integral capacity of the specimen as permitted by ASTM E1820 [40], i.e. the lesser of b0rY/10 or BrY/10, where b0 is the original uncracked ligament, B is the original specimen thickness, and rY is the material flow stress. These Jmax values represent lower-bound fracture initiation toughness measurements. It is important to note that the difference in Jmax between noncharged welds tested at 293 and 223 K (Table 3) was primarily due to the difference in specimen thickness and the corresponding dependence of the maximum J-integral capacity on this dimension. Non-charged specimens tested at 223 K were thinner than those tested at 293 K, while hydrogen-precharged specimens were all of the same thickness. Regardless, crack blunting and a lack of noticeable crack extension were observed for non-charged specimens tested at both 293 and 223 K. The measured fracture initiation toughness of hydrogen-precharged welds represents a reduction of greater than 59% compared to the lower-bound fracture initiation toughness of noncharged welds. Hydrogen-precharged specimens met the dimensional requirements in ASTM E1820 [40] to define conditional JQ as JIC. To reflect the hydrogen-precharged condition, fracture initiation toughness is denoted JIH and stress-intensity factor is KJIH. Fracture initiation toughness with internal hydrogen is essentially the same at 293 and 223 K. At the initiation of crack growth (Fig. 3b), the J–R curves for both temperatures are indistinguishable. However, at longer crack extensions (Fig. 3a), welds tested at the lower temperature exhibit lower crack growth resistance dJ/dDa. This divergence in J–R curve slope occurs at a crack extension of about 0.8–1 mm from the precrack tip. This distance was found to correspond to a change in orientation and morphology of austenite grains and d-ferrite dendrites at a depth in the fusion zone where two weld passes overlapped (Fig. 4). Crack paths were tortuous in general, and any crack deflection or differences in fracture appearance due to microstructural differences in the weld pass overlap region were not visibly distinguished from that in other regions. 3.3. Fractography
Fig. 3. Crack growth resistance (J–R) curves for 304L/308L GTA welds with and without precharged hydrogen, tested at 293 K and 223 K. Up to four J–R curves per condition are plotted. (a) At longer crack extensions, welds tested at 293 K have greater crack growth resistance than at 223 K. (b) Immediately following crack initiation, J–R curves were indistinguishable at 293 versus 223 K, as were JIH values.
Fracture features are significantly different for specimens fractured with and without hydrogen; for specimens fractured in hydrogen, a different distribution of fracture features is seen when test temperature is decreased from 293 to 223 K. Effects of hydrogen and test temperature on fracture features are compared in Fig. 5 and in higher magnification in Fig. 6. Non-charged welds at both 293 and 223 K exhibited significant crack blunting followed by a slight amount of ductile crack extension. These fracture surfaces featured only equiaxed and U-shaped dimples (Fig. 5a and b), indicating microvoid coalescence (MVC). Under these conditions, microvoids nucleate at fine spherical precipitates. Hydrogen exposure altered the fracture mode of austenitic welds (Fig. 5c and d). In hydrogen-precharged welds fractured at 293 K, the fracture surface was dominated by features which were elongated and aligned parallel to the crack propagation direction, which was also the weld solidification direction (Figs. 5c, 6a and b). The dendritic morphology of these features (denoted by D in Figs. 5c and 6a) resembled the underlying d-ferrite microstructure (Fig. 2), with smooth, relatively flat features interspersed with tear ridges (denoted by T in Fig. 6a), i.e. vertical steps joining two flat
214
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
Table 3 Fracture initiation and crack growth toughness of 304L/308L GTA welds Non-chargeda,b
Hydrogen-precharged 2
a b c
0.5
JIH (kJ m )
KJIH (MPa m
223 K
144 123 109
175 161 152
171 175 181
Mean
125
163
176
293 K
110 123 132 111
153 161 167 154
209 190 196 242
Mean
119
159
209
)
dJ/dDa
c
2
1
(kJ m mm
)
2
Jmax (kJ m
)
dJ/dAac (kJ m2 mm-1) 0.5
KJ (MPa m
)
dJ/dDa
304
254
965
625
364
965
c
(kJ m2 mm1)
Mean of 3 replicate tests 6 of non-charged specime. B of non-charged specimens was 6.35 mm for tests at 50 °C and 12.7 mm for tests at 20 °C. Slope of J–R curve, measured over 0.5 mm of crack extension immediately following blunting.
Fig. 4. A change in microstructural orientation and morphology was observed where two weld passes overlap, about 0.8–1 mm from the precrack tip and 2.3–2.5 mm from the base metal/weld interface. Backscattered electron image.
Fig. 5. Fracture surfaces of hydrogen precharged and non-charged welds, tested at 223 or 293 K. Without hydrogen, at (a) 293 K or (b) 223 K, fracture surfaces had only dimpled rupture. With hydrogen at (c) 293 K, the fracture surface was dominated by elongated features with a dendritic morphology (D) and aligned with the crack growth direction, while at (d) 223 K, the dominant features were elongated (E) and equiaxed (Q) dimples, with few dendritic-shaped features. Crack growth direction is from bottom to top in all images.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
215
Fig. 6. Fracture features at matching locations on both halves of hydrogen-precharged fracture surfaces. At 293 K (a and b), the dominant features were dendritic (D) with tear ridges (T) between dendritic features. At 223 K (c and d), few regions had dendritic features (D). The dominant features (e and f) were planar arrays of elongated (E) dimples and patches of equiaxed (Q) dimples. Crack growth direction is from bottom to top in all images.
areas of dimpled rupture, with steps also exhibiting dimples. Most of the fracture surface exhibited features related to the underlying dendritic structure of ferrite and the inter-dendritic regions of austenite. In hydrogen-precharged welds fractured at 223 K (Figs. 5d, 6c, and d), considerably fewer regions of the fracture surface showed evidence of the dendritic microstructure (denoted by D in Fig. 6c and d). Most regions showed either elongated or equiaxed dimples (denoted by E and Q respectively in Figs. 5d and 6e). Elongated dimples occurred in nominally parallel and coplanar arrays (Fig. 6e and f). Parallel arrays of elongated dimples formed steps on the fracture surface (Figs. 5d and 6e). Matching locations on both fracture surfaces of hydrogen-exposed welds (Fig. 6) were analyzed by EDS to provide a basis for inferring the phase that was present at the surface based on the approximate Ni and Cr concentrations, which were consistent with the values for austenite or ferrite in Table 2. EDS analysis alone may be inconclusive if the ferrite dimensions are very thin and ferrite evades detection by the EDS probe. By comparing elemental analyses and fracture features on both halves of the fracture surface, the crack path could be identified as fracture within ferrite, at the austenite/ferrite phase boundary, or within austenite. All
three types of crack path were observed in various locations at both temperatures. In the present work, no distinction was made between cracks at c/c boundaries and cracks within a c grain, although both may occur. For this reason, it cannot be conclusively stated that d was absent at a given fracture location and that the crack path was entirely contained within the c phase. Planar arrays of elongated dimples, seen only at 223 K, were associated with fracture within the austenite phase; such crack growth could be either transgranular or at c/c interfaces. Previously [20,21], it was established that for hydrogen-precharged austenitic welds at 293 K, tear ridges and equiaxed dimples were associated with fracture within austenite; the flat areas containing dendritic features were associated either with fracture at the austenite/ferrite phase boundary or within ferrite. At both temperatures, fracture surfaces exhibited significant out-of-plane crack growth and tortuous crack topographies. With hydrogen at both 293 and 223 K, crack extension evolved as multiple discontinuous microcracks that joined up to form the main crack path. At 293 K, the microcracks formed on planes parallel to the Mode I crack growth plane, i.e. normal to the tensile axis [21]. At 223 K, microcracks formed primarily on planes inclined
216
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
to the Mode I crack growth plane, as confirmed from stereo pair fractographs. 3.4. Fracture profile metallography of hydrogen-exposed welds Fracture profiles parallel and transverse to the crack growth direction in hydrogen-exposed welds were imaged by backscattered electron microscopy (Figs. 7 and 8). Atomic number contrast distinguishes d-ferrite as the dark grey dendritic phase and austenite as the light grey matrix. Localized deformation bands appear as bright parallel streaks in the light grey austenite matrix. Fracture profiles (Fig. 7) suggested that, although microvoids/ microcracks initiated near ferrite at both 223 and 293 K in hydrogen, crack growth occurred by different mechanisms at the two temperatures. At 293 K, microcracks were predominantly associated with dferrite (Fig. 7a and b). Microcracks nucleated either within the ferrite phase or at the austenite/ferrite phase boundary (Fig. 7a) and propagated parallel to ferrite dendrites (denoted by d in Fig. 8a). These microcracks often created coarse cracks (Fig. 7b) parallel to the main crack, which were located up to several hundred micrometers above or below the fracture surface (Fig. 8c). At 223 K, there was no evidence for large microcracks away from the fracture surface. Fine microvoids 1 lm or less in diameter (Fig. 7d and f) were the main type of microstructural damage ob-
served. With an equiaxed cross-section, the three-dimensional shape of these microvoids could be either spherical or tubular. Some fine porosity was randomly distributed in the as-welded microstructure, however the microvoids that were associated with damage accumulation and fracture had a distinctive arrangement. Specifically, these microvoids tended to be aligned and to coincide with the intersections between deformation bands (Fig. 7f). The predominant microvoid nucleation site was adjacent to d-ferrite (Fig. 7c–f), while a smaller fraction formed elsewhere in austenite grains. Microcracks evolved via the nucleation and coalescence of closely-spaced microvoids (Fig. 7c, d, and f). Fracture profiles reveal a second key difference between fracture at 293 K and at 223 K: the macroscopic crack path. At both 293 and 223 K, fractures had significant out-of-plane crack growth and tortuous crack topographies (Fig. 5c and d). This is reflected in cross-sections as macroscopic steps (denoted by S in Fig. 8a,c,d). At 293 K, steps corresponded to microcracks growing along dferrite on various planes ahead of the crack tip that were nominally parallel to each other and perpendicular to the tensile axis (Fig. 8a and c). The main crack grew when these microcracks joined along planes parallel to the tensile axis, facilitated by intense shear in the ligaments between them (denoted by I in Fig. 8a). This out-of-plane crack growth corresponds to that seen on fracture surfaces (Fig. 5c).
Fig. 7. Fracture profiles show different mechanisms of microvoid nucleation at 293 versus 223 K. At 293 K (a and b), microcracks originate at and propagate along d-ferrite. At 223 K (c–f), microvoids nucleate where localized deformation bands intersect other deformation bands, and microvoids are concentrated near phase boundaries. Microcracks form via the coalescence of multiple microvoids (c, d, and f). Crack growth direction is from left to right in all images except for a, in which crack growth is normal to the page.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
217
Fig. 8. Fracture profiles reveal that at both temperatures, the crack path is marked by macroscopic steps (S). At 293 K (a and c), cracks propagate along ferrite dendrites (d) but can link up through intense shear (I) in the ligaments between them. At 223 K (b and d), crack paths are primarily within austenite and do not clearly follow ferrite.
At 223 K, macroscopic steps and out-of-plane crack growth were also observed. However, microcracks were observed to form on planes inclined to the tensile axis (Fig. 6c–f). In cross-section, the crack path does not clearly follow the ferrite phase (Fig. 8b and d) and is primarily within the austenite phase.
4. Discussion As shown in previous studies [20,21], one primary role of hydrogen in degrading fracture resistance and altering fracture mechanisms in austenitic stainless steel welds at room temperature is enhancing planar deformation in austenite. This phenomenon produces stress concentrations where localized deformation bands intersect d-ferrite. These stress concentrations facilitate microcracking of the elongated d-ferrite grains, which are aligned with the direction of crack growth. This dominant role of ferrite in the initiation and propagation of microcracks in austenitic welds at room temperature is illustrated in Figs. 6a, 6b, 7a, andb and documented in several studies [17–21]. Extending the study on 304L/308L welds at room temperature [21], we found that low-temperature (223 K) exposure had no effect on the fracture initiation toughness of hydrogen-precharged welds (Table 3), although the fracture appearance was clearly changed (Fig. 5c and d). In contrast to room-temperature fracture, at 223 K fracture surfaces and fracture profiles showed a diminished role of d-ferrite in facilitating crack propagation (Figs. 6 and 7), relative to room temperature. We propose that low temperature alters the competition between potential fracture initiation modes in hydrogen-precharged austenitic welds. The two main effects of low temperature are that (1) low temperature localizes plastic deformation in austenite, rendering this phase more susceptible to hydrogen-enhanced localized plasticity, and (2) low temperature increases the hydrogen concentration in austenite near phase boundaries through its effects on the thermodynamics of hydrogen trapping. In the following sections, we discuss effects of low temperatures on plastic
deformation and hydrogen distribution in welds and how these changes govern the fracture mode. 4.1. Sequence of crack initiation and propagation in hydrogenprecharged welds at 223 K In the presence of hydrogen, low temperature (223 K) produced a significantly different fracture surface than seen at room temperature (293 K). At 293 K, a mixture of ferrite cleavage and phase boundary separation was observed. Microcracks nucleated and propagated preferentially at d-ferrite (Figs. 7a and b), resulting in elongated, dendritic features on fracture surfaces (Figs. 5c, 6a, and 6b) that resemble the underlying ferrite morphology (Fig. 2). These microcracks formed in the ferrite phase itself or at austenite/ ferrite interfaces, facilitated by high stress concentrations at these sites. At 223 K, neither ferrite cleavage nor phase boundary separation is an important fracture mechanism, based on the observed fracture surface features (Fig. 6c–f). Rather, the dimpled surface indicates that fracture was caused by MVC within the austenite phase. Microcracks originated as microvoids at intersections between localized deformation bands in austenite (Fig. 7c–f). Because deformation bands are parallel, intersections between deformation bands occur in rows, as do the microvoids that form at these intersections. The non-equiaxed growth of these aligned microvoids produced coplanar arrays of elongated fracture dimples (Figs. 5d, 6e, and 6f). In limited locations, the fracture appearance at 223 K resembled the dendritic features associated with microcracks in d-ferrite or at austenite/ferrite boundaries at 293 K (denoted by D in Fig. 6c and d). An important feature of the proposed low-temperature fracture mechanism is that, although deformation band intersections occurred throughout the austenite bulk, microvoids were not randomly distributed but were concentrated near austenite/ferrite phase boundaries (Fig. 7c–f). We can infer that in welds or two-phase microstructures, two conditions promote microvoid nucleation in austenite: (1) deformation bands inter-
218
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
sect and (2) the intersection is in close proximity to an austenite/ferrite phase boundary. The role of hydrogen in facilitating void formation near phase boundaries will be discussed in the final section. Despite the differences in fracture appearance, the fracture process at 223 K had a similar sequence of steps compared to the fracture process at 293 K [20,21]. First, planar deformation in austenite developed as localized bands. The formation of these deformation bands is facilitated by the hydrogen enhanced localized plasticity (HELP) mechanism [44–46]. In austenite, HELP results in greater slip planarity [4,38,39], exacerbating the effects of metallurgical factors predisposing an alloy to planar slip, such as low stacking fault energy (SFE) [31–33]. In metastable 300-series alloys such as 304L and 308L, SFE decreases as nickel content decreases [47,48]. It follows that the effects of low SFE, low temperature, and hydrogen would produce more severely localized slip than any factor acting independently. As slip becomes more localized, other planar deformation modes such as twinning and e-martensite become activated at lower strains [33,35,49–53]. The next step in the process was the impingement of localized deformation bands on microstructural obstacles, thus producing stress concentrations. Austenitic stainless steels contain various microstructural features that can act as obstacles to slip or twinning, including grain boundaries, secondary phases, and annealing twin boundaries. At 293 K, the d-ferrite phase could serve as a barrier to propagation of planar deformation bands, and the resulting stress concentration facilitated microcrack initiation either as cleavage of ferrite or decohesion at the austenite/ferrite phase boundary [21]. At 223 K, the resulting microstructural damage was instead a series of discrete microvoids at deformation band intersections near phase boundaries. Hence, low temperature changed the dominant mode of damage accumulation, leading to a different mode of microcrack initiation relative to that at room temperature. The final step was crack propagation via the growth and coalescence of microcracks. At 293 K, microcracks formed from cleavage in ferrite or decohesion of austenite/ferrite interfaces, then these microcracks linked by intense shear in the remaining ligaments (Figs. 7a, 7b, 8a, and 8c). At 223 K, microvoids nucleated predominantly adjacent to weld ferrite, while ferrite cleavage and austenite/ferrite interface decohesion were infrequently observed. Instead, under low-temperature conditions, decohesion was not a dominant fracture mode; an alternative mode of crack propagation intervened, and multiple microvoid nucleation and coalescence events were required for microcracks to form (Fig. 7c–f). Similar to the process at room temperature, individual microcracks ahead of the crack tip at low temperature linked up via shear in the ligaments between them, producing coarse steps on the fracture surfaces (Fig. 8b and d). 4.2. Effect of temperature on crack initiation: localized deformation In this study, temperature governed the mode by which hydrogen-assisted cracks initiated and propagated. At both 293 and 223 K, planar deformation and microstructural obstacles played a prominent role in the initiation of hydrogen-assisted cracking in austenitic stainless steel welds. We propose that low temperature altered fracture mechanisms in hydrogen-precharged welds in part by conditioning the austenite for more severe planar slip that is further exacerbated by hydrogen. The planar deformation bands that evolve increase the severity of stress concentrations at intersecting deformation bands near the austenite/ferrite phase boundary, perhaps elevating stress concentrations at these sites relative to the stress concentrations at intersections of deformation bands with d-ferrite.
Consequently, void nucleation in austenite near phase boundaries is favored over microcracking in ferrite or at austenite/ferrite interfaces. Low temperature exacerbates localized deformation in several ways. First, low temperature increases the tendency toward planar slip in austenite by decreasing SFE [33–37]. More intense planar slip on [1] planes in fcc austenite gives rise to deformation twins, bcc a0 -martensite, and hcp e-martensite at lower strains [33,35,49–53]. These deformation modes promote the confinement of deformation in discrete bands, leading to more pronounced strain incompatibility and enhanced stress concentration at obstacles. After deforming 301LN stainless steels in tension at temperatures between 233 and 353 K, Talonen and Hanninen [33] observed that, as temperature decreased, a greater volume fraction of a0 martensite formed, and at lower levels of strain. They proposed that low SFE at low temperature facilitates the martensitic transformation. In contrast, at higher temperature, the overlapping of stacking faults is less regular, hindering the formation of martensite nuclei. Deformation at lower temperature produces more twins and martensite – the results of planar deformation modes – than at room temperature. Second, low temperature could also change the types of deformation products that preferentially form. In addition to decreasing SFE, low temperature could alter the free-energy differences between the fcc, bcc, and hcp phases, and thus the thermodynamic driving force for transformation to the hcp or bcc martensite phases. During an investigation of tensile deformation of austenitic steels containing 16-18 wt% Cr and 11–13 wt% Ni at temperatures between 77 and 523 K, Lecroisey and Pineau [35] observed deformation bands containing both deformation twins and e-martensite, with a higher proportion of twins at higher deformation temperatures. They proposed that the tendency to nucleate austenitic twins versus e-martensite at a given temperature depends on the relative values of extrinsic and intrinsic stacking fault energies as well as the volume contraction due to the fcc-to-hcp transformation. Temperature probably determines the relative proportions of hcp or bcc martensite and fcc twins that compose deformation bands. Although insufficient data exists to evaluate effects of deformation band makeup on the severity of stress concentrations, it is reasonable to presume that deformation products such as twins and e-martensite confine deformation in discrete bands, promoting strain incompatibility and stress concentration at obstacles [21]. Third, more severe planar deformation has been correlated with physically thinner deformation twins and e-martensite plates and, consequently, higher stress concentrations where they impinge on obstacles. Based on an investigation of uniaxial tensile fracture of nitrogen-bearing austenitic steels at temperatures between 4 and 298 K, Mullner and coworkers [54,55] attributed fracture initiation to stress concentrations at deformation twin intersections. Lower temperature decreased the thickness of twins and the dislocation density, and they asserted that such conditions rendered twins effective obstacles against propagation of impinging twins, creating stress concentrations high enough to initiate microcracks at twin intersections. Based on fractographs and fracture profiles of low-SFE, highMn austenitic steels fractured in Charpy tests at 77 K, Takaki et al. [56] inferred that intersections between e-martensite plates and the associated stress concentrations led to the formation of parallel tunnel-like voids at these intersections. They invoked this mechanism to explain the dimpled planes and steplike ridges they observed on fracture surfaces. It is plausible that propagation of both twins and e-martensite plates could be blocked by the austenite/ferrite phase boundary as well. Thinner deformation twins and e-martensite plates produced at lower
219
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
temperatures would lead to more severe stress concentrations than at room temperature. While the preceding discussion supports the notion that low temperature conditions austenitic stainless steels for more localized deformation, which can then be exacerbated by elevated hydrogen, leading to elevated stress concentrations at intersecting deformation bands, one detail still requiring attention is the observation that at low temperature, microvoids form predominantly near austenite/ferrite phase boundaries. One possible explanation for this observation is that deformation is disproportionately localized near austenite/ferrite phase boundaries, where Ni depletion locally decreases SFE. Brooks et al. [18,57] attributed the Ni concentration profile observed for AISI 309 GTA welds to partitioning during the diffusion-controlled solid-state transformation from ferrite to austenite following primary ferrite solidification. The authors attributed Ni enrichment at the centers of austenite grains (i.e. the boundaries of solidification cells) to secondary austenite solidification due to Ni enrichment of the liquid. EDS concentration profiles of the present weld suggested that Ni concentration was 2.5–3 wt% lower near the phase boundaries than in the centers of austenite grains (i.e. the boundaries of solidification cells), where the 12– 13 wt% Ni exceeded the nominal Ni concentration in weld austenite, about 10.2 wt% (Table 2). Such concentration gradients could bias localized deformation toward the austenite/ferrite phase boundaries. In summary, low temperature enhances the formation of planar deformation bands in austenite, especially near austenite/ferrite phase boundaries. One potential role of hydrogen in linking deformation and fracture is that hydrogen magnifies planar slip. This intensified planar slip or the deformation products that it stimulates (e.g. twinning, e-martensite) leads to stress concentrations at obstacles such as intersecting deformation bands. The observation that microvoid nucleation in austenite near phase boundaries out-competes microcrack formation in ferrite could be due in part to increased stress concentrations at deformation band intersections near phase boundaries relative to those at intersections between deformation bands and d-ferrite. The difference in damage initiation modes gave rise to the significant difference in fracture surface appearance. While the fracture modes at 293 and 223 K differed significantly, temperature had no notable effect on fracture initiation toughness (Table 3). We propose that the similar JIH values at 293 and 223 K are a coincidence. In other words, it is coincidental that the dominant fracture processes at these two temperatures are characterized by a similar magnitude of the local fracture criterion, e.g., a critical local strain. For example, at cryogenic temperatures, crack nucleation would likely occur by cleavage of d-ferrite due to its ductile–brittle transition [25]. Other fracture modes might dominate at different sub-ambient temperatures, altering crack initiation toughness. On the other hand, dJ/da, a measure of crack growth resistance, was slightly lower at 223 K than at 293 K. According to Ritchie and Thompson [58], metallurgical factors affecting the local fracture strain have a greater effect on crack growth than on crack initiation, owing to a weaker strain singularity ahead of a stable growing crack versus a stationary crack. Metallurgical factors dictating ductility include local microstructure and the degree of planar deformation. In the present welds, the J–R curves for tests at 293 and 223 K coincided near initiation but diverged after cracks extended into a region of overlap between weld passes and a change in microstructural orientation and morphology (Fig. 4). This microstructural change and the increase in planar deformation at 223 K had a stronger influence on dJ/da at 223 K than at 293 K.
4.3. Effect of temperature on crack initiation: local hydrogen distribution We propose that low temperature alters the thermodynamics of hydrogen trapping in austenitic stainless steel welds in a way that redistributes hydrogen from the interiors of d-ferrite grains to phase boundary trap sites. At 223 K, this could contribute to hydrogen-assisted fracture initiation near phase boundaries rather than within d-ferrite. The total concentration of dissolved hydrogen consists of hydrogen in interstitial lattice sites and hydrogen trapped at microstructural features such as grain boundaries, phase boundaries, dislocations, and impurities [59]. Luppo et al. [60] provided evidence that austenite/ferrite phase boundaries in welds serve as hydrogen trap sites. According to Hirth [61], trap sites become more densely populated as temperature decreases, as expressed in following equation:
hT EB ¼ hL exp½ 1 hT kT
ð2Þ
where hT is the trap site occupancy, hL is the lattice hydrogen concentration (in this formulation, hL 1), Eb is the binding energy of hydrogen in the trap site, k is Boltzmann’s constant, and T is temperature. Because a hydrogen-precharged specimen can be considered a closed system with a fixed total hydrogen concentration, the hydrogen demanded by trap sites at lower temperature must be supplied by lattice sites. Conversely, when temperature increases, trap occupancy decreases, with hydrogen migrating into lattice sites. From a kinetic standpoint, the d-ferrite phase is more likely than austenite to supply hydrogen to trap sites at phase boundaries. Based on the values recommended by Perng and Altstetter [62] for ferritic stainless steel 29Cr–4Mo–2Ni and those recommended by San Marchi et al. [42] for 300-series austenitic stainless steels, the diffusivity of hydrogen in ferrite and austenite were determined at 293 and 223 K and are summarized in Table 4. Hydrogen diffusivity is four to six orders of magnitude faster in bcc ferrite than in fcc austenite at the test temperatures. Even though the lattice solubility of hydrogen in austenite is about 100 times that in ferrite [63], and thus the concentrations will differ between the two phases, the overall diffusional flux will still be greater in d-ferrite. The present analysis demonstrates that hydrogen diffusivity in d-ferrite is rapid enough to satisfy the increased demand for hydrogen at traps as induced by low temperature. The hydrogen diffusion distance x varies with diffusivity D and time t according to Eq. (3), derived from the solution to the transient diffusion equation:
pffiffiffiffiffiffi x 2 Dt
ð3Þ
Specimens were exposed to the test temperature for at least 30 min and for up to several hours prior to mechanical testing. The hydrogen diffusion distance in d-ferrite during a typical expoTable 4 Hydrogen diffusivity D and diffusion distance x in ferrite and austenite phases at 293 and 223 K a
(m2 s1)
x (lm)
T (K)
D
t = 30 min
t=3h
Ferrite
223 293
1 1013 8 1012
30 200
60 600
Austenite
223 293
2 1019 2 1016
0.04 1
0.1 3
a Based on the values recommended by Perng and Altstetter [62] and San Marchi et al. [42].
220
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
sure (Table 4) is significantly greater than the maximum required migration distance, i.e. from the center of a ferrite grain to trap sites at phase boundaries, about 0.5 lm. At low temperatures, hydrogen migration from lattice sites in dferrite to traps at phase boundaries both depletes hydrogen from ferrite grains and enriches phase boundaries with hydrogen. This tilts the competition away from microcrack formation in ferrite and toward microvoid nucleation near phase boundaries. As a result of this hydrogen enrichment at phase boundaries that is induced by low temperature exposure, we propose that hydrogen has two roles in concentrating damage near phase boundaries. First, hydrogen in concert with low temperature localizes deformation and increases stress concentrations near phase boundaries, consistent with the concepts described in the previous section. Second, hydrogen could have a direct role in facilitating microvoid nucleation near phase boundaries through several possible mechanisms. Vacancies are generated by intense localized plastic deformation and stabilized by hydrogen, according to the model of hydrogen-stabilized vacancy formation proposed by Nagumo [64]. Building upon this model, Neeraj et al. [65] propose a nanovoid coalescence (NVC) micromechanism of hydrogen embrittlement for ferritic steels. In the NVC mechanism, nanovoids nucleate at a critical excess vacancy concentration. They proposed that NVC gave rise to the nanoscale (10–20 nm) dimples they observed on quasi-brittle facets on fracture surfaces. In the previous section, we linked the formation of microvoids in austenite to intersections between planar deformation bands that may consist of e-martensite plates [56] and deformation twins [54,55]. It is plausible that hydrogen could directly promote fracture by encouraging void formation at these intersections, in addition to hydrogen’s indirect role in fracture, i.e. localizing deformation. In the context of the present work, we propose that hydrogen enrichment at phase boundaries, particularly at low temperature, makes near-phase-boundary austenite the preferred site for damage nucleation. At low temperature, this damage proceeds in a manner consistent with the aforementioned mechanisms, in which hydrogen stabilizes the vacancies that eventually form microvoids or hydrogen facilitates microvoid nucleation at the intersection of localized deformation bands consisting of e-martensite and deformation twins. At room temperature, the damage is predominantly in the form of microcracks in ferrite or at phase boundaries. 5. Conclusions Elastic–plastic fracture mechanics tests were conducted to evaluate effects of low temperature on hydrogen-assisted crack propagation in austenitic stainless steel welds fabricated from AISI 304L base metal and 308L filler metal and which had been thermally precharged with hydrogen gas: 1. Hydrogen degrades fracture resistance by >59% relative to the lower-bound fracture initiation toughness of non-charged welds. At room temperature (293 K), hydrogen alters the dominant mode of damage accumulation in austenitic welds, thereby altering fracture mechanisms by (1) enhancing localized planar deformation in austenite and (2) facilitating microcracking of d-ferrite or its interfaces. The dendritic microstructure of weld ferrite, which is aligned with the direction of crack growth, governed the crack path. 2. Low temperature (223 K) exposure altered fracture mechanisms in hydrogen-precharged welds. At 223 K, fracture initiation was dominated by microvoid formation at deformation band intersections in austenite near phase boundaries, while
microcracking of ferrite and its interfaces did not play a significant role in crack growth. Low temperature was presumed to (1) lower SFE and condition the austenite for more localized deformation, which was further exacerbated by hydrogen, and (2) enrich phase boundaries with hydrogen by increasing the occupancy of hydrogen at trap sites relative to lattice interstitial sites in ferrite. The more localized deformation in austenite, particularly near phase boundaries, could lead to more severe stress concentrations at deformation band intersections relative to those at intersections of deformation bands with ferrite. 3. Additionally, we propose that elevated hydrogen concentration near phase boundaries may directly promote localized void nucleation in austenite, consistent with the mechanisms of hydrogen-stabilized vacancies proposed by Nagumo [64] and vacancy-induced nanovoid coalescence (NVC) proposed by Neeraj et al. [65]. Alternately, hydrogen facilitated microvoid formation at intersections between localized deformation bands, consistent with literature reports of stress concentrations and microvoid nucleation at intersections between e-martensite plates [56] and deformation twins [54,55]. 4. Exposure to 223 K had no effect on fracture initiation toughness of precharged welds, despite the difference in the dominant mode of damage accumulation. It is likely coincidental that the dominant fracture modes at the two test temperatures (microvoid formation at intersecting deformation bands at 223 K vs. microcracking in ferrite or its interfaces at 293 K) had similar local fracture criteria, resulting in similar fracture initiation toughness values.
Acknowledgments The authors are grateful to J. Campbell for hydrogen pressure systems support, A. Gardea for metallographic preparation, and J. Chames and R. Nishimoto for SEM imaging. Sandia is a multiprogram laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the US Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
References [1] A.A. El-Yazgi, D. Hardie, The embrittlement of a duplex stainless steel by hydrogen in a variety of environments, Corros. Sci. 38 (1996) 735–744. [2] W. Zheng, D. Hardie, The effect of hydrogen on the fracture of a commercial duplex stainless steel, Corros. Sci. 32 (1991) 23–36. [3] Y. Mine, T. Kimoto, Hydrogen uptake in austenitic stainless steels by exposure to gaseous hydrogen and its effect on tensile deformation, Corros. Sci. 53 (2011) 2619–2629. [4] A.W. Thompson, I.M. Bernstein, The role of metallurgical variables in hydrogen-assisted environmental fracture, in: M.G. Fontana, R.W. Staehle (Eds.), Advances in Corrosion Science and Technology, Plenum Press, New York, NY, 1980, pp. 53–175. [5] D. Delafosse, T. Magnin, Hydrogen induced plasticity in stress corrosion cracking of engineering systems, Eng. Frac. Mech. 68 (2001) 693–729. [6] S.P. Lynch, Metallographic contributions to understanding mechanisms of environmentally assisted cracking, Metallography 23 (1989) 147–171. [7] S.P. Lynch, A fractographic study of hydrogen-assisted cracking and liquidmetal embrittlement in nickel, J. Mater. Sci. 21 (1986) 692–704. [8] S.P. Lynch, A fractographic study of gaseous hydrogen embrittlement and liquid-metal embrittlement in a tempered-martensitic steel, Acta Metall. 32 (1984) 79–90. [9] Y. Mine, K. Hirashita, M. Matsuda, M. Otsu, K. Takashima, Effect of hydrogen on tensile behaviour of micrometre-sized specimen fabricated from a metastable austenitic stainless steel, Corros. Sci. 53 (2011) 529–533. [10] M. Koyama, E. Akiyama, T. Sawaguchi, K. Ogawa, I.V. Kireeva, Y.I. Chumlyakov, K. Tsuzaki, Hydrogen-assisted quasi-cleavage fracture in a single crystalline type 316 austenitic stainless steel, Corros. Sci. 75 (2013) 345–353. [11] R.J. Walter, W.T. Chandler: effects of high-pressure hydrogen on metals at ambient temperature, Report no. NASA-CR-102425, Rocketdyne, Canoga Park, CA, 1969.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221 [12] H.G. Nelson, Hydrogen embrittlement, in: C.L. Briant, S.K. Banerji (Eds.), Treatise on Materials Science and Technology: Embrittlement of Engineering Alloys, Academic Press, New York, NY, 1983, pp. 275–359. [13] G.R. Caskey Jr., Hydrogen effects in stainless steels, in: R.A. Oriani, J.P. Hirth, M. Smialowski (Eds.), Hydrogen Degradation of Ferrous Alloys, Noyes Publications, Park Ridge, NJ, 1985, pp. 822–862. [14] N.R. Moody, S.L. Robinson, W.M. Garrison Jr., Hydrogen effects on the properties and fracture modes of iron-based alloys, Res. Mech. 30 (1990) 143–206. [15] C. San Marchi, B.P. Somerday: technical reference on hydrogen compatibility of materials, Report no. SAND2008-1163, Sandia National Laboratories, Livermore, CA, 2008.
221
[37] J. Singh, Influence of deformation on the transformation of austenitic stainless steels, J. Mater. Sci. 20 (1985) 3157–3166. [38] D.G. Ulmer, C.J. Altstetter, Hydrogen-induced strain localization and failure of austenitic stainless steels at high hydrogen concentrations, Acta Metall. Mater. 39 (1991) 1237–1248. [39] K.A. Nibur, D.F. Bahr, B.P. Somerday, Hydrogen effects on dislocation activity in austenitic stainless steel, Acta Mater. 54 (2006) 2677–2684. [40] ASTM, Standard test method for measurement of fracture toughness, ASTM International, West Conshohocken, PA, 2009. [41] ASTM, Standard test method for J-integral characterization of fracture toughness, ASTM International, West Conshohocken, PA, 1996. [42] C. San Marchi, B.P. Somerday, S.L. Robinson, Permeability, solubility and diffusivity of hydrogen isotopes in stainless steels at high gas pressures, Int. J. Hydrogen Energy 32 (2007) 100–116. [43] D.J. Kotecki, T.A. Siewert, Wrc-1992 constitution diagram for stainless steel weld metals: a modification of the wrc-1988 diagram, Weld. J. 71 (1992) 171– 179. [44] C.D. Beachem, A new model for hydrogen-assisted cracking (hydrogen ‘‘embrittlement’’), Metall. Trans. 60 (1972) 437–451. [45] H.K. Birnbaum, P. Sofronis, Hydrogen-enhanced localized plasticity - a mechanism for hydrogen-related fracture, Mater. Sci. Eng. A A176 (1994) 191–202. [46] D.P. Abraham, C.J. Altstetter, Hydrogen-enhanced localization of plasticity in an austenitic stainless steel, Metall. Mater. Trans. A 26A (1995) 2859–2871. [47] C.G. Rhodes, A.W. Thompson, The composition dependence of stacking fault energy in austenitic stainless steels, Metall. Trans. A 8A (1977) 1901–1906. [48] R.E. Schramm, R.P. Reed, Stacking fault energies of seven commercial austenitic stainless steels, Metall. Trans. A 6A (1975) 1345–1351. [49] P.L. Mangonon Jr., G. Thomas, The martensite phases in 304 stainless steel, Metall. Trans. 1 (1970) 1577–1586. [50] J.F. Breedis, L. Kaufman, Formation of hcp and bcc phases in austenitic iron alloys, Metall. Trans. 2 (1971) 2359–2371. [51] L. Rémy, A. Pineau, Observation of stacked layers of twins and e martensite in a deformed austenitic stainless steel, Metall. Trans. 5 (1974) 963–965. [52] G. Olson, M. Cohen, A general mechanism of martensitic nucleation: Part i. General concepts and the fcc ? hcp transformation, Metall. Trans. A 7A (1976) 1897–1904. [53] L. Rémy, The interaction between slip and twinning systems and the influence of twinning on the mechanical behavior of fcc metals and alloys, Metall. Trans. A 12A (1981) 387–408. [54] P. Müllner, C. Solenthaler, P.J. Uggowitzer, M.O. Speidel, Brittle fracture in austenitic steel, Acta Metall. Mater. 42 (1994) 2211–2217. [55] P. Müllner, On the ductile to brittle transition in austenitic steel, Mater. Sci. Eng. A A234–236 (1997) 94–97. [56] S. Takaki, T. Furuya, Y. Tokunaga, Effect of si and al additions on the low temperature toughness and fracture mode of fe–27mn alloys, ISIJ Int. 30 (1990) 632–638. [57] J. Brooks, J. Williams, A. Thompson, Microstructural origin of the skeletal ferrite morphology of austenitic stainless steel welds, Metall. Trans. A 14A (1983) 1271–1281. [58] R. Ritchie, A. Thompson, On macroscopic and microscopic analyses for crack initiation and crack growth toughness in ductile alloys, Metall. Trans. A 16A (1985) 233–248. [59] A. Turnbull, Hydrogen diffusion and trapping in metals, in: R.P. Gangloff, B.P. Somerday (Eds.), Gaseous Hydrogen Embrittlement of Materials in Energy Technologies, Woodhead Publishing, Oxford, UK, 2012, pp. 89–128. [60] M.I. Luppo, A. Hazarabedian, J. Ovejero-Garcia, Effects of delta ferrite on hydrogen embrittlement of austenitic stainless steel welds, Corros. Sci. 41 (1999) 87–103. [61] J.P. Hirth, Effects of hydrogen on the properties of iron and steel, Metall. Trans. A 11A (1980) 861–890. [62] T.-P. Perng, C.J. Altstetter, Effects of deformation on hydrogen permeation in austenitic stainless steels, Acta Metall. 34 (1986) 1771–1781. [63] M.R. Louthan Jr., R.G. Derrick, Hydrogen transport in austenitic stainless steel, Corros. Sci. 15 (1975) 565–577. [64] M. Nagumo, Hydrogen related failure of steels – a new aspect, Mater. Sci. Technol. 20 (2004) 940–950. [65] T. Neeraj, R. Srinivasan, J. Li, Hydrogen embrittlement of ferritic steels: observations on deformation microstructure, nanoscale dimples and failure by nanovoiding, Acta Mater. 60 (2012) 5160–5171.
Contents lists available at ScienceDirect
Corrosion Science journal homepage: www.elsevier.com/locate/corsci
Effect of low temperature on hydrogen-assisted crack propagation in 304L/308L austenitic stainless steel fusion welds H.F. Jackson ⇑, C. San Marchi, D.K. Balch, B.P. Somerday Sandia National Laboratories, Livermore, CA 94550, USA
a r t i c l e
i n f o
Article history: Received 5 February 2013 Accepted 5 August 2013 Available online 11 August 2013 Keywords: A. Stainless steel B. SEM C. Hydrogen embrittlement
a b s t r a c t Effects of low temperature on hydrogen-assisted cracking in 304L/308L austenitic stainless steel welds were investigated using elastic–plastic fracture mechanics methods. Thermally precharged hydrogen (140 wppm) decreased fracture toughness and altered fracture mechanisms at 293 and 223 K relative to hydrogen-free welds. At 293 K, hydrogen increased planar deformation in austenite, and microcracking of d-ferrite governed crack paths. At 223 K, low temperature enabled hydrogen to exacerbate localized deformation, and microvoid formation, at austenite deformation band intersections near phase boundaries, dominated damage initiation; microcracking of ferrite did not contribute to crack growth. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Austenitic stainless steels are used extensively in the nuclear power, chemical processing, and oil and gas industries, in applications where they are susceptible to stress corrosion cracking (SCC) [1–3]. SCC begins with an electrochemical process, such as local anodic dissolution, however synergistic processes of anodic dissolution and cathodic hydrogen damage are thought to contribute to SCC in stainless steels [4,5]. Hydrogen damage is typically characterized by a significant loss of tensile ductility. In various alloy/ solution systems, interactions between dislocations at a crack tip and corrosion products, such as hydrogen, can play a key role in SCC. Detailed fractographic and microstructural studies support the premise that adsorbed or absorbed hydrogen promotes crack growth by localized plastic flow [3,6–10]. This suggests that effects of a corrosive environment and sustained tensile stress are made worse by the presence of hydrogen. Understanding the contribution of hydrogen to deformation and fracture processes is therefore central to developing a mechanistic understanding of SCC. Austenitic stainless steels are highly resistant to embrittlement promoted either by hydrogen or low temperature, retaining significant ductility and fracture toughness in these environments [4,11–15]. Austenitic stainless steel welds can be more susceptible than base materials to fracture, corrosion, and other degradation. The compositions of austenitic welds are typically selected to promote primary solidification as d-ferrite and solid-state transformation to austenite to suppress solidification cracking [16]. The as⇑ Corresponding author. Present address: Structural Integrity Associates, Inc., San Jose, CA 95138, USA. Tel.: +1 408 833 7201. E-mail address: [email protected] (H.F. Jackson). 0010-938X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.corsci.2013.08.004
welded microstructure usually retains several volume percent of the bcc ferrite phase at room temperature, resulting in a duplex austenitic/ferritic microstructure. Studies of duplex stainless steels emphasize their attractive mechanical and corrosion properties, intermediate between those of austenite and ferrite. High hydrogen fugacities at the metal surface promote hydrogen pickup, especially in applications with cathodic protection or sour gas or oil containing significant H2S [1,2]. For hydrogen-exposed austenitic welds and other duplex (two-phase) austenite/ferrite microstructures, the detrimental influence of ferrite on room-temperature ductility and fracture toughness has been noted in several studies [17–23]. Hydrogen transport is complicated by the lower solubility but higher diffusivity and permeability of hydrogen in the ferrite phase [24]. Austenitic steels are not as readily embrittled by hydrogen as ferritic steels [2]. Low temperature can induce a ductile– brittle transition in ferrite and localizes deformation in austenite, yet few studies are reported for hydrogen-assisted fracture of austenitic stainless steel welds at low temperature, particularly studies that quantify crack growth resistance by fracture mechanics methods. Mills [25] reviewed cryogenic fracture toughness of AISI 304 and 316 base metals and their welds at temperatures as low as 4 K, however, studies at less severe sub-ambient temperatures are needed. The mechanism of environment-assisted crack propagation is known to differ depending on the specific combination of metallurgical, environmental, and mechanical variables [5,26]. For example, features of hydrogen damage in austenitic and duplex stainless steels differ when they result from exposure to high pressure hydrogen gas versus cathodic charging in an aqueous solution; or upon tensile straining after precharging with hydrogen versus in situ straining in a hydrogen environment [1,27–30]. Austenitic al-
211
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
loys’ susceptibility to embrittlement has been correlated with a propensity for localized deformation, which can arise due to metallurgical factors such as low stacking fault energy (SFE) [31–33] or environmental factors such as low temperature [33–37] or hydrogen exposure [4,38,39]. Hydrogen-assisted crack growth is sensitive to the combination of metallurgical, environmental, and mechanical variables, hence it is important to understand effects of relevant combinations of material and environmental variables (absorbed hydrogen, low temperature, ferrite/austenite phase distribution) on fracture of austenitic welds. In the present manuscript, we have isolated effects of a high-purity gaseous hydrogen environment. However, hydrogen derived from aqueous sources (e.g. corrosion reactions at a stress-corrosion crack tip) is thought to affect alloys in a similar manner [4,14,38]. The present study expands on the results of a previous investigation of room-temperature hydrogen-assisted crack propagation in gas-tungsten arc (GTA) welds having a 304L base metal and 308L filler [21]. The objectives of the present study are to characterize hydrogen-assisted crack propagation of these welds at low temperature. Fracture mechanics specimens were thermally precharged in hydrogen gas, and fracture initiation toughness and crack growth resistance curves were measured at 223 K (50 °C). The effects of low temperature on hydrogen-assisted crack initiation and propagation mechanisms were assessed via electron microscopy of fracture surfaces and fracture profiles. 2. Experimental The processes of weld fabrication, hydrogen precharging, and fracture testing of welds at room temperature were detailed by Jackson et al. [21] and are summarized here. 2.1. 304L/308L Welds The weld, referred to as 304L/308L throughout this work, was fabricated from a round bar (64 mm diameter) of annealed 304L stainless steel having a longitudinal U-shaped groove. The groove was filled with 1.1 mm diameter 308L filler wire using 9 gas-tungsten arc (GTA) weld passes. Bulk elemental compositions of the bar, wire, and weld fusion zone are reported in Table 1. The volume fraction of ferrite in the weld fusion zone was estimated by magnetic measurement (Feritscope, Helmut-Fischer GmbH, Sindelfingen, Germany) at several distances from the weld root along the centerline. The as-welded microstructure was imaged using optical and scanning electron microscopy (SEM, JSM-840, JEOL, Tokyo, Japan). The elemental compositions of the ferrite and austenite phases within the weld were quantified by electron probe microanalysis (EPMA, JXA-8200, JEOL, Tokyo, Japan). Variation in tensile properties as a function of depth in the 304L/308L weld was characterized by tensile tests of subsized specimens which were extracted from the root, mid-height, and top of the weld. Upon going from the root to the top of the weld, yield strength decreases
from 420 MPa to 350 MPa, while strain at failure increases from 40% to 50% [21]. 2.2. Preparation of fracture specimens Disk-shaped compact-tension (CT) specimens, illustrated schematically relative to the weld in Fig. 1, had a width W (distance from the load-line to the back face of the specimen) of 40.6 mm, thickness B of 6.4 mm, and net thickness Bn (between the sidegrooves) of 4.8 mm, consistent with ASTM E1820 [40] and ASTM E1737 [41]. Specimens tested at room temperature in the noncharged condition had the same W but with a B of 12.7 mm and Bn of 9.9 mm. In order to locate the fatigue precrack tip in a similar microstructure in all specimens, the precrack was grown to the same distance into the weld fusion zone, about 1.5 mm. To accommodate variation in the distance from the notch tip to the root of the weld in the machined specimens, the final crack length-to-width ratio varied between 0.63–0.68. Precracks were grown in air along the radial direction of the welded bar along the weld centerline at 10 Hz under a load ratio of 0.1 and final maximum stress intensity factor Kmax of 30 MPa m1/2. Residual stresses in precracked specimens are minimized by conducting precracking in accordance with standard procedures in ASTM E1820 [ref]. Specimen dimensions conformed to allowable dimensions (with respect to specimen geometry) of the starter notch, final precrack, and amount of fatigue crack growth. The Kmax applied during precracking was kept well below the material fracture toughness that is measured during subsequent fracture testing. During the precracking procedure, the displacement is kept constant, hence the Kmax and DK range decrease progressively as the crack grows. Thermal precharging of CT specimens was in 99.9999% hydrogen gas at 138 MPa and 573 K for a minimum of 39 days and up to 61 days. The charging time and temperature were selected to achieve at the specimen midthickness a minimum hydrogen concentration of 90% of the equilibrium hydrogen concentration at the surface. After hydrogen precharging but before mechanical testing, specimens were stored at below 250 K to minimize hydrogen outgassing. The specimen was equilibrated at the test temperature, either 293 or 223 K, for at least 30 min prior to mechanical testing. The hydrogen concentration in the fusion zone was 140 wppm (0.8 at.%), as measured by inert gas fusion (Wah Chang, Albany, OR). This concentration is consistent with that predicted based on the thermal precharging parameters and the hydrogen solubility of 300-series alloys [42]. 2.3. Fracture mechanics testing Elastic–plastic fracture mechanics tests of the fatigue-precracked disk CT specimens were conducted according to ASTM E1820 [40]. Three or four replicate specimens were tested for each
Table 1 Elemental compositions (wt%) of base metal, filler, and weld fusion zone (balance Fe).
a
304L 308La 304L/308L weld b,c 304L/308L weldb,d a b c d e
Cr
Ni
Mn
Si
Mo
C
S
P
N
Cu
Nb
Creqe
Nieqe
19.85 20.5 21.27 20.75
10.73 10.3 10.23 10.19
1.6 1.56 1.69 1.65
0.57 0.5 0.51 0.51
0.17 <0.01 0.05 0.04
0.030 0.028 0.04 0.02
0.001 0.012 0.02 0.011
0.005 0.006 0.006 0.005
0.02 0.055 0.047 0.048
– 0.018 0.02 0.02
– <0.005 0.01 –
19.85 20.51 – –
12.18 12.38 – –
Per manufacturer certification. By emission spectroscopy. Analysis at mid-height of weld. Analysis at root of weld Nieq = Ni + 35C + 20N + 0.25Cu, and Creq = Cr + Mo + 0.7Nb per Kotecki and Siewert [43].
212
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
tiation of crack extension following blunting was identified as the second slope change in the DCPD vs. COD record, which is consistent with observations from interrupted fracture mechanics tests of stainless steels [22]. The fracture initiation toughness was identified from the intersection of the 0.2 mm offset blunting line with the R curve. 2.4. Post-testing characterization Fracture surfaces and fracture profiles parallel to and normal to the crack growth direction were characterized by SEM. Elemental analysis of compositional partitioning by energy dispersive X-ray spectroscopy (EDS) was used to infer phases in the weld fusion zone. Fig. 1. Optical image of the 304L/308L weld in cross-section, with the orientation and location of disk-shaped CT specimens illustrated schematically.
3. Results 3.1. As-welded microstructure
condition. Room-temperature tests were conducted in the ambient laboratory air. Tests at 223 K were conducted within an environmental chamber mounted in the load train of a servo-hydraulic test frame (MTS Corporation, Eden Prairie, MN). Evaporated liquid nitrogen was used as a coolant, with a resistive heater and temperature controller maintaining the test temperature of 223 K. Testing was conducted at a constant (monotonic) displacement rate of 0.4 mm min1. The crack-opening displacement (COD) at the load-line was measured with a clip gauge, and crack extension was monitored continuously using the direct current potential drop (DCPD) method of ASTM E1737 [41]. The test was terminated manually after a greater than 10% load drop from peak load and an increase in DCPD by at least 15% from its initial value, or when load-line displacement exceeded the measurement capacity of the clip gauge (4–5 mm). Following the test, CT specimens were heat-tinted at 623 K for 60 min to mark the extent of crack growth during the test. The specimens were then broken apart in liquid nitrogen, and the final crack length and precrack length were measured from the fracture surfaces. The calculated crack length based on DCPD data was linearly scaled so that the final calculated and measured crack lengths were equal. Prior to correction, the discrepancy between the final measured and calculated crack lengths was less than 1.5%. The J-integral crack growth resistance (J–R) curve was constructed from the load vs. COD and DCPD vs. COD records. The initial J vs. crack growth increment (Da) relationship was assumed to be a linear crack blunting response following ASTM E1737. The ini-
The microstructure of the fusion zone consisted of primary dferrite in a matrix of austenite (c), as seen in Fig. 2. The ferrite number (FN) as measured at distances of 2, 6, 10, and 14 mm from the weld root along the centerline was 9.4, 10.1, 10.7, and 10.1, respectively. These values represent the mean of five measurements per location. The FN approximates the volume percent ferrite, and the measurements are consistent with values of 8–10 predicted from the WRC-1992 diagram [43], based on the calculated Creq and Nieq for the weld metal (Table 1). Table 2 shows the elemental compositions of the austenite and ferrite phases within the weld, as measured by EPMA spot analyses at two locations: at mid-height of the weld and near the weld root. Table 2 also compares the compositions of the individual phases to the average composition of the weld fusion zone, as measured by emission spectroscopy in the same locations. In the fusion zone, the average composition by emission spectroscopy closely matched that of the original 308L filler wire. Weld austenite had approximately the same wt% Ni and Cr as in the 304L and 308L starting materials, while the approximately 8–10 vol% of d-ferrite was depleted in Ni and enriched in Cr. 3.2. Fracture toughness Compared to the non-charged condition, precharged hydrogen markedly decreased the fracture initiation toughness of 304L/ 308L welds at both 293 and 223 K, as shown by crack growth resis-
Fig. 2. (a) Optical and (b) backscattered electron images of the fusion zone in the 304L/308L weld near the fusion zone/base metal interface at the weld centerline. The solidification direction is nominally from bottom to top of the image. The microstructure contains skeletal ferrite (d) in an austenite (c) matrix.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221 Table 2 Elemental compositions (wt%) of ferrite and austenite phases within the weld. Phase
a b c
Location
Cr
Ni
Weld ferrite
a
Weld mid-height Weld roota
28.17 27.86
4.41 4.38
Weld austenite
Weld mid-heighta Weld roota
20.02 19.73
10.24 10.40
Weld average
Weld mid-heightb Weld rootb 308L filler wirec
21.27 20.75 20.5
10.23 10.19 10.3
Base metal
Outside fusion zonea 304L annealed barc
18.44 19.85
10.21 10.73
EPMA, mean of three measurements for ferrite and nine for austenite. Emission spectroscopy. Per manufacturer certification.
tance (J–R) curves (Fig. 3) and fracture initiation toughness (Jmax or JIH) values (Table 3). Also reported in Table 3 are the corresponding stress-intensity factor KJ values, determined by:
KJ ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffi EJ 1 m2
ð1Þ
where E is Young’s modulus, J is fracture initiation toughness, and m is Poisson’s ratio [40]. The J–R curve slope over the first 0.5 mm of crack extension immediately following blunting is reported as crack growth resistance dJ/dDa.
213
Welds not exposed to hydrogen and tested at both 293 and 223 K exhibited significant crack tip blunting, and tests were terminated manually before the onset of noticeable crack extension, as confirmed by post-test analysis. Crack extension in non-charged specimens was insufficient to permit calculation of JIC, and instead we report as Jmax the lesser of two values, either (1) the maximum J value recorded during the test or (2) the maximum J-integral capacity of the specimen as permitted by ASTM E1820 [40], i.e. the lesser of b0rY/10 or BrY/10, where b0 is the original uncracked ligament, B is the original specimen thickness, and rY is the material flow stress. These Jmax values represent lower-bound fracture initiation toughness measurements. It is important to note that the difference in Jmax between noncharged welds tested at 293 and 223 K (Table 3) was primarily due to the difference in specimen thickness and the corresponding dependence of the maximum J-integral capacity on this dimension. Non-charged specimens tested at 223 K were thinner than those tested at 293 K, while hydrogen-precharged specimens were all of the same thickness. Regardless, crack blunting and a lack of noticeable crack extension were observed for non-charged specimens tested at both 293 and 223 K. The measured fracture initiation toughness of hydrogen-precharged welds represents a reduction of greater than 59% compared to the lower-bound fracture initiation toughness of noncharged welds. Hydrogen-precharged specimens met the dimensional requirements in ASTM E1820 [40] to define conditional JQ as JIC. To reflect the hydrogen-precharged condition, fracture initiation toughness is denoted JIH and stress-intensity factor is KJIH. Fracture initiation toughness with internal hydrogen is essentially the same at 293 and 223 K. At the initiation of crack growth (Fig. 3b), the J–R curves for both temperatures are indistinguishable. However, at longer crack extensions (Fig. 3a), welds tested at the lower temperature exhibit lower crack growth resistance dJ/dDa. This divergence in J–R curve slope occurs at a crack extension of about 0.8–1 mm from the precrack tip. This distance was found to correspond to a change in orientation and morphology of austenite grains and d-ferrite dendrites at a depth in the fusion zone where two weld passes overlapped (Fig. 4). Crack paths were tortuous in general, and any crack deflection or differences in fracture appearance due to microstructural differences in the weld pass overlap region were not visibly distinguished from that in other regions. 3.3. Fractography
Fig. 3. Crack growth resistance (J–R) curves for 304L/308L GTA welds with and without precharged hydrogen, tested at 293 K and 223 K. Up to four J–R curves per condition are plotted. (a) At longer crack extensions, welds tested at 293 K have greater crack growth resistance than at 223 K. (b) Immediately following crack initiation, J–R curves were indistinguishable at 293 versus 223 K, as were JIH values.
Fracture features are significantly different for specimens fractured with and without hydrogen; for specimens fractured in hydrogen, a different distribution of fracture features is seen when test temperature is decreased from 293 to 223 K. Effects of hydrogen and test temperature on fracture features are compared in Fig. 5 and in higher magnification in Fig. 6. Non-charged welds at both 293 and 223 K exhibited significant crack blunting followed by a slight amount of ductile crack extension. These fracture surfaces featured only equiaxed and U-shaped dimples (Fig. 5a and b), indicating microvoid coalescence (MVC). Under these conditions, microvoids nucleate at fine spherical precipitates. Hydrogen exposure altered the fracture mode of austenitic welds (Fig. 5c and d). In hydrogen-precharged welds fractured at 293 K, the fracture surface was dominated by features which were elongated and aligned parallel to the crack propagation direction, which was also the weld solidification direction (Figs. 5c, 6a and b). The dendritic morphology of these features (denoted by D in Figs. 5c and 6a) resembled the underlying d-ferrite microstructure (Fig. 2), with smooth, relatively flat features interspersed with tear ridges (denoted by T in Fig. 6a), i.e. vertical steps joining two flat
214
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
Table 3 Fracture initiation and crack growth toughness of 304L/308L GTA welds Non-chargeda,b
Hydrogen-precharged 2
a b c
0.5
JIH (kJ m )
KJIH (MPa m
223 K
144 123 109
175 161 152
171 175 181
Mean
125
163
176
293 K
110 123 132 111
153 161 167 154
209 190 196 242
Mean
119
159
209
)
dJ/dDa
c
2
1
(kJ m mm
)
2
Jmax (kJ m
)
dJ/dAac (kJ m2 mm-1) 0.5
KJ (MPa m
)
dJ/dDa
304
254
965
625
364
965
c
(kJ m2 mm1)
Mean of 3 replicate tests 6 of non-charged specime. B of non-charged specimens was 6.35 mm for tests at 50 °C and 12.7 mm for tests at 20 °C. Slope of J–R curve, measured over 0.5 mm of crack extension immediately following blunting.
Fig. 4. A change in microstructural orientation and morphology was observed where two weld passes overlap, about 0.8–1 mm from the precrack tip and 2.3–2.5 mm from the base metal/weld interface. Backscattered electron image.
Fig. 5. Fracture surfaces of hydrogen precharged and non-charged welds, tested at 223 or 293 K. Without hydrogen, at (a) 293 K or (b) 223 K, fracture surfaces had only dimpled rupture. With hydrogen at (c) 293 K, the fracture surface was dominated by elongated features with a dendritic morphology (D) and aligned with the crack growth direction, while at (d) 223 K, the dominant features were elongated (E) and equiaxed (Q) dimples, with few dendritic-shaped features. Crack growth direction is from bottom to top in all images.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
215
Fig. 6. Fracture features at matching locations on both halves of hydrogen-precharged fracture surfaces. At 293 K (a and b), the dominant features were dendritic (D) with tear ridges (T) between dendritic features. At 223 K (c and d), few regions had dendritic features (D). The dominant features (e and f) were planar arrays of elongated (E) dimples and patches of equiaxed (Q) dimples. Crack growth direction is from bottom to top in all images.
areas of dimpled rupture, with steps also exhibiting dimples. Most of the fracture surface exhibited features related to the underlying dendritic structure of ferrite and the inter-dendritic regions of austenite. In hydrogen-precharged welds fractured at 223 K (Figs. 5d, 6c, and d), considerably fewer regions of the fracture surface showed evidence of the dendritic microstructure (denoted by D in Fig. 6c and d). Most regions showed either elongated or equiaxed dimples (denoted by E and Q respectively in Figs. 5d and 6e). Elongated dimples occurred in nominally parallel and coplanar arrays (Fig. 6e and f). Parallel arrays of elongated dimples formed steps on the fracture surface (Figs. 5d and 6e). Matching locations on both fracture surfaces of hydrogen-exposed welds (Fig. 6) were analyzed by EDS to provide a basis for inferring the phase that was present at the surface based on the approximate Ni and Cr concentrations, which were consistent with the values for austenite or ferrite in Table 2. EDS analysis alone may be inconclusive if the ferrite dimensions are very thin and ferrite evades detection by the EDS probe. By comparing elemental analyses and fracture features on both halves of the fracture surface, the crack path could be identified as fracture within ferrite, at the austenite/ferrite phase boundary, or within austenite. All
three types of crack path were observed in various locations at both temperatures. In the present work, no distinction was made between cracks at c/c boundaries and cracks within a c grain, although both may occur. For this reason, it cannot be conclusively stated that d was absent at a given fracture location and that the crack path was entirely contained within the c phase. Planar arrays of elongated dimples, seen only at 223 K, were associated with fracture within the austenite phase; such crack growth could be either transgranular or at c/c interfaces. Previously [20,21], it was established that for hydrogen-precharged austenitic welds at 293 K, tear ridges and equiaxed dimples were associated with fracture within austenite; the flat areas containing dendritic features were associated either with fracture at the austenite/ferrite phase boundary or within ferrite. At both temperatures, fracture surfaces exhibited significant out-of-plane crack growth and tortuous crack topographies. With hydrogen at both 293 and 223 K, crack extension evolved as multiple discontinuous microcracks that joined up to form the main crack path. At 293 K, the microcracks formed on planes parallel to the Mode I crack growth plane, i.e. normal to the tensile axis [21]. At 223 K, microcracks formed primarily on planes inclined
216
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
to the Mode I crack growth plane, as confirmed from stereo pair fractographs. 3.4. Fracture profile metallography of hydrogen-exposed welds Fracture profiles parallel and transverse to the crack growth direction in hydrogen-exposed welds were imaged by backscattered electron microscopy (Figs. 7 and 8). Atomic number contrast distinguishes d-ferrite as the dark grey dendritic phase and austenite as the light grey matrix. Localized deformation bands appear as bright parallel streaks in the light grey austenite matrix. Fracture profiles (Fig. 7) suggested that, although microvoids/ microcracks initiated near ferrite at both 223 and 293 K in hydrogen, crack growth occurred by different mechanisms at the two temperatures. At 293 K, microcracks were predominantly associated with dferrite (Fig. 7a and b). Microcracks nucleated either within the ferrite phase or at the austenite/ferrite phase boundary (Fig. 7a) and propagated parallel to ferrite dendrites (denoted by d in Fig. 8a). These microcracks often created coarse cracks (Fig. 7b) parallel to the main crack, which were located up to several hundred micrometers above or below the fracture surface (Fig. 8c). At 223 K, there was no evidence for large microcracks away from the fracture surface. Fine microvoids 1 lm or less in diameter (Fig. 7d and f) were the main type of microstructural damage ob-
served. With an equiaxed cross-section, the three-dimensional shape of these microvoids could be either spherical or tubular. Some fine porosity was randomly distributed in the as-welded microstructure, however the microvoids that were associated with damage accumulation and fracture had a distinctive arrangement. Specifically, these microvoids tended to be aligned and to coincide with the intersections between deformation bands (Fig. 7f). The predominant microvoid nucleation site was adjacent to d-ferrite (Fig. 7c–f), while a smaller fraction formed elsewhere in austenite grains. Microcracks evolved via the nucleation and coalescence of closely-spaced microvoids (Fig. 7c, d, and f). Fracture profiles reveal a second key difference between fracture at 293 K and at 223 K: the macroscopic crack path. At both 293 and 223 K, fractures had significant out-of-plane crack growth and tortuous crack topographies (Fig. 5c and d). This is reflected in cross-sections as macroscopic steps (denoted by S in Fig. 8a,c,d). At 293 K, steps corresponded to microcracks growing along dferrite on various planes ahead of the crack tip that were nominally parallel to each other and perpendicular to the tensile axis (Fig. 8a and c). The main crack grew when these microcracks joined along planes parallel to the tensile axis, facilitated by intense shear in the ligaments between them (denoted by I in Fig. 8a). This out-of-plane crack growth corresponds to that seen on fracture surfaces (Fig. 5c).
Fig. 7. Fracture profiles show different mechanisms of microvoid nucleation at 293 versus 223 K. At 293 K (a and b), microcracks originate at and propagate along d-ferrite. At 223 K (c–f), microvoids nucleate where localized deformation bands intersect other deformation bands, and microvoids are concentrated near phase boundaries. Microcracks form via the coalescence of multiple microvoids (c, d, and f). Crack growth direction is from left to right in all images except for a, in which crack growth is normal to the page.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
217
Fig. 8. Fracture profiles reveal that at both temperatures, the crack path is marked by macroscopic steps (S). At 293 K (a and c), cracks propagate along ferrite dendrites (d) but can link up through intense shear (I) in the ligaments between them. At 223 K (b and d), crack paths are primarily within austenite and do not clearly follow ferrite.
At 223 K, macroscopic steps and out-of-plane crack growth were also observed. However, microcracks were observed to form on planes inclined to the tensile axis (Fig. 6c–f). In cross-section, the crack path does not clearly follow the ferrite phase (Fig. 8b and d) and is primarily within the austenite phase.
4. Discussion As shown in previous studies [20,21], one primary role of hydrogen in degrading fracture resistance and altering fracture mechanisms in austenitic stainless steel welds at room temperature is enhancing planar deformation in austenite. This phenomenon produces stress concentrations where localized deformation bands intersect d-ferrite. These stress concentrations facilitate microcracking of the elongated d-ferrite grains, which are aligned with the direction of crack growth. This dominant role of ferrite in the initiation and propagation of microcracks in austenitic welds at room temperature is illustrated in Figs. 6a, 6b, 7a, andb and documented in several studies [17–21]. Extending the study on 304L/308L welds at room temperature [21], we found that low-temperature (223 K) exposure had no effect on the fracture initiation toughness of hydrogen-precharged welds (Table 3), although the fracture appearance was clearly changed (Fig. 5c and d). In contrast to room-temperature fracture, at 223 K fracture surfaces and fracture profiles showed a diminished role of d-ferrite in facilitating crack propagation (Figs. 6 and 7), relative to room temperature. We propose that low temperature alters the competition between potential fracture initiation modes in hydrogen-precharged austenitic welds. The two main effects of low temperature are that (1) low temperature localizes plastic deformation in austenite, rendering this phase more susceptible to hydrogen-enhanced localized plasticity, and (2) low temperature increases the hydrogen concentration in austenite near phase boundaries through its effects on the thermodynamics of hydrogen trapping. In the following sections, we discuss effects of low temperatures on plastic
deformation and hydrogen distribution in welds and how these changes govern the fracture mode. 4.1. Sequence of crack initiation and propagation in hydrogenprecharged welds at 223 K In the presence of hydrogen, low temperature (223 K) produced a significantly different fracture surface than seen at room temperature (293 K). At 293 K, a mixture of ferrite cleavage and phase boundary separation was observed. Microcracks nucleated and propagated preferentially at d-ferrite (Figs. 7a and b), resulting in elongated, dendritic features on fracture surfaces (Figs. 5c, 6a, and 6b) that resemble the underlying ferrite morphology (Fig. 2). These microcracks formed in the ferrite phase itself or at austenite/ ferrite interfaces, facilitated by high stress concentrations at these sites. At 223 K, neither ferrite cleavage nor phase boundary separation is an important fracture mechanism, based on the observed fracture surface features (Fig. 6c–f). Rather, the dimpled surface indicates that fracture was caused by MVC within the austenite phase. Microcracks originated as microvoids at intersections between localized deformation bands in austenite (Fig. 7c–f). Because deformation bands are parallel, intersections between deformation bands occur in rows, as do the microvoids that form at these intersections. The non-equiaxed growth of these aligned microvoids produced coplanar arrays of elongated fracture dimples (Figs. 5d, 6e, and 6f). In limited locations, the fracture appearance at 223 K resembled the dendritic features associated with microcracks in d-ferrite or at austenite/ferrite boundaries at 293 K (denoted by D in Fig. 6c and d). An important feature of the proposed low-temperature fracture mechanism is that, although deformation band intersections occurred throughout the austenite bulk, microvoids were not randomly distributed but were concentrated near austenite/ferrite phase boundaries (Fig. 7c–f). We can infer that in welds or two-phase microstructures, two conditions promote microvoid nucleation in austenite: (1) deformation bands inter-
218
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
sect and (2) the intersection is in close proximity to an austenite/ferrite phase boundary. The role of hydrogen in facilitating void formation near phase boundaries will be discussed in the final section. Despite the differences in fracture appearance, the fracture process at 223 K had a similar sequence of steps compared to the fracture process at 293 K [20,21]. First, planar deformation in austenite developed as localized bands. The formation of these deformation bands is facilitated by the hydrogen enhanced localized plasticity (HELP) mechanism [44–46]. In austenite, HELP results in greater slip planarity [4,38,39], exacerbating the effects of metallurgical factors predisposing an alloy to planar slip, such as low stacking fault energy (SFE) [31–33]. In metastable 300-series alloys such as 304L and 308L, SFE decreases as nickel content decreases [47,48]. It follows that the effects of low SFE, low temperature, and hydrogen would produce more severely localized slip than any factor acting independently. As slip becomes more localized, other planar deformation modes such as twinning and e-martensite become activated at lower strains [33,35,49–53]. The next step in the process was the impingement of localized deformation bands on microstructural obstacles, thus producing stress concentrations. Austenitic stainless steels contain various microstructural features that can act as obstacles to slip or twinning, including grain boundaries, secondary phases, and annealing twin boundaries. At 293 K, the d-ferrite phase could serve as a barrier to propagation of planar deformation bands, and the resulting stress concentration facilitated microcrack initiation either as cleavage of ferrite or decohesion at the austenite/ferrite phase boundary [21]. At 223 K, the resulting microstructural damage was instead a series of discrete microvoids at deformation band intersections near phase boundaries. Hence, low temperature changed the dominant mode of damage accumulation, leading to a different mode of microcrack initiation relative to that at room temperature. The final step was crack propagation via the growth and coalescence of microcracks. At 293 K, microcracks formed from cleavage in ferrite or decohesion of austenite/ferrite interfaces, then these microcracks linked by intense shear in the remaining ligaments (Figs. 7a, 7b, 8a, and 8c). At 223 K, microvoids nucleated predominantly adjacent to weld ferrite, while ferrite cleavage and austenite/ferrite interface decohesion were infrequently observed. Instead, under low-temperature conditions, decohesion was not a dominant fracture mode; an alternative mode of crack propagation intervened, and multiple microvoid nucleation and coalescence events were required for microcracks to form (Fig. 7c–f). Similar to the process at room temperature, individual microcracks ahead of the crack tip at low temperature linked up via shear in the ligaments between them, producing coarse steps on the fracture surfaces (Fig. 8b and d). 4.2. Effect of temperature on crack initiation: localized deformation In this study, temperature governed the mode by which hydrogen-assisted cracks initiated and propagated. At both 293 and 223 K, planar deformation and microstructural obstacles played a prominent role in the initiation of hydrogen-assisted cracking in austenitic stainless steel welds. We propose that low temperature altered fracture mechanisms in hydrogen-precharged welds in part by conditioning the austenite for more severe planar slip that is further exacerbated by hydrogen. The planar deformation bands that evolve increase the severity of stress concentrations at intersecting deformation bands near the austenite/ferrite phase boundary, perhaps elevating stress concentrations at these sites relative to the stress concentrations at intersections of deformation bands with d-ferrite.
Consequently, void nucleation in austenite near phase boundaries is favored over microcracking in ferrite or at austenite/ferrite interfaces. Low temperature exacerbates localized deformation in several ways. First, low temperature increases the tendency toward planar slip in austenite by decreasing SFE [33–37]. More intense planar slip on [1] planes in fcc austenite gives rise to deformation twins, bcc a0 -martensite, and hcp e-martensite at lower strains [33,35,49–53]. These deformation modes promote the confinement of deformation in discrete bands, leading to more pronounced strain incompatibility and enhanced stress concentration at obstacles. After deforming 301LN stainless steels in tension at temperatures between 233 and 353 K, Talonen and Hanninen [33] observed that, as temperature decreased, a greater volume fraction of a0 martensite formed, and at lower levels of strain. They proposed that low SFE at low temperature facilitates the martensitic transformation. In contrast, at higher temperature, the overlapping of stacking faults is less regular, hindering the formation of martensite nuclei. Deformation at lower temperature produces more twins and martensite – the results of planar deformation modes – than at room temperature. Second, low temperature could also change the types of deformation products that preferentially form. In addition to decreasing SFE, low temperature could alter the free-energy differences between the fcc, bcc, and hcp phases, and thus the thermodynamic driving force for transformation to the hcp or bcc martensite phases. During an investigation of tensile deformation of austenitic steels containing 16-18 wt% Cr and 11–13 wt% Ni at temperatures between 77 and 523 K, Lecroisey and Pineau [35] observed deformation bands containing both deformation twins and e-martensite, with a higher proportion of twins at higher deformation temperatures. They proposed that the tendency to nucleate austenitic twins versus e-martensite at a given temperature depends on the relative values of extrinsic and intrinsic stacking fault energies as well as the volume contraction due to the fcc-to-hcp transformation. Temperature probably determines the relative proportions of hcp or bcc martensite and fcc twins that compose deformation bands. Although insufficient data exists to evaluate effects of deformation band makeup on the severity of stress concentrations, it is reasonable to presume that deformation products such as twins and e-martensite confine deformation in discrete bands, promoting strain incompatibility and stress concentration at obstacles [21]. Third, more severe planar deformation has been correlated with physically thinner deformation twins and e-martensite plates and, consequently, higher stress concentrations where they impinge on obstacles. Based on an investigation of uniaxial tensile fracture of nitrogen-bearing austenitic steels at temperatures between 4 and 298 K, Mullner and coworkers [54,55] attributed fracture initiation to stress concentrations at deformation twin intersections. Lower temperature decreased the thickness of twins and the dislocation density, and they asserted that such conditions rendered twins effective obstacles against propagation of impinging twins, creating stress concentrations high enough to initiate microcracks at twin intersections. Based on fractographs and fracture profiles of low-SFE, highMn austenitic steels fractured in Charpy tests at 77 K, Takaki et al. [56] inferred that intersections between e-martensite plates and the associated stress concentrations led to the formation of parallel tunnel-like voids at these intersections. They invoked this mechanism to explain the dimpled planes and steplike ridges they observed on fracture surfaces. It is plausible that propagation of both twins and e-martensite plates could be blocked by the austenite/ferrite phase boundary as well. Thinner deformation twins and e-martensite plates produced at lower
219
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
temperatures would lead to more severe stress concentrations than at room temperature. While the preceding discussion supports the notion that low temperature conditions austenitic stainless steels for more localized deformation, which can then be exacerbated by elevated hydrogen, leading to elevated stress concentrations at intersecting deformation bands, one detail still requiring attention is the observation that at low temperature, microvoids form predominantly near austenite/ferrite phase boundaries. One possible explanation for this observation is that deformation is disproportionately localized near austenite/ferrite phase boundaries, where Ni depletion locally decreases SFE. Brooks et al. [18,57] attributed the Ni concentration profile observed for AISI 309 GTA welds to partitioning during the diffusion-controlled solid-state transformation from ferrite to austenite following primary ferrite solidification. The authors attributed Ni enrichment at the centers of austenite grains (i.e. the boundaries of solidification cells) to secondary austenite solidification due to Ni enrichment of the liquid. EDS concentration profiles of the present weld suggested that Ni concentration was 2.5–3 wt% lower near the phase boundaries than in the centers of austenite grains (i.e. the boundaries of solidification cells), where the 12– 13 wt% Ni exceeded the nominal Ni concentration in weld austenite, about 10.2 wt% (Table 2). Such concentration gradients could bias localized deformation toward the austenite/ferrite phase boundaries. In summary, low temperature enhances the formation of planar deformation bands in austenite, especially near austenite/ferrite phase boundaries. One potential role of hydrogen in linking deformation and fracture is that hydrogen magnifies planar slip. This intensified planar slip or the deformation products that it stimulates (e.g. twinning, e-martensite) leads to stress concentrations at obstacles such as intersecting deformation bands. The observation that microvoid nucleation in austenite near phase boundaries out-competes microcrack formation in ferrite could be due in part to increased stress concentrations at deformation band intersections near phase boundaries relative to those at intersections between deformation bands and d-ferrite. The difference in damage initiation modes gave rise to the significant difference in fracture surface appearance. While the fracture modes at 293 and 223 K differed significantly, temperature had no notable effect on fracture initiation toughness (Table 3). We propose that the similar JIH values at 293 and 223 K are a coincidence. In other words, it is coincidental that the dominant fracture processes at these two temperatures are characterized by a similar magnitude of the local fracture criterion, e.g., a critical local strain. For example, at cryogenic temperatures, crack nucleation would likely occur by cleavage of d-ferrite due to its ductile–brittle transition [25]. Other fracture modes might dominate at different sub-ambient temperatures, altering crack initiation toughness. On the other hand, dJ/da, a measure of crack growth resistance, was slightly lower at 223 K than at 293 K. According to Ritchie and Thompson [58], metallurgical factors affecting the local fracture strain have a greater effect on crack growth than on crack initiation, owing to a weaker strain singularity ahead of a stable growing crack versus a stationary crack. Metallurgical factors dictating ductility include local microstructure and the degree of planar deformation. In the present welds, the J–R curves for tests at 293 and 223 K coincided near initiation but diverged after cracks extended into a region of overlap between weld passes and a change in microstructural orientation and morphology (Fig. 4). This microstructural change and the increase in planar deformation at 223 K had a stronger influence on dJ/da at 223 K than at 293 K.
4.3. Effect of temperature on crack initiation: local hydrogen distribution We propose that low temperature alters the thermodynamics of hydrogen trapping in austenitic stainless steel welds in a way that redistributes hydrogen from the interiors of d-ferrite grains to phase boundary trap sites. At 223 K, this could contribute to hydrogen-assisted fracture initiation near phase boundaries rather than within d-ferrite. The total concentration of dissolved hydrogen consists of hydrogen in interstitial lattice sites and hydrogen trapped at microstructural features such as grain boundaries, phase boundaries, dislocations, and impurities [59]. Luppo et al. [60] provided evidence that austenite/ferrite phase boundaries in welds serve as hydrogen trap sites. According to Hirth [61], trap sites become more densely populated as temperature decreases, as expressed in following equation:
hT EB ¼ hL exp½ 1 hT kT
ð2Þ
where hT is the trap site occupancy, hL is the lattice hydrogen concentration (in this formulation, hL 1), Eb is the binding energy of hydrogen in the trap site, k is Boltzmann’s constant, and T is temperature. Because a hydrogen-precharged specimen can be considered a closed system with a fixed total hydrogen concentration, the hydrogen demanded by trap sites at lower temperature must be supplied by lattice sites. Conversely, when temperature increases, trap occupancy decreases, with hydrogen migrating into lattice sites. From a kinetic standpoint, the d-ferrite phase is more likely than austenite to supply hydrogen to trap sites at phase boundaries. Based on the values recommended by Perng and Altstetter [62] for ferritic stainless steel 29Cr–4Mo–2Ni and those recommended by San Marchi et al. [42] for 300-series austenitic stainless steels, the diffusivity of hydrogen in ferrite and austenite were determined at 293 and 223 K and are summarized in Table 4. Hydrogen diffusivity is four to six orders of magnitude faster in bcc ferrite than in fcc austenite at the test temperatures. Even though the lattice solubility of hydrogen in austenite is about 100 times that in ferrite [63], and thus the concentrations will differ between the two phases, the overall diffusional flux will still be greater in d-ferrite. The present analysis demonstrates that hydrogen diffusivity in d-ferrite is rapid enough to satisfy the increased demand for hydrogen at traps as induced by low temperature. The hydrogen diffusion distance x varies with diffusivity D and time t according to Eq. (3), derived from the solution to the transient diffusion equation:
pffiffiffiffiffiffi x 2 Dt
ð3Þ
Specimens were exposed to the test temperature for at least 30 min and for up to several hours prior to mechanical testing. The hydrogen diffusion distance in d-ferrite during a typical expoTable 4 Hydrogen diffusivity D and diffusion distance x in ferrite and austenite phases at 293 and 223 K a
(m2 s1)
x (lm)
T (K)
D
t = 30 min
t=3h
Ferrite
223 293
1 1013 8 1012
30 200
60 600
Austenite
223 293
2 1019 2 1016
0.04 1
0.1 3
a Based on the values recommended by Perng and Altstetter [62] and San Marchi et al. [42].
220
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221
sure (Table 4) is significantly greater than the maximum required migration distance, i.e. from the center of a ferrite grain to trap sites at phase boundaries, about 0.5 lm. At low temperatures, hydrogen migration from lattice sites in dferrite to traps at phase boundaries both depletes hydrogen from ferrite grains and enriches phase boundaries with hydrogen. This tilts the competition away from microcrack formation in ferrite and toward microvoid nucleation near phase boundaries. As a result of this hydrogen enrichment at phase boundaries that is induced by low temperature exposure, we propose that hydrogen has two roles in concentrating damage near phase boundaries. First, hydrogen in concert with low temperature localizes deformation and increases stress concentrations near phase boundaries, consistent with the concepts described in the previous section. Second, hydrogen could have a direct role in facilitating microvoid nucleation near phase boundaries through several possible mechanisms. Vacancies are generated by intense localized plastic deformation and stabilized by hydrogen, according to the model of hydrogen-stabilized vacancy formation proposed by Nagumo [64]. Building upon this model, Neeraj et al. [65] propose a nanovoid coalescence (NVC) micromechanism of hydrogen embrittlement for ferritic steels. In the NVC mechanism, nanovoids nucleate at a critical excess vacancy concentration. They proposed that NVC gave rise to the nanoscale (10–20 nm) dimples they observed on quasi-brittle facets on fracture surfaces. In the previous section, we linked the formation of microvoids in austenite to intersections between planar deformation bands that may consist of e-martensite plates [56] and deformation twins [54,55]. It is plausible that hydrogen could directly promote fracture by encouraging void formation at these intersections, in addition to hydrogen’s indirect role in fracture, i.e. localizing deformation. In the context of the present work, we propose that hydrogen enrichment at phase boundaries, particularly at low temperature, makes near-phase-boundary austenite the preferred site for damage nucleation. At low temperature, this damage proceeds in a manner consistent with the aforementioned mechanisms, in which hydrogen stabilizes the vacancies that eventually form microvoids or hydrogen facilitates microvoid nucleation at the intersection of localized deformation bands consisting of e-martensite and deformation twins. At room temperature, the damage is predominantly in the form of microcracks in ferrite or at phase boundaries. 5. Conclusions Elastic–plastic fracture mechanics tests were conducted to evaluate effects of low temperature on hydrogen-assisted crack propagation in austenitic stainless steel welds fabricated from AISI 304L base metal and 308L filler metal and which had been thermally precharged with hydrogen gas: 1. Hydrogen degrades fracture resistance by >59% relative to the lower-bound fracture initiation toughness of non-charged welds. At room temperature (293 K), hydrogen alters the dominant mode of damage accumulation in austenitic welds, thereby altering fracture mechanisms by (1) enhancing localized planar deformation in austenite and (2) facilitating microcracking of d-ferrite or its interfaces. The dendritic microstructure of weld ferrite, which is aligned with the direction of crack growth, governed the crack path. 2. Low temperature (223 K) exposure altered fracture mechanisms in hydrogen-precharged welds. At 223 K, fracture initiation was dominated by microvoid formation at deformation band intersections in austenite near phase boundaries, while
microcracking of ferrite and its interfaces did not play a significant role in crack growth. Low temperature was presumed to (1) lower SFE and condition the austenite for more localized deformation, which was further exacerbated by hydrogen, and (2) enrich phase boundaries with hydrogen by increasing the occupancy of hydrogen at trap sites relative to lattice interstitial sites in ferrite. The more localized deformation in austenite, particularly near phase boundaries, could lead to more severe stress concentrations at deformation band intersections relative to those at intersections of deformation bands with ferrite. 3. Additionally, we propose that elevated hydrogen concentration near phase boundaries may directly promote localized void nucleation in austenite, consistent with the mechanisms of hydrogen-stabilized vacancies proposed by Nagumo [64] and vacancy-induced nanovoid coalescence (NVC) proposed by Neeraj et al. [65]. Alternately, hydrogen facilitated microvoid formation at intersections between localized deformation bands, consistent with literature reports of stress concentrations and microvoid nucleation at intersections between e-martensite plates [56] and deformation twins [54,55]. 4. Exposure to 223 K had no effect on fracture initiation toughness of precharged welds, despite the difference in the dominant mode of damage accumulation. It is likely coincidental that the dominant fracture modes at the two test temperatures (microvoid formation at intersecting deformation bands at 223 K vs. microcracking in ferrite or its interfaces at 293 K) had similar local fracture criteria, resulting in similar fracture initiation toughness values.
Acknowledgments The authors are grateful to J. Campbell for hydrogen pressure systems support, A. Gardea for metallographic preparation, and J. Chames and R. Nishimoto for SEM imaging. Sandia is a multiprogram laboratory operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the US Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
References [1] A.A. El-Yazgi, D. Hardie, The embrittlement of a duplex stainless steel by hydrogen in a variety of environments, Corros. Sci. 38 (1996) 735–744. [2] W. Zheng, D. Hardie, The effect of hydrogen on the fracture of a commercial duplex stainless steel, Corros. Sci. 32 (1991) 23–36. [3] Y. Mine, T. Kimoto, Hydrogen uptake in austenitic stainless steels by exposure to gaseous hydrogen and its effect on tensile deformation, Corros. Sci. 53 (2011) 2619–2629. [4] A.W. Thompson, I.M. Bernstein, The role of metallurgical variables in hydrogen-assisted environmental fracture, in: M.G. Fontana, R.W. Staehle (Eds.), Advances in Corrosion Science and Technology, Plenum Press, New York, NY, 1980, pp. 53–175. [5] D. Delafosse, T. Magnin, Hydrogen induced plasticity in stress corrosion cracking of engineering systems, Eng. Frac. Mech. 68 (2001) 693–729. [6] S.P. Lynch, Metallographic contributions to understanding mechanisms of environmentally assisted cracking, Metallography 23 (1989) 147–171. [7] S.P. Lynch, A fractographic study of hydrogen-assisted cracking and liquidmetal embrittlement in nickel, J. Mater. Sci. 21 (1986) 692–704. [8] S.P. Lynch, A fractographic study of gaseous hydrogen embrittlement and liquid-metal embrittlement in a tempered-martensitic steel, Acta Metall. 32 (1984) 79–90. [9] Y. Mine, K. Hirashita, M. Matsuda, M. Otsu, K. Takashima, Effect of hydrogen on tensile behaviour of micrometre-sized specimen fabricated from a metastable austenitic stainless steel, Corros. Sci. 53 (2011) 529–533. [10] M. Koyama, E. Akiyama, T. Sawaguchi, K. Ogawa, I.V. Kireeva, Y.I. Chumlyakov, K. Tsuzaki, Hydrogen-assisted quasi-cleavage fracture in a single crystalline type 316 austenitic stainless steel, Corros. Sci. 75 (2013) 345–353. [11] R.J. Walter, W.T. Chandler: effects of high-pressure hydrogen on metals at ambient temperature, Report no. NASA-CR-102425, Rocketdyne, Canoga Park, CA, 1969.
H.F. Jackson et al. / Corrosion Science 77 (2013) 210–221 [12] H.G. Nelson, Hydrogen embrittlement, in: C.L. Briant, S.K. Banerji (Eds.), Treatise on Materials Science and Technology: Embrittlement of Engineering Alloys, Academic Press, New York, NY, 1983, pp. 275–359. [13] G.R. Caskey Jr., Hydrogen effects in stainless steels, in: R.A. Oriani, J.P. Hirth, M. Smialowski (Eds.), Hydrogen Degradation of Ferrous Alloys, Noyes Publications, Park Ridge, NJ, 1985, pp. 822–862. [14] N.R. Moody, S.L. Robinson, W.M. Garrison Jr., Hydrogen effects on the properties and fracture modes of iron-based alloys, Res. Mech. 30 (1990) 143–206. [15] C. San Marchi, B.P. Somerday: technical reference on hydrogen compatibility of materials, Report no. SAND2008-1163, Sandia National Laboratories, Livermore, CA, 2008.
221
[37] J. Singh, Influence of deformation on the transformation of austenitic stainless steels, J. Mater. Sci. 20 (1985) 3157–3166. [38] D.G. Ulmer, C.J. Altstetter, Hydrogen-induced strain localization and failure of austenitic stainless steels at high hydrogen concentrations, Acta Metall. Mater. 39 (1991) 1237–1248. [39] K.A. Nibur, D.F. Bahr, B.P. Somerday, Hydrogen effects on dislocation activity in austenitic stainless steel, Acta Mater. 54 (2006) 2677–2684. [40] ASTM, Standard test method for measurement of fracture toughness, ASTM International, West Conshohocken, PA, 2009. [41] ASTM, Standard test method for J-integral characterization of fracture toughness, ASTM International, West Conshohocken, PA, 1996. [42] C. San Marchi, B.P. Somerday, S.L. Robinson, Permeability, solubility and diffusivity of hydrogen isotopes in stainless steels at high gas pressures, Int. J. Hydrogen Energy 32 (2007) 100–116. [43] D.J. Kotecki, T.A. Siewert, Wrc-1992 constitution diagram for stainless steel weld metals: a modification of the wrc-1988 diagram, Weld. J. 71 (1992) 171– 179. [44] C.D. Beachem, A new model for hydrogen-assisted cracking (hydrogen ‘‘embrittlement’’), Metall. Trans. 60 (1972) 437–451. [45] H.K. Birnbaum, P. Sofronis, Hydrogen-enhanced localized plasticity - a mechanism for hydrogen-related fracture, Mater. Sci. Eng. A A176 (1994) 191–202. [46] D.P. Abraham, C.J. Altstetter, Hydrogen-enhanced localization of plasticity in an austenitic stainless steel, Metall. Mater. Trans. A 26A (1995) 2859–2871. [47] C.G. Rhodes, A.W. Thompson, The composition dependence of stacking fault energy in austenitic stainless steels, Metall. Trans. A 8A (1977) 1901–1906. [48] R.E. Schramm, R.P. Reed, Stacking fault energies of seven commercial austenitic stainless steels, Metall. Trans. A 6A (1975) 1345–1351. [49] P.L. Mangonon Jr., G. Thomas, The martensite phases in 304 stainless steel, Metall. Trans. 1 (1970) 1577–1586. [50] J.F. Breedis, L. Kaufman, Formation of hcp and bcc phases in austenitic iron alloys, Metall. Trans. 2 (1971) 2359–2371. [51] L. Rémy, A. Pineau, Observation of stacked layers of twins and e martensite in a deformed austenitic stainless steel, Metall. Trans. 5 (1974) 963–965. [52] G. Olson, M. Cohen, A general mechanism of martensitic nucleation: Part i. General concepts and the fcc ? hcp transformation, Metall. Trans. A 7A (1976) 1897–1904. [53] L. Rémy, The interaction between slip and twinning systems and the influence of twinning on the mechanical behavior of fcc metals and alloys, Metall. Trans. A 12A (1981) 387–408. [54] P. Müllner, C. Solenthaler, P.J. Uggowitzer, M.O. Speidel, Brittle fracture in austenitic steel, Acta Metall. Mater. 42 (1994) 2211–2217. [55] P. Müllner, On the ductile to brittle transition in austenitic steel, Mater. Sci. Eng. A A234–236 (1997) 94–97. [56] S. Takaki, T. Furuya, Y. Tokunaga, Effect of si and al additions on the low temperature toughness and fracture mode of fe–27mn alloys, ISIJ Int. 30 (1990) 632–638. [57] J. Brooks, J. Williams, A. Thompson, Microstructural origin of the skeletal ferrite morphology of austenitic stainless steel welds, Metall. Trans. A 14A (1983) 1271–1281. [58] R. Ritchie, A. Thompson, On macroscopic and microscopic analyses for crack initiation and crack growth toughness in ductile alloys, Metall. Trans. A 16A (1985) 233–248. [59] A. Turnbull, Hydrogen diffusion and trapping in metals, in: R.P. Gangloff, B.P. Somerday (Eds.), Gaseous Hydrogen Embrittlement of Materials in Energy Technologies, Woodhead Publishing, Oxford, UK, 2012, pp. 89–128. [60] M.I. Luppo, A. Hazarabedian, J. Ovejero-Garcia, Effects of delta ferrite on hydrogen embrittlement of austenitic stainless steel welds, Corros. Sci. 41 (1999) 87–103. [61] J.P. Hirth, Effects of hydrogen on the properties of iron and steel, Metall. Trans. A 11A (1980) 861–890. [62] T.-P. Perng, C.J. Altstetter, Effects of deformation on hydrogen permeation in austenitic stainless steels, Acta Metall. 34 (1986) 1771–1781. [63] M.R. Louthan Jr., R.G. Derrick, Hydrogen transport in austenitic stainless steel, Corros. Sci. 15 (1975) 565–577. [64] M. Nagumo, Hydrogen related failure of steels – a new aspect, Mater. Sci. Technol. 20 (2004) 940–950. [65] T. Neeraj, R. Srinivasan, J. Li, Hydrogen embrittlement of ferritic steels: observations on deformation microstructure, nanoscale dimples and failure by nanovoiding, Acta Mater. 60 (2012) 5160–5171.