be identified from optical data by correlating multiple images with different lighting conditions and from multiple layers (Abdelrahman et al. 2017). With the help of supervised machine learning, it is also possible to extend the detection capability of cameras beyond LOF defects (Gobert et al. 2018), although it remains difficult to differentiate LOF, porosity, cracks, and inclusions. Mechanical properties such as density and ultimate and yield strengths can also be inferred by optical images captured during in situ mon- itoring of the selective laser melting (SLM) process (Lu et al. 2019). Besides measuring material properties, optical imaging can also provide new insight into the physical phenomena that occur during the AM process. High-speed imaging with microsecond resolution can help in establishing new under- standings of the competition between the volatilization of ele- mental species and the surrounding gas flow to regulate the dynamics of powder motion in powder bed systems away from the molten pool (Matthews et al. 2016), as shown in Figure 4. Research has demonstrated that by including spatters as the process signature driver, a significant increase in the capa- bility to detect under-melting and over-melting conditions is possible (Repossini et al. 2017). Ultrasonic Testing UT is widely used for materials characterization and can be used to evaluate material characteristics such as grain size and the presence and quantity of inclusions and porosity, along with material properties such as elastic modulus and (directly or indirectly) the material’s hardness, strength, and fracture toughness (Nanekar and Shah 2003). Wave speed, attenua- tion, backscatter amplitude, and critical angles are commonly used metrics in UT. In the context of AM, porosity can be cor- related with ultrasonic wave speeds, and the resolution limit of such a method is ~0.5 (Slotwinski et al. 2014). This presents certain difficulties for the materials scientist, where accurate measurement of porosity fractions below 0.5 may be desired. Laser ultrasound is a noncontact UT technique that induces a laser-induced thermal stress that is sufficient to generate 100 μm +0 μs +0 μs +66 μs +132 μs +200 μs +200 μs 100 μm 100 μm 100 μm 100 μm 100 μm Figure 4. High-speed imaging of melt track progression and powder movement under the influence of the hot vapor Bernoulli effect (reused from Matthews et al. [2016] under Creative Commons Attribution License [CC BY]). Distance from array center (mm) LIPA at 3 MHz with XCT overlay AM component Build plate XCT center XCT area Indication area 0 –1 –2 –3 –4 –5 –6 –7 –8 24 20 16 12 8 4 0 –20 –15 –10 –5 0 5 10 15 20 Figure 5. Normalized TFM image using shear–shear wave arrival (reused from Pieris et al. [2020] under Creative Commons Attribution License [CC BY]). ms–1 3400 3300 3200 3100 3000 2900 2800 2700 1 mm Figure 6. Image of TiLG685 showing the internal structure within the large grains. The crystallites are clearly visible, and spatial resolution is about 50 μm (reused from Smith et al. [2014] under Creative Commons Attribution License [CC BY]). ME | AMNDEOVERVIEW 54 M A T E R I A L S E V A L U A T I O N • A P R I L 2 0 2 2 Distance from surface (mm)

ultrasonic waves within a sample. Laser ultrasonic methods are suitable to perform in situ or in the online inspection of parts with very complex geometry in a high-temperature environment (Lévesque et al. 2016). Recent progress in laser induced phased arrays (LIPA) (Pieris et al. 2020) has demonstrated that LIPA is a viable remote, nondestructive, UT technique capable of being implemented as part of an online inspection of AM as seen in Figure 5. It is worth noting that the LIPA system described by Pieris et al. (2020) had some difficulty sizing the defects, but that the positional accuracy was quite good. This issue of sizing may be improved by either optimizing the shear wave frequency or through more sophisticated data processing, perhaps through models that can handle multiple modalities. Spatially resolved acoustic spectroscopy (SRAS) is an acoustic technique that uses surface acoustic waves to map the grain structure of a material (Smith et al. 2014), includ- ing local crystallographic orientation and texture. The use of surface acoustic waves has been correlated with build quality of SLM parts (Smith et al. 2016b). In some respects, SRAS results provide a high spatial assessment of the materi- al’s state, and thus can serve as so-called ground truth when other (cheaper) NDE methods are used and, potentially, fused. Figure 6 provides an example of SRAS grain size and orienta- tion measurement. Acoustic Emission Testing Acoustic emission testing (AE) is an NDE method that measures the elastic energy released in the form of acoustic waves in materials that undergo some type of change (such as plastic deformation, cracking, or rupture) (Ida and Meyendorf 2019). Passive monitoring of acoustic signatures has been performed for a directed energy process, showing variations in acoustic emission signatures that correlate with varying process parameters. Because the technique is passive, little modification is required for integration with AM systems, while exhibiting good sensitivity to crack-like events (Koester et al. 2018). One of the exciting demonstrations of the appli- cation of UT to AM involves the assessment of a type of hybrid AM, where the material state has been tuned through the nonuniform application of a secondary peening process (Sotelo et al. 2021). This work shows that UT can be used to spatially assess differences in the material state, providing a promising pathway for future efforts where the composi- tion and material state may change within a single unitized structure. The attenuation map shown in Figures 7a and 7b suggests that the microstructure of these samples is mostly homogeneous, despite the known heterogeneity introduced by the AM process, and Figure 7c exhibits a pronounced cyclic behavior, which is primarily attributed to microstruc- tural changes imparted by the hybrid process. Electromagnetic Testing From low frequency to high frequency, this family of NDE techniques comprises alternative current potential drop (ACPD), eddy current testing (ECT), and microwave and millimeter wave techniques as well as Terahertz measurement technology. ECT is arguably the most promising technique of these four candidates for metal powder–based AM processes because it offers a noncontact and high-speed way to inspect surface and near-surface features of samples under test. Due to the skin effect that depends on the working frequency and the material’s electrical properties, it is very difficult for ECT to probe deep features (for example, 20 mm deep cracks) for fer- romagnetic materials. However, such depth measurements are possible for nonferromagnetic materials if a special coil design is used (Janousek et al. 2005). Traditional coil-based ECT systems have been proven applicable for surface and near- surface (depth = 1.2 mm, minimum length = 0.2 mm, material = Ti64) cracks in an AM manufacturing environment (Du et al. 2018). Advancement in magnetometer technology has helped to improve the performance of ECT in terms of minimum detectable defect size. A heterodyne ECT system based on a magnetoresistive sensor has been shown to be able to detect surface defects in the order of 100 μm (Ehlers et al. 2020), as seen in Figure 8. Eddy current in array form (ECA) has recently been used for AM process monitoring due to its superior performance compared to its single-channel counterparts. ECA techniques can detect discontinuities, surface irregularities, and undesir- able metallurgical phase transformations in magnetic and non- magnetic conductive materials additively manufactured using laser powder bed fusion (Todorov et al. 2018). Electromagnetic techniques in general are sensitive to bulk electrical properties of the samples under testing, Scan axis (mm) Build direction (+z) 2.4 2.2 2 4 2 0 0 5 10 15 20 2.2 2 1.8 4 2 0 0 5 10 15 20 2.2 2 1.8 4 2 0 0 5 10 15 Figure 7. Attenuation, α (Np/m), maps for: (a) wrought (b) AM and (c) hybrid AM samples. Note the differences in scale (reprinted with permission from Sotelo et al. [2021]). A P R I L 2 0 2 2 • M A T E R I A L S E V A L U A T I O N 55 Index axis (mm) Index axis (mm) Index axis (mm)