Semiatin et al. 2004). And, while much is known about the dynamic nature of the AM process, other research is leading to new insights into the formation and evolution of defects (Kenney et al. 2021 Quintana et al. 2021), the importance of fluid dynamics (Tammas-Williams et al. 2015 Hojjatzadeh et al. 2019) on the molten pool and presence of any keyhole, and the competition between molecular flow of gas and the vapor- ization of elemental species and their combined effect on the proximal powder (Yoder et al. 2021 Ahsan and Ladani 2020). These new physics are being discovered in sophisticated experimental facilities, including high-energy beam lines, where both high spatial and temporal data can be obtained (micrometer and microsecond). The state-of-the-art mea- surements are beginning to be correlated with some NDE approaches, as these are promising methods to correlate the physical mechanisms associated with AM with signals that can be measured during the AM process. It is clear that for both the realization of many of the promises of AM as well as the determination of different physical domains that the NDE approach has an important role to play. This paper consists of two primary components. First, it provides a brief review of some of what is known about the composition–process–materials state–performance rela- tionships in AM. Elements of this first section will include some aspects of NDE techniques, as relevant. New connections between aspects of the materials state and NDE techniques will be presented. Second, it provides a review of the applica- bility of different NDE techniques for both ex situ and in situ assessments of the materials state, and by extension, initial metrics of the quality of the as-manufactured materials and components. A Review of Additive Manufacturing Jim Williams, an internationally renowned expert on titanium, physical metallurgy, and microstructure-property relationships, as well as a former dean at both Carnegie Mellon University and Ohio State University, once provided the most pithy yet useful definition of AM: “It is the opposite of subtractive man- ufacturing.” In addition to its brevity, this “definition” is useful for two reasons. First, it is implicitly broad, as it does not invoke any of the prototypical details that are typically invoked yet constrict our perspective, such as the heat source (such as laser), geometries of the material that is added (such as molten pool), or incoming material type (such as powder). Second, it implies a capability that is important for the NDE community: the addition of volumes of material means that those volumes can be probed in a manner paralleling (following) the AM technique itself, providing a highly detailed perspective of the materials state. The intellectual property history of AM can most clearly be understood based upon this definition. From a certain perspective, civilization’s earliest methods of manufacturing involved AM, as exemplified by coil pots, which permitted indi- viduals to make clay pottery prior to the advent of the potter’s wheel. However, from a modern perspective, the earliest technical basis for metal-based AM is found in a 1920 patent by Ralph Baker (1920), who patented a method to produce dec- orative articles using electric arc welding to deposit beads of material onto previously deposited beads of the same metal. While this method was cited in other welding techniques in the 1960s, the next notable patent came in 1979 from Brown et al. (1979) while working at the United Technologies Corp. on a US Navy–funded project. In 1979, the inventors disclosed a process for the subsequent deposition of metallic layers that would be capable of producing bulk, rapidly solidified metals. In their work, they termed this technique “LAYERGLAZE,” and in their patent, they included the possibilities of multiple heat sources (including both electron beams and lasers) and of multiple material forms (including both powder and wire). While this was not pursued fully at the time, there is a direct connection between this patent and Sandia National Laboratories’ work on a directed energy deposition system with a powder-blown delivery system and a laser energy source known as Laser Engineered Net Shaping (LENS™), which resulted in the first commercial company for metals-based AM, Optomec, and the first commercial sale in 1998 to Ohio State University. Other key technology patents in the 1980s that have benefited the AM community are rooted in polymeric materials, including the work of Hideo Kodama in 1981 (Kodama 1998), Charles Hull’s work in stereolithography in 1984 (Hull 1984), and Charles Hull’s first 3D printer in 1987 (3D Systems 2021). In the earliest days of metals-based AM using LENS systems, there were simultaneous efforts to understand the processing-property space, including the first appearance of metrics that combined key “feed and speed” parameters into energy density terms (Yin and Felicelli 2010 Hofmeister et al. 2001) understand the composition–microstructure–property space, including the use of elemental blends (Schwender et al. 2001 AlMangour et al. 2017) and produce the initial work into producing composi- tionally graded structures (Zhang and Bandyopadhyay 2019 Bandyopadhyay and Heer 2018 Obielodan and Stucker 2013 Balla et al. 2009). Industries and agencies began to fund work to develop the first processing–structure–property databases and began to place AM metallic parts into service (Collins et al. 2014, 2016). Within the past decade, there have been sustained efforts in developing and integrating computational tools to predict the geometry (including distortion and residual stress), microstructure, properties, and performance of AM parts (Smith et al. 2016a King et al. 2015). The level of sophistication and availability of machines is now sufficiently robust that in 2019, it was even shown that it was possible to 3D print and “fly” a certain superhero suit (All3DP 2021). It is difficult to bound the variations of AM systems. The scale of the systems ranges from aerosol jet-like processes, which have submicrometer resolution and are used to manu- facture functional devices, to large-area AM, which produces parts with dimensions of multiple meters (Lim et al. 2012 Williams et al. 2016). While most metals-relevant AM systems involve fusion (pools of liquid metal), there are other inno- vative AM techniques that are solid state (such as the MELD 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 47

process [Yoder et al. 2021 Griffiths et al. 2019, 2021]) and rely upon frictional or ultrasonic methods of joining. The heat sources for fusion-based techniques include lasers, electron beams, plasma sources, and, most recently, resistance-based techniques (such as Joule heating [Huang et al. 2014 Batista et al. 2020]). The incoming material to be added is prototypi- cally wire or powder, but can also include thin sheet or ribbon (Kobryn et al. 2022 Hascoet et al. 2014). The atmospheres can be equally varied, ranging from vacuum to inert shield gas to deposition in controlled atmosphere glove boxes. This variability has an impact on the composition of the as-deposited materials in fusion-based systems. AM systems can be additive only, or hybrid involving the recursive oper- ation of both additive and subtractive (machining), or other techniques, such as laser peening for local control of the residual stress (Hackkel et al. 2018 Madireddy et al. 2019). Systems can be equipped to deliver material from a single feed source or from multiple feed sources to enable the spatial control of the composition in a preprogrammed manner (Kelly et al. 2021 Schwartz and Boydston 2019). The as-de- posited structures can be free form or supported using lattice structures such as in powder beds (Collins et al. 2016 Davis et al. 2009 Zalameda et al. 2013 Vaissier et al. 2019 Hussein et al. 2013). Similarly, the architectures can be designed to be fully dense, lattices, or with variations of controlled internal cavities (Juechter et al. 2018 Wang et al. 2018 Tao 2016 Gardan and Schneider 2015). Figure 1 provides a broad overview of the types of structures, systems, and processes that exist. Considering these capabilities in aggregate, and imagining future systems where the “material effectors” are selectable to achieve a particular materials state with a particular function in a particular location (a logical fusion of concepts found in AM and metamorphic manufacturing [Xie et al. 2016 Daehn and Taub 2018 Feucht et al. 2020 Love et al. 2013 Good and Landau 2017]), one can begin to conceive of new materials and desirable topologies across a range of length scales. In general, the types of physical processes of fusion-based AM techniques involve: the transfer of energy from the heat source into the material, which includes consideration of the material’s reflectivity to the particular energy wavelength and adjusted for particular configurations that correspond to the efficiency and redun- dancy of energy-impingement events the heat-transfer mechanisms within the part, including radiation, conduction, and convection (Raghavan et al. 2013 Gutowski et al. 2017) the heat-transfer mechanisms associated with the material’s thermodynamics, including phase transformations and any attending enthalpy of mixing of different species (Kumara et al. 2020 Zhang et al. 2019b Kenel et al. 2017) the fluid dynamics of the liquid, including Marangoni convection, buoyancy, gravity, and other complex melt-pool dynamics (Hojjatzadeh et al. 2019 Gan et al. 2017 Khairallah et al. 2016) possible secondary processes in proximity to the molten pool, including capillarity, wetting, sintering, and thermal grooving (Blank et al. 2019 Mullins 1957, 1958) dynamics mediated by the liquid/vapor interface, including volatilization or gettering of elemental species, and the gas dynamics, including kinetic motion of molecules (Sato and Kuwana 1995 Semiatin et al. 2004 Collins et al. 2014) physical processes at the liquid/solid interface, including melting and solidification solid-state phase transformations and the current evolution (and retention) of elastic-plastic defor- mation processes induced by either phase transformations, phase evolution, or transients in thermal gradients. These processes are further complicated by the motion of the heat source, inducing so-called thermal gyrations into the part whose magnitude and frequency are functions of the part geometry and the build scan strategies (G-code). Each of these is sufficiently complex so as to merit their own treatment, and as such lies beyond the scope of this paper. However, an understanding of such broad categorizations is helpful to understand the types of measurements that will be useful in understanding key metrics of the materials state that influ- ence the properties and performance of the material. Table 1 provides correlations between these general types of physical processes and a hierarchy of materials state parameters that govern the properties and performance of materials. For some materials, these correlations have been quantified, and the most important factors have been determined. Important Materials State Factors From the perspective of failure in most metallic systems (assuming a reasonable level of ductility), the principal aim is to understand the presence of defects and damage and their evolution during service. Thus, concepts such as fatigue and fracture go hand in hand with materials state awareness and any attempt to link NDE methods with AM, whether ex situ or in situ. In AM materials, the dominant macroscopic defects include porosity, lack of fusion (LOF), cracking/tearing, and balling. After the macroscopic defects are considered, two other variables demand our attention. The first variable, residual stress, can couple with defects, leading to unexpected failure. Explicitly, it is necessary to state that residual stress is the gradient in the local density of dislocations, which are the atomic-scaled line defects responsible for an alloy’s ductil- ity, and which can lead to considerable strengthening of the material. Thus, a material can have a low average dislocation density but a high residual stress (large gradients), a high dis- location density but a low residual stress (small gradients), or other permutations. The second variable involves grains, their size, and any preferential crystallographic orientation, also known as texture. Texture is increasingly recognized as having an important role on the properties and performance of materials. Texture is very common in AM materials, owing to ME | AMNDEOVERVIEW 48 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