Process development and heat flow analysis for thin walls made of aluminum using extreme high-speed laser material deposition (EHLA3D)

Additive manufacturing is becoming increasingly important as more and more areas of application are being entered and the number of patents and parts produced is increasing rapidly.^1,2 In the industry, laser material deposition has become an established process due to its versatile deployment for coatings, repair applications, and additive manufacturing.^3,4 Modified process variants are being developed to meet the industry's increasing demands in terms of cost and resource efficiency. One of these variants is the EHLA process (extreme high-speed laser material deposition).

For the EHLA process, the powder particles are already melted above the melt pool on the substrate’s surface during the interaction between powder particles and the laser beam (cf. Fig. 1). This eliminates the time required to melt the powder particles in the melt pool and drastically reduces the time required to produce a defect-free layer, which in turn, enables higher process speeds. Compared to conventional laser material deposition (LMD), the achievable deposition speed in the range of a few meter per minutes is drastically increased to typical feed rates of up to 200 m/min,³ and simultaneously, the minimum achievable layer thickness is decreased from several hundred micrometers to as low as 25 μm. Furthermore, the heat input into the component as well as the dilution zone are reduced. Difficult-to-weld materials and many combinations of materials with dissimilar thermophysical properties can be processed.^5,6 The wide range of process feed rates (20–200 m/min) allows cooling rates to be suited to the material and to achieve cooling rates similar to Laser Powder Bed Fusion (LPBF).³ Because of these advantages, the EHLA process has become established in various industries, such as the coating of brake disks or piston rods.^6,7

Up to now, applications for EHLA have mainly been limited to wear or corrosion protection or repair of rotationally symmetric components due to the challenges associated with achieving the high relative speeds required when machining free-form surfaces.⁸ With highly dynamic kinematics, the benefits of EHLA can be extended to additive manufacturing, and free-form coatings and repairs (EHLA3D). To implement the high translational process feed rates, a tripod parallel kinematic system was developed by ponticon GmbH, offering a maximum feed rate of up to 200 m/min.⁹ Using this system technology, numerous materials have already been validated for the EHLA3D process at the Fraunhofer ILT. These include 316L,¹⁰ M2,¹¹ and IN718.¹² Typical deposition rates are in the range of 3 kg/h for 316L,¹⁰ while structures with a wall thickness of 300 μm and a minimum dilution zone around 10 μm have been demonstrated for IN718.¹² 

However, process development at feed rates >20 m/min becomes challenging when depositing complex three-dimensional geometries beyond cubes and walls. This is due to the varying process conditions that arise from increasing geometrical deviations over the built height. To compensate these deviations, previous approaches have pursued process adjustments via local parameter adaptations. However, no understanding of the cause-and-effect relationship between the locally applied energy density and induced heat in the workpiece is established. Consequently, a time-consuming and costly parameter study is carried out for each individual geometry.

This study uses an experimental approach to vary the applied energy density and determine the process-induced heat in the workpiece at a feed rate of 30 m/min. Walls with a thickness of approximately 1 mm are considered, which are built up from aluminum using EHLA3D. Thermal analysis focuses on the global heating of the workpiece over several layers and not on the area of the melt pool.

All experiments are carried out on the pE3D tripod kinematics from ponticon GmbH. The system enables maximum feed rates of up to 200 m/min and a maximum acceleration of 50 m/s².⁹ The size of the cylindrical workspace is Ø700 and 800 mm in height. A Twin-150 from Oerlikon Metco is used as a powder feeder. The beam source used is a Trumpf TruDisk 4002 disk laser (maximum laser power 4 kW) with a wavelength of λ_L = 1030 nm and a beam quality of BPP = 8 mm mrad. The beam is guided by a fiber with a core diameter of d_f = 200 μm and shaped by a Trumpf Beo D70 welding optics with a motorized collimation. A coaxial powder nozzle of type HighNo 4.0 by HD Sonderoptiken GmbH with a nominal standoff distance of 9 mm (distance between the nozzle tip and the powder focus lane) and a powder focus of d_P = 0.95 mm (measured for 316L powder at a mass flow of m_P = 3 g/min) is utilized. It can be assumed that a larger and slightly offset powder focus is to be expected for the material and mass flow used. The initial distance between the nozzle and the substrate is set to 10 mm, while the powder focus standoff is measured at 9 mm.

AlSi10Mg powder with a specified particle size distribution of 20–63 μm from Fehrmann Materials GmbH is used. The chemical composition corresponds to EN-AC 43000 (Ref. 13) and is listed in Table I.

Hot-rolled EN AW-5083 aluminum plates with a thickness of 10 mm are used as substrates. Before welding, the substrates were sandblasted and cleaned with ethanol.

In an initial review of the existing process parameters, single tracks with a length of 40 mm were welded to provide a foundation for the subsequent buildup of thin walls. The parameters used are provided in Table II. For all experiments, the laser beam diameter is kept constant at d_L = 1.2 mm, the shielding gas flow at Q_s = 12 l/min, and the carrier gas flow at Q_c = 6 l/min.

A metallographic analysis is conducted to evaluate the characteristics of the tracks. The tracks are cut perpendicular to the welding direction and analyzed microscopically. The quality is defined as follows:

• Minimal porosity,

• sufficient bonding,

• minimal dilution zone size, and

• No cracks.

Thin walls are builtup using the single-track parameters. As a build-up strategy, a bidirectional approach with fly-in and fly-out movements according to Fig. 2 is applied. In order to achieve a constant standoff between the substrate and the nozzle, the layer thickness is determined iteratively.

In this study, the following influential variables are varied and the effect on workpiece quality and heat development is examined.

• Geometric dimension of the workpiece

• Line energy

  • o Overall line energy per layer,

  • o Local line energy during laser-on time,

• Return time to the center of the workpiece.

Since the fly-in and fly-out movements as well as the pause time contribute significantly to the thermal behavior of the workpiece, the duration of laser-on time $t O n$ and laser-off time $t O f f$ is considered. Due to the high acceleration of the tripod and based on previous research, the shares of the acceleration paths up to the process feed rate and the layer steps are considered to be small enough and, therefore, not relevant for the general and comparative statements on temperature development within this experimental study. For the variation of local line energy, the laser power P_L is varied. After the buildup, the walls are metallographically examined for wall thickness and porosity. For this, cross sections are prepared for two positions of the walls.

To measure the process-induced heat during thin-wall deposition, the workpiece is observed with an ImageIR 8300 thermal camera by Infratec. The detector features a resolution of 640 × 512 IR pixels, with a calibrated measuring range of 200–600 °C at a measuring accuracy of ±1 °C or ±1%. The emission coefficient is set to ɛ = 1. The precise emissivity was not determined. For this reason, the measured temperature is given in arbitrary units. As shown in Fig. 3, the camera faces the long side of the wall at a distance of 0.76 m. For a single image, the camera is triggered after every second layer by a transistor-transistor logic (TTL) signal immediately after the laser was turned off.

The temperature is evaluated using measuring points that were placed in the wall center at a theoretical distance from the top edge of the wall, as shown in Fig. 4. The real distance depends on the variation of the standoff between the nozzle and the wall during buildup. The distance between the measuring points as well as between the points and the nozzle remains constant.

Metallographic cross sections of parameter validation for the single tracks are shown in Fig. 5. In general, tracks with a sound bonding are deposited for feed rates of 30 m/min. Pore formation occurs scattered in the sample welded with a laser power of P_L = 1760 W. No cracking can be observed in the investigated parameter ranges.

In a direct comparison, the track welded with a higher laser power shows a slightly increased build-up height and a larger dilution zone depth. With a maximum dilution zone depth of 110 μm for a laser power of P_L = 1920 W, a high dilution zone can be observed considering the EHLA process.¹² According to the literature,¹² the width of the deposit is mainly determined by the diameter of the laser beam. Since a laser beam diameter of d_l = 1.2 mm is used, this assumption can be confirmed by the width of the tracks.

After depositing single tracks, the process is transferred to thin walls with a targeted deposition of 300 layers. First, the layer height is determined iteratively. For this, the average single-track height (100 μm) is set as the initial Δz-increment and then reduced in subsequent experiments until a good process stability is observed at a Δz-increment of 75 μm.

The parameter variations and resulting return-to-center times as well as effective line energies are shown in Table III.

The lengths of the fly-ins are chosen so that the influence of the effective line energy as well as the return time between the wall lengths is comparable. A minimum fly-in length of 40 mm is selected to reliably exclude the possible influence of machine vibrations. The process is stopped if there is visible process instability.

In the first layers, a good process stability is observed for the buildup of the walls during all experiments. However, the standoff between the nozzle and the substrate changes at the start of the process, as the height built up is greater than the programmed z-offset. The standoff stabilizes when the buildup is sufficiently high so that the standoff of the process reaches the level of the powder focus. Samples of the walls created are shown in Fig. 6. The metallographic cross section in Fig. 6(c) reveals that the wall is thinner up to a certain height until the wall thickness increases sharply and maintains almost constant up to the top of the wall.

A complete buildup with 300 layers cannot be achieved for every wall. In these cases, at a certain height, the buildup is too shallow and the distance between the nozzle and the substrate increases again. This effect intensifies over a small number of layers and, eventually, causes the process to fail.

As can be seen in Fig. 6(b), the defective buildup is particularly pronounced in the center of the wall. Before the process is stopped, it is observed that the buildup is lower in the center of the wall than at the side edges. It can be assumed that less material builds up in the center than at the edges, as the heat accumulates in the edge areas and, thus, more powder is melted. Due to the shallower buildup in the center, an excessive workpiece standoff initially leads to instability there. In general, a longer laser-off time results in a higher buildup for all samples.

The cross section in Fig. 6(c) shows that the walls exhibit a rough surface with loose, adhering particles. This surface could also be observed for other materials at similar feed rates.¹² It can be assumed that this phenomenon can be reduced by suitable parameter adjustments. The accumulation of particles on one side of the wall can be explained by a slightly asymmetrical powder jet.

The thickness of the walls and the porosity are measured by metallographic analysis. A comparison of the porosities for the walls with lower and higher laser powers is shown in Fig. 7.

This study is not designed to optimize porosity. As a result, the porosity is significantly higher than expected from the literature.⁵ The highest porosities are found at a laser power of P_L = 1760 W for the shortest fly-in lengths with maximum porosities of 18%. A greater laser-off time results in a reduction in porosity to 10% for the 35 mm walls and 14% for the 70 mm walls, respectively. Similar results can be seen for the higher laser power (P_L = 1920 W) with overall lower porosities between 9% and 13%. Similarly, there is a decrease in porosity as the effective line energy decreases. However, with identical effective line energies, both an increase and a decrease in porosity can be seen as the return-to-center time increases. A further study to reduce porosity is required.

The evaluation of the wall thickness at different heights is illustrated in Fig. 8. For the lower laser power (P_L = 1760 W), the average wall thickness in the bottom area measures 780 μm, in the mid-section, 1073 μm, and in the top-section, 1119 μm. For the higher laser power (P_L = 1920 W), 813 μm in the bottom area, 1087 μm in the mid-section, and 1129 μm in the top-section, respectively. Thus, a slight increase in the wall thickness with a higher laser power is observed.

In the cross sections (see Fig. 6), the samples show a sharp increase in thickness below a height of 5 mm. It can be assumed that this difference in the wall thickness is caused by the described correction of the process standoff. The continuous increase in the wall thickness can be explained by an increasing workpiece temperature and a resulting increase in the melt pool size. This effect can be confirmed by walls with a length of 35 mm, where the difference in wall thickness between the middle and top sections decreases as the fly-in length or laser-off time increases.

Figure 9 shows the effect of wall length on process-induced heat. For both wall lengths, a complete buildup of the programmed 300 layers fails. Due to the short fly-in of 40 mm, both walls reach almost identical maximum temperatures (corresponding to 500 a.u. in Fig. 9) at the top edge. At this point, a drop in temperature is observed, which resulted from the flattening of the buildup and, therefore, P1 measuring above the workpiece.

This is confirmed by the comparison with the measurements at point P2, where the temperature continues to rise. The drop in temperature can similarly be observed between layer 18 and layer 32 at a wall length of 35 mm, which is caused by a process-induced correction of the workpiece standoff (10 mm) to the powder focus standoff of the nozzle (9 mm). Due to an excessive buildup, P1 measures further away from the top edge, resulting in a colder measuring temperature. As the process time increases, the temperature continues to rise. A direct comparison shows that the 70 mm long wall initially becomes significantly warmer due to the longer energy input, while the 35 mm long wall reaches the maximum temperature more quickly, which leads to earlier process instability. Therefore, it is assumed that the shorter return-to-center time for the 35 mm long wall has a large impact, while the higher effective line energy for the 70 mm long wall has a smaller effect on process instability.

Figure 10 illustrates the comparison of different fly-in lengths and a pause between the individual layers. It is evident that a longer laser-off time leads to a lower temperature. A fly-in length of 57.5 mm leads to a further buildup, but the 300 layer buildup can only be achieved with a fly-in length of 97.5 mm or a longer laser-off time. Process instability and the resulting process termination can similarly be observed at a temperature corresponding to 500 a.u. Similar findings can be made for a wall length of 70 mm.

The previously described effect of the standoff correction can be derived from the temperature measurements for all fly-in lengths of the walls with a length of 35 mm.

Figure 11 shows the comparison of two walls with the same return-to-center time, but different line energy per layer due to varying wall lengths. Although the energy density of the wall with a length of 70 mm is twice as high as that of the wall with a length of 35 mm, only an additional build-up height of approximately 35 layers is achieved. For both walls, a uniform temperature increase from layer 50 is observed, with the lower line energy wall showing a temperature difference corresponding to 40 a.u.

To identify the effect of the return-to-center time, two walls with the same effective line energy per layer but different return-to-center times are compared (35 mm wall with 40 mm fly-in and 70 mm wall with 80 mm fly-in). The results are shown in Fig. 12.

A direct comparison of different return-to-center times shows that the effective line energy has only a minor influence on the process-induced temperatures. Comparable heating rates are visible from layer 30 up to about layer 75 at point P1. From this height, the heat rate is maintained by the 35 mm wall, while the temperature of the 70 mm wall exhibits a degressive profile. This difference results in a complete buildup for the 70 mm wall, while the process for the 35 mm wall already becomes unstable at layer 180.

Using the finding that an increase in fly-in length has a significant impact on the feasible build height, the targeted build height is increased. The number of layers is raised to 2000, resulting in a programmed height of 150 mm. The built-up walls are shown in Fig. 13. The buildup is successful for both wall lengths. Therefore, it is sufficient to adjust the laser-off times in each layer to eliminate the need to adjust the laser power.

From a height of about 120 mm, a slight waviness can be observed on the surface of both wall lengths. It can be assumed that this is not caused by the process, but by component vibration. A possible cause could be the clamping of the substrate in combination with the small substrate size.

A thermographic analysis is shown in Fig. 14. A degressive temperature profile can be observed with an almost constant temperature from layer 750.

In this study, an experimental approach for the systematic variation of the heat input in the EHLA3D process was given. For this purpose, thin walls made of AlSi10Mg were built and the resulting temperature was measured. Subsequently, process adjustments for a uniform buildup were presented.

Initially, single tracks were deposited, and the following key findings were obtained:

• Tracks with sound bonding and a minimum dilution zone of 42 μm can be achieved for a feed rate of 30 m/min.

• Pore formation occurs scattered for the sample welded with a lower laser power (P_L = 1760 W).

After validating the single tracks, thin walls were builtup by using a parameter set with a layer thickness of 75 μm. The wall lengths, fly-in lengths, and the laser power were modified to make the effective line energy and the return-to-center time comparable. The walls were examined metallographically, and the porosity and the wall thickness were determined. The global process-induced workpiece temperature was measured in the wall center by thermographic analysis. The following conclusions were drawn.

• The programmed build height of 300 layers could not be achieved for all walls. Local overheating of the workpiece changed the process initial conditions so that thinner and wider layers were deposited, and the process standoff increased. This resulted in a local thickening of the wall and process instability.

• Porosities tend to decrease with longer fly-in lengths, but a generally high level of porosity of >10% can be observed. A further study to decrease porosity is necessary.

• The wall thickness increases with build height and workpiece temperature. A more uniform wall thickness can be achieved through longer fly-in distances and longer laser-off time.

• By using thermography, the effects of local overheating of thin walls were demonstrated.

• Process instabilities occurred when the temperature of the top edge exceeded a critical value (corresponding to 500 a.u. in the graphs presented).

• Shorter walls overheat faster, whereas longer walls initially heat up more rapidly due to the higher effective line energy per layer.

• The laser-off time has a significant impact on the achievable deposition height. Heating of the workpiece stagnates faster, resulting in a degressive temperature development.

• With identical return-to-center times, a wall with half the effective line energy only achieved a gain in build height of 35 layers. The heat-up rate was similar.

• For identical effective line energies and different return-to-center times, a significantly lower maximum temperature is seen for a longer return time and a significantly higher buildup is feasible.

• The temperature development approaches a constant value, what was shown by increasing the deposition height to 150 mm. Therefore, a laser power adjustment over the build height is not necessary when varying the laser-off times.

In future studies, the thermal analysis for EHLA3D can be extended to further geometries and materials. A comparison of the thermal analyses with simulations is promising in order to allow extrapolation to other workpiece dimensions or more complex geometries. With this knowledge, an optimized path planning can be developed that provides thermally optimized paths and takes advantage of the improved thermal conditions. As the powder mass flow remains constant even during the laser-off time, the powder efficiency for extended fly-in lengths decreases significantly. However, by adjusting the path planning, local processing of different workpiece areas can cancel out this effect. With the help of such path planning, time-consuming and costly parameter studies can be avoided, and the deposition of complex workpieces can be simplified.

The authors have no conflicts to disclose.

Cedric Hauschopp: Conceptualization (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead). Mariana Borba de Souza: Investigation (equal). Ricardo Kaierle: Investigation (supporting); Visualization (supporting). Adrian Häussler: Investigation (supporting); Visualization (supporting). Thomas Schopphoven: Supervision (equal). Wilhelm Meiners: Supervision (equal). Constantin Häfner: Supervision (supporting).