Environmentally Assisted Cracking Behavior of 18Ni 250 Maraging Steel Welds Obtained with and Without Flux in Chloride Media

The 18Ni 250 grade maraging steel, possessing an excellent combination of strength and fracture toughness, is a preferred choice of material for aerospace and military applications.¹  In particular, this steel is a material of construction for a large number of solid rocket motor casings for satellite launch vehicles.²  The fabrication of these motor casing involves gas tungsten arc welding (GTAW). Post welding, the large-sized casings are subjected to peak aging treatment. The fabricated structures in aged conditions are stored in atmospheric conditions, during which a risk of weldment corrosion exists. As a result, these welds are periodically inspected until these are put into service.

Several studies relating to the general corrosion and stress corrosion cracking (SCC) behavior of maraging steels have been conducted.^3-5  Dean and Copson³  showed that SCC of maraging steels was largely influenced by their process history. In the standard annealed and peak maraged condition (815°C for 1 h + 485°C for 3 h) with a fine grain structure 18Ni 250 steel showed no cracking in 3.5 wt% NaCl solution in a U-bend test even after 30 d exposure, while a coarse grain and thermally-embrittled structure revealed cracks within 4.8 h of exposure. The SCC susceptibility of various grades of maraging steel have been found to increase with their tensile strength.^5-6  It is also known that irrespective of the grades, the SCC susceptibility of these steels increases in the order: solution annealed < overaged < peak aged < underaged.^6-8 

However, there is disagreement on the cracking mechanism operating in these alloys in chloride medium in freely corroding conditions. Some researchers suggested that in maraging steels environmentally assisted cracking (EAC) occurred by anodic path dissolution (SCC),^9-11  while a few others have proposed hydrogen embrittlement (HE) mechanism.^12-13  Rubin⁹  proposed anodic dissolution mechanism for 18 Ni maraging steel in 3 wt% NaCl. For this steel, Syrett¹⁰  suggested anodic dissolution mechanism at potentials more noble than −600 mV vs. saturated calomel electrode (SCE) and HE in the potential range −600 mV_SCE to −800 mV_SCE, in 3.5 wt% NaCl. It is easily understandable because the open-circuit potential (OCP) of this metal was about −580 mV_SCE and the above two conditions distinctly represent anodic and cathodic polarization of the metal. Perkins and Haney¹¹  through their study in acidified 0.6 N NaCl (pH 2.2) solution suggested that EAC susceptibility of 18Ni 250 maraging steel was due to SCC, not HE, despite the solution being acidic in nature. On the other hand, Stavros and Paxton¹²  suggested HE as the EAC mechanism occurring in 18Ni 250 maraging steel freely exposed to 3 wt% NaCl, both in near-neutral (pH 6.3) and in acidified (pH 1.7) conditions. Hayden and Floreen¹³  also suggested HE mechanism in this steel while studying the failure mechanism under various modes of loading in 3.5 wt% NaCl.

Moreover, published literature on SCC behavior of maraging steel weldments is limited. Kenyon, et al.,⁴  studied the SCC behavior of GTAW of lower-strength maraging steels (grades 180 and 200) in seawater using the U-bend technique. Both of the steels in aged conditions did not crack even after 3 y of exposure. However, their welds were found to be less resistant to cracking compared to the base metal. The grade 180 welds did not crack for 1 y, while some of the grade 200 welds failed within a short span of two months of exposure. Nevertheless, annealing treatment at 816°C for 1 h prior to the aging significantly improved the cracking resistance of these weldments. Sastry, et al.,¹⁴  reported that 18Ni 250 GTA peak-aged welds having composition either similar to or modified from that of the base metal, showed lower K_1SCC values (36 MPa√m and 40 MPa√m, respectively) than that of the base metal (50 MPa√m) in 3.5 wt% NaCl. Reddy and Rao¹⁵  have shown that homogenization (1,150°C for 1 h) of maraging steel GTA welds prior to aging improved their SCC resistance. However, homogenization of large welded assemblies runs the risk of causing dimensional changes. Additionally, not much research has been directed to understand the role of weld structure on the electrochemical corrosion and SCC especially of 18Ni 250 welds.

Gupta, et al.,¹⁶  have shown earlier that the addition of CaF₂ flux to GTAW eliminated the need to repair motor casing welds, a process integral to manufacturing sound welded components, thereby making the welding process very efficient. Also, these welds showed similar or better tensile strength and fracture toughness compared to those welds produced without flux addition. In a recent publication, Gupta and Raja¹⁷  have reported SCC and electrochemical corrosion behavior of heat-affected zones (four of them), highlighting the influence of their respective microstructures on EAC behaviors. Therefore, there is a need to understand the EAC behavior of the weld fusion zone vis-à-vis the microstructure of the weld produced through flux addition to know the overall performance of welded motor casing. Thus, this work concerns with understanding the EAC behavior of GTAW metal of the 18Ni 250 maraging steel, obtained with and without CaF₂ flux addition, in their peak-aged condition. Additionally, the electrochemical corrosion of the weld metal in relation to its microstructure has been studied.

Solution-annealed 18Ni 250 grade maraging steel plate of 8 mm thickness was used for the study. The chemical composition of the plate is given in Table 1. To obtain weld specimens, 300 mm × 500 mm × 8 mm thick coupons were welded by machine GTAW process using modified composition filler metal^18-19  as mentioned in Table 1. Flux was introduced to welding through application of a slurry (1 g CaF₂ powder + 5 mL of laboratory-grade acetone) on weld surface root pass onward, during coupons welding. To qualify the welds for further studies, dye penetrant, ultrasonic, and x-ray radiographic techniques were used. All of the specimens (base metal and welds) were machined before aging which was performed at 758±5 K (485±5°C) for 3 h 20 min in an electrical resistance furnace followed by grinding using 1000 grit SiC emery paper prior to testing.

Round tensile specimens of 2.5 mm diameter, as per Figure 8 of ASTM E8M²⁰  were used for conducting the slow strain rate tensile (SSRT) tests, as per ASTM G129.²¹  Specimens for base metals were sectioned along rolling direction of the plate, while those for welds were sectioned longitudinally along fusion zone. Tests were conducted at a constant crosshead speed of 0.0017 mm/min corresponding to an initial strain rate of 2.26 × 10⁻⁶ s⁻¹ on 20 kN capacity, United Calibration Corporation STM 20^† model tensile machine. Five different test environments namely, (a) glycol, (b) air, (c) 0.1 M NaCl (pH 6.7) solution, (d) acidified 0.1 M NaCl (0.1 M NaCl + 0.063 M HCl, pH 1.2) solution, and (e) 0.1 M NaCl solution with cathodic charging at −1.3 V_SCE, were used for testing. Cathodic charging tests were conducted on Solartron SI 1287^† corrosion work station, using a three-electrode system comprised of a platinum counter electrode, SCE as the reference electrode, and the SSRT specimen as the working electrode. Tests were performed using a set of three specimens for each test condition. The failed specimens were examined microscopically to study the mode of failures.

Specimens were mounted in a resin in such a way that the thickness section of the base metal specimens and the cross section of the welds could be exposed during testing. Mounted specimens were provided with an electrical connection using an insulated copper wire attached to the rear face of the specimen. The mounted specimens were ground to 1000 grit SiC emery paper and then polished with 3 μ diamond paste prior to testing. The polished specimens were ultrasonically cleaned in methanol. A three-electrode system comprised of a platinum counter electrode, SCE as a reference electrode, and the specimen as a working electrode was used for the tests. The tests were conducted using a Biologic SP-300^† corrosion workstation.

Microstructures of the base metal and welds sections, in their aged conditions, were studied under an optical microscope (Olympus GX51F^†). Specimens were etched by the modified Fry’s reagent (28 g FeCl₃ + 8 g CuCl₂ + 20 mL HNO₃ + 400 mL HCl + 400 mL ethanol + 400 mL H₂O). Electron probe microanalysis (EPMA) using Cameca SXFive^† model wavelength dispersive spectroscopy (WDS) was performed on welds for the compositional analysis. Specimens exposed to 0.1 M NaCl solutions and subjected to SSRT were examined under a Quanta 200^† scanning electron microscope (SEM) equipped with energy dispersive spectroscopy (EDS).

To analyze the nature of corrosion attack, ground and polished specimens were immersed in 0.1 M NaCl (pH ∼6.7) and acidified 0.1 M NaCl (pH 1.2) solutions for 48 h and 16 h, respectively, cleaned in methanol ultrasonically, and were investigated by microscopic examination.

The hydrogen content of 5 mm × 5 mm × 5 mm sized specimens exposed to neutral and acidified 0.1 M NaCl solutions for 14 d was measured using LECO-ONH 836^† gas analyzer. Specimens were exposed for such a long duration deliberately to ensure that any slow entry of hydrogen could also be measured. Specimens were also examined for their hydrogen content prior to the exposure in NaCl solutions. The specimens were ground until 1000 grit SiC emery papers and cleaned in demineralized water ultrasonically prior to the tests.

Typical microstructures of 18Ni 250 maraging steel and its welds (without and with flux) post the aging treatment are shown in Figures 1 and 2, respectively. The base metal showed blocks of lath martensite (Figure 1). The microstructures of the welds had complex features comprised of dendritic cells of light etched martensite surrounded by regions of dark etched martensite with white isolated pools formed at intersections of dendrites (Figure 2). These white pools were identified as soft austenite phase, as reported by Gupta, et al.¹⁶  The dendritic arm spacing measured transverse to the direction of dendrites was found to be 15 µm to 22.5 µm and 22 µm to 32.5 μm in the welds obtained with and without flux addition, respectively.

The segregation of Ni, Mo, and Ti at the boundaries in GTAW of this alloy has been reported in the literature^19,23  in the context of austenite reversion during aging; therefore, the possible existence of compositional variation in the present weld was studied using WDS-EPMA. X-ray maps (Figure 3) and the compositional analysis obtained from this study (Table 2), confirmed notable segregation of Ni, Mo, and Ti at the boundaries, which was expected due to their lower partition ratio (k < 1).¹⁹  Interestingly, significant segregation of Al (Table 2) was also noticed at the boundaries, which has not been reported earlier in the literature.

As Mo and Ti significantly influence the alloy strength and the welds contain lower amounts of these elements than in the base metal (Table 1) (adding Al contribution same as Mo),²⁵  welds were expected to have about 40 ksi (275 MPa) lower strength than that of the base metal.

The effect of environments, including the air (it is considered to be mildly corrosive due to humidity), has been assessed by comparing the respective mechanical properties with that obtained in glycol (considered to be inert). Notably, the air did not cause any significant loss in mechanical properties from that of in glycol. A significant loss in ductility (21% to 23%) in base metal and both the welds was observed in 0.1 M NaCl although the UTS values remained the same. Acidified 0.1 M NaCl solution caused a significant drop in ductility both in base metal and welds, showing an apparent failure within elastic regime, with a marginal reduction in their strengths that was 5% and 12%, respectively. The cathodic charging of the specimens in 0.1 M NaCl also caused the failure in elastic regime, albeit losses in their strengths were more significant (23% and 32% in base metal and welds, respectively). Notably, vigorous hydrogen evolution occurred over the tensile specimens exposed to acidified 0.1 M NaCl, while the same was not visible in near-neutral 0.1 M NaCl solution. During the test, the pH of acidified 0.1 M NaCl solution was found to vary from 1.2 to 1.5.

The fractured specimens in glycol, air, and 0.1 M NaCl showed cup-cone feature, typical of macroscopic feature of a ductile failure. In all three cases (base metal, weld with and without flux), no discernible features in their respective microvoid coalescence structures in glycol, air, and neutral 0.1 M NaCl were seen. Even high-magnification images of the fractured surfaces did not reveal any perceptible variation due to 0.1 M NaCl. However, as expected, welds comprised of dendritic cast structures showed relatively coarse and nonuniform dimpled structures than that found in the base metal (Figure 5). Between the welds, the one with the flux addition revealed nearly uniform-sized dimples which were marginally finer and larger in number than that in the case of the weld without flux (Figure 5). Such variation could be attributed to the refined dendritic microstructure due to flux addition (Figure 2). Gauge sections of the specimens in 0.1 M NaCl revealed the presence of few fine cracks, which seemed to initiate from the emergent slip steps; for representation purpose, only the weld without flux specimen is shown here (Figure 5[d]). It is worthwhile to comment on the observation that despite both the welds and the base plate showing significant loss in ductility due to exposure to 0.1 M NaCl solution, no significant variation in the microscopic fracture features are noticed. While no clear explanation can be offered to such a behavior, it can be mentioned that appearance of microvoids does not preclude a material being subjected to EAC. In fact, the adsorption-induced dislocation emission (AIDE) mechanism suggested by Lynch²⁶  explains the formation of microvoids even while a metal suffers EAC.

Fractured surfaces of the base metal and the welds, tested both in acidified 0.1 M NaCl and cathodically-charged condition, exhibited distinct brittle fracture over a large area in the cross section of the specimens (Figures 6 through 8). The base metal exhibited a mixed-mode of failure having intergranular (IG) and transgranular (TG) cracks in both acidified and cathodically charged conditions (Figure 6). The unfluxed weld displayed IG fracture in acidified 0.1 M NaCl and TG fracture under cathodic charging (Figure 7). On the other hand, the fluxed weld displayed TG fracture both in acidified 0.1 M NaCl and cathodically charged condition (Figure 8). This indicated the possibility of the dendritic boundaries being anodically attacked in acidified NaCl in the unfluxed welds, while the fluxing reduced such an attack. The fact that the unfluxed weld cracked intergranuarly in acidified NaCl contrary to TG cracking under cathodic charging suggested then preferential anodic attack on the boundaries in former case.

Slow strain rate data and the fractography suggested the following behavior of the welds with and without flux addition, with regard to EAC:

• In neutral 0.1 M NaCl, welds were prone to EAC, showing 23% loss in ductility with ductile failure, similar to that of the base metal.

• Both the acidified 0.1 M NaCl solution and cathodic charging caused brittle fractures in welds with a substantial drop in ductility. Their tensile properties were significantly lowered than the base metal, thereby suggesting that under these conditions, welds were more susceptible to EAC than the base metal.

• Flux addition did not influence the EAC behavior of the weld, although fracture features of both the welds differed perceptibly, especially in the acidified condition.

The susceptibility of the alloy was found to differ widely in all of the test conditions and, thus, to further understand the EAC mechanism, their electrochemical corrosion behaviors were studied.

The electrochemical corrosion behaviors of welds were studied both in neutral and acidified NaCl electrolytes. For comparison, base metal was also examined in the same conditions. The potentiodynamic polarization curves of the welds with and without flux addition and the base metal, in freely exposed neutral and acidified 0.1 M NaCl solutions, are shown in Figures 9(a) and (b), respectively. In latter case, scan rate was increased to avoid fluctuations observed during polarization due to high corrosion rate. Corrosion rates obtained from the polarization curves are plotted in Figure 10. In neutral NaCl, both the welds exhibited a lower corrosion rate when compared to that shown by the base metal (Figure 10[a]). However, the same was found to be reversed in the acidified condition, where the welds showed much higher corrosion rates than that of the base metal (Figure 10[b]). The addition of flux did not result in a significant difference in the weld corrosion behavior, although the scatter in the corrosion rates of the fluxed welds was notably lower as against the wide scatter displayed by the weld without flux.

The anodic branch of the polarization curves of all specimens showed active dissolution with no apparent passivity (Figures 9[a] and [b]). The cathodic polarization curves in neutral solution showed activation controlled cathodic reaction close to their corrosion potentials (E_corr), which became diffusion-controlled at high applied potentials (Figure 9[a]). In acidified chloride solution (Figure 9[b]), welds showed a notable shift in their E_corr values toward negative potential from that in the neutral solution with the average E_corr being: −0.26 V_SCE and −0.36 V_SCE in neutral and acidified solutions, respectively. Though the average E_corr of the base metal was −0.33 V_SCE and −0.34 V_SCE, in neutral and acidified solutions, respectively, they showed no appreciable change. As expected, all specimens showed higher cathodic kinetics in acidified solution from that in neutral solution, where the increase shown by the welds was higher than that of the base metal.

Metallographic investigation of base metal exposed in 0.1 M NaCl solutions exhibited a fairly uniform attack, however, at higher magnification preferential attack along the martensite laths and at the boundaries of the martensite blocks/prior austenite (Figure 11) was discernible. On the other hand, both the welds revealed preferential dissolution of dendritic cells over the interdendritic boundaries in neutral NaCl (Figure 12). In acidified NaCl, the weld with addition of flux showed more uniform attack across the matrix, albeit the weld without flux addition showed the presence of typical local attacks apparently along the boundaries (Figure 13), matching with their respective fractographic features where the former revealed cleavage failure (Figure 8[a]) and the latter showed presence of IG attack (Figure 7[a]).

Electrochemical corrosion studies of the welds in various NaCl media could be summarized as below:

• In neutral NaCl, welds showed lower corrosion rate than that of the base metal. However, in acidified conditions, these welds showed a significant increase in corrosion from that of the base metal.

• Fluxed weld displayed less scatter in their corrosion rates than that shown by unfluxed weld. In acidified solution, the latter revealed the presence of local attacks contrary to the former revealing fairly uniform attack.

With respect to welds EAC and electrochemical corrosion behaviors shown in NaCl medium, the possible failure mechanism under freely corroding condition has been discussed as below.

To ascertain the possible failure mechanism of EAC as anodic dissolution (or HE), it was necessary to delineate the possible role of hydrogen on EAC of this weld. Therefore, the hydrogen content of the base metal and welds before and after exposure to 0.1 M NaCl and acidified 0.1 M NaCl solutions (for sufficiently long duration 14 d), was measured and is summarized in Table 4. Under cathodic charging at −1.3 V_SCE potential, the corresponding cathodic current densities were found to be sufficiently high (on order of 10⁻² A/cm²) to cause hydrogen evolution on alloy surface (as per Figure 9[a]).

The unexposed base metal, as expected, has a very low level of hydrogen, whereas the weld has significant hydrogen in the unexposed condition. The high hydrogen content of the weld might be due to both the welding and the filler metal (Table 1). Notably, both the base metal and the welds did not pick up any hydrogen due to exposure in 0.1 M NaCl. On the contrary, both the welds and the base metal contained 25 ppm to 30 ppm of hydrogen upon exposure to acidified 0.1 M NaCl, indicating the occurrence of hydrogen evolution reaction (HER) in the acidified solution.

Putting and E_H+/H2 = respective E_corr as reported above in the Electrochemical Corrosion Behavior section, in neutral solution (pH 6.7) the weld surfaces were expected to be exposed to about 10⁻¹⁴ atm pH₂ implying that hydrogen evolution and its entry into the metal surface was not possible, while in acidified 0.1 M NaCl (pH 1.2) solution, pH₂ was estimated to be about 74 atm, indicating the occurrence of hydrogen evolution on the surface of the weld, agreeing with hydrogen measurement results in exposed samples.

Thus, in neutral solution, the hydrogen evolution on the alloy surface was not found to be possible, indicating that EAC apparently followed SCC. The fact that the addition of sodium arsenate, a cathodic poison, to 3.5 wt% NaCl did not influence the EAC behavior of 18Ni 250 steel,¹⁰  whereas the application of specific cathodic current lowered the EAC susceptibility of the alloy in NaCl,⁹  suggested that this alloy was not susceptible to HE in neutral sodium chloride medium. The same was indeed supported by SSRT results in the present investigation. The fractographs showing the presence of secondary cracks on the gauge surface suggest the breakdown of the surface film due to emerging slip steps and subsequent corrosion by media, leading to anodic dissolution of material.^10-11  In an alloy of such high toughness, it is possible that both AIDE and slip-assisted active dissolution mechanism may operate in 0.1 M NaCl. The lower corrosion rate of welds than that of the steel shown in electrochemical corrosion tests could be attributed to the difference in their compositions (Table 1). Apparently, the higher Co content in welds (by about 4%) from that of the base metal reduced anodic dissolution of the weld in neutral solution.²⁷  Haigh,²⁷  in his doctoral work, demonstrated that Co was beneficial to corrosion resistance, in comparison with Fe in 18Ni 250 maraging steel. However, it is necessary to point out that Co is known to restrict the cross slip in steel.²⁸  As a consequence, the weld metal is expected to be more susceptible to EAC than that of the base metal. However in 18% maraging steel, its embrittling effect is reported to be compensated by the high content of Ni.²⁹  Another factor that can complicate the behavior of the weldment is its cast structure as opposed to the wrought structure of the alloy, making it difficult to know the influence of the added Co on EAC behavior of the alloy. Notably, no perceptible difference between welds and base metals could be seen with respect to their EAC behavior observed in neutral solution. Further, in welds, the preferential attack within dendritic cells over the cell boundaries could be attributed to the compositional variations observed from their compositional analysis through WDS-EPMA (Figure 3[a] and Table 2), mainly the depletion of Ni within the cells, and the enrichment of same at the boundaries, resulting in this typical attack.

Contrary to the neutral condition, occurrence of HER in the acidified solution was corroborated by the fact that in this solution copious hydrogen was evolved with bubbling observed on the tensile specimens. Also, the unstressed specimens freely exposed in the acidified solution had significant amount of hydrogen. The SSRT results (Table 3) showing sharp reduction in ductility and brittle fracture features (Figures 6 through 8) in acidified 0.1 M NaCl suggested HE failure mechanism. Similarities of these results with the SSRT results for the specimens with cathodic charging at −1.3 V_SCE (where HER was assumed to occur on the metal surface) further supported that the predominant failure mode was due to HE. Although, the contribution due to anodic dissolution mechanism cannot be ignored in acidified solution. The higher susceptibility of the welds to HE as revealed by greater loss in mechanical properties and higher cathodic kinetics in acidified conditions (Table 3 and Figure 9[b]) could be attributed to their coarse dendritic structures,^4,30  from that of the fine-grained steel. Post aging, all heat affected zones (HAZs) comprised of peak-aged microstructures are expected to behave similar to the base metal as against different EAC susceptibility shown by HAZ II, III, and IV, in as-welded condition in an earlier study.¹⁷  Thus in a peak-aged GTAW, weld fusion zone is the most susceptible region to EAC compared to the base metal or HAZs.

The effect of flux in weld has been reflected by the prevention of IG mode of cracking (Figures 7[a] and 8[a]) and reduction in the selective interdendritic attack (Figures 13[a] and [b]) in acidified conditions, indicating the possibility that long-term exposure might render the grain boundaries more vulnerable to cracking in case of unfluxed welds. The retardation of anodic attack at the cell boundaries and the lowered scatter in corrosion rates in fluxed weld apparently is due to its refined microstructure. The role of CaF₂ flux in weld refinement was brought out in previous work.¹⁶  It should be pointed out that the corrosion in neutral sodium chloride solution may lead to localized acidification in crevices and pitting, in which case the welds can become susceptible to brittle failure. Thus, the changes brought by the flux addition in welds microstructural features are noteworthy.

Environmentally assisted cracking and electrochemical corrosion behaviors of gas tungsten arc welds of 18Ni 250 maraging steel, with and without the addition of CaF₂ flux in NaCl medium, were studied. Following are the salient findings.

• Both the welds, with and without the flux addition, revealed a 21% to 23% loss in ductility, in neutral 0.1 M NaCl, similar to that of the base metal, showing susceptibility to EAC. The absence of hydrogen in the exposed welds as determined experimentally and the lack of thermodynamic tendency for hydrogen evolution on the alloy at E_corr suggested that the ductility loss suffered in neutral chloride solution was apparently due to anodic dissolution or adsorption-induced dislocation emission mechanism.

• In acidified 0.1 M NaCl, both types of welds failed within elastic regime, showing brittle fracture features. Copious hydrogen evolution on the specimens seen during testing and the significant amount of hydrogen found in the exposed specimens suggested that the predominant mode of failure was due to hydrogen embrittlement. Both types of welds behaved similarly even in the neutral 0.1 M NaCl under the application of a more negative potential. While in these conditions, welds have shown higher susceptibility than that of the base metal.

• The unfluxed weld displayed intergranular fracture in acidified NaCl solution at OCP and transgranular fracture under cathodic charging. However, the fluxed weld displayed transgranular fractures in both the conditions. These fractographic features suggested a possibility of the dendritic boundaries being anodically attacked in acidified condition, and were apparently retarded with the addition of flux. Such a transition from intergranular to transgranular cracking might affect long-term service behavior of the welds to corrosion and EAC, though it is not reflected in the short-term tests.

• With respect to electrochemical corrosion, the unfluxed weld showed a large scatter in the corrosion rate and preferential attack at the interdendritic cell boundaries. The addition of flux significantly lowered the scatter and prevented the preferential attack, due to microstructural refinement.

One of the authors, Renu N. Gupta, acknowledges the support extended by Mr. Roy M. Cherian and Mr. Chidanand Magadum, Vikram Sarabhai Space Center (VSSC), Trivandrum, India in providing 18Ni 250 maraging steel for this research.