Summary

The laser welding of titanium alloys is increasingly being considered for developing near net shape structural components. Processing parameters for the laser welding of titanium alloy Ti-6Al-4V have clearly been identified for producing repeatable and high quality joints on a variety of material thicknesses and joint configurations. Less information is available on metastable beta titanium alloys especially the recently emerged titanium alloy Ti-5553. As this alloys gains acceptance in the aerospace industry, it can be predicted that the laser welding process will play a crucial role in its wider implementation hence it is necessary to address its weldability by producing high quality defect-free welds with good mechanical performance.

61 Chapter 3. Effect of Defocusing Distance and Weld Speed on Laser

Welding of Ti-5553

3.1. Introduction

There are many advantages associated with laser welding such as a low and precise heat input, ease of automation, and rapid processing rates. In addition, laser welding offers additional flexibility when welding titanium alloys since it offers the possibility of welding either autogenously or with filler material in the form of a wire or powder. For autogenous continuous wave Nd:YAG laser welding, the power, defocusing distance, and welding speed are amongst the most important parameters that influence how the energy is applied to the joint and ultimately, the quality of the welds. Hence, it is crucial that these parameters be optimized to fit within an optimum processing window that satisfies the aerospace specification tolerances. The work presented in this chapter deals with the effect of defocusing distance and welding speed on the welding quality of Ti-5553 autogenous welds.

3.2. Experimental Procedure

Ti-5553 material, received in ingot form, was sectioned to obtain weld coupons of 76 mm in length x 38 mm in width x 3.1 mm in thickness. Prior to welding, the weld coupons were (1) solution treated at 815.5°C for 45 min. in vacuum followed by an argon quench and (2) aged at 621°C for 8 hrs in argon partial pressure followed by an argon quench.

The faying surfaces and neighbourhood of all the specimens were brushed and then cleaned with methanol to remove surface oxides and other contaminants prior to

62 clamping. Butt joints were welded in the direction along the length of each coupon using a 4 kW CW Nd: YAG laser system (manufactured by TRUMPF, Germany) equipped with an ABB robot and magnetic holding fixture system. A collimation lens of 200 mm, focal lens of 150 mm, and a fiber diameter of 0.6 mm were employed to produce a laser beam with a spot diameter of approximately 0.45 mm. Titanium, in its molten state and at temperatures above 300°C, is reactive with most atmospheric gases such as oxygen, nitrogen, carbon, and hydrogen ^{8, 60}. Therefore, it is necessary to take adequate measures to shield the weld region until the weldment is cooled below the reactivity temperature.

High purity argon at a flow rate of 23.6 l/min was used to shield the top surface of the work-piece. The trail on the top surface and the bottom of the work-piece was shielded using helium at a flow rate of 66.1 l/min, which according to AWS C7.2 minimizes the HAZ size during welding of titanium alloys. The laser power (P) used was kept constant at 4 kW while the defocusing distance (∆z) and welding speed (v) were adjusted to obtain fully penetrated autogenous butt welds. Table 5 shows the processing parameters used.

After welding, the specimens were examined by radiography to detect any cracks and/or porosity in the weld regions. The surface quality of all laser welds was visually evaluated and macroscopically recorded using an Olympus SZ40 stereoscope. For the examination of the transverse section and microstructure of the welds, two metallurgical specimens were cut from each joint and then mounted using cold-setting epoxy resin. The specimens were prepared using standard metallographic techniques and final polishing was performed using 0.04 micron colloidal silica with 10% hydrogen peroxide to produce a mirror-like finish. Etching to reveal the microstructure was accomplished using Kroll’s

63 reagent (1-3 mL HF + 2-6 mL HNO3 + 100 mL H2O) for ~ 10 seconds. Microstructural examination was carried out using an inverted optical microscope (Olympus GX71) equipped with digital image analysis software (AnalySIS Five). The microhardness profiles across the welded joints were measured at a testing load of 500 g and a dwell time of 15 seconds using a Vickers microindentation machine (Struers Duramin A300) with an automated testing cycle. For each weld condition, three hardness profiles across the weld joint near the top, center, and root height of the joint were made with an indent interval of 0.3 mm. Three tensile specimens per weld having a standard sub-size geometry of 25 mm in gage length, 6 mm in width and 3.1 mm in thickness were machined in accordance with ASTM E8M-01. All specimens were tested at room temperature using a 250 kN MTS 810 tensile machine equipped with an Aramis 3D deformation measurement system. The Aramis system is a non-contact optical system that automatically measures strains along the gage length of the tensile sample during the testing. The system comprises of two CCD cameras capable of recoding 15 frames per second (fps), a trigger box and a high performance PC system. It is noteworthy that the functionality of the Aramis deformation system depends on the quality of the speckle pattern applied on the tensile sample, i.e. a good contrast between fine black dots against a white background is required. For each sample, the quality of the applied pattern was verified before mechanical property evaluation to ensure strain recording along the entire gage length. After examination for pattern recognition, tensile property evaluation was conducted using displacement control at a rate of 2 mm/min up to the rupture with Aramis set at an acquisition rate of 3 fps. After tensile testing, the fracture surface of the specimens was examined using a Hitachi S-3000N scanning electron microscope.

64 Table 5. Processing parameters used for Ti-5553

Welding # Weld speed

Visual examination of the weldments revealed a silver color for the crown and root surfaces of all joints indicating that the welds were adequately shielded during welding and cooling. Also, stereoscopic examination showed no surface porosity or cracking. The effect of defocusing distance (at 4 kW and 6.0 m/min) on the surface morphology and transverse sections is shown in Figures 22 and 23, respectively. Full penetration was achieved at defocusing distances ranging from −1 mm to +1 mm. Incomplete/critical penetration occurred at −2 mm indicating that the energy was insufficient to fully penetrate the joint. At a defocusing distance of +1 mm, the root width became narrow indicating that further increases in the defocusing distance above +1 mm would result in a lack of penetration (LOP) defect. As shown in Figure 24, the influence of the defocusing

65 distance between -1 mm and 0 mm on the FZ area, HAZ area and FZ width is relatively minor. Therefore, the optimum defocusing conditions was between −1mm and 0 mm.

a)

b)

Crown Root

Figure 22. Stereoscopic micrographs showing the crown and root surface morphologies at defocusing distances a) −2 mm and b) +1 mm at 6.0 m/min. The welding direction is from left to right.

Figure 25 shows the effect of welding speed on the transverse sections. Fully penetrated joints were produced up to a welding speed of 6.0 m/min at a defocusing distance of −1 mm. Lower welding speeds resulted in larger FZ and HAZ areas, and FZ widths (Figure 26). This is due to the higher heat input at lower welding speed which results in the melting of more material. Also, from the structure of the FZ shown in Figure 25 (e-f), it was observed that the relative size of the columnar β grains increased appreciably with increasing heat input, which is related to the primary dependence of the β grain size on the heat input during welding of titanium alloys.

1 mm Lack of penetration

1 mm

1 mm 1 mm

66

(a) −2 mm (b) −1 mm

(c) 0 mm (d) +1 mm

Figure 23. Effect of defocusing distance on the cross sections at 6.0 m/min.

0

Figure 24. Effect of defocusing distance on the FZ and HAZ areas and widths at 6.0 m/min.

67

(a) 2.25 m/min (b) 3.0 m/min

(c) 4.5 m/min (d) 6.0 m/min

(e) 2.25 m/min (f) 3.0 m/min

Figure 25. Effect of weld speed on the transverse sections at −1 mm defocusing.

2 mm

2 mm 2 mm

HAZ

HAZ BM

BM FZ

200 μm 500 μm

2 mm

68

Figure 26. Effect of weld speed on FZ and HAZ areas and widths at defocusing distance of −1 mm.

3.3.2. Defects

The two most commonly encountered weld defects were underfill and porosity as seen in the transverse sections (Figures 23 and 25). Underfill appears as a depression on the top or bottom surface extending below the adjacent surface of the base metal. At a constant weld speed of 6 m/min, Figure 27a shows the effect of defocusing distance on the maximum underfill depth, which was measured by the maximum linear depth of the depression from the crown or root surface of the welds. The largest underfill depth occurred at a defocusing distance of 0 mm and the lowest at –2 mm. Considering that the defocusing distance influences the power density, at a defocusing of 0 mm, the focal point of the laser is located on the top surface of the weld and hence the maximum power density is exerted on the top surface. This may cause some expulsion/spatter, and evaporation of some alloying elements, leading to the formation of underfill defects.

69 Figure 27b shows the influence of welding speed on the maximum underfill depth at a defocusing distance of −1 mm. Underfill was found on both the top and bottom surfaces of the welds for all the welding speeds. The maximum underfill depth increased with increasing welding speed, reaching a maximum at 4.5 m/min, after which it decreased.

Generally speaking, there are two main mechanisms that are probably responsible for the formation of underfill: (1) the expulsion and/or evaporation of molten metal which is usually dominant at lower weld speeds and (2) high welding speeds that do not allow sufficient time for the molten metal to refill the depressions ⁶¹. In the second case, the center of the weld bead obtains a higher peak and a more convex shape ⁶². Underfill reduces the sheet thickness and creates stress concentrations, causing a decrease in fatigue life and tensile strength of the joints. Therefore, it is important to avoid or minimize the underfill defect. According to AWS D17.1, the specification for fusion welding for aerospace applications stipulates a maximum underfill depth of 0.07T (where T is the thickness or 3.1 mm in the present work) for Class A welds. Hence, over the range of welding conditions examined, the laser welds in Ti-5553 satisfy Class A requirements in that the underfill discontinuity remained below 0.22 mm (0.07T).

70

Figure 27. Effect of (a) defocusing distance and (b) weld speed on the maximum underfill depth.

Apart from underfill discontinuity, porosity was another common defect observed in the welded joints, as indicated by the various types of porosities shown in Figure 28. Also, the porosity was observed to vary in terms of the shape and location depending on the origin. Mostly the porosity exhibited a spherical shape and was most commonly found at the FZ/HAZ boundary (Figure 28a-b), although some pores were randomly distributed in the weld (28c). In some instances for both the high and low welding speeds, interdendritic shrinkage was observed, as seen in Figures 28d. Figure 29 shows the effect of welding speed on the total area and the percentage of porosity in the FZ. Overall, more porosity was observed at lower welding speeds, probably due to the greater time available for the growth and coalescence of pores. In all cases, however, the measured porosity was found to be below 1% of the FZ area with a maximum occurring at 4.5 m/min. Previous work has reported that pores appearing at the HAZ/FZ interface of titanium weldments are related to hydrogen because the solubility of hydrogen increases with decreasing temperature in titanium alloys ⁶³.

Acceptable limit Acceptable limit

71

(a) (b)

(c) (d)

(e) (f)

Figure 28. Various types of porosities in the welds (a-c) gas porosity near FZ/HAZ interface, (d) cluster of porosity near the bottom of weld and (e, f) shrinkage porosity.

200 μm

100 μm 500 μm

100 μm 500 μm

50 μm FZ

HAZ FZ

HAZ

FZ FZ

FZ FZ HAZ

72

Figure 29. Effect of welding speed on the total porosity area and % porosity in the FZ.

3.3.3. Microstructure

Figure 30 displays a series of microstructures from the base metal to the end of the heat affected zone in the as-welded joint. The base metal has a bimodal structure (Fig 30b), consisting of globular/equiaxed primary α phase particles in a β matrix with fine secondary α acicular platelets. The volume percentage of the globular primary α phase in the BM is approximately 20 ± 5 % (average of 20% and standard deviation of 5).

Dissolution of most of the secondary α phase platelets progressively occurs in the HAZ, where the temperatures experienced are below the liquidus of the alloy but are still high enough for the solid-state reactions to occur. Figure 31 is a plot of the volume percentage and average diameter of the globular primary α phase versus distance from the center of the weld. It appears that the primary α phase globules are also progressively affected in that the fine particles dissolve, whilst the larger particles decrease in size (Fig. 30).

However, this effect is masked in the overall calculation of the α phase particle diameter evolution, most probably due to the HAZ next to the BM having very coarse and very fine globules (larger particle size distribution), whilst the HAZ close to the FZ has a

73 nearly homogeneous particle size as illustrated in Fig. 30e. The FZ which experiences temperatures above the liquidus of the alloy consists entirely of the metastable/retained β phase that is formed when the alloy is melted and then quenched rapidly. Coarse columnar β grains in the FZ epitaxially grow from semi-melted grains in the HAZ towards the central weld pool opposite the heat extraction direction as seen in Figure 32a.

Within the FZ, a cellular dendritic microstructure is observed (Figure 32b).

The effect of defocusing distance on FZ microstructure is seen in Figure 33. No clear change in the β grain size and dendrite arm spacing was observed over defocusing distances ranging from −1 mm to + 1 mm. However, the β grain size for a defocusing distance of −2 mm is slightly smaller than that observed for the other defocusing distances. This is most likely due to the lower heat, as indicated by the lack of penetration defect, which results in a higher cooling rate. Some equiaxed grains near the weld center region were observed at −2 mm, which is indicative of a columnar-to-equiaxed transformation (CET), as shown in Figure 33a. This was not observed for any other welds. As the welding speed was increased (or the heat input decreased), the β grain size as well as the cellular-dendritic structure became finer. The decrease in the β grain size with increasing weld speed can be reasoned on the basis of the lower heat input as explained previously, while the refinement of the cellular-dendritic structure is related to the cooling rate, which is effectively greater at higher welding speeds.

74

(a) (b) BM

(c) region 1 of HAZ (d) region 2 of HAZ

(e) region 3 of HAZ (f) region 4 of HAZ

Figure 30. Series of optical micrographs from the BM to the end of the HAZ for a welding speed of 2.25 m/min and defocusing distance of −1 mm

2 3 4

20 μm 20 μm 2 mm

BM 1

20 μm

20 μm 20 μm

75

Figure 31. Volume percentage and particle diameter of primary α across the welded joint

(a) (b)

Figure 32. Typical microstructures of (a) the FZ/HAZ boundary and (b) the FZ for a welding speed of 2.25 m/min and a defocusing distance of −1 mm

(a) −2 mm (b) +1 mm

Figure 33. Effect of defocusing distance at −2 and +1 mm on the FZ microstructure

100 μm

76 3.3.4. Hardness

The variations in microstructure during laser welding are well reflected in the hardness distribution of the weld joints. Figure 34a shows a typical distribution of microindentation hardness values taken at the top, middle, and bottom of a weld joint. It is seen that the FZ hardness is lower than that of the BM. The BM has a Vickers hardness ranging from approximately 316 to 340 with an average of 327 ± 8 HV (average of 327 and standard deviation of 8). The FZ has a Vickers hardness ranging from 280 to 305 HV with an average of 289 ± 9 HV. Compared to the BM, this represents an average decrease of approximately 11 % in hardness. This decrease in the HAZ hardness is directly related to the dissolution of α (HCP) phase (primary and secondary). Specifically, the strengthening mechanism in metastable β titanium alloys relies heavily on α phase precipitation for which the volume fraction, morphology, and location strongly influence the mechanical properties of the alloys ⁴⁶. It is noteworthy that the hardness in the FZ does not significantly vary with defocusing distance and welding speed (Figure 34b), albeit changes in the size of the FZ are reflected in the hardness profile.

200

Figure 34. Microindentation hardness profiles at (a) 3.0 m/min) and (b) 2.25 and 6.0 m/min (top profile)

77 3.3.5. Tensile Properties

The global tensile properties of the laser welds are summarized in Figure 35. The UTS increased with increasing welding speed from approximately 845 MPa to 929 MPa. A similar tendency was observed for the YS. The elongation at fracture showed no trend but remained below 4 %. The lowest elongation value was obtained at 4.5 m/min, which was the specimen that showed the highest porosity area. In literature, Ti-5553 samples in the solution treated and aged condition consisting of a bimodal microstructure had average values of 1236 MPa for the UTS, 1174 MPa for the YS, and 13 % for the elongation ⁵. In this work, the properties of the as-received ingot material were also tested and determined to be 1046 MPa for the YS, 1107 for the UTS and 14% for the elongation. Considering that the material in the gage section of the tensile sample consists of both the bimodal microstructure of the base metal as well as a retransformed beta microstructure in the FZ, the mechanical properties of weldment should range in between the reported values and the data obtained in the present work. However, as indicated by the mechanical properties of the weldment, the joint efficiency in terms of UTS ranged from 68 % to 75%

(depending on welding speed) when comparing with the properties of the bimodal microstructure and 76% to 83% with that of the ingot microstructure (beta grain structure with acicular alpha). The overall decrease in mechanical properties for the laser welded Ti-5553 can be reasoned in terms of the microstructural changes, which for Ti-5553 have been reported to be particularly sensitive to the volume fraction, size, and distribution of the primary and secondary alpha phases ⁴⁶.

78 In all cases the specimens failed in the HAZ region as shown in Figure 36. Examination of the fracture regions revealed shallow ductile dimples (indicating ductile fracture) as well as porosity (Figure 37). The increase in the UTS and YS with increasing welding speed is most likely due to the refinement of the prior-β grain size and the cellular dendritic microstructure, which are an outcome of the lower heat input and higher cooling rate. Moreover, this trend is consistent with the hardness results in that, although the Vickers microhardness values did not change with increasing welding speed, the greater extent of material affected thermally (i.e. FZ and HAZ size from the hardness trough) rendered an overall greater decrease in the strength of the weldment. Also, an effect of the maximum underfill depth on the strength could not be discerned from the results, which is inevitably related to the discontinuity size that has an impact on the properties only above 0.07T according to AWS D17.1. However, it is clear that the strain is concentrated at the HAZ, which was narrower at 6.0 m/min than at 2.5 m/min as depicted in Figure 38 from the digital images taken by the Aramis system prior to failure. From this Figure, it can be seen that in the gauge length of the material the strain can reach 5%

to 6% in the HAZ prior to localization that is clearly evident in Figure 38a and was found

to 6% in the HAZ prior to localization that is clearly evident in Figure 38a and was found