Friction stir processing of austenitic stainless steel cold spray coating deposited on 304L stainless steel substrate: feasibility study

In this work, the feasibility test of friction stir processing (FSP) of 1.5-mm-thick austenitic stainless steel cold spray coating deposited on 304L stainless steel substrate was performed using tungsten carbide tool. Applied FSP parameters (advance speed 50 mm/min, rotation speed 300 rpm, axial force 20 kN, tilt angle 1.5°) allowed to perform FSP treatment with a higher depth than the coating thickness. As a result, the material mixing at the coating-substrate interface was observed. The microstructure observation revealed that the coating microstructure in the stir zone was significantly modified. EBSD analysis confirmed that full material recrystallization during FSP allowed the formation of dense and uniform fine-grained structure with the mean grain size of 1.9 mm. Average coating microhardness was decreased from 406 to 299 HV. Further FSP parameter optimization should be carried out in order to improve the process reliability and avoid any coating failure during treatment.


Introduction
Austenitic stainless steels are widely used for different industrial applications thanks to their excellent corrosion resistance in various aggressive environments and high mechanical properties at elevated temperatures. For example, 304L and 316L stainless steels are used for the production of pressure tubes in thermal and nuclear power plants and connection pipes and other parts in chemical factories. The microstructure and mechanical properties of the austenitic stainless steels in the welding joints could be modified by micro-alloying and local heat treatment [1,2]. Austenitic stainless steel parts could be joined with dissimilar material such as aluminum alloys using alternative welding techniques such as explosive welding [3]. However, after some service time, some component damage may occur due to wear, fatigue, or other causes. Damaged stainless steel components could be replaced or repaired, depending on the nature of failure, the component dimensions, and complexity. Nowadays, different material deposition techniques involving high-temperature processes like welding and laser cladding are widely used for stainless steel component repair [4][5][6]. The main advantages of these methods are high quality of the interface between the part and deposited material, high process robustness, and acceptable cost. However, in some cases, the application of these methods is difficult due to significant thermal impact. In particular, in case of repair of parts with complex geometry, the extensive heating of the deposition site induces high thermal stresses that could distort the repaired part.
Cold spray is the solid-state material deposition process providing an alternative to the high-temperature technologies. In this process, the supersonic jet of preheated gas formed by converging-diverging nozzle accelerates particles of metal powder. The particle adheres to the surface due to local material recrystallization induced by high-speed plastic deformation during impact. As a result, the deposit consisting of highly deformed particles forms [7][8][9]. Nowadays, cold spray is applied for the repair of parts produced from the heat-sensitive materials like aluminum and magnesium alloys [10]. For example, successful application of cold spray for restoration of aircraft skin was reported [11,12]. However, it is important to note that the properties of cold spray deposit significantly differ from the reference values measured for the bulk material produced in traditional way. In particular, due to specific structure of the particle-particle interface, the ductility of cold spray deposits is very low [9,13]. In addition, the elevated porosity diminishing the tensile strength of deposit is observed in case of deposition of hard materials [13]. Results of successful cold spray deposition of austenitic stainless steel are available in literature [14][15][16][17][18][19][20][21][22][23][24][25][26][27][28]. It was found that the deposition efficiency of austenitic stainless steel coatings depends on the initial powder microstructure [16]. Obtained coating could have elevated porosity in case of spraying with nitrogen [14,15,20,21]. Fragile behavior of the coating at tensile strength tests, typical for cold spray deposits, was also reported [18]. Obtained cold spray coatings could be applied as corrosionresistant coatings in nuclear industry [23] and biomedical industry [24,25].
Previous studies showed that heat treatment or hot isostatic pressing could significantly improve the mechanical properties of cold spray deposits [24][25][26][28][29][30][31][32]. In these processes, the high temperature enhances the material diffusion through the particle-particle interface and leads to formation of joint grains between neighbor particles. For example, Al-Mangour et al. [24] showed that annealing of austenitic stainless steel coating annealing at 1000°C allowed formation of fully recrystallized uniform structure. Similar conclusion was made by Yin et al. [28]. The mechanical properties of stainless steel cold spray deposits such as ductility and toughness significantly increase after heat treatment or HIP [28]. However, in the case of restoration of large-scale parts, these post-treatment techniques are not always applicable due to dimension restriction of heat treatment and HIP equipment.
In this regard, application of friction stir processing (FSP) offers a promising alternative to the thermal post-processing methods. In FSP process, a rotating tool inserted into the material generates the frictional heating contact. During tool movement along the contact line, plastic material occurs at a temperature below the melting point as for cold spray. The plastic flow transfers the material from the advancing side to retreating side, leading to the formation of a stirred region [33,34]. From a physical point of view, the FSP process is very close to friction stir welding process where the plastic material flow induced by rotating tool allows strong material joining even in case of non-similar alloys [35,36].
Formation of a fine grain structure during FSP of metals is attributed to the dynamic recrystallization phenomena combined with thermomechanical treatment [37,38]. Besides the process parameters such as tool geometry, rotation speed, and normal force, the grain size of the material after FSP treatment could be also controlled by localized modification of material composition. For example, introducing ceramic nanoparticles in the stir zone allows formation of ultrafine structure in copper [39].
Nowadays FSP is used to refine the surface microstructure of various bulk materials such as bronze [40], aluminum, and magnesium [41,42]. Application of FSP for the treatment of harder materials like steels and nickel-based alloys is complicated. In the case of proper parameter selection, the FSP treatment can increase the microhardness on the austenitic stainless steels' surface by formation of nanosized grains [43]. At present, many articles devoted to the friction stir processing of cold spray despots are available in literature [44][45][46][47][48][49][50][51][52][53]. The major part of these works deal mainly with FSP treatment of relatively soft materials like aluminum and copper alloys [44][45][46][47][48][49]. It was shown that friction stir processing allowed to completely modify the microstructure of cold spray deposits. In particular, full material recrystallization and grain refinement were observed [44][45][46]. No particle-particle interface is visible in the deposits after FSP [46][47][48]. The improvement of tensile strength and ductility was also reported [48,49]. As it was mentioned earlier, the friction stir processing of steels and other metals with high mechanical properties is difficult due to high tool load required for proper material deformation during tool rotation. However, some experiments with FSP of titanium and stainless steel cold spray coatings are also presented in the literature [50][51][52][53]. In these works, the authors successfully demonstrated the feasibility of FSP treatment of hard material deposits. Significant microstructure modification and material grain refinement were observed. It is important to note that in all reported cases with FSP treatment of stainless steel deposits, the depth of the coating layer affected by FSP treatment did not exceed 200-300 mm [54,55]. However, from practical point of view, in the case of application of cold spray for the stainless steel part repairing, it could be necessary to perform the FSP treatment of the deposits with significantly higher thickness. The main purpose of this work was the feasibility study of friction stir processing of austenitic stainless steel cold spray deposits with the depth affected by the treatment equal to 1 mm or more. The parameter selection of FSP process and its influence on the structure and the properties of the deposit after treatment were also considered.

Cold spray: materials and deposition parameters
Commercially available austenitic stainless steel powder Diamalloy 1003 supplied by Oerlikon Metco was used for coating deposition. The SEM images of the powder are presented in Fig. 1. Powder composition is provided in Table 1.
The chemical composition of the Diamalloy 1003 powder was close to the standard composition of 316L austenitic stainless steel, except the percentage of silicon. Increased content of silicon in the powder leads to the higher oxidation and sulfidation resistance, but on the other hand decreases ductility of the alloy.
The cold sprayed coating deposition was performed with the commercially available system CGT KINETIKS 4000 equipped by the nozzle OUT-1 provided by Impact Innovation (Germany). Nitrogen was used as the working gas. The process parameters used in this study are summarized in Table 2. In the table, P and T denote the gas stagnation pressure and temperature, respectively. V s is the scanning velocity of the nozzle, d is the distance between two passes, and n is the number of layers. These parameters allowed to deposit the Diamalloy 1003 coatings with deposition efficiency about 75%. The coating thickness was in the range between 1.4 and 1.5mm. The strategy of the nozzle displacement as well as the geometry of the final samples is presented in Fig. 2. The rectangular 304L stainless steel plates were used as the substrates. The composition of the plates is presented in Table 3.
Initially, the tests with coating deposition were performed using three different surface preparation techniques: acetone cleaning (i), grit blasting (ii), and chemical machining (iii). In the case of grit blasting, alumina grains with particle size mesh 16 accelerated by air jet with stagnation pressure of 10 bars were used for surface roughening. In accordance to surface roughness measurements performed by Alicona InfiniteFocus profilometer, the mean roughness of the grit blasted substrate Sa was equal to 17.6 μm. The chemical machining was performed using ferric chloride solution. Approximately 200 μm of the material was chemically removed from the top surface

Friction stir processing: tool and processing parameters
The FSP experiments were performed using a dedicated Friction Stir Welding gantry machine (MTS-ISTIR) with a tungsten carbide tool. As shown in Fig. 3, the tool geometry is composed of a 15-mm diameter at shoulder with a 1.4-mm-long tapered pin. The length of the pin was chosen to be sufficient for stirring all the thickness of the Diamalloy 1003 cold spray coating down to the interface with the 304L substrate. Figure 4 shows the positioning of the specimens and the clamping conditions as well as the tool during the plunging phase of the pin. Pure argon was used as shielding gas in order to prevent the high temperature oxidation of the tool and material during processing. The experiments were conducted at force control mode. At this mode, the normal force is maintained at constant value during treatment, and the penetrating depth depends on the normal force, material properties, and the tool geometry.

Deposition of cold spray coatings
Preliminary tests revealed that in the case of spraying on the acetone-cleaned and grit-blasted substrate, the coating immediately delaminated from the substrate surface after deposition due to weak adhesion strength and high residual stresses. However, in the case of deposition on the substrate after chemical machining, the coatings without visible defects were produced (Fig. 5).
Higher coating adhesion on the substrate prepared by chemical machining could be explained by the lower hardness value at the substrate surface. The chemical machining allowed to remove the material from near-surface zone affected (hardened) by rolling process and machining during substrate manufacturing. As a consequence, the spraying is performed on softer surface than in case of raw or grit-blasted substrates. The results of nanohardness measurements on the substrate before and after chemical machining are presented in Fig. 6. Figure 6 shows that the nanohardness in the nearsurface zone of the substrate dropped from more than 4GPa for the raw plate to 3.3 GPa for the substrate after chemical machining. The SEM images of the top surfaces of the raw substrate after acetone cleaning and chemical machining are  Fig. 7. One can see that in contrast to the raw substrate, the grains are well visible on the surface of the plate after chemical machining. This observation confirmed that the chemical machining allowed to remove the superficial layer of the plate that leaded to adhesion improvement.
The detailed study of the physical mechanism of coating adhesion improvement by application of chemical machining was not the purpose of this study. However, it is important to note that the deposition of the final batch of samples for further FSP treatment was done using chemically machined substrates.

Determination of FSP parameters
In the case of processing of stainless steel, the proper selection of FSP parameters is a complex task. Miles et al. [54] and Selvam et al. [55] considered an advancing velocity V between 50 and 150 mm/min, whereas the rotating speed N achieved 250 rpm for a pin length greater than 6mm in the case of processing of bulk stainless steel. Hajian et al. [56] used a tool without pin with a rotating speed up to 1800 rpm by advancing at 20 mm/min for 316L FSP treatment. For the processing of 2-mm-thick plates made of the same 316L steel, Chen et al. [57] applied the tool rotation speed in the range between 200 and 315 rpm with advancing velocity taken equal to 63mm/min The FSP process parameters were defined during the series of preliminary tests. In these tests, the rotation speed and advancing speed were varied from 50 to 100 mm/min and 200to 500 rpm correspondingly. Tilt angles 0°, 1.5°, and 3°were tested. The normal force F z was varied from 15 to 25 kN. The cold spray coating was extremely sensitive to the process parameters. Fragile coating failure in the zone of treatment was observed in most of the cases. It was found that processing at F z = 20 kN, V = 50 mm/min, N = 300 rpm, and tilt angle i =1.5°allowed avoiding the total failure of coating in the treatment zone. However, local coating delamination and failure at the non-treated zone at sample periphery were observed. At F z = 20 kN, V = 50 mm/min, N = 300 rpm, and tilt angle equal to 0°, the coating integrity was also preserved, but the coatings had open defect related to the lack of mixing on the advancing side (Fig. 8).
After a series of preliminary tests, the FSP parameter set allowing to perform the coating treatment without coating delamination was established (Table 4). Further structure analysis was performed for the samples produced using these FSP parameters.

Cross-sectional analysis of as-sprayed and FSPtreated deposits
The cross-sectional micrographs of as-sprayed coatings taken at different magnifications are presented in Fig. 9. Microscope observations did not reveal any cracks or other macroscale Fig. 5 Overview of the Diamalloy 1003 cold spray coating deposited on the 304L substrate Fig. 6 Results of nanohardness measurements performed in the near-surface zone of the substrate defects in the coating. The interface between the coating and substrate is well distinguishable; however, no signs of coating delamination were observed. At the same time, the coating had high residual porosity well visible in the lowmagnification cross-sectional image. The porosity value was not measured in this study; however, high porosity of stainless steel cold spray deposits (up to 22%) sprayed using nitrogen was previously reported by Villa et al. [14], Sova et al. [21], and Adachi et al. [58].
High-magnification images revealed that the coating had a typical structure for cold spray coating consisting in highly deformed stainless steel particles (splats) with visible interface. High porosity of the coating is explained by insufficient particle deformation and, as a consequence, weak bonding between splats.
Friction stir processing significantly changed the structure of deposit. Typical cross-sectional image of the coating after FSP is provided in Fig. 10. A difference in gray level allows to clearly identify the substrate and coating material. Crosssectional observations show that the depth of material affected by FSP was significantly higher than the coating thickness. Taking into account the length of the pin (1.4mm), it is well seen that the stirring zone is deeper than the penetration depth of the pin. Therefore, the coating-substrate interface was affected by the FSP, and the material transfer from the substrate to the coating is well visible. The material transfer was especially intensive in the advancing side where substrate material reached the top of the coating. Practically no defects were observed on the coating-substrate interface. At the same time, some cracks in the coating-substrate interface were detected in the regions not affected by FSP. The coating structure in the stirring zone was completely modified due to intense material deformation. No splats and inter-splat porosity are visible. It was also observed that the interface between the affected and non-affected zones was well distinguishable on the cross-sectional images after strong chemical etching, especially at advancing side (Fig. 11).

EDS analysis
The element composition of as-sprayed and FSP-treated coating was performed by EDS analysis. The measurements were performed in spots located at the middle of the coating, at coating-surface interface, and at the substrate at the distance near 200 μm from interface. The results are presented in Fig.  12. As expected, the compositions of as-sprayed coating and the substrate were close except the Mo and Si contents.
As it was mentioned above, the FSP treatment induced intensive material flow; however, the EDS analysis did not reveal any important diffusion of Mo from the coating to the substrate. Its content remained significantly higher in the coating zone than in the substrate (1.85 wt% and 0.4wt% Fig. 7 SEM images of the substrate surfaces; a raw substrate after acetone cleaning, b substrate after chemical machining Fig. 8 The coating overview after single pass of FSP treatment at V =50 mm/min, N = 300 rpm, F z = 20 kN, and tilt angles 1.5°and 0°2 correspondingly). However, the new elements such as cobalt and tungsten were detected in the coating and under the coating-substrate interface after FSP treatment. The apparition of Co and W is explained by material diffusion from the WC-Co FSP tool to the coating material due to high temperature at the tool-coating contact zone. Apparition of these elements indicates an important wear of FSP tool during treatment.

Microhardness
The hardness map of coating is presented in Fig. 13a. One can see that the coating microhardness varied from less than 100 to 550HV. Such hardness variation could be explained by significant coating porosity. The maximum measured value (598HV) indicates strong work hardening of the particles during high-velocity impact and further plastic deformation. Elevated microhardness values in cold sprayed austenitic stainless steel coatings were previously reported by Chen et al. [59]. The substrate microhardness was almost constant. The measured value was in the range 250-300 HV, in good correlation with nanohardness measurements of the substrate before coating deposition.
The microhardness map after FSP is presented in Fig. 13b. The coating microhardness was significantly decreased and homogenized in the FSP-affected zone. Hardness in almost all zone lies in the range 300-350 HV. High microhardness values equal to 400-420 HV were measured in the area where the material transfer from substrate to coating was especially pronounced. However, these peak values were significantly lower than the maximum hardness measured in as-sprayed deposit (598 HV). The microhardness in the non-affected parts of the coating was higher than that in the FSP affected zone (with the peak values~550 HV). Figure 14 presents the EBSD measurements on the Cross Stirring Direction (CSD)-Normal Stirring Direction (NSD) plane in the as-sprayed coatings which confirmed the presence of high number of small grains inside the splats that is typical for cold spray deposits. The interface between substrate and coating is well distinguishable on the EBSD image due to significantly higher grain size and absence of sub-grains in the substrate material as presented in Fig. 14b. In addition, Fig. 14c reveals the presence of sub-grains in the coating through local EBSD map in the coating corresponding to the white dashed region in Fig. 14b that could be explained by strong particle material deformation at impact.

EBSD analysis on CSD-NSD plane
EBSD maps made on different zones of the coating after FSP are shown in Fig. 15. The images were obtained on the intersection plane perpendicular to the advancing direction. A very clear microstructure made of small grains with grain size of about 1.9 mm with almost no substructure nor twin boundaries is revealed, supporting the idea that the material ig. 9 Cross-sectional optical images of as-sprayed Dimalloy 1003 coatings on the 304L substrate at different magnifications: a overview; b ×200; c ×500  Cross-sectional micrographs of Diamalloy 1003 coatings on the 304L substrate after FSP taken at the interface between affected and nonaffected zones of the coating dynamically recrystallizes during FSP. The grain structure in the treated coating and in the substrate was very similar with the consequence that the coating-substrate interface could not be clearly identified as the respective grain sizes of these regions are similar: the average grain sizes measured in the affected zone of the coating and in the substrate were 1.9 mm and 2.5 mm, respectively.
Pole figures (f100g, f110g, and f111g planes) obtained from EBSD maps are illustrated in Fig. 16 in the reference frame of FSP, i.e., stirring direction, cross stirring direction, and normal stirring direction. Texture differences were not found between the half-thick of coating and the interface with the substrate. This observation confirms that recrystallization process was similar across the coating thickness.

Discussion
The experiments presented above demonstrated the feasibility of FSP of austenitic stainless steel cold spray deposits with thickness superior to 1 mm. At the same time, several important issues influencing the process reliability were identified. The fragile failure of the coating at the periphery was regularly observed. This phenomenon could be explained by the Fig. 12 Element composition of as-sprayed and FSP-treated coating measured by EDS analysis Fig. 13 Microhardness of the assprayed (a) and FSP-treated (b) Diamalloy 1003 coatings on the 304L substrate structure of cold spray deposit (Fig. 17). Any metallic cold spray deposit has specific two-level structure where severely deformed metal particles (splats) are in close contact. Each splat has a normal grain structure, with larger grains at the splat center and smaller grains at the splat periphery. The interface between splats has complex structure containing sub-grains, dislocations, pores, and oxides [8,9]. It is also known that the cohesive strength of the cold spray coating is mainly defined by the quality of the contact between splats. The micro-tensile tests revealed that the ductility of the material in the splat-splat contact is very low in comparison with the primary material in the splat center. In addition, the ultimate tensile strength of the core splat material is also higher than the cohesive strength between splats [60]. Consequently, during the tensile and compressive tests in macroscale, the fragile failure of deposits at the splat boundaries is regularly observed.
During the FSP treatment, high normal load and shear stress in the stir zone rapidly induce the viscoelastic flow of material in the splats. The fragile failure in the stir zone is avoided due to rapid material recrystallization, erasing the splat boundaries and forming uniform small-grained structure (Fig. 18). The coating adhesion to the substrate is also improved by intensive material mixing and recrystallization at the coatingsubstrate interface. At the same time, the coating periphery also experiences strong rise of the stress induced by the tool (lateral forces, torque, as well as thermal stresses). However, in contrast to the stir zone, the material recrystallization at the periphery does not occur, and the deposit keeps the initial two-level structure. As a consequence, if the tool-induced stresses become higher than the tensile strength of the splat-splat and splat-substrate contact, the coating failures. Two types of failures can happen and have been experimentally observed: cohesive failure at the splat-splat interface characterized by crack Fig. 18 Schematics of coating failure during FSP process Fig. 19 Consecutive deposition and further FSP treatment of the single lines of coating propagation in the coating and adhesive failure at the splatsubstrate interface noticed as a delamination of the whole coating.
The obvious way to avoid the coating failure is to increase the coating tensile and adhesive strength. It is known that the mechanical properties of the cold spray coating depend on the particle impact velocity and temperature. In general, increase of these parameters allows to improve the coating UTS and adhesive strength. Assadi et al. [9] introduced the parameter η characterizing the ratio between critical velocity and particle impact velocity: Critical velocity v critical is the minimal velocity that the particle should have in order to bond to the substrate. Authors also showed that the highest possible cohesive and adhesive strength of the coating corresponding to the strongest splat-splat contact is achieved if it lies in the range 1.5-1.8 [9]. It is also known that in the case of spraying of 316L powder with nitrogen as the working gas at the maximum possible spraying parameters (P=50 bars, T=800°C), the η parameter does not exceed 1.25 [9]. Thus, significant improvement of adhesive and cohesive strength of the austenitic stainless steel cold spray coating is not achievable if the nitrogen is used as the working gas. Usage of helium instead of nitrogen allows to increase the impact particle impact velocity almost in two times under similar gas stagnation temperature and pressure [7]. The UTS of the coating sprayed with helium could be similar to the reference value obtained for bulk material. For example, Coddet et al. [18] reported that UTS of 304L stainless steel coating sprayed with helium at 28 bars and 500°C was near 600MPa. However, the usage of the helium significantly increases the cost that reduces the industrial interest to this option [10].
Another approach that potentially could minimize the risk of coating failure is the progressive coating deposition with consecutive FSP treatment. One can propose that the FSP treatment should be applied at moment when the coating width does not exceed the diameter of FSP tool. The coating width could be further increased by the consecutive deposition and immediate FSP treatment of single coating lines attached to the FSP-treated zone. This approach is schematically illustrated in Fig. 19.

Conclusion
In this study, the feasibility study of friction stir processing of 1.5-mm austenitic stainless steel cold spray coatings deposited on 304L stainless steel substrate was performed. Taking into account the obtained results, the following conclusions could be drawn: & Friction stir processing of relatively thick (1.5 mm) austenitic stainless steel cold spray coating is feasible. Approximate parametric window of FSP treatment was experimentally determined. However, further parameter optimization as well as the treatment strategy should be done in order to avoid the coating failure during treatment. The issue with the tool heating during treatment should be also resolved by application of proper cooling system. & The depth of the zone affected by FSP treatment was slightly higher than the coating thickness. The mixing of the substrate and the coating material at the interface was observed. This effect should significantly influence on the adhesive strength of the coating. However, the detailed analysis of the adhesive strength was not performed in this study. & Material recrystallization in the stir zone allowed to modify the coating microstructure. Instead of splat structure with significant amounts of pores and defects, a uniform fine-grained structure was formed. The microhardness maps confirm remarkable improvement of the uniformity of coating microstructure in comparison of as-built coating. & Further research should be devoted to the investigation of the coating material properties after FSP treatment. In particular, the mechanical properties of the coating as well as the bonding strength should be analyzed.
Availability of data and material The authors declare that all data supporting the findings of this study are available within the article. No supplementary data are available.