https://ntrs.nasa.gov/search.jsp?R=20120014531 2019-04-11T08:26:12+00:00Z Mitigating Abnormal Grain Growth for Friction Stir Welded Al-Li 2195 Spun Formed Domes Po-Shou Chen, Carolyn Russell Abstract Formability and abnormal grain growth (AGG) are the two major issues that have been encountered for Al alloy spun formed dome development using friction stir welded blanks. Material properties that have significant influence on the formability include forming range and strain hardening exponent. In this study, tensile tests were performed for two 2195 friction stir weld parameter sets at 400 °F to study the effects of post weld anneal on the forming range and strain hardening exponent. It was found that the formability can be enhanced by applying a newly developed post weld anneal to heat treat the friction stir welded panels. This new post weld anneal leads to a higher forming range and much improved strain hardening exponent. AGG in the weld nugget is known to cause a significant reduction of ductility and fracture toughness. This study also investigated how AGG may be influenced by the heating rate to the solution heat treatment temperature. After post-weld annealing, friction stir welds were strained to 15% and 39% by compression at 400 °F before they were subjected to SHT at 950 °F for 1 hour. Salt bath SHT is very effective in reducing the grain size as it helps arrest the onset of AGG and promote normal recrystallization and grain growth. However, heat treating a 18 ft dome using a salt bath is not practical. Efforts are continuing at Marshall Space Flight Center to identify the welding parameters and heat treating parameters that can help mitigate the AGG in the friction stir welds. I. Introduction Cryogenic fuel tanks used in rocket vehicles are of diameters in the range of 12 feet or greater. Large rocket cryogenic tank domes have typically been fabricated using Al-Cu based alloys like Al-Cu alloy 2219. The use of aluminum-lithium based alloys can reduce weight because aluminum-lithium alloys have lower density and higher strength than Al-Cu alloy 2219. However, Al-Li alloys have rarely been used to fabricate fuel tank domes because of the inherent low formability characteristic that make them susceptible to cracking during the forming operations. Joining Al-Li alloys with fusion welding adds to the complexity and cost of cryogenic tank fabrication also because of the susceptibility to cracking and defect formation during welding. Friction stir welding (FSW) as a solid state joining process has proven to be a perfect solution for manufacturing cryogenic tanks with these alloys. NASA is interested in developing the technology that would enable a single piece, net-shaped dome for cryogenic fuel tanks from Al-Li 2195 alloy using spin forming. This application requires large plates with diameters greater than 18 ft. Commercial plate is not available in sizes large enough to fulfill this requirement, but large blanks can be fabricated by joining smaller plates with FSW. Several technical challenges must be overcome when spin forming Al-Li alloy 2195 using friction stir welded blanks. First, the ability to form metal by spin forming without excessive thinning or necking depends on the strain hardening exponent “n”. A high strain hardening exponent is beneficial to a material’s ability to uniformly distribute the imposed strain. Also, FSWs that have undergone solution heat treatment (SHT) can develop abnormal grain growth (AGG) in the nugget zones, which significantly reduces the mechanical properties. AGG can also arise during post-weld annealing and is sensitive to plastic deformation of the weld. The as-welded blank is not suitable for spin forming because there is an unequal hardness distribution in the weld nugget, the thermo-mechanically affected zone, the heat affected zone and the base material. This leads to an inhomogeneous forming behavior and uneven material thinning between the nugget and the base material. Therefore the as-welded blank must be heat treated prior to spin forming. To spin form an aluminum lithium alloy dome blank joined by FSW, the hardness distribution across the spin blank must be equalized by a heat treatment called “post weld anneal”. More importantly, the formability for such blanks should not be adversely affected by this post weld annealing treatment. MSFC has developed a novel annealing process that can achieve a work hardening exponent on the order of 0.27 to 0.29, which is approximately 50% higher than what is typically obtained for Al-Li alloys using the conventional method. AGG mitigation for the spun-formed 2195 dome is complicated because the forming degree can vary between 5 to 40% along the dome axis. Another main objective of this study is to identify a post forming heat treatment method that can mitigate AGG and improve the mechanical properties. The AGG mitigation efforts for FSW 2195 is continuing at Marshall Space Flight Center and the results of the present study will be presented. II. Experimental Procedures 1. Material and Friction Stir Welding The alloy selected for this study is AA2195 alloy that belongs to the Weldlite (Al-Cu-Li) family. The nominal composition of AA2195 is shown in Table 1. AA2195 is one of the most complicated alloys due to the precipitation of a large number of second phase particles. Details of the microstructure evolution and precipitation behavior are published elsewhere [1-3]. Table 1. Nominal chemical composition of Al-Li 2195 Alloy Cu Li Ag Mg Zr Others 2195 4.0 1.0 0.3 0.35 0.10 <0.05 each Conventional friction stir welding was used to join the plates creating the spin forming blank. That is, the applied forging load was reacted by a steel anvil. Weld parameters were chosen to achieve a desired grain size in the nugget, rather than to optimize mechanical properties. Previous work indicated that larger grains in the nugget of conventional friction stir welds were less susceptible to abnormal grain growth after subsequent processing operations than small grains. The pin tool design was a standard configuration used at MSFC. An optimized weld nugget shape was not evaluated as part of this effort, although it is believed to have an effect on subsequent spun formed mechanical properties. The C-FSW butt welds were produced with the welding direction parallel to the plate rolling direction. Two welds, weld 1 and MT weld were used for the formability study. Weld 1 was produced at Marshall Space Flight Center and MT weld was furnished by MT Aerospace. Weld nugget grain size was determined using electron backscatter diffraction (EBSD) analysis using a FEI Quanta 600F field emission scanning electron microscope. The grain size is approximately 7.0 µm for weld 1 and 10.0 µm for MT weld. 2. Heat treatment for formability study and microhardness testing After welding, all welds were annealed using three annealing treatments (1) PW anneal 1; (2) PW anneal 2; (3) MT anneal. PW anneal 2 is a new formability enhancing heat treatment method developed at NASA-Marshall Space Flight Center [4]. According to this method, the product is first heated to a temperature within the range of 204° to 343° C. (400° to 650° F), and held at the temperature reached for up to 16 hours. Thereafter, the product is subjected to a controlled heating process at a heating rate of 3° C/hour or higher to a second temperature within the range of 371° to 482° C. (700° to 900° F), held at the temperature reached for up to 12 hours. Then the product is subjected to a controlled cooling process at a cooling rate of 3° C/hour or higher to a third temperature within the range of 204° to 343° C. (400° to 650° F), and held at the temperature reached for up to 4 hours. Then the product is cooled to room temperature. Figure 1 graphically depicts the heat treating process of PW anneal 2. The schedule for MT anneal is proprietary. Figure 1. A schematic graph illustrating the heat treating process of PW anneal 2. The weld hardness was measured using a microhardness tester to examine the local variations in the hardness in and around the weld zone before and after the post weld anneal. The microhardness tests were performed at room temperature, approximately 72 °F. 3. Tensile Testing Tensile testing was performed to determine the mechanical properties and strain hardening exponent. Tensile specimens were taken along the T direction from the welded plates. Specimens with a double reduced gage section were used to ensure that the specimen would neck and break in the nugget of the weld. The specimen configuration is shown in Figure 2. Tensile test were performed at 200° C (396 °F) with a strain rate of 0.03 in/in/minute. The weld nugget is completely inside the 0.5” gauge section. A 0.3” extensometer was used to record the strain within the 0.5” gage section of the specimen. Figure 2. The tensile specimen with a double reduced gage section. The weld nugget is completely inside the 0.5” gauge section. The value of the strain hardening exponent (n) was determined using true stress-true strain diagram by Eq. 1. . ln   K  n n    (Eq.1)  ln  The tensile test matrix for weld 1 and MT weld is shown in Table 2 and 3, respectively. Table 2. The tensile test matrix for weld 1 Weld 1 Post Weld Anneal Testing Temperature Specimen location Orientation PW Anneal 1 200° C (396 °F) weld LT PW Anneal 2 200° C (396 °F) weld LT Table 3. The tensile test matrix for MT weld MT Weld Post Weld Anneal Testing Temperature Specimen location Orientation MT Anneal 200° C (396 °F) weld LT PW Anneal 2 200° C (396 °F) weld LT 4. Compression Straining at 200°C (396°F) To simulate the forming strains incurred during spin forming, the C-FSWs were subjected to compression straining. Cylindrical compression specimens measuring 1 in.-tall x 0.375 in.- diameter were machined from the C-FSW panel with the specimen axis oriented transverse to the plate rolling direction with the weld seam bisecting the specimen length (see Figure 3). Compression tests were performed at 200 °C (396 °F) in air with a strain rate of 0.03 in/in/minute. The cylindrical specimens were axially loaded in compression to various strain level increments as detailed in the test matrix shown in Table 4. The total deformation strain of 39% approximates the maximum forming degree in the dome membrane region. Figure 1. Schematic of FSW blank showing orientation of the compression specimens. 5. Anneal and Solution Heat Treatment Following the compression straining, the specimens were annealed and solution heat treated to study the microstructural evolution of the weld. To investigate the effects of heating rate to the SHT temperature on AGG, three different anneal/SHT schedules were used as shown in Table 4 and Table 5. One of these schedules, anneal/SHT 1, was conducted in a conventional air furnace, while the other two schedules, anneal/SHT 2 and anneal/SHT 3, were conducted in a salt bath. Following solution heat treatment, the weld specimens were sectioned, mounted, polished and etched for metallurgical examination. Table 1. Compression deformation and heat treatment test matrix. 1st Strain 2nd Strain 3rd Strain Total Heat Weld Anneal Anneal Increment Increment Increment Strain Treatment Weld 1 15% no - no - 15% Anneal/SHT 1 Weld 1 15% no - no - 15% Anneal/SHT 2 Weld 1 15% no - no - 15% Anneal/SHT 3 Weld 1 15% no 15% no 15% 39% Anneal/SHT 1 Weld 1 15% no 15% no 15% 39% Anneal/SHT 2 Weld 1 15% no 15% no 15% 39% Anneal/SHT 3   Table 5. The heat treatment schedule used for compression strained specimens. Schedule Heat Treatment Anneal/SHT 1 R/C heating to 800°F + 800°F/4h + R/C heating to 950°F + 950°F/1h + WQ Anneal/SHT 2 Immersion in a 950°F salt bath + 950°F/1h + WQ Anneal/SHT 3 R/C heating to 800°F + 800°F/4h + immersion in a 950°F salt bath + 950°F/1h + WQ R/C – ramp controlled A/C – air cooling III. Results and Discussion 1. Nugget zone microstructure Macroscopic views of the nugget zone after post weld anneal for weld 1 and MT weld are shown in Figure 4 and 5, respectively. The nugget shape of weld 1 clearly shows the weld parameters used resulted in a non optimized nugget. Although bulging is readily seen, internal voids did not develop. The typical onion ring structure can be clearly seen in weld 1 after post weld anneal. The nugget zone is not clear after MT anneal, but was revealed nicely after PW anneal 2. This finding suggests that the temperature for MT anneal is lower than that for PW anneal 2. (a) (b) Fig. 4. Macro images of the nugget zones for weld 1 after (a) PW anneal 1 and (b) PW anneal 2. (a) (b) Fig. 5. Macro images of the nugget zones for MT weld after (a) MT anneal and (b) PW anneal 2. 2. Microhardness Microhardness testing was conducted to examine the local variations in the hardness around the friction stir weld zone. The microhardness profile around the weld for weld 1 and MT weld is shown in Figure 6. MT weld with MT anneal has the highest hardness in the weld nugget and shown an unequal hardness distribution in the weld nugget and base metal. The weld/parent metal hardness was more equalized using PW anneal 1 and PW anneal 2 for both weld 1 and MT weld. This finding indicates that the temperature for MT anneal is lower than that for PW anneal 1 and PW anneal 2. Fig. 6. Comparison of hardness profile across the weld for Weld 1 and MT weld after various post weld anneals. In general, the welded blank is not suitable for forming operations when there is an unequal hardness distribution in the weld nugget, the thermo-mechanically affected zone, and the heat affected zone and the base material. The unequal hardness distribution would result in an inhomogeneous forming behavior and uneven material thinning between the nugget and the base material. Therefore the as-welded blank must be annealed prior to spin forming. A more  homogenous forming behavior in the welding area can be achieved with the use of PW anneal 1 and PW anneal 2, although there is still a small variation in microhardness across the weld nugget. 3. Tensile Properties and Forming Range The effects of post weld anneal on the tensile properties are shown in Table 6. In general, a lower value of yield strength and a higher value of tensile strength gives greater formability. Therefore, the forming range, which is the difference between the tensile strength and yield strength, was used to compare the formability. In most cases a higher value of forming range would be expected to lead to a greater formability [5]. Table 6. Tensile properties of weld 1 and MT weld with different post weld anneals. Yield Stress Tensile Ductility *Forming Formability Material Anneal (ksi) Stress (ksi) (%) Range (ksi) Ranking Weld 1 PW anneal 1 11.5 16.6 65.9 5.1 3 Weld 1 PW anneal 2 9.3 18 63.3 8.7 1 MT Weld MT anneal 17.7 22 58.2 4.2 4 MT Weld PW anneal 2 11.5 20 60.8 8.6 2 *: forming range denotes tensile strength - yield strength For weld 1, PW anneal 2 is the better heat treatment than PW anneal 1 for forming operations as it leads to a lower yield strength and a higher ultimate tensile strength. Accordingly, the forming range increases from 5.1 ksi (by PW anneal 1) to 8.1 ksi by the use of PW anneal 2, a more than 60% improvement. PW anneal 2 is also superior to MT anneal for MT weld. The forming range for MT weld increases from 4.2 ksi (by MT anneal) to 8.6 ksi with the use of PW anneal 2, a more than 80% improvement. The overall formability ranking based on the forming range is shown in Table 6. 4. Strain Hardening Exponent (n) Figures 7 is an illustration of true stress vs. true strain curves for weld 1 and MT weld after three post weld anneals. It can be seen that the annealing schedule has profound effects on the tensile behavior. MT anneal for MT weld gives high yield strength but accelerated strain softening, which are signs of poor formability. Anneal 2 is the most effective in increasing the strain hardening for both MT weld and weld 1. Fig. 7. The true stress-true strain curves for Weld 1 and MT weld after different anneals. The strain hardening exponent, n, which indicates the ability of the material to distribute the strain uniformly prior to diffuse necking, was calculated for all tensile tested samples using Eq. 1. A material with a higher value of n indicates more uniform strain distribution and hence greater formability [6]. It can be seen from Table 7 that the strain hardening exponent is higher when annealed using PW anneal 2 for both weld 1 and MT anneal. With the use of PW anneal 2, the strain hardening exponent for weld 1 was increased from 0.16 to 0.21 in the 1% to 5% strain range, and from 0.1 to 0.15 in the 5% to 9% strain range. Table 7. Strain hardening exponent of weld 1 and weld 2 with different post weld anneals. Strain Hardening, n Strain Hardening, n Material Anneal Formability Ranking (1 to 5% strain) (5 to 9% strain) Weld 1 PW anneal 1 0.16 0.1 3 Weld 1 PW anneal 2 0.21 0.15 1 MT Weld MT anneal 0.08 0.01 4 MT Weld PW anneal 2 0.19 0.15 2 *: forming range denotes tensile strength - yield strength PW anneal 2 also works very well for MT weld. The strain hardening exponent after MT anneal was only 0.08 in the 1% to 5% strain range, and dropped to 0.01 in the 5% to 9% strain range. By applying PW anneal 2, the strain hardening exponent was drastically increased from 0.08 to 0.19 in the 1% to 5% strain range, and from 0.01 to 0.15 in the 5% to 9% strain range. Based on the strain hardening exponent values, PW anneal 2 is the most promising among the three post weld anneals for 2195 friction stir welds. The formability ranking based on the strain hardening exponent is shown in Table 7. It is significant to note the use of forming range and strain hardening exponent to rank the formability yields the same order. The similarities in strain hardening exponent values for both the weld 1 and MT weld subjected to post weld anneal 2 indicates the nugget grain size used for this study is not a factor in determining the strain hardening exponent. The strain hardening exponent as a function of true strain for weld 1 and MT weld with different anneals is shown in Figure 8. For MT weld with MT anneal, the strain hardening exponent starts with approximately 0.05 and quickly drops to zero at 7% true strain. A drastic improvement in the strain hardening exponent was observed when the MT weld was treated with anneal 2. The onset of necking was delayed from 3.5% (with MT anneal) to approximately 13.5%. PW anneal 2 also leads to a noticeable increase in the strain hardening exponent for weld 1.