Grain size manipulation by wire laser direct energy deposition of 316L with ultrasonic assistance

Additive manufacturing (AM) is substantially shifting the paradigm in global manufacturing. The possibilities of AM open the door to unique opportunities in part design like multimaterial builtups in functional components, hybrid components, lightweight design, or functionally graded materials.¹ Among the laser-based AM processes, direct energy deposition (L-DED) with wire has demonstrated immense potential for the fabrication of near-net shape, large-scale industrial components. This can be attributed to the access to a standardized feedstock material, efficient feedstock utilization, and weaker safety requirements compared to its powder-based variant.² 

Overcoming the inherent thermal conditions of this manufacturing process could yield additional exploitation capabilities of its potential. Typically, the rapid cooling of the melt pool coupled with a highly directional heat flux via the substrate material results in a steep temperature gradient throughout the part. Consequently, epitaxial growth of crystals during solidification is triggered, and the formation of coarse columnar grain structures parallel to the build direction is induced.^3,4 Grains can extend over several deposition layers and may develop a distinct texture corresponding to the fastest crystallographic growth direction (<100> for cubic and <1010> for hexagonal alloys⁵). Additionally, heterogeneity in solute distribution and strong element segregation originating from directional solidification contributes to the formation of brittle and low melting point second phases in interdendritic regions. A prime example constitutes the IN718 alloy that forms the low melting Laves-phase in these regions.^6,7 It is evident that the aforementioned process environment severely compromises the performance of as-built components and causes strong anisotropy of the mechanical properties.^3,4 The ability to effectively and locally tailor grain size and morphology and to eliminate the texture is, therefore, a highly desirable goal. This could not only give rise to materials with higher fatigue resistance, strength, toughness, and ductility, but it is also proven that grain refinement can significantly reduce the cracking susceptibility, for instance, in Al or Ni alloys.^8–10 Furthermore, novel design approaches like on-demand grading of mechanical properties may become possible.

While the phenomenon responsible for the formation of columnar grains has been comprehensively investigated, appropriate approaches to counteract this issue are still lacking. In the past, the formation of coarse solidification structures has been efficiently tackled by either thermomechanical processing^11–13 or introduction of nucleant particles into the melt pool.^14,15 However, the machining of intricate AM-components does require complex and impractical process setups. Introducing nucleant particles into the melt imply a change in composition of the parent alloy and could cause clusters that are not fully melted during fabrication.^15,16 It is also to mention that the above listed methods do not offer the option for controlled grain structure grading. Therefore, industry and academia still lack a standard procedure that provides tools to local and in situ control and modify the microstructure of AM-parts on-demand without chemical adjustments of the alloy compositions, nucleating agents, or additional processing.

The utilization of ultrasonic (US) excitation of melt pools in casts or welds already proved to be a promising alternative for the promotion of equiaxed grains with a weakened texture. Li et al., for example, achieved a reduction in grain size of chill casted 2219 Al ingots by introducing three 20 kHz ultrasonic generators into the melt pool with a peak-to-peak amplitude of 20 μm.¹⁷ Nabahat et al. demonstrated a similar effect on microstructural features during ultrasound-assisted tungsten inert gas welding of AISI 321 stainless steel.¹⁸ Recently, Todaro et al. adopted the method for a powder L-DED process.¹⁹ It was found that a substantial grain refinement of Ti6Al4V and IN625 specimens was realized by vibrational excitation of the substrate plate with a 20 kHz ultrasound. The same research group analyzed the response of AISI 316L steel to ultrasound excitation during powder-DED.²⁰ The study revealed that ultrasound assistance leads to columnar-to-equiaxed transition of grain structure and to the weakening of the typical <100>-fiber texture. Furthermore, Yuan et al. have proven that the grain structure of Ti6Al4V and ER321 austenitic steel can successfully be influenced during L-DED with wire.^21,22 They used a sonotrode probe guided behind the wire process head with an offset to transmit vibrations into the workpiece surface and excite the molten pool. Those ultrasonic waves with sufficiently high intensity induce cavitation, and acoustic streaming in fluids has been well established.^23,24 Yet, whether these effects are triggered in metal melts during AM and to which extend it affects solidification are still topics extensively discussed and investigated in the literature. One effect that has been proposed as a major factor and tries to elucidate the effect of ultrasonic excitation on grain formation is the fragmentation theory. Da Shu et al. suggested that mainly shockwaves created by the collapse of cavitation bubbles lead to the fragmentation of dendrite arms.²⁵ More recent studies from Wang et al. also propose that quasi-steady state oscillating bubbles in the vicinity of dendrite arms play a significant role.²⁶ These bubbles induce an alternating bending motion on dendrite arms and initiate a fatigue fragmentation effect. In both cases, the fragments will ultimately act as nucleation sites and form new grains ahead of the solidification front. That way directional solidification is interrupted, and a more isotropic, equiaxed, and finer structure is generated. Furthermore, Wu et al. found that acoustic streaming might support stirring of the melt pool and promote a more homogenous distribution of elements, and hence, it suppresses major segregations.²⁷ 

In the study proposed in this paper, the use of high-intensity ultrasound-assisted L-DED with wire was explored, and the influence of ultrasound amplitude on grainsize, -morphology, and -texture was investigated to gain further insight into the possibilities for microstructural tailoring. The substrate was directly linked with the US-device and agitated in its entirety. Unlike powder L-DED, the feedstock material and the agitated melt pool in this process are in constant contact during processing, resulting in a relative motion of feedstock and agitated melt. This might assist stirring of the melt pool and could offer additional exploitation potential. The combination of wire L-DED and holistic US-assistance described here presents a unique new approach to investigate the influence of US. Because of its industrial significance and commercial accessibility, AISI 316L steel was used as a benchmark substrate and feedstock material.

The wire L-DED experiments were carried out in a modified 5-axis AM-machine of the 19389 MTH (Arnold Ravensburg) type. The general setup and the high-intensity ultrasonic system are presented in Fig. 1(a). A COAXwire deposition system from Fraunhofer IWS was utilized. The process head comprises a centrically aligned wire feeding system and three integrated coaxial partial laser beams, which are focused toward the feedstock material. This offers improved geometrical freedom and directionally independent deposition. The laser source used in this study was a 4 kW LDF4030 diode laser (Laserline GmbH Germany).

The US-system was custom-developed in cooperation with the company BANDELIN electronic GmbH & Co.KG. It is essentially composed of three components: the transducer for the translation of the electrical signal from the US-generator (400 W) into mechanical oscillation, the booster for amplitude regulation, and the sonotrode that represents the substrate material in this setup. Therefore, printed parts and the sonotrode are directly linked, and vibrations are transmitted with minimal attenuation into the melt pool.

Three single track wall samples were printed analogical to the strategy shown in Fig. 2(b). Each sample was manufactured with different ultrasonic amplitudes (A = 0, 12, and 21 μm). The wall sample consisted of 10 (A = 12 and 21 μm) to 13 (A = 0 μm) layers. Therefore, the height of the walls varied between 5 and 7 mm. To compensate for the additional heat input from the acoustic energy, laser power was incrementally (first layer, second layer, and fifth layer) reduced during printing (600, 550, and 500 W). Start and stop of the wire feeding for each layer were performed with an offset of 0.1 mm per layer. In this way, it was ensured that the material was only delivered when the process head was above the previous printed layer and a collision of the wire material and previous layers could be avoided.

As a result, the track length was reduced successively with each layer creating an overall reduction of 0.2 mm per layer. The process parameters used are summarized in Table I. Figure 2(c) exemplifies the fabricated wall samples.

AISI 316L austenitic steel was chosen as a substrate and wire material. Substrate plates had a thickness of 20 mm and a diameter of 70 mm. The diameter of the wire material was 1 mm. The chemical composition of the material is given in Table II. The microstructural assessment of the samples was conducted in their as-built state.

The metallographic analysis was performed by means of the scanning electron microscope (SEM) JEOL JSM-7800F (Akishima, Japan) including electron backscattered diffraction (EBSD) using a NordlysNano EBSD detector (Oxford Instruments, Abingdon, UK) together with light microscope Olympus GX51 (Tokio, Japan). For microstructural investigations, the cladded samples were cut and mounted using a cold mounting epoxy resin. The samples were ground with a SiC abrasive paper, wet-polished using a diamond polishing suspension, and finally vibratory polished for multiple hours using Al and Si oxide polishing suspensions. For light microscopic investigations, the samples were etched using an etching solution Kalling II consisting of 100 ml H₂O, 100 ml HCl (32%), and 5 g CuCl₂ of etching additives.

EBSD scans were performed on the samples tilted by 70° with a step size of 1.5 μm, and the acquired data were analyzed with AZtecCrystal software (Oxford Instruments, Abingdon, UK). For the grain size analysis, the misorientation threshold was set to T ≥ 10°, and very small grains with less than 100 pixels were excluded from calculations.

The optical micrographs of the deposited wall specimens after etching are depicted in Fig. 3. It was observed that due to the additional energy input by ultrasound excitation, the geometry of the built samples was affected. This is reflected in the increased width of the wall specimens. The single layers of the deposited wall specimens are clearly visible in the depicted cross sections of Fig. 3. For samples manufactured without ultrasound assistance, some coarse columnar grains extending over several deposition layers in build direction can be observed on the etched specimen surface. However, the excessive growth of such grains was not found in the samples manipulated by the ultrasound.

Regardless of the degree of ultrasound excitation, all investigated samples exhibited a dendritic microstructure after deposition (Fig. 4). Exemplary EBSD phase-analysis of the specimen produced with ultrasound stimulation at A = 21 μm reveals a dual phase structure, Fig. 4(d).

It consists of γ-austenite (fcc) and residual δ-ferrite (bcc), which is typical for AISI 316L steel due to its ferrite-austenite solidification mode.^32,33 As a consequence of high cooling rates during the L-DED process, diffusion-controlled δ-ferrite dissolution is hampered. This results in a relatively high δ-ferrite content of approximately 4% for all investigated samples determined by means of EBSD analysis.

No significant influence of ultrasonic excitation on morphology and distribution of dendritic networks was observed by means of SEM analysis. Considering the impact of ultrasound excitation on the grain structure and the texture of the manufactured samples, they were analyzed in detail by means of EBSD technique. The orientation maps of the samples produced with and without ultrasound excitation are shown in Fig. 5. All samples were analyzed in the middle of the deposited wall specimens. Figure 5(a) highlights that for the sample produced without ultrasound stimulation, a significantly pronounced elongation of columnar grains in the build direction can be observed. By increasing the ultrasound amplitude from A = 12 μm [Fig. 5(b)] to A = 21 μm [Fig. 5(c)], the grain structure became finer and more equiaxed. This can be further reinforced by grain size analysis from the acquired EBSD-data.

Figure 6(a) represents the area-weighted grain size distribution of the samples manufactured with and without ultrasound assistance. Specifically, for the specimen produced with a higher ultrasonic amplitude of A = 21 μm, it is evident that the fraction of large grains with equivalent circle diameter d > 300 μm disappears.

The area-weighted mean grain size decreases from d_m = 284.5 μm for specimen without ultrasound excitation to d_m = 182.5 μm for specimen with A = 12 μm and further to d_m = 130.4 μm for specimen with A = 21 μm. Attributed to the heat flux toward the substrate, solidification of wire L-DED wall specimens is highly directional. An aspect ratio of grain shapes can provide a good insight into the possible effect of ultrasound assistance. The calculated data provided in Fig. 6(a) reveal that only a slight decrease in the aspect ratio for the specimen with A = 21 μm was observed; however, for the specimen with A = 12 μm, no significant changes have been noticed and even a slight increase from a = 2.5 to a = 2.6 has been recorded.

The texture analysis of the acquired EBSD data reveals a generally great impact of ultrasound excitation on the solidification behavior of the investigated AISI 316L steel compared to the characteristic microstructure without ultrasonic assistance.^4,34 Figure 6(b) shows pole figures of the samples in different conditions. The sample manufactured without ultrasound excitation exhibits a strong cube texture. This so called <100>-fiber texture is typical for fcc-metals in the case of directional solidification, which also applies for L-DED manufactured wall specimens in current investigations.³⁵ Specifically, for specimen with the highest ultrasonic amplitude of A = 21 μm, a significant randomization of the grain orientation can be observed.

Todaro et al. used a similar setup to investigate the effects of ultrasonic excitation during powder L-DED of AISI 316L.²⁰ Notably, they observed significantly smaller grains without application of ultrasound, with d_m= 52 μm; this is likely due to different process parameters.²⁰ Smaller laser spot size, lower laser power, and higher welding speed result in significantly higher cooling rates.³⁶ Additionally, cuboid samples result in a bigger area available for heat conduction due to overlapping tracks, improving the cooling.³⁴ The ultrasonic excitation with an amplitude of 30 μm and a frequency of 20 kHz in 20 resulted in d_m = 16 μm,²⁰ a reduction in grain size of ∼70% compared to ∼54% achieved in this work. This, however, matches well with the observation of a more pronounced effect with an increasing amplitude.

Other works differ in the deposited material, the type of feedstock used, the method of ultrasonic excitation, the sample geometry, the utilized process parameters, or the examined properties. Therefore, a direct comparison proves difficult. Despite these differences, Todaro et al.¹⁹ and Yuan et al.^21,22 have shown the same general results of reduced grain size, lower aspect ratio, and suppression of the <100>-fiber texture.^, To the best of the authors' knowledge, no other work on ultrasonically assisted laser-DED has presented the degree of ultrasonic influence in relation to the amplitude shown here.

In the current study, a new setup for ultrasonic assisted L-DED with wire was established, and the influence of ultrasound excitation on the microstructure of the processed AISI 316L steel has been analyzed. The following conclusions based on the key results can be drawn:

• It could be confirmed that the holistic excitation of the substrate material by direct linking of the US-device and substrate provides an effective method to transmit ultrasonic vibrations into components during L-DED. However, unique process challenges like controlled wire conveying, additional acoustic energy, and structural damping effects become prevalent utilizing this US excitation strategy.

• All produced specimens exhibited a dendritic microstructure, composed of γ-austenite matrix and residual δ-ferrite, which is typical for fast solidified AISI 316L steel. The ultrasound excitation does not have a significant impact on dendritic solidification and δ-ferrite network size and distribution.

• The ultrasound excitation induces structural changes from a coarse and strongly columnar microstructure to a finer and more equiaxed structure. For the highest ultrasound amplitude, the smallest grain size could be obtained.

• The ultrasound excitation had a great impact on the texture of the manufactured samples. A randomization of the predominant <100>-fiber texture can be realized by increasing the amplitude of ultrasound excitation.

In summary, a positive effect of ultrasound-assisted L-DED processing of AISI 316L wire on grain refinement and prevention of strong fiber texture has been demonstrated. The impact became more pronounced with the increased amplitude of ultrasonic excitation. A more thorough investigation of the responsible mechanism is required to better understand the principles of grain refinement for the investigated material and system setup. The “global” excitation of the whole substrate and, therefore, the AM part is an easy and reliable way to verify the US-effects. However, the implementation of this method for large industrial components might be a tall order. Structural attenuation and changing resonance frequencies of the deposited material can pose a problem. A variant for local ultrasonic excitation of the melt pool should be assessed in detail in the future. This could push the potential of US assistance in AM even further.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The authors have no conflicts to disclose.

Maximilian Heidowitzsch: Data curation (equal); Validation (equal); Visualization (equal); Writing – original draft (lead). Leonid Gerdt: Formal analysis (equal); Investigation (equal); Methodology (equal); Visualization (equal); Writing – original draft (equal). Conrad Samuel: Investigation (equal); Project administration (equal); Writing – review & editing (equal). Jacob-Florian Maetje: Investigation (equal); Methodology (equal); Validation (equal); Visualization (equal); Writing – review & editing (equal). Jörg Kaspar: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Supervision (equal); Writing – review & editing (equal). Mirko Riede: Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Elena López: Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Frank Brueckner: Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Christoph Leyens: Funding acquisition (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available within the article and from the corresponding author upon reasonable request.