Adjusting the grain structure of laser beam welds

The grain structure of a weld is of specific interest to increase its strength or ductility or to reduce its susceptibility to the formation of hot cracks.^1-4 A conventional strategy to benefit from this effect is to influence solidification by optimizing the alloy composition. To avoid high costs and limited flexibility, which coincide with the application of non-standard alloys, strategies were recently developed that improve the grain structure of a weld by a smart optimization of the process parameters.

Laser beam welding enables adjusting the energy input by an almost independent adaption of the different process parameters such as laser power P, welding velocity v, and focal diameter d_f. This allows the formation of geometrically identical weld seams with strongly different process parameters. Each parameter set marks corresponding solidification conditions that cause different grain structures.

Figure 1 shows an example with the horizontal section on the left and the cross-sections on the right of two welds with almost identical geometry, but achieved via significantly different process parameters and therefore completely different grain structures. Anodic etching and the illumination with polarized light allows us to distinguish between the individual grains in the microstructure of a weld. The comparison of Figures 1a and 1b shows an obvious difference in their microstructure despite their geometrical correspondence: The grain structure of the weld in Figure 1a contains solely equiaxed dendritic grains, whose refined grain size is associated with enhanced and isotropic mechanical characteristics. On the contrary, no equiaxed grains are present in the weld shown in Figure 1b.

The metallographic analysis of the cross sections shown in Figure 1 demonstrates the potential to adjust the grain structure of laser beam welds without any optimization of the alloy composition, but with an optimization of the process parameters. The phenomena have been investigated at the IFSW of the University of Stuttgart within the framework of several open-funded projects and in collaboration with industrial partners to benefit from this possibility.

Results

One main result was the analytical derivation of a parameter-based criterion for the formation of a refined equiaxed dendritic grain structure.⁵It describes the minimum power per welded depth:

which must be absorbed during the welding process, with the material-specific properties liquidus temperature 𝑇_{𝐿𝑖𝑞𝑢𝑖𝑑𝑢𝑠}, heat conductivity 𝜆_𝑡_ℎ, density 𝜌, and heat capacity 𝑐_𝑝 in an ambient temperature 𝑇_𝑎𝑚𝑏.⁵ The alloy-specific ratio G/R_eqx ≅ 3 Ks/mm² between temperature gradient G and solidification rate R describes the required conditions for equiaxed grain growth during solidification of the investigated AA6016 aluminum alloy. It’s dependent on alloy composition and welding parameters.⁶

The derived criterion is compared with experimental results in Figure 2. The green vertical line represents the critical power per depth 𝑃_{𝐷𝑒𝑝𝑡}_ℎ,eqx ≈ 677 W/mm calculated with the equation above for the case of welding an AA6016 alloy with an accuracy of about ±10%. The data points and their error bars represent the share of equiaxed grains 𝜂_𝑒𝑞𝑥 = w_eqx/w_weld, which represents the ratio between the width of the equiaxed zone and the total width of the weld, as introduced in Figure 1a. The widths 𝑤_𝑒𝑞𝑥 and 𝑤_{𝑤𝑒𝑙𝑑} were measured in the metallographic analysis of welds in overlap configuration of AA6016 sheets with welding depths ranging from 1.2 to 2.9 mm, welding velocities ranging from 0.25 to 30 m/min, laser powers ranging from 0.38 to 16 kW, and laser beam diameter on the surface of the sample from 50 to 630 µm. The orange squares represent measurements resulting from an additional superimposition of beam oscillation (wobble technique) during welding.⁷

FIGURE 2. Share of equiaxed grains in rectilinear welds (blue) and welds welded with beam oscillation (orange) as a function of depth-specific power [5].

The comparison of the experimental results and the calculated threshold in Figure 2 proves the validity of the theoretically derived criterion. The share of the equiaxed grains in the weld increases strongly with an excess of the power per depth over the threshold—𝑃_{𝐷𝑒𝑝𝑡}_ℎ,eqx. In the case of welding with a power per depth below this threshold, no equiaxed dendritic grains can be identified in the metallographic sections—that is, η(P/s < 𝑃_{𝐷𝑒𝑝𝑡}_ℎ,eqx) = 0.

When welding with a power per depth that significantly exceeds the threshold, the share of equiaxed grains converges toward a value of η(P/s < 𝑃_{𝐷𝑒𝑝𝑡}_ℎ,eqx) ≅ 0.5.

Approaches in process optimization

Since an equiaxed dendritic grain structure is associated with smaller grain sizes and isotropic mechanical characteristics, it is desirable to increase the power per depth above the determined critical value. However, an increase of the power per depth requires the increase of the absorbed laser power without the increase of the penetration depth of the capillary. This goal can be achieved with two basic strategies:

Increase the width of the weld. This leads to an increased volume, which must be molten at the same time. The resulting requirement for higher energy per time requires the targeted increase of the power per depth.

Increase the welding velocity. To achieve the same depth and width of a weld seam at a higher velocity, the same volume must be molten in a shorter time. The resulting requirement for the same energy in a shorter time requires the targeted increase of the power per depth.

Figure 3 presents the consequence of these two basic strategies, which can be described by the hyperbolic relation between the width of the weld and the welding velocity for a constant power per depth.⁵ The solid black curve in Figure 3a represents the hyperbolic relation for the critical value 𝑃_{𝐷𝑒𝑝𝑡}_ℎ,eqx ≈ 677 W/mm. The data points result from the same experiments, as described above.

FIGURE 3. Process strategies to exceed the threshold for an equiaxed dendritic grain structure (a) and the resultant horizontal sections in case of welding with (b) or without (c) transverse beam oscillation [5,7].

The width of the weld can be systematically increased either by increasing the laser beam diameter (green arrow), or by an additional transverse beam oscillation (orange arrow). The comparison of the horizontal sections in Figures 3b and 3c shows the expanded presence of equiaxed dendritic grains in the case of a sinusoidal transverse beam oscillation with an amplitude of 0.75 mm at a frequency of 100 Hz.

The horizontal purple arrow in Figure 3a represents exceedance of the critical power per depth by an increase of the velocity. To achieve comparable widths at higher velocities, the increase in velocity is connected with an increase in the focal diameter—this requires an additional increase in the power per depth.

Both strategies—the increase of the width of the weld and the increase of the velocity—are essentially limited by the available laser power. High-power laser beam sources are mandatory for the implementation of such strategies.

The knowledge about the influence of process parameters on the grain structure even enables the modulation of the grain structure in a laser weld.⁸Figure 4 shows the horizontal section of such a weld seam, generated with a periodical modulation of the absorbed power per depth between 550 and 2500 W/mm.

FIGURE 4. Interruption of hot crack propagation by an active, laser-parameter-driven local modulation of the grain structure [8].

This modulation strategy (see Fig. 4) was successfully applied to interrupt the propagation of centerline cracks by a local change of the grain structure at the position of the white arrow.

The combination of the derived equations with the RDG model yields the description of the influence of the laser welding parameters on the hot cracking sensitivity of laser beam welds.^9,10 The results identify the line energy per depth, P/(v∙s), as the key influencing parameter on the critical maximum strain rate, which the solidifying weld can resist without the formation of a centerline hot crack.¹⁰ The calculated critical strain rate as a function of the line energy per weld depth was validated with experimental results,^3,11 which were measured using digital image correlation with the method presented in Optics and Lasers in Engineering.¹²

Those results prove that the increase of the applied line energy per depth yields an increase in the critical strain rate for hot crack formation, which results from an optimized grain structure.

Derived models and criteria are an excellent tool for the optimization of welding processes and explain how to design process strategies, which have a deterministic effect on the grain structure and lead to a successful reduction of hot crack formation.^3,5,7,8,10

REFERENCES

1. H. Fujita and T. Tabata, Acta Metall., 21 (1973).

2. J. W. Wyrzykowski and M. W. Grabski, Philosophical Magazine A, 53 (1986).

3. C. Hagenlocher et al., Sci. Technol. Weld. Join., 24 (2019).

4. Z. Tang and F. Vollertsen, Weld. World, 58 (2014).

5. C. Hagenlocher et al., Mater. Des., 174 (2019).

6. C. Böhm et al., Metall. Mater. Trans. A (2021).

7. C. Hagenlocher et al., Mater. Des., 160 (2018).

8. C. Hagenlocher et al., Proc. CIRP, 74 (2018).

9. M. Rappaz et al., Metall. Mater. Trans. A, 30 (1999).

10. C. Hagenlocher et al., Metall. Mater. Trans. A, 50 (2019).

11. D. Weller et al., Proc. CIRP, 74 (2018).