Effects of Laser-Powder Alignment on Clad Dimension and Melt Pool Temperature in Directed Energy Deposition

Directed Energy Deposition (DED) is a rapidly emerging additive manufacturing (AM) technology for repairing and producing near-net shape components. While DED encompasses many material delivery techniques (e.g., wire-fed, cold spray), powder-blown DED has become popular due to its multi-axis capabilities as well as its ability to fabricate bi-metallic or gradient material structures [1]. Major parameters affecting the DED process include but are not limited to laser (power and beam shape), powder particle (size and shape distributions), powder flow (feed rate, powder profile), laser path (scanning speed and geometry), and environmental parameters. The comprehensive set of process parameters creates not only opportunities and degrees of freedom but also challenges for process uniformity and repeatability [2].

Due to localized material processing, DED involves rapid heating, melting, potential vaporization, and rapid solidification of the powder into the deposited material [3,4]. Specifically, DED includes fast cooling rates and is controlled by non-equilibrium thermodynamics and related kinetics, which are not fully understood through existing theories. This lack of understanding of DED processes has led to the development of process maps and optimization studies that can be used to successfully (or more efficiently) manufacture desired components. Bax et al. [5] developed process maps for nickel alloy IN718 by varying laser power, scan speed, and powder feed rate on a commercial Lasertec 65 3D hybrid machine and utilized dilution ratio and powder efficiency to determine the highest quality single clad for use in repair. Guo et al. [6] utilized the response surface method for the optimization of process parameter selection in multi-layer cladding structures to increase geometric accuracy. The relative error between the predictive model and specimen geometry was within 9%, confirming the effectiveness of the response surface method. Other works have successfully used machine learning techniques to identify valid processing windows in DED [7]. These processing window studies, however, fail to quantify the effect of numerous ancillary processing parameters, such as the degree of alignment between the powder and laser profiles.

In situ monitoring techniques have also been used to correlate online process parameters with clad dimensions. For example, Bennett et al. [8] investigated the relationship between process parameters, solidification rate, and the volume change of the clad by utilizing an in situ infrared camera to monitor the thermal history of the deposited clad. Fathi et al. [9] developed in situ monitoring of clad height using a CCD camera and a feedforward controller by manipulating the scanning speed. Zhou et al. [10] also developed a height control system using a CCD camera as an in situ monitoring source and a dynamic powder splitter as the control input. These studies focus on the monitoring or feedback response for nominal cladding conditions and do not investigate the effect of off-nominal deposition conditions, such as a laser misalignment, on in situ readings.

The powder flowrate and powder flow distribution have been studied through both numerical simulations and experimental methods due to their influence on the melt pool and catchment efficiency, which drive the clad dimensions. Experimental powder nozzle measurements have included the convergence angle, or the angle at which powders flow out of the nozzle tip to meet near the center line of the nozzle, the powder “waist,” or the width of powder flow at the narrowest point [11], particle velocities [12], and powder flow distributions [13]. Eisenbarth et al. [13] developed a method to measure the 3D shape of the powder flow using a precision balance and pinhole, emphasizing the importance of standoff distance and alignment between the laser and powder. Tan et al. [14] created an analytical model to show the relationship between standoff distance and powder flow, where the layer thickness is affected within a multi-layer cladding part due to changes in the powder flow as the nozzle standoff distance changes.

Once the nature of powder flow in DED is established, one may then study its effect on melt pool flow and clad geometry. For instance, a multi-physics model of powder stream flow and melt pool formation by Bayat et al. [15] showed the effect of powder flow on the height/width ratio of tracks, and that higher powder flow rates influence the heat and fluid flow conditions in the melt pool. Previous works have shown that both vertical misalignment and transverse asymmetry in power and powder application influence the clad shape. For example, Zheng et al. [16] investigated the effect of laser power, laser focus position, scan speed, and powder feed rate on the microstructure and porosity content in different geometries. They found that the laser focus position within the powder stream critically influences the surface quality of the components, the growth of the melt pool, and the degree of lack of fusion defects. Kriczky et al. [17] have pointed out that the alignment of the laser to the powder nozzle can alter the melt pool geometry from their optical thermal imaging studies. Other works [18] have studied the effect of angular misalignment between the deposition head axis and the surface normal and found that the resulting asymmetry in the projected laser power causes asymmetry in the clad shape. Despite the extent of these works, there have been very few experimental studies on how local interactions between the laser and powder distributions affect the clad shape. Due to the individually established effects of local powder flow and laser power on melt pool temperature and flow, it was hypothesized that inducing a translational misalignment in the powder and laser profiles would locally affect these conditions in the melt pool, altering the resulting clad geometry. Clad geometry is a critical factor in the success of a build (e.g., the potential mismatch of layer height and designed toolpath as well as the resulting geometric accuracy).

In this paper, the effects of the laser-powder profile alignment on the melt pool temperature and clad dimensions of single-clad squares were characterized to demonstrate the significant effect of laser-powder alignment on melt pool temperature and clad morphology. Through this method, the uniformity and repeatability of builds can be improved, and the importance of laser beam and powder profile characterization is established.

Experiments were performed on an open-architecture custom-built DED system (Additive Rapid Prototyping Instrument, ARPI) [19] that includes (1) 1 kW CW laser (λ = 1070 nm, IPG Photonics, Oxford, MA), (2) process laser head (Precitec, Wixom, MI), (3) powder delivery system (Coax-8 nozzle & GTV powder feeder; Fraunhofer, USA), and motion control stages (PRO165LM, PRO225LM, PRO280 LM; Aerotech, Pittsburgh, PA). The DED process laser head includes a coaxial ring nozzle to deliver both laser and metal powders to the target substrate (low-carbon steel, AISI 1018, McMaster-Carr, Elmhurst, IL) as shown in Fig. 1(a). The metal powder (Inconel 718; 45–106 µm; spherical; AP&C, USA) is carried by Argon (Ar) gas and their flow out of the ring nozzle creates a cone shape that results in a Gaussian distribution at the substrate surface. The process laser beam also has a Gaussian distribution and is typically located at the center of the powder distribution as indicated in Fig. 1(b).

Melt pool temperature was monitored by using on-axis photodiode-based Planck thermometry (PDPT) (PrintRite3D, Sigma Labs, Inc.) [20,21]. PDPT uses two color wavelengths to determine melt pool temperature using two Avalanche Photodiodes (APD410A; Thorlabs) at 100 kHz bandwidth, similar to how a two-color pyrometer operates. The APDs are installed coaxially to the optical laser path so that the emitted light from the melt pool is directly delivered to the APDs. An optical profilometer (InfiniteFocus G4, Alicona) was used to characterize the clad width, height, and peak position. The resolutions of the profilometer are 2 μm (vertical) and 10 μm (lateral). It is noted that this resolution cannot capture fine surface roughness changes; however, the dominant mechanism for surface roughness in powder-blown DED is powder particles adhering to the heated clad, which tends to obscure these finer changes regardless of the measurement technique.

A set of experiments was designed to investigate the effects of laser-powder alignment. First, the laser beam and powder flow profiles were characterized, and a method to align them was established. Then, three concentric squares consisting of a single clad were deposited on the same substrate under different alignment conditions, including (1) aligned, (2) misaligned (500 μm offset), and (3) realigned as shown in Figs. 2(a) and 2(b). In the misaligned case, the laser center was shifted in the desired direction and the shift distance was measured with a laser beam profilometer. After the deposition with misalignment, the laser and powder were realigned at the same point. This condition is termed the “realigned” condition as shown in Fig. 2(b). The process parameters remain the same as shown in Table 1 except for alignment and cladding direction conditions.

The effects of laser-powder alignment are characterized using the melt pool temperature, the clad width and height, and the location of the center of the clad cross section. To characterize the laser beam, a scanning-slit optical beam profiler (BP-209, Thorlabs) was used to measure the beam diameter (1/e² width assuming a Gaussian beam) and its center. A guide beam (λ = 658 nm, <1 mW, Gaussian-shaped) was used instead of the process laser due to the maximum irradiance limit of the beam profiler. It is noted that the guide beam does not directly represent the laser beam diameter; however, it is used here to approximate the process laser beam size since a relationship between the melt pool size and guide beam size can be established. For powder flow characterization, a weigh module (WMF204, METTLER TOLEDO) was used to measure the powder feed rate within a 2D deposition plane at a given standoff distance. A plate with a precision-drilled central pinhole of 0.5-mm diameter was placed between the weigh module and the laser process head [13]. The 0.5-mm hole was selected to maintain the same effective pinhole area using a mean powder diameter of 75 µm for this study as was used and defined by Eisenbarth et al. [13]. Then, the laser process head moved in a raster pattern above the pinhole. The carrier gas and powder flowed through the pinhole as the nozzle moved over it. The powder flowing through the pinhole was collected on the weigh module, and the local powder feed rate (within the powder flow stream) was calculated based on the nozzle movement speed and pattern. Thus, both the spatial and temporal distributions of the powder, or the powder mass flow rate profile, can be determined. For a schematic of the powder profile measurement, please refer to the original publication [13].

The powder mass flowrate profile was calculated using the slope of a linear fit from weigh module data. The Ar gas flow was at a steady-state and added a constant offset to the mass measurement; therefore, there was no significant effect on the measurement of the powder flow rate. Powder particles bouncing out of the collection crucible were also minimal. It is also expected that erroneous powder bounces at the pinhole surface would contribute little to the error in the powder flow rate measurement since the nozzle has a constant stream of argon that directs the powders in a conical shape; therefore, powders impact the pinhole surface at an angle and their rebound “bounce” at the pinhole surface would cause them to typically rebound away from the pinhole. It is noted that the measured amount of powder flowing through the nozzle at any given time can change based on the local size distribution of the powder particles, the maintained pressure in the carrier gas flow, and uncertainty in the weigh module. For further discussion on measurement error, the reader is referred to the study by Eisenbarth et al. [13] where the method was developed.

The measured powder profile is shown in Fig. 3. The total mass flow rate of the powder feeder was set at 13 g/min. The raw data from the measurement include the powder mass amount and the position of X and Y. The powder mass amount was calculated as the derivative with respect to time (marked as powder feed rate) as shown in Fig. 3(a). In order to retain as much point-by-point information in the powder profile as possible, a Gaussian fit for each 2D scan line in the X- and Y-directions was performed and then combined in 3D coordinates. This leads to the preservation of a mildly elliptical powder profile with different “diameters” in the X and Y directions. The method used for fitting also leads to a less smooth fit than typically reported. The 1/e² diameters of the powder flow profile were 3.93 mm and 4.34 mm for the X-axis and Y-axis, respectively. The 1/e² diameter of the laser beam was 2.24 mm, as characterized by the beam profilometer.

In addition to the individual characterizations of the laser and powder feed, the alignment between them must also be determined. This was accomplished by (1) placing the pinhole at the center of the laser beam, (2) scanning a powder flow profile, (3) fitting the powder flow profile to a Gaussian distribution to find its center, (4) adjusting the laser beam position within the powder distribution, and (5) repeating these steps until the center of both the laser and powder flow profiles were coincident within 0.1 mm. An offset of 500 μm was chosen considering the diameter of our laser beam (2.24 mm) and powder profile (3.93 mm/4.34 mm). A 500 μm offset leads to an approximately 12.5% change in distance within the powder profile but a much larger change in the local specific energy density since the powder profile is Gaussian. Additionally, 500 μm can be spatially resolved with the utilized beam profiler. Further offset values were not pursued in this study due to the time-intensive process of aligning and re-aligning the laser beam with the powder profile.

First, the effect of alignment on the center positions of single clads is observed. In each of the misalignment cases with an offset of 500 μm, the center position of the single-clads is moved by 483 ± 27 μm. This result shows a straightforward relationship between the laser position and the corresponding deposited position of single clads. Figure 4 illustrates the melt pool temperatures under different alignment conditions and cladding directions, either Clockwise (CW) or Counter-Clockwise (CCW). Three observations can be made. First, there is no significant change in the average recorded temperatures between the CW and CCW-deposited clads. Second, the offset direction matters as indicated in the comparison between the temperature profiles for the X- (Figs. 4(a) and (c)) and Y-direction misalignments (Figs. 4(b)–4(d)). Third, the alignment procedures are reliable since the temperature distribution in the realigned case (the smallest concentric square on each substrate) is comparable with that of the initial aligned experiment (the largest concentric square).

It was thought that an offset in either the X or Y direction would lead to different “leading” or “trailing” conditions between the laser/melting area and powder flow area in the clockwise (CW) and counterclockwise (CCW) cases. For example, if the laser was offset in the positive X-direction and the nozzle was traveling in the positive X-direction, then the laser would be “leading” the powder flow; however, if the nozzle was traveling in the negative X-direction, the laser would be “trailing” the powder flow. The small offsets used in this study did not make a significant difference in the average melt pool temperature when considering nozzle travel direction in relation to the offset direction by comparing the CW and CCW cases. For X-direction misalignment, the changes in temperature from the aligned case were 82.2 K and 62.8 K for the CCW and CW scans, respectively. For Y-direction misalignment, the CCW and CW cases resulted in respective temperature changes of 12.9 K and 11.5 K. Clearly, the direction of misalignment produces a more substantial effect on the melt pool temperature than does the scanning direction. However, note that within the X-direction offset cases, the temperature increase was most pronounced when the laser was leading the powder flow (bottom clad in 4a), top clad in 4c)). When the laser misalignment was behind or transverse to the direction of deposition, the temperature increase was still observed, but to a lesser extent. This leading/lagging/transverse misalignment dependence was less pronounced in the Y-direction misalignment cases, due to the smaller amount of laser misalignment relative to the dimension of the powder stream in the Y-direction. There is a higher temperature along a single edge of the square substrate in Figs. 4(b) and 4(d), and this is mainly due to the substrate alignment. There were four separate substrates mounted in the chamber, and the experiments were conducted concurrently. The local level difference within each substrate may cause this temperature difference because this happens repeatedly in the same substrate.

The detailed temperature values on each side of each substrate are plotted in Fig. 5. The measured temperature changes between the aligned and misaligned cases are up to 100 K for the X- and 40 K for the Y-direction offsets. Misalignment causes higher melt pool temperatures compared to the aligned case because less powder is delivered at the point where the laser beam is focused. This will be discussed later through energy density estimations. The measured temperature changes are significant when it is considered that the offset distance was only 500 μm. Based on our experience, an average melt pool temperature increase of 20 K is induced when laser power is increased to 600 W (20% increase) from 500 W (under the conditions in Table 1 except for laser power). Zhang et al. [22] found that a laser power increase of 20% in the deposition of Ni-based superalloy causes an increase in melt pool temperature of up to 100 K, which is significant for process control. Considering the many efforts on controlling melt pool temperatures during the process [23–26] in which laser-powder alignment was not considered, the presented results strongly suggest that the alignment between the laser and powder profiles needs to be considered as a process parameter or at least a considerable factor when powder-blown DED machines are characterized.

The different phenomena for either X- or Y-direction offsets can be interpreted by the shape of the powder profile. Notably, the change in melt pool temperature was less significant in the case of Y-direction misalignment compared to the case of X-direction misalignment. The powder profile measured in Fig. 3 is elliptical, with semi-minor and semi-major axes b and a of 1.965 and 2.170 mm, respectively. This corresponds to an eccentricity of e = 0.42, where e takes its usual definition of $e=1−b2/a2$⁠. Powder profile axis a is aligned with the Y-axis and b is aligned with the X-axis. The difference in semidiameters is 410 μm, comparable in scale to the induced misalignment of 500 μm. Consequently, the misalignment was lower relative to the powder diameter in the Y-direction, compared to that in the X-direction. The lower misalignment in Y relative to the size of the powder stream could be a strong contributing factor to the reduced temperature and clad dimension change in Y-direction misalignment, when compared to X-direction misalignment.

Figure 6 illustrates the correlation between changes in temperature and in clad dimensions. The relative dimension changes under different alignment conditions are calculated based on the measured dimensions in the aligned case, i.e., the width of 1.6 ± 0.1 mm and height of 1.1 ± 0.1 mm. The overall trend shows that a larger melt pool temperature change leads to larger dimension changes in both width and height. Consistently, the results in the X-offset cases show larger changes in both melt pool temperature and dimensions as shown in Figs. 6(a) and 6(b) due to the shorter diameter in the X-axis, resulting in a larger mismatch between laser power and powder flow.

As all the experiments employed the same process parameters except for the alignment conditions, the nominal energy density values are the same, i.e., 117.7 J · s/mm² · g. To understand the alignment effects, a mesh grid of 4 mm × 4 mm is generated, and the 2D Gaussian profiles are applied to laser power (P) and powder mass flow rate (⁠$m˙$⁠) based on the characterizations conducted as shown in Fig. 3. The estimated energy density distributions are plotted in Fig. 7. Note that the dark streaks appear in Fig. 7 due to sharp discontinuities in the experimentally measured powder profile used for energy density calculation. The peak energy density distribution becomes wider for the +X offset case (bright spots appear more frequently in Fig. 7(b)) compared to the aligned and +Y offset cases (Figs. 7(a)–7(c), respectively). To provide quantitative estimates of the differences between the aligned and misaligned cases, the standard deviation of specific energy density is calculated. The standard deviation values are 4.05, 4.45, and 3.88 J · s/mm² · g for aligned, laser spot shifted to +X direction by 500 μm, and to +Y direction by 500 μm, respectively. The results show a 10% difference in the standard deviation of specific energy density between the aligned and misaligned cases. These findings are consistent with the previous discussion, i.e., that (1) misalignment leads to higher melt pool temperatures because less powder is delivered where the laser beam is formed, and (2) the Y-axis diameter is 410 μm larger than the X-axis diameter, which is comparable to the offset amount of 500 μm. This difference leads to lower temperature changes in the Y-axis compared to the X-axis due to the offset.

While this study evaluated changes in the location of the laser beam within a powder profile, the inverse problem of the powder profile changing with respect to the laser beam can also be understood through the presented analysis. For example, a powder profile in the case of a nozzle blockage will become skewed and asymmetric where less powder is delivered to one side (or more) of the laser beam area. The decrease in powder delivery will lead to changes in melt pool temperature as well as a significant change in clad dimension. It is also noted that standoff distance can influence both powder flow and laser beam profiles [28], which will lead to changes in clad dimension and melt pool temperature as discussed in this study.

This study investigated the effects of laser-powder alignment on melt pool temperature and single-clad dimensions to demonstrate the importance of the mutual alignment of their profiles. When comparing aligned and misaligned geometries, up to 20% of dimension changes in both width and height occur and significant melt pool temperature changes of up to 100 K, which are considerable changes even for a small 500 µm offset. The change in height is particularly critical to the process as failure to correctly predict layer height leads to error accumulation in the nozzle-clad standoff distance for a fixed vertical traverse between layers.

The specific energy density with consideration of 2D profiles in both laser and powder distributions was applied to understand laser-powder alignment, and this study shows that it is important to pay attention to the alignment of the laser and powder profiles when reviewing process parameters. With these considerations, process maps can be fully understood, and the effects of nozzle blockages or laser misalignment can be evaluated in advance.

This research received funding from the DEVCOM Army Research Laboratory under Cooperative Agreement Numbers W911NF-20-2-0292 and W911NF-21-2-02199. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing official policies, either expressed or implied, of the Army Research Laboratory or the US Government. The US Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation herein. This work was also supported by the National Institute of Standards and Technology (NIST)—Center for Hierarchical Material Design (CHiMaD) under Grant No. 70NANB19H005.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.