Towards Multilayered Coatings of 304L Stainless Steels Using Friction Surfacing

Previous studies have shown the feasibility of the process for multilayered coatings; however, to the best of the authors’ knowledge, a comprehensive study on measurable process outputs and the coating properties of multilayer friction surfacing 304L onto 304L does not exist. In-process measurement of forces (X, Y, Z), flash temperature, flash geometry, layer temperature, and post-process measurement of layer geometry, microhardness, and microstructure are used to characterize changes in the friction surfacing process as more layers are deposited.

Friction surfacing is an emerging solid-state technology that produces fine-grained coatings with excellent surface and corrosion properties [1,2]. The concept of friction surfacing was first introduced in a patent published by Klopstock [3]; however, a growing interest in the technology has materialized recently. A schematic illustration of friction surfacing is shown in Fig. 1(a). A rotating solid rod is pressed against the substrate under an applied axial load and is traversed along a defined toolpath. Frictional heating at the local contact zone between the consumable rod and the substrate due to the elastoplastic deformation of the rod generates a viscoplastic rubbing interface at the rod tip (Fig. 1(b)). The material exiting the rubbing interface splits with some curling away from the substrate (flash) and the rest rolling onto the substrate (coating). An interdiffusion process is enabled by the pressure and temperature at the interface between the coating and the substrate, as a function of time, resulting in forming of a metallurgical bond. The viscoplastic material is continuously deposited onto the substrate surface by applying a traveling movement. The difference in the tangential velocity of the consumable rod to the linear translation velocity of the consumable rod with respect to the substrate was also reported as the cause for coatings to detach [4–6].

Friction surfacing is often compared to other solid-state friction-based manufacturing processes in literature like friction stir welding, friction welding, and additive friction stir deposition. Friction surfacing uses a consumable rod, part of which becomes the coating, compared to the friction stir tool, which aside from nominal wear, is not deposited. Additionally, the boundary conditions between friction surfacing and friction stir welding are very different and impact heat transfer when the material is exposed to the atmosphere and the level of mixing between materials. Significant forging and extrusion of materials from both sides of the weld occur in friction stir welding. In contrast, the bond between the coating and substrate in friction surfacing is believed to form by a diffusion bonding process with a small amount of relative motion between the surfaces. The additive friction stir process is a closer relative of friction surfacing (MELD Manufacturing, Christiansburg, VA). In additive friction stir, a solid feedstock is constrained by a hollow tool with a shoulder. Recent studies have also used other feedstocks, such as powder [7–10]. Unlike in friction surfacing, additive friction stir adds additional downward pressure on the plastically deformed material as it rolls onto the substrate. This process also provides the additional advantage of semi-continuously feeding the consumable material, allowing upscaling capabilities. Friction welding can appear very similar to friction surfacing when comparing the thermal boundary conditions and material flow, particularly during the plunge phase of both. However, friction surfacing includes lateral motion and is a deposition process while the latter is a joining process with uniaxial motion.

Considerable research has been conducted on friction surfacing of austenitic stainless steel. One of the first studies reported the aerial coverage by friction surfacing of 304 and 316 austenitic stainless steels on mild steel substrates [11]. The authors studied the impact of providing inclination to the consumable rod during the process. Owing to the dynamic nature of phase transformation in stainless steel, emphasis has been given to microstructural changes and material flow occurring during the process. Chandrasekaran et al. [12] reported the formation of “discrete laminar” layers during friction surfacing of 304 stainless steels onto 5083 aluminum 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. Khalid Rafi et al. [13] reported a reduction in average grain size from 40 to 5 µm due to dynamic recrystallization when depositing 304 stainless steel onto low carbon steel with friction surfacing. Reduction in grain size was also observed by Puli and Janaki Ram [14], who also reported that the average grain size increased from 4.8 µm at the coating-substrate interface to 9.4 µm near the top of the coating. They attributed this change to the reduction in the cooling rate at the top of the coating. The authors also found necklacing of subgrains around existing grains, which is a classic sign of discontinuous dynamic recrystallization (dDRX). dDRX was also observed by Guo et al. [15], who reported changes in strain rate and temperatures during the process with changing spindle speeds critical for different modes of recrystallization. The solid-state nature of the friction surfacing process provides it with significant mechanical and corrosion advantages over traditional fusion-based deposition methods. Due to these advantages, friction surfacing is now a potential repair process. Earlier studies conceptualized friction surfacing for the potential repair of turbines and railway tracks [16,17]. More recently, friction surfacing has been demonstrated for crack repair with potential applications in nuclear, marine, and transportation industries [18–21].

Friction surfacing has been applied as an additive manufacturing method by applying multiple layers of the coating. Batchelor et al. [22] studied the effect of metal type on multilayer friction surfacing using consumable rods made of brass, aluminum, and stainless steel on mild steel substrates. They were only able to bond stainless steel on the mild steel substrates. They reported the thermal conductivity of the coating material as one of the important factors to ensure a good deposit. Tokisue et al. [23] deposited 2017 aluminum alloy on 5052 aluminum alloy substrates. They reported a higher efficiency of coating in multilayer configuration compared to monolayer coatings. They also reported higher tensile strength in multilayer coatings. Similar studies were conducted on mild steel rods and substrates by Dilip et al. [24], who machined the surfaces of coatings prior to depositing the next layer, to remove oxides. They reported excellent bonding between individual tracks. The authors also investigated multimaterial coatings using alternate layers of 316 and 410 stainless steels. The microstructure revealed that multilayer friction surfacing can involve significant reheating of previous layers. The authors also reported the impact of reheating on heat-treatable aluminum alloys such as AA 2014 aluminum alloy when deposited over AA 5083 aluminum alloy [25]. They reported over-aging of strengthening precipitates due to the process. Multilayer friction surfacing has also been used to produce functionally graded coatings and metal matrix composites. Gandra et al. [26] used multilayered coatings to enable tailored coating compositions to achieve predefined gradients. They achieved this by reinforcing SiC particles in AA 6083 aluminum alloy rods in increasing concentration per layer. In another study, Karthik et al. [27] were able to uniformly disperse reinforced titanium particles in AA 5083 aluminum alloy consumable rods and deposited on substrates of the same material. They reported the formation of very fine equiaxed grains and well-dispersed second phase particles. In another study, the authors have shown the feasibility of performing multilayer depositions using 6060 aluminum alloy on a substrate of the same material, while investigating observable process changes [28]. Even though previous studies have shown the process’s feasibility for multilayered coatings, to the best of the authors’ knowledge, a comprehensive study on measurable process outputs and the coating properties of multilayer friction surfacing 304L onto 304L is still missing.

All substrates were 5-mm-thick 304L stainless steel sheet (Walzwerke Einsal GmbH, Germany), quenched and drawn with a tensile strength of 814 MPa and a mean hardness value of the 322 HV1. The 304L stainless steel rod (referred to as consumable rod) of 10 mm diameter (Viraj Profiles Limited, India) was cold drawn and polished and has a tensile strength of 702 MPa and a mean hardness value of 230 HV1. Table 1 shows the chemical compositions of the substrate and consumable rod provided by the manufacturers.

Friction surfacing experiments were conducted on a three-axis CNC milling machine (HAAS, VF2, USA). Figure 2 shows the experimental setup. All experiments were conducted using a position control mode of operation. The substrates together with the force measurement system were fixed to the machine table using a mounting plate. The other measuring equipment (mirror reflex camera, two-color pyrometer optics, and infrared camera) were installed inside the machine enclosure to observe the process.

In the friction surfacing tests, the consumable rods were clamped in a manually tightened ER 32 collet chuck holder and had a protruding length of 40 mm. The consumable rod was positioned with an initial plunge (below the point of contact, z_p) of 1 mm at a constant plunge rate (v_zp) of 20 mm/min. After a dwell time (t_dwell) of 0.25 s, the rod was traversed along the substrate with a lateral traverse feedrate (v_x) of 150 mm/min and a constant axial (plunge) feedrate (v_z) of 60 mm/min. The consumable rod was rotated at a spindle speed (n) of 4000 rpm. The length of all coatings (l_d) was 40 mm (center to center). The parameters were selected based on an initial parametric study performed by the authors to determine suitable processing condition for single-layer deposits based on mechanical characterization. The initial parametric study varied the spindle speed from 1000 to 4000 RPM, traverse feedrate (v_x) from 60 to 210 mm/min, and axial (plunge) feed rate (v_z) from 40 to 150 mm/min. All the tests were done in position control mode on the three-axis CNC mill. The chosen experimental parameters for this study (Table 2) were found to be the most effective in providing uniform single-layer depositions.

The consumable rods as well as substrates were degreased before the tests. Figure 3 pictorially shows the evolution of the friction surfacing process. For multilayered coatings, five layers were deposited using identical process parameters. The temperatures mentioned in the figure are the peak temperatures in the flash around the consumable rod and have been measured at the bottom of the flash. More information on temperature measurement is presented in the following sections. The rods were plunged at the same location for all the layers. The deposited layers were allowed to cool down to room temperature before depositing successive layers. No shielding was used and no post-processing of coatings (e.g., machining, polishing) was conducted.

The bonding rate is of interest because it describes the amount of usable solid material available in the deposition. In most applications, the unbonded edges of the layers will need to be removed to create a functional part.

The workpieces were mounted atop a three-axis piezoelectric force dynamometer (Kistler model 9139AA, Switzerland), which measured the transient process forces. The temperature was measured with two different methods: a two-color pyrometer (IMPAC model IGAR12-LO MB13, Germany) via a fiber optic cable and focusing optic and an infrared camera (OPTRIS model PI400, Germany). The two-color pyrometer was used to measure the temperature of the flash. The pyrometer measures emitted radiation from a 1-mm-diameter spot at wavelengths of 1.28 µm and 1.65 µm, has a temperature sensing range of 350–1300 °C, and a measurement uncertainty of approximately ±6 °C at 1000 °C. The infrared camera was used to monitor the surface temperatures on the retreating side of the different layers during friction surfacing. The camera has a 0.075 °C thermal sensitivity, a spectral range of 8–14 µm, and is set to a temperature range from 200 to 1500 °C with an 80 Hz image acquisition frequency. The emissivity was set as 0.9 according to the literature [30]. The photo measurements of the flash formation were made with a digital single-lens reflex (DSLR) camera (NIKON model D750, Japan) in series image capture mode. The camera has an image size of 6016 × 4016 pixels and was set to an exposure time of 1/20 s, aperture of F/8, and ISO value of 500.

Micro-indentation tests were conducted on a microhardness tester (EMCO, Austria) using a Vickers diamond indenter set to HV1 (load = 9.81 N) with a load application time of t = 15 s. The hardness measurements were performed on cross sections with indents made from the top of the coating across the interface into the substrate with a minimum indent spacing of 250 µm. The measurements on the consumable rod were carried out in the same way.

In addition to hardness measurements, micrographs were made to evaluate the grain size and phase distribution. The friction surfacing coatings (layers) were cross-sectioned in the lateral Y–Z-plane (perpendicular to the traverse direction), in the middle of the coating (∼20 mm traverse distance) to observe the microstructure when the process is in a steady state. The process is at a steady state once the process forces in the z-axis become steady. The cross-sectioned samples were ground successively with 240, 400, 600, 800, 1200, and 2500 grit SiC paper, followed by polishing with 9 µm, 3 µm, and 1 µm diamond slurry, and finally 0.05 µm colloidal silica solution for 15 min. Two different etchants were used to visualize the microstructure. Lichtenegger-Bloech (LBI) etchant (100 mL distilled water; 20 g ammonium hydrogendiflouride; 0.5–1 g potassium disulfite; heated to 37 °C) was used to add color to the images and highlight different phases: e.g., martensite, delta ferrite. Glyceregia (three parts glycerol; two parts hydrochloric acid; one part of nitric acid) was used to highlight the areas with large dislocation density (highly deformed, boundary layers) as well as the regions where pressure and heat had a significant influence (heat affected zone). The samples were imaged using a white light optical metrology system (Alicona, InfiniteFocus^® G4, Austria) and an optical microscope (Zeiss, Axio Imager, Germany).

Friction surfacing coatings deposited in the single and multilayered configuration are compared in this section. A detailed discussion on the deposition morphology, variation in process dynamics, and microstructure are presented. Table 3 provides an overview of the measured physical quantities and calculated performance metrics that are discussed in this section.

Figure 4 shows a cross-sectional view for a single-layer deposition of 304L onto 304L using friction surfacing. As seen from the image, the unbonded regions at the outer edges of the coating account for approximately 34% of the total coating (Table 3). These regions potentially need to be machined off post-process to improve mechanical and corrosion performance of the coatings or access the bonded geometry that is desired. Figure 4 also shows a cross-sectional image of a consumable rod after deposition, where the curling up of flash is observed. The deposition efficiency for single-layer coatings (also the first layer in multilayer coatings) was close to 31% with overall coating efficiency being 20% (Table 3). This means that more of the consumable rod becomes flash than is deposited as a coating on the substrate and brings about the need to find methods for reusing or recycling the material that evolved as a flash. The cross-sectional images in Fig. 4(b) were used to measure the width and height of the flash.

Figure 5 shows a cross-sectional view of multiple layers deposited by friction surfacing where the rod was traveling into the page. All five layers were deposited with the same processing parameters (Table 2). Since the deposited layers are not machined before successive coatings, oxide-rich (black) regions can be observed. These oxide-rich coatings can potentially reduce the mechanical and corrosion properties of deposition, and probably need to be machined off between layers if an inert atmosphere cannot be used during deposition. From Fig. 5 and Table 3, it can be observed that the total and bonded widths remain similar across all layers leading to a relatively consistent joining efficiency. However, the thickness of each successive layer decreases from 0.9 mm for layer 1 to 0.5 mm for layer 5. Additionally, the deposition of material in successive layers shifts laterally towards the retreating side. The change in layer thickness is reflected in the reduction in deposition efficiency from 31% for layer 1 to 17% for layer 5. By observing the coating efficiency, which reflects the fraction of the consumed rod that is bonded to the substrate, a nearly constant behavior over the first three layers is seen (Table 3). The fourth and the fifth layers are then dropping to a value at the fifth layer that is nearly half of the value of the first layer. This indicates that the deposition process is changing, and process parameters would have to be adapted in order to create multilayer coatings with more uniform layer morphology.

Figure 6 shows cross-sectional binary images of 304L consumable rods used to create layers 1 through 5 in one repetition of the experiments conducted in this study. The width as the layers progress, the flash on the consumable rods becomes wider and shorter in height (Table 3). With layers being deposited further away from the large heat sink that is the substrate and associated thermal pathway (boundary condition) changes, it can be hypothesized that a greater amount of the generated heat conducts into the consumable rod and is reflected in the decreasing deposition efficiency and deposition rate.

Figure 3 shows the evolution of flash through a sequence of still images captured by a DSLR camera. As the rod (tool) is consumed more flash is generated, which is seen as a continuous increase in the flash height during deposition. Figure 7 shows the change in flash height as a function of deposition length for all five layers. The slope of flash height differs from the axial feedrate (v_z). The rate of flash growth ranges between 25 and 30 mm/min across all layers. The rate of flash height growth is approximately half the axial feedrate (60 mm/min). This difference is attributed to the curling of the flash and an increased thickness as the deposition progresses. This linear slope of flash height as a function of time is similar across the five layers, suggesting that this variable is not sensitive to the process changes that manifest themselves in thinning and offset layers that are seen in Fig. 5 and Table 3.

Figure 8 shows the initial and undeformed LBI color etched microstructure of both the 304L sheet and rod used in this study. In Fig. 8(a), a section of the 304L substrate (far from the friction surfacing coatings) is shown, with a relatively high content of martensite seen by the areas with a light blue color. The horizontal texture is caused by the high degree of forming (drawing after quenching) that occurred during the production of this 304L sheet. In Fig. 8(b), the initial state of the 304L rod (i.e., consumable tool) with its slightly reduced martensite content is shown. Again, some textures in the microstructure in a vertical orientation (drawing direction) are visible. Even though both materials have a similar chemical composition (Table 1), prior studies show that during high deformation of austenitic stainless steel the creation of an α-phase is boosted and a phase transformation from γ–ɛ–α′ or even directly from γ–α′ could take place [31–33]. In contrast to an ɛ-martensitic microstructure, an α′-martensitic structure shows ferromagnetic properties. Furthermore, the mechanical properties of α′-martensite have a higher strength than the austenitic phase for an identical chemical composition [34].

The hot working conditions during friction surfacing coupled with high strain rates and process dynamics lead to microstructural changes in the consumable rod and deposited material. Figure 9 shows a cross-sectional view of a consumable rod near its centerline. An area of high deformation is clearly visible. There is visible deformation of the initial vertical texture of the drawn rod. When the yield point exceeds the flow stress, due to the high plastic deformation, the formation of dislocations in the material occurs. This is the case in addition to the forming process of the initial material condition as well as due to the friction surface process. The deformation follows the flash formation in the rod. The change in direction occurs at a height of approximately 500 µm from the surface. α′-martensitic areas can even be found in the flash, however, very little evidence of martensite is found in the hot working region of the rod (bottom of Fig. 9). Based on the micrographs, it is hypothesized that within 100 µm of the end of the rod (rubbing interface from Fig. 1) the recrystallization temperature is exceeded resulting in the low content of α′-martensite in this region.

Figure 10 shows an etched cross section of a single-layer coating (i.e., the first layer deposited directly on the 304L sheet). The grain size varies across the deposited layer with the smallest equiaxed grain found at the top of the layer and the largest in the center. The grains at the bottom of the layer, where the bond to the substrate is formed, are smaller than the center but not as small as near the top. There is a significant reduction in both the grain size and size of martensitic inclusions in the austenitic mix in the top and bottom of the layer. These regions extend approximately 100 µm from either edge. In the center of the layer, the grain size reduction is smaller, an observation that has been previously reported [14]. Along with the larger grains are larger martensitic sites in the center of the layer. This variation can be attributed to the difference in strain rates and therefore local temperatures as well as cooling rates experienced in different regions of the deposited layer. α′-martensitic transformations are strongly dependent on these factors. These variations will be investigated in future work using electron back-scattering and x-ray diffraction techniques.

Figure 11 shows cross-sectional views of an LBI etched multilayered coating across all layers. Images were taken from the center of each deposit to focus on the interfacial grain map and martensitic content. The general transition in grain structure is similar for all layers with grains being finer at the top and bottom of each layer as compared to the center.

The intermediatory layers 1–4 follow a very similar grain structure. Due to the thermal and plastic effects of the process, slip bands have formed in the grains, as well as deformation-induced martensite in places (bright blue color). Delta ferrite is present only in very small amounts (non-etched areas that are white in color). At the top of each layer, a high degree of deformation is visible in the microstructure. This shows up as a lamellar structure in the horizontal direction. The transition zones between successive layers show up as very dark-colored horizontal bands that probably contain very fine equiaxed grains and are rich in oxides. The presence of oxides is most likely a result of: (1) these experiments being conducted in the air without any shielding gas and (2) no post-processing (e.g., machining, cleaning) of the deposited layers being conducted.

In the topmost layer (layer 5), which is also the thinnest layer, horizontal lamellar grains are visible. This region of significant deformation extends to a depth of approximately 400 µm from the top of the 500-µm-thick layer. The formation of α′ martensite is visible (light blue color). In the lower part of the layer, larger amounts of martensite are present. This decreases significantly in size towards the top of the layer. The lamellar austenite grains in the top portion of the layer are surrounded by line-shaped white areas. Generally, when using the LBI etchant, the unaffected white regions are either δ-ferrites or carbides. Given the peak temperatures observed during the experiments (<1200 °C), the formation of δ-ferrites is highly unlikely. Previous friction surfacing studies conducted using 304 stainless steel consumable rods have also reported the absence of δ-ferrites in the deposited layers [35,36]. The authors hypothesize these white regions to be grain boundaries enriched with carbides. An in-depth chemical and dispersive analysis will be completed as part of a future investigation to better understand the compositional variations in multilayered coatings of 304L stainless steel.

Figure 12 shows the microhardness measurements for single-layer and multilayer friction surfacing, including the consumable rod and substrate. The background in Fig. 12 is created from micrographs of surfaces that were etched with Glyceregia, showing the heat-affected zone in the substrate. The substrate (304L sheet) and consumable rod (304L rod) have mean bulk hardness values of 322 HV1 and 230 HV1, respectively. The hardness decreases in the hot working area of the consumable rod and the heat-affected zone of the substrate. The deposited layers have an average hardness of 205 HV1 (measured close to the process centerline), which is lower than the initial hardness of the consumable rod. This is primarily related to the lower martensite content in these regions compared to the feedstock material. Prior research on friction surfacing of steels has reported higher hardness values in the deposited layers as compared to the consumable rod, which was primarily attributed to the Hall–Petch effect, i.e., increasing hardness with decreasing grain size [14,15,26]. The explanation for the difference in trends is due to the relatively high martensite content of the feedstock (consumable rod) and substrate (sheet) used in this study as compared with the prior research and the fact that the influence of this phase has a greater influence on the mechanical properties than grain size.

Figure 13 shows the X, Y, and Z process forces and surface temperature of the flash during deposition of a single layer of 304L using friction surfacing. All experiments were conducted using a position control mode of operation on a three-axis CNC mill. The axial (Z) forces are the highest during the initial plunging phase of the process when the rod and substrate are still warming up. Initially, the pressure builds up to increase the amount of frictional heat generation. After the maximum force is achieved, it quickly begins to decrease as the rode and substrate warm up and their yield strength decreases, allowing the material to plastically deform more easily. As the rod traverses, plastic deformation becomes the major source of heat generation, and the axial force stabilizes. This region is therefore termed in the literature as the steady-state region during friction surfacing. The Z-force during steady-state was approximately 3400 N during deposition of the first layer in these experiments. Figure 13 also shows the surface temperature of the flash, which does not vary significantly throughout the plunge or traverse stages of the process. The peak temperature achieved during the deposition is 1150 °C, which is 80% of the solidus temperature for 304L stainless steel.

The transverse (X and Y) forces are an order of magnitude smaller than the axial (Z) forces. The oscillations in X and Y forces have been studied by creating X–Y-plots as a function of deposition time (Fig. 14). During the plunging stage, the amplitude of fluctuations in the X and Y forces is larger. This is a result of the substrate and rod being cold. Visual observations of the process also indicated the rod “hunting” after initial contact with the substrate, as a slender drill bit does. As the rod starts to traverse, the amplitude of force oscillations decreases, with the X–Y-plot (Fig. 14(b)) showing a circular pattern with a magnitude of approximately 350 N in the steady-state region.

The axial (Z) forces for a multilayer coating have been plotted in Fig. 15. With increasing friction surfacing passes the onset of a steady-state region of the deposition (during traverse) is delayed. For the first and second layers, the steady-state region begins very close to the start of the traversing stage. The third and fourth layers do not reach a steady state until approximately 6 s after the traverse begins. The fifth layer does not show a visible steady-state region within the duration of the experiments conducted for this study. The average steady-state force also increases by 400 N from the first to the fifth layer.

Figure 16 shows the X–Y-plots for process forces during the traverse phase of deposition of layers 2–5. Stability, or a steady-state region, is indicated by circular patterns. Layers 2–4 show larger oscillations earlier in the traverse (blues) and a smaller circular pattern towards the end (yellows). In contrast, layer 5 never shows a circular pattern because the process does not reach a steady state within the length of the experiments conducted in this study. Another observation is that the center of the circles appears to be offset from X = 0, Y = 0.

The exact reasons for these changes in the force dynamics during deposition as well as the modest increase in average forces during traverse are not understood at this time. However, they do show some correlation with changes in the layer morphology and microstructure that have been observed. This indicates that process force dynamics could be part of a system to alert the user to changes in deposition quality.

The temperature of flash forming on the consumable tool was recorded using the two-color pyrometer and recorded as a function of time for all five layers. The temperature was measured at the bottom of the flash, where the flash is separating from the material and begins to curl up. The values reported in Fig. 17 are the measured peak values from the pyrometer. The temperature variation trends remain consistent among all the layers, suggesting that the right conditions, including temperature and stress, to create consistent flash are achieved in the rod. The peak and average temperatures increase slightly with successive layers. Whether the measured changes in flash temperature are sensitive enough to changes in the deposition to be useful as a process monitoring tool requires further investigation.

Figure 18 shows the variation in temperatures on the side of each layer and substrate with successive depositions. These temperatures were measured using the IR camera halfway along the deposited length on the retreating side (i.e., right side of Fig. 5). The temperature of the layer being deposited as well as preceding layers increases with successive depositions. Every layer represents a thermal conduction resistance as well as a thermal mass. As more layers are deposited, the thermal conduction resistance and thermal mass increase. Therefore, whether one analyzes the heat transfer as transient or steady state, for an assumed constant heat input it will lead to higher peak temperatures and take more time for the deposited energy to diffuse away from the top layer that is being deposited. For instance, the temperature of the fifth layer during the fifth friction surfacing pass (980 °C) is larger than the temperature of the fourth layer in the previous pass (960 °C). At the same time, the temperature of the preceding (fourth layer) in the fifth pass (950 °C) is larger than the temperature of the third layer in the fourth pass (890 °C). This shows that even after a couple of ∼1 mm thick layers, the thermal resistance through the layers is increasing enough to measure a change in temperatures on the outside of the build.

Multilayer (i.e., five layers) friction surfacing was performed using stainless steel 304L consumable rods over the substrate of the same material. The process parameters were consistent across all layers, and in-process measurements of forces and temperatures were recorded. Based on the various mechanical, microstructural, and morphology studies conducted, the significant findings from this study are:

• Cross-sectional analyses of multilayered coatings showed consistency in the total width and bonded width of deposited layers. However, the coatings’ thickness and the flash height were reduced with subsequent layers, resulting in reduced deposition efficiency, coating efficiency, and bonding rate.

• Microstructural analyses showed grain refinement across all deposited layers with grain sizes being smaller at the top and bottom of each layer compared with the center.

• The presence of an oxide layer between consecutive layers was observed.

• Measurement of in-process forces showed an increase in the average process forces with subsequent layers. The X–Y force plot also showed increasing instability in the process with upscaling.

• The peak temperatures measured near the interface in the flash around consumable rod remained consistent. However, the changing thermal boundary conditions were visualized with temperature measurements in the coatings and substrate with the layer buildup getting warmer with subsequent depositions.

This study shows that friction surfacing is a viable technology for producing multilayered coatings with 304L stainless steel. However, due to changes in the thermal boundary conditions, even within the first five layers, process parameters need to be adjusted for successive layers. In addition, the efficacy of shielding gas at mitigating the oxides between layers needs to be studied. Going forward, an in-depth study of microstructural changes, mechanical properties, and process stability of multilayered coatings will be useful in understanding the feasibility of manufacturing these structures for commercial applications.

The authors would like to acknowledge the partial support of this work by the U.S. Department of Energy (Grant DE-NE0008801), the Austrian Marshall Plan Foundation endowed professorship (FFG Project Number 846946), the members of the Institute of Production Engineering and Photonic Technologies at TU Wien who helped setting up the experiment’s setup, and the Institute of Materials Science and Technology at TU Wien for their advice and help with the etching.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.

(See Sec. 2.3 for a detailed explanation of each term)

n =

spindle speed in rpm

d_rod =

consumable rod diameter in mm

h_d =

thickness of deposition in mm

h_f =

height of flash in mm

l_d =

deposition length in mm

t_dwell =

dwell at end of plunge in seconds

v_x =

lateral traverse feedrate in mm/min

v_z =

axial feedrate in mm/min

v_zp =

plunge rate in mm/min

w_b =

bonded width of deposition in mm

w_d =

total width of deposition in mm

w_f =

width of flash in mm

z_p =

initial plunge in mm

A_d =

cross-sectional area of coating in mm²

A_r =

cross-sectional area of rod in mm²

B_rate =

bonding rate in kg/h

C_eff =

coating efficiency in %

C_rate =

consumption rate in kg/h

D_eff =

deposition efficiency in %

D_rate =

deposition rate in kg/h

J_eff =

joining efficiency in %

ρ =

density in g/mm³