Nickel-containing superalloy laser weld qualities and properties

Autogenous laser welding could be a promising technique for high productivity, low distortion joining of selected superalloys. For example, recent work¹ has identified that Yb fiber laser melt runs can be made in 3.35–5 mm Inconel (IN) 718, over a range of heat inputs, with levels of underfill and internal porosity meeting stringent requirements of an industrially relevant standard/benchmark, AWS D17.1, so long as appropriate choices in beam focusing and optics are made.

Nevertheless, this same work also identified that slower speed (and thus higher heat input) laser welding appeared necessary to avoid heat affected zone (HAZ) microcracking in IN718 and satisfy minimum weld width criteria (not stipulated in AWS D17.1, but of interest to some aerospace fabricators).

Having to reduce welding speed obviously impacts on some of the productivity and distortion advantages of laser welding, and questions remain around whether such stringent quality criteria are necessary to meet property requirements in all circumstances.

The current work therefore not only examines the qualities that can be achieved in IN718 butt welds as well as melt runs, at speeds more representative of laser welding, to AWS D17.1 criteria, but also examines other laser welded superalloys and measures selected mechanical properties, even in instances where the combination of the alloy being welded and the conditions used are known to give rise to small numbers of HAZ microcracks.

The following materials were laser welded:

• IN718, 2 mm thickness wrought sheet

• Haynes alloy (HA) 188, 2 mm thickness wrought sheet

• IN718, 2.7 mm thickness cast sheet

• C263, 2.7 mm thickness wrought sheet

• IN600, 3 mm thickness wrought sheet

• IN718, 5 mm thickness wrought sheet.

Immediately prior to welding, all materials had their upper and lower surfaces acetone degreased, then abraded using a Scotch-brite^TM abrasive pad, then redegreased. Plate edges were also dry machined square (without lubricant) and degreased before being abutted, aiming for a close fitting, flush, butt joint. Welding typically took place within 48 h of machining. All of the welds made in this work were autogenous, i.e., additional filler wire was not used.

A continuous wave output 5 kW Yb fiber laser beam source was used in the welding trials.

The laser beam was delivered via a 0.15 mm core diameter flexible optical fiber to a process head containing either a 160 mm or 300 mm focal length focussing optic. With the 160 mm focal length collimating optic used, this provided either a 0.15 or 0.28 mm nominal diameter spot at focus, respectively.

The laser power at the work piece was measured prior to welding using a water cooled power meter.

The focussing head was manipulated robotically, at a constant stand-off from the materials to be welded, which were clamped in a steel butt welding jig.

Argon shielding gas was used for top bead and underbead shielding.

Tensile tests were carried out in this work using a 100 kN electromechanical test apparatus with hydraulic water-cooled grips, a high temperature extensometer with alumina rods, and an electric furnace (for high temperature tests). Stress-strain curves were recorded during strain-rate controlled testing. In particular, strain (ε) was recorded using either the extensometer (ε < 1.5%) or via the cross-head motion (ε ≥ 1.5%). Strain rates of 2.5 × 10⁻⁴/s (ε < 2%) or 6.7 × 10⁻³/s (ε ≥ 2%) were used. Temperatures during high temperature testing were monitored using three K-type thermocouples at the top, center, and bottom of each specimen gauge length, following heating to, and 10–15 min stabilization at, each test temperature.

Low cycle fatigue (LCF) tests were carried out using a 200 kN servohydraulic test frame fitted with oxide dispersion strengthened (ODS) grips, a high temperature extensometer and an electric furnace (for high temperature tests). LCF tests were executed in strain control, using a triangular waveform, with a strain rate of 1%/s, and an R ratio of either −1 (in tests on IN718) or 0.5 (in tests on thinner IN600 test pieces, to minimize the risk of buckling). Strain ranges of 0.6%–2.5% were used, to obtain LCF lives of 10²–10⁵ cycles. The stress-strain hysteresis for each cycle was recorded, from which the maximum and minimum stress and strain per cycle and the Young's moduli during loading and unloading could be extracted. Temperatures during testing were monitored using a thermocouple at the center of each specimen gauge length, following heating to, and 10–15 min stabilization at, each test temperature.

The targeted joint configurations of the welding trials carried out are summarized in Table I.

Melt run on plate (melt run) trials were made first, prior to welding any butt joints, to reduce unnecessary materials usage. Melt run trials began on the two 2 mm thickness materials, examining the effects on weld quality of:

• Laser power: in the range of 1.0–2.8 kW

• Welding speed: in the range of 1–8 m/min

• Focused spot diameter: of either 0.15 or 0.28 mm

• Defocus position: in the range of −4 (laser focused 4 mm below material surface position) to +2 mm

• Laser preheating immediately prior to welding

• Laser postheating immediately after welding.

Melt run trials were also carried out on the thicker materials, with welding speeds, laser beam powers, and power densities being adjusted to produce full penetration conditions.

Suitable melt run conditions for each material and thickness were selected based on visual inspection and then radiographic examination and metallographic sectioning in selected cases. These conditions were then used to weld butt joints, as detailed previously in Table I. Visual inspection, radiographic examination, and metallographic sectioning of selected butt welds were then carried out, to ascertain the quality that could be produced.

Following completion of the welding trials, further laser butt welded coupons were made using selected welding conditions. A number of mechanical tests were then carried out on test pieces removed from these coupons or equivalent parent material for comparison purposes. These tests are summarized in Table II. All IN718 coupons were subjected to a postweld heat treatment (PWHT) prior to these tests. The IN600 coupons were not heat treated.

Cross-weld tensile test pieces contained the full thickness of the welded joint, including the weld bead (2 mm thickness materials and 5 mm thickness when tested at 750 °C) or only the central 2.5 mm thickness (5 mm thickness, tested at ≤550 °C). Tests were carried out to ISO 6892–1/-2.

Cross-weld LCF test pieces contained the central 2 mm thickness of the welded joint (3 mm thickness materials) or the central 2.5 mm thickness (5 mm thickness materials), i.e., the weld beads were machined off in both cases. Tests were carried out to ISO 12106.

From melt runs, two potentially suitable conditions were identified as giving consistent full penetration, with the lowest levels of spatter and underfill:

• 2 kW laser power, 0.15 mm focused spot diameter, 4 m/min welding speed, −1 mm defocus position

• 2 kW laser power, 0.28 mm focused spot diameter, 3 m/min welding speed, −4 mm defocus position.

Radiography indicated typically <5 pores of diameter of 0.1–0.2 mm, along a 300 mm length of melt run made with either of these conditions. The porosity contents of a pair of butt joints welded with these two conditions were slightly higher. Radiography indicated that the butt weld made with the 0.28 mm spot had the highest porosity content, to only class B of AWS D17.1. HAZ microcracking was also detected in the HAZs of cross sections of butt welds. Figure 1 shows an example of this.

Higher heat input melt runs were also made at 1–2 m/min. Nevertheless, a minimum beam power of 1.5 kW was found necessary to guarantee weld stability and avoid an excessive porosity content. Welding at 1 m/min using this 1.5 kW beam appeared to reduce the amount of HAZ microcracking, but could not be guaranteed to eliminate it completely.

A +180 mm defocused, 2 kW, laser beam (using a 160 mm focal length focusing lens) was considered suitable at 0.9 m/min to preheat the material immediately prior to welding, whilst avoiding excessive oxidation. This approach, however, was also unsuccessful in eliminating HAZ microcracking. This same approach was also used to postheat a laser weld immediately after it had been made, but this was also unsuccessful. Altogether, these results indicated that a suitable welding condition for 2 mm HA188 was not found in this work.

Two series of melt runs were made, centered around the same two conditions selected for 2 mm HA188. These melt runs appeared slightly more underfilled than those made in 2 mm HA188. At this stage, only the melt runs made using the 0.15 mm spot were radiographed. These appeared to contain similar levels of porosity to those made in 2 mm HA188. A pair of butt joints welded with the two conditions was also radiographed. These also appeared to contain similar levels of porosity to the melt runs, but these levels were not quantified at this stage in the work.

HAZ microcracking was once again detected in cross sections of these welds. This was in keeping with previous IN718 results when using similar welding conditions.¹ Nevertheless, the IN718 cracked to a much lesser extent than the HA188.

In common with 2 mm HA188, welding at 1 m/min using a 1.5 kW beam also appeared to reduce the amount of HAZ microcracking, but could not be guaranteed to eliminate it completely either. Laser preheating was similarly unsuccessful in eliminating microcracking. On the basis of these results, neither laser preheating nor postheating was pursued further.

Full penetration melt runs were made in both of these materials, using conditions selected from the preceding trials on 2 mm wrought IN718. Cross sections of two melt runs detected ∼2–3 microcracks in the HAZ in the IN718. Microcracks were not detected in the HAZ in the C263. A cross section of a dissimilar butt weld made between these two materials once again detected microcracks, but only in the HAZ on the IN718 side of the joint.

Laser melt runs were made using laser powers of either 2 or 3 kW. Full penetration was achieved at ≤1.5 m/min (0.28 mm spot) or ≤2 m/min (0.15 mm spot), when using a 2 kW beam. Higher speeds of 2–4 m/min could be reached using a 3 kW beam. However, these resulted in a greater amount of internal porosity (using 0.28 mm spot) or spatter and underfill (using 0.15 mm spot) than when welding with a 2 kW beam.

HAZ microcracking was not detected in a cross section of a 2 kW beam weld (Fig. 2). However, a degree of HAZ grain coarsening was observed in this section (arrowed in the figure). The weld cap was also slightly underfilled, to a maximum depth of ∼0.1 mm. AWS D17.1 permits undercut to a maximum depth of only 0.05 mm.

A visually acceptable butt weld was achieved using a 5.1 kW beam at 2.6 m/min, with focusing optics giving a 0.15 mm spot focused 2 mm below the material surface. X-ray radiography indicated that this weld was to class A of AWS D17.1, in terms of both the size and content of its internal porosity, containing a maximum accumulated porosity length of 0.6 mm per 76 mm. Figure 3 shows a cross section of this weld.

Approximately 2–3 HAZ microcracks were detected in a cross section of this weld. The largest example is shown in Fig. 4. The detection of microcracks in this weld was again in line with previous IN718 results when using similar welding conditions.¹ These were of a similar length, ∼50 μm, as those that had been detected previously in laser welds in 2 mm IN718.

As reported above, laser welds completely free of HAZ microcracks were not produced in 2 mm IN718 in the current work. A small number of microcracks up to 50 μm long were detected, even in the best cases. Mechanical testing of these laser welds was deliberately carried out to determine what effect, if any, these would have on the mechanical properties of the welds.

2 mm IN718 butt welded coupons were therefore made using the 0.28 mm spot conditions described previously. This spot diameter and its slower welding speed (of 3 m/min) were found to be slightly more tolerant of any small fit-up gaps still present after fixturing the coupons, than when welding using a 0.15 mm spot at 4 m/min.

X-ray radiography detected pores up to 0.3 mm in diameter (well within class A of AWS D17.1) and accumulated lengths (in 76 mm) up to 0.52 mm (also well within the requirements of class A). Figure 5 plots these lengths for the three welds made against the AWS D17.1 class A/B cut-off.

A cross section through one of these welds detected underfill (Fig. 6) and an HAZ microcrack of similar length, ≤50 μm, to those detected previously (Fig. 7). The weld was also slightly underfilled, to a maximum depth of ∼0.07 mm. This value is slightly greater than those measured previously,¹ but these earlier values were measured from melt runs, which are always likely to have lower levels of underfill than butt welds between two pieces of material. These test welds were then subjected to PWHT.

Butt welded test pieces were made using the 2 kW beam condition at 1 m/min developed previously. X-ray radiography detected pores up to 0.4 mm in diameter (well within class A of AWS D17.1) and accumulated lengths (in 76 mm) up to 4.05 mm. Class A of AWS D17.1 allows a maximum accumulated length of 4 mm. Figure 8 plots these lengths for the five welds made. Figure 2, shown earlier, provided a cross section typical of these butt welds. These welds were not subjected to any PWHT.

In common with 2 mm IN718, laser welds completely free of HAZ microcracks were not made in this work, albeit that any cracks present were similarly small and few in number. Butt welded test pieces were made using the 5.1 kW beam condition at 2.6 m/min developed previously, to determine what effects these cracks would have on subsequent mechanical test performance.

X-ray radiography of these detected pores up to 0.6 mm in diameter (well within class A of AWS D17.1) and accumulated lengths (in 76 mm) up to 3.9 mm (also well within the requirements of class A). Figure 9 plots these lengths for the five welds made. A cross section of one of these welds once again detected HAZ microcracks. These test welds were then subjected to PWHT.

Laser welds in 2 mm IN718 had a cross-weld ductility and strength similar to parent material, up to at least 550 °C, in spite of the presence of some underfill and presence of HAZ microcracks in the welds. The test pieces themselves consistently failed outside the weld zone. Examples of this are shown in Fig. 10 (weld line positions arrowed).

Laser welds in 5 mm IN718 also had a cross-weld ductility and strength comparable with parent material, consistently failing outside the weld zone, up to 550 °C. At 750 °C, failure appeared to occur in the middle of the gauge length, but for both welded and parent test pieces. Voids detected in the fracture faces suggested creep was beginning to play a role in failure at this temperature.

The stress ranges experienced per cycle during the 3 mm IN600 LCF tests, Δσ, were evaluated, as a function of test temperature, the strain range applied during the test, Δε_t, and the ratio (N/N_f) of the number of test cycles elapsed (N) versus the number of cycles to failure (N_f). In general, Δσ increased rapidly at the beginning (N/N_f < 0.1) of all the tests, before then decreasing slightly, more gradually, as the tests continued.

Test results were analyzed further following procedures described in ISO 12106. The monotonic stress (⁠$σa1/4$⁠) versus strain (⁠$εa1/4$⁠) behavior was calculated from the stress-strain data obtained during the first quarter cycle of each test, where

and $εa1/4$ = max. strain in first quarter cycle, $σa1/4$ = max. stress in first quarter cycle, E = Young's modulus, K = monotonic strength fit coefficient, and n = monotonic strain-hardening fit coefficient.

Cyclic stress (σ_a) versus strain (ε_a) behavior was also calculated, from

where $εa=Δεt/2$ = strain amplitude at half-life, $σa$ = stress amplitude at half-life, where $σa=Δσ/2$⁠, E = Young's modulus, K′ = cyclic strength fit coefficient, and n′ = cyclic strain-hardening fit coefficient.

Figure 11 shows an example of the comparison made, at 550 °C, of these monotonic and cyclic stress-strain behaviors.

In Fig. 11, the points are experimentally determined data and the solid lines those calculated from Eqs. (1) and (2). Figure 11 shows:

• The IN600 laser welds have higher monotonic yield strength than parent material.

• The monotonic strain hardening behavior of the IN600 welds is similar to parent material.

• The cyclic yield strengths are higher than the monotonic.

• The IN600 welds have higher cyclic yield strength than parent material (albeit the cyclic strain hardening rate of the welds is thereafter lower than that of the parent material).

The LCF properties of the welds and parent material were then calculated by fitting the experimental data to the Coffin–Manson relationship

where variables are as defined for Eqs. (1) and (2) and $εea=Δσ/2E$ = elastic strain amplitude at half-life, $εpa$ = plastic strain amplitude at half-life, $σf′$ = fatigue strength fit coefficient, b = fatigue strength fit exponent, $εf′$ = fatigue ductility fit coefficient, c = fatigue ductility fit exponent, and N_f = number of cycles to failure.

Figure 12 shows the $εa$ and ε_pa values for LCF tests performed on welded or parent IN600 test pieces at 550 °C. In Fig. 12, the points are experimentally determined data, and the solid lines those calculated from Eq. (3). Trends in elastic strains (ε_ea) with N_f are not shown, but were similar for both types of test piece.

Figure 12 implies that parent material has been able to resist higher levels of plastic strain better, for a given LCF life. Conversely, for a given applied strain amplitude, parent material has a longer LCF life. Figure 13 plots these LCF lives against stress amplitudes. Although laser welds, for a given life, resist less total strain amplitude, owing to their higher cyclic yield strengths (Fig. 11), these lower strain amplitudes can correspond to higher stress amplitudes than those experienced in parent material. The longer lives achieved at 25 °C are also shown, for welded test pieces.

The test pieces failed in the middle of their gauge lengths in virtually all of these tests, including those of parent material. Whether there exists a clear relationship between failure location, weld location and LCF life remain the subject of ongoing research.

As with the preceding LCF tests on 3 mm IN600, the stress ranges experienced during the 5 mm IN600 LCF tests were evaluated. In general, a gradual decrease in Δσ took place as the tests proceeded, particularly at 550 °C. Comparisons were once again made between the monotonic and cyclic stress-strain behaviors. Figure 14 shows an example of this comparison at 550 °C (cf. Fig. 11).

In Fig. 14, the points are experimentally determined data, and the solid lines calculated from Eqs. (1) and (2). Figure 14 shows:

• The monotonic behaviors of welded IN718 and parent test pieces are very similar.

• The cyclic yield strengths are lower than the monotonic (albeit the cyclic strain hardening rate of the IN718 laser welds then exceeds that of the parent material slightly).

In a manner similar to Fig. 12, Fig. 15 shows the ε_a and ε_pa values for LCF tests performed on welded or parent IN718 test pieces at 550 °C.

As Fig. 15 shows, once again parent material can withstand a larger plastic strain, for a given life.

Figure 16 shows these LCF lives plotted against stress amplitudes. The trend for laser welded test pieces is now similar to parent material. This result is in line with the similarity in cyclic yield stress values for the two types of test piece (Fig. 14). Weld results at 25 °C are also shown.

The majority of tests pieces did not fail in the middle of their gauge lengths, in contrast with IN600 results. Nevertheless, clear correlations were not found between failure location and N_f, in these IN718 tests.

The welding trials in this work have confirmed that, with appropriate choice of parameters, full penetration fiber laser welds can be made in superalloys of different thicknesses.

Pore diameters and porosity contents can be achieved acceptable to the class A requirements of AWSD17.1.

In some cases, however, weld underfill up to 0.1 mm in depth has been observed, although this could still be tolerated in applications if postweld machining is carried out or, potentially, corrected with an appropriate wire filler addition during welding.

These trials have also indicated, however, that certain alloys can be susceptible to either HAZ grain coarsening (IN600) or HAZ microcracking (HA188). The HAZ microcracking tendency of IN718 has also been reconfirmed.

Nevertheless, the tensile testing carried out suggests that any weld underfill, HAZ microcracks, and pores that may be present do not necessarily lead to a significant deterioration in tensile properties, in the case of IN718, 2–5 mm in thickness, tested to 550 °C. This is exemplified, for yield stress values, in Fig. 17.

The LCF behavior of the welded materials examined requires careful interpretation.

The different effects the laser welding has on the monotonic behaviors of IN600 and IN718, seen in Figs. 11 and 14, respectively, can be explained by the different strengthening mechanisms of each alloy.

IN718 is precipitation-strengthened, while IN600 is solid-solution strengthened. The rapid cooling immediately after laser welding can increase dislocation density. In IN600, these dislocations offer significant resistance to plastic flow, and the yield strength is increased compared with parent material. Following yielding, the flow behaviors and stress-strain curves of both materials then appear similar. By contrast, in IN718, precipitates have a much greater effect on yield behavior, and the PWHT appears to remove or reduce the effect of any dislocations present. Overall, from Fig. 14, it would seem that after PWHT the monotonic yielding of the welded and parent IN718 is essentially the same.

During fatigue testing, IN600 work hardens, resulting in the cyclic hardening observed. The cyclic yield strengths of both welded and parent materials increase, but the higher dislocation density initially present in the welded material may explain its higher than parent cyclic strength also.

By contrast, IN718 softens during testing. This phenomenon can be caused by the formation of easier glide paths through the distribution of strengthening precipitates, these paths developing through the passage of many dislocations during testing. One influence of laser welding on IN718 can be seen in Fig. 15, where the “double curve” characteristic of the parent material S–N plot, posited to be due to a change in the slip system behavior,³ appears eliminated by welding. Nevertheless, a transmission electron microscopy study would be needed to elucidate why the slip behaviors of laser welded test pieces are different to those of parent material.

Comparison of the trends in LCF life with applied strain amplitude shows that both the laser welded IN600 and IN718 tolerate lower plastic strains, in general, than equivalent parent material. As a result, the lives of the welded test pieces, as a function of the applied strain, are shorter than parent test pieces. These differences in LCF lives may be related to the differences in the cyclic stress-strain behaviors of both types of weld, when compared with their corresponding parent materials.

Interestingly, when LCF life is considered on the basis of the stress level, the lives of the welded test pieces can then be the same as (IN718) or longer than (IN600) the parent test pieces. The reasons for this are most likely the similarity in cyclic yield stress between laser welded test pieces and parent material, for IN718 (Fig. 14), and the higher cyclic yield strength of laser welded test pieces than parent material, for IN600 (Fig. 11), respectively.

These differences in LCF lives could have important consequences for laser welded superalloy component designs in fatigue. If the stress level is well below the yield strength, laser welds can exhibit a similar (IN718) or even higher (IN600) fatigue life than parent material. In this design case, reductions of fatigue properties due to the weld are not necessarily anticipated. However, when the stress level is above the yield strength, the fatigue life of the laser welds can be substantially shorter than those of the parent material. Under such circumstances, it is not possible to design with the fatigue properties of the parent material, as these are not sufficiently conservative.

Overall, the conclusions of this work for fiber laser welded Ni-containing superalloys are:

• Laser welding is capable of producing welds with internal porosity contents meeting AWS D17.1 class A in 2–5 mm thickness IN718, 2.7 mm thickness C263, and 3 mm thickness IN600.

• Conversely, satisfactory welding of 2 mm HA188 does not appear possible, given its greater propensity to HAZ microcracking than IN718.

• In spite of microcracks, the static tensile properties of welded IN718, in the thickness range of 2–5 mm, are comparable with parent material, up to at least 550 °C. At 750 °C, creep appears to play a role in tensile failure of both materials.

• LCF testing of 3 mm IN600 indicates welded joints are less capable of withstanding large plastic strains. LCF lives when based on strain amplitude are therefore shorter than those of parent material. Contrarily, the longer LCF lives when based on stress amplitude are in line with the higher cyclic yield stresses of the welds during testing. Work remains to correlate failure locations with respect to the positions of the welds.

• LCF testing of 5 mm IN718 also indicates welded joints are less capable of withstanding large plastic strains, leading to shorter lives based on strain amplitude. Lives based on stress amplitude are similar to parent material, stemming from similarities in cyclic yield stresses between welded and parent material.

Chris has worked at TWI since 2002, specializing in laser materials processing for aerospace, power generation, transport, oil and gas, shipbuilding, and defense clients, in addition to managing national and international collaborative projects.

Rob has worked at TWI since 2013, with a primary interest in metal joining process developments. His background has focused on metallurgy, with postgraduate research work on nickel superalloy design for the aerospace industry.

After achieving a doctorate in materials science, Thijs has worked as a researcher at M2i and NLR specializing in high temperature materials & testing, and electron microscopy. Currently, he holds a research position at Applied Environmental Chemistry, TNO.