Microstructure and Corrosion Behavior of Friction Surfaced 304L Austenitic Stainless Steels

1.1 Friction surfacing of 304L stainless steel

The concept of friction surfacing was first introduced in a patent by Klopstock et al. [1] in 1941. Figure 1 shows a schematic illustration of the friction surfacing process. A rotating consumable rod is pushed against a stationary substrate under an applied load. Under appropriate rotating speeds and pressures, the frictional heat generates a viscoplastic boundary layer at the rod tip. The evolution in temperature and pressure conditions leads to an interdiffusion process resulting in a metallurgical bond between the plasticized material and the substrate. As shown in Fig. 1b, heat conduction into the substrate enables the plastically deformed material to consolidate near the interface, and as such, the viscoplastic shearing interface is formed. As the rotating tool is translated across the substrate, material at the frictional interface either becomes a flash or will roll onto the substrate and form the coating [2–4].

There are published reports discussing friction surfacing of 304L consumable rods over ferrous and non-ferrous materials. Rafi et al. [5] showed discontinuous dynamic recrystallization in friction surfaced 304L stainless steel coatings on low alloy steels, and the coatings were free of carbides and secondary phases. Chandrasekaran et al. [6] reported the formation of 'discrete laminar' layers during friction surfacing of 304 stainless steel on 5083 Al alloy. The delaminated layers from the consumable rod rolled over the substrate before attaining sufficient plasticity to flow on the substrate, leading to a filament-like structure. Shariq et al. [7] studied the feasibility of friction surfacing 304L stainless steel rods over aluminum and copper substrates. The authors successfully deposited stainless steel coatings over aluminum using a start-up plate of mild steel, however, coatings on copper substrate were unsuccessful. To the authors' knowledge, however, there is no available data in the literature that discusses the microstructural evolution during friction surfacing of 304L stainless steel rods onto a substrate of the same material.

1.2 Corrosion performance of friction surfacing

Austenitic stainless steels generally show high corrosion resistance at room temperature, even under oxidizing environments. However, their susceptibility to corrosion at sensitizing temperatures and in the presence of highly passive corrosion products is well documented. Various theories have been introduced regarding mechanisms of intergranular corrosion in austenitic stainless steels, including: (1) chromium depletion due to precipitation of chromium-rich carbides (M₂₃C₆) during sensitization, (2) strain energy associated with grain boundaries as a driving force for intergranular corrosion; which is often accompanied by knife-edge attack, (3) electrochemical potentials setup at the carbide-matrix interface leading to M₂₃C₆ being cathodic and nobler with respect to the matrix, and (4) segregation in the solid solution [8]. These corrosion mechanisms strongly depend on the corroding solution and inherent impurities present in the material. Traditional deposition methods such as fusion welding led to the formation of δ-ferrite, which promote subsequent metallurgical transformation to brittle intermetallic phases, such as sigma (σ) and chi (χ) [9], which are prone to preferential corrosion attack. Due to the mechanical and microstructural advantages that friction surfacing has over traditional fusion-based deposition methods, it has emerged as a promising repair technology in the transportation and energy sectors [10–12]. In certain applications, corrosion resistance of repaired components can be an important factor. For example, the corrosion performance of 304L stainless steel canisters used for used nuclear fuels storage determines the service life. These canisters are fusion-welded during fabrication to create a leak-tight system. However, the high temperatures and phase transformations during fusion welding (e.g., ferrite and sigma phases, sensitization) make these canisters susceptible to corrosion and eventually stress corrosion cracking (SCC). For this applications, the authors demonstrated the efficacy of friction surfacing to produce leak-tight repairs on 304L austenitic stainless steels with simulated SCC [13]. Rafi et al. [14] showed that during friction surfacing of 304 stainless steel, performed in the temperature regime of sensitization, there was no presence of the δ-ferrite phase. They also observed that the pitting corrosion resistance of friction surfaced specimens was considerably higher than gas tungsten arc welded (GTAW) specimens. After electrochemical anodic polarization tests, the friction surfaced specimens showed similar corrosion potential to the as-received 304 stainless steel. In addition, 316L stainless steel friction surface coatings showed the absence of elemental dilution around the grain boundaries [15]. In an extensive microstructural study conducted by Guo et al. [16], it was found that friction surfacing with 316L stainless steel resulted in fragmentation of MnS inclusions, which reduced the size of pits generated after immersion tests in chloride solutions. The authors also reported higher susceptibility to corrosion attack on the advancing side of the coating due to larger strain energy and stronger < 111 > texture. Salt spray tests were used to show the effectiveness of friction surfacing in producing corrosion-resistant coatings with 316L stainless steel [17].

1.3 Objective and novelty

The present work examines microstructural transformations during friction surfacing of 304L stainless steel rods over a sensitized 304L stainless steel substrate. In addition, the corrosion performance of the friction surfaced coatings in chloride-rich environments was evaluated. It was shown that friction surfacing led to strain-induced martensite formation within the austenite matrix, with intense plastic deformation resulting in grain refinement. Corrosion studies revealed that friction surfaced coatings deposited on sensitized substrates improved the corrosion performance of the material.

2.1 Friction surfacing

Friction surfacing was performed using a 3-axis CNC milling machine (HAAS, TM-1, USA). The substrates were 3-mm-thick 304L stainless steel sheets (North American Stainless, USA). The consumable rod (i.e., consumable tool) 12.7-mm in diameter was also 304L stainless steel (Walsin Lihwa Corp., Taiwan). Table 1 shows the chemical composition of the substrates and rods. The substrate was sensitized by heating in a furnace at 600°C for 48 hours in ambient air to simulate a heat-affected zone (HAZ) in fusion welding. The rods and substrate were polished using 600 grit SiC paper to normalize the surface condition and remove the surface oxide layer and contaminants. In the friction surfacing process, the consumable rod was positioned with an initial plunge (below the point of contact) of 1 mm at a 20 mm/min constant plunge rate. After a dwell time of 0.25 seconds, the rod was traversed along the substrate with a lateral traverse feed rate (V_x) of 150 mm/min and a constant axial (plunge) feed rate (V_z) of 60 mm/min. The consumable rod was rotated at a spindle speed of 4,000 RPM. The depositions were 40 mm long. Table 2 summarizes the parameters used during the friction surfacing experiments. 

As shown in Fig. 2, the unbonded regions in the as-deposited coatings were machined off to minimize crevice corrosion near the unbonded edges during corrosion tests and the top of the coating (~ 200 µm) was polished with 600-grit size abrasive paper, to reduce surface roughness and remove surface oxides formed during the deposition. No coating damage such as cracking and delamination were identified after these post-deposition steps. More information on the experimental setup can be found in the authors' previous publication [13].

2.2 Microstructural characterization

Friction surfaced samples were sectioned perpendicular to the advancing direction of the tool at a location two-thirds from the start of the coating to examine the steady-state region. The process was deemed to be at a steady state once the process forces in the z-axis became steady. The cross-section was ground successively with 240, 400, 600, 800, and 1200-grit SiC paper, followed by polishing with diamond slurry of 9 µm, 3 µm, 1 µm, and 0.25 µm abrasive sizes. This was followed by vibratory polishing using colloidal silica solution at 0.05 µm for 3 hours. The polished samples were etched using aqua-regia solution (3-parts hydrochloric acid and 1-part nitric acid). Select samples were also etched using Lichtenneger-Bloech (LBI) solution to perform color etching. The solution distinguishes the different phases in stainless steel by preferential etching; the austenite phase turns yellow/brown, the martensite turns blue while the ferrite remains white. The solution was prepared by adding 20 g of ammonium hydrogen difluoride and 0.5-1 g potassium disulfide to 100 ml DI water while maintaining the solution temperature at 37 °C.

The cross-sectioned samples were imaged using a white light optical metrology system (Alicona, InfiniteFocus® G4, Austria) and scanning electron microscopy (SEM, Zeiss LEO-1 microscope) operated at a 5 kV accelerating voltage. Energy-dispersive x-ray spectroscopy (EDS, Thermo Noran, USA) was used to analyze chemical composition across the coating-substrate interface at 10 kV accelerating voltage. Phase identification of the substrate and coatings was performed using Bruker D8 Discovery X-ray diffraction (XRD) with Cu Kα radiation (wavelength 1.5404 Å). The diffracted beam was acquired from 20° to 100° with coupled 2θ mode.

X-ray diffraction patterns were also collected between 2θ = 20–120° using a Malvern Panalytical Empyrean diffractometer in the θ-θ geometry with a Cu Kα source (λ = 1.5404 Å) for residual stress measurements and phase identification. Diffracted beams were captured using a scanning line detector. Residual stress measurements were conducted using the sin²ψ technique assuming a biaxial stress state and plane stress condition per ASTM-E2860 standard. For this method, a higher angle 2θ peak is desirable, therefore, (211) peak at 90.25° was selected for analysis. Five measurements were taken in the + ψ direction at tilt angles between ψ = 0–40°. A custom MATLAB program was used to strip the Kα2 contribution, remove background, apply corrections for Lorentz polarization and absorption, and fit the peaks to a pseudo-Voigt function. Measurements were repeated at sample rotation angles of φ = 0° and 90° to evaluate the biaxiality of the stress state.

Micro-indentation tests were conducted on a microhardness tester (Beuhler, Micromet, USA) using a Vickers diamond indenter with a 50 g load. The hardness measurements were taken from the top of the coating across the interface into the substrate with an indent spacing of approximately 30 µm. The indentation marks were observed using SEM to identify any cracking around the indents as a qualitative measure of the brittleness of the coatings.

2.3 Corrosion testing

The friction surfaced coatings and reference 304L substrate were polished using 600-grit SiC paper to normalize the surface condition prior to corrosion tests. Two corrosive media were used to understand the pitting corrosion mechanisms of friction surfaced coatings. First, pitting corrosion behavior was studied per ASTM G48, Method-A using ferric chloride solution at room temperatures for 72 hours. To allow for corrosion of only the coated surface, all the other faces were masked using a lacquer/ liquid maskant Pealseal green (Maskcoat LLC, USA). Second, the samples were immersed in boiling MgCl₂ solution for 48 hours and maintained at 155 ± 1 °C per ASTM G36-94 standards. Only the region of interest of the specimen was only exposed by masking other areas using engine enamel (VHT SP128, Sherwin-Williams, USA). After the boiling MgCl₂ test, the formation of corrosion by-products adhered to the corroded region were observed. To image regions underneath the corrosion product, the by-products were removed using the ASTM G1 cleaning procedure. The samples were cleaned in 10% HNO₃ acid solution heated at 60 °C for 20 minutes. At the end of each corrosion test, the samples were cleaned ultrasonically in de-ionized water for 20 minutes. The morphology of the corroded area was imaged using SEM, and phase changes due to the corrosion process were studied using XRD.

The ASTM G48 method (FeCl₃ solution) is usually utilized to characterize resistance to pitting corrosion. Pitting corrosion resistance is a critical property for friction surfacing applications if the surface is exposed to sea-water or chloride-rich environments. To predict the resistance to SCC, the ASTM G36 method (boiling MgCl₂ solution) has been used. In this method, samples are subjected to a tensile stress using C-ring or U-bend geometries. However, a high process force of friction surfacing makes it challenging to perform friction surfacing on curved surfaces. The ASTM G36 tests still hold value for flat friction surfaced specimens. Since pitting is a precursor to SCC in boiling MgCl₂ solution, the corrosion behavior of the flat specimens in boiling MgCl₂ indicates the susceptibility to SCC.

3.1 Microstructure

Figure 3 shows the friction surfaced coating and consumable rod images after the deposition. The depositions are continuous, and no visible defects were identified on the surface. The average width of the coating was 13.3 ± 1 mm, and the bonded width was 12.2 ± 1 mm. As mentioned earlier, these unbonded regions are machined-off for further analyses of the coatings in the region of interest. The coatings were approximately 500 µm thick, and 200 µm was polished-off from the surface to remove an effect of the oxidized layer in the subsequent tests. As shown in Fig. 3b, a flash was formed during the process around the consumable rod.

Figure 4 shows the cross-sectional SEM images in different zones such as the coating, substrate near the coating/substrate interface, and the middle of the substrate (along the height, without friction surfacing effect) after the chemical etching. The coating showed finer grains compared to the substrate, indicating dynamic recrystallization of the material due to intense plastic deformation of the rotating consumable rod. Previous studies have also reported the evolution of finer grains around existing grain boundaries (necklacing) in the deformed microstructure. Nucleation of new grains at the existing grain boundaries occurs, and a continuous strand of new grains is formed [5]. This phenomenon concurs with the theories of discontinuous dynamic recrystallization (DDRX). The grains right below the coating/substrate interface ('near interface') in the substrate were larger than the bulk grains. It is believed that thermal energy accumulated at the interface during the process induced grain relaxation and growth, as shown in Fig. 4b.

Figure 5 shows the SEM image of the coating/substrate interface showing the interface to be smooth and devoid of any visible defects (e.g., voids, cracks). The EDS elemental map for Cr (Fig. 5b) does not show segregation of chromium in the region of the coating-substrate interface after friction surfacing. Dilution and chromium depletion in the weld/deposition is known to be one of the mechanisms for poor corrosion properties in fusion-based methods in chloride-rich environments [8].

The XRD peaks of the reference substrate and friction surfaced coating are compared in Fig. 6. After friction surfacing, a BCC (111) peak was evident in the coating while it was weak in the substrate. The BCC peak is attributed to the strain-induced martensite transformation occurring in the friction surface coating due to the severe plastic deformation. The presence of martensite was also confirmed by color etching of the cross-sections of the coating. Figure 7 shows the etched macrograph of the cross-section. As mentioned before, the LBI solution colors the martensite blue, and the surrounding austenitic matrix yellow/brown. These blue-colored martensitic sites are observed in Fig. 7. No other phases were identified in the coating. Residual stresses measured using the sin²ψ XRD technique showed compressive stresses of 620 ± 89 MPa on the surface of friction surfaced coatings. It is believed that the high compressive stresses combined with sub-solidus process temperatures provide friction surfacing mechanical and microstructural advantages over fusion-based deposition methods. Fusion welding methods like gas tungsten arc welding result in tensile residual stresses in the weld zones due to melting and solidification processes.

Figure 8 shows the microhardness profile in the cross-section of the friction surfaced specimen. In general, the hardness values in the coating are higher than the substrate, and the values decrease from the top of the coating to the coating-substrate interface. The trend can be correlated to the dynamic recrystallization of the coating. Grain refinement led to increased hardness, as stipulated by the Hall-Petch effect [18]. The high hardness of the coating suggests good wear resistance of friction surfaced deposits. Figure 9 shows the SEM images of the indentation marks in the coating (approximately 50 µm from the top surface), near the interface, and in the substrate. The images show no cracking around the corners of the indents, suggesting that the coatings are well-bonded to the substrate and have good toughness. It is interesting to note that the indentation is smaller for the rhombus's upper half than the lower (Fig. 9a), further indicating a gradient in hardness of the coating from top to the bottom, as shown in Fig. 8.

3.2 Pitting Corrosion Behavior

Figure 10 shows the SEM images of the surface of 304L stainless steel substrates and friction surfaced coatings after 72 hour exposure to FeCl₃ solution per ASTM G48 standards. The images are from localized corroded regions in reference and friction surfaced specimens. Friction surfaced coatings showed smaller size pits. The average pit area in the reference sample was approximately 5000 µm², while it was 500 µm² in the coating. However, the number density of pits was much larger compared to those observed in the reference sensitized 304L samples. Furthermore, the pits observed on friction surfaced coatings are shallower, and there is a presence of smaller crevices within a pit. The small pit diameter may be related to the fine grain structure in the friction surfaced samples. It is believed that grain refinement during friction surfacing played an important role in the pitting corrosion behavior. In addition, grain boundary per unit area increases with decreasing in grain size. This promotes the diffusion of chromium and improves the stability of the passive film [19, 20]. It is noted there are additional factors influencing pitting corrosion resistance. Texture can strongly influence pitting during friction surfacing. Guo et al. [16] showed pitting was favored to occur along < 111 > grains. These grains were present more on the advancing side of the coating compared to the center and retreating side. Segregation of impurities can also affect pitting [21] and the compositional homogenization induced by friction surfacing may mitigate segregation. Further microstructural analyses using TEM and EBSD techniques must be performed for a comprehensive understanding of the pitting corrosion mechanisms of the 304L friction surfaced samples.

To evaluate the corrosion resistance of friction surfaced coatings at elevated temperatures (e.g., 155°C), the samples were exposed to boiling MgCl₂ solution per ASTM G36 standard. As mentioned earlier, only flat friction surfaced specimens were exposed to the solution due to the difficulty of performing friction surfacing with curved specimens due to the high process forces. Figure 11 shows SEM images of the sensitized 304L stainless steel reference sample and the friction surfaced coating after 48 hours of exposure to the solution. ASTM G1 procedure was used to remove the corrosion products from the surfaces. Although the specimens were cleaned for 20 minutes, the corrosion product still remained on the uncoated 304L substrate, which is thicker than the friction surfaced specimen. As shown in Fig. 11, the corrosion products on the uncoated substrate are larger, darker, and localized, whereas the corrosion products in the friction surfaced coatings are smaller and distributed over the surface, which is consistent with the result from the FeCl₃ test. 

Figure 12 shows higher magnification images of uncoated substrate and friction surfaced specimen after removing the corrosion by-product via ASTM G1. The image shows preferential corrosion at the grain boundaries of the samples. It is hypothesized that the fine grain structure of friction surfaced coating is one factor influencing the distinct corrosion behavior in the two corrosive media compared to the sensitized reference sample. The effect of grain refinement on pitting corrosion was mentioned earlier, and grain refinement can also improve the performance of samples against SCC. SCC is known to initiate at the triple junction of grain boundaries and then propagate along the grain boundary to another junction, where it is arrested. The reduction in grain sizes means an increase in the number of triple grain boundary junctions, increasing the probability of crack arrest [22].

The reference sample used in this study was sensitized at 600 °C for 48 hrs. Sensitization leads to the precipitation of chromium carbides at the grain boundaries. This leads to the formation of chromium-depleted bands adjacent to the grain boundaries that are highly susceptible to intergranular corrosion [23]. The thermal cycles in friction surfacing are rapid enough to potentially not cause any sensitization in the deposited material. In addition, ferrite transformations are absent since the temperatures during friction surfacing are below the solidus, as confirmed in Fig. 6. The absence of ferritic phase in friction surfacing is an important advantage over fusion welding methods such as gas tungsten arc welding, where ferrite phase formed is prone to preferential Cl-induced corrosion [14].

XRD was performed on the friction surfaced coating after corrosion tests, and the results are shown in Fig. 13. The diffracted beams were obtained at both corroded and uncorroded regions on the sample surface. After the exposure to boiling MgCl₂ solution, the corroded regions in the specimen showed less intensity of the BCC peak than the uncorroded regions, suggesting preferential corrosion on martensitic sites. The strain-induced martensite is known to have poor corrosion resistance due to selective anodic dissolution [24].