Microwave Reflectometry for Online Monitoring of Metal Powder Used in Laser Powder Bed Fusion Additive Manufacturing

This study presents the results of using a millimeter-wave reflectometer system, operating at 150 GHz, for demonstrating the basic efficacy of measuring electromagnetic scattering of metal powder used in laser powder bed fusion (LPBF) additive manufacturing (AM). Metal spatter (spatial) properties—particles ejected during laser interaction with metal powder—are potential indicators of process deviations (from a prescribed manner) or defect formation in a printed part. Electromagnetic modeling of scattering properties of metal powder has shown to be a potentially viable tool for assessing metal powder cloud spatial distribution and other properties. This work takes the next natural step by measuring the scattering properties of a cloud of metal powder. This investigation begins with samples of stationary powder, demonstrating a strong correlation between packing density and the measured output voltage of the reflectometer. The study progresses into detecting the flow of relatively large metal particles (i.e., solder balls) in air and measuring responses of flowing metal powder blown inside a nitrogen-filled chamber. Results crucially confirm that this method can distinguish a cloud of metal powder from the baseline condition where no powder is present. While promising, this investigation represents an initial step in the long journey toward optimizing millimeter-wave methods for integration into real-world LPBF AM systems.

SECTION I. Introduction

Laser powder bed fusion (LPBF) is a widely used metal-based additive manufacturing (AM) technique that selectively melts metal powder layers to create parts with complex geometries. While it offers important benefits, such as low cost and relatively high speed of production, achieving consistent quality in the final part remains a major challenge [1], [2]. This highlights the need for developing online monitoring methods to detect defects or potential for defect generation [3].

In [4] and [5], an online monitoring method for LPBF was proposed, focusing on analyzing the millimeter-wave (30–300 GHz) electromagnetic (EM) scattering properties of spatters—particles ejected during laser interaction with the powder bed. The model operates on the premise that the EM scattering properties of spatters are influenced by their spatial distribution, and therefore should be detectable using a suitable reflectometer (or scatterometer). Deviations from optimal AM processing parameters or defect formation during LPBF AM have been shown to impact the spatial distribution of these spatters [6], [7]. Consequently, monitoring their EM scattering properties presents a promising approach for identifying undesired manufacturing processing deviations that are likely to generate defects in the final printed part. In [5], it was the proposed EM model demonstrated the potential for detecting changes in the processing parameters by analyzing the millimeter-wave EM scattering properties of the spatters, particularly identifying conditions that lead to the formation of defects such as keyholes.

This initial yet foundational study was conducted to evaluate the experimental capability of detecting metal powder both in stationary and flowing forms. A reflectometer operating at 150 GHz, using a lens antenna [8], was employed to measure the response to various metal powder sizes and spatial conditions. The lens antenna provided for a focused beam of ~245 mm away from the lens (focal depth) and with a spatial resolution of ~6 mm. Using this measurement setup allows for focusing the lens at the location of a metal powder (stationary or flowing). The investigation begins with stationary powder samples prepared using different metal particle sizes, revealing a strong correlation between powder packing density, ascertained by an optical method, and the measured output voltage of the reflectometer. Next, the method is used to measure the flowing powder of relatively large particles in the air. Finally, the reflectometer is integrated into a nitrogen-filled chamber to detect flowing metal powder. This step-by-step measurement approach provides preliminary experimental results toward developing a future online monitoring technique.

SECTION II. Measurement Procedure

A. Measurement of Stationary Metal Samples

Fig. 1 shows the schematic of the setup used to measure the reflectometer response from several stationary powder samples. The reflectometer output voltage is proportional to the measured reflected power from a formation (i.e., cloud) of metal particles [8]. A Keysight Synthesized sweeper (8340A) provided a continuous-wave (CW) signal at ~16.6 GHz with an output power of ~5 dBm, which was then fed into an STE-SF906-00-S1, $\times 9$ full-band frequency extender from Eravant, thereby generating the 150-GHz incident signal. The output voltage of the reflectometer was measured using a relatively fast-response Schottky diode detector and recorded using the NI USB-6366 data acquisition (DAQ) system [8].

FIGURE 1.

Schematic of the measurement setup for stationary samples.

Samples were positioned at the focal depth of the lens and within its beamwidth. The reflectometer output voltage was sampled at a rate of 1 Mps ($10{^{{6}}}$ samples per second), and 1000 points were collected at each location as a scanning table moved the sample along the x- or y-direction.

Fig. 2 shows the measured reflectometer output voltages from a single 2-mm diameter metallic ball placed on a small piece of transparent double-sided tape for securing the ball in place during scanning table movement. The results indicate a ~0.05 mV difference when the 2-mm diameter metallic ball was placed in front of the lens compared to the signal produced by the tape (background) alone.

FIGURE 2.

Measured reflectometer output voltage for a single 2-mm diameter metallic ball, and clear tape (background). The scanning table moved along the x-direction.

To investigate the efficacy of this method for detecting smaller metallic particles, similar samples were prepared using metal particles of different sizes. Each sample consisted of a 1 cm $\times$ 1 cm piece of transparent double-sided tape, positioned at the center of the lens beam and at a distance equal to its focal depth. The procedure began by pouring 0.6-mm diameter solder balls onto the tape surface, followed by 0.2-mm diameter solder balls and actual metal powder (X-P44-X8) a cobalt and nickel-based alloy used in actual LPBF AM with an average particle diameter size of $75~\mu$ m. The results for these samples and a clear tape (i.e., devoid of metal particles/balls) are shown in Fig. 3(a)–(e). The results clearly demonstrate that this reflectometer is readily capable of detecting these metal particles/balls.

FIGURE 3.

Samples made of: (a) 0.6-mm diameter solder balls, (b) 0.2-mm diameter solder balls, and (c) metal powder (X-P44-X8), and measured reflectometer output voltages for: (d) samples shown in (a) and (b), and (e) sample shown in (c), and the 1 cm $\times$ 1 cm clear tape. The scanning table moved along the x-direction.

During these measurements, a potential correlation between the packing density of the samples and the measured voltage was observed. To establish a ground truth, a set of similar samples using the X-P44-X8 metal powder with varying packing densities was prepared. Optical images of these samples were produced using an AM4113ZT(R4), a high-resolution digital microscope from Dino-Lite. The images were processed using MATLAB to determine the packing density of each sample, defined as the ratio of the area of particles to the total area of the sample (i.e., 5 mm $\times$ 10 mm). Additionally, the measured reflectometer output voltage from the samples was also obtained using the setup shown in Fig. 1.

Fig. 4 presents the calculated packing densities from the microscopic images on the left axis, alongside the measured reflectometer output voltage on the right axis. The results clearly indicate a strong correlation between the 150-GHz reflectometer measurements and the optical measurements.

FIGURE 4.

(Top) Images of the three samples (5 mm $\times$ 10 mm), and (bottom) packing density of the samples calculated using optical images on the left vertical axis, and reflectometer measured output voltage on the right axis.

B. Measurement of Flowing Solder Balls in Air

Fig. 5 shows the measurement setup used to detect flowing solder balls in the air. Microwave absorbers were placed in front of the lens and behind the flow to reduce unwanted clutter (i.e., reflections from the surroundings).

FIGURE 5.

Measurement setup to detect the flow of solder balls in air.

The solder balls were slowly poured into the air, centered within the beamwidth of the lens, through a glass funnel. The reflectometer output voltage was recorded for a few seconds without pouring the flow in front of the lens, followed by pouring solder balls for several seconds. The measured voltage data was later “smoothed” using a moving average filter with the length of 0.5 s, as shown in Fig. 6, revealing that the flow of two different-sized solder balls is clearly detectable. The recorded signal level for the flow of 0.2-mm diameter solder balls is lower compared to that of the 0.6-mm diameter balls, primarily due to their smaller size, as expected. Additionally, the measured voltage duration is shorter, which is attributed to the smaller poured total volume of the 0.2-mm diameter solder balls relative to the 0.6-mm diameter.

FIGURE 6.

Measured reflectometer output voltage from the flow of 0.6- and 0.2-mm diameter solder balls.

C. Measurement of Flowing Metal Powder Inside Nitrogen-Filled Chamber

The setup shown in Fig. 7 was prepared for measuring blown metal powder (X-P44-X8). Given the very small size of the metal particles, the flow is susceptible to being ignited and catching on fire. Therefore, safety measures had to be taken to reduce this possibility and to reduce safety concerns. To this end, a chamber was designed and built, within which oxygen was replaced by running nitrogen gas. To dispense the powder within this chamber, the RODOS dry powder dispenser and a vibratory feed, both from Sympatec, were placed inside the chamber, as shown in Fig. 7. The reflectometer (i.e., lens) was then securely mounted on one side of the chamber looking normal to the direction of the flowing metal powder. It must be noted that in the actual LPBF AM process, oxygen is vacated from the printing chamber and replaced with an inert gas (e.g., argon) for the same safety reason [1].

FIGURE 7.

Measurement setup used to detect cloud of metal powder blown inside the nitrogen-filled chamber.

Fig. 8(a) and (b) presents the measured reflectometer output voltage for flowing 0.6-mm diameter solder balls, and three different trials of actual metal powder (X-P44-X8) inside the chamber, respectively. In both figures, the recorded data have been smoothed using a moving average filter with a 0.5-s window, as before. The red line in each figure represents the signal recorded when no material was flowing inside the chamber, serving as a measurement baseline reference.

FIGURE 8.

Measured voltage with and without flowing: (a) 0.6-mm diameter solder balls and (b) metal powder (X-P44-X8).

As shown on Fig. 8, flowing metal balls/powder is clearly detectable. Furthermore, Fig. 8(b) demonstrates that the measured reflectometer output voltage converges back to the baseline as the flow ceases, indicating the end of the powder flow.

SECTION III. Conclusion

In this foundational study described in this short paper, we presented a potential and systematic measurement procedure as a first step toward developing an online metal powder monitoring method for LPBF AM by using a lens-focused reflectometer at 150 GHz. The results clearly demonstrated the capability to detect stationary metal balls/particles, flow of relatively large metal (solder) balls in air, and flowing clouds of metal powder blown inside a nitrogen-filled chamber. Additionally, a strong correlation was shown between the packing density of metal powder samples, calculated using optical images of the samples, and the reflectometer output voltage.

It should be noted that while the measurement method successfully detected metal powder clouds within the nitrogen chamber, it does not exactly replicate the dynamic conditions of the LPBF process. This investigation serves as the first of its kind for illustrating the potential for measuring scattering from flowing metal powder—hence a foundational study. Future efforts should consider optimizing the reflectometer properties (i.e., frequency) and consider other types of scatterometers that can look through the Plexiglas (or glass) walls of the chamber without the need to be physically secured into the chamber construction. Additionally, significantly more measurements have to be conducted with respect to powder size, type, and volume content to establish the ultimate efficacy of the method. These will be some of the critical areas of interest for future work. Furthermore, multiple measurements will be required to establish the statistical detection characteristics of the powder. These efforts would help refine the method and pave the way for its eventual integration into practical LPBF AM applications.

ACKNOWLEDGMENT

The authors would like to sincerely thank Joseph Philbert for his extensive contributions to designing and constructing the nitrogen-filled chamber and for his dedicated assistance throughout the measurement procedures.