Laser powder bed fusion (LPBF) is a highly promising technique that affords significant advantages in mitigating the high costs and lengthy procedures associated with manufacturing precise and complex components in the aerospace industry. However, the printing process encounters uncontrollable issues, such as fluctuations in laser energy, unstable airflow, and damage to the recoater. These issues can lead to uneven powder spreading thickness, causing deposition defects that critically impact part quality. To improve the formation quality, the deposition defects caused by abnormal powder thickness must be monitored. Despite rapid advancements in online monitoring technologies, the complexity of signal data and its unclear correlation to actual part defects present significant challenges. Establishing the relationship between the deposition defect and monitoring signal for different powder thicknesses is necessary to address the issues related to powder spreading anomalies. Moreover, developing rapid and effective diagnostic methods is crucial to providing a foundation for the feedback control of defects. This study demonstrates the use of an online monitoring system that integrates proprietary photodiodes and high-speed cameras to collect and analyze data across various powder thicknesses. We establish a foundation for the online monitoring and real-time diagnostics of defects by investigating the evolution patterns of part surface quality and internal defects.

In the experimental study, substrates pre-treated with milling are customized with designs of grooves with different depths ranging from 30 μm to 300 μm in 30-μm steps (Fig.4). The powder is spread across these grooves, and a recoater is used to ensure that each groove reaches the designated thickness. Single-layer laser exposure is performed in different areas using different laser parameters (Table 1). The light intensity and melt pool area are monitored online at a frequency of 10 kHz during the printing process by using three off-axis photodiodes and a coaxial high-speed camera. After printing, the surface morphology and internal defects of the samples are characterized using confocal laser scanning microscope and scanning electron microscope. The impact of powder thickness on deposition defects is investigated by integrating online monitoring signals with offline material characterization data.

The mapping images of light intensity and melt pool area distribution (Fig.5) reveal that, as powder thickness increases under the same process parameters, the light intensity gradually decreases, and the melt pool area increases. Additionally, under the same powder-layer thickness, the average intensity of the melt pool decreases with decreasing energy density (Fig.6). The surface roughness increases with powder thickness (Fig.8). For instance, with the laser power at 200 W and scanning speed at 1000 mm/s, the surface roughness increases from 4.16 μm to 117.86 μm as the powder thickness increases from 30 μm to 300 μm. The surface morphology (Fig.9) and internal porosity defects (Fig.10) indicate the following three stages of the melt pool with increasing powder thickness: 1) smooth surface with clear melt tracks and uniform melt pool depth; 2) continuous melt tracks on the surface but large fluctuations in melt pool size (width, height, and depth); 3) melt track discontinuities with the emergence of large balling defects over 150 μm, leading to porosity between melt tracks and at the bottom. Monitoring signals under different powder thicknesses can be categorized into three corresponding stages based on melt pool conditions. The balling state requires the device to immediately detect anomalies and quickly respond in subsequent layers. Receiver operating characteristic (ROC) curve analysis shows that selecting 0.21 V as the threshold for low values and 7.14% as the threshold percentage yields a model with a good capability to identify deposition defects (Figs.12 and 13).

In summary, printing characteristics under different powder thicknesses during the LPBF process are investigated. When the powder thickness increases in LPBF, the surface quality worsens, and internal porosity defects occur. When the powder thickness exceeds 90 μm, large balling defects can exceed 150 μm in size. The relationship between the light intensity collected by photodiodes and melt pool area captured by high-speed cameras is analyzed. The monitoring of the melt pool light signal is highly sensitive to deposition defects due to powder spreading anomalies. As the powder-layer thickness increases, the light intensity decreases, while the melt pool area increases. As the thickness increases from 30 μm to 300 μm, the average intensity of the melt pool decreases from 0.6 V to 0.2 V. On the basis of these results, a novel diagnostic method for deposition defects is devised by employing threshold percentages derived from optical monitoring signals. When the proportion of light intensity values less than 0.21 V exceeds 7.14%, it can be diagnosed as abnormal powder spreading with a true positive rate of 97.22%.