Mechanical and Microstructural Characterization of Plasma Welded Joints of 18 Ni-300 Maraging Steels Subjected to Repairs

Department of Materials, Faculty of Engineering of Guaratinguetá-Paulista State University“Júlio de Mesquita Filho”, Brazil Department of Aerospace Science and Technology-DCTA, Aeronautics and Space Institute-IAE, Brazil Institute for Advanced Studies-IEAv, Department of Aerospace Science and Technology-DCTA, Brazil Applied Physics and Mathematics, Post-Graduation in Space Sciences and Technologies-PG/CTE, Technological Institute of Aviation, ITA, Brazil


Introduction
The material used to manufacture a rocket motor envelope must have specific functional and technical characteristics with high performance under severe conditions. It also needs to combine excellent mechanical properties, mainly resistance and tenacity, with a low specific mass (Rajan and Narasimhan, 2002). In the case of satellite launchers, for example, the payload reaches less than 5% of the total weight (Bruhn, 1967), making it essential to reduce the vehicle's mass.
Among the candidate materials for aerospace use are composites and metallic materials. Composite materials (filament-reinforced plastic) have lower density and lower cost, however, metallic materials are more robust, have better ductility, can flow before failure and withstand higher working temperatures (700 to 1000°C).
Maraging steels can be used in the transport, manufacturing, tools, electromechanical sector and, mainly, for the nuclear and military industry, being one of the most suitable for the aerospace sector (Sun et al., 2020;Reis et al., 2015). These steels are ultra high strength (yield strength above 1380 MPa), have good corrosion resistance, good creep resistance and heat treatment facility to optimize mechanical properties (Chakravarthi et al., 2017;Schmidt and Rohrbach, 1991;AMS 2759AMS /3E, 2008AMS 6514G, 2010;Vasudevan et al., 1990). Hall and Slunder (1968) report the use of maraging steel for both small rocket engines with a diameter of 406.4 mm and walls with 1.78 mm (HARP Project-High Altitude Research Program) and for large motor envelopes, as tested by NASA, with a diameter of 6604 mm and thicknesses from 10 to 15 mm. In the Brazilian case, the last satellite launch vehicle designed has one of the propulsion stages, called S43, with a diameter of about 1000 mm, 6500 mm in length, the thickness of the steel walls is about 3.3 mm. These are formed by calendered plates and plasma welded to form the cylinder of the motor envelope. This structure has the specification for a tensile strength limit of 1765 MPa and a hardness of 52 to 55 HRC (Sakai, 2015).
Several studies have been indicating for decades the superiority of two types of steels for rocket application, D6AC and Maraging. These alloys show similarities in the properties of specific strength, fracture toughness, specific stiffness, machinability, cost and coefficient of thermal expansion, however, with advantages for Maraging steel when considering corrosion resistance and stress corrosion (Rajan and Narasimhan, 2002;Poinke and Garland, 1973;Matsuo and Kawaguchi, 1995;Suresh, 2008).
For the aerospace sector, the combination of high strength-to-weight ratio and good fracture toughness of Maraging is essential and proves to be superior when compared to other high-strength steels (ESA, 2004;ASM, 2019). Due to its low carbon content, it presents good weldability and dimensional stability (Galindo-Nava et al., 2016). One of the important factors that increases the hardness and resistance of Maraging 18Ni steel, without marked loss of ductility, is the aging hardening process. Aging occurrs mainly in the range of 455 to 510°C, below the reversion temperature for austenite (Reis et al., 2016;Pardal et al., 2006), allowing the uniform precipitation of various intermetallic compounds such as Ni3Ti, Ni3Mo and Fe2Mo (Fanton et al., 2014;Dos Reis et al., 2015).
The welding process is not free from defects, nondestructive analyzes performed by Xrays and dyes must analyze the welded region and eventually, in cases of defects not acceptable by technical standards, the element must be rejected or if possible repaired. However, the repair process involves removing part of the previous weld, with the defect and applying a new weld layer, involving a new thermal cycling and alteration of the region's microstructure (Sakai et al., 2015).
Complex processes are activated when carrying out repairs, as they involve new fusion and solidification with formation of dendritic structure, redistribution of solutes and transformation of phases. The heterogeneities in the chemical composition with the enrichment of elements such as Ni, Co, Mo and Ti, can lead to the formation of retained austenite, with a localized reduction in the properties of steel (Pardal, 2012). There is also the phenomenon of grain growth and austenite reversal, which reduce hardness and mechanical properties. In regions where the temperature reaches values close to 500°C there is the precipitation of intermetallic, with a localized increase in the hardness value (Reis et al., 2016). Due to all these phenomena, it is essential to analyze in detail the microstructure and the influence of the thermal cycles of each repair used Sakai et al., 2016;de Lima et al., 2019). Marcisz et al. (2017) studied the influence of temperature and aging time on the mechanical properties of Maraging 18Ni 350 steel, this same aging will also occur due to the thermal cycling of the welding process or welding repairs. Depending on the distance of the microstructure to the fusion zone, a corresponding temperature range will be developed and the localized phenomena depend on the chemical composition and the times and temperatures developed in the region (Blondeau, 2008).
This study was carried out to try to reduce the gap in the literature regarding the effects of welding repairs carried out on aerospace structures. Some works with weld repairs for other types of materials and applications, such as stainless steel, DH36 steel, aluminum alloys and titanium show the technical feasibility of carrying out repairs, provided that respected the peculiarities of the welding process and subsequent heat treatments, however, losses in resistance and ductility properties are generally observed (Teng et al., 2017;Kanne Jr et al., 1999;Reimann et al., 2017;He et al., 2017). Few articles were found for Maraging steel, (Sinha et al., 1993) used the Maraging 250 and performed up to two repairs per weld, finding a loss in mechanical strength of 5 to 6%. Ahmed et al. (2012), using the same steel, observed a 10% reduction in strength after making two repairs and aging the alloy. In this study, we sought to deepen the understanding of the microstructural processes involved in the repaired region. After welding, up to three repairs were made to the weld bead of Maraging 18Ni steel (300) and the changes in the microstructure and mechanical properties resulting from this interference were analyzed.

Materials
The material used was Maraging 18Ni steel (300) produced by BÖHLER in the form of a 3.35 mm thick plate, which was hot rolled and solubilized (hot rolled, annealed). In the welding process, the MAR300 steel alloy in the form of wire addition material with a diameter of 0.889 mm, according to the AMS 6463D standard (AMS 6463D, 2009). Table 1 shows the chemical composition of Maraging steel sheet and filler metal used in welding and carrying out repairs.

Welding
From the sheet received from Maraging 300 steel, pieces 400 mm long and 100 mm wide were cut in the transverse direction. The surfaces next to the weld were cleaned with a cloth and acetone to remove contaminations and oxides.
The welding was carried out similarly to what is done for the manufacture of the cylindrical sections of the metallic motor envelopes of the Institute of Aeronautics and Space IAE (Brazil) called longitudinal welding (90° with the lamination direction). This is the most critical position, because the weld supports the greatest loads during a launch. The process was carried out continuously with filler metal in a single pass and in the flat position. In Fig. 1a some cut sheets are shown and in Fig. 1b the strips prepared to receive plasma welding.
For the welding, PAW was used with the technique called keyhole and butt welding. The welding equipment was used in direct non-pulsing current. The parameters used are shown in Table 2.

Weld Repairs
After the welds, the repairs were carried out as shown in Fig. 2a, with an length of 80 mm. This is sufficient to machine four specimens according to the (AWS B4.0, 2007) standard. Among the regions that received repairs, a space was left to remove specimens to verify the effect of the heat emanated from the repairs carried out.
The opening in the weld bead for the repair was carried out with an abrasive disc with a thickness of approximately 3 mm (Fig. 2b). The depth of 2.3 mm (70% in depth) was removed leaving 1.0 mm of the original cord. Subsequently, the channels were prepared with a rotary sander (60 mesh) and cleaned with acetone. The repair was made with the filler metal indicated in Table 1 by GTAW with a current of 50 A and a flow rate of the shielding gas (argon) of 8 l/min.
The samples were prepared with one, two or three repairs at these locations ( Fig. 2), afterwards, the microstructural changes were analyzed and specimens were made to evaluate the mechanical properties.

Heat Treatment and Mechanical Properties
From the welded and inspected joint specimens were manufactured for the tensile tests according to the (AWS B4.0, 2007) standard by the wire EDM process. In Fig. 3a the coupon drawing is shown and in Fig. 3b the specimen cut by EDM, leaving the weld bead in the center of the specimen. Four specimens were tested for each condition studied.
The specimens, without repair and with the repairs, were submitted to the heat treatment of solubilization and aging. Following recommendations from (AMS 6521C, 1991). Solubilization was carried out at a temperature of 815±5°C with a 60 min stay in the oven in an uncontrolled atmosphere. For the thermal treatment of aging the steel was heated to a temperature of 480±5°C with a period of 3 h in the oven in an uncontrolled atmosphere. In both processes, the heating rate was 200°C/h and cooling was done in air up to 25°C.
For the performance of the tensile tests, the servohydraulic traction machine type MTS model 810.23 M with 250 kN of capacity was used with the test speed of 1 mm/min. At least three coupons were tested for each condition studied.
Vickers Hardness measurements (HV) were also made by micro-indentation on samples of all studied conditions. The equipment used was the FUTURE TECH microhardness model FM-700 with a load of 300 gf and with a penetration time of 15 sec. Figure 4 shows a schematic of the measurements taken in two directions: (a) Verticalalong the bead from the root to the top of the weld (line A-A) and (b) horizontal-from the center of the weld bead in towards the Base material (BB line). The space between two consecutive measurements was 0.25 mm.

Microstructural Characterization
For the microstructural analysis. samples were taken from the welding and repair regions. These were embedded in bakelite, sanded with granulometries up to 800 mesh and then polished with alumina solution (1 μm) and with colloidal silica solution (0.1 μm). For the exams, the optical microscope model AXIO IMAGER 2 from ZEISS was used.
For microscopic analysis, different reagents were used depending on the purpose of the examination. Table 3 shows the different solutions used.
The geometric measurements of the weld bead were taken from the embedded samples using a Holtermann LH-10/20x magnifying glass and the OPTIV Classic VM 3020 profile projector.

Characteristics of Maraging Steel before Welding
Starting Maraging 18Ni (300) steel was analyzed for its microstructure and mechanical properties before welding. repair or heat treatment processes. Table 4 shows the initial properties of the steel with respect to grain size evaluated according to the (ASTM E112-13, 2013). hardness and tensile properties. the initial condition of the steel being solubilized. The properties of the steel were also evaluated. as shown in Table 4. After receiving the thermal treatment of aging. an increase in the mechanical properties of hardness and tension is noted. with a reduction in elongation.

Characteristics of Welded Steel
Microstructural Characterization after Welding Figure 5a shows the Fused Zone (FZ). the Heat Affected Zone (HAZ) and the Base Material (BM). In FZ there is the formation of dendritic structure due to solidification and in HAZ there is a change in the microstructure due to phase transformation or grain growth, both with changes in mechanical properties. In the BM there are no microstructural changes maintaining the original properties of the material. Images 5.b and 5.c show the FZ of the steel before and after the aging treatment, respectively. After the thermal aging treatment homogenization occurs in the microstructure and properties of the BM and HAZ regions.
In Fig. 5b it is shown how the measurements of the cord dimensions were made. The average values found were: Height (A-A). 4.0±0.2 mm; top width (B-B). 5.4±0.3 mm; and root width (C-C). 2.7±0.4 mm. After repairs, these measures were redone.

Microstructural Effects Due to Repairs
After removing parts of the fusion zone (70% in depth). the welds were redone, simulating 1 repair. 2 repairs and 3 repairs in the weld bead. Subsequently. the geometric and microstructural changes that these repairs caused were evaluated, because, in addition to the addition of the material, there is the effect of the repetition of thermal cycling in each repair performed. In Fig. 6 is shown the microstructures of the weld bead after repairs and the heat treatment applied. an increase in the width of the fusion zone after repairs is noted, mainly at the top of the weld. where the repair is applied.
To increase the mechanical properties it was necessary to perform the thermal treatment of aging, the images in Fig. 7 illustrate the weld bead morphology after the treatment, there are no dimensional changes due to the heat treatment.
When analyzing the weld geometry. As shown in Fig. 5b. it is noted that there is an increase in the deposition of the material after the repairs. as shown in Table 5. The increase in the dimensions of the cord (reinforcement) are not removed when used in the Brazilian rocket motor envelope. Contributing to increase the effort supported by the cylinder.
When the steel is welded, there is the formation of a dendritic region in the FZ. As shown in Fig. 8a at the interface FZ and HAZ (without repair) and 8b (2 repairs). with enlargement of the interface marked in Fig. 7b. With black rectangle. The welds made for repair produce similar dendritic microstructures in the region. However. due to thermal cycling. close to the FZ interface, grain growth occurs. It is also noted that. Among the martensitic dendrites, there is the formation of the austenitic phase ɣ (white color), this fact occurs due to the enrichment of the elements Ni. Ti and Mo. stabilizer of this phase. of the last liquid fractions to solidify (Hall and Slunder, 1968). Fanton et al. (2014), performing laser welding of maraging 300 steels, estimated a percentage of 17% of phase ɣ in the FZ. The high temperatures and prolonged times in the treatment of aging also favor the reversion of the martensitic to austenitic phase.
A considerable increase in the initial grain size of the steel was observed in the vicinity of the FZ. Figure 9 illustrates this increase for the welded condition without repairs, it is noted that the average grain size increased about 4 times. in the vicinity of the FZ. according to ASTM. they went from a No. 9 grain to a No. 4 grain. The grains will decrease until reaching the original size of the base material. about 1 mm from the FZ. Venkateswara Rao et al. (2010;Fanton et al., 2014) found a similar increase in the region close to the FZ interface in maraging steels. This same behavior was observed when making 1. 2 or 3 repairs, indicating that the thermal cycling, although producing phase transformations in the HAZ, does not lead to an increase in the grains higher than that observed due to plasma welding.

Microhardness in the Vertical Direction
The direction that coincides with the A-A axis in Fig. 4. From the root to the top of the weld was considered vertical. Figure 10 shows that the variations in the hardness of the weld bead, from the root to the top of the weld (vertical), before the aging treatment-HT (Fig. 10a) and after aging (Fig. 10b). The various repairs carried out did not affect the hardness in the melted area of the material producing little variation in this region.
The aging treatment increased the levels of hardness for all cases studied by about 200 HV. The graph in Fig.  10b and Table 6 show the average values measured before and after the heat treatment and the variations due to the repairs carried out. After repairs and aging, there is an increase in hardness in the samples of just over 30 HV. This increase is due to the aging of the as-welded Maraging steel, which is responsible for precipitate driven pinning of dislocations.

Microhardness in the Horizontal Direction
The horizontal direction is the direction indicated by line B-B of Fig. 4. The hardness measurements were taken from the center of the Fusion Zone (FZ). passing through HAZ. and continuing towards the Base Material (BM). In Fig. 11a hardness profile is shown for each condition studied. before receiving the aging Treatment (TT).
The red reference line indicates the hardness of 349 HV. Hardness of the base material before heat treatment. It is observed that. although there is a similarity in the microhardness values in the FZ and in the transition band (ZT). there are significant changes due to the new thermal cycling due to repairs. It is noted that there is a small reduction in the hardness in the Transition Zone (ZT). a fact related to the increase in grain size, however, the hardness remains close to that of the base material (martensitic).
The higher increases in hardness values are related to the phenomenon of aging associated with the heat of the weld. In the case of unrepaired weld, the greatest effect appears at a distance between 7 and 8 mm from the center of the weld, where the heat has reached a temperature sufficient to cause precipitation of intermetallics, but it was not high enough for austenitization. In regions between 2 and 7 mm the temperatures were higher than 623°C. austenitization temperature, according to . with subsequent formation of martensite in cooling with a hardness similar to that of the base material. Similar results were observed by (Sakai et al., 2016;Fanton et al., 2014) working with different types of weld.
When carrying out the first repair. it is noted that the heat displaced the precipitation region for greater distances, above 10 mm, changing the previously aged band due to the heat imposed during the repair, sufficient to produce alteration of phases with reduction in hardness.
When the second repair is performed, the profile changes again, showing a region between 4 and 6 mm of hardness growth, however, not as high as the distance above 8 mm. In this intermediate region. the temperature is such that it allows partial transformation of phases (austenite-martensite) and partial aging of the grains that have not undergone transformation. For 3 repairs, the structure is more homogeneous, with a more accentuated aging for distances greater than 11 mm, indicating that the higher temperatures, which allow austenitization, have reached greater distances.
It is also noted in FZ some points with a drop in hardness, this fact is associated with the dendritic structure and the formation of untransformed austenite due to the increase of solute atoms in the last liquid portions to solidify, or reversed, due to reheating, a phenomenon also observed by some other authors Sakai et al., 2016;de Lima et al., 2019).
Subsequently, the effect of the aging treatment on the hardness of the samples in FZ. ZT and HAZ was observed. The graphs in Fig. 12 show that aging increased the hardness. for the region of the base material and of the HAZ, to similar levels of the maraging steel without welding (red reference line -568 HV) for all the studied conditions. However, it is noted that in the regions where the presence of austenite was observed in the microstructure (FZ and ZT) there is a drop in the hardness values, because the increase in hardness due to precipitation, occurs more effectively in the martensitic phase. There is also a minimum value in the transition zone because, in addition to the presence of austenite, there is also an increase in grain size in this region.

Tensile Test
The graph of Fig. 13 shows the average curves obtained in the tensile tests. of the Plasma-Welded (PAW) and aged specimens. Deformation was measured with a strain gauge placed over the FZ and HAZ. Table 7 shows the values of the limit of resistance, flow and the elongation found.
Looking at Fig. 13 and Table 7 it can be seen that the fact that steel is welded reduces the strength limit by around 7.5%, compared to non-welded steel. After making repairs to the weld. this reduction remains at similar levels. the following reductions in the resistance limit were observed, according to the number of repairs: 1R-5.7%; 2R-9.2% and 3R-6.0%. The yield limit levels of the repaired coupons remain high and the variation in elongation between the welded samples is small. Figure 14 shows the broken region of the welded specimens, with and without repairs. Broken in traction after heat treatment of solubilization and aging. Note that there is no macroscopic difference with respect to the rupture site, similarly to all conditions, with or without repairs. the fracture tends to start at the periphery of the molten zone or in the molten zone and propagates through this region. There is a coincidence of the fracture site with the positions of less hardness, located in the transition region between FZ and HAZ. due to the increase in grain size and the presence of untransformed austenite. Although the hardness in the FZ after heat treatment may be below the value found in the transition region in some cases (Fig. 12) due to the increase in the height of the cord, a larger resistant area is offered (Table 5). However, it is a region of possible fracture propagation. The fact that there is a discontinuity between the top of the weld and the HAZ also favors the initiation of the fracture. as it creates a concentration of stresses in the region. For all samples presented in Fig. 14, the tensile strength just before the rupture was 1.8±0.1 GPa.

Fracture Surface and Microscopic Aspects
The broken surfaces were analyzed by SEM to identify the fracture micro-mechanisms. The unrepaired welded steel sample, Fig. 15b, shows an intense formation of dimples on the fracture surface with different sizes indicating that the activated micro-mechanism was the ductile fracture.
The material submitted to a repair also shows a surface full of alveoli (15d), but the macrography (15c) presents more irregularity of the surface in relation to the material without repair. For the case of the sample with two repairs, images in Fig. 15e and 15f, indicates that the crack has crossed the molten area and there are macroscopic irregularities indicating the presence of plastic deformations. In the image shown in Fig. 15f, it can be seen that the predominant fracture micro-mechanism was also due to dimples, similarly to the previous case. Finally, on the surface of the welded coupon with three repairs shown in Fig. 15g. there are samples of plastic deformations and in Fig. 15h dimples with reduced size are observed, both aspects related to ductile fractures.

Conclusion
The welds carried out by the plasma process allowed a good union in Maraging 18 (Ni) steel. The performance of repairs (one. two and three repairs) showed that. as the number of repairs increases, there is an increase in the dimensions of the Fused area (FZ) and the Heat Affected area (HAZ), without prejudice to the mechanical properties.
Before the thermal treatment of aging it is noted that the hardness measurements of the FZ. ZT (Transition Zone FZ-HAZ) show a drop, associated with the presence of austenite in the microstructure. In the ZT. in addition to the austenite phase, grain growth was observed, which went from ASTM 9 to 4. In HAZ, there are certain regions where there was an increase in hardness due to the effect of aging, these bands of greater hardness moved for FZ, as the number of repairs increased.
After the heat treatment of aging, the hardness values of the steel increased. At HAZ, even after the repairs were applied. the values reached the same levels as the aged base material. However, although the hardness of FZ and ZT increased, they still showed hardness values below the expected value for the aged base material, an effect related to the presence of austenite in the region.
The mean values of the limit of tensile strength obtained varied from a lower value of 1795 MPa. in the sample with 2 repairs, to the highest value of 1864 MPa sample with 1 repair. These values, considering the standard deviation of about 30 MPa, are similar to the average value observed in welded and aged steel without repairs: 1829 MPa. Comparing these values with the resistance limit of the base material aged before welding (1977 MPa), it is noted that the loss of resistance is less than 10%.
The analysis of the steel fracture surface after welding, repairs and aging, showed that the specimens broke in the region between FZ and ZT. This fact is associated with the presence of austenite in these regions and the phenomenon of grain growth in the ZT interface (FZ-HAZ). As a rule, the fracture started in the region close to the melting line and spread into the weld bead. Nevertheless, the fracture micromechanisms were ductile, with the formation of dimples across the fracture surface indicating that. for all welding conditions, without repair or with up to 3 repairs, no fragile fracture occurred. A factor that indicates viability plasma welding on maraging steel and the possibility of carrying out repairs, if necessary, without loss of strength or ductility.