Process-microstructure-corrosion of additively manufactured steels: a review

Abstract

Among the alloy systems for the fabrication of metallic components using additive manufacturing (AM), steels as a broad family of alloys attract significant attention owing to their established mechanical and corrosion properties under various loading and environmental conditions. In an erosive/corrosive environment, the possible failure mechanisms necessitate consideration in materials selection and design. Although functional alloying elements are deliberately added to steels to improve corrosion resistance, the crucial role of initial microstructure and post-process heat treatments are equally critical and cannot be neglected. Specifically, the proper selection and optimization of AM process parameters are essential steps in minimizing microstructural defects. This comprehensive review provides a consolidated treatise of powder atomization techniques, AM process parameter control, rapid solidification principles, their roles on the as-built microstructure. The role of different phases and microstructures formed during post-process heat treatments, and corrosion behavior of various grades of AM steels are also presented. The review presents the interplay between process-microstructure-corrosion properties of the AM steels through the entire fabrication process - from powder feedstock to final AM product exposed to harsh environments. In the first section of this review paper, the authors comprehensively explain the important factors, which affect the processability and quality of AM steel components. In the second section, the authors focus on the microstructure of both as-printed and heat-treated AM steel products, the parameters which determine variations of the microstructure, the effect of microstructural features such as the grain size and the dendrite arm spacing on the mechanical properties, and the detrimental effect of inclusions embedded into the matrix. Finally, the third part of the paper is a comprehensive review of the corrosion of different types of stainless steels (316 L austenitic stainless steel, 420 martensitic stainless steel, martensitic precipitation-hardened stainless steels of 15-5 PH, 17-4 PH, and Corrax) fabricated by AM techniques.

Keywords:

• Additive manufacturing
• corrosion
• stainless steels
• powder atomization
• powder bed fusion

Additive manufacturing (AM) is a family of processes to fabricate near-net-shape components with complex geometries and minimal waste generation – directly from 3D models.^[1] With a growing demand for both AM development and applications in the aerospace, marine, and automotive industries, numerous studies on metal additive manufacturing (MAM) have been recently conducted.^[1,2] According to Scopus’ database, MAM that mostly includes laser powder bed fusion (LPBF), directed energy deposition (DED), and wire-arc additive manufacturing (WAAM), has attracted tremendous attention over the past decades – as shown in Figure 1.^[3]

Figure 1. Research outputs on the topic of “metal additive manufacturing” (a) documents by year, and (b) documents by subject area (Adapted with permission from^[3]).

As with any manufacturing technique, AM has its own advantages and disadvantages – that require an understanding of process sequences and physical metallurgy. For instance, while AM is able to produce a dense component with a unique geometry, some process parameters e.g., powder atomization atmosphere, atomization velocity, laser power, laser scan speed, etc. directly determine the level of structural defects such as porosity and cracks, which causes a significant deviation in density and other properties. Besides, the effect of other parameters such as initial chemical composition, transformation temperatures, and thermal conductivity of the material should not be overlooked. Therefore, to minimize any unexpected differences between the properties of an AM component and the ones theoretically reported, an overview on both powder production techniques and the AM process parameters is presented in section one.

Materials selection, powder production technique, powder characteristics like sphericity, and size distribution, AM technique and its process parameters directly/indirectly change the microstructural features.^[4–6] The so-called “as-built” microstructure can be investigated in terms of grain orientation, size, morphology (columnar, dendritic, equiaxed), solidification pattern (planar, non-planar), phases, and level of porosity and microsegregation. Such features are related to interconnected thermophysical factors e.g., diffusion coefficient, interatomic potential, temperature gradient, and solid/liquid interface velocity; with rapid solidification during AM essential to be understood. For instance, rapid solidification accompanied by a steep temperature gradient during AM introduces an anisotropic fine microstructure that contains mostly elongated columnar grains; however, reheating cycles can allow the formation of fine equiaxed grains.^[6,7] Depending on how fast solid/liquid interface movement and cooling ahead of solidification front are, microsegregation can take place only within a very narrow intergranular area, while the majority of alloying elements is trapped into the solid lattice.^[7,8] As other structural features, defects such as lack of fusion (LOF), gas and shrinkage porosities can be named, as their formation is highly dependent on the AM process parameters. Besides, the AM products, particularly those vertically printed, show a considerable magnitude of residual stresses, which is high enough in some cases to act as a driving force for partial recrystallization during post-process heat treatment.^[9] For industrial purposes, the steel product is rarely used in its as-built condition and is usually undergone to post-process heat treatment for residual stress relieving, homogenizing, hardening, and strengthening through solutionizing/aging treatments. In this regard, various steel grades present significant differences in phases formation, hardening tendency, and segregation healing, etc., where all recalls the importance of understanding the primary microstructure, as well as thermal history and applied heat-treatment cycles. Section two explains all these factors, which ultimately determine the corrosion resistance of the AM steels.

Numerous researchers have reported that grain size, grain orientation, porosity, segregation, phase fraction, residual stresses, and surface roughness have crucial roles on corrosion properties – particularly pitting – in AM steels.^[10–12] Generally, an increase in grain size or volume fraction of porosity and segregation, can be detrimental toward pitting of AM components. However, the effect of grain size on general corrosion resistance is still unclearly defined and related to many factors including environmental conditions and stability of the passive layer. While porosity is reported to be a preferential pit nucleation site whether inside the grain or at grain boundaries (if the porosity volume fraction is > 1%), segregation causes a microgalvanic coupling between the solute-rich zone and adjacent depleted matrix, influencing stress corrosion cracking (SCC) in the presence of as tensile stress, and localized corrosion at the melt pool boundary.^[10,12] Electrolyte stagnation in concaved areas is reported as a resultant of surface roughness, which may also influence localized corrosion – as reported in AM 17-4 PH steel.^[10] In AM 316 L stainless steel, a high level of residual stresses and an anisotropic grain structure along the building direction are both reported to decrease corrosion resistance and SCC susceptibility, respectively.^[10] Finally, matrix phases like martensite and austenite with different resistance to corrosion, and those formed as inclusions such as MnS, Fe₂O₃, and Cr₂O₃, vary the corrosion behavior of AM steels through different manners.^[11] Therefore, to deeply comprehend the effect of each parameter on corrosion properties, the principles, and mechanisms of corrosion in AM steels are required to be discussed, as presented in section three of this review.

In this study, we present a critical review of literature that encompasses a comprehensive analysis and evaluation of scholarly publications, books, articles, and other relevant sources. Our review focuses on the process parameters of steel components fabricated through additive manufacturing (AM) techniques, as well as microstructural and corrosion studies on stainless steels produced by AM processes. To achieve these objectives, our literature review follows a systematic approach with distinct components. Firstly, we established the objective and search strategy, aiming for a comprehensive search of references related to steel components fabricated using different AM techniques. We particularly emphasized the impact of various process parameters on the production of defect-free and high-quality steel components. Our search encompassed a comprehensive study of relevant databases, exploring key factors influencing the processability of additive manufacturing for steels. We delved into powder production processes, ensuring the suitability of powders for diverse AM machines used for metallic component printing. Additionally, we investigated the correlation between process parameters and the resulting microstructure and mechanical properties of the components. Understanding that the microstructural characteristics of AM components differ from those produced by traditional methods, we focused on studies examining the microstructural features developed after the AM process on steel parts, along with the effects of post-heat treatment processes. Furthermore, considering the close relationship between microstructural features and corrosion behavior, we dedicated our final step to seeking relevant references on the corrosion behavior of steel components, with a specific emphasis on stainless steels fabricated through AM techniques. For this review paper, we sourced information from various reliable and pertinent literature published in scientific database sites, including Scopus, Google Scholar, and ScienceDirect. These sources provided a comprehensive array of studies and insights necessary for our analysis.

In conclusion, through our critical review of literature, we conduct a comprehensive analysis and evaluation of existing scholarly publications, books, articles, and other relevant sources. By focusing on process parameters, microstructural studies, and corrosion behavior of steel components fabricated using AM techniques, we have gained an in-depth understanding of the current state of knowledge while identifying research gaps. Our findings and insights aim to guide future research in this area.

1. Key factors affecting the processability in additive manufacturing of steels

1.1. Powder production and properties

The properties of the metal powder employed in additive manufacturing (AM) need to be determined, to correctly select powder for fabricating components with predictable properties.^[13] To achieve acceptable integrity in the microstructure of complicated parts via powder fusion AM, powder characteristics play an important role, as they determine the relative density of the product. To this end, particle size, sphericity, wettability, oxidation, chemical composition, flowability, thermal conductivity, and surface roughness are the most prominent properties of metal powders.^[14] A suitable selection of metal powder not only directly affects the physical and mechanical properties of the as-built product, but also allows the manufacturer produce hierarchical microstructures in terms of chemical composition, phases, and structural morphology, called microstructure architecting.^[15] Therefore, understanding the powder fabrication methods, properties, and defects is a breakthrough, as discussed in the following sections.

1.1.1. Powder atomization techniques

There have been different methods to produce metal powder used in AM such as water atomization,^[16,17] gas atomization,^[17] ultrasonic atomization,^[18,19] centrifugal atomization,^[20] and plasma rotating electrode process.^[21] Regarding materials selection, powder size distribution, and consolidation ability, each method has its own technical and economic advantages and drawbacks. In this section, the methodologies, technical considerations, limitations, and applications of each method are presented.

1.1.1.1. Water atomization (WA) technique

This technique is known as a cost-effective approach to produce a high volume of metal powder, where the liquid metal is poured into a tundish located on top of an atomization chamber.^[22] An interplay between low pressure in the chamber and the melt hydrostatic pressure leads to the melt flow within a nozzle at the bottom of the tundish. To continue the process with a consistent production rate, the melt level in the tundish as well as the chamber’s internal pressure should be kept stable.^[22,23] As shown in Figure 2, the atomization zone is placed right under the melt nozzle in which the melt stream collides with high-pressure water jet/spray exited out of the water nozzles. The apex angle is an angle between the melt and water steams, which is a key factor determining the powder’s geometrical and physical characteristics. In conventional WA systems, there have been a number of water nozzles symmetrically placed on the top corners of the atomization chamber. Once the operating pressure increases, the continuous stream of water breaks up to separate droplets, interacting with the melt stream to produce liquid metal droplets. Subsequently, the liquid metal droplets start to solidify while passing through the bottom of the atomization chamber. Their attachment and deposition at the bottom of the chamber provide the metal powder to be collected, dehydrated, filtered, annealed, classified, and packaged as the powder feedstock.

Figure 2. Schematic design of (a) water atomization, and (b) gas atomization processes of metal powder (Reproduced with permission from^[22,31]).

In terms of materials selection, there is a limitation accompanied by the WA process; as an induction furnace is used to melt the raw material, some alloying elements like carbon and manganese cannot be removed in steel processing.^[24] Therefore, carbon and manganese remain in the final product, necessitating more attention, particularly for low carbon feedstocks. Controlling the manganese, on the other hand, is important to decrease the oxidation severity taking place during the WA process.^[25] Another limitation coming from the WA process is the shape and morphology of the resulting metal powder. The powder produced is highly irregular compared to the other methods, which results from high cooling and spheroidization rates. For instance, the cooling rates in the WA are 10–100 times larger than the gas atomization process.^[23,25]

To correlate the mass median particle size, $d_{50},$ to processing parameters, e.g., water pressure, $P_W,$ water velocity, $V_W,$ and nozzle apex angle, $\alpha ,$ Persson et al.^[26] presented a revisited model based on the Bergquist model^[27] as follows (1) $$d_{50}=k\frac{{\gamma }_M^{0.80}{\mu }_M^{0.93}{D}_M^{0.21}}{P_W^{0.98}\hspace{0.17em}\sin \hspace{0.17em}\alpha }{\left(\frac{m_W}{m_M}\right)}^{-0.04}$$(1) where $\gamma ,$ $\mu ,$ $D,$ and $m$ are surface tension, viscosity, diameter of stream, and mass flow rate, respectively, $k$ is constant, while $M$ and $W$ subscripts stand for liquid metal and water materials. Equation (1)(1) $$d_{50}=k\frac{{\gamma }_M^{0.80}{\mu }_M^{0.93}{D}_M^{0.21}}{P_W^{0.98}\hspace{0.17em}\sin \hspace{0.17em}\alpha }{\left(\frac{m_W}{m_M}\right)}^{-0.04}$$(1) for two types of nozzles and atomizer setups can be rewritten as^[28] (1a) $$d_{50}=114P_W^{-0.58}\mathrm{ }\left(\text{in annular nozzle}\right)$$(1a) (1b) $$d_{50}=68P_W^{-0.56}\mathrm{ }(\text{in open V}-\text{jet nozzle})$$(1b) in which the finer metal powder can be produced via the open V-jet nozzles.

1.1.1.2. Gas atomization (GA) technique

The most common way to produce the metal powder used in AM is the gas atomization (GA) technique, where a high energy flow of an inert gas atomizes the melt upon pressure.^[29] The spherical droplets of the atomized liquid metal solidify once the temperature drops down the solidus temperature. As shown in Figure 2, the atomization setup is very similar to the WA process; however, a GA nozzle may be categorized as a close couple, De Laval, and free-fall types. In a close couple nozzle, pressurized inert gas flows only a minimal distance before a collision with the melt droplets, which is far more efficient in atomization than a free fall nozzle.^[30] In addition, the gas used in GA method has a profound effect on the efficiency of this process. Helium (He) proves to be a significantly superior gas for atomization in comparison to Argon (Ar). Helium possesses the ability to absorb and transfer heat at a much faster rate than Argon. This characteristic enables the droplets to cool more rapidly, resulting in more effective outcomes.^[31]

Generally, the GA process can be categorized in terms of the method of melting the metal, or nozzle design, as illustrated in Figure 3. In the first category, the melting mechanism can be induction heating, plasma torch, plasma atomization, or electrode induction melting gas atomization (EIGA). In the induction heating and plasma torch, a crucible is used to keep the melt at consistent temperature; however, in the plasma atomization technique, a metal wire passes through a pair of plasma torches and melts to liquid metal droplets.^[29] In the EIGA process, the rod-shaped metal is heated via an induction heating system, and subsequently, the liquid droplets fall into the atomization chamber with no contact with the other parts. Hence, the melted material obtained by the plasma atomization and EIGA techniques is usually less contaminated by ceramic materials, as there is no crucible used in these processes.^[32]

Figure 3. Gas atomization techniques based on melting mechanisms and gas nozzle design (Reproduced with permission from^[29]).

In the second category, the GA process is identified by various types of nozzles. In the free-fall nozzle design, the extrinsic force to push the liquid metal through the nozzle is gravity, where the main disadvantage of this technique is the slow flow rate of the melt.^[29] The metal powder collected is more than 50 $\mathrm{\mu }\mathrm{m}$ in diameter that is too coarse for the AM processes.^[25] In the close-coupled nozzle design, the liquid metal is pulled down by the atomization gas and the flow rate is adjustable via changing the atomization gas flow. In this configuration, the melt flow rate is much higher, and fine particles with 10 $\mathrm{\mu }\mathrm{m}$ diameter are achievable.^[25] In the De Laval nozzle, the melt flow is accelerated to supersonic speed, thanks to a laminar gas flow.^[29] The particle size in this technique is between 15 and 45 $\mathrm{\mu }\mathrm{m},$ which also proposes a narrow distribution of the particle diameter.^[33] Gas consumption in the De Laval nozzle is much lower than the other techniques.^[29] As an important consideration in the free-fall nozzle and close couple nozzle, freezing the melt stream is possible inside the nozzle, and consequently, the process abruption can take place.^[29]

The melting process in the GA technique is implemented either under a protective atmosphere or vacuum, which differs from the melting process in the WA technique. Therefore, the liquid metal in the GA contains less contamination, e.g., oxides and nitrides than in the WA technique.^[23] Regarding lower oxidation rate during the process, there is no necessity to dictate more manganese content to mitigate severe oxidation of the metal.^[25] During the GA process, air, nitrogen, argon, and helium are the most common gases that entered into the atomization chamber and interacted with the melt stream.^[23] Despite the WA process in which the diameter of the particles is highly correlated to water jetting pressure, the gas/metal ratio is the controlling factor affecting the particle size in the GA process. However, the prevalent disadvantage of the GA technique is the gas entrapment in the melt droplets resulting in porosity formation particularly when the argon gas is used for the atomization.^[25] During the atomization, the gas is probably formed due to mechanical entrapment taking place during particle collision.^[25]

In terms of $d_{50},$ Lubanska^[34] proposed Equation (2)(2) $$d_{50}=d_{0}k{\left[\frac{{\vartheta }_m}{{\vartheta }_{g}We}\left(1+\frac{m_m}{m_g}\right)\right]}^{0.5}$$(2) to reveal the advantages of GA technique for the production of fine metal powder. (2) $$d_{50}=d_{0}k{\left[\frac{{\vartheta }_m}{{\vartheta }_{g}We}\left(1+\frac{m_m}{m_g}\right)\right]}^{0.5}$$(2) where $We$ is the Webber number and is defined as $We=\left({\rho }_m{d}_0{\left(\Delta \vartheta \right)}^2\right)/{\sigma }_m,$ $\Delta \vartheta$ is the ratio of inertial force correlating to the relative velocity between the melt and gas, ${\rho }_m$ is the density of the melt, ${\sigma }_m$ is the melt surface tension, and $d_0$ is the melt stream diameter. Besides, $k$ is a constant, while ${\vartheta }_m,$ ${\vartheta }_g,$ $m_m,$ and $m_g$ stand for the kinematic viscosities and mass flow rates of the melt and atomizing gas, respectively, $d_{50}$ is defined as the mean particle diameter.

An increase in supersonic gas velocity shows its influence on the Webber number expression by changing $\Delta \vartheta$ term. This term is the controlling factor to promote the melt stream thinning and ligamentation during the atomization process.^[35] To obtain sufficient gas velocity to remarkably improve the particle size refinement, Anderson and Terpstra^[30] showed that the atomization gas pressure for either argon or nitrogen only needs to be almost 3 $\mathit{MPa},$ which can be easily supplied by the most commercial GA setup.

1.1.1.3. Ultrasonic atomization (UA) technique

As schematically shown in Figure 4, the main concept of this technique is the way to atomize the liquid metal utilizing an ultrasonic vibration at the gas/liquid interface.^[36] A thin film of liquid metal is completely spread over a vibrating surface ($\ge 20\mathrm{ }\mathrm{kHz}$ vibration frequency) to break apart into fine droplets.^[19] There have been two hypotheses reported for the mechanism of liquid disintegration during the process: (1) capillary wave, and (2) cavitation hypothesis. The first one states that the capillary waves at the gas/liquid interface generated by the ultrasonic excitation leads to pinch-off the droplets from the wave peaks.^[37] This theory is established on Taylor instability criterion in which there is a relationship between mean droplet size and capillary wavelength.^[38] On the other hand, in the cavitation theory, the ultrasound wave from a transducer passes through the liquid metal, and subsequently, the local pressure of the fluid during the rarefaction phase of the sound wave drops below the vapor pressure of the liquid that forms the vapor bubbles. These bubbles violently implode during the compression phase leading to hydraulic shocks. Those near to the surface of liquid rupture the gas/liquid interface causing atomization.^[37,39] The cavitation mechanism is usually dominant when both the vibration frequency (16 $\mathrm{kHz}$–2 $\mathrm{MHz}$) and energy intensity are high.^[39] As another theory called conjunction theory, Bouguslavski^[40] proposed that a synergy between the periodic hydraulic shocks and the finite amplitude capillary waves excite the liquid metal to form the droplets.

Figure 4. (a) Capillary waves causing the liquid stream atomization on the top surface of the transducer Horn, and (b) a layout of the UA setup (Reproduced with permission from^[36]).

Cavitation shock is a random phenomenon causing nonuniformity of droplet distribution due to an irregular random disintegration of the thin liquid film, while the ultrasonic vibration applies capillary waves more regular in nature resulting in better uniformity of droplet size distribution. An interplay between the nature of these two phenomena gives support to the fact that each of the presented hypotheses (capillary wave and cavitation) cannot be solely responsible for liquid droplet formation in the UA process. Due to the random nature of the cavitation phenomenon, its presence and influence on the atomization is usually monitored via semi-quantitatively energy balance method, where the measurement of the droplet ejection velocity and validation based on the melt decomposition are both taking into account.^[41] Regarding the regularity of the ultrasonic vibration, there has been an equation developed to interpret the effect of process parameters on the median particle diameter as follows: (3) $${\mathrm{d}}_{50}=0.34{\left(\frac{8\mathrm{\pi }\mathrm{\sigma }}{\mathrm{\rho }{\mathrm{f}}^2}\right)}^{0.33}$$(3) where $f,$ $\sigma ,$ and $\rho$ are excitation frequency ($1/s$), surface tension ($N/m$), and density of the liquid ($kg/m^3$). Equation (3)(3) $${\mathrm{d}}_{50}=0.34{\left(\frac{8\mathrm{\pi }\mathrm{\sigma }}{\mathrm{\rho }{\mathrm{f}}^2}\right)}^{0.33}$$(3) in its general form is reported by Rajan and Pandit,^[39] where the dimensionless numbers in Equation (4)(4) $$d_{50}={\left(\frac{\pi \sigma }{\rho f^2}\right)}^{0.33}\left[1+A{\left(N_{We}\right)}^{0.22}{\left(N_{Oh}\right)}^{0.166}{\left(N_{In}\right)}^{-0.0277}\right]$$(4) are incorporated in physicochemical properties of the atomizing liquid as (4) $$d_{50}={\left(\frac{\pi \sigma }{\rho f^2}\right)}^{0.33}\left[1+A{\left(N_{We}\right)}^{0.22}{\left(N_{Oh}\right)}^{0.166}{\left(N_{In}\right)}^{-0.0277}\right]$$(4) in which $N_{We},$ $N_{Oh},$ and $N_{In}$ are modified Weber number, modified Ohnesorge number, and intensity number, respectively. These dimensionless numbers can be measured as (4a) $$N_{We}=\frac{fQ\rho }{\sigma },N_{Oh}=\frac{\mu }{fA{m}^2\rho },N_{In}=\frac{f^{2}A{m}^4}{CQ}$$(4a) here $Q,$ $\mu ,$ $A,$ and $m$ are volumetric flow rate ($m^3/s$), liquid viscosity ($Pa.s$), surface area of the droplet ($m^2$), and mass of the droplet ($kg$), respectively, $C$ stands for the velocity of sound in liquid medium ($m/s$). It is important to mention that to break the liquid stream into separate droplets using an ultrasonic vibration, the setup amplitude should be above a threshold value calculated by Equation (5)(5) $$A_{\mathit{crit}}=\left(\frac{2\eta }{\rho }\right){\left(\frac{\rho }{\pi \sigma f}\right)}^{0.333}$$(5) . It has been reported that the droplet size distribution becomes more compact as the liquid viscosity increases, i.e., better size uniformity.^[41] (5) $$A_{\mathit{crit}}=\left(\frac{2\eta }{\rho }\right){\left(\frac{\rho }{\pi \sigma f}\right)}^{0.333}$$(5)

In addition, there have also been other equations presented, based on the Rayleigh,^[42] Walzel,^[43] and Davies^[44] models, to predict the mean particle diameter during the UA process. The most important advantage of the UA technique compared to others is the necessity of lower liquid discharge pressure to disintegrate the liquid metal film compared to gas/water jet atomizing the liquid metal stream.^[41] Particles by the UA technique also show smaller size, less impurities, and better uniform size distribution for electronic applications.^[36]

1.1.1.4. Centrifugal atomization (CA) technique

The centrifugal atomization (CA) technique, Figure 5, refers to the atomization of a liquid metal stream by high-speed spinning the melt – where the centrifugal force drives the melt away from the axis of a rotating disk (rotating disk atomization) or a cup (rotating cup atomization). Subsequently, a continuous liquid film flows over a disk/cup surface until reaching the rim, then after leaves the disk/cup and breaks into droplets.^[45,46] The CA technique produces a narrow size range of powder, operates continuously, and uses very little energy compared to the GA and WA processes.^[46] As its disadvantage, the equipment size is determined by the distance necessitated for the melt droplets to fly before the solidification, which makes the apparatus very large and the capital costs fairly high.^[46] Rotating electrode process (REP) is also categorized as another commercial CA method in which the spinning element is a consumable electrode.^[25] A rod-shape solid-state electrode is rotated at almost 15,000 $\mathrm{rpm},$ while an arc fuses the consumable electrode, simultaneously. Centrifugal ejection of the liquid metal in the form of small droplets and the solidification process take place in an inert-gas-filled chamber.^[25] The fusion process of the consumable electrode can be conducted via either electric arc applied from a Tungsten tip or Helium plasma arc, where the second one is also called as Plasma REP.^[25]

Figure 5. (a) Schematic layout of the rotating disk atomization of liquid stream, and (b) droplet formation during the CA process (Reproduced with permission from^[47]).

The liquid metal flow regime on a rotating disk/cup depends on three factors of (a) flow rate, (b) the speed of rotation, and (c) the distance from the axis that determines the radius velocity.^[45] As the flow rate or speed of rotation increases, the instability and waviness of the liquid thin film will increase at both vicinity of the rotation axis of the disk/cup and far away from the axis. Also, an increase in the distance from the rotation axis leads to intensify the film waviness and transform a smooth waveless flow to a laminar-wave regime, then after to turbulent regime.^[45] Finally, the liquid metal ligaments detach from the disk and turn into small droplets. The CA process has attracted attention to fabricate a broad range of metal powder, e.g., Sn, Pb, Al, Fe, Mg, Ti, Co, Zn, and their multicomponent alloys.^[48] As previously mentioned, this technique is mostly used when the finished product needs to be within a narrow size distribution range and is desired in large volumes with no specification changes.^[46] The REP process is particularly applied to produce Ti alloys powder, since the molten Ti is corrosive intrinsically. Since this method is implemented in solid-state condition, the addition of contaminations from external sources is prevented, thus the metal powder with exact standards of cleanliness is achievable.^[21,25] In one of the most recent efforts to develop an equation presenting the median particle size from the CA technique, Shemyakina et al.^[49] proposed an empirical Equation (6)(6) $${\mathrm{d}}_{50}=\frac{4.27\times {10}^6}{{\left(\frac{2\mathrm{\pi }\mathrm{\eta }}{60}\right)}^{0.95}{\mathrm{d}}^{0.61}}{\left(\frac{\mathrm{\theta }}{\mathrm{\rho }}\right)}^{0.42}\left(\frac{{\mathrm{Q}}^{0.12}}{{10}^6}\right)$$(6) , presented as (6) $${\mathrm{d}}_{50}=\frac{4.27\times {10}^6}{{\left(\frac{2\mathrm{\pi }\mathrm{\eta }}{60}\right)}^{0.95}{\mathrm{d}}^{0.61}}{\left(\frac{\mathrm{\theta }}{\mathrm{\rho }}\right)}^{0.42}\left(\frac{{\mathrm{Q}}^{0.12}}{{10}^6}\right)$$(6) where $\eta ,$ $d,$ and $\theta$ are designated as the disk/cup rotation rate (rpm), the disk diameter (m), and the surface tension ($N/m$), respectively. Also, $\rho$ and $Q$ stand for the liquid density ($kg/m^3$) and the liquid flow rate ($m^3/s$).

1.1.2. Alloy constituents

In recent years, various types of steel powder have been successfully utilized to fabricate AM products. For instance, the powder of maraging-type precipitation-hardened martensitic stainless steels (also called PH steels, e.g., 17-4 PH/1.4542 and 15-5 PH/1.4545),^[50,51] carbon-free maraging steels (particularly 18Ni-300/1.2709),^[52] austenitic stainless steels (316 L/1.4404 and 304 L/1.4307),^[53,54] C-bearing tool steels (H13),^[55] transformation/twinning induced plasticity (TRIP/TWIP) steels (X30Mn22),^[56] martensitic stainless steels (410 L/1.4003 and X30CrMoN15-1),^[57,58] duplex stainless steels (SAF2507),^[59] oxide-dispersed strengthened (ODS) steels (PM2000),^[60] Fe-Si and Fe-Co electromagnetic steels,^[61,62] and Fe-Ni invar steels^[63] have been broadly used in the AM processes. Regarding the applications of the AM steels, the type and content of the alloying elements are varied over a wide range, which directly/indirectly affect the thermophysical properties of the feedstock as well as the final products.

In the austenitic stainless steels, Cr is added at high percentage (up to 17–18 $wt.\mathrm{\%}$) to improve the corrosion resistance of the alloys, while Ni addition increases the tendency of the microstructure being austenitic.^[64] In contrast, there is no report on the AM 200-series austenitic stainless steels in which Mn is added to stabilize the austenite phase. Segregation of Cr and Mo as the ferrite stabilizer elements into the interdendritic regions of the microstructure leads to formation of a thin layer of ferrite phase (up to 9 $\mathit{vol}.\mathrm{\%}$), where the rest of the microstructure usually shows the austenite phase.^[65,66] In the AM 304 L steels, lower Ni content compared to the AM 316 L stainless steels is intentionally considered to reduce the price, as well as the austenite stability.^[67] Wang et al.^[54] studied the influence of chemical composition on the austenite stability in direct energy deposited (DED) 304 L, where it was reported that a decrease in Ni content, and consequently, reducing the austenite stability leads to the promotion of strain hardening induced by TRIP effect. As a result, both ultimate tensile strength (UTS) and elongation to failure sharply enhance. The effect of alloy composition, cooling rate, and thermal cycles on the solidification mode and the residual δ-ferrite in rapidly solidified austenitic stainless steels are always of interest.^[68,69] In this regard, the sequences of phases formation through the equilibrium Fe-Ni-Cr ternary phase diagram and the ratio between Chromium equivalent and Nickel equivalent (i.e., ${Cr}_{eq}/{Ni}_{eq}$) are key parameters which determine the microstructure obtained by the phase changes. Regarding the sequences of phases formation during the equilibrium solidification, Di Shino et al.^[68] used the equations developed by Hammer and Svensson to schematically show the solidification mode of the austenitic stainless steels. Based on that, four various modes of solidification can be written as,

Mode A: $L\to L+\delta \to \delta \to \delta +\gamma ;$ when ${Cr}_{eq}/{Ni}_{eq}>2.01$

Mode B: $L\to L+\delta \to L+\delta +\gamma \to \delta +\gamma \to \gamma ;$ when $1.51<{Cr}_{eq}/{Ni}_{eq}<2.00$

Mode C: $L\to L+\gamma \to L+\gamma +\delta \to \gamma +\delta \to \gamma ;$ when $1.38<{Cr}_{eq}/{Ni}_{eq}<1.50$

Mode D: $L\to L+\gamma \to \gamma ;$ when ${Cr}_{eq}/{Ni}_{eq}<1.37$

where L, δ, and γ represent the liquid, the ferrite, and the austenite phases, respectively. In these modes: (7) $${Cr}_{eq}=\left(\mathrm{\%}Cr\right)+1.37\left(\mathrm{\%}Mo\right)+1.5\left(\mathrm{\%}Si\right)+2.0\left(\mathrm{\%}Nb\right)+3.0\left(\mathrm{\%}Ti\right)$$(7) (8) $${Ni}_{eq}=\left(\mathrm{\%}Ni\right)+22.0\left(\mathrm{\%}C\right)+14.2\left(\mathrm{\%}N\right)+0.31\left(\mathrm{\%}Mn\right)+\left(\mathrm{\%}Cu\right)$$(8)

While the equations above are only governed under the equilibrium solidification, the metal powders produced using atomization techniques, as well as the AM products introduce different phases in the microstructure of the austenitic stainless steels, as they both undergo nonequilibrium rapid solidification. Therefore, the role of cooling rate during solidification and solid-state phase transformation should not be neglected.^[70] Tiller et al.^[71] defined (δ) ferrite-to-austenite transformation “massive” in nature, which occurs by the migration of a disordered interface. The presence of trans-interphase interface diffusion and mitigation of long-range diffusion in a massive transformation produce the massive product which inherits the parent phase composition. Based on that, the diffusion paths of Cr and Ni in austenite are too long for a diffusion-controlled transformation during the rapid solidification. At high cooling rates during the AM and the powder atomization processes, the solidus line can remarkably deviate from its equilibrium line, resulting in changes in solid-state phase transformations, and subsequently, phases differences in the as-produced microstructure. In this regard, the atomized powders with smaller sizes are more prone to these changes as they solidify at higher cooling rates during the atomization compared with coarser particles. The role of colling rate on the final microstructure of the AM component was also confirmed by Hung et al.^[72] where they reported almost 10–15% retained δ-ferrite in the microstructure of LPBF austenitic stainless steel (22Cr-14Ni-2.5Mo), while ${Cr}_{eq}/{Ni}_{eq}$ was 1.66. Consequently, the as-produced product contained a duplex microstructure containing austenite and retained ferrite. Tarasov et al.^[73] also reported that the heat input used in electron beam additive manufacturing has a threshold at which the δ-ferrite reaches to a maximum content in AM 304 austenitic stainless steels.

In another study, Heiden et al.^[74] showed that 316 L stainless steel powder interacting with the energy source such as the laser beam, yet not consolidated into an AM product can experience dynamic heating/reheating interactions, causing variable morphological, microstructural, and magnetic properties. Reused powder feedstock was reported to have a combination of fully ferritic particles and fully austenitic particles both single crystalline in nature; however, virgin powder feedstock solely contained polycrystalline austenitic particles, as shown in Figure 6. Magnetic susceptibility was increased in reused powder feedstock, as they showed the ferrite phase in some particles.

Figure 6. (a) the Fe-Ni-Cr Isopleth at 19% Cr showing various solidification paths,^[68] (b) solidification modes in austenitic stainless steels,^[68] (c and d) EBSD phase maps and corresponding inverse pole figure maps of virgin and reused 316 L stainless steel powders (Reproduced with permission from^[74]).

Very few studies have been conducted on AM martensitic stainless steels (400 series), which contain 12–14 $wt.\mathrm{\%}$ Cr, but very low concentrations of C and Ni austenite stabilizers.^[57,58] Therefore, they do not show the austenite phase at ambient temperature. Boes et al.^[58] showed that N and C are responsible for ${\mathrm{M}}_2\mathrm{N}$ and ${\mathrm{M}}_{23}{\mathrm{C}}_6$ precipitate formation in the heat-treated AM nitrogen-alloyed martensitic stainless steels. In the duplex (austenitic/martensitic) stainless steels, the microstructure mostly contains a combination of $\delta -$ ferrite and austenite, which necessitates high content of Cr ferrite-stabilizer (24–26 $wt.\mathrm{\%}$), while Ni concentration is the same as the level in austenitic stainless steels.^[64] In superduplex grades (e.g., steel 2507/1.4410), nitrogen element is also added to increase the pitting and crevice corrosion.^[75] The empirical pitting resistance equivalent (PRE) for the aforementioned steel grades can be given by Equation (9)(9) $$\mathit{PRE}=\mathrm{\%}Cr+3.3\times \mathrm{\%}Mo+16.0\times \mathrm{\%}N$$(9) : (9) $$\mathit{PRE}=\mathrm{\%}Cr+3.3\times \mathrm{\%}Mo+16.0\times \mathrm{\%}N$$(9)

In TRIP/TWIP steels, the as-produced microstructure mainly contains austenite, ${\alpha }^{\prime }$-, and $\epsilon$-martensite. Kies et al.^[76] produced a DED high-Mn steel (X30Mn22) at various Al contents. As Al percentage increased in the feedstock, the stacking fault energy of the as-built material came up, resulted in TRIP to TWIP transition in deformation mechanism, as well as a suppression in austenite to martensite transformation in the as-produced microstructure. In another study, Niendorf et al.^[77] investigated a commercial TWIP steel X-IP 1000 (X60Mn22) produced by laser powder bed fusion (LPBF) technique. The powder feedstock was chemically modified by Ag addition, where Ag meliorates the electrochemical dissolution rates proposed for biodegradable implants applications.

Precipitation-hardened (PH) stainless steels consist of Cr (16–18 $wt.\mathrm{\%}$), Ni, and Cu (3–5 $wt.\mathrm{\%}$ each) as the main alloying elements. They possess an excellent combination of corrosion resistance and mechanical strength in which Cr in the form of passive layer of ${\mathrm{Cr}}_2{\mathrm{O}}_3$ is responsible for the corrosion resistance.^[64] Ni as an austenite stabilizer enhances the corrosion resistance as well; however, Cu has a dominant role in mechanical strengthening through the precipitation hardening mechanism in aged AM products. 15-5 PH steels contain lower Cr concentration and minor addition of Mo compared to the 17-4 PH steels. Among the PH steel powder, some grades like 17-7 PH (1.4568) and commercial EOS CX stainless steel are alloyed by Al instead of Cu.^[78] In these cases, the precipitation hardening is taking place through the formation of ${\mathrm{Ni}}_3\mathrm{A}\mathrm{l}$ precipitates instead of Cu-rich ones.^[79]

The chemical composition of the maraging steel powder includes various alloying elements such as Ni, Ti, Al, and Mo, mainly due to increase the ability of precipitation hardening of the AM product by the formation of ${\mathrm{Ni}}_3\mathrm{X}\mathrm{ }(\mathrm{X}=\mathrm{Ti},\mathrm{ }\mathrm{Al},\mathrm{ }\mathrm{Mo})$ and ${\mathrm{Fe}}_7{\mathrm{Mo}}_6$ phases during the post-processing aging treatment.^[80] As Dehgahi et al.^[7] declared, the high concentration of Ni (18–19 $wt.\mathrm{\%}$) helps the microstructure keep the austenite phase at room temperature, which eases the formation of martensite phase as the major phase in the as-built product. However, there have been few regions in the microstructure maintained austenitic. An increase in Ti content in 18Ni-300 maraging steels leads to no difference in martensite/austenite volume ratio in the powder; however, a notable variation of the phases volume fraction in the LPBF products, where the one has higher Ti content shows higher percentage of the retained austenite.^[7] Also, compared with the low-Ti 18Ni-300 samples, the ones produced by laser fusion of high-Ti powder show ultra-fine CoNi metastable phases in the as-built microstructure.^[7] Jägle et al.^[81] indicated that the laser metal deposited (LMD) 18Ni-300 maraging steel shows Ti and Mo as austenite stabilizer once they have been partitioned in interdendritic regions.

In the carbon-bearing tool steels, the coexistence of Cr, Mo, and V leads to form stable MC-type and ${\mathrm{M}}_7{\mathrm{C}}_3$ carbides during the aging treatment, as these elements show high affinity to react with carbon.^[82] Compared with other steels series previously discussed, high level of C content in these steels brings higher hardness values in the as-LPBF-, and as-DED-produced samples (51–59 $\mathit{HRC}$).^[64] Same as the microstructure of the AM maraging steels, these alloys’ microstructure also mainly contains the martensite phase, as well as a low volume fraction of retained austenite formed in the interdendritic zones, where the alloying elements of Si, V, Mo, and Cr have been segregated.^[55]

The oxide dispersion strengthened (ODS) steels are among the stainless steels that contain no austenite stabilizer. Cr content in these steels are varied between 12 and 23 $wt.\mathrm{\%},$ which allows to have the martensitic, martensitic/ferritic, or fully ferritic microstructure.^[64] Besides, high value of Cr (and Al in some grades) increases the corrosion resistance, while the superior creep resistance at high temperatures is attributed to dispersion of fine oxide particles such as ${\mathrm{Y}}_2{\mathrm{Ti}}_2{\mathrm{O}}_7,$ ${\mathrm{Y}}_4{\mathrm{Al}}_2{\mathrm{O}}_9,$ and ${\mathrm{Y}}_2{\mathrm{O}}_3$ in the microstructure.^[60,83] The ODS steel powder is usually decorated with the oxide particles through the mechanical alloying process, (i.e., high-energy ball milling), and subsequently used as a feedstock in the laser melting processes.

As the last categories of Fe-rich alloys in this section, the iron-silicon (Fe-Si) alloys well-known as highly electromagnetized steels usually contain 6.5–7.0 $wt.\mathrm{\%}$ Si (and sometimes B, Nb, Cu addition), where the silicon addition brings near-zero magnetostriction, and near-zero magnetocrystalline anisotropy.^[62] The Fe-Si steels are difficult to process through traditional subtractive or forming methods, as the high content of Si leads to material embrittlement. Hence, to less expose the material to the embrittlement zone during the solidification, the rapid solidification such as the one applied during the LPBF technique is recommended.^[62] Cu addition causes clustering and forming Cu-rich phases that play a role as potent nucleation site for $\alpha -$ (FeSi) nanocrystals formation.^[84] Nb, on the other side, has a growth restricting effect on these nanocrystals. A synergy between the effect of Nb and Cu addition results in a fine as-built microstructure with much lower coercivity than the Fe-Si alloys.^[85] Apart from the Fe-Si alloys, the Fe-Co-based soft magnetic alloys powder is usually mechanically pre-alloyed by other alloying elements, e.g., Ni, Nb, Si, B, Cu, and Mo before its application in the laser AM.^[61] Borkar et al.^[84] reported that the magnetic properties of the AM Fe-Co-based alloys are dependent on the as-built microstructure relied on the $\mathrm{Si}/\mathrm{B}$ ratio. B and Si are known as elements enhancing the glass-forming ability and electrical resistivity, respectively.^[86] Finally, Invar and its variants are Fe-Ni alloys in which Fe and Ni concentrations are 63–64 and 35–36 $wt.\mathrm{\%},$ respectively, while others like Mn and C can be added as trace elements.^[63] The main property of these alloys is their low coefficient of thermal expansion (CTE) for temperatures up to 200 °C, also named as Invar effect.^[63] This effect is very susceptible to variation in percentage on ferromagnetic elements, e.g., Fe, Ni, and Co, where any deviation from the 64Fe-36Ni composition drastically reduces the Invar effect.^[87]

1.1.3. Powder shape and geometry

In AM, powder morphology is responsible for determining how well the particles lay or pack together to consequently realize the minimum part layer thickness and density.^[13] The powder morphology plays a significant role on flowability, packing density, and the AM component properties.^[88] According to the ASTM B243 standard,^[89] the metal powder takes various morphologies mainly based on production process, process parameters during production and packing, as listed in Table 1.

Table 1. Various types of powder morphology in ASTM B243.^[89]

Morphology	Definition
Acicular	Needle-shaped
Flake	Flat or scale-like, whose thickness is small compared with other dimensions
Granular	Almost equidimensional, nonspherical shapes
Irregular	Lacking symmetry
Needles	Elongated and rod-like
Nodular	Irregular, having knotted, rounded, or similar shaped
Platelet	Composed of flat particles having considerable thickness
Plates	Flat particles of metal powder having considerable thickness
Spherical	Globular-shaped

As noted, there are different methods to produce metal powders, where some methods can provide a more regular powder morphology than others. For instance, it has been reported that employing gas atomization (GA) allows more spherical particles than water atomization (WA), as revealed in Figure 7. Since the powder solidification rate in the GA technique is lower than the WA technique, more spherical particles can be formed bringing better flowability and apparent density.^[14] However, the product is highly prone to another defect called satelliting effect, i.e., small particles attachment on the surface of large metal powder.^[90] Herein, a decrease in satelliting effect brings more sphericity, and subsequently, better flowability due to lower interparticle friction. The flowability is also influenced by the size of particles solidified during the atomization process, where the larger particles ($>50\text{ μm}$) show good flowability, while those between 5 and 50 $\mathrm{\mu }\mathrm{m}$ in diameter present lower flowability as the interparticle van der Waals’ forces increase.^[91] In terms of powder fabrication process, Sames et al.^[92] reported that the best sphericity with less surface defects are achieved in the powder produced by the plasma atomization (PA) and plasma rotating electrode process (PREP). On the other side, the powder produced by the WA, GA, and rotating-disk centrifugal atomization (CA) techniques is susceptible to irregularity, satelliting, and nodularity, respectively.^[92]

Figure 7. SEM micrographs and optical Appearance (in the insets) of 316 L stainless steel powder produced by (a) GA-, and (b) WA-techniques (Reproduced with permission from^[93]).

Besides fabrication process parameters, thermophysical properties of the liquid metal, chemical composition, powder purity, and post-processing alloying have important roles determining the morphology of the metal powder. Regarding the satelliting effect, an inevitable rendezvous of fine and coarser particles takes place during their flight downstream in the atomization chamber.^[94] Since the coarser powder shows a delay to fully solidify rather than fine particles, their surfaces that are still semi solid or mushy act as suitable substrates on which the fine solidified particles impinge/weld. In this regard, the powder with slower solidification rate in the GA process is more exposed to this phenomenon. Another mechanism called alternative satellite attachment considers spray chamber design in which an entrainment of “clouds” of fine solidified powder is conducted into the atomization spray exterior “cone.” As a result, they attach themselves to form larger particles.^[95]

In terms of thermophysical properties of the liquid and initial chemical composition, thermal conductivity of alloying elements, oxidation affinity, electrical conductivity that determines surface tension forces between fine solidifying particles, and the presence of alloying elements that can result in a variation in freezing range during the peritectic reaction (in steels) and the other transformations (in other Fe-based alloys) should be considered. For instance, Dehgahi et al.^[7] presented that an increase in Ti content in 18Ni-300 maraging steels powder leads to broaden the freezing range. This is more serious when the particle size increases, because wider solidification temperature range can intensify satelliting effect. Oxidation degree of the metal powder is also remarkable, and can directly affect the powder morphology, since an irregular oxide layer can be formed on the powder surface.^[96] In this regard, Hyrha et al.^[97] reported that the morphology and surface quality of the metal powder are highly affected by the surface composition, where the surfaces of gas atomized 316 L stainless steel recycled powder are covered by Cr-bearing oxides, as shown in Figure 8. Tunberg and Nyborg^[98] also reported that the surface oxidation in water atomized 304 L stainless steel powder is remarkably influenced by the cooling rate each particle experiences during the solidification. An increase in particle size accompanied by slowing the cooling rate leads to an increase in average surface oxide thickness, while the nature of oxide changes from silicon-rich to iron-, and chromium-rich oxides. Finally, various alloying elements with different manners in partitioning can vary the solidification rate, which cause solidification defects such as lack of interdendritic feeding, detachment, porosity formation, and voids coalescence, as marked in Figure 8.^[99]

Figure 8. Gas atomized 316 L stainless steel (a) virgin powder, (b) recycled powder after EBM cycle,^[97] and (c,d) small satellites and surface defects in 304 L stainless steel powder (Reproduced with permission from^[99]).

It is worth noting that the powder morphology can be considered from another view, i.e., the microstructural features of the metal powder. Regarding the process parameters, thermophysical properties and chemical composition of the liquid metal, the microstructure of the solidified powder can be cellular, columnar-dendritic, and equiaxed.^{[7,74,100–104]} Opatová et al.^[101] reported a dendritic microstructure in the virgin and reused 18Ni300 maraging steel powder, where the interdendritic regions are suitable sites for Ti and Mo segregation during the solidification. The coarser powder experienced slower cooling during the atomization mostly elucidate more severe segregation.^[101] Reused metal powder often shows wormy shape cavities inside that can be assigned to lack of interdendritic feeding during the solidification. The virgin powder, on the other side, mostly shows internal spherical cavities caused by gas entrapment during the atomization process.^[101] The initial chemical composition of the powder also acts as important role on determining the microstructure morphology. In this case, Dehgahi et al.^[7] analyzed the morphology of the same-size 18Ni300 steel powder at two Ti contents. As the titanium concentration increases, the area fraction of equiaxed grains enhances, which is ascribed to an increase in constitutional undercooling. However, the rest of the cross-section area shows columnar-dendritic morphology in both cases.^[7] The equiaxed morphology has also been observed in the gas-atomized powder of 316 L, H13, and P20 steels.^[100]

1.1.4. Size distribution of powder

Among the powder characteristics that affect the physicomechanical behavior of the bulk powder and the subsequent as-built AM product, particle size distribution (PSD) is a key factor that directly determines powder rheology, packing density, compactability, sinterability. The powder PSD has also been linked to the mechanical properties of components.^[105] Reports indicate that PSD has can affect the layers density in laser powder bed fusion (LPBF), where a wide PSD inclined toward fine powder, also called multi-modal PSD, results in higher layer density.^[106] In terms of the bulk powder physical property such as thermal conductivity, the PSD also plays an undeniable role. It has been reported that the thermal conductivity of the solid material is almost two times greater than the well-packed bulk powder in which this difference enhances as the PSD mostly biased toward either fine or coarse particles.^[107] Karapatis et al.^[106] derived several criteria for tailoring the PSD to the powder bulk density and concluded that to achieve a high value of powder bulk density, a broad random distribution of particles is necessary. To quantify the PSD, sieve analysis, microscopy, and laser diffraction (LD) are commonly used. The LD is the most reliable technique employed based on the Fraunhofer diffraction theory in which the particle size is directly proportional to intensity of the light scattered by particle.^[108] However, quantitative image processing technique on the electron micrographs are hired in many studies.^[109,110] Although a 2D approach to measure the particle size is not the most accurate technique, it is acceptable in the AM industry due to the ability to both quantitatively and qualitatively assess the powder particles in terms of shape, size, and surface roughness.^[111] The sieve analysis is among the oldest techniques to measure the particle size, albeit it is widely used because of the cost-effective approach and simplicity. However, this technique is not well applicable in the AM processes, since the collected particles are usually coarser than the ones needed in the AM. A conventional setup contains a stack of sieves with top-to-bottom arrangement in mesh size. As the sieves stack is shook or mechanically vibrated, each mesh retains the powder comprised of particles coarser than the mesh size. To report the PSD, the mass of powder retained on each mesh is calculated and addressed to the size range or bin size.^[88,111] Table 2 gives a summary of powder size measurement techniques used in the AM.

Table 2. Conventional techniques for powder size measurement relevant to AM.^[88,111]

Technique	Materials	Resolution	Particle size range	Advantages	Disadvantages
Sieve analysis	Solids	Bin size: Dependent on size separation of sieves	20 $\mathrm{\mu }\mathrm{m}$–125 $\mathrm{mm}$	Applicable to broad size range, Minimal sample preparation, Low cost	Not able to analyze the shape, Longer measurement time, Prone to blinding, Broad uncertainties in particles with large aspect ratio
Microscopy	Depends on microscope	$d=\frac{0.61\lambda }{NA}$ d: resolution distance λ: light source wavelength NA: numerical aperture	Determined by system resolution	Allows qualitative and quantitative observation of particle shape, Great flexibility for particle size and shape analyses	Costly in fine particles, Requirement of SEM and TEM sample preparation, More efforts than the optical microscopy
LD	Solids	Dependent on instrument design and data analysis algorithm	0.04 $\mathrm{\mu }\mathrm{m}$–8000 $\mathrm{\mu }\mathrm{m}$	Short analysis time, Highly repeatable results, No need for skilled labor	Emergence of error in non-spherical particles, Dependency of measurements on instrument design, Difficulty in detection of agglomerates

The particle size distribution is usually presented by cumulative mass or volume distribution percentage vs. particle diameter. The output is usually reported as $D_i\mathrm{ }\left(\text{μm}\right),\mathrm{ }i=10,\mathrm{ }50,\mathrm{ }90,$ which means the particle diameter at i% cumulative mass or volume distribution percentage.^[109] There has usually been an interest to present a Gaussian curve within a narrow range of particle size, i.e., containing both fine and coarse particles, while the median and average values are close to each other. The Gaussian distribution provides an optimum packing density, i.e., the level of powder compaction together with no applied compressive load.^[105] Consequently, the maximum relative density of the AM product can be guaranteed if other processing parameters have been properly selected. Any deviation of the normally distributed data to smaller or larger quantities brings less packing density, where finer particles are prone to clumping or agglomeration effect, while coarser ones lead to higher vacant fraction due to a remarkable decrease in surface area contact.^[90] In a random Gaussian distribution, the vacant areas between coarse powder are well filled by the fine particles. Most of commercial steel powder suppliers report 15–45 $\mathrm{\mu }\mathrm{m},$ and 40–105 $\mathrm{\mu }\mathrm{m}$ particle size distribution as optimum windows to have a proper packing density in LPBF and electron beam manufacturing (EBM), respectively.^[112–114] As discussed earlier, reused powder is usually larger in size than the virgin powder, shifting the size distribution curve to the right side and causing less packing density of the feedstock.^[74] A normal Gaussian size distribution of the 17-4 PH stainless steel powder is demonstrated in Figure 9. The powder production technique is another consideration determining the size and distribution of particles. Some production techniques such as the rotating electrode process (REP) and rotating disk atomization present a narrower window of particle size distribution, while the water and gas atomization produce particles through a broad range of size. Using the vacuum atomization (VA) shows a limitation in providing fine metal powder sized below a specific particle diameter.^[90]

Figure 9. (a) Particle size distribution of gas atomized AISI H13 tool steel powder, and (b) the SEM-SE micrograph (Reproduced with permission from^[109]).

1.1.5. Density and flowability of powder

The powder flowability and packing density are categorized as the powder characteristics directly affected by the PSD and other powder properties, e.g., morphology, impurities, and moisture. As revealed in Figure 10, the bulk powder packing density, i.e., the level of powder compaction together with no applied compressive load, is identified using various parameters such as apparent density (DIN ISO 3923), tap density (DIN ISO 3953), and Hausner Ratio (HR) (ASTM D7481).^[115–117] The apparent density provides the mass per unit volume of loose packed powder, where the low-density value is designated to fine particles, while the high level is mostly used to identify the coarse particles.^[116] The internal pores of a bulk powder but not the interstitial volume between powder packed together are considered to report the apparent density.^[116] Several methods are employed to determine the apparent density, where the Scott volumeter (ASTM B329) and the Arnold meter (ASTM B703) are more favorable.^[118,119] As illustrated in Figure 11, in the Scott volumeter technique, metal powder is poured into a series of funnels, and then followed by traveling through a baffle series into the last funnel placed under the baffle box. In the end, the collected powder is poured into a collection cup with a specific volume upon exiting the final funnel. To determine the mass, and subsequently, the apparent density, the collection cup is placed on a balance. An Arnold meter setup includes a steel block with a powder delivery sleeve and cavity in the middle in which the sleeve is located on either side of the die cavity with a specific volume. Powder is poured into the sliding sleeve that allows them to fall through and fill the die cavity. The sleeve sliding back across the cavity results in flattening the amount of powder flush with the steel block. Subsequently, the mass and density of the powder pile in the die cavity are measured.

Figure 10. Bottom-up diagram presenting the effect of powder properties on the properties of the bulk powder, in-process performance, and the as-built product (Reproduced with permission from^[114]).

Figure 11. Illustration of two techniques that measure the apparent density: (a) a Scott volumeter, and (b) an Arnold meter (Reproduced with permission from^[13]).

Tap density of the bulk powder is the maximum density of the bulk powder when it is vibrated or tapped under a specific condition such as a defined period of time.^[117] This type of density presents the random dense packing of the powder. The common process to measure the tap density is pouring a specified mass of powder into a graduated cylinder, where the cylinder is loaded into a tapping apparatus. This apparatus applies tapping (frequency of 100–300 taps per minute) against a firm base. Once no further tap-induced decrease in the bulk powder volume takes place, the achieved volume and known mass of the specimen are used to measure the tap density.^[13] Another method to measure the packing density is the Hausner Ratio presented as the ratio of the tap density over the apparent density.^[115] However, the HR is reported to be too far away from the actual condition in the AM form, because it is highly related to adhesiveness, PSD, and powder morphology.^[88,114] In regard to the Hausner Ratio value, $\mathrm{HR}=1$ means an incompressible bulk solid. Karapatis et al.^[106] declared that a wide PSD is necessary for a high powder bulk density.

Powder flowability is an “umbrella term” demonstrating the powder complex behavior, once it is mobilized or subjected to stresses, and not a comprehensive property of bulk powder.^[114] As shown in Figure 12, there have been multiple characteristics confirming that the flowability is not an inherent powder property.^[114] Despite the independency of flow properties on the measurement equipment, the term “flowability” always deals with the way of testing and measurement techniques. The most common method to spread the powder over a precise layer height is utilizing a metallic blade, roller, or a rake, followed by powder flowability testing methods such as the Hall flowmeter funnel (ASTM B213), the Carney funnel (ASTM B964), and Hausner Ratio (HR) (ASTM D7481).^{[115,120,121]} In both Hall flowmeter funnel and Carney funnel, metal powder is timed as it flows through the orifice with different diameter. The Hall flowmeter funnel is narrower with an opening diameter of 2.5 $\mathrm{mm},$ while the size in the Carney funnel is almost 5.08 mm.^[122,123] Although Spierings et al.^[124] reported the Hall flow test as an acceptable technique used for the AM processes, they still argue that this is not the best suited technique due to its limitation only to superior flowing powder, not more cohesive ones. As a result, the Hall flow test is widely used for the EBM powder characterization rather than the LPBF feedstock, as the powder in the LPBF is finer and more cohesive.^[114] In the HR technique, since both the apparent and tap densities are dependent on the interparticle friction forces, it directly affects the flowability. Table 3 shows different degrees of flowability regarding the HR values.^[88] Once the HR comes below 1.25, the term “free-flowing” is designated for the powder. The angle of repose (AOR) is a promising technique for free flowing of slightly cohesive homogeneous powder.^[114] In this technique, the metal powder is allowed to free-fall under a controlled manner and deposit onto a smooth surface. The angle between the surface and the slope of the powder pile is defined as the angle of repose, α.^[88] This angle is directly related to the interparticle friction forces in which higher angle of repose shows greater powder cohesivity.

Figure 12. An interplay among the terms “flowability,” “flow properties,” and respective parameters (Reproduced with permission from^[114]).

Table 3. Degrees of powder flowability in terms of the Hausner ratio.^[88]

Flow description	Excellent	Good	Fair	Passable	Poor	Very poor	Non-flowable
HR	1.00–1.11	1.12–1.18	1.19–1.25	1.26–1.34	1.35–1.45	1.46–1.59	≥1.60

There have been some other techniques such as powder rheometer utilizing, the dynamic avalanche angle measurement, Round Robin testing, Gustavsson flowmeter funnel (ISO 13517), and density-based flow measurement, which are employed to quantify the powder flowability.^[114,125] As an agreement between most of the mentioned techniques, the following relations are summarized as:^[114]

• Narrowing the window of the PSD increases the powder flowability.

• The coarser powder shows better flowability.

• As the moisture content increases until saturation with liquid, the flowability reduces.

1.1.6. Gas entrapment

Gas entrapment during AM processes is mainly resulted from two sources: trapped gas in powder feedstock, and trapped gas during powder fusion under protective atmosphere. Powder rapidly solidified under inert gas environment usually shows gas content on the order of several atomic parts per million.^[126] The entrapment probability increases, as the particle size increases, where the coarse powder often contains gas bubbles. Since most techniques of powder production employ an inert gas environment such as argon and helium to refuse oxidation and promote convective cooling, gas entrapment during the process is almost inevitable. The solubility of most inert gases in the metal is vanishingly small, which create micropores and/or fine interdendritic porosity.^[126] In Table 4, the level of porosity formed during various atomization techniques of some 304 stainless steel powders is reported regarding the particles size and the technique used.

Table 4. Gas concentration and porosity level in rapidly solidified 304 SS metal powder.^[127]

Atomization technique	Atomization gas	Helium content $(\mathrm{at}.\mathrm{ppm})$	Argon content $\left(\mathrm{at}.\mathrm{ppm}\right)$	Nitrogen content $(\mathrm{at}.\mathrm{ppm})$	Oxygen content $(\mathrm{at}.\mathrm{ppm})$	Mean particle size $(\mathrm{\mu }\mathrm{m})$	Porosity
CA	He	8	<0.1	ND	600	80	Low
IGA	Ar	ND	2.5	250	700	70	High
IGA	He	1.5	<0.1	ND	1100	60	None
IGA	Ar	ND	4.6	1200	900	50	High
IGA	N	ND	<0.1	5800	850	50	None

CA: centrifugal atomization; IGA: inert gas atomization; *ND: Not determined.

During the atomization of liquid stream, the liquid droplets can be either broken up or solidified containing an internal trapped gas. According to Equation (10)(10) $$We=\frac{d{\vartheta }^2{D}_0}{\gamma }$$(10) , break-up of the liquid droplets at high relative gas velocity can take place over various regimes of the Weber number in which $12<We<50.$^[90,126] This scenario is similar to bubble blowing that bursts the bag, leading to a large number of fine droplets. Break-up condition usually results in droplets with relatively low viscosity and surface tension compared with the liquid metals. As the viscosity increases, the droplets break-up concurrently happens with solidification. On the other side, it is probable for highly viscous liquid droplets not to be broken up but folded as the solidification proceeds. The formed hollow sphere with gas inside has a thick wall, as illustrated in Figure 13. (10) $$We=\frac{d{\vartheta }^2{D}_0}{\gamma }$$(10) where $d,$ $\vartheta ,$ $D_0,$ and $\gamma$ stand for the flow field density, the relative viscosity, the initial particle diameter, and the surface tension, respectively.

Figure 13. Schematic Illustration of the bag break-up or folding mechanisms during the powder production (Adapted with permission from^[126]).

Another possible explanation for the gas entrapment is the difference between gas solubility in the solid and liquid, where the solubility level in the solid is much lower than the liquid during solidification. The insoluble gas may expand to form a large gas pore entrapped into the solidified particle; however, this mechanism is unlikely since it necessities proper solubility in the liquid on the order of measured gas contents.^[90,126] At in-service condition, equilibrium solubility of the inert gas is several times lower than experimental measurements even in liquid state. Besides, solidification initiates on one or more nucleation sites and does not follow a simple trend from the shell to core of the solidifying droplets. The last possible mechanism for gas entrapment in powder feedstock is gas atoms absorption on the surface of liquid droplet followed by thermally induced migration of vacancies along the temperature gradient. Sufficient solidification time to allow long-range diffusion is a key factor, which cannot be usually satisfied in the AM powder atomization techniques.^[126]

Gas entrapment during the metal powder fusion is almost inevitable as it depends on many factors such as powder moisture, inversion of Marangoni flow, feedstock oxidation, void pockets between powder particles, etc.^[128] Since the residual gas porosity is very hard to eliminate from the common AM builds, diminishing the atomization gas entrapment in the powder feedstock is the best way to mitigate this type of porosity.^[90] To this end, the use of an atomization technique with lower energy to persuade the “bag breakup and collapse” mechanism is required.^[90] Also, pores reduction via hot isostatic press (HIP) of the powder is another recommendation.^[128]

1.2. Processing parameters

Processing parameters as independent variables affecting thermophysical-mechanical-microstructural properties of the AM product, e.g., the microstructure morphology, microstructural defects, thermal expansion, fatigue, and tensile strength are required to be known and optimized. According to Ishikawa diagram shown in Figure 14, many processing parameters can control laser additive manufacturing (LAM), particularly the LPBF. To precisely develop processes, it is crucial to recognize the influence of each parameter; however, many studies have reported that the parameters in the LAM processes exhibit a high degree of interdependency and interacting effects on the quality of as-built products.^[55,129] Hence, obtaining precision in these processes is usually accompanied by difficulties. Among the parameters reported in Ishikawa diagram, the most significant ones can be classified as effective layer thickness, $t,$ laser power, $P,$ hatch spacing, $h,$ scanning speed, $v,$ scan strategy, point distance, $PD,$ exposure time, $\theta ,$ and environment of the printing chamber. According to Equation (11)(11) $$E_V=\frac{P}{(v\times h\times t)},\mathrm{ }v=\frac{PD}{\theta }$$(11) , the laser energy density, $E_V,$ is defined as^[130] (11) $$E_V=\frac{P}{(v\times h\times t)},\mathrm{ }v=\frac{PD}{\theta }$$(11) where $P,$ $v,$ $t,$ $h,$ and $E_V$ are in ($W$), ($mm/s$), ($mm$), ($mm$), and ($J/{mm}^3$), respectively.

Figure 14. Ishikawa diagram of processing parameters controlling the properties of LPBF product (Reproduced with permission from^[131]).

To melt a certain mass of powder particles, a level of energy density, E_m, needed is estimated by Equation (12)(12) $$E_m=C\rho (T_m-T_a)$$(12) .^[130] In this correlation, C is the specific heat capacity (J/Kg.K), ρ is the material density (Kg/mm³), E_m (J/mm³), T_m and T_a are the melting and ambient temperatures (K), respectively. A dimensionless ratio, e, describing the relationship between the energy density of the laser and the one required for melting a powder layer is shown via Equation (13)(13) $$e=\frac{E_V}{E_m}$$(13) : (12) $$E_m=C\rho (T_m-T_a)$$(12) (13) $$e=\frac{E_V}{E_m}$$(13)

1.2.1. Thickness of powder layer

In the LPBF, the layer thickness is assigned to the actual thickness of either powder particles spread on the build plate via a recoater or build layer after fusion and solidification of particles, respectively, i.e., effective layer thickness (ELT), and the nominal layer thickness (NLT), where the NLT is lower than the ELT. The ELT is reported to be influenced by many factors such as particle size, distribution, morphology, and purity, while the NLT is mostly affected by thermophysical properties of the printing material like thermal expansion, temperature gradient, density, capillary effect, and processing parameters coming up in the next section. The ELT is an important criterion determining the level of fusion in which the lack of fusion (LOF) defect is defined as the ratio between the melt pool depth and the ELT.^[132] As depicted in Figure 15(a), the minimum depth of the overlapped semi-circles or semi-ovals melt pools is shown as ${\mathrm{M}}_{\mathrm{d}}^0$ for a given melt pool. To guarantee to avoid the LOF defect between successive layers, the ELT must be smaller than ${\mathrm{M}}_{\mathrm{d}}^0.$ This quantity is directly affected by the hatch spacing dictating a maximum ELT. A reduction in the ELT and hatch spacing prolongs the beam exposure time linearly and is easier to control rather than the temperature profile through the melt pool. Therefore, it is recommended to select ELT/hatch spacing according to the appropriate energy density shown in Figure 15(b).

Figure 15. (a) Complete-fusion criterion relied on melt pool geometry, where $M_d$ and $M_d^0$ are the melt pool depth and the minimum ELT to have complete fusion (Adapted with permission from^[132]), (b) a relationship between the AM product quality and the optimum laser energy density,^[130] (c) transverse section of a single-track cladding layer of 1Cr18Ni9Ti stainless steel at ELT of 60, 80, and 100 $\mathrm{\mu }\mathrm{m},$^[137] (d) bridge-like sample before and after detachment from the substrate.^[140]

Shamsaei et al.^[133] reported the role of layer thickness on product quality in the DLD process. Since the conventional DLD is programmed by a CAD part slicing into parallel layers from bottom to top within a finite thickness of deposition layers, those surfaces whose normal vectors are not either parallel ($\theta =0°$) or perpendicular ($\theta =90°$) to the build direction can be only deposited. A designed part shape, e.g., curved surfaces are typically assembled by a series of parallel layers in which the material shrinkage of a single layer during the solidification creates slanted walls around the layer itself at the expense of dragging the previously deposited layers. A combination of geometric approximation and layer dragging causes a surface showing step-like features, also known as the staircase effect. The staircase effect leads to poor surface quality and requires post-processing machining, and consequently, more overall process time. The magnitude of this defect is dependent on the ELT and the angle between the surface normal vector and the build direction. In a method called multi-axis processing, a $90°$ rotation in the slicing direction of an overhang structure is applied instead of the classical parallel slicing.^[134] However, to control the ELT, changing the powder mass feed rate is usually taken into account, where the larger feed rate causes enlarging the ELT at the expense of a higher porosity volume fraction.^[135] Shamsdini et al.^[136] reported the same evidence in the LPBF maraging steels printed at two different ELT in which the more ELT, the more porosity volume fraction. They also observed higher ultimate tensile strength and elongation percentage in a sample fabricated at smaller ELT whose microstructure contained lower fraction of retained austenite phase. As seen in Figure 15(c), an increase in the ELT enhances the probability of micropores formation in the selective laser melting (SLM) 1Cr18Ni9Ti stainless steel.^[137]

The magnitude of the ELT and NLT is reported to be influential on other characteristics such as relative density in the LPBF steel alloys in which higher ELT causes a decrease in relative density. In terms of the microstructure features, higher NLT changes the as-built H13 tool steel microstructure by coarsening the width of primary dendrite arms.^[138] In a recent case, a synergy between high NLT, low scanning speed, and high beam power can be much more effective in coarsening the microstructure morphology. It has also been reported that the ELT in the binder jetting (BJ) process is the most crucial factor determining the integrity of the product and the possibility to design the smallest feature size. It is recommended that the optimized BJ end-product properties necessitate a tradeoff between applying a small ELT and increasing the building time because a large ELT causes unbounded layers, while a thin layer leads to pushing away of deposited layers as the next powder layer spreads on it.^[139]

As another characteristic affected by the ELT, distortion caused by residual stresses during layer-by-layer AM is considerable. Safronov et al.^[140] derived a mathematical model presenting a relationship between curvature radius, $R,$ the height of the beam, $H,$ and effective layer thickness, $t,$ in an as-built sample. As schematically shown in Figure 15(d), a bridge-like sample is usually produced to assess the magnitude of residual distortion generating during the powder fusion. Residual stresses stored in the as-fabricated sample are mostly released after sample/substrate detachment resulting in distortion of the beam. In Equation (14)(14) $$R=\frac{1}{3e}\frac{H^2}{H-t}$$(14) , by supposing a linear $\sigma -\epsilon$ behavior through the beam distortion, i.e., no nonlinear elastic and plastic deformation, and constant elastic properties along with the build direction, the curvature radius can be measured as (14) $$R=\frac{1}{3e}\frac{H^2}{H-t}$$(14) where $e$ is linear thermal strain estimating from the thermal expansion coefficient and a difference between the temperature of 3D printer chamber and liquidus temperature of the fuzing powder.

1.2.2. Laser power and beam distribution

In LAM for both curing and fuzing processes, the energy is directly transferred to material leading to either photochemical reaction in the curing process or photothermal reaction in thermal sintering and melting. A laser is usually used as an energy source due to the high-intensity beams irradiated on a surface of the printing material with maximum absorption efficiency with no transfer medium. In commercial 3D printing machines, the beam sources are CO₂ lasers, Nd-YAG lasers (neodymium-doped yttrium aluminum garnet), Yb-fiber lasers (ytterbium-doped), and excimer lasers, where the Yb-fiber lasers are usually employed in direct metal laser sintering (DMLS) and SLM machines. Table 5 reports the specifications of these beam sources.

Table 5. Specifications of beam sources used in AM.^[141]

Energy source	CO2 laser	Nd-YAG laser	Yb-fiber laser	Excimer laser
Application	SLA, SLM, SLS, LENS	SLM, SLS, LENS	SLM, SLS, LENS	SLA
Efficiency	5–20%	Lamp pump: 1–3% Diode pump: 10–20%	10–30%	1–4%
Output power ($\mathrm{CW}$)	Up to 20 $\mathrm{kW}$	Up to 16 $\mathrm{kW}$	Up to 10 $\mathrm{kW}$	Avg: 300 $\mathrm{W}$
Operation wavelength	9.4, 10.6 $\mathrm{\mu }\mathrm{m}$	1.06 $\mathrm{\mu }\mathrm{m}$	1.07 $\mathrm{\mu }\mathrm{m}$	193, 248, 308 $\mathrm{nm}$ (ArF, KrF, XeCl, respectively)
Operation mode	CW and pulse	CW and Pulse	CW and pulse	Pulse
Pump source	Electrical discharge	Flashlamp or laser diode	Laser diode	Excimer recombination via electrical discharge
Pulse duration	Hundreds $\mathrm{ns}$-tens $\mathrm{\mu }\mathrm{s}$	Few $\mathrm{ns}$-tens $\mathrm{ms}$	Tens $\mathrm{ns}$-tens $\mathrm{ms}$	Tens $\mathrm{ns}$
Beam quality factor ($\mathrm{mm}.\mathrm{mrad}$)	3–5	0.4–20	0.3–4	160 × 20 (Vertical × Horizontal)
Fiber delivery	Impossible	Possible	Possible	Possible
Maintenance periods	2000 $\mathrm{h}$	200 $\mathrm{h}$ (Lamp life) 10,000 $\mathrm{h}$ (Diode life)	25,000 $\mathrm{h}$	10^8–10^9 pulses (Thyratron life)

SLA: stereolithography; SLM: selective laser melting; SLS: selective laser sintering; LENS: laser engineered net shaping.

Lasers parameters can be specified in terms of average power, power stability, beam diameter, beam quality, central wavelength, spectral bandwidth, pulse duration, pulse energy, and repetition rate.^[142] However, the laser power as a determining characteristic varying the laser energy density is considered as an interplay among the operating wavelength, pulse frequency, and laser intensity. The absorptivity of each alloying element in a multi-component system is dependent on the operating wavelength of the beam source, where the shorter the wavelength, the better the beam absorptivity. For instance, the absorptivity of Fe element in a loose powder state is 64 and 45%, when Nd-YAG laser (operating wavelength = 1064 $nm$) and CO2 laser (operating wavelength = 10.6 $\mu m$) are respectively utilized.^[143] The laser intensity, i.e., the laser power per unit area, is remarkably dependent on the process throughput and material selection. For the materials showing a high thermal diffusivity or a high reflectivity, e.g., aluminum and copper, high laser intensities are necessitated to overcome the slow rate of temperature increase.^[142] In terms of the laser operation mode, except excimer lasers only operated in pulsed mode, most lasers used in DMLS and SLM machines are able to operate in both pulsed and continuous modes. In laser additive manufacturing, the pulsed mode can provide high thermal energy with minor dissipation to surrounding material; however, in the continuous wave mode, the energy level is hard to reach the threshold energies since there has always been a portion of beam energy diffusing into the surrounding materials. The more pulse frequency, the higher energy input interacted with the metal powder.^[142]

The beam spot size as a process parameter affecting the size and shape of melt pools is investigated in terms of beam distribution, e.g., Gaussian, inverse Gaussian, and flat-top (Tophat) distributions, shown in Figure 16(a).^[144] In a Gaussian power-density distribution, the consolidation zone is relatively wide and inclined to the powder layer plane; however, in both Tophat and inverse Gaussian, the consolidation zone is narrower and perpendicular to the powder layer.^[144] The depth of each fused track, on the other side, is observed larger at the center of the track than the peripheral regions in the Gaussian beam, while a more uniform depth of fusion is observed in other power-density distributions. Even at low laser powers, the high energy density at the center of a Gaussian beam (circular) guarantees adequate melting that avoids the LOF defect compared to a Tophat and inverse Gaussian (elliptical) distributions; however, the peripheral zones in a Gaussian distribution usually show lower energy density and consequently, more probability of the LOF defect at large beam shape, Figure 16(b).^[145] Roehling et al.^[145] reported that a Gaussian distribution provides a circular beam shape, while the Tophat and inverse Gaussian beams result in an elliptical beam shape with rough and discontinuous tracks, containing more equiaxed grains in the microstructure of 316 L stainless steels, Figure 16(c).

Figure 16. (a) Various types of power-density distribution,^[144] (b) energy density vs. laser power at three types of melt zone profile, 0: no deposition, 1: low substrate wetting between a melt bead and the substrate, 2: good substrate wetting by a melt bead showing no substrate penetration, 3: shallow substrate penetration with conduction mode laser melting, 4: Intermediate substrate penetration, 5: deep substrate penetration with key-hole laser melting, and (c) solidification microstructure of the melt pool at different energy densities and powers, 0: no fusion, 1: equiaxed, 2: equiaxed-columnar, 3: columnar (Reproduced with permission from^[145]).

It is reported that to increase the deposition rate in the LPBF process, the spot size needs to be increased with power; however, this results in a coarser surface finish.^[146] In terms of the laser power, Narvan et al.^[55] reported a decrease in the surface roughness of the LPBF-H13 steels, as the laser power increases with respect to other unchanged process parameters. The same improvement of the surface quality was observed in thin wall structures when the spot size of the beam was increased in the electron beam melting (EBM) process.^[147] The beam interaction with the powder feedstock is also considered in some research in which a defocused beam decreases the melt pool surface temperature and leads to a columnar grain morphology within the melt pools.^[148,149]

At high laser power, irregular morphology of a single track is observed, particularly when fast scanning speed is applied. An increase in laser power also enhances the maximum temperature of the powder bed, bringing a steeper temperature gradient, especially in materials with low thermal conductivity.^[150] There has been a close relationship between the laser track width and the values of laser power and scanning speed at which a high magnitude of scanning speed accompanied by low values of laser power drastically decrease $E_V,$ leading to an inadequate fusion of the irradiated powder and subsequently, tightening the track width toward the laser spot size. At constant $E_V,$ an increase in laser power makes the melt pool deeper and more stable in which the range of the stable zone becomes narrower in high thermally conductive materials.^[151] In terms of structural defects in the AM product, the product exposed to higher laser power during powder deposition is more prone to thermal stresses and hot cracking, as an increase in laser power intensifies the thermal shrinkage rate.^[152]

1.2.3. Laser scanning speed

Scanning speed as the velocity of laser beam movement, while it passes across a scan vector, is an important process parameter affecting surface quality, microstructure, and mechanical properties of AM products.^[153] In powder fusion AM, it has been reported that the maximum temperature of a melt pool slightly reduces with an increase in laser scanning speed; moreover, the liquid lifetime, i.e., the time interval between the onset of powder melting in a specific area and its complete solidification, is reduced as the scanning speed increases.^[154] A decrease in the liquid lifetime tightens the solidification duration resulting in a fine microstructure since the rejected alloying elements during solidification are able to diffuse within the short paths and thermal undercooling is high. A drop in maximum temperature of the melt pool at faster scanning velocity makes the temperature gradient, $G,$ gentler persuading the morphology more dendritic/equiaxed than columnar.^[7,155] It is worth noting that a decrease in a liquid lifetime, on the other side, intensifies the solid/liquid interface velocity, $R,$ during solidification, contributing to columnar to dendritic/equiaxed transition, as $G/R$ decreases.^[7,156] The role of scanning speed and laser power on the grain size and microstructure morphology has been reported inversely proportional in which a rise in laser power brings coarser microstructure, longer liquid lifetime, and subsequently, more columnar morphology.^[154]

There have been some contradictory reports on the effect of $E_V$ on melt pool shape and geometry. Zhang and Coddet indicated that the melt track dimensions (depth, height, width) are increased with increasing $E_V,$ where the depth and width are further prone to the magnitude of $E_V.$^[157] However, Cunningham et al.^[158] reported that an increase in laser power or a decrease in scanning speed, that brings higher $E_V,$ increases the depth of a melt pool, while the width becomes narrower. In another study, once the laser scanning speed as a variable in $E_V$ equation increases, the melt pool length increases, while its width reduces.^[151] According to Rombouts et al.,^[159] in the presence of high scanning speed, the melt pool becomes more elongated and loses the circular shape, eventually taking a comet-like shape. At the onset of the Plateau-Rayleigh instability governed by increasing the laser scanning speed, the melt pool tends to break into droplets, forming the beads. That is why the balling effect – which is the shrinking and breaking up the molten track into a row of spheres due to capillary instability decreasing the surface free energy – is dominant at high scanning speeds.^[150]

The higher the scanning speed, the higher the melt viscosity (i.e., less fluidity) due to a reduction in the maximum temperature of the melt pool. In terms of the melt viscosity, its magnitude should be balanced so that it becomes high enough to prevent the balling effect, but low enough to assure appropriate spreadability of the melt over a previously deposited layer.^[154] As another characteristic, the melt flow represented by the sign of the surface tension gradient between minimum and maximum temperatures of a melt pool, $d\gamma /dT,$ elucidates the shape of the melt pool; a shallow/broad one ($d\gamma /dT<0$) and a deep/narrow one ($d\gamma /dT>0$).^[159] A melt pool at high laser scanning speed or low laser power tends to show $d\gamma /dT<0,$ i.e., the peripheral zones with a temperature higher than the central zone in a melt pool. The melt flow instability and splashing emerge as the scanning speed goes up.^[160] It is worth noting that during SLM of steels and ferrous alloys, surface oxidation can dramatically reduce the surface tension close to the edge, causing a melt flow from the cooler liquid close to the edge to central zones of the melt pool.^[159]

The AM product shows higher tensile strength when a higher laser scanning speed is applied.^[154] The reason behind that can be interpreted by changes in both grain size and dislocations’ mobility. According to the Hall-Petch relationship, as the grain size reduces, the AM components show higher strength due to grain boundary strengthening. An increase in scanning speed might cause less sufficient time to diffuse the alloying elements during solidification, leading to narrow microsegregation and severity of solute trapping into the lattice structure.^[7] Less tendency of segregation or high susceptibility of solute trapping raises the shear strength of the lattice, generating a higher magnitude of the Peierls-Nabarro friction stress.^[156] The friction stress acts as an obstacle against dislocations’ mobility, resulting in higher dislocations’ density and strengthening. Higher scanning speed and/or ELT causes a reduction in relative density that is attributed to increasing the porosity volume fraction, particularly at low laser power.^[161] Consequently, the macro-hardness of the AM product is lower at a high scanning speed. There is an optimum level of $E_V$ in which both laser power and scanning speed are high so that the samples at this condition show a small portion of porosity mostly circular in shape, balling effect, and high thermal stress cracking. At low laser power and scanning speed, un-melted powder usually fills the porosity due to insufficient melting.^[162] Figure 17 illustrates the effect of laser scanning speed on various microstructural and mechanical parameters.

Figure 17. (a) How the scanning speed affects the deposition shape of a single track in 316 L stainless steel,^[150] (b,c) Macrohardness and relative density variations in 18Ni-300 steel samples at various scanning speeds and layer thicknesses (t) (Reproduced with permission from^[161]).

1.2.4. Hatch spacing

Hatch spacing as the center-to-center distance between two adjacent single tracks is another process parameter affecting the density, microstructure, and mechanical properties of the AM component; however, there is limited information in the literature investigating this parameter with respect to other unchanged process parameters. The optimal hatch spacing is measured via Equation (15)(15) $$h=0.7w$$(15) .^[163] (15) $$h=0.7w$$(15) where $w$ is the beam waist (or beam focus), i.e., the location along with the beam irradiation in which the beam radius size is minimum.

It was observed that a small change in hatch spacing has a negligible effect on the cooling rate through a melt pool as long as the melt pools of adjacent hatches sufficiently overlap.^[164] Mukherjee et al.^[164] showed that as the hatch spacing in LPBF 316 stainless steel increases from 70 to 110 $\mathrm{\mu }\mathrm{m}$ (at 1000 $\mathrm{mm}/\mathrm{s}$ scanning speed), the cooling rate decreased almost $1.5\times {10}^5\mathrm{ }\mathrm{K}/\mathrm{s}$ (∼3.5% reduction); however, the hatch spacing ranged between 35 and 70 $\mathrm{\mu }\mathrm{m}$ has a negligible influence on the magnitude of cooling rate. A decrease in cooling rate after a critical hatch spacing is attributed to separating the hatches, which reduces the heat transfer from the melt pool. As a result, the microstructure becomes coarser containing a higher fraction of columnar structure. In the presence of over-critical hatch spacing, the as-built product is susceptible to the LOF defect due to improper fusion bonds between successive hatches and layers, as depicted in Figure 18(a). However, the LOF voids are not only dependent on the value of hatch spacing. At a given hatch spacing, increasing the overlapping of fused regions via employing lower laser scanning speed minimizes the LOF voids shown in Figure 18(b). In this figure, the percentage of LOF voids was measured as a ratio of the un-melted area to the total area of the product cross-section. In conclusion, the LOF defect depends on hatch spacing, scanning speed, ELT, and size of the melt pool in which an increase in hatch spacing, scanning speed (a decrease in the melt pool size), and ELT intensifies the LOF defect vulnerability.

Figure 18. Effect of hatch spacing on (a) cooling rate of the melt pools, and (b) LOF voids percentage in 316 L stainless steel (Reproduced with permission from^[164]).

1.2.5. Laser scanning strategy

The scan strategy is defined as a combination between the laser pattern during scanning the powder layer, scan count, i.e., the scanning frequency of a layer prior to recoating the next fresh powder layer, and the laser parameters designated to a specific zone.^[165,166] This process parameter has been reported to affect temperature distribution, surface roughness, residual stresses, microstructure, mechanical strength, and relative density. As schematically revealed in Figure 19, there have been various scan strategies in terms of the direction of scanning over a certain powder layer or the scanning rotation between consecutive deposited layers. In island scanning, 9 isolated islands (2 × 2 $\mathrm{mm}$) cover the entire scanning domain, where the completion steps of a single island are randomly selected, and the subsequent layers are deposited through a scanning direction rotation. Line scanning without isolated islands (horizontal, vertical, or 45°) is back and forth scanning with the same scanning patterns over consecutive layers categorizing based upon the oriented angle with respect to the X-axis. The only difference between rotate scanning and line scanning is related to scanning pattern over consecutive layers in which the rotate scanning uses various oriented angles from one layer to another. In-out and out-in scanning are spiral scanning patterns with reverse direction from inside to outside of the scanning domain.^[165]

Figure 19. Layout of different scanning strategies (Reproduced with permission from^[165]).

Scan strategy plays an important role in the localized temperature evolution through the scanning domain, which determines the probability of deformation defects, e.g., LOF, key-hole defect, and hot cracking. It was shown that both line and rotate scanning patterns have comparatively lower minimum substrate temperature than the island scanning. A laser needs more time to scan a single layer in the island scanning, introducing more heat to the sublayers. The island scanning shows the highest magnitude in both minimum substrate temperature and maximum temperature compared to the other strategies, resulted from a longer time to completely scan a layer and residual heat effect from a short scanning path over a single island.^[165] The maximum temperature is hard to reach the steady-state condition in the island scanning due to periodic switch between scanning directions, path angles, and length.

Figure 20 illustrates the contours of residual stresses and deformation over the entire scanning domain. As shown, the regions close to the corners of deposited layers are subjected to the compressive residual stresses, where S11 and S22 stresses are maximum adjacent to the edges and gradually decrease toward the central region. In all scan strategies, the edge/interface locations are susceptible to cracking due to the localization of residual stresses. The out-in strategy shows the largest S11 and S22 values, while 45°-line scanning presented the smallest ones. This might be due to better uniformity in temperature gradient variance in 45°-line scanning patterns. The maximum difference between S11 and S22 stresses is belonged to the line scanning case in which the value is almost 160 $\mathrm{MPa}.$ Among three angular rotation cases (45, 67, and 90°), the 67°-rotate scanning presents the smallest difference between S11 and S22 stresses. As depicted in Figure 20(c) and Figure 20(e), the smallest deflection in build direction is achievable in 45°-line scanning case, while the in-out spiral strategy shows the largest deformation; almost two times more than the 45°-line scanning one. All angular rotation cases show a similar magnitude of deformation that is comparable with the 45°-line scanning. It is worth mentioning that the magnitude of deformation is simulated for three deposited layers; however, in real laser processing, the printed samples are exposed to much larger deformation.

Figure 20. (a–c) Residual stresses and deformation contours, (d) max S11 and S22 stresses, (e) max deflection magnitude at various scanning strategies (Reproduced with permission from^[165]).

In terms of the role of scan strategy in AM of steels, Bhardwaj and Shukla^[113] used bidirectional (0° line scanning) and cross-directional (90° rotate scanning) strategies to assess their effects on physical and mechanical properties of the DMLS 18Ni-300 maraging samples. They mentioned that the sample printed via cross-directional scanning shows a better surface finish and higher relative density. This sample contains a much lower fraction of retained austenite (10.3% $\gamma$) compared with the bidirectionally scanned sample (60% $\gamma$), as a change of scanning direction in cross-directional strategy leads to a multidirectional heat flux resulting in a higher solidification rate of the melt pool. The EBSD analysis of the samples confirmed that both strategies lead to the microstructure including columnar cells followed by epitaxial grain growth. However, the crystallographic texture is different in which the bidirectionally scanned sample reveals strong dominance of $〈111〉$ and $〈001〉$ textures along with the side surface and the top surface, respectively, while the cross-directionally one shows no dominance in texture due to a rotation in heat flux over successively deposited layers. Regarding the mechanical properties, the cross-sectionally scanned sample has higher yield and ultimate tensile strength, higher microhardness, and 2% lower ductility resulted from a strong interlayer bonding. The lower tensile strength of the bidirectionally scanned sample is attributed to the nucleation and coalescence of microvoids under the tensile loading, which leads to early crack formation.

In another study on the mechanical and microstructural dependency of AM steels on the scan strategy, Kudzal et al.^[167] analyzed different patterns to build AM 17-4 stainless steel. The authors noted that elongated scan lines aligned the load direction leads to an almost two-fold increase in the fraction of retained austenite compared with short scan lines perpendicular to the load direction. They used line scanning (unidirectional along the width (90 BF-F), unidirectional along the length (0 BF-F), back and forth along the width (90 BF-T), back and forth along the length (0 BF-T), Hexagon scanning, and concentric scanning strategies. The relative density was measured almost the same for all cases. However, the average fraction of austenite and tensile properties differed from one case to another is presented in Table 6.

Table 6. Summary on the effect of scan strategy on the properties of AM steels.^[167]

Material	Scan strategy	YS ($\mathrm{MPa}$)	UTS ($\mathrm{MPa}$)	El (%)	HV ($\mathrm{kg}/{\mathrm{mm}}^2$)	Avg. austenite (vol%)
DMLS 18Ni-300 steel^[^95^]	0°	720 ± 7	1021 ± 28	19 ± 0.70	380 ± 7	60
	90°	753 ± 10	1082 ± 62	17.5 ± 1.0	399 ± 4	10.3
LPBF 17-4 stainless steel^[^148^]	90 BF-F	810 ± 15	948 ± 36	4.8 ± 1.1	350.2 ± 11.7	27.2 ± 6.5
	0 BF-F	873 ± 10	951 ± 17	5.3 ± 2.7	346.7 ± 22.4	43.6 ± 7.9
	90 BF-T	773 ± 3	1043 ± 2	17.6 ± 2.6	355.3 ± 19.8	25.5 ± 2.6
	0 BF-T	866 ± 10	935 ± 8	3.3 ± 0.6	350.3 ± 18.8	69.9 ± 9.7
	Hexagon	798 ± 21	798 ± 21	15.8 ± 3.5	346.3 ± 13.8	58.3 ± 2.1
	Concentric middle	824 ± 30	824 ± 30	4.2 ± 2.4	356.1 ± 012.8	82.4 ± 5.3
	Concentric edges					50.3 ± 7.6

1.2.6. Environment of chamber

The chamber atmosphere in which the AM process is being conducted plays an important role in surface quality, metal transfer, dimensional accuracy, and structural defects of the product, e.g., porosity, entrapped oxide particles, etc. Shielding gas is primarily purged to avoid oxidation and atmospheric contamination, as powder is recoated, and solidification evolves. As the temperature enhances, the reaction rate between depositing metal and the surrounding atmosphere is accelerated, so the volumetric flow rate of shielding gas needs to be adequately high to constrain chemical reaction at melt/gas interface.^[168] In laser-based processes, the machine chamber is filled and recirculated via Ar, N₂, He, CO₂ or mixed gases relied on the depositing material’s properties. In terms of protective atmosphere, the oxygen level and chamber pressure need to be monitored online, where a trace oxygen sensor is used for this purpose. In this regard, chamber pressure fluctuation has been reported as an influential parameter on the powder feeding rate and the flow rate of processing gas.^[169]

Besides oxidation prevention, the gas flow contributes to an undisturbed operation by removing process by-products from the fusion zone. As depicted in Figure 21(a), the process by-products are mainly categorized as welding plume (plasma plume and metal vapor) and spatter/ejected powder. The plasma plume refers to the ionization of the gas directly above the interaction zone, particularly at high laser powers. The evaporation of some alloying elements in the center of the laser spot caused by a high energy density of the focused beam leads to metal vaporization ruptured out of the melt pool. Under the inappropriate operation of shielding gas, the vaporized metal is rapidly cooled down, forming particles with 10–150 $\mathrm{nm}$ in diameter. Based upon the duration of the condensation process, these particles agglomerate to various orders of magnitude.

Figure 21. (a,b) Schematic Representations of process by-products and spherical particles formed through redeposition of them,^[170] and (c–e) the effect of normal gas flow velocity on porosity level, melt pool width and penetration (1–9 correspond to location numbers on the build plate given in legend) (Reproduced with permission from^[172]).

The spatter powder is referred as liquid particles expelled during metal fusion, which is shown to be spherical and chemically identical to the raw material.^[170] The SLM process is very prone to spatter powder formation due to many factors such as high absorptivity of powder particles, unsteady feeding of the powder into the melt, abrupt changes in layer thickness, the operation mode of the laser, and surface contamination of the powder particles.^[171] As another process by-product, powder near the melt pool can be ejected from the powder bed and redeposited, as the welding plume dynamics have a key role in the powder bed instability shown in Figure 21(b). Ladewig et al.^[170] reported that to optimize the gas flow in the SLM process, it needs to be homogenous and directly cover the entire build plate. The flow rate should be constant and high enough to avoid redeposition and remove the process by-products. To reduce the potential disturbance of the laser beam by process by-products, the flow separation and turbulence needs to be minimized in the upward direction. Furthermore, the stream of shielding gas should be near the build surface in which the laser/by-products interaction can be minimized. Reijonen et al.^[172] indicated that the flow rate of shielding gas has effects on porosity level and melt pool geometry of LPBF 316 L stainless steel, where an increase in flow rate over a certain threshold decreases the instability during the LPBF process providing better penetrability, the lower volume fraction of porosity, and a decrease in melt pool width, as given in Figure 21(c–e).

Among the protective gases, H₂ is reported to reduce the available oxygen in the chamber preventing metals oxidation; however, it induces embrittlement in most alloys, particularly at high temperatures. For Ti-based alloys and other nonferrous ones, Ar and He are recommended, while Ar is more commercial due to its lower in price (3–4 times). He with much higher ionization potential (24.58 $V$) than that of Ar (15.75 $V$) also provides higher arc energy density in Gas Metal Arc Welding (GMAW).^[173] In Fe-based alloys and steels, Ar/N₂ is mostly used, as it also induces strengthening during heat treatment of steels. Pure Ar is another protective gas used to fabricate AM steels.^[168] N₂ addition to shielding gas mixtures brings benefits in terms of austenite stabilizing and corrosion resistance of the AM product.^[168] Except for chemical affinity of protective gas, other factors also determine gas/metal reaction severity in laser AM. The molecular mass of a gas is one affecting the heat balance. In this regard, He (0.1786 $\mathrm{kg}/{\mathrm{m}}^3,$ standard atmosphere) is significantly lighter than Ar (1.784 $\mathrm{kg}/{\mathrm{m}}^3,$ standard atmosphere) resulting in better properties in hot isostatic pressed AM products.^[174] The viscosity, heat exchange coefficient of convection, thermal conductivity, and thermal capacity of a gas have negligible effects on heat balance since the shear stresses applied by a gas are much lower than surface stresses due to the high thermal conductivity of the fuzing metal, and thermocapillary flow. The higher chamber interior gas pressure, the less oxygen entered into the chamber; however, in AM processing accompanied by no metal evaporation, the gas pressure shows insignificant trace on the material parameters and the process.^[175] Finally, the surface free energy of liquid/protective gas has an influence on the melt pool geometry and balling behavior.^[176]

The role of shielding gas composition on surface quality and mechanical properties of LPBF 304 and 17-4 PH stainless steels were investigated.^[177,178] It was reported that the mixed Ar/CO₂ gas led to the worst surface finish due to the higher phase fraction of martensite induced by carbon absorption during 304 SS powder deposition. On the other side, both N₂ and Ar showed much better surface finish and similar Ra values, while their thermal conductivities are different, 0.026 and 0.018 $W/mK,$ respectively. This confirmed that despite the gas chemical composition, its thermal conductivity has a negligible effect on the surface quality of the product. Regarding the dependency of mechanical properties on the gas composition, the LPBF 17-4 PH steels built under Ar and Ar/N₂ shielding gases were tested through tensile loading, where Ar/N₂ shielded samples showed higher tensile strength, hardness, and fracture strain. This was due to finer microstructure formed at high cooling rates provided by N₂ atmosphere.

1.3. Summary

The aerospace, marine, and automotive industries have shown a growing interest in additive manufacturing (AM), resulting in numerous studies on metal additive manufacturing. However, like any manufacturing technique, AM has advantages and disadvantages that require an understanding of process sequences and physical metallurgy. Although AM allows for the production of components with unique geometries, specific process parameters such as powder atomization atmosphere, atomization velocity, laser power, and laser scanning speed can impact the occurrence of structural flaws like porosity and cracks, resulting in variations in density and other properties. Additionally, factors such as the initial chemical composition, transformation temperatures, and thermal conductivity of the material also have an influence. To minimize unexpected deviations between the properties of AM components and the expected theoretical outcomes, this section provides an overview of powder production techniques and AM process parameters. The characteristics of the powder used in powder fusion AM play a vital role in ensuring the desired quality of microstructures in complex parts. These characteristics determine the relative density of the final product. Key properties of the powder include particle size, sphericity, wettability, oxidation, chemical composition, flowability, thermal conductivity, and surface roughness. The selection of an appropriate metal powder not only directly affects the physical and mechanical properties of the printed product but also enables the creation of complex microstructures with control over chemical composition, phases, and structural morphology, which is known as microstructure architecting. Therefore, comprehending the fabrication methods, properties, and defects of the powder is a significant advancement, as elaborated in this section. Apart from that, to optimize the properties of AM products, including their thermophysical, mechanical, and microstructural characteristics, it is crucial to understand and optimize the processing parameters, particularly effective layer thickness, laser power, hatch spacing, scanning speed, scan strategy, and environment of the printing chamber. These parameters encompass factors such as the morphology of the microstructure, microstructural defects, thermal expansion, fatigue, and tensile strength. To effectively develop AM processes, it is necessary to grasp the impact of each parameter; however, achieving precision in these processes can be challenging.

2. Microstructure of additively manufactured steels in as-built and heat-treated conditions

2.1. Microstructure of as-built components

The microstructure of an AM component and therefore its mechanical properties are strongly dependent on processing parameters, e.g., scanning speed, laser power, gas flow rate, powder feed rate, etc. During the non-equilibrium solidification in laser additive manufacturing (LAM), coexistence of high classical temperature gradient ($G$), fast solid/liquid (S/L) interface velocity ($R$), capillary effect or Marangoni convection, recoil pressure, surface tension-driven Plateau-Rayleigh instability, evaporative and radiative cooling makes the nature of solidification very complicated in the aggressively dynamic melt. Consequently, microstructural features, e.g., grain size, morphology, orientation, and phase fraction show significant changes by altering the alloying elements, AM processing techniques, and even from one section to another section of an as-built component. Melt pool shape and geometry as other factors changing the microstructural features are also subjected to these uncertainties and complexities during AM processing. Hence, understanding the principles of rapid solidification, phenomena taking place during phase transformation, and heat/mass transfer through a solidifying melt pool are useful to better correlate the microstructural and mechanical properties of an AM product. In this chapter, the microstructure of as-built-, and heat-treated AM steels and iron-based alloys is discussed in more detail.

2.1.1. Microscale characteristics of melt pool

Melt pool shape is reported to have a determining effect on the microstructure and texture of AM products since it directly changes $G,$ $R,$ and the thermocapillary effect. As previously discussed, melt track characteristics, i.e., root penetration depth, bead height, centerline roughness, contact angle, and track continuity, are proportionally related to the factors determining the energy density, as well as size and shape of the laser beam (cf. Section 3.2). While these are categorized as the macroscale characteristics, the ones designated to microscale are usually considered at two levels of the grain morphology and the solidification pattern.^[145] The grain morphology can be changed from columnar-dendritic to equiaxed, while the solidification pattern varies from quasi-planar (banded structure) to cellular to non-planar (dendritic).^[179]

Columnar grains with constrained nucleation at the fusion boundary usually epitaxially grow and form elongated grains, while equiaxed grains experience unconstrained nucleation anywhere in the fusion zone. In terms of solidification pattern, despite dendrites growth through various crystallographic directions (those with secondary and ternary arms), cells develop normal to the S/L interface (antiparallel to the heat extraction direction in a melt).^[145] Banded structure is usually formed over separate narrow regions parallel to the S/L interface, where the interface perturbations is hindered due to rapid $R$ value.^[179] Matthews et al.,^[180] reported that the laser beam and circular profiles remarkably change the melt pool shape and geometry. For instance, the elliptical beam leads to shallower melt pool and higher area of fusion boundaries compared with the Gaussian one, as revealed in Figure 22(a,b). Also, Figure 22(c–g) shows that laser beam profile has an apparent influence on melt track width, and subsequently, thermophysical properties of the solidifying melty, e.g., $G,$ $R,$ cooling rate, $\dot{T},$ natural convection in the melt, etc. The annular and transverse elliptical beams result in wide melt tracks presenting gentler temperature gradient, while the Gaussian beam acts vice versa. The longitudinal elliptical beam shows a moderate effect on the melt pool shape; however, the Bessel beam provides a confined temperature distribution next to the surface and shallow melt pools. An increase in spot size and laser power also makes the melt track wider for each beam profile.^[181] As follows, an interplay between these factors and energy density parameters directly changes the grain morphology and solidification pattern.

Figure 22. (a,b) Melt pool shape in the LPBF SS 316 L printed via Gaussian and elliptical beam profiles, respectively, (c–g) ALE3D-assisted FE simulation of a melt track fused using Gaussian, transverse elliptical, longitudinal, annular, and Bessel beam profiles, respectively (mean beam diameter: 100 $\mathrm{\mu }\mathrm{m},$ laser power: 550 $\mathrm{W},$ scanning speed: 1800 $\mathrm{mm}/\mathrm{s},$ domain size: 250 × 250 × 750 ${\text{μm}}^3$) (Reproduced with permission from^[180]).

Figure 23(a–c) presents the effect of laser power on the fraction of equiaxed grains in the LPBF SS 316 L at constant energy density, Q, and circular-medium size (C-M) beam profile, where M = 187 $\mu m.$ At constant Q, laser scanning speed increases with laser power. The velocity of interface is correlated to the laser scanning speed via Equation (16)(16) $$R=\mathit{vCos}\alpha$$(16) . (16) $$R=\mathit{vCos}\alpha$$(16) in which $\alpha$ is an angle between the directions of solidification and laser scanning. This means that an increase in laser power speeds up the solidification rate (at constant $Q$), and consequently, decreases $G/R$ magnitude, and develop the S/L interface perturbation persuading columnar-dendritic solidification.

Figure 23. Grain morphology over a cross-section of melt track roots at (a-c) various laser powers (C-M: circular-medium size beam profile) and (d-f) various beam profiles where the equiaxed area is 2%, 28%, and 77% for the C-M, LE-M, and T E-M, respectively (C: circular, LE: longitudinal elliptical, TE: transverse elliptical) (Reproduced with permission from^[145]), (g,h) optical micrographs of melt pools in samples scanned parallel and perpendicular to shielding gas direction, respectively, and (i,j) their corresponding IPF maps (Reproduced with permission from^[183]).

The role of beam profile in grain morphology and substrate penetration depth at constant beam size, laser power, and Q is also shown in Figure 23(d–f). Equiaxed solidification at low laser power takes place over a larger area when the elliptical and Bessel beam profiles are used instead of the circular one. Although the homogenous nucleation of equiaxed grains is suppressed due to very high $G,$ non-stochastic nucleation mechanism induced by melt mixing is responsible for equiaxed solidification. Natural convection due to thermocapillary and recoil pressure effects causes a melt vortex associated with a melt track topological depression, i.e., stir of hot melt from the depression to high-velocity transition region with lower temperature. The flow velocity of melt is higher within melt pools fused by elliptical laser profile. Fragmentation and redistribution of dendrite tip ahead of solidification front as a result of high-velocity melt flow introduce a huge number of potential nucleation sites for equiaxed grains.^[182] In terms of beam profile, the transverse elliptical profile is reported to be the most suitable for equiaxed solidification^[145,180]

An angle between the shielding gas flow and scanning direction is reported as another factor changing the melt pool shape and crystallographic texture.^[183] Andreau et al.^[183] studied the melt pool shape, size, and grain texture of the LPBF SS 316 L scanned parallel and perpendicular to shielding gas direction at constant laser power, scanning speed, and Q. Interestingly, the sample printed perpendicular to gas direction shows higher height/width ratio of melt pool, which confirms deeper melt pool penetration accompanied by keyhole transition welding mode. However, the other sample with shallower melt pools experiences a conduction mode with less vulnerability to microstructural defects, as revealed in Figure 23(g,h). In terms of grain orientation and texture, the sample printed along the gas direction shows very coarse <101> grains epitaxially grew through at least 40 successive layers, while the other sample contains more grains randomly oriented and higher area fraction of <001> ǁ BD columnar grains, Figure 23(i,j). Despite a remarkable difference in grains’ orientation and average grain size (laser scanning along the gas direction results in a two-fold increase in grain size), both elucidates a similar Goss texture with small difference in counter plot intensity.

2.1.2. Alloying elements and phases

Regarding chemical composition, cooling condition, size and geometry of the component, and process parameters, various phase constitutions can be observed in the final microstructure. In this section, phase constitutions of different types of steel, i.e., austenitic stainless steels, duplex stainless steels, martensitic stainless steels, precipitation-hardening (PH) stainless steels, tool steels, and oxide dispersion strengthened (ODS) steels, are discussed.

2.1.2.1. Austenitic stainless steels

The grades SS 316 L and SS 304 L in this series have received much attention in AM processes since high Cr content (17–18 $wt.\mathrm{\%}$) makes them a suitable candidate in corrosion-resistant applications in marine, oil and gas, and aerospace industries.^[184,185] High Ni content (8–14 $wt.\mathrm{\%}$) stabilizes the austenite phase at ambient temperature, resulting in a high fraction of austenite in the as-built microstructure. Solidification of both grades can start with either primary austenite or primary δ-ferrite. Direct energy deposition (DED) of these steels is usually accompanied by microsegregation of ferrite stabilizers (Cr and Mo) in the intercellular regions.^[66] Consequently, the DED microstructure mostly contains an austenitic matrix with a small volume fraction of fine ferritic films (up to 9 $\mathit{vol}.\mathrm{\%}$), as depicted in Figure 24. Although solute partitioning also occurs within narrow distances during the LPBF processing, Cr and Mo segregated in intercellular regions are insufficient to stabilize ferrite, so the LPBF microstructure is fully austenitic.^[186] Transformation-induced plasticity (TRIP)-, and twinning-induced plasticity (TWIP)- steels are like the austenitic stainless steels, since their microstructure is usually fully austenitic; however, the presence of other phases has also been reported. In this regard, Haase et al.^[56] reported that the as-produced microstructure of an LPBF high-manganese steel (X30Mn22) mostly contains austenite, together with $\epsilon$-, and ${\alpha }^{\prime }$ martensite.

Figure 24. A Schematic layout of typical microstructures obtained after LPBF and DED processing of various steels, ret.: retained, GB: grain boundary, $\alpha :$ ferrite – bcc, ${\alpha }^{\prime }:$ martensite – bcc/bct, $\gamma :$ austenite – fcc (Adapted with permission from^[64]).

2.1.2.2. Duplex stainless steels

A combination of ferrite and austenite in the microstructure of these steels is due to the presence of a very high concentration of Cr (24–26 $wt.\mathrm{\%})$ as a ferrite stabilizer, and high Ni concentration comparable with Ni content in austenitic stainless steels. However, the austenite phase usually forms during post-processing heat treatment in which its fraction can be as much as the ferrite volume fraction. Hence, the as-built microstructure is exclusively ferritic (primary δ-ferrite forms during solidification) with a small fraction of austenite along the grain boundaries (cf. Figure 24). During the LAM process, by increasing the energy density, a higher fraction of austenite can be observed in the as-built microstructure that is attributed to more time at which the austenite can grow at elevated temperatures. Unwanted Cr-rich $\sigma$ phase that has a deleterious effect on ductility is usually absent in the as-built microstructure, while post-processing heat treatment above 1200 °C to form austenite leads to form this intermetallic phase.^[187,188] In superduplex steels, N is also added into the initial composition, which leads to forming a small amount of chromium nitride (${Cr}_{2}N$) at grain boundaries of the as-built microstructure. In this regard, Eriksson et al.^[189] used the wire arc additive manufacturing (WAAM) process to produce Zeron 100X steel and observed a high fraction of chromium nitride and austenite in the as-produced microstructure ascribed to remarkably slower $\dot{T}$ in WAAM compared with the LPBF process. It is worth noting that the duplex stainless steels are widely used in harsh environments, particularly where high resistivity against pitting and crevice corrosion is required.

2.1.2.3. Martensitic and PH stainless steels

Martensitic stainless steels are within 400 series stainless steels containing zero or low percentage of C and Ni that is inadequate to stabilize austenite at ambient temperature, while Cr content is comparable with the one in austenitic stainless steels (almost 12–14 $wt.\mathrm{\%}$). DED, LPBF, and laser cladding can successfully produce AM martensitic stainless steels with suitable density; however, there are limited works available on the processing of these grades. So far, the grades 410, 420, and Nitrogen-alloyed X30CrMoN15-1 have been produced using the mentioned techniques.^[58,190,191] Although the microstructure is usually exclusively martensitic, DED-manufactured and laser cladded parts are reported to contain a low fraction of eutectic δ-ferrite.^[191] The LPBF SS 420 at high laser powers and LPBF X30CrMoN15-1 show retained austenite as the second phase into the martensitic matrix.^[58,192] PH stainless steels with promising mechanical properties, weldability, and corrosion resistance attract much attention for AM production. The microstructure can be either fully martensitic or austenitic/martensitic (cf. Figure 24), among which two famous grades of 17-4 PH and 15-5 PH are categorized as fully martensitic grades.^[193] However, the as-produced microstructure of LPBF 17-4 PH can consist of a high fraction of austenite (up to $f_{\mathit{Austenite}}$=1) based upon the process parameters and type of shielding gas used during the process.^[194] Nevertheless, the as-produced microstructure of WAAM 17-4 PH contains a small fraction of retained austenite embedded into a dense δ-ferritic/martensitic matrix.^[195] In the PH stainless steels, Cr content (16–18 $wt.\mathrm{\%})$ is added to enhance the corrosion resistance, while Ni and Cu (3–5 $wt.\mathrm{\%}$ each) act as elements that stabilize the austenite, and induce precipitation strengthening, respectively. As another grade of PH stainless steels, 17-7 PH, Uddeholm Corrax, or EOS CX stainless steel can be named, in which Cu is replaced by Al to induce precipitation strengthening through Ni₃Al phase formation.^[196] Shahriari et al.^[78] reported that the solidification of LPBF-produced Corrax steel starts with nucleation of δ-ferrite then followed by austenite, and finally, martensite formation. They observed pockets of martensite containing a large number of fine parallel laths, and a small fraction of retained austenite in between. Although a thermophysical tool predicted the presence of primary ferrite in the as-produced microstructure, TEM results did not confirm this result. They also reported two types of carbides (M₂₃C₆ and M₇C₃) in the as-built microstructure.

2.1.2.4. Tool steels

Carbon-bearing tool steels and carbon-free maraging steels are two kinds of tool steels used in AM, and the final microstructure of both is martensitic in as-produced condition. Martensite in carbon-bearing tool steels is hard and brittle, so post-processing tempering is necessary to form carbides and recover some ductility.^[52] On the other side, the martensite in as-produced maraging steels is relatively ductile and soft, which require aging heat treatment to obtain strength through a precipitation-strengthening mechanism.^[52] Due to rapid cooling during LAM and high hardenability of these steels, an exclusively martensitic microstructure (with/without retained austenite) is almost inevitable (cf. Figure 24), which induce thermal-stresses cracking in carbon-bearing tool steels; however, maraging steels with a ductile microstructure show invulnerability to this defect.^[197]

The most well-known grades of carbon-bearing tool steels are H11 and H13 hot working tool steels, the X65MoCrWV3-2 cold working steel, and the M2 high-speed steel, which have attracted attention in AM.^[64] For instance, the grade H13 can be produced using LPBF, EBM,^[198] and DED, in which preheating the powder bed ($T>100°\mathrm{C}$) results in a dense crack-free AM component.^[199] In both LPBF-, and DED-produced microstructures, the retained austenite is reported to be in interdendritic regions assigned to the partitioning of Cr, Mo, V, and C in these areas.^[200] In the case of C partitioning, carbide precipitates can also form in the as-built material.^[201] The as-produced microstructure of the LPBF H11 is similar to H13 grade, where the fraction of retained austenite can be as much as 15 $wt.\mathrm{\%},$ and reduces to less than 3 $wt.\mathrm{\%}$ after post-processing tempering.^[202] A low amount of M₃C carbides is also observed in the as-produced state. In the case of high preheating temperature ($T>500°\mathrm{C}$), the as-built microstructure contains retained austenite, carbides, and upper bainite (instead of martensite).^[203] Similarly, the microstructure of the M2 grade contains martensite, retained austenite, and carbides (M₆C and M₄C₃); however, a eutectic phase as an uninvited guest is also reported in interdendritic/intercellular regions, which is due to the segregation of the alloying elements with low distribution coefficient.^[204]

The most prominent maraging steels used in AM processes are 18Ni-300, 18Ni-250, and 14Ni-200 grades.^[64] During cooling, prior austenite transforms to martensite in which the size of martensite lath blocks is dependent on the size of prior austenite grains. In this regard, the prior austenite grains of a DED sample were reported to be coarser than a LPBF one, where the grain diameter can increase up to 1 $\mathrm{mm}$ in DED material. (Note that the cell size in DED-, and LPBF-produced material is nearly 0.3–2 $\mathrm{\mu }\mathrm{m},$ and 5 $\mathrm{\mu }\mathrm{m},$ respectively^[64]). Regardless of AM process, the final microstructure of maraging steels also shows retained austenite (up to 6–11%) that often surround the martensite lath blocks. The retained austenite along cell boundaries is attributed to the segregation of alloying elements like Ni in interdendritic/intercellular regions.

2.1.2.5. Oxide dispersion strengthened (ODS) steels

ODS stainless steels with applications mainly at elevated temperatures, e.g., membrane reactors, and fission/fusion power plants, are either fully ferritic or martensitic (martensitic/ferritic), since their chemical composition has at least 12 $wt.\mathrm{\%}$ Cr and no austenite stabilizers (cf. Figure 24). Their suitable corrosion resistance is due to high Cr content (can be up to 23 $wt.\mathrm{\%}$), as well as Al addition in some grades, while their satisfactory creep resistance comes from fine oxide particles (mostly Y₂O₃) homogenously distributed through the matrix (the stoichiometry of Y₂O₃ may change to Y₂Ti₂O₇ or Y₄Al₂O₉ during AM processing). Some prominent grades of this series are Fe-19Cr-5.5Al-0.5Ti-0.3Y₂O₃ (PM2000 or MA956), Fe-14Cr-1W, Fe-18Cr-2W-0.5Ti-0.3Y₂O₃, which can be produced using LPBF and EBM techniques.^{[60,205–207]} The earliest work on LPBF ODS steels dates back to 2009 when Walker et al.^[60] used mechanically alloyed PM2000 powder to synthesize single-wall specimens. They reported that at low laser powers, and importantly, slow scan speeds, the oxide precipitates coarsen, even coarser than the ones in conventionally produced PM2000 (48–61 $nm$ vs. 30 $\mathrm{nm}$). However, optimized process parameters used to fabricate the steel Fe-14Cr-1W resulted in the formation of finer oxides in the range of 10–40 $\mathrm{nm}.$^[205] Notwithstanding the mechanical alloying of powder to synthesize ODS steels, the oxide precipitates can be formed in-situ during the LPBF process. To this end, the oxygen concentration in the melt is deliberately increased using either an oxidized powder feedstock (provided by powder atomization under an oxidizing gas atmosphere), or an oxygen-enriched atmosphere in the printer chamber. In the latter case, an increase in oxides’ number density is expected. The hypothesized mechanism behind the in-situ formation of oxide precipitates is that the superheated melt dissolves the oxides during AM processing. A portion of debonded oxygen can exit from the melt pool surface as spatter, and/or float in the melt pool, and/or be entrapped in the melt, forming coarse oxide inside the bulk. The rest of debonded oxygen seemingly precipitates from the liquid steel to form the small round oxide precipitates.^[64]

2.1.3. Morphology of grains

The theory of directional solidification by Kurz and Fisher^[208] indicates that melt pool boundaries are heteronucleation sites, where solidification in AM processes often starts and is directed inward. $G$ and $R$ are the major factors determining the morphology of grains; however, the effect of other factors such as partitioning coefficient, $k_0,$ constitutional undercooling, ${\Delta T}_C,$ and convection due to thermocapillary forces should be also considered.

A strong morphological and crystallographic texture is usually observed through a solidified melt pool due to the unidirectional heat transfer and cooling where heat is vertically conducted away through the part already printed and the substrate.^[209] In the absence of grain nucleation-induced techniques such as inoculant addition and ultrasonic treatment during solidification, the morphological texture of an AM product usually shows columnar and equiaxed grains along the lateral section.^[210] As schematically shown in Figure 25, due to a high $G/R$ magnitude at the beginning, solidification starts with nucleation and growth of columnar grains from melt pool boundaries. The evolution of rapid solidification results in establishing a short-length solute diffusion zone ($\delta$) ahead of S/L interface. Within this zone, the alloying elements, mostly those containing low $k_0,$ are rejected from a solid into a liquid. Subsequently, a decrease in liquidus temperature, also called ${\Delta T}_C,$ facilitates frequent grain nucleation with equiaxed morphology ahead of the grown columnar grains. Besides, a decrease in $G$ values contributes to the unconstrained formation of these grains, as they form easier than columnar grains due to an interplay between ${\Delta T}_C,$ thermal-, and curvature undercooling caused by the Gibbs-Thomson effect.^[211] Consequently, the solidification of the top central zone of a single melt pool is accompanied by nucleation and growth of equiaxed grains. Ghoncheh et al.^[156] mentioned that at high $R$ values in LAM, the $R$-dependent distribution coefficient approaches to unity ($k\to 1$), causing a very narrow $\delta ,$ and short-range solute segregation. This phenomenon – solute trapping – is responsible for another feature called banded microstructure, i.e., alternate structures or phases developed parallel to the transformation front due to an unstable-to-oscillation manner of the S/L interface, as presented in Figure 25.^[212] Natural convection due to the thermocapillary forces in the melt has a synergistic interaction with high $R$ values on abating the segregation, as it conducts better redistribution of solute, and subsequently, further narrowing $\delta .$^[213] To sum up, at $R$ values (0.1–1.0 $m/s$) during solidification of LAM components, there has been a mixed morphology of columnar, equiaxed, and banded structure over a single melt pool.

Figure 25. (a) A schematic layout of grain morphology evolution in a solidifying single melt pool (Adapted with permission from^[156]), (b) a lateral section of a direct-energy-deposited specimen of SS 316L (Reproduced with permission of [215]), (c, d) solute banding lines (red arrows) in the LPBF SS 316L (Reproduced with permission from^[212]).

The layer-by-layer deposition is usually accompanied by partial remelting of the last deposited layers in which the severity of remelting and the number of layers exposed to reheating cycles are strongly dependent on the process parameters and thermal conductivity of the material. The partial remelting persuades heterogeneous grains nucleation and their growth through $〈001〉$ preferred direction toward the pre-grown columnar grains. In the case of high laser beam power or slow scanning speed where the depth of fusion zone is enough for partial remelting, the growth of columnar grains can spread over the melt pool boundaries, introducing coarse elongated grains epitaxially grown on each other.^[216] The size of elongated grains and the continuity of epitaxial growth are dependent on the columnar-, and equiaxed grains volume fraction ($f_v$) and their size in the pre-deposited layers. In the case of low $f_v^{\mathit{equiaxed}},$ the primary columnar grains act as heterogeneous nucleation sites for further columns nucleation and growth, facilitating the epitaxial growth beyond the melt pool boundaries, as shown in Figure 26(a). As $f_v^{\mathit{equiaxed}}$ increases, the equiaxed grains in the layers underneath that survive the partial remelting intercept epitaxial growth of columnar grains parallel to the building direction. In this case, these grains act as nucleation sites for newly formed columnar grains within the freshly deposited layer, as elucidated in Figure 26(b).^[210]

Figure 26. (a,b) IPF maps of AM SS 316 L printed under 1000 and 400 $W$ beam powers, respectively,^[210] (c,d) effect of ultrasound treatment on ${\Delta T}_C$ magnitude (CS zone), epitaxial columnar growth, cavitation effect, and columnar-to-equiaxed transition time ($T_E,\mathrm{ }{T}_A:$ equilibrium liquidus-, and actual temperatures, ${\mathrm{\Delta }T}_C,{\mathrm{\Delta }T}_n:$ constitutional-, and nucleation undercooling, $t:$ time),^[219] (e–g) IPF maps of AM Fe-Ti alloy containing 2, 5, 7.5 wt.% Ti, respectively (Reproduced with permission from^[222]).

Epitaxial growth is undesirable as it introduces a strong anisotropy in the tensile behavior of AM component. Besides, the microstructure containing the elongated columnar grains is expected to be more susceptible to shrinkage porosity, and hot cracking due to a drastic drop in the intercolumnar liquid pressure, particularly near the root of columns.^[217] As the solidification evolves, the viscous liquid cannot be properly sucked into the intercolumnar channels, resulting in porosity formation, hot tear nucleation, and propagation due to the residual stresses applied by the thermal gradient, solidification shrinkage, and phase transformation.^[218] Hence, the more fraction of equiaxed grains, the better mechanical properties; however, optimizing the process parameters and controlling the dynamics of melting and solidification to achieve this goal are always big challenges. An increase in grain nucleation frequency by inoculant-induced heteronucleation or ultrasonic treatment during the solidification is another technique that has attracted increasing attention.^[219,220]

Todaro et al.^[219] employed a high-intensity ultrasound technique controlling the solidification process of the direct energy deposited (DED) SS 316 L. The ultrasound-assisted DED technique in which the component is developed layerwise on a vibrating probe introduces accelerated solidification. The melt pool remains molten for only 0.01–0.1 s due to the ultrasound irradiation-induced solidification. As illustrated in Figure 26(c,d), they found that not only does ultrasonic treatment promote columnar-to-equiaxed transition and the formation of random crystallographic texture, but it also results in a nine-times increase in the nominal density of grains. Besides, the high-intensity ultrasound provides higher contents of ${\Delta T}_C$ by lowering $G$ value over the entire melt pool. While an increase in ${\Delta T}_C$ is responsible for the equiaxed grains nucleation, the fine morphology of the grains is due to the acoustic cavitation effect where the formation, growth, and collapse of bubbles occurs abruptly in liquid metal (∼ 3.0E-5 s). Acoustic cavitation applies intense energy-matter interactions, with hot spots up to 5000 °C, pressure of 1.0E6 $\mathit{kPa},$ heating and cooling rates at 1.0E11 $°\mathrm{C}/s$ inside the bubbles.^[221] In another study, Ikehata et al.^[222] reported that sub-micron Ti(O,N) particles properly act as a grain refiner in AM ferritic steel of Fe-Ti produced by LPBF. They proposed inoculant-induced heterogeneous nucleation as the main mechanism controlling the grain size and morphology, which is in accordance with some other reports.^[214,223] In this regard, nanoparticles with a small lattice misfit with the matrix, e.g., TiO and TiN are more efficient for grain refinement. The microstructure of the LPBF Fe-Ti at various Ti contents printed under Nitrogen gas is shown in Figure 26(e–g).

2.1.4. Dendrite arm spacing

Microstructural features, i.e., grains, dendrite arm spacing, inclusions, etc., play a critical role in the mechanical and corrosion properties of AM builds. While in many cases, fine microstructural features are of interest, AM component in some applications like creep needs coarser grains with low area fraction of high angle grain boundaries. Hence, the size of microstructure features is a critical characteristic varied by many factors, e.g., thermophysical condition, manufacturing processes, working atmosphere, etc., and needs to be monitored precisely. Here is an overview of microstructure features size in different AM processes and some factors proportionally correlated with it.

Grain size as a characteristic that determines toughness and hardness of AM build is reported to be changed by $\dot{T}$ during solidification, $G/R$ magnitude, presence of heteronucleation sites, etc. By considering a columnar-dendritic grain as a colony of dendrite arms, dendrite arm spacing (DAS) is a term determining the fineness of dendrites, and subsequently, the grains. There have been some theoretical models that show a correlation between DAS and other thermophysical factors. The initial model proposed by Hunt^[224] was based on an assumption that the dendrite tip is spherical; however, the values obtained for the PDAS (P: primary) were inconsistent with experimental results. Later, Kurz and Fisher^[225] derived another model in which the dendrite tip radius was considered equal to the instability wavelength at the solid/liquid interface. Their results were found to over predict the PDAS. To have a better consistency between theoretical and experimental results, Trivedi^[226] proposed another model, Equation (17)(17) $$\mathit{PDAS}=2.83{\left(\frac{1}{G}\right)}^{0.5}{\left(\frac{\Delta T_{0}L{k}_{0}D\mathrm{\Gamma }}{R}\right)}^{0.25}$$(17) , that is in accordance with experimental data obtained by electron beam AM.^[227] (17) $$\mathit{PDAS}=2.83{\left(\frac{1}{G}\right)}^{0.5}{\left(\frac{\Delta T_{0}L{k}_{0}D\mathrm{\Gamma }}{R}\right)}^{0.25}$$(17) where $\mathrm{\Gamma },$ $D$ in $m^2/s,$ $k_0,$ and $\Delta T_0$ in $K$ are the Gibbs-Thomson coefficient, diffusion coefficient in liquid, distribution coefficient, and equilibrium solidification range. Also, $G,$ $R,$ and $L$ stand for temperature gradient, solid/liquid interface velocity, and the quadratic function of harmonic perturbation. While this model correlates PDAS to micro-, and atomic-scale thermophysical properties, some research was focused on how to roughly estimate the SDAS (S: secondary) in $\mu m$ using $\dot{T}$ in $K/s.$ In this regard, Knapp et al.^[228] used Equation (18)(18) $$\mathit{SDAS}=A{\left(\dot{T}\right)}^B$$(18) to measure SDAS in powder-blown AM SS 316 L. They reported that as the laser power increases from 1000 to 2500 $W,$ SDAS is enlarged from 2.51 to 3.27 $\mathrm{\mu }\mathrm{m}.$ (18) $$\mathit{SDAS}=A{\left(\dot{T}\right)}^B$$(18) in which $A$ and $B$ are constants equal to 50 and −0.4, in this case. However, these constants are different in other alloying systems as well as PDAS measurements. For instance, Tian et al.^[229] reported PDAS = 0.42 ± 0.12 $\mathrm{\mu }\mathrm{m}$ in LPBF M789 steels, when $\dot{T}$ was 1E06 $K/s.$ Therefore, this empirical equation needs to be rederived and validated for each case study using experimental measurements. For this purpose, image processing techniques conducted through micrographs are usually employed.

Some efforts were also carried out to find a relationship between grain size and $\dot{T}$ in AM components. Ma et al.^[230] proposed mathematical equations for SS 316 L manufactured via selective laser melting (SLM) and laser cladding deposition (LCD). As seen in Figure 27, width, $\lambda ,$ and length, $L,$ of columnar grains versus $1/\sqrt{\dot{T}}$ follow different trends regarding the manufacturing techniques. In the LCD SS 316 L, these terms can be measured using Equations (19)(19) $$\lambda =36.6+\frac{6.6\times {10}^2}{{\dot{T}}^{0.5}}$$(19) and (20)(20) $$L=432+\frac{2.3\times {10}^4}{{\dot{T}}^{0.5}}$$(20) : (19) $$\lambda =36.6+\frac{6.6\times {10}^2}{{\dot{T}}^{0.5}}$$(19) (20) $$L=432+\frac{2.3\times {10}^4}{{\dot{T}}^{0.5}}$$(20)

Figure 27. Role of $\dot{T}$ in determining (a) width, and (b) length of columnar grains in SS 316 L alloy produced by various AM techniques (Reproduced with permission from^[230]).

On the other hand, for the samples produced using SLM technique, Equations (21)(21) $$\lambda =1.6+\frac{4.2\times {10}^4}{{\dot{T}}^{0.5}}-\frac{2.6\times {10}^7}{\dot{T}}+\frac{5.9\times {10}^9}{{\dot{T}}^{1.5}}$$(21) and (22)(22) $$L=-117.5+\frac{8.2\times {10}^5}{{\dot{T}}^{0.5}}-\frac{6.2\times {10}^8}{\dot{T}}+\frac{1.7\times {10}^{11}}{{\dot{T}}^{1.5}}$$(22) are proposed. (21) $$\lambda =1.6+\frac{4.2\times {10}^4}{{\dot{T}}^{0.5}}-\frac{2.6\times {10}^7}{\dot{T}}+\frac{5.9\times {10}^9}{{\dot{T}}^{1.5}}$$(21) (22) $$L=-117.5+\frac{8.2\times {10}^5}{{\dot{T}}^{0.5}}-\frac{6.2\times {10}^8}{\dot{T}}+\frac{1.7\times {10}^{11}}{{\dot{T}}^{1.5}}$$(22)

As a consequence of coupling the equations proposed for DAS and grain size as functions of $\dot{T},$ a correlation between microstructural features size can be obtained; however, this estimation is rough since $\dot{T}$ is not constant during solidification and drastically varied by thermophysical properties previously mentioned. It is noted that for AM processes using heteronucleation-induced techniques, an interdependence model has been proposed to measure the grain size.^[231]

The effect of inclusion as another microstructural feature on the properties of AM builds can be either detrimental (stress localization sites due to a difference in thermal expansion coefficient with matrix, brittle nature, sharp corners, improper wettability, etc.) or beneficial (ODS steels). While the oxygen and nitrogen content in powder feedstock is an important factor that indirectly affects the volume percentage and distribution of inclusions in the final product, AM process parameters are crucial factors determining the fraction of inclusions in-situ formed during manufacturing. In SLM maraging steels, there is a broad variety in size of inclusion, as illustrated in Figure 28(a–e). Thijs et al.^[232] called the oxide layer at the periphery of melt pools surface as a reason for large inclusions formation. An abundance of oxides at this location is due to an outward Marangoni flow in melt pools with a negative surface tension gradient. However, the oxides on the melt pools surface with a positive tension gradient are prone to be detached by an inward Marangoni flow and drought into the melt. In maraging steels, Ti and Al show a severe affinity to oxygen, which contributes to oxide layer formation. Stability of the oxides (Al₂O₃ and Ti₃O₅) is more than the nitrides in liquid steel; however, a portion of Ti content also reacts with nitrogen (in atmosphere and shielding gas) to form small cubic TiN inclusions. Hence, inclusions with a wide size distribution are observed in the SLM maraging steels, where the small ones are formed by oxide entrapment into the melt or in-situ reaction of trapped gases with alloying elements, while the irregular coarse ones mostly form on top of each deposited layer.

Figure 28. (a–d) Top and side views of the LPBF 18Ni-300 maraging steel produced via laser remelting under pure N₂ atmosphere. Dark and white arrows respectively show the white parent powder embedded into the dark grey oxides, and TiN inclusions, (e) Top-view SEM micrograph of the LPBF 18Ni-300 maraging steel that shows heavily cracked inclusions. Single melting under oxygen rich N₂ atmosphere was employed,^[232] (f) role of $\dot{T}$ in Sn-Mn inclusions size in conventionally cast, and LMD SS 316 L (Reproduced with permission from^[235]).

Other grades of AM steels show inclusions with different sizes and chemical compositions. For instance, Si-, and Mn-rich oxide inclusions in two sizes of ≤50 $\mathrm{nm}$ and 100–300 $\mathrm{nm}$ are reported in laser metal deposited (LMD) SS 316 L.^[233,234] Besides the role of mentioned factors in the size of inclusions, $\dot{T}$ is also reported as another parameter that drastically changes the size of the inclusions.^[235] In this regard, as $\dot{T}$ increases, the mean radius of inclusions decreases, as presented in Figure 28(f).

2.1.5. Grain coarsening

Grain coarsening during LAM is a phenomenon caused by heating/reheating cycles, uneven temperature distribution along the building direction and changes in process parameters. This phenomenon does not imply grain growth during post-processing annealing treatment. Regarding the Hall-Petch relationship, grain coarsening results in a decrease in both hardness and yield strength of the material; however, for high-temperature applications, the coarse-grain microstructure is recommended to improve creep life. Grain coarsening also brings a high vulnerability of partially melted zone (PMZ) at the grain boundaries. Kou^[236] reported that the PMZ is less ductile, as the grains become coarser. Also, less area of grain boundaries due to grain coarsening leads to more concentration of low-melting-point phases and impurities at intergranular regions, intensifying hot cracking susceptibility through the PMZ. Hence, except in some high-temperature applications, the coarse-grain microstructure is unfavorable; however, its occurrence is almost inevitable. Even for high-temperature applications, there have been many efforts conducted through the creep life improvement by grain boundary engineering rather than the usage of components with coarse-grain microstructure.^[237]

While $G/R$ determines the grain morphology in the microstructure, $G\times R$ or $\dot{T}$ is used to describe the grain size in the microstructure. As schematically shown in Figure 29(a), a single melt track scanned with a scanning speed equal to $V$ shows maximum $R$ at the centerline (CL) and minimum $R$ at the fusion line (FL) (cf. Equation (16)(16) $$R=\mathit{vCos}\alpha$$(16) ). Nevertheless, due to the elongated shape of the melt pool and the greater distance between the hottest spot of the pool ($T_{\max }$) and the pool boundary ($T_L$), maximum and minimum $G$ occurs at FL and CL, respectively. Consequently, (23) $$\{\begin{array}{c}{G}_{CL}<G_{FL}\\ R_{CL}\gg R_{FL}\end{array}$$(23) (24) $${\left(\frac{G}{R}\right)}_{CL}\ll {\left(\frac{G}{R}\right)}_{FL}$$(24) (25) $${(G\times R)}_{CL}>{(G\times R)}_{FL}$$(25)

Figure 29. (a–c) Top and side views of a melt track showing variation in $G,$ $R,$ solidification mode, and grain size across fusion zone (Adapted with permission from^[236]), (d) cross-section micrographs of WAAM HSLA showing grain coarsening along the building direction (Q-PF: quasi-polygonal ferrite) (Adapted with permission from^[238]).

These equations indicate that it is expected to have a transition in solidification mode across the fusion zone, where the grains near the FL are mostly cellular, while those that are formed at the CL are dendritic/equiaxed. It is worth noting that the planar mode usually takes place through a very narrow region near the FL in welding processes; however, in AM processes like LPBF, it repetitively happens through the melt pool to form a banded structure.^[179] Also, as the cooling rate increases from the FL to CL, the DAS and grain size are expected to become finer. Variations in grain morphology and size are presented in Figure 29(b,c). Successive layer deposition during AM processes makes the thermal nature of the solidifying component complex in terms of heat extraction rate, temperature distribution, and interlayer temperature. During fresh layer deposition, the layers underneath the top layer are exposed to partial melting and/or solid-state reheating cycles. Based on the magnitude of heat input into these layers, either solid-state grain growth due to grain boundary migration within the heat-affected zone (HAZ) or heteronucleation of new grains in PMZ can take place. Rodrigues et al.^[238] monitored the interlayer temperature in WAAM high strength low alloy (HSLA) steel. They reported that during layer-by-layer deposition, initial layers experience high $\dot{T}$ and low interlayer temperature. By distancing from the substrate, $\dot{T}$ goes down, causing higher interlayer temperatures, and subsequently, coarser grains, as illustrated in Figure 29(d). Hence, the wall shows a gradient in grain size ranging from fine to coarse along the building direction.

The role of process parameters in the grain size of AM builds was comprehensively studied by Kohen et al.^[239] They employed various platform preheating (200–800 °C), scanning speeds (550–950 $\mathrm{mm}/\mathrm{s}$), and scan strategies to compare grain size, geometrically necessary dislocations (GND) density, texture, and cells size in the LPBF X30Mn21 austenitic advanced high strength steel (AHSS). They figured out an increase in scanning speed results in fine grains, reduced texture intensity, and high GND density while using higher preheating temperatures encourages the formation of coarse columnar grains, remarkable texture intensity, and low GND density. In terms of scan strategy, the samples printed using Mark&Sleep (M&S) strategy present finer microstructure than those subjected to bi-directional continuous strategy. As shown in Figure 30, except for the sample M&S (no preheat, scanning speed of 750 $mm/s$), the rest were printed using the bi-directional continuous scan strategy. High scanning speed accompanied by M&S strategy facilitate partial columnar to equiaxed transition that introduce fine microstructure.

Figure 30. Effect of scanning speed, scan strategy, and preheating temperature on grain size and other microstructural features of the LPBF X30Mn21 AHSS (Adapted with permission from^[239]).

2.2. Heat-treated microstructure

2.2.1. Steel classifications

A critical part of AM process is post-processing, which is the final step of the fabrication process. Although AM process has many advantages, the drawbacks of this process could limit its full applicability in many industrial sectors. It is reported that repetitive thermal cycles during the LPBF process of metallic components results in high-stress concentration, inappropriate surface roughness, high residual stresses, inaccurate dimensions, and large porosity.^[240] Despite all the advantages that AM has over traditional manufacturing methods, the post-processing operations are therefore required to enhance the mechanical and functional aspect of materials fabricated by this emerging method. Some common post-processing operations that have been applied previously include heat treatment, electrochemical polishing, laser polishing, shot peening, and grinding.^[241–244] In this section, we aim to provide a review of the heat-treatment processes applied on steel components made by the fusion-based AM. Our focus is to study the influence of heat treatment on microstructural changes, mechanical and electrochemical properties of the as-built steel parts. In addition, a major part of this section is limited to the LPBF and directed energy deposition (DED) as fusion-based AM techniques to fabricate metallic components, particularly steel parts.

Steel is one of the main industrial alloys and almost 80% of all metallic engineering components are fabricated from. The conventional fabrication processes such as casting, hot and cold working are the main processes to produce different industrial steel parts.^[245] Although AM technique has been achieved a successful trend in the production of components from different types of metallic alloys, this technique still needs more development and adoption to different steel grades. The most common steels adopted to AM processes based on the fusion mode are classified and reviewed in this section:

1. Stainless steels including austenitic, martensitic/precipitation hardening, and duplex grades are corrosion resistant and have general longevity under harsh environmental conditions.^[246]

2. Tool steels, which have a high resistance to abrasion due to their high strength are attractive in the tool and die industry. The maraging steels like 18Ni-300, which are almost free from carbon, are the most suitable choices of tool steels in AM manufacturing.^{[138,247,248]}

3. Oxide dispersion strengthened (ODS) steels with a few studies in terms of AM process are a steel grade usually composed of a ferritic matrix with a dispersion of reinforced particles.^[249]

2.2.2. Heat treatment of stainless steels

2.2.2.1. Austenitic stainless steels

Austenitic stainless steels are one of the more applicable categories of steel in different industry sectors. This type of stainless steel has excellent corrosion resistance, biocompatibility, and ductility. Different industrial parts are fabricated from this type of stainless steel by AM process for aerospace, defense, and petrochemical sections.^[250] 316 L stainless steel is almost one of the more commonly used austenitic stainless steel and there is much literature about the different aspects of this material fabricated by AM processes. The high corrosion resistance of this type of steels returns the presence of high content of chromium (17–18 wt.%), while the austenite stabilizer element in this alloy is nickel, which stables the austenite phase through a broad temperature range down to ambient temperature.^[64]

Considering the wide application of 316 L, particularly biomedical applications, its wear resistance needs careful consideration. Tascioglu et al.^[251] carried out a study to understand the effect of various heat treatment temperatures on the microstructural and mechanical performance of the as-built 316 L components fabricated by LPBF. They considered three different heating temperatures of 600, 800, and 1100 °C for a soaking time of 2 h followed by air cooling. The higher heating temperature resulted in the formation of new grains and disappearing the melt pool boundaries, as shown in Figure 31.^[251] Moreover, both hardness and porosity were found to decrease with increasing heat treatment temperature. It was also observed that structural characteristics determined by XRD, including intensities and peak widths, were influenced by the heat treatment temperature. In the aforementioned study, it was demonstrated that the wear resistance of the as-built 316 L was strongly affected by porosity, rather than hardness. Heat treatment at a higher temperature led to more reduction of the porosity level inside the microstructure, leading to a higher wear resistance of the heat-treated samples.^[251]

Figure 31. The microstructural features of the as-built and heat-treated LPBF 316 L samples (Reproduced with permission from^[251]).

The improvement of wear resistance of 316 L parts after solution treatment was also supported by Benarji et al.,^[252] who observed an increase in the rate of wear for the heat-treated samples. In their research, heat treatment was conducted at temperatures of 800 and 1000 °C for about 2 h with water quenching as the final step. It was found that columnar dendrites developed in the as-built microstructure were dissolved after heat treatment at 800 and 1000 °C (Figure 32(b,c)). In addition, a uniform microstructure without a preferential growth direction was observed after heat treatment. The SEM micrographs shown in Figure 32(d–e) also confirm that the phase fraction of ferrite reduces when the temperature of heat treatment increases.^[252] They also reported the reduction of microhardness after the heat-treatment processes, which was similar to the results presented by Tascioglu et al.^[251] From automatic ball indentation (ABI) examinations, it is observed that the maximum displacement of 0.11 mm was obtained for the sample heat-treated at 1000 °C. In comparison with the as-built samples, the plasticity retaining capacity of 316 L sample after heat-treatment process at temperatures of 800 and 1000 °C was also increased by 40.4% and 72.8%, respectively (Figure 32(f)). Grain coarsening and reduction in ferrite phase were reasoned for this behavior in heat treatment temperature.^[252]

Figure 32. Microstructure of the LPBF 316 L at conditions of (a) as-built, (b,c) solution-annealing treatment at 800 °C and 1000 °C, respectively, (d,e) SEM images of the samples solution treated at 800 °C and 1000 °C, respectively, and (f) load-displacement curves plotted via automatic ball indentation (ABI) testing (Reproduced with permission from^[252]).

An increase in corrosion resistance and specific wear rate of 316 L samples were reported by Benarji et al.^[252] An improvement in corrosion resistance after heat-treatment process was ascribed to a reduction in the phase fraction of the ferrite after the heat-treatment process. Furthermore, an increase in the specific rate of wear in the heat treated 316 L samples was observed. Grains coarsening accompanied by reduction in hardness resulted in an increase in the coefficient of friction (COF) after heat treatment. The heat-treated samples showed a low phase fraction of ferrite, which affects the mechanism of wear. The major mechanism of wear for the heat-treated samples was considered to be plastic deformation with spalling and delamination. The SEM micrographs of the worn surfaces of as-built and heat-treated samples after wear testing are shown in Figure 33. The wear debris with plowing grooves was detected for the as-built samples parallel to the sliding direction (Figure 33(a)). The wear mechanism for the as-built sample was abrasive wear, while an abrasive-adhesive mechanism was corresponded to heat-treated samples (Figure 33(b,c)).^[252]

Figure 33. SEM micrographs of worn surfaces in (a) as-built sample, and (b, c) heat-treated samples tested at wear-test temperatures of 800 °C and 1000 °C, respectively (Reproduced with permission from^[252]).

Chen et al.^[253] performed a heat-treatment process at temperatures of 1000 and 1200 °C for 1 h on 316 L samples fabricated by gas metal arc additive manufacturing (GMA-AM). Their results showed that heat treating at 1000–1200 °C for 1 h does not clearly affect the morphology of grains, but remarkably affects the contents of σ and δ phases, as shown in the EBSD orientation-imaging maps of as-deposited and heat-treated steels and inverse pole figure (Figure 34). In as-built steel, the grains morphology is columnar (Figure 34(a)). When the sample was heat-treated at temperatures of about 1000–1200 °C for 1 h, the columnar grains had no obvious change compared to the as-deposited state (Figure 34(b–d)). The complete elimination of σ phase at a higher heat treatment temperature (1200 °C) resulted in a reduction in ultimate and yield tensile strength of the components, and an increase in the amount of elongation and reduction of area. Although they stated that the σ phase has a better strengthening effect than δ phase, the σ phase could also degrade ductility and increase the susceptibility to cracking in the steel. Furthermore, they showed that limiting the volume fraction of both σ and δ phases through heat treatment can improve the corrosion resistance of steel.^[253]

Figure 34. Orientation-imaging maps of the as-deposited and heat-treated (HT) GMA-AM 316 L: (a) as-deposited, (b) HT at 1000 °C for 1 h, WQ, (c) HT at 1100 °C for 1 h, WQ, and (d) HT at 1200 °C for 1 h (Reproduced with permission from^[253]).

Riemer et al.^[184] also investigated the effect of heat treatment of stress-relieving at 650 °C for 2 h on fatigue performance of 316 L fabricated by the LPBF technique. It was found that the microstructure of the heat-treated samples was very similar to those in the as-built condition. Moreover, they reported high ductility of LPBF 316 L in heat-treated condition. The behavior was attributed to the high cycle fatigue (HCF) condition and the crack growth behavior, which were not significantly affected by defects induced by the process, i.e., the pores and internal stresses^[184] Transformation of austenite to duplex austenite-ferrite during the heat-treatment process at temperatures higher than 900 °C was observed by Saeidi et al.^[254] It was shown that the reduction of the dislocations’ concentration occurred for the samples heat-treated at 800 °C for 1 h, while the cellular structure and cell size remained the same (Figure 35(b)). Moreover, the cellular-structure disappeared after heat treatment at 900 °C (Figure 35(c)), and images at higher magnifications showed the traces of them in some areas (Figure 35(d)).^[254]

Figure 35. A comparison between cellular structures in AM 316 L steel: (a) as-built, (b) heat treated at 800 °C for 1 h, (c, d) SEM and bright field TEM images of the sample heat treated at 900 °C, (e,f) EBSD phase map and grain orientation map of the sample heat treated at 1100 °C, and (g) tensile stress–strain curves of the as-built and heat-treated (HT) samples (HT was annealing at 1100 °C for 1 h) (Reproduced with permission from^[254]).

In the sample heat-treated at 1100 °C, further merging of the non-cellular sub-grains into coarser sub-grains was observed. The EBSD phase map (Figure 35(e)) confirmed the coexistence of ferrite and austenite (austenite is shown as the main phase with green color, while ferrite in red color). Substantial energy stored in the as-built steel within a high cooling rate of the LPBF provides a driving force for recrystallization and partial grain growth. Furthermore, they showed that the tensile strength for the 316 L as-built samples was higher than those heat-treated at 1100 °C. The stress-strain plots related to the as-built and heat-treated specimens are presented in Figure 35(g). This behavior was ascribed to the fine columnar structure along with micro sub-grains, high dislocations’ concentration and pure single-phase austenite formed in the as-built condition in comparison with the heat-treated state, which showed a larger grain size, less dislocation’s density, and stress relief. Therefore, synergetic effects of grain size, strain relief, and phase differences were reasoned for the difference in strength.^[254]

2.2.2.2. Martensitic stainless steels

Martensitic stainless steels are mainly produced according to a Fe-Cr-C ternary system. The phase fraction of martensite formed in this type of steels is controlled by the amount of undercooling, where the Koistinen–Marburger equation is used to predict the amount of martensite formed during the quenching.^[243] The crystalline structure of the martensite is a body-centered tetragonal (BCT) with restricted slip systems. Therefore, martensitic stainless steels show high resistance to deformation, and a brittle behavior. Moreover, the formation of a saturated solid solution from carbon contributes to a part of strengthening the martensite. Therefore, a heat-treatment process such as tempering is conducted on this type of steels to achieve a good combination of mechanical strength, and toughness.^[255,256] Among different types of martensitic stainless-steel alloys, AISI 420 is more suitable for AM processing due to medium carbon content (>0.15%) in its chemical composition, which provides good weldability within the AM process.^[257] In addition to this group, precipitation hardening (PH) stainless steels have shown suitability for production through AM due to a low C content and high weldability. PH stainless steels are developed in either martensitic or austenitic grades.^[193] 17-4 PH and 15-5 PH called “fully martensitic” grades are more applicable to AM processing. The main alloying elements in most of PH stainless steels are chromium (about 16–18 wt.%), nickel, and copper (3–5 wt.% each). High corrosion resistance is obtained in this type of steels due to the presence of chromium, which results in the formation of a passive layer.^[258] Nickel as an austenite stabilizer has also played an important role to enhance corrosion resistance. Cu precipitates formed during the aging heat-treatment process result in the precipitation hardening of these alloys. The content of chromium in 15-5 PH steel is slightly lower than that of 17-4 PH; however, molybdenum is added to 15-5 PH stainless steel. Also, the other types of this category are developed based on adding Al or another alloying element instead of the copper for the precipitation hardening proposes. One of these new developments is EOS CX or Uddeholm Corrax.^[196] The addition of Al leads to precipitation of NiAl phase, which increases the strength by the precipitation hardening mechanism.^[64] Although there are different types of martensitic stainless steels as well as the austenitic PH stainless steels, the martensitic PH stainless steels are discussed more in the literature in terms of AM processes.^[259] Therefore, we focus on only PH stainless steels in this section due to a huge attention to them in AM.^[250]

The influence of the heat-treatment process on the mechanical performance of 420 martensitic stainless steel fabricated by DED process was studied by Alam et al.^[260] Their study indicated that the mechanical behavior of the as-built specimens was different in longitudinal and transverse directions due to an anisotropy developed in the microstructure during the DED process. They also performed a heat-treatment process to eliminate the anisotropic issue. Accordingly, the heat treatment at a temperature of 565 °C for 1 h effectively removed the anisotropic features of the DED 420 stainless steel samples. Figure 36 showed the as-built and heat-treated microstructures of the 420 stainless steel samples. After the heat-treatment process, the inter-diffused areas, as well as the beads, were eliminated from the AM samples cut from the transverse direction (Figure 36(c)). Also, the tempered martensite obtained after the heat-treatment process includes a lower volume fraction of retained austenite and delta-ferrite (Figure 36(d)).^[260] These microstructural changes during the heat-treatment process result in a more homogeneous microstructure, which eliminates the anisotropic structural features developed in the as-built condition. It is an important outcome of the heat-treatment process, which should be considered for designing functional components. The results of mechanical properties after the heat-treatment process performed on a transverse sample at a temperature of 565 °C for 1 h showed an improvement by about 30% in the yield strength, and 11% in ductility (%TE). However, the tensile strength was decreased about 30%. The type of fracture also was changed after the heat-treatment process, in which the heat-treated samples showed a ductile fracture as compared to the as-built samples.^[260]

Figure 36. Cross-sectional SEM micrographs of (a) as-built, and (b) heat-treated samples, and (c, d) higher magnifications of Figure 36(b) (etched by ralph reagent) (Reproduced with permission from^[260]).

Similar results were also obtained by Saeidi et al.^[256] They found that printing the 420 stainless steel samples followed by tempering at 400 °C increases the ultimate tensile strength from 1670 to 1800 MPa_, and yield strength from 600 to 1400 MPa. Also, the elongation of the samples was increased from 3.5% in the as-built condition to about 25% after the tempering. They disclosed that an increase in the volume fraction of austenite had a profound impact on the mechanical behavior of the samples after tempering. According to their findings, the heat-treated samples show higher tensile strength and ductility in comparison with those of the as-built samples due to the formation of a finer grain structure accompanied by the formation of austenite and occurring a transformation induced plasticity (TRIP).^[256] In addition, Nath et al.^[257] reported an improvement in mechanical properties of the LPBF 420 stainless steel samples after adding Nb and Mo accompanied by tempering at a temperature of 315 °C for 2 h followed by air cooling. After heat treatment, the UTS of 420 stainless steel components with Nb and Mo improved to 1750 ± 30 MPa, and elongation to 9.0 ± 0.2%, much higher than the properties previously reported for LPBF and wrought 420 stainless steels. It was stated that the formation of nanoscale NbC particles after tempering results in an increase in the mechanical properties of the samples.^[257]

Microstructural heterogeneities are induced due to the unique heating, melting, and solidification conditions occurred during AM process. Heat-treatment processes are designed for the AM products to control the volume fraction of some phases, and to remove the residual stresses produced during the AM process.^{[241,261,262]} It is previously reported that 17-4 PH stainless steel parts produced by the AM process show a non-equilibrium microstructure with different texture components developed in parallel and perpendicular to the build direction.^{[102,263–265]} Also, it is detected that in some conditions, the volume fraction of retained austenite developed in the 17-4 PH components fabricated by the AM process is noticeable.^[266,267] The evolution of a fine-grain structure, the presence of large strains developed at grain boundaries, and added alloying elements such as nitrogen from powder or within the fabrication process are the main reasons for the retention of austenite after AM solidification process.^[263,265] A higher volume fraction of austenite was also revealed in the as-built 17-4 PH samples fabricated under nitrogen atmosphere compared to those fabricated in an Argon control atmosphere process.^[268] Different post-processing procedures are designed to modify microstructural features in 17-4 PH AM components, e.g., changes in volume fractions of the austenite and martensite, and elimination of uneven distribution of grain size.^[266,267] In an investigation performed by Cheruvathur et al.^[50] the as-built AM 17-4 PH parts were homogenized at 1150 °C to achieve uniform microstructures. The homogenization annealing resulted in eliminating the dendritic structures produced during the solidification and 5-fold reduction in phase fraction of the retained austenite after the heat-treatment process. However, microstructural inhomogeneities were still exhibited in the heat-treated components. Although the standardized heat treatments have been established for the PH stainless steels fabricated by the conventional manufacturing processes such as casting, there is still uncertainty about the effect of the standard heat-treatment processes on microstructures and the phases developed in additively manufactured PH stainless steel samples. Sun et al.^[269] studied the impact of a standard heat treatment procedure for the wrought samples of 17-4 PH stainless steel on LPBF 17-4 PH parts. In their work, the as-built AM parts were heat-treated in a solution annealing and aging process H900. Accordingly, the solution annealing process was conducted at a temperature of 1038 °C for 4 h, and then quenching was performed in air. They named the solution annealing step of the heat treatment as ‘Condition A’. After the solution annealing process, samples were aged at 482 °C (900 F) for 1 h with subsequent air quenching. The image orientation and the corresponding grain boundary maps were taken from the solution heat-treated sample, as shown in Figure 37(b,e), respectively. The EBSD maps show that the microstructure of as-built samples converts to a much finer and more homogeneous structure after the solution annealing treatment (HT), and the pole figure of the aged sample is swept, cf. Figure 37(c). Moreover, the EBSD results showed a higher fraction of low angle grain boundaries (LAGBs), about 51%, after the heat-treatment process. Accordingly, it was concluded that the as-built microstructure was converted to a conventional martensitic structure after the solution annealing process. Also, significant changes in the grain structure of the AM sample were not detected after the subsequent H900 treatment (Figure 37(c)).

Figure 37. EBSD orientation maps of (a) as-built, (b) solution-annealed, (c) solution-annealed-aged samples, (d) distribution of grain boundary corresponding to the areas shown in Figure 37(a,e) grain boundary maps of the areas shown in Figure 37(b) (color Code for the grain boundaries are: Green: 2°–15°, red: 15°–50°, and blue: 50°–180°) (Reproduced with permission from^[269]).

Sun et al.^[269] reported a reduction in phase fraction of austenite after the solution annealing process; however, the volume fraction of austenite was slightly increased after the subsequent aging process. Furthermore, it was revealed that after the solution annealing-aging process, nanoscale Cu-rich precipitates were distributed in the martensitic matrix of 17-4 PH samples produced by AM process. Also, they did not observe any changes in grain structure after the heat-treatment process. Their results also demonstrated that despite some differences between the microstructural features developed in AM samples and those of the wrought parts, the solution annealing and aging treatment led to the development of similar microstructure and phases in both AM and wrought 17-4 PH samples.^[269] Nezhadfar et al.^[270] also studied the changes in the microstructural features and mechanical performance of the 17-4 PH stainless steel components produced by LPBF after the heat treatment. They designed five heat-treatment processes, with and without primary solution annealing cycle, i.e., Condition A (CA), for the LPBF specimens in both as-built and machined-surface conditions. They studied the influence of heat treatment on fatigue and tensile behavior of the samples. Their findings depicted that the solution annealing at 1050 °C for 30 min (CA) considerably improves the fatigue strength of the LPBF 17-4 PH samples. A more homogenized microstructure was formed after the CA thermal cycle. In addition, the fatigue strength of the parts was increased after machining and polishing processes. Similar results on the effect of the heat-treatment process on fatigue performance of the LPBF 17-4 PH parts printed in the horizontal and vertical directions were declared by Yadollahi et al.^[264] Solution annealing for 30 min at 1040 °C, and then air cooling (AC) to room temperature was considered as Condition A. Also, precipitation hardening for 1 h at 482 °C and then air quenching to room temperature was named “Condition H900 or peak-aging” for the as-built 17-4 PH samples. Monotonic tension tests and fully reversed (R = −1) strain-controlled fatigue tests were performed on both the as-built and heat-treated specimens. Fatigue test results obtained from their study indicated that solution annealing and subsequent aging treatment (CA-H900) are useful for fatigue life in a low cycle and detrimental for the fatigue life in a high cycle in LPBF 17-4 PH samples. In addition, Figure 38 reveals that the specimens after the heat-treatment process (HT) showed a lower high-cycle fatigue life than that of the as-built (AB) counterparts. The heat treatment resulted in the evolution of a harder and more sensitive microstructure to impurities that are sourced from the LPBF process. However, higher low cycle fatigue strength was revealed for heat treated specimens in regimes of short life, where defects’ sensitivity was lower.^[264]

Figure 38. Fatigue life in the LPBF 17-4 PH samples at various heat treatment and manufacturing conditions (Reproduced with permission from^[264]).

Sarkar et al.^[271] performed a study of the effect of heat treatment on mechanical performance and corrosion resistance of LPBF 15-5 PH stainless steel components. Their results showed that the mechanical properties, as well as corrosion resistance, improved after standard aging condition (H900¹) due to the evolution of nanoscale spherical ε (Cu-rich) precipitates with an average size of ∼15 nm. However, the distribution of Cu-rich precipitates in the martensitic matrix of 15-5 PH samples leads to the development of a brittle structure with lower impact energy as observed in SEM micrographs taken from the fracture surfaces after heat-treatment processes (Figure 39).^[271] It was stated that residual stresses are developed in the as-built samples. Also, due to the different thermal cycles applied to different sections of the as-built parts within the LPBF process, different microstructural features are developed in the top and bottom surfaces. Accordingly, different levels of the residual stresses are formed, and an inhomogeneous microstructure is developed throughout the specimen. The interconnections between layers are affected by the microstructural heterogeneities resulting in an increase in energy absorption during the impact test. The observations of the fracture surface of as-built specimens can confirm the above reasons (Figure 39(a)). After impact testing, fracture surfaces showed relatively less homogenous ductile behavior, whereas the solution annealed LPBF specimens showed more homogenous pull-out features (Figure 39(b)). It has been detected that not only does aging treatment increase the yield strength and hardness of the LPBF sample, but it also leads to the formation of a brittle structure. Fracture surface after the aging process showed no sign of ductile pull-out (Figure 39(c,d)). When the aging temperature was increased to 621 °C, named as H1150² condition, the formation of coarser ε (Cu-rich) precipitates accompanied by an increase in volume fraction of retained austenite lead to a more ductile behavior of the samples, as shown in Figure 39(e,f).^[271] Finally, it was concluded that solution annealing can reduce anisotropy in mechanical properties due to the development of a more homogenized microstructure. However, aging at higher temperatures and longer soaking time does not have a significant impact on the mechanical properties, but it deteriorates the corrosion performances.^[271]

Figure 39. SEM Fractographs of the 15-5 PH samples after impact testing: (a) as-built, (b) solution-annealed, (c,d) H900, and (e,f) H1150 conditions (Reproduced with permission from^[271]).

Nong et al.^[272] compared the microstructural and mechanical performances of the as-built and heat-treated 15-5 PH samples fabricated by LPBF process with their wrought counterparts. Their EBSD analysis revealed that the phase fraction of austenite is reduced after the heat-treatment process, which increases the strength of the material. TEM image of the heat-treated LPBF sample is also presented in Figure 40(a), where the accumulation of dislocations adjacent to the inclusions is displayed with white-colored arrows. They reported that an increase in the material’s strength is due to the pile-up of dislocations around the inclusions, which can retard the movement of newly formed dislocations under force. X-ray residual stress analyzer (μ-X360) was used to measure the level of residual stress after the heat-treatment process for the LPBF 15-5 PH samples. An average residual stress of almost 127 MPa was measured for the heat-treated samples. They showed that the high density of dislocations and heterogeneities developed in the LPBF 15-5 PH parts leads to the generation of residual stresses. A higher density of dislocations at fine grain leads to an enhancement of the dislocations’ resistance to slip transfer and an increase in strength (Figure 40(a)). TEM images with a high resolution showed that the dislocations in heat-treated samples are piled up at the grain boundaries (Figure 40(b) – red arrows). However, the accumulation of dislocations observed in the heat-treated wrought samples was not similar to those of the LPBF heat-treated samples (Figure 40(c)). The HE-XRD profile of phase transformation in heat-treated 15-5 PH samples produced by the LPBF technique within the tensile test is shown in Figure 40(d). It was observed that the diffraction intensity peaks of austenite such as (111), (220), and (311) decrease when the strain increases. This reduction was attributed to an induced strain transformation in which the retained austenite transforms to martensite, increasing the tensile strength and plastic stability.^[272]

Figure 40. Bright-field TEM micrographs of (a,b) heat-treated LPBF 15-5 PH, (c) heat-treated wrought 15-5 PH, (d) HE-XRD profiles of heat-treated LPBF 15-5 PH samples at different applied strains (in these profiles, austenite and martensite are marked in green and red colors, respectively) (Reproduced with permission from^[272]).

Recently, CX stainless steel was developed by EOS GmbH, which is in the category of PH 13-8 Mo,^[273] and of which is compatible with the LPBF process. It was reported that CX stainless steel parts without defects could be produced by the LPBF process.^[196,274] The study conducted by Hadadzadeh et al.^[79] aimed to study the microstructural and mechanical performance of LPBF CX in the as-built and heat-treated conditions. Also, they tried to determine the strengthening mechanisms for this alloy. The effect of various heat treatments on the hardness and microstructure of LPBF CX samples were assessed in their study. Two different heat-treatment processes were designed for the LPBF CX parts. One of the heat-treatment processes was annealing at 900 °C for 1 h, air quenching, and then aging at a temperature of 530 °C for 3 h followed by quenching in air. The other heat-treatment was aging at 530 °C for 3 h with subsequent quenching at air without prior annealing step. Their results showed a detrimental effect of austenization-aging treatment on the strength of LPBF CX. This behavior was attributed to the growth of the martensite laths and retardation of the formation of nanoscale precipitates. However, the aging process without the annealing step resulted in an increase in strength due to the evolution of nanometric and coherent β-NiAl precipitates in a fine martensite structure. Moreover, they discussed the role of the dislocation networks that were remained in the microstructure and had a key effect in the improvement of the strength after the aging process without the annealing step.^[79] Zhang et al.^[275] reported a different behavior in mechanical strength of the heat-treated LPBF CX samples to those observed by Hadadzadeh. et al.^[79] They performed three heat-treatment cycles including solution annealing treatment (ST), aging (AT), and solution annealing and subsequent aging (ST + AT) on as-built CX samples fabricated by the LPBF technique. Their results indicated that solution annealing at 900 °C for 1 h results in a more homogeneous microstructure in which the microsegregation was diminished, and the retained austenite was transformed into martensite. Also, the dissolution of nickel and aluminum as the alloying elements in the matrix leads to the formation of supersaturated solid solution. Consequently, the solution annealing treatment resulted in a reduction of the strength; however, the aging process (AT) resulted in a partial conversion of the martensite to austenite, which had a detrimental impact on the strength of the alloy. In contrast to ST and AT, the solution annealing and aging process (ST-ST) led to the elimination of the retained austenite and the precipitation of nanoscale β-NiAl particles. Finally, the ST-AT resulted in a rapid increase in the tensile strength and the surface hardness of the LPBF CX samples as compared to the ST and AT processes.^[275] Shahriari et al.^[276] also studied the influence of traditional heat-treatment processes on microstructural evolution as well as, corrosion behavior of LPBF CX samples in which they compared the corrosion performance of this alloy with maraging stainless steel fabricated by a conventional process (420 stainless steel). It was found that after heat treating the LPBF parts, the recovery of solidification structure has occurred, and the cellular/fine subgrains are disappeared, as shown in Figure 41(a–c). They showed that the heat-treated LPBF SS CX has a superior corrosion resistance than the wrought martensitic tooling steel, as shown in potentiodynamic curves obtained from the samples exposed to 3.5 wt.% NaCl solution (Figure 41(d–e)). This behavior is associated with the absence of austenite and Cr-depletion zones because their presence in the wrought martensitic steel results in promoting pitting nucleation zones and destabilizing the passive layers.^[276]

Figure 41. SEM micrographs of the LPBF CX parts in (a,b) as-built, (c) heat-treated conditions, (d,e) cyclic polarization measurements of heat-treated LPBF CX, and quench-tempered 420 stainless steels, respectively (Reproduced with permission from^[276]).

2.2.2.3. Duplex stainless steels

The studies on duplex stainless steel (DSS) are limited and remained largely unexplored in AM processes. Ferrite and austenite are phases developed in these steels. Therefore, mechanical and corrosion performances of this type of steels are affected by the combined effects of the ferritic and austenitic structures. The ductility and strength of the dual-phase steels are higher than the ferritic stainless steels. Due to a higher amount of Cr and Mo, the corrosion resistance of DSS is higher than that of austenitic and ferritic mono-phase stainless steels.^[277,278] Despite having a similar concentration of Ni (austenite-stabilizing element) as austenitic stainless steels have, ferrite stabilizing elements, particularly Cr (24–26 wt.%) are added to stabilize ferrite in their microstructures. δ-ferrite is the primary phase formed in these steels after solidification; however, austenite is developed within the heat-treatment process.^[64] Hengsbach et al.^[279] studied the microstructural and mechanical performances of UNS S31803 duplex stainless steel fabricated by LPBF in both as-built and heat-treated conditions. Their findings showed that due to process-related high cooling rates within the LPBF process, the formation of austenite was almost suppressed. Therefore, they designed various annealing heat treatments at different temperatures between 900 °C and 1200 °C for a soaking time of 5 min to achieve a desired austenitic-ferritic microstructure. The microstructure of the as-built and heat-treated samples was compared (Figure 42), where the columnar grains oriented in the building direction (BD) are developed in the as-built specimens, as shown in Figure 42(a). Also, the volume fraction of ferrite was estimated to be about 99% for the as-built specimen, as seen in corresponding phase map (Figure 42(e)). They designated this to rapid cooling rates within the LPBF process in which the melt is entirely solidified into δ-ferrite according to the following sequences: Liquid → δ-ferrite + Liquid → δ-ferrite. Also, the formation of austenite and precipitates are almost suppressed during the solidification in the LPBF process.^[279] They also observed recrystallization occurred within the annealing process, as shown in Figure 42(b–d). Furthermore, secondary austenite in the post-treated conditions was detected using phase maps (Figure 42(f–h)).^[279]

Figure 42. (a–d) inverse pole figure maps, (e–h) corresponding phase maps of the as-built and solution annealed samples, and (i) stress-strain curves of the as-built and annealed S31803 DSS samples (Reproduced with permission from^[279]).

Tensile response of as-built and heat-treated samples displayed that heat treatment has a significant impact on mechanical properties (Figure 42(i)). The ferritic as-built specimens showed a higher tensile strength with a lower elongation to failure as compared to the heat-treated samples. The mechanical behavior of as-built sample was associated with the formation of a high density of dislocations and the nitride phase, which retards the movement of dislocations, enhancing the strength. Moreover, the elongation at fracture is significantly affected by the annealing process. Two main factors to decrease the elongation after annealing at low temperature are the formation of sigma phase and nitride precipitates. Also, it was revealed that the volume fraction of austenite was decreased after annealing at 900 °C. In addition, the volume fraction of retained austenite was significantly decreased with an increase in the annealing temperature, e.g., annealing at a temperature of 1200 °C led to a predominant reduction in volume fraction of austenite, resulting in the lower elongation at fracture. Conclusively, it was stated that desirable mechanical performance is obtained after heat treatment at temperatures in which a higher amount of retained austenite can be achievable.^[279] Davidson et al.^[280] studied the fabrication of a DSS component using the LPBF technique to optimize the process parameters. Besides, they investigated the role of the heat-treatment process on microstructural changes in the as-built samples.^[280] The designed heat-treatment process was annealing at temperature of 1040 °C for a soaking time of 1 h in a vacuum furnace followed by cooling at a rate of 1.8 °C/s, which is higher than critical cooling rate to avoid the formation of precipitates (about 0.8–0.9 °C/s). Their results indicated that the as-built samples contained ferrite as the main phase with a low volume fraction of austenite formed along the grain boundaries, or as Widmanstatten plates. Also, the heat-treatment process resulted in the formation of near-equilibrium duplex phases. After the heat-treatment process, an optimum austenite-ferrite ratio of 45.5:54.5 was obtained.^[280] In another study, Saeidi et al.^[187] revealed the σ phase as a precipitate formed in Cr-rich steels, which is detrimental on ductility. Due to the predominate presence of ferritic microstructure in the as-built DSS samples fabricated by the LPBF process, their yield and ultimate strength were much higher than those of the conventionally produced material. Also, a moderate elongation to failure of almost 8% was measured for the as-built samples in a ductile fracture mode. Also, the desired mixture of austenite and ferrite phases was formed after heat treatment at 1200 °C; however, besides these phases, σ phase was also precipitated.^[187]

2.2.3. Heat treatment of tool steels

2.2.3.1. Maraging steels

Maraging steels with low carbon content, e.g., grade or slight modifications of this alloy, are widely produced using AM and welding.^[281,282] EOS company has produced the powders of this type of alloy named as MS1 to be used in the AM process.^[283,284] Mooney et al.^[284] produced the 18Ni-300 steel alloy fabricated by the LPBF process and conducted different heat-treatment processes to estimate the effect of heat treatment time and temperature on plastic anisotropy and mechanical properties. They scheduled the aging treatment at different temperatures from 460 °C to 600 °C and different soaking times within 1–16 h. They designed a time-temperature contour map (Figure 43) by applying different heat-treatment processes on three samples, which were produced in three different AM build orientations of 0°, 45°, and 90°.^[284] Figure 43 clearly shows that the effect of aging parameters on the mechanical performance of LPBF MS1 parts. High strength and hardness are obtained after aging at temperatures ranging from 460 to 525 °C (Figure 43(a–d)); however, ductility (and toughness) are significantly decreased in this condition (see Figure 43(c)). Aging at temperatures above 525 °C resulted in an increase in ductility; however, a loss in strength and hardness was predicted due to the over aging and austenite reversion. Also, the contour plot obtained from their study depicts that ΔR (planar anisotropy) changes are less than 0.25 at 1.5% axial strain for all treatments (Figure 43(e)). In addition, planar anisotropy was minimized through under-aged to peak-aged condition. They proposed that the non-uniform transverse straining is mitigated due to blocking the gliding dislocations by stiff precipitates. Comparatively, a higher planar anisotropy was detected for the samples, which were overaged and austenitized. In this case, due to the formation of a higher amount of ductile austenite phase and fewer dislocations’ obstacles, ΔR deviates from zero.^[284] Moreover, their study showed that the aging process at 490 °C for 6 h was not the optimal aging treatment. Using their comprehensive experimental data and the constructed contour maps, it is suitable to select proper heat treatment parameters of time and temperature to achieve a desired strength-ductility-anisotropy for this alloy.^[284]

Figure 43. Contour curves of the LPBF MS1 samples mechanically tested (a) R_p0.2: yield strength; (b) R_m: tensile strength; (c) a_t: total % of elongation; (d) HV: Vickers hardness; and (e) ΔR: planar anisotropy estimated at 1.5% axial strain (Reproduced with permission from^[284]).

According to Yao et al.,^[285] the effects of different heat treatments on microstructure evolution and mechanical properties of DED MS1 parts were explored after 1 h of solid solution treatment at 830 °C and aging treatments at 490 °C for 1, 4, 7, and 10 h. It was found that with the extension of aging time, the directional columnar/dendritic crystals are transformed into martensitic laths and the number of stomas decreased considerably. Meanwhile, the microhardness and tensile properties of the forming parts are improved significantly after the heat treatments.^[285] Furthermore, the formation of precipitates in a martensitic matrix leads to high strength and toughness in maraging steels after the aging process. The precipitation sequence within the aging process of LPBF maraging samples is comparable to their conventionally fabricated counterparts.^[286–288] It includes spherical Ni₃X precipitates (η-phase) as the first state of precipitation (X: Ti, Al, Mo),^[81,289] which is followed by Fe₇Mo₆ (μ-phase) formation,^[81,289] as shown in Figure 44. Additionally, Santos et al.^[290] investigated the fatigue behavior of LPBF MS1 samples after the post-manufacturing heat-treatment process. They applied a heat-treatment process that included a very slow heating of the samples up to a temperature of 635 °C. Then samples were held at this temperature for 6 h, followed by a controlled cooling in the oven for 3 h to reach 360 °C. Finally, samples were air-cooled to room temperature. This heat treatment was selected based on the standard heat-treatment process applied to molding parts in mold industry.^[290]

Figure 44. Atom probe tomography of the precipitates Ni₃Ti (η type) and Fe₇Mo₆ (μ phase) formed in the LPBF MS1 maraging steel aged for 2 h at 510 °C (Reproduced with permission from^[80]).

They displayed the impact of heat treatment on the crack growth of LPBF MS1 samples after applying tensile overload under a constant loading amplitude and plotted da/(dN-ΔK) curves (Figure 45). The observation of a stable crack propagation, i.e., before overloading, was a beneficial effect of the heat treatment. After overloading, a retardation effect similar to those observed in cast materials was also obtained. This retardation was usually associated with the crack closure effect. Figure 45(a) shows that the heat-treated LPBF specimen has a lower retardation effect than that of the as-build material due to having a lower value of ΔK baseline. The influence of the number of overload cycles on the transient crack growth behavior in the heat-treated samples is also presented in Figure 45(b). Their study stated the heat treatment led to main changes on both microstructure and crack path and increased the resistance to the propagation of the fatigue cracks.^[290]

Figure 45. Plots of da/dN vs. ΔK at different (a) heat treatments, and (b) number of overload cycles (Reproduced with permission from^[290]).

2.2.3.2. Carbon-bearing tool steels

Carbon-bearing steels are divided into a board category of hardening and case hardening steels. Hardened steels include almost 0.8–1.2 wt.% carbon, where the content of carbon is uniform through all sections of these steels. This type of steels is hardened by quenching and tempering. Comparatively, the content of carbon in the case-hardened steels is low, typically 0.1–0.4 wt.%. The heat-treatment process for the case-hardened steels includes a process in which the steel is exposed to high temperatures of 870–980 °C. This heating process is performed in an atmosphere, which can diffuse back carbon atoms to the steel’s surface. Consequently, the steel is carburized and after quench-tempering process, a core/shell structure with a hard surface “case” as shell and softer bulk as core can be achieved with hardness of 58–64 HRC and 25–50 HRC, respectively.^[291] M2, HS 6-5-3-8, X65MoCrWV3-2, H11 and H13 are the most applicable bearing tool steels for AM processes.^[292–294] Among these alloys, the tool steel H13 attracts more attention for AM processing, so the heat-treatment processes of this alloy produced by AM process is given in this section.

Åsberg et al.^[295] showed that the microstructure of directly solidified colonies of prior austenite in the LPBF products disappears after standard hardening-tempering (SR + HT), and hot isostatic pressing (SR + HIP + HT) treatment. They applied the first post-treatment cycle, which was isothermal heating at a temperature of 650 °C for a soaking time of about 8 h to eliminate residual stresses. After that, the specimens were austenitized at 1020 °C for 70–75 min, followed by nitrogen quenching and then secondary tempering at temperature of about 585 °C for 2.25–3 h. After this heat treatment, specimens were named SR + HT. Also, they designed the other post-treatment cycle, which was the HIP process for 6 h at 1130 °C and 100 MPa pressure after the SR process. Then HT was conducted in the same way as for the SR + HT. The microstructure of the SR sample is presented in Figure 46(a). The micrograph showed the evolution of prior austenite grains as elongated with carbides. Also, a high-volume fraction of carbides with different morphologies is revealed inside the grains. After the SR + HT process, the microstructure of the samples is changed (Figure 46(b)), in which the dissolution of the prior austenite grain boundaries occurs. Therefore, the microstructure is solely tempered martensite. In addition, fine round-shape carbides are distributed along martensite needle boundaries. The SR-HT and SR + HIP + HT specimens show a similar microstructure, so it is concluded that HIP does not change the microstructure remarkably (Figure 46(c)).^[295]

Figure 46. SEM micrographs of the samples subjected to different post treatments of (a) SR, (b) SR + HT, and (c) SR + HIP + HT, (d) large porosity in SR sample (Reproduced with permission from^[295]).

According to their study, it was mentioned that the lack of fusion and spherical gas porosity in SR and SR + HT specimens led to a remarkable scatter in the elongation values to break. The HIP process also resulted in reaching the highest strength values in comparison with those obtained for the H13 samples conventionally fabricated and heat-treated in the same way. This increment in strength after the HIP process was attributed to the elimination of porosity and lack of fusion defects.^[295] Deirmina et al.^[296] studied the microstructure and mechanical properties of H13 steel fabricated using the LPBF technique and then heat treated through direct tempering and quench/tempered conditions. In their research, two types of heat-treatment processes were considered including austenization at 1020 °C for 15 min, quenching to room temperature followed by tempering at different temperatures (i.e., 450 °C, 500 °C, 550 °C, 600 °C, and 650 °C) for 2 h, and subsequent quenching. Also, direct tempering without austenization step at same tempering temperatures was carried out.^[296] Their results showed that a partial recovery of the solidified microstructure occurred within the austenitizing treatment for 15 min at 1020 °C. The observation of an equiaxed martensitic structure after the quenching process confirmed partial recovery/recrystallization phenomenon. Quenching caused an evolution of a martensitic microstructure with less than 2% retained austenite (RA). In addition, the decomposition of retained austenite and precipitation of carbides occur within direct tempering at temperatures above 550 °C. Decomposition of a large volume fraction of retained austenite during the direct tempering led to a shift in secondary hardness peak in heat-treated samples to a higher temperature as compared to the samples that were austenitized and tempered. This feature elucidates the effect of direct tempering on the LPBF H13 samples, which is breakthrough for the applications under harsh environment where high hardness is required.^[296] Observations of retained austenite decomposition within the post-heat treatment at temperature of 500 °C and elimination of the cellular microstructure after tempering at temperatures above 600–700 °C were reported by Mazumder et al.^[297] and Krell et al.^[298] Their study indicated that the austenization-quenching caused the formation of a martensitic structure without retained austenite, which is very similar to that observed in conventionally produced materials subjected to austenitizing and quenching process (Figure 47).^[298]

Figure 47. SEM micrographs of LPBF H13 samples after various heat-treatment processes of (a,b) tempering processes at 600 °C and 700 °C, respectively, and (c) austenitizing at temperature of 1040 °C followed by quenching in oil (Reproduced with permission from^[298]).

2.2.4. Heat treatment of ferritic steels

The primary phases in the oxide dispersion strengthened (ODS) steels are ferrite and nanoscale particles of yttrium oxides dispersed in the matrix of the alloy. These steels are thriving structural material for high-temperature applications where, high tensile and creep strength are required.^[298–300] Although this alloy is produced mostly by powder metallurgy, controlling the nanoparticle dispersion and size is important as it influences the performance of the final product.^[301] Consequently, a fabrication method consists of colloidal nanoparticles fragmented by laser fragmentation in liquids (LFL) technique, followed by pH-controlled dielectrophoretic supporting of the steel powder is suggested to better control the nanoparticle dispersion and size.^[302] In some cases, micro-size powder of ODS is produced for DED and LPBF techniques of manufacturing.^[303] The commercial alloy PM2000 (Fe–19Cr-5.5Al-0.5Ti-0.3Y₂O₃), or the close family of that (MA956) is the most attractive ODS alloys in AM processing. Walker et al.^[60] reported the first study on LPBF processing of ODS parts, while another study used the same manufacturing technique on mechanically alloyed PM2000 or MA956 powders.^[207] The microstructure of ODS alloys was similar after the fabrication process. They contain a fully ferritic matrix with a strong fiber texture along the <001> direction that is parallel to the build direction, and fine oxides uniformly distributed in the matrix. In some cases, due to the agglomeration of nanoscale oxide particles, coarser reinforced particles are also observable.^[207,304] Y₂Ti₂O₇ and sometimes Y₄Al₂O₉ are reported as the nanoscale particles dispersed in the matrix.^[83] HIP process is one of the main post-treatment processes applied to this type of alloys to control the structure and composition of oxides. Shi et al.^[305] fabricated an ODS Fe-9Cr alloy with a nominal chemical composition of Fe-9Cr-1.5W-0.3Ti-0.3Y (wt.%) using laser engineered net shaping (LENS) at various laser power. To significantly reduce the size and density of micropores, and refine the grain structure, they recommended HIP as the post-treatment process for the as-printed ODS alloy fabricated at low laser power. Also, the HIP process was applied to increase the density of nanoscale oxides (Y₂TiO₅ and Y₂Ti₂O₇). All these changes led to a significant increase in tensile properties, where the mechanical properties of the AM components were comparable to those fabricated by conventional techniques.^[305] They presented the microstructure of the ODS alloys fabricated using different laser power (Figure 48). The figures showed that the microstructure of alloys is martensitic and some micropores (dark spots) with a size ranged between 10 and 200 μm are also detected. Also, they showed that with an increase in laser power, the density and size of micropores decreased as shown in Figure 48(a,b), which agrees with results obtained from the relative density calculations. For the ODS sample fabricated at low laser power + HIP, no obvious micropores were detected. This was due to the HIP processing at 1373 K, and pressure of 200 MPa for 2 h.

Figure 48. OM images of ODS samples produced at different laser powers of (a,b) 1200 W, (c,d) 1600 W, (e, f) 2000 W, and (g, h) 1200 W – HIP. TEM micrographs of the oxides in ODS alloy produced using various laser powers of (i) 1200 W, (j) 1600 W, (k) 2000 W, and (l) 1200 W – HIP (Reproduced with permission from^[305]).

The distribution, morphology, and composition of the reinforced oxides in the LPBF + HIP ODS samples are illustrated via TEM analysis (Figure 48(i–l)). A high density of nanoscale oxides is seen in the sample fabricated at low power density (1200 W) (Figure 48(i)). However, the density of oxides significantly decreases when the laser power increases from 1200 to 2000 W, as shown in Figure 48(i–k). Compared with the sample fabricated at low laser power density, the area density of nanoscale oxides in the samples fabricated at low laser power density + HIP increases, while the average size is reduced. This behavior is attributed to re-precipitation of finer oxides during HIP (Figure 48(l)).^[305] Boegelein et al.^[304] also investigated the mechanical response and deformation mechanism of ferritic ODS samples that contained a fine dispersion of nanoscale Y(Al, Ti) particles. They used the LPBF technique to consolidate as-mechanically alloyed PM2000 powder (Fe–19Cr–5.5Al–0.5Ti–0.5Y₂O₃, wt.%), and produced solid and thin-wall builds at various thicknesses. Their findings stated that the yield strength of as-grown thin-wall builds is lower than that of conventional PM2000 alloy (recrystallized), but 1 h post-build annealing treatment at a temperature of 1200 °C significantly increases the yield strength of the as-built sample. They reported an increase in yield strength (YS) of the annealed ODS samples due to the evolution of a fine dispersion of ODS particles come from remaining Y(Al, Ti). A higher number of fine dispersoids acts as an obstacle against dislocations’ mobility and subsequently, increases the YS. Annealing (post-build or during deposition) is also known to change the particle morphology from multiphase particles to predominantly single-phase Al–Y oxides.^[304]

2.3. Summary

The microstructure and mechanical properties of an additive manufacturing (AM) component are significantly impacted by various processing parameters like scanning speed, laser power, gas flow rate, and powder feed rate. In laser additive manufacturing (LAM), the solidification process is non-equilibrium and complex due to factors such as high temperature gradient, rapid solid/liquid interface velocity, capillary effects, recoil pressure, surface tension-driven instability, and evaporative and radiative cooling in the dynamic melt. As a result, microstructural characteristics like grain size, morphology, orientation, and phase fraction can exhibit significant variations depending on alloying elements, AM processing techniques, and even within different sections of a component. The shape and geometry of the melt pool, which also influence microstructural features, are subject to uncertainties and complexities during AM processing. Thus, it is essential to comprehend the principles of rapid solidification, phase transformation phenomena, and heat/mass transfer in the solidifying melt pool to establish a meaningful relationship between microstructure and mechanical properties of AM products. This section provided a more detailed examination of the microstructure in both as-built and heat-treated AM steels and iron-based alloys. To this, the “as-built” microstructure can be analyzed in terms of grain orientation, size, morphology (columnar, dendritic, equiaxed), solidification pattern (planar, non-planar), phases, porosity, and microsegregation. These features are influenced by thermophysical factors like diffusion coefficient, interatomic potential, temperature gradient, and solid/liquid interface velocity. Rapid solidification during AM leads to a fine anisotropic microstructure with elongated columnar grains, but reheating cycles can produce fine equiaxed grains instead. Microsegregation occurs within a narrow intergranular area, with most alloying elements being trapped in the solid lattice. Defects such as lack of fusion, gas porosities, and shrinkage porosities depend on the AM process parameters. Additionally, AM products, particularly those printed vertically, experience significant residual stresses that can induce partial recrystallization during post-process heat treatment. Industrial steel products commonly undergo post-process heat treatment to alleviate residual stress, homogenize the material, and enhance strength through solutionizing/aging treatments. Different steel grades exhibit variations in phase formation, tendencies for hardening, and segregation, underscoring the importance of understanding the primary microstructure, thermal history, and applied heat treatment cycles. This section provided further explanations of these factors, which ultimately impact the corrosion resistance of AM steels.

3. Corrosion of additively manufacturing steels

3.1. Introduction

A persistent issue in many industry sectors is the corrosion of metallic components, which causes considerable expense. Applying paints, using corrosion resistant alloys, the repairing and replacing of components result in increased costs.^[306] A solution to decrease the additional costs is the selection of a suitable and economical corrosion prevention strategy.^[306] In such a context, additive manufacturing has a role to play, with unique advantages in the fabrication of metallic components in comparison with conventional fabrication methods. For example, the local thermal cycles, which occur within the AM processes, can cause the evolution of unique microstructural features with fine grain structures.^[307,308] The corrosion mechanism and performance of the AM components are affected by the formation of un-melted powders, micro-cracks, porosities, and balling. Consequently, the unique microstructural features determine the lifetime of the metallic structures.^[309] Nowadays, the different types of stainless steels are the most applicable metallic alloys in the various parts of the industry. The formation of a passive layer results in stainless characteristics of the stainless steels. This chromium-rich oxide layer is an invisible and adherent film, which can heal itself in the presence of oxygen. Other elements such as nickel, manganese, molybdenum, copper, titanium, silicon, niobium, aluminum, sulfur, and selenium are added to the chemical composition of stainless steel. Moreover, the carbon content is kept in a range of less than 0.03% to almost 1.0% in the specific grades.^[310] Laser powder bed fusion (LPBF) as one of the AM techniques is most frequently used to fabricate the stainless-steel components.^[311] Although a high efficiency of the material utilization and fabrication of components with the complex geometries are obtained by the LPBF, the corrosion behavior of the AM stainless steel components is still unknown, which needs in-depth studies.^[312] Different types of stainless steels are produced by LPBF techniques such as 316 L austenitic stainless steel, 420 martensitic stainless steel, maraging stainless steels of 15-5 PH, 17-4 PH, and Corrax.^{[78,254,271,313–319]} In addition to more applicable stainless steel, few studies of the AM processes have been performed on austenitic-ferritic stainless steels, twinning-induced plasticity (TRIP/TWIP) steels, and oxide dispersion strengthened (ODS) steels.^[64]

3.1.1. Theoretical considerations of corrosion behavior of stainless steels

The passivity of stainless-steel results in their useful corrosion resistance in different environments. Figure 49 shows the general form of anodic polarization curves of stainless steel in acidic solutions. If the cathodic branch of the system intersects the curve at passive region, the stainless steel is passive, and the film should be healed even if damaged. This presents a situation in which stainless steel can operate safely. Once the cathodic curve intersects the active region and transpassive region between the anodic and passive regions of the curve, the passivity is unstable and any break in the film would result in a rapid metal solution. It is well-known that the cathodic reactions occur at lower potentials in the acidic solutions in comparison with those of oxidizing solutions, which consequently create a favoring condition for the corrosion. The oxidizing solutions are favoring for passivity and formation of the passive layer. However, if the cathodic curve intersects the anodic branch at the area of transpassivity, the passive film is rendered unstable by oxidation. This type of curve is a convenient method to estimate the corrosion behavior of stainless steel in different media and study the effects of steel composition, heat treatment, and microstructure on corrosion resistance.^[320,321] Also, the presence of halide ions in the solution can result in an intersection at a lower potential than the oxygen evolution potential (as shown in Figure 49), as named the breakdown potential or the pitting potential. The pitting potential is not an absolute value, and it depends on different parameters such as the environment. It is well known that bromide and chloride are the most aggressive halides, which affect the pitting potential. The amount of any oxidizing agents such as dissolved oxygen as well as, halide salts (e.g., ferric chloride, cupric chloride) are also important parameters, which affects the passivity of the stainless-steel components.

Figure 49. A scheme of cathodic, anodic, passive and transpassive regions used to identify the localized corrosion parameters (Reproduced with permission from^[321]).

In terms of chemical composition, different types of alloying elements affect the pitting corrosion resistance of stainless steels and the effect of a number of alloying elements are expressed by the “pitting resistance equivalent number” (PREN) as follows:^[320] (26) $$\mathit{PREN} =\mathit{\text{ \% Cr }}+ 3.3\mathit{\text{ x \% Mo }}+ 16\mathit{\text{ X \% N}}$$(26)

An improvement in the corrosion resistance of stainless steels’ is obtained by adding molybdenum and nitrogen as alloying elements.

3.1.2. The passive state

The formation of an interfacial solid layer on the surface of metallic components, which occurs spontaneously or due to a anodic driving force, is known as passivity. The interaction between the environment and metallic substrate is limited by the formation of a thin and protective passive layer. The stability of many metallic structures depends on their state of passivity. The passive layer forms with a thickness in the range of ∼1–10 nm and are created by oxidation of the surface. The corrosion rate of metallic components decreases with the formation of the passivating oxide film. In the absence of a passive layer, many metallic components will rapidly corrode. Furthermore, corrosion may occur if passivity is disrupted, of the natural regeneration of the passive layer is hindered due to the formation of pits or cracks.^[320] The formation of a chromium-rich oxide on the surface of stainless-steel results in high corrosion resistance across a broad range of pH and environments. Although the thickness of the passive layer formed on stainless steels is less than 5 nm, this layer is strongly adherent, chemically stable, and protective. The passive layer, which forms on the surface is susceptible to pitting in a certain medium, especially in solutions containing chloride ions in which this oxide layer can breakdown, and lead to the local dissolution of the substrate.^[322] The passive film formed on stainless steel displays a bi-layer structure that contains an outer layer of hydroxides and an inner layer of enriched chromium and iron oxides.^[323–326] The thickness and composition of the passive layers in the form of the oxide and the hydroxide depends on the chemical composition of the alloy and has a close correlation with the external factors such as the chemical potential for passivation, the solution’s nature (mainly pH), and the heat treatment processes applied on stainless steel components.^[322] In addition, the passive layers formed on stainless steels are ascribed to semiconductors, which generally consist of an inner layer as p-type and an outer layer in form of n-type. The p-type semiconductivity is associated with the presence of Cr₂O₃ and cation vacancies or oxygen surplus in the inner layer, and the n-type semiconductivity is related to Fe⁺² ions in interstitial zones and to some oxygen vacancies or hydrogen atoms trapped in the outer layer in some cases.^[327–330]

3.1.3. Localized forms of corrosion

Localized attacks in stainless steels refer to specific types of corrosion that occur in localized areas rather than uniformly across the entire surface. These attacks can lead to the formation of pits, cracks, or other forms of damage. Localized forms of the corrosion occurred on the stainless steels are also categorized as pitting, crevice corrosion, intergranular corrosion and also environmental cracking.^[331]

3.1.4. Pitting corrosion

Pitting is the common type of corrosion on stainless steels in acidic or halide containing media and has been the most intensively studied to date. In general, halides such as chloride, bromide, and iodide are the primary ions that accelerate pitting corrosion on stainless steel. However, there are a few other ions that can also contribute to pitting corrosion to some extent such as sulfate ions (SO₄⁻²) which can promote pitting corrosion on stainless steel, particularly in the presence of chloride ions. Also, thiocyanate ions (SCN⁻) at higher concentrations could accelerate pitting corrosion on stainless steel. Nitrate ions (NO₃⁻) have generally a passivating effect on stainless steel, while at elevated temperatures and in the presence of other aggressive ions, they can contribute to pitting corrosion.^[320] The corrosion of AM steels remains debated in some respects, as a result of the fact that many studies on corrosion of AM steels have studied steels fabricated on different instruments, with different post-processing, and in different environments. The corrosion of AM stainless steels was recently review by Schneider et al.,^[332] who noted that some trends across studies – as more studies emerge – are becoming apparent. As a consequence, a consolidated review such as that herein looks further to seek unified trends from the merging literature.

Much like conventional steels, the corrosion of AM steels is a function of the environment and loading conditions, such that crevice corrosion, stress corrosion cracking, inter-granular corrosion, and corrosion fatigue, are all possible – which significantly affects the service life of the industrial structures.^[333,334] The formation of microscopic holes/cavities on the surface of metals/alloys due to microstructural heterogeneities of the surface and/or the localized deterioration of the protective passive layer is known as pitting. The pitting happens on alloys, which are also able to create a stable self-healing, thin, and adherent oxide film. Pits form as holes in hemispherical or polygonal shapes on the passivated surfaces due to localized dissolution. Some anions, such as halides (e.g. Cl⁻) serve as aggressive agents in the media and results in localized attack of the passive layer. The diameters and depths of pits could range from very small to larger, which are also filled with corrosion products. Small pits may have diameters in the range of micrometers (µm) to millimeters (mm), while larger pits may have diameters in the range of millimeters to centimeters. Similarly, the depth of pits can vary widely. It can range from a few micrometers to several millimeters or more, depending on the severity and duration of corrosion. The localized corrosion-induced pits can act as the stress concentration zones and helps with nucleation and propagation of stress corrosion or fatigue cracks. Therefore, the formation of these pits decreases the service life of the stainless-steel components. It is estimated that ∼11% of the failure of industrial metallic parts is estimated to be due to pitting corrosion;^[334] however, there is a more general requirement to the importance of understanding this phenomenon – because pitting is the initial form of dissolution that evolves to propagation in the form of inter/intra-granular corrosion, SCC or corrosion fatigue.

3.1.5. Metallurgical variables and pitting corrosion

The susceptible regions to localized corrosion are known to be any heterogeneities such as grain boundaries, mechanical scratches, and slip steps emerging at surfaces. Furthermore, the alloying elements in solid solution, the presence of the second phases such as sigma, chi, manganese sulfide, carbides, etc., significantly affect the passivity and resistance to pitting of the conventional fabricated stainless steels parts.^[334]

Grain boundaries in nitrogen-bearing austenitic stainless steels were preferential sites to pits’ nucleation due to the precipitation of complex carbides, resulting in depletion of chromium and other alloying elements in this region.^[335] Furthermore, a detailed study of the sensitized microstructure of the stainless steel indicated that intergranular corrosion is a consequence of pitting corrosion.^[336] It was also reported that a reaction between carbon and chromium can happen during the exposure of the stainless steel to high temperatures of 425–875 °C through the welding or fabrication process. The precipitation of carbides enriched in chromium such as (Fe,Cr)₂₃C₆ takes place preferentially at the grain boundaries, which causes the depletion of chromium in these areas. Consequently, the areas adjacent to the carbide precipitates are prone to pitting and an increase in corrosion rate.^[337]

Inclusions in stainless steels are typically oxides and sulfides and the literature reporting the initiation of localized corrosion at inclusions is available.^[338–340] Manganese sulfide inclusions present in commercial stainless steels are often attributed to pitting. The sulfide inclusions are prone to pitting because of their high conductivity, which results in adsorption of chloride ions and facilitation of anodic dissolution.^[339] Furthermore, the corrosion potential of a passive stainless steel in an aqueous solution containing chloride is between 0 and 200 mV (SHE³) in which the sulfide inclusions are thermodynamically unstable and have a tendency to dissolution.^[338] Therefore, when the surface of the metal is exposed to a solution containing aggressive agents such as chloride ions, the sulfide inclusions can be dissolved, and depending on the dissolution and mass transfer condition, the pits growth can continue. Regarding pitting adjacent to inclusions, three different stages are suggested for nucleation and growth of pits: (i) dissolution of active inclusion and initiation of micro-cavities, (ii) incubation stage, which is accompanied by the accumulation of chloride ions at, and (iii) pits’ nucleation and growth at the cavities.^[338]

3.1.6. Crevice corrosion

Crevice corrosion, similar to pitting corrosion, begins when the protective oxide layer of stainless steel breaks down, resulting in the formation of shallow pits. However, unlike pitting corrosion, crevice corrosion takes place in narrow gaps or crevices rather than on the exposed surface. In fluid systems, these crevices are commonly found between tubing and tube supports or clamps, between adjacent tubing sections, and underneath accumulated dirt and deposits. It is challenging to completely eliminate crevices in tubing installations, and tight crevices pose a significant threat to the integrity of stainless steel. Crevice corrosion occurs when seawater penetrates into a crevice, creating a chemically aggressive environment where corrosive ions are trapped and cannot easily escape. In such situations, the entire surface within the crevice can corrode rapidly. It is important to note that crevice corrosion can occur at lower temperatures compared to pitting corrosion because it requires less effort to initiate a “pit” within a geometric crevice.^[341]

3.1.7. Intergranular corrosion (IGC)

Intergranular corrosion refers to the selective dissolution that occurs along the boundaries between grains in a material. This type of corrosion is influenced by the precipitation of particles at the grain boundaries and the depletion of solutes near these boundaries. A well-known example of intergranular corrosion is observed in stainless steels, where chromium-rich carbides precipitate within a temperature range of 450–850 degrees Celsius (commonly found in the heat-affected zones of welded joints) along the grain boundaries. Consequently, chromium-depleted zones form in the vicinity of these boundaries. Since chromium is the crucial alloying element responsible for the passivation ability of iron-chromium alloys, the chromium-depleted zones become active and preferentially dissolve. The remaining passive surface of the material acts as a large cathode, driving the dissolution of the grain boundary zones.^[341]

3.1.8. Environmental cracking

Environmentally induced cracking corrosion of stainless steels refers to the phenomenon where stainless steel materials undergo cracking in the presence of specific environmental conditions. This type of corrosion occurs due to a combination of factors such as the presence of corrosive agents, mechanical stress, and material susceptibility. There are several types of environmentally induced cracking corrosion that can affect stainless steels, including:^[320]

• Stress Corrosion Cracking (SCC): SCC is a form of cracking that occurs when stainless steels are exposed to a corrosive environment and are under tensile stress. It typically happens in the presence of chloride ions, such as in marine or industrial environments. SCC can propagate rapidly and lead to catastrophic failure of the material.

• Chloride Stress Corrosion Cracking (Cl-SCC): Cl-SCC is a specific type of SCC that is caused by the combined effects of tensile stress and chloride ions. It is particularly common in austenitic stainless steels and can occur at relatively low temperatures.

• Hydrogen-Induced Cracking (HIC): HIC, also known as hydrogen embrittlement, occurs when stainless steels are exposed to hydrogen gas. Hydrogen can permeate into the material and cause the formation of internal cracks, leading to brittle behavior and potential failure.

• Sulfide Stress Cracking (SSC): SSC is a type of cracking that occurs when stainless steels are exposed to environments containing hydrogen sulfide (H₂S). The presence of H₂S, along with applied or residual tensile stress, can lead to the initiation and propagation of cracks in the material.

3.1.9. Synergetic effects of different types of corrosion

Stainless steel can be susceptible to various types of corrosion, and in certain scenarios, multiple types of corrosion can occur simultaneously or sequentially. Some combinations of different types of corrosion that can affect stainless steel. pitting corrosion is the localized corrosion that forms small pits on the surface of stainless steel, while crevice corrosion occurs in confined spaces or crevices. These two types of corrosion often occur together because crevices provide an ideal environment for pitting to initiate and propagate. The corrosive attack can start with crevice corrosion and then progress to pitting corrosion, causing localized damage to the material. Also, SCC, which results from the combined effects of tensile stress and a corrosive environment, can occur concurrently with pitting corrosion. Tensile stress can promote the initiation and propagation of pits, leading to localized cracking in the presence of a corrosive medium. The combination of SCC and pitting corrosion can significantly compromise the integrity of stainless steel. Moreover, intergranular Corrosion (IGC) is the preferential corrosion along the grain boundaries of stainless steel, often caused by the precipitation of carbides or other phases at the boundaries. In the presence of a corrosive environment, the grain boundaries susceptible to IGC can also become sites for pitting corrosion initiation. The combined action of IGC and pitting corrosion can lead to severe deterioration of the material. Also, the combinations of erosion corrosion and pitting could further exacerbate the damages on the surface of stainless steels’ components. Erosion corrosion occurs when the surface of stainless steel is subjected to the combined effects of fluid flow and corrosion. In areas where fluid flow is turbulent or impinging, such as in piping systems, erosion corrosion can cause material loss and surface damage. Pitting corrosion can further exacerbate the damage, as the already weakened surface is prone to localized pitting attack.^[320]

3.2. Corrosion of stainless-steel alloys fabricated by LPBF

This section reviews the electrochemical behavior of stainless steels fabricated by the LPBF technique. Different types of stainless steels are fabricated using the LPBF as one of the AM techniques, where their corrosion performances are comparable with the ones conventionally produced. However, the breakdown of the passive layer in form of pits on the surface of stainless steel is one of the main problems in these alloys. The microstructure of stainless steels produced by LPBF is distinguished from the counterparts conventionally fabricated due to the formation of pores, subgrains entangled with the dislocations, and fine inclusions. Also, the corrosion behavior of them has a close correlation with their microstructures. The estimation of the service life of LPBF stainless steels is one of the important measurements undertaken in different sectors of industry. In this regard, a study on the effect of the unique microstructural features developed within the LPBF process on the passivation and pitting corrosion is important to achieve an estimation of the service life for the LPBF stainless steel components.

3.2.1. Role of pores

The defect formation is still a common problem in AM processing, and the presence of defects such as pores are considered to affect the corrosion behavior of AM samples. However, the lack of direct experiments is a barrier to validate the role of defects in nucleation and propagation of corrosion pits.^[342] The pores induced within the LPBF are divided into two groups including those that are around the un-melted powders, and others that are created by trapped gas inside the powder during the gas atomization process.^[309] The process parameters of the LPBF technique, such as laser energy, scanning rate, and scanning direction can control the porosity levels in LPBF parts. Increasing laser power and decreasing scanning rate were reported to sufficiently decrease the porosity of the 316 L stainless steel fabricated by LPBF.^[343] The highest metastable pit frequency and highest cumulative charge as indexes of low resistance to pitting correspond to the as-built specimens with the highest percentage of porosity. Also, LPBF samples with the lowest level of porosity show the lowest metastable pit frequency as well as, the lowest cumulative charge, which are presented in Figure 50.^[343] Sander et al.^[343] declared that the presence of sub-surface pores in LPBF 316 L stainless steel components and a relatively rougher surface impede the repassivation on stable pit, which is nucleated within pores. Duan et al.^[344] reported that pitting corrosion of LPBF 316 L stainless steel initiates in gas pores, where transformation of metastable pits to stable pits easily occurs in gas pores with metal covers in comparison with open pores. They have indicated a pitting mechanism induced by gas pores, schematically shown in Figure 51. Two types of gas pores are defined based on three geometrical parameters of a pore: “r” is the cross-sectional radium, “R” is the spherical radium, and “h” is the depth of the spherical cap, as shown in Figure 51(a). According to the definition, the spherical-shape cap is defined as an open gas porosity if h < R (Figure 51(a,b)), and a closed gas porosity if h > R (Figure 51(c,d)).

Figure 50. Cumulative charge of metastable pit vs. the frequency of the metastable pit (sample 4 has the highest fraction of porosity, while sample 7 has the lowest value. Wrought sample has no porosity) (Reproduced with permission from^[343]).

Figure 51. A scheme of the effect of gas porosity on the pitting mechanism in the LPBF 316 L stainless steel (Reproduced with permission from^[344]).

H⁺ and Cl⁻ ions are the main agents required to the aggressive pit media, where the hydrolysis of the metallic cations results in creating H⁺. Also, the anions such as Cl⁻ moves into pits to maintain electroneutrality. The electrolyte’s harshness inside the pits is assessed by the concentration of the metallic cations at the surface of the pit (C_surf). Therefore, the pitting nucleation stage is related to the moving of metallic cations from the surface of the pit to the pit interior shown in Figure 51(b). Also, the nucleation stage of pitting is followed by the formation of pits in a metastable state (Figure 51(c)). Accordingly, the presence of metallic cations results in a decrease in pH, leading to a reduction in the thickness of the passive film. In other words, a reduction in the passive film’s thickness helps to increase the cation diffusion rate. With increasing accumulation of metal cations and decreasing pH in the gas pore, the cation concentrations reach the critical value in which the thickness of the passive layer approaches zero. The process continues with starting the active dissolution of metal rather than the passive layer’s dissolution at the gas pores located on the surface. Therefore, the active dissolution of the matrix and the movement of metallic cations control the rate of pits’ growth. When the shape of a gas pore converts to a hemisphere, (h > R) or (h ≤ R), the active dissolution stage becomes more severe, as shown in Figure 51(d,e).^[344]

Another study of the pitting mechanism in LPBF 316 L stainless steel was performed by Geenen et al.^[345] Their studies showed that the porosity induced during the LPBF process is responsible for lower corrosion resistance of the LPBF components in comparison with LPBF parts that are post-treated by hot isostatic pressing (HIP).

They described the corrosion mechanisms of stainless steels produced via powder metallurgy in various stages (Figure 52), which are similar to those for LPBF stainless steel components. According to their description, first, the formation of a stable passive film on the surface of steel and the pore surface (Figure 52(a)). This passive layer is subsequently dissolved, and the dissolution begins toward the interior of the pore (Figure 52(b)). Secondly, the dissolution progresses toward the inside of pores due to the formation of a microscale active-passive cell, (Figure 52(c)). Subsequently, a general disintegration of the material occurs, shown in Figure 52(d). They claimed that HIP post-processing after the LPBF process led to worse corrosion resistance due to an increase in grain sizes, and the formation of former lamellar oxides, which were spheroidized.^[345]

Figure 52. Corrosion mechanisms of stainless steels produced via powder metallurgy: (a) initial state with an intact passive layer on the pore’s surface, (b) beginning of the dissolution, (c) further dissolution as active-passive cells are formed, and (d) further progression of dissolution (Reproduced with permission from^[345]).

The effect of process-induced pores on pitting corrosion of LPBF 316 L stainless steel and the role of building direction were also studied by Kazemipour et al.^[346] The effect of process-induced porosity along two different orientations of the as-built samples was investigated (planes parallel and perpendicular to build direction). Their research showed that the pitting potential and corrosion behavior of the LPBF samples were not affected by the process-induced porosity during the initial immersion times, although the electrochemical results revealed that the process-induced porosity had a more detrimental impact on reduction of the electrochemical stability through the surface of LPBF 316 L. The formation of crevice-like porosity with larger size and higher density was associated with its low electrochemical stability. SEM micrographs taken from the surface plane perpendicular to building direction (Top sample) and immersed for a long time (20 days) confirmed that the remaining porosity are the preferential pitting spots (Figure 53(c,d)). The process-induced pores showed a larger size and more crevice-like shape on the top side with respect to the side plane (Figure 53(a) and (b)). Furthermore, a few pits separated from each other inside the melt pools were corresponded to oxide inclusions detected interior of the pits, as shown in Figure 53(e). According to their findings, oxide inclusions can be the main cause of pitting in LPBF components.

Figure 53. SEM micrographs of the surfaces in the LPBF 316 L stainless steel immersed into a solution at different times: (a) surface perpendicular to build direction (top sample) after a short time of immersion, (b) the surface parallel to build direction (side sample) after a short time of immersion, (c) the top sample after 20-day immersion, (d) the side sample after 20-day immersion, and (e) EDS maps of an area Pitted (Reproduced with permission from^[346]).

The effect of various laser scan speeds on void evolution during solidification of LPBF 316 L stainless steel components and its consequent impact on corrosion behavior were studied by Xiaoqing et al.^[347] Their study showed that the fraction of porosity increases at higher scanning speeds. Also, the presence of silicon and un-melted powders was detected inside the porosity, which has an adverse impact on the pitting corrosion resistance of LPBF parts. In other words, with an increase in scanning speed, the susceptible sites to pitting increase, and porosity acts as a suitable site for nucleation of the pits. Also, their results indicated that the formation of inclusions decreases during the LPBF process, which has a correlation with the pitting resistance. The pitting potential of the as-built 316 L stainless steels was almost 300 mV more than that of the heat-treated sample as a result of the inclusions’ modification during the LPBF process.

Laleh et al.^[348] have investigated the pitting resistance of LPBF 316 L stainless steel to estimate its erosion-corrosion behavior. According to their research, the alloy possessed a superior pitting corrosion resistance in comparison with that of a conventionally fabricated counterpart. However, unexpectedly, the LPBF samples had a lower erosion-corrosion resistance than those of the commercial 316 L stainless steel. This behavior is associated with the weaker ability of LPBF components to re-passivation, i.e., ability of the metallic components to reproduce the passive layer after being removed due to an impact applied by the sand particles during the erosion-corrosion processing. The presence of pores formed during the LPBF process results in changes in chemical composition of the solution that diffuses into the pores, and difficulties in repassivation. This situation is similar to the active corrosion state in which the growth of the pits continues. Therefore, although the density of the LPBF sample is higher than 99.5%, there are still some small porosities in the microstructure, which have a remarkable impact on the erosion-corrosion resistance. Similar results of the effect of porosity induced by the LPBF process were reported by Suryawanshi et al.^[349] Their research showed that the presence of pores accelerates corrosion, as the pores provide preferential migration sites for the chloride ions and suitable places for the ions’ accumulation, which cause breakdown of the passivation film. They showed that the pits formed in LPBF 316 L stainless steel are bigger than those formed in conventionally manufactured stainless steel samples subjected to corrosion testing.

According to Yusuf et al.,^[350] corrosion behavior of LPBF 316 L stainless steel after high-pressure torsion (HPT) confirmed the undisputed role of process-induced porosity on corrosion performance. The HPT process on LPBF stainless steels brought a significant increase in corrosion resistance in chloride medium due to the substantial removal of pores and defects. Their results indicated that applying few numbers of HPT cycles leads to downsizing large pores and closing the other pores. They explained that Cl⁻ ions attack, and rupture, the initially formed stable passive. The process-induced porosity helps to the formation of an active-passive cell between the internal part of porosity and the matrix interface, causing accumulation of the Cl⁻ ions.

3.2.2. Role of inclusions and secondary phase particles

Numerous investigations have stated that the defects though the underlying bulk can be the preferential sites for pit initiation within passive layers on the surfaces. According to that, the initiation of localized corrosion occurs on structural heterogeneities.^[338,340] Due to the high conductivity of sulfide compounds in comparison with the oxide layer, the adsorption of chloride ions occurs on the surface of sulfide inclusions in stainless steels^[339] Moreover, the anodic dissolution of sulfide inclusions is accelerated by the adsorption of the chloride ions, which is observed in the austenitic and ferritic stainless steels.^[351] Manganese sulfides and its other compounds are the main inclusions formed in the austenitic and ferritic stainless steels; however, the formation of CrS is thermodynamically possible in steels with a low manganese content. Also, at some levels of manganese, an iron-manganese spinel is the stable sulfide phase formed in steel, which more contributes to promote the initiation of pits in comparison with the CrS. Some studies showed that the inclusion size is in accordance with the current noise, which is generated due to inclusions’ dissolution and the size of active inclusions corresponded to the pitting potential.^[351] Furthermore, it is noticed that to have suitable resistance-to-pitting, the size of the inclusions should be kept below ∼1 µm in stainless steels. This was previously revealed using simple immersion tests.^[351]

The potential-pH diagram of MnS-H₂O-Cl⁻ indicates that the potential range in which the sulfide is thermodynamically unstable is within the corrosion potential range of the stainless steel’s surface, which is passivated in an aqueous solution containing chloride ions^[338] A surface of the bare metal is exposed to the solution by the dissolution of the sulfide inclusions. The growth of the pits is depended on the dissolution and mass transfer that occur on the surface. In addition to sulfides, oxides, carbides, silicates, and oxy-sulfides are the preferential zones for the localized corrosion, particularly pitting. The electron microprobe analysis revealed that corrosion pits were mostly initiated on oxides, sulfides, and oxy-sulfides inclusions.^[334]

Some researchers have reported that stainless steels fabricated using the LPBF process have a higher corrosion resistance performance in comparison with the wrought material.^[343,349] According to Sander’s research,^[343] the resistance of the LPBF 316 L stainless steel to pit nucleation is more than that of the wrought 316 L, while its repassivation ability is more inferior once a stable pit forms. Although Cr-depletion zones formed adjacent to MnS inclusions are known to be suitable areas for pit nucleation, their formation probability decreases in the LPBF process, resulting in higher resistance to pit nucleation in LPBF components in comparison with conventionally fabricated counterparts.^[234,343] Chao et al.^[352] also elucidated the absence of MnS inclusions in the microstructure of 316 L stainless steel fabricated using the LPBF process, which leads to higher corrosion resistance compared with wrought structures. The fast solidification in AM process hindered the formation of MnS inclusions and corresponding Cr-depletion areas as strong pit initiators. However, the evolution of nano-scale oxides/oxynitrides containing a homogenous Cr composition result in a distinct resistance to pitting.^[352]

Microstructural analysis of AM 17-4 PH stainless steel showed that NbC precipitates form within a finer size range compared with those form during a conventional manufacturing process, introducing a higher pitting resistance (nearly 10 mV higher in E_pit).^[351] Besides, an enhancement in pitting resistance of LPBF 316 L stainless steel in simulated body fluid (SBF) was studied, where Man et al.^[353] reported that higher densities of grain boundary, subgrain boundary, and dislocation led to a thicker passive film created in LPBF 316 L stainless steel compared with the one formed in a wrought sample subjected to the SBF solution. Furthermore, a nobler pitting potential in the LPBF sample than that of the wrought sample in the SBF was due to dislocation accumulation around oxide inclusions and the absence of sulfide in the LPBF components. Also, the size of inclusion in the LPBF component was 50–200 nm, which was much finer than that in the wrought specimen, resulting in formation of a more noble pitting potential of the LPBF 316 L stainless steel. Similar to Man’s findings, Lodhi et al.^[354] also reported that the LPBF 316 L stainless steel shows a superior corrosion resistance, especially pitting corrosion, compared with the commercial counterpart in an acidic solution due to suppression of MnS particles formation.^[354] Accordingly, the higher corrosion resistance was attributed to the development of fine sub-granular structures, which improves the stability of the passive oxide film. A reduction of MnS inclusions due to rapid solidification rate (approx. 10⁷ K/s) is another factor contributing to improved corrosion resistance in LPBF samples. Similar results were also observed for the biocompatible additively manufactured 316 L stainless steel in a physiological environment.^[355] Since the microstructure of the LPBF 316 L stainless steel contain refined sub-grains within each coarse grain and the formation of micro-inclusions such as MnS is restricted, the LPBF 316 L stainless steel shows higher pitting resistance compared with the wrought counterpart.^[355]

A comparison between the re-austenitized 17-4 PH stainless steel produced by the LPBF technique and a wrought martensitic sample of 17-4 PH stainless steel showed that the re-austenitized LPBF steel had a remarkably higher corrosion resistance. This behavior is due to the lower sulfur content, which results in a lower level of MnS inclusion in the re-austenitized condition. Destabilization of passive layer is promoted due to MnS inclusions dissolution, which leads to formation of a sulfur-rich layer in adjacent areas. The wrought steel shows a deteriorated corrosion resistance because it includes a higher content of MnS inclusions.^[319] Investigation on additively manufactured graded material including 316 L stainless and 431 stainless steels indicated that the higher weight percent of 316 L stainless steel resulted in a reduction in carbide precipitates at grain boundaries. Also, it resulted in an increase in compactness of passive film and pitting corrosion resistance of the material. The conditions with more than 50 wt.% 316 L stainless steel introduce similar potential and pitting susceptibility as the pure 316 L stainless steel has.^[356]

The passivity behavior of as-received and solutionized-annealed (SA) samples of the LPBF 316 L stainless steel components was also investigated by Kong et al.^[357] Due to precipitation of particles with a size range between 100 and 300 nm at the grain boundaries of the SA samples, a poor passivation, and re-passivation ability were observed in these samples. The precipitates are mainly composed of Si and Al oxides.^[358] Besides, minor precipitation of MnS inclusions around the oxide particles can also be as pit nucleation sites, which reduces the pitting potential for the SA-LPBF 316 L samples. Accordingly, the intergranular corrosion is increased at the presence of impurity segregations into the grain boundary of the SA LPBF 316 L samples.^[359] Laleh et al.^[360] also reported deterioration of corrosion resistance in the heat-treated LPBF 316 L stainless steel. Their findings showed an unexpected drastic reduction in resistance-to-pitting of thermally post-processed LPBF 316 L stainless steel parts due to formation of deleterious MnS inclusions. Figure 54 exhibits SEM images of some inclusions before and after 48 h exposure of heat-treated LPBF 316 L stainless steel to a ferric chloride solution. Accordingly, there was no changes in the manganese silicates and manganese chromite inclusions, resulting in their higher dissolution resistance (Figure 54(a–c)). However, Figure 54(d) displays that the interface of the MnS inclusions and matrix was locally attacked by the corrosive media because the interface acts as a pit initiation zone. It was also shown that heat-treatment at temperatures above 1000 °C, changed the inclusions, particularly MnS in the as-built LPBF 316 L stainless steel. This type of inclusion was supposed to be the main reason for the severe reduction in resistance of the LPBF 316 L stainless steel to pitting after heat-treatment at temperatures higher than 1000 °C. It was reported that the as-built LPBF 316 L stainless steel has substantially superior resistance to pitting than its commercial counterpart. Also, the high-temperature heat-treatment results in a decrease in the pitting resistance.

Figure 54. SEM micrographs of the inclusions before Exposing to 6 wt. % ferric chloride solution (left column) and after that (right column). the chemical composition of the inclusions is shown schematically in the middle column in which green color represents the manganese chromite, blue color is manganese silicate, yellow color is silicon oxide, and red color is MnS (Reproduced with permission from^[360]).

Although it was found that the resistance to pitting of the LPBF 316 L stainless steel drastically decreases after heat-treatment due to the formation of MnS inclusions, its intergranular corrosion (IGC) resistance is improved compared with the commercial counterparts.^[361] Laleh et al.^[361] explained this unusual behavior due to no precipitation of Cr-rich particles in the LPBF samples since the LPBF specimens displayed a finer grain structure composed of high-density twin boundaries and low-angle grain boundaries, resulting in no probability for the evolution of localized Cr depletion regions. They claimed that the degree of sensitization has an inverse correlation with the grain.

According to Laleh et al.,^[361] the microstructures containing the finer grain size provide a higher grain boundaries’ surface area, which results in a high number of potential sites for the nucleation of the precipitates. However, the precipitates should reach a critical size to end sensitization. Since the LPBF alloys are composed of a fine-grained structure, this phenomenon leads to the presence of more potential areas for the nucleation of precipitates. However, due to a higher area that is susceptible to precipitation, the distribution of carbon is limited, avoiding the precipitation of chromium carbides. Furthermore, the presence of highly coherent coincidence site lattice (CSL) boundaries resulted in a reduction of intergranular corrosion. Figure 55 shows FIB images of the specimen fabricated by a conventional method exhibited dissolution at to grain boundaries (GB) (after the DLEPR test). However, the dissolution at GB areas for the LPBF samples are much lower in comparison with those of their commercial counterparts.

Figure 55. FIB cross sections that show the typical corrosion morphology at grain boundaries in (a–c) commercial, and (d–g) LPBF 316 L stainless steels. The high-magnified subfigures (b,c,f,g) are corresponding to the white rectangles marked in subfigures (a,d,e) (Reproduced with permission from^[361]).

Similar results on the effect of inclusions on pitting resistance of the LPBF stainless steel were also revealed in a study on LPBF 15-5 PH stainless steel. The development of NbC-(Mn, Si) O duplex particles formed after solution annealing-aging was appointed as the main reason for reduction of pitting resistance and the stability of the passive layer formed in this alloy^[317] Moreover, a study on the effect of prolonged aging on the corrosion behavior of 15-5 PH stainless steel indicated that the formation of Cr₂₃C₆ precipitates occurred after a long soaking time deteriorates the corrosion properties.^[271] It was claimed that the precipitation of Cu-rich precipitates with an outer shell enriched in Ni, Mn and Si improves the corrosion resistance of the aged samples.

3.2.3. Role of different phases

Pitting generally can happen at or near the micro-scale inhomogeneities such as inclusions, precipitated phases, dislocation, and other defects. The chemical changes in and around these sites can affect the stability of the passive film locally.^[362] These disordered regions result in high diffusivity for the movement of cation vacancies from the barrier layer/solution (bl/s) interface to the metal/barrier layer (m/bl) interface. Excess vacancies, which cannot be swooped by cation injection from the substrate metal, accumulate and form a blister that subsequently results in the nucleation of a metastable pit based on the Point Defect Model (PDM).^[363] Regarding the effect of different phases and the solid-state phase transformation theory, the contents of Cr, Ni and Mo are different in austenite and martensite phases, and these elements are expected to affect the passivity and corrosion resistance of stainless steels.^[364] Therefore, it is expected that martensite and austenite phases show different resistance to corrosion, or specifically, dissolution. Furthermore, it is reported that the nucleation of pits preferentially occurs at the interfaces of the austenite-delta (δ) ferrite or inside the austenite’s dendrite cores in the stainless steels, which are composed of austenite and ferrite phases.^[334] A lower cooling rate in the DED process than that of the LPBF process results in the formation of considerable level of retained δ, and microsegregation in 304 L stainless steel, which was observed by Melia et al.^[365] Their results showed that the formation of δ-ferrite affects the corrosion resistance due to the higher amount of Cr in δ-ferrite and formation of Cr-depletion zones adjacent to the δ/γ boundaries. Also, due to the lower cooling rate of the DED process than that of the LPBF, microsegregation of S and P can take place at δ/γ interface, which promotes nucleation of pits. Consequently, preferential dissolution of the γ phase in DED 304 L stainless steel results in a finely dispersed δ, creating web-like lacy pit covers and nest-like pits, which are susceptible to nucleation and propagation of pits. In addition, more aggressive pit chemistry can be created in the presence of residual δ stringers.^[365]

The effect of different heat-treatment processes on microstructural evolutions and corrosion behavior of 15-5 PH stainless steel was studied. It was reported that the evolution of reversed austenite within the heat-treatment process has a positive impact on the corrosion performance of the LPBF sample.^[317] The results showed that when the samples are aged at 500 °C for 10 h, the bulk/thin austenite still exists at the bottom of the melt pool/martensite lath boundary. The SKPFM result (Figure 56) displays that the potential surface of the austenite distributed at the bottom of the melt pool is 15 mV higher than that of the martensite matrix. As indicated by the potentiodynamic polarization curves and EIS results, the austenite leads to an increase in the resistance of martensitic steel to dissolution, which is in accordance with the literature reporting the beneficial effects of austenite on corrosion resistance.^[317] The results revealed that austenite formed at the bottom of the melt pool, which is rich in nickel, ultimately affects the crystallinity of the passivation film and the content of Cr₂O₃ in the inner layer of the passivation film. The SKPFM result showed that there is an improvement of 15 mV at the surface potential of the austenite comparing with the martensite. It has been reported that the as-built 17-4 PH components produced by the LPBF technique primarily includes the δ-ferrite phase due to high solidification rates during this manufacturing technique. This microstructure reveals higher pitting corrosion resistance compared with the martensitic microstructure of the LPBF re-austenitized parts.^[319] This behavior is related to higher area of grain boundaries and dislocations density in the re-austenitized parts. However, the general corrosion behavior of both structures is similar. Therefore, presence and volume fraction of ferrite and martensite do not have a significant impact on corrosion resistance.^[319] In addition, it was shown that the formation of the σ-phase in the LPBF duplex stainless steels can deteriorate the resistance of the alloy to pitting corrosion. The results showed that the duplex-stainless steel with a ratio of almost 59.5:40.5 (α:γ) shows higher resistance-to-corrosion characteristic. The precipitation of a few σ phases nucleated at the α/α, α/γ, γ/γ, α/σ and γ/σ boundaries lead to Cr depletion through these areas, which subsequently reduce the corrosion resistance.^[366]

Figure 56. The EDS and SKPFM measurements of the LPBF 15-5 PH parts: (a) microstructure of the at sample, (b) higher magnification of the area marked in (a), (c) the SKPFM profile corresponding to b), and (d) the potential-distance data in different areas (Reproduced with permission from^[317]).

3.2.4. Role of melt pool boundaries

One of the characteristics of the LPBF microstructure is melt pool boundaries (MPBs) in which the elemental segregations, thermal stresses, and non-equilibrium phase formation take place.^[367,368] MPBs are divided into two forms: layer by layer and track by track boundaries.^[369] Macatangay et al.^[370] showed that MPBs in LPBF components are more susceptible to corrosion in aggressive solutions. Similarly, the presence of non-equilibrium phases at MPBs is the main reason for localized corrosion in the LPBF 316 L parts used in proton exchange membrane fuel cells.^[371] The evidence of unstable passive current density detected from potentiodynamic tests is attributed to the non-equilibrium structures at MPBs.^[371] Moreover, MPBs of 316 L stainless steels are susceptible to corrosion when the material is subjected to FeCl₃ solution, as reported by Ni et al.^[372] Zhou et al.^[373] investigated the effect of subcritical heat treatment on corrosion of austenitic 316 L stainless steel in which heat treatment at subcritical temperature (950 °C) improves the resistance-to-corrosion in the LPBF components due to eliminating the MPBs reported as the susceptible areas to pitting. MPBs are interfacial defects, which can provide suitable areas for the nucleation of pits and intergranular corrosion. Their results showed that the heat-affected zones (HAZs) with a cellular microstructure, and remelted zones with a coarse-grain microstructure form at MPBs. Besides, the rapid solidification during the LPBF process results in a local temperature gradient, bringing a surface-tension gradient and local heat accumulation in the MPBs.^[374] Higher interfacial free-energy at MPBs compared with the one through the matrix and grain boundaries is also due to the local heat accumulation, which makes the MPBs vulnerable to localized corrosion in LPBF parts.^[373] Different types of MPBs, and pits formation on MPB are presented in Figure 57(a,b), respectively.^[369,373]

Figure 57. (a) SEM image of MPBs in the LPBF 316 L sample, and (b) a pit formed on a MPB (Reproduced with permission from^[369,373]).

Spherical pores and different microstructures are observable at each side of the MPBs in the as-received LPBF 316 L stainless steel.^[350] As seen in Figure 58, area “A” shows a fine elongated structure, while area “B” depicts an equiaxed cellular structure. The formation of the cellular substructure is attributed to fast and non-equilibrium solidification of the melt during the LPBF process. The cell boundaries are brighter than the cells, which might be due to partitioning of heavy elements into boundaries during solidification. Within the cell boundaries, severe microsegregation of Cr and Mo, and lower segregation tendency of Ni are detected compared with intercellular regions. Therefore, it is claimed that the segregations of the alloying elements such as Cr and Mo into the cell boundaries can affect the corrosion behavior of LPBF parts. In addition, the formation of distinguishable microstructure at each side of the MPBs finally leads to formation of areas vulnerable to localized corrosion. Indeed, the localized corrosion at MPBs (instead of the grain boundaries) is an important finding in the literature.^[350]

Figure 58. (a) Distinguishable areas of “A” and “B” at each side of MPBs, and (b) BSE image with higher magnification of cellular subgrains in the LPBF 316 L sample (Reproduced with permission from^[350]).

3.2.5. Role of surface roughness

The surface roughness of a metallic component has a pronounced impact on pitting corrosion. It is pointed out pitting potential of the stainless steels with rougher surfaces is lower than those smoother ones. Also, the metastable pitting incidence is reduced for a smoother surface by decreasing the area fraction of areas susceptible to metastable pitting growth.^[375] It is worth noting that different processes of surface finishing applied on stainless steel surfaces affect the chromium content in the oxide film.^[376] The correlation between surface roughness and pitting corrosion has been investigated in several studies. The effect of roughness on electrochemical activity of the metallic surfaces is discussed as follows:^[377]

1. The surface roughness produced by grinding and bead blasting can introduce surface defects such as internal stresses, plastic deformation, and micro-strains, resulting in fragmentation of grains beneath the surface and structural heterogeneity. Subsequently, the stored energy as a source of activation energy, the nucleation of pits, and corrosion rate increase.^[378,379]

2. Grinding produces deep grooves that act as the micro-reaction sites through which aggressive agents are accumulated, and the corrosion products are trapped in their roots, which results in enhancement of the pits growth. Nevertheless, the activated sites for the pits’ nucleation are decreased on a smoother surface.^[380,381]

3. The area of interfacial zones in a rough surface is high, which leads to enhancement of the corrosion rate under the corrosive environment.

4. The hydrodynamic and the mass-transfer boundary layers, as well as the fluid motion, are affected by rough surfaces, causing high sensitivity of the corrosion processes to fluid velocity.^[382,383] Flow mode can determine uniform or localized corrosion, particularly the initiation and propagation of pits. The corrosion corresponding to fluid flow is named as flow corrosion and categorized into corrosion-erosion and flow-accelerated-dependent corrosion.^[384,385]

The LPBF components show higher surface roughness than the counterparts fabricated using conventional techniques. Therefore, the impact of their rough surfaces on corrosion resistance is detrimental.^[386] The surface roughness of LPBF 316 L stainless steels was reported about 12-16 μm.^[387,388] In addition, the surface quality of the AM components depends on the surface orientation within the building process. Anisotropy in surface roughness was observed in AM-fabricated parts. Some studies showed that changing the process parameters leads to produce components with smoother surfaces. Chen et al.^[389] found that increasing the building orientation results in a reduction of roughness in both top and side surfaces and decreasing the anisotropy in the surface roughness. Additionally, some post-processing techniques such as abrasive blasting, electrochemical polishing, electropolishing, and plasma spraying are applicable to increase the surface quality of the LPBF components.^[390,391] The effect of laser polishing on corrosion performance of the LPBF 316 L stainless steel indicated that the laser polishing improves the corrosion resistance. A combination of grain refinement and a decrease in surface roughness after laser polishing are the reasons for an improvement in resistance-to-corrosion property in the LPBF samples.^[311] The corroded surfaces of the as-printed and laser-polished specimens are presented in Figure 59(a–f). The surface of the as-built samples shows severe corrosion and the corrosion reactions’ products, while the laser polished surface displays fewer and smaller corrosion pits. Defects on the surface, such as residual fragments of powder, partially melted particles, and asperities are reduced after laser polishing, bringing reduction of the roughness surface. These defects are barriers against evolution of a stable and integrated passive layer. Additionally, the rougher surfaces expose a larger surface area to a corrosive environment. Therefore, the elimination of the surface defects decreases the most vulnerable zones to pitting, as shown in Figure 59(g,h). Furthermore, a fine uniform grain structure can be produced on the surface of LPBF 316 L stainless steels due to remelting and subsequent rapid cooling process during the laser polishing, which can result in an evolution of a more stable passive layer, as seen in Figure 59(h).^[311]

Figure 59. SEM images of corroded surfaces in (a–c) as-built sample, (d–f) laser polished sample, immersed into 0.4 Mol/L HCl solution (various magnifications), (g,h) corrosion mechanism in the as-built and laser polished samples, respectively (Reproduced with permission from^[311]).

3.2.6. Role of anisotropy

Anisotropy in the LPBF parts is a result of various solidification rates along different directions with different heat conductivity. The building direction usually shows faster heat conductivity than the other directions due to presence of pre-deposited metal layers.^[392] Moreover, it is known that anisotropy in microstructural features leads to anisotropic mechanical properties and corrosion behavior.^[393] The most common microstructure developed in the LPBF austenitic stainless steel is composed of cells, columnar dendrites, and the equiaxed grains rarely formed due to a high-temperature gradient in the AM process.^[392] The development of different solidification structures, grain structures, and textures through the LPBF process is also highly dependent on the process and the chemical composition of the alloys, which can affect the corrosion performance in a harsh environment.^[392] The effect of a unique crystallographic lamellar microstructure (CLM) developed in LPBF 316 L stainless steel was investigated. It showed that the CLM texture containing a major orientation of ⟨011⟩ aligned with the build direction results in higher pitting corrosion resistance than that of the conventionally manufactured counterparts.^[313] Columnar cells with an angle of approximately ±45° with respect to the build direction are shown with green arrows in Figure 60. As seen from Figure 60(a,b), there are also the elongated cells parallel to the building direction (red arrows) at the center of the melt-pools. The {001} pole figures shown in Figure 60(d–f) exhibit that the direction of the elongated cellular microstructure is almost ⟨001⟩. Accordingly, the CLM are formed through ⟨011⟩ + ⟨001⟩ orientation with respect to build direction. The results of corrosion tests on the CLM of LPBF 316 L stainless steel samples are also depicted in Figure 60(g).

Figure 60. (a,b) BSE images of the columnar cells’ boundaries and the melt pool traces through the y-z plane, (c) corresponding Crystal orientation map along the z-axis, (d–f) {001} pole figures of the points C, D, and E in (b), respectively, where the arrows indicate the direction of cell orientation (green: ±45°-oriented cells, red: vertically-oriented cells), and (g) corresponding potentiodynamic polarization curves of the CLMs and the wrought counterpart (reference plate) immersed into 0.9 wt.% NaCl solution at 37 °C (Reproduced with permission from^[313]).

Compared with pitting potential of ∼0.5 V in the wrought specimen, the pitting potential in the CLM produced via LPBF technique is significantly high (∼1.2 V). The pitting potential of 1.2 V is close to the potential in a transpassive state in which the instability of the oxide of chromium occurs thermodynamically. It reveals that the LPBF process results in a superior pitting potential in 316 L stainless steel, although the underlying mechanisms still remain unclear.^[313] Therefore, the presence of a unique texture can improve corrosion behavior. However, it is possible to have other textures in the LPBF components with different electrochemical behaviors. One of the studies showed that the laser power affects the preferred orientation developed in the LPBF components. The preferred orientation of ⟨110⟩ direction is formed in the LPBF 316 L components fabricated at higher laser power, as shown in Figure 61. The electrochemical impedance spectroscopy (EIS) and the potentiodynamic tests in SBF solution demonstrate that the LPBF samples produced at high laser power have a nobler pitting potential and inferior rate of corrosion than the conventionally fabricated 316 L parts. It can be concluded that there are preferential orientations to decrease the corrosion rate in the LPBF stainless steels.^[394]

Figure 61. IPF maps of 316 L stainless steels at conditions of (a) Quenched, (b) SLM-120 W, (c) SLM-150 W, (d) SLM-195 W, (e) SLM-220 W, and (f) corresponding grain size distribution maps (Reproduced with permission from^[394]).

Residual stresses are other factors of anisotropy in the LPBF parts, which can distribute unevenly over the entire structure. Successive remelting and rapid cooling during the LPBF process induce the residual stresses, which can be represented as dislocations’ tangles.^[395] According to Cruz et al.,^[396] investigations on residual stresses during the LPBF process confirmed that compressive residual stresses can improve the resistance of the LPBF 316 stainless steel to pitting.

Cruz et al.^[396] discussed that if the compressive stresses can be relieved during the post-processing heat treatment, it results in concurrent increase in donor density and point defect concentration within the passive film in 316 L stainless steel. It is hypothesized that an increase in point defect concentration, such as that of oxygen vacancies created by compressive stress relief, can promote destabilization of the passive film by facilitating Cl⁻ adsorption/ingress into the passive film. Interestingly, the compressive residual stresses are found to detrimentally influence the E_rep values in 316 L stainless steel. Mott-Schottky plots on samples with different levels of residual stresses (Figure 62) showed that the as-built sample with the highest value of compressive residual stresses has much lower donor density, which confirms that the passive layer is more compact and intact compared with the bulk underneath.^[396] As a postulation, if both compressive and tensile residual stresses are generated in the LPBF samples, some areas with compressive residual stresses will show a cathodic behavior versus the other areas with an anodic behavior, which finally deteriorates the corrosion resistance. Consequently, this phenomenon can result in the formation of micro-galvanic cells in the structure.

Figure 62. Mott-Schottky plot at frequency of 1 kHz measured in the cathodic direction. The area between two dotted lines shows a difference between the slope of plots through the passive region. Inset shows a difference in n-type region for different samples (Reproduced with permission from^[396]).

The anisotropy in grain size and crystallographic orientation was also revealed in the LPBF components of 316 L stainless steel.^[397] Figure 63(a–c) shows the location of three different planes and their value of porosity, where all samples show almost equal level of porosity. Figure 63(d–g) show the EBSD results of three planes with image orientation maps. When the location of scanning is away from the bottom plane (Figure 63(d–f)), the number of grains through [001] direction decreases from 18.7% in plane “A” to about 12.1% in plane “C,” while the grains aligned with [111] direction show an insignificant increase in quantity from 3.0% in plane “A” to about 3.4% in plane “C.” The distribution map of grain size in three planes was also calculated and showed in Figure 63(g). Plane “C” contains a higher percentage of small grains than the other planes. Moreover, an average grain diameter related to the plane “A,” plane “B,” and plane “C” was calculated as about 8.74, 8.44, and 8.36 μm, respectively. It was stated that due to more heating cycles applied to bottom sections of the sample, growth of grains is more pronounced in these areas compared with those located in top sections. Also, the potentiodynamic polarization plots of three planes in the LPBF 316 L stainless steel are exhibited in Figure 63(h). According to the potentiodynamic curves, with an increase in the distance from the bottom, the curve shifted to the positive direction, resulting in a promoted corrosion resistance.^[397]

Figure 63. (a–c) Optical micrographs of three planes “A,” “B,” and “C” in the LPBF 316 L stainless steel along the building direction, (d–f) corresponding IPF maps of planes “A,” “B,” and “C,” respectively, (g) the grain distribution map, and (h) potentiodynamic polarization plots of planes “A,” “B” and “C.” (Reproduced with permission from^[397]).

All planes are fully dense with a lower percentage of porosity, where the austenite is detected as the main phase in the microstructure. However, the number of grains through two different directions is observed different, as the distance from the bottom increases. Accordingly, the number of crystals along [200] orientation gradually decreases, while it increases along [111] orientation. Additionally, the size of grains is slightly decreased by distancing from the bottom. Therefore, the different levels of corrosion resistance in the samples are attributed to differences in microstructural features, such as the orientation of crystals and the average grain size along the building direction of the LPBF 316 L samples.^[397] Moreover, an anisotropy in corrosion behavior of LPBF samples was also investigated in CX stainless steel reported by Shahriari et al.^[78] Two different planes of cuboidal samples, including side planes parallel and top planes perpendicular to the direction of building, were selected to compare their properties. The electrochemical and microstructural analyses indicated that microstructural variations between the planes are due to the different thermal cycles occurred in top and side planes during the LPBF process. The side planes showed slightly coarser grain size (as shown in Figure 64), which can result in a lower dislocation density due to presence of low angle grain boundaries. Also, it was assumed that the level of residual stresses accumulated in the side planes is low, which has an important contribution to improve its corrosion resistance as compared to that of the top plane.^[78]

Figure 64. (a,b) Misorientation maps of grain boundary in the LPBF CX sample obtained from the top and side planes, respectively, (c) corresponding grain size distribution map, and (d) the CPP curves of the top and side planes (Reproduced with permission from^[78]).

3.3. Unique corrosion issues in AM-fabricated stainless steels

The effect of hydrogen charging on the microstructure and durability of the wrought and the LPBF 316 L stainless steels in a proton exchange membrane fuel cell (PEMFC) was studied by Kong et al.^[398] The LPBF technique has been expanded to design the novel PEMFC parts, which provides an improvement in the system efficiency, stability, durability, and a reduction in manufacturing and running costs. The suitable corrosion resistance, fatigue endurance, and economic efficiency of the LPBF 316 L stainless steel introduce them as a suitable candidate for bipolar plate materials in the PEMFC. Estimation of the stability and durability of the LPBF stainless steels used in the PEMFC, particularly under hydrogen charging, is crucial in this end. Hydrogen damage is a severe problem that contributes to deterioration of the anti-corrosive behavior and mechanical properties of a metal. The stacking fault energy (SFE) is decreased by the diffusion of hydrogen into the lattice, where the materials with low SFE usually shows higher tendency of dislocations’ dissociation resulting in formation of excessive stacking faults.^[398] Those generated stacking faults can serve as nuclei for the subsequent formation of ε-martensite and/or α’-martensite. In general, the martensite phase has poor corrosion resistance compared with the austenite phase and is preferentially corroded. Hydrogen charging tests are carried out under a constant current mode with two electrodes: a Pt plate with a 3.3 cm² area as a counter electrode, and a stainless-steel plate with a 1 × 1 cm² area as a working electrode. The hydrogen charging process is conducted through the wrought and the LPBF 316 L stainless steels at 50 mA/cm² for 1 and 4 h in 0.5 M H₂SO₄ and 0.25 g/L thiourea (CH₄N₂S) solutions, respectively, where the thiourea is a recombination inhibitor that enhances the solute hydrogen concentration at the solution/specimen interface. The hydrogen-charged specimens are outgassed for 5 days for the microstructural and electrochemical characterizations. Figure 65(a) is a scheme of a PEMFC with complex components fabricated using the LPBF technique.

Figure 65. (a) Scheme of a PEMFC with a complex structure,^[398,399] (b,c) bright-field TEM of the surface in the wrought 316 L stainless steel after hydrogen charging at 50 mA/cm² for 4 h, (d) the electronic diffraction pattern of the area marked in (b), (e,f) bright-field TEM of the surface in the LPBF 316 L stainless steel after hydrogen charging at 50 mA/cm² for 4 h (Reproduced with permission from^[398]).

Hydrogen charging of the LPBF 316 L stainless steel for 4 h at 50 mA/cm² led to an infinitesimal fraction of martensite transformed from the austenite, which was associated with high density of dislocations and fine cellular microstructure. Phase transformations in the wrought 316 L stainless steel under hydrogen charging contained the steps of γ → ε → α’. Figure 65(b–f) display the TEM results of the wrought and the LPBF 316 L stainless steels after hydrogen charging. It is observed that there are many nanotwins in wrought 316 L stainless steel (Figure 65(b,c)) and that martensite formed along the nanotwin boundary (Figure 65(d)). It is found that α′-martensite nucleates and grows on twin boundaries; thus, the presence of hydrogen-induced twinning has a great effect on the formation of martensite. However, in this work, except a difference in movement of some dislocations, austenite to martensite transformation due to the hydrogen charging was negligible, resulting in an excellent resistance to corrosion (Figure 65(e,f)). An improvement in their corrosion resistance compared with the wrought counterparts is ascribed to zero-to-low percentage of martensite induced by hydrogen charging in the LPBF samples.

In another study, the corrosion behavior of the LPBF 316 L stainless steel reinforced by amorphous Fe-base alloy powder was studied and compared with the LPBF 316 L stainless steel.^[400] In this research, high pure commercial powder of 316 L stainless steel and amorphous Fe_43.7Co_7.3Cr_14.7Mo_12.6C_15.5B_4.3Y_1.9 (at. %) powder were used as the matrix and reinforcement particles, respectively. The research showed that the production of particle-reinforced composites using the LPBF process is a suitable technique for developing high-performance hard parts. The mechanical strength and wettability of LPBF stainless steel were increased by adding the amorphous Fe-base particles.^[401] It was found that austenite is the only phase formed in the LPBF parts; however, the composite components showed a microstructure of the austenite phase and dispersed Y₂O₃ embedded in the top surface. However, the Y₂O₃ particles were removed during polishing operation of the composite, indicating that the Y₂O₃ particles were only dispersed through the composite’s surface. Also, it was observed that the presence of the reinforcement particles results in grain refinement of the microstructure shown via inverse pole figure (IPF) maps (Figure 66(a,b)). The coarse grains in LPBF composite are only 50 μm in size, while the size of coarse grains in LPBF stainless steel is almost ∼100 μm (Figure 66(c,d)). Besides, the average grain size in the composite is estimated to be almost 6.47 μm, which is lower than that of the LPBF 316 L samples (∼12.72 μm). This difference is attributed to the presence of Y₂O₃ particles, which induces growth restriction to the austenite grains in the composite. The difference in corrosion behavior of the particle-reinforced composite and LPBF 316 L parts was also evaluated by the cyclic potentiodynamic polarization (CCP) testing (Figure 66(e)). The CCP curves showed that the passive area and pitting potential of the composite sample is higher than those of the LPBF 316 L part. The repassivation potential value of the composite is also nobler than its corrosion potential, indicating that the passive layer can be reproduced even at the potentials in which the passive layer breaks down. However, repassivation potential value of the LPBF 316 L part is lower than its corrosion potential, confirming that the formed passive film is not able to be reproduced over the entire test after it is broken down at pitting potential. Accordingly, it was stated that once the pitting occurred, the passive layer formed on the composite could be re-passive easier than that of the LPBF 316 L part. The difference between the electrochemical behavior of the LPBF 316 L part and the particle-reinforced composite was due to the formation of passive layers with different chemical compositions. In addition, the formed yttrium oxide, elements of B, C, Co, Cr, and Mo had a positive effect on improvement the anti-corrosion properties of the composite part. Furthermore, the evolution of a finer grain structure in the composite part resulted in the evolution of a thicker-integrated passive layer on the composite surfaces. This behavior was associated with an increase in electrons’ activities in the grain boundaries induced by grain refinement. Also, the formation of a passive layer was accelerated by the Cr and Mo alloying elements in the composite. The standard electrode potential ($E_H^{°}$) of Cr or Mo was lower than that of iron, which led to their ability to accelerate the formation of passive films compared with that of the iron. Besides, the passive layer on the composite part showed a solid solution structure composed of the molybdenum-, chromium-, and iron oxides. It was implicated that Mo also increased the Fe and O reactions and avoided the diffusion of chloride ions into the passive layer.^[400]

Figure 66. (a,b) Inverse pole figures of the LPBF stainless steel and particle-reinforced composite produced by the LPBF process, respectively, (c,d) the grain size distribution of the LPBF stainless steel and the LPBF composite, respectively, and (e) corresponding CPP curves of both Materials (Reproduced with permission from^[400]).

One of the issues in the LPBF austenitic stainless steels is the stress corrosion cracking (SCC) susceptibility. Lou et al.^[402] investigated the SCC susceptibility for the LPBF 316 L stainless steel in high-temperature water. The effect of un-recrystallized anisotropy and recrystallized equiaxed grains on SCC susceptibility of the alloy was investigated by applying different heat treatment cycles. Also, different parameters, which affect the SCC such as oxidizing normal water chemistry (NWC), stress intensity factor (K), reducing hydrogen water chemistry (HWC), crack propagation orientation, cold work (CW), retained un-recrystallized grains, and porosity were investigated. Their results showed that the SCC growth rates on the heat-treated LPBF parts are similar to that in wrought counterparts. They stated that cold working resulting in subgrains formation could act as a driving force for the SCC. Cold working perpendicular to the building direction resulted in a rapid growth of transgranular cracks. After a fully recrystallization, the intergranular crack morphology was similar for both the cold-worked LPBF parts and the wrought counterparts. At high-temperature annealing condition, effect of the retained un-recrystallized grains on the corrosion behavior of LPBF parts was insignificant in both NWC and HWC states. Also, when the volume fraction and size of pores increased, the growth rate of SCC was increased for NWC and HWC conditions. Figure 67(a) exhibits the grain boundary map of a crack taken from the EBSD analysis. The morphology of crack is changed from intergranular to transgranular in the annealed parts; however, the underlying mechanism for this transition is still unclear. It is assumed that in the absence of cold working, the resistance to intergranular cracking increases in the annealed LPBF parts. A comparison between the growth rates of SCC at different conditions of NWC and HWC is shown in Figure 67(b). Despite having 70% un-recrystallized microstructure in NWC and HWC conditions, the SCC behavior of the partially-recrystallized sample was similar to that of the fully-recrystallized material.^[402]

Figure 67. (a) Grain boundary map of a SCC in the LPBF 316 L steel, and (b) SCC crack growth of un-recrystallized (stress-relieved at 650 °C), partially-recrystallized (heat-treated at 955 °C), and fully-recrystallized (HIP + SA) LPBF components (all tests are through the x-z orientation with no primary cold working) (Reproduced with permission from^[402]).

Austenitic stainless steel (SS) is a widely utilized material in advanced nuclear power systems due to its exceptional combination of high-temperature mechanical properties and long-term resistance to corrosion. The emergence of Laser-powder bed fusion (LPBF) as an additive manufacturing technique has garnered significant attention in the nuclear power sector, offering notable advantages in producing intricately shaped components. However, a critical issue affecting stainless steel components in nuclear power plants is Irradiation Assisted Stress Corrosion Cracking (IASCC), which can lead to severe safety incidents and reactor shutdowns. IASCC is a complex phenomenon that arises from the combined effects of material damage induced by radiation, the corrosive environment, and stress. It is crucial to acknowledge the importance of the unique microstructure of LPBF austenitic stainless steel in terms of its resistance to irradiation when employing additive manufacturing techniques. In a study conducted by Jiang et al.,^[403] the researchers investigated the microstructure induced by irradiation and the corrosion performance of proton-irradiated 304 L stainless steel fabricated via LPBF in the primary circuit of a simulated pressurized water reactor (PWR).^[403] Figure 68(a,b) illustrate the cross-sectional transmission electron microscopy (TEM) images displaying the microstructures of irradiated samples produced through laser-powder bed fusion (LPBF) and traditional manufacturing (TM), respectively. Upon irradiation, both samples exhibited irradiation-induced dislocation loops, appearing as black dot damage. However, the LPBF sample demonstrated a lower generation of dislocation loops in the peak damage region compared to the TM sample. Additionally, Figure 68(c) presents a quantitative comparison of the characterization of the irradiation-induced dislocation loops and black dots between the LPBF and TM samples. From Figure 68(c), it is evident that the sizes of dislocation loops in the irradiated TM 304 L sample are considerably larger than those in the irradiated LPBF 304 L sample, which can be attributed to the unique microstructural features. Cellular sub-grains and nanoprecipitates have been identified to have a beneficial effect in mitigating irradiation damage in LPBF stainless steels. Dislocations, acting as preferred sinks for interstitial defects, can absorb interstitials through dislocation climbing, thus serving as unsaturated defect sinks.

Figure 68. Bright-field transmission electron microscopy (TEM) images of (a) LPBF and (b) TM 304 L stainless steel after irradiation. Different depths of the cross-section are shown for the LPBF sample (a1–a3) and the TM sample (b1–b3). (c) Illustration of frank dislocation loops observed in LPBF and TM 304 L stainless steel following irradiation at the peak damage region. The size distribution of dislocation loops in LPBF and TM 304 L stainless steel, irradiated with 2 MeV protons to a dosage of 0.18 dpa at 360 °C, is presented (Reproduced with permission from^[403]).

The LPBF samples exhibit numerous cellular substructures with a high density of dislocations at the cell boundaries (CBs), while the intracellular cells have minimal dislocations. The relatively fast diffusion and preferential sink for interstitial defects in LPBF samples lead to the annihilation of more defects near the CBs. As a result, the concentration of interstitial defects within the cell interior is lower compared to the TM samples. Figure 69 illustrates the surface morphologies of corrosion products formed on all specimens after exposure to simulated primary water in a pressurized water reactor (PWR) for 454 h at 320 °C. On the outer surface, faceted oxide particles are observed with a continuous layer formed beneath the dispersed particles. When comparing the irradiated samples to the unirradiated ones, both the size and number of oxide particles increase. The diameter of the oxide particles in the irradiated samples ranges from approximately 2–3 μm, with some larger particles exceeding 3 μm. This is in contrast to the unirradiated samples, where the oxide particles are around 0.1 μm in size. The overall size of the irradiated TM sample is similar to that of the LPBF sample, except for some slightly smaller oxide particles in TM samples. Notably, the number of oxide particles on the TM samples is significantly higher than that of the LPBF samples, especially when comparing Figure 69(c,d). The smaller size and lower number of oxide particles in the LPBF samples indicate better corrosion resistance. In this study, both the size and number of outer oxide particles, as well as the thickness of the inner oxide layer, significantly increase in the irradiated samples. It is worth mentioning that the oxide particles become finer and denser in the irradiated TM samples (Figure 69(b)). This can be attributed to the increased presence of irradiation defects, which serve as preferred nucleation sites for crystallite formation, resulting in a higher number of crystallites during the initial oxidation stages.

Figure 69. Surface morphology of the oxide scale formed on (a1,a2) irradiated LPBF samples, (b1,b2) irradiated TM samples, (c1,c2) unirradiated LPBF samples, and (d1,d2) unirradiated TM samples after corrosion for 454 h in simulated primary water of a pressurized water reactor (PWR) at 320 °C (Reproduced with permission from^[403]).

The higher density of dislocation loops in the irradiated TM samples provides more nucleation sites for oxide particles, resulting in the formation of finer and denser oxide particles compared to the LPBF samples. Irradiation causes an imbalanced flow of cationic and anionic species, leading to the formation of a porous inner oxide layer. This porous layer facilitates the diffusion of Ni and Fe ions toward the outer surface, leading to the precipitation of outer oxide particles through a dissolution-precipitation mechanism. Additionally, microstructural defects induced by irradiation in the matrix enhance diffusion by creating more vacancies for inward oxygen ions and outward metal cations toward the oxide/matrix interface. As a result, irradiation-induced defects accelerate corrosion by providing a fast diffusion pathway, leading to the formation of a thicker and more porous oxide film. Moreover, irradiation-induced point defects in the matrix enhance the diffusion of metal atoms to the oxide/matrix interface, providing sufficient Fe, Cr, and Ni atoms for the oxidation reaction at the oxide/substrate interface. The difference of 151 nm in oxide film thickness between irradiated TM and LPBF samples after corrosion can be attributed to the synergistic effect of microstructure and irradiation. The TM samples exhibit more point defects after irradiation, which in turn create additional diffusion channels for cations toward the oxide layer/matrix interface, accelerating the diffusion of anions into the matrix and resulting in a thicker inner oxide layer.

Comparing the unirradiated LPBF and TM samples, there is a 148 nm difference in oxide film thickness, indicating the excellent corrosion resistance of LPBF samples. The presence of small-sized cellular sub-grains in LPBF samples promotes the rapid formation of a compact and thin protective layer during the initial stages of corrosion, acting as a protective barrier against long-term corrosion. This results in a thinner and more effective oxide film with improved corrosion resistance.

3.4. Summary

Corrosion of metallic components poses a significant problem in various industries, leading to increased expenses for applying protective coatings, using corrosion-resistant alloys, and repairing or replacing components. To mitigate these costs, selecting an appropriate and cost-effective corrosion prevention strategy is crucial. Additive manufacturing (AM) offers a potential solution, as it provides distinct advantages over conventional fabrication methods in the production of metallic components. AM processes generate localized thermal cycles that can result in the development of unique microstructural features, including fine grain structures. Researchers have observed that various microstructural factors play a significant role in the corrosion properties of AM steels, particularly in relation to pitting. Factors such as grain size, grain orientation, porosity, segregation, phase fraction, residual stresses, and surface roughness have been found to have notable effects. Generally, an increase in grain size or the presence of higher porosity and segregation volumes tend to have negative impacts on pitting corrosion in AM components. However, the influence of grain size on corrosion resistance is not yet well-defined and depends on factors like environmental conditions and the stability of the passive layer. Porosity is often identified as a favored site for pit nucleation, whether within grains or at grain boundaries, especially when the porosity volume fraction exceeds 1%. Segregation creates microgalvanic coupling between regions rich in solutes and adjacent depleted matrix, affecting stress corrosion cracking (SCC) under tensile stress and localized corrosion at the boundaries of the melt pool. Surface roughness can cause electrolyte stagnation in concave areas, potentially influencing localized corrosion, as observed in AM 17-4 PH steel. In AM 316 L stainless steel, high levels of residual stresses and an anisotropic grain structure along the building direction have been found to decrease corrosion resistance and increase susceptibility to SCC, respectively. Additionally, matrix phases like martensite and austenite, as well as inclusions such as MnS, Fe2O3, and Cr2O3, influence the corrosion behavior of AM steels differently due to their varying resistance to corrosion. Therefore, to comprehensively understand the effect of each parameter on corrosion properties, a discussion of the principles and mechanisms of corrosion in AM steels, as presented in section three of this review, is necessary.

4. Outlook

Additive manufacturing (AM) of steels – inclusive of stainless steels – has received significant research attention in the past five years, particularly owing to the importance of such alloys in the marine, energy, oil and gas, shipbuilding, defense, automotive, and aerospace industries. Several AM techniques including laser powder bed fusion, direct energy deposition, and wire arc additive manufacturing have been used to successfully produce AM steels. Among the available powders and process parameter sets, maraging steels (C250, C300, and C350), maraging stainless steels (13-8 PH, 15-5 PH, and 17-4 PH), austenitic stainless steels (316 L, 316, and 304), and stainless steels including duplex, ferritic, and ODSs have also been of particular interest.

In the review herein, a consolidation of the current status of these alloys from the perspective of powder production, solidification behavior, microstructure control, post processing, and corrosion performance has been presented. The unification of these inter-related topics has provided a holistic insight. Several limiting factors in the production of steels using AM techniques were also highlighted and discussed. These factors include cracking from exceeding a threshold concentration of carbon, cooling-rate related phenomena leading to a distribution of performance of final products, and high quantity powder production. In addition, the effects of process parameters on microstructural defects and inhomogeneities of AM steels were reported in detail. Aqueous corrosion and the associated electrochemistry of what are considered to be novel alloys – in that AM steels are not widely studied for corrosion performance to date – were also reported including results from testing in harsh environments. In the context of ferrous alloys, corrosion performance is of the utmost importance in materials selection, directly associate with the specific application for which such AM steels are being used. From a materials selection and design perspective, it remains critical to consider the importance of alloying elements in achieving both the desired mechanical properties (whereby mechanical properties includes numerous properties therein) and the requisite corrosion performance of AM steels. Subsequently, a broad analysis of the influence of associated production parameters and the influence and limitations of different post-processing procedures (including heat treatment and hot isostatic pressing) on the microstructure and corrosion performance of stainless steels were investigated and discussed. A synopsis of the field indicates that the rapid heating and cooling within the AM process have an important role in influencing microstructure and hence, electrochemical performance across a range of testing environment. For example, the production of components with fine grain structure, and elimination of detrimental inclusions such as MnS in austenitic stainless steels, was shown to be beneficial in increasing the corrosion performance of AM prepared components in aqueous media – including in the presence of chloride ions.

Ferrous alloys are broadly used across many industries and readily available for AM procedures. What remains evident however, is that there is significant prospect for the families of ferrous alloys to grow – as the potential of AM steels alloys is greater than the feedstock currently adopted in the AM market. The unique properties that have been demonstrated from AM steels (such as high strengths from ultra-fine-grained microstructures) and improved corrosion performance from defect annihilation, reveal the prospect of developing new steels/stainless steels that can be utilized in harsh and demanding environments. Similarly, such novel AM steels may be able to replace higher grades of conventional alloys – at a lower cost. There are also great opportunities to develop new heat treatment procedures for AM stainless steels. The non-equilibrium solidification behavior for AM alloys requires new or yet-to-be-developed heat treatment regimens relative to those of conventional alloy counterparts. Various phenomena including phase transformations and precipitation hardening of AM steels can be investigated and in light of new exploration, offering the prospect that improved properties may also be achieved.

Hybrid additive-conventional processing procedures have also been recently explored in order to decrease the final price of AM steel components, or to develop additive repair procedures. Such hybrid additive repair procedures for stainless steel components in remote locations or on-board vessels in the ocean, can be transformational for several industries. Need-less to mention, operational challenges such as process control and post processing for remote sites will require its own concerted research effort. To date, Hybrid AM procedures have already been demonstrated in advanced tool and dies (maraging steels) and aerospace (Inconel) applications. The successful adoption of hybrid AM steels will require in depth studies regarding the performance of interfaces, including nano hardness measurements along and analytical electron microscopy. Additively manufactured steels have shown to have unique properties and have also been readily adopted in industrial applications – albeit at an introductory scale thus far. Overcoming challenges with producing high quality powders and developing process parameter sets will provide assurance to accelerate the wider adoption of AM steels.

CRediT authorship contribution statement

M.H. Ghoncheh: Investigation, Writing – original draft (Section 1 (powder production and process parameters), Section 2 (microstructure of as-built AM products)), A. Shahriari: Investigation, Writing – original draft (Section 2 (microstructure of heat-treated AM products), Section 3 (corrosion of AM steels)), N. Birbilis: Writing – review & editing, M. Mohammadi: Writing – review & editing.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Solution annealing at 1038 °C for 30 minutes and air cooling followed by aging at 482 °C for 1 h and subsequent air cooling to room temperature.^[271]

2 Solution annealing at 1038 °C for 30 minutes and air cooling followed by aging at 621 °C for 4 h and subsequent air cooling to room temperature.^[271]

3 Standard hydrogen electrode.