Corrosion of Additively Manufactured Alloys: A Review

With a growing demand for higher complexity in design and more efficient use of raw materials, additive manufacturing (AM), more commonly called 3D printing, has grown exponentially in both development and application.^1-2  The advantages of AM over traditional manufacturing methods are apparent; AM facilitates the fabrication of near-net-shaped structures with minimal waste generation and enables production of complex designs in less time. The influence of AM has challenged customary approaches to design and maintenance. In the aerospace industry alone, AM has become a large disruptor, being able to reduce the production lead time of a jet engine from months to weeks.^3-4  As an example, GE industries have announced their intention to put jet engines that are 35% produced by AM into service by 2020.⁵  In the last decade, the capabilities of AM have expanded to a wide range of alloys, many incapable of production by other routes, often revealing enhanced mechanical properties.^6-8  However, there remain several obstacles before the impact of metal AM can be fully realized; both standardization of printing methods and the consistent production of qualified components require significant further development. Although there have been major strides in understanding the influence of AM on mechanical properties, the understanding of the corrosion performance of AM-produced metals and alloys remains in its infancy. The review herein seeks to critically consolidate the relevant studies to date and identify key areas of future work.

Metal AM processes are primarily divided into two main categories: powder bed fusion (PBF) processes, which include selective laser melting (SLM) and electron beam melting (EBM);⁹  and direct laser deposition (DLD), in which both the material and energy are supplied simultaneously to the build surface. The latter is often also often interchangeably termed direct energy deposition (DED). Although various other AM processes for metal fabrication exist (for example, welding and cold spray), this review will focus only on SLM, EBM, and DLD, which are currently the most predominant metal AM methods, and those which involve the consolidation of metal or alloy powder.

SLM operates in a powder bed, building components layer-by-layer by rastering a high-powered laser as directed by a computer aided design (CAD) model. A typical schematic diagram for an SLM system is presented below in Figure 1(a),¹⁰  along with its major features. After each layer is completed (i.e., consolidated via laser rastering), the build platform is lowered and an additional layer of powder is added via a recoater arm, capable of accurately producing layers of ∼20 μm to 100 μm in thickness.¹¹  The build process occurs within an inert build chamber under a nitrogen or argon atmosphere to limit oxidation.¹²  Furthermore, in order to reduce oxygen contamination, the chamber is kept under positive pressure.^13-14  SLM is suitable for producing highly complex parts with high dimensional tolerance. The build resolution is generally dictated by the powder size used, which is typically the range of 10 μm to 45 μm in diameter.¹⁵  Components manufactured by SLM nominally achieve a surface roughness (R_a) of 9 μm to 16 μm as a result of SLM’s high cooling rates, which are as high as ∼6 × 10⁶ °C/s.^16-19  High cooling rates can also result in isotropic, metastable, and refined microstructures in SLM fabricated samples.^19-21 

Similar to the SLM process, EBM is performed using a powder bed. However, EBM uses an electron beam as the energy source, which consequently requires that the process be operated in an evacuated build chamber so that a high-power electron beam can be generated to subsequently melt metal powder.²²  Vacuum pressure levels can vary, but levels near 10⁻⁶ torr are not uncommon. The chamber vacuum ensures that a high-quality electron beam is produced and contamination or oxidation is minimized in the case of reactive powders (i.e., titanium). To date, EBM has not been widely applied to many alloy systems, largely studied only in the case of titanium (Ti) alloys.²¹ 

In every AM process, the resulting microstructure (and hence the mechanical properties and, indeed, corrosion behavior) are highly dependent on the build parameters used. For each production method used, several key parameters have been identified that have a significant influence. Most critically, the input energy density (which is a parameter that is influenced by numerous input variables) will dictate the local temperature and local melt pool size (including shape and constancy), also influencing the cooling rate following solidification—all of which impact the resulting microstructure.^23-27  In PBF systems, the major parameters that affect the input energy density are the laser/electron beam energy power, laser spot size, and traverse speed.^28-29  The parameters that affect its thermal history are hatch spacing (the distance between two consecutive laser scans), scan pattern, layer thickness, and bed temperature.³⁰ 

In contrast to PBF systems, DLD supplies its material to the build surface via blown powder simultaneously with its energy source.^31-32  Wire arc additive manufacturing (WAAM) is an AM method similar to DLD in which the energy input is supplied by electric arc and metallic wires are utilized as feedstock.³³  The DLD system operates on a 5 to 6 axis system, which makes it an ideal candidate for repair and cladding work, but also for the manufacture of large components. In the case of DLD, the powder size range utilized is between 50 μm and 150 μm and thus is comparatively larger than the powder size range used in the SLM process (10 μm to 50 μm).^34-35  The laser used for DLD is typically an Nd:YAG diode or CO₂ laser, while the metal powder is fed through a coaxial or multijet nozzle.³⁶  In order to maintain an inert atmosphere, an argon shielding gas may be introduced via the powder feed as shown in Figure 1(b).^15,37-38  Unmelted powder can be recycled depending on the operation. In the case of DLD systems, the energy density is affected by laser power, laser spot size, traverse speed, and powder input rate; and the thermal history by the scan pattern and hatch spacing.^39-42 

In view of growing understanding toward AM processes, extensive studies have been performed (noting that corrosion was not characterized in most cases) to determine the properties of the following alloys, which are traditionally weldable alloys:

• SLM: stainless steels (Types 316L [UNS S31603⁽¹⁾]^43-44  and 304L [UNS S30403]⁴⁵ ), Ti-6Al-4V,⁴⁶  Inconel 625^† (UNS N06625),⁴⁷  AlSi10Mg,⁴⁸  Al-Zn-Mg-Cu⁴⁹ 

• DLD: stainless steel Type 316L,^50-51  Ti-6Al-4V⁵² 

• EBM: Ti-6Al-4V,^53-54  Inconel 718^† (UNS N07718)⁵⁵ 

Each AM process presents relative advantages and disadvantages, some of which are presented in Table 1.

The microstructure of alloys produced by AM are significantly different from those produced using conventional production methods due to a variety of AM process variables.^21,56  For SLM, the most widely varied process variables include laser power, scanning speed, hatch distance, and layer thickness (all of which are typically varied by a trial-and-error process to optimize component density). The metal powder undergoes rapid heating (usually to temperatures >2,000°C) followed by rapid solidification during SLM to form the “build” as the laser beam rasters onto the metal powder.⁵⁶  The build is also subjected to several thermal cycles due to local heat transfer emanating during SLM. The rapid heating and fast cooling rates coupled with thermal cycling induces the formation of unique alloy microstructures that typically have refined grain structures,^21,56  dislocation cell sub-structures,^21,56  and residual stresses.⁵⁶  They also cause the formation of metallurgical defects, including hot tear cracks,^57-58  entrapped gas porosity,⁵⁶  lack of fusion porosity,⁵⁹  and dendritic growth within the build.^3,57  The AM process variables therefore govern the nature of the microstructures and defects formed in the build. The influence of such AM process parameters on the nature of the build are briefly reviewed below.

The surface roughness (R_a) of components manufactured by SLM is typically high. The R_a values for metal parts produced by SLM are normally in the range 10 μm to 30 μm, which is significantly higher than that produced by methods such as milling (∼1 μm to 2 μm).⁶⁰  Wang, et al.,⁶⁰  studied the effect of laser energy density (ω) on the surface roughness of Type 316L manufactured by SLM. It was reported that when ω is <75 J/mm³, the R_a value is ∼15 μm and the surface roughness is mainly attributed to sporadic lack of melting. However, when ω is between 100 J/mm³ and 170 J/mm³, the R_a was decreased below 10 μm. Furthermore, R_a was observed to increase with ω > 180 J/mm³, possibly arising from excessive melting of the constituent powders. The phenomenon of “balling” was found to occur predominantly when ω was between 75 J/mm³ and 120 J/mm³;⁶⁰  balling is described as the formation of metallic droplets in opposition to a desired uniform spread of liquid metal on the molten surface.⁵⁹  Although there was some correlation between a number of SLM process variables and the surface roughness of the manufactured parts, the underlying mechanistic aspects are, to date, not clearly understood.

Sander, et al.,⁶¹  studied the effect of the laser power and laser scanning speed on the porosity of Type 316L specimens manufactured by SLM. A direct correlation between specimen porosity and the laser power/scanning speed was not derived in that study. Shang, et al.,⁴³  studied the influence of laser scanning speed on the porosity of Type 316L specimens (manufactured by SLM). In that work, the laser power was kept fixed at 195 W, hatch spacing at 0.09 mm, and layer thickness at 0.02 mm, while the scanning speed was varied between 800 mm/s and 1,083 mm/s. They observed that the porosities of the Type 316L specimens increased with the scanning speed, and proposed that at higher scanning speeds, the lack of melting might cause porosity. Similarly, Li, et al.,⁴⁴  investigated the influence of scanning speed on the porosities of Type 316L specimens. The laser power was kept fixed at 100 W, whereas the laser scanning speed was varied between 90 mm/s and 120 mm/s. They also observed that the porosity of the Type 316L specimen increased with increasing scanning speed. The effect of ω on the porosity of Type 316L was also studied by Cherry and co-workers,⁵⁹  who observed that at low ω such as ∼40 J/mm³, the porosity of the produced Type 316L specimens reached as high as 8.8% due to lack of melting of the metal powder and lack of sintering between the different layers. At higher ω, the porosity was found to remain comparatively high (∼6.5%). It was suggested that the larger melt pools that were formed at higher ω may be susceptible to solidification microshrinkage porosity. Furthermore, the formation of gas voids could also result in the formation of spherical pores in the build. Undoubtedly, a mechanistic understanding of the effects of the laser parameters on the specimen porosity is still being clarified and is essential to understand on the basis (as elaborated further below) that porosity can influence corrosion, namely processes such as pit repassivation.⁶¹ 

The formation of dislocation cell sub-structures is considerably common in the alloy microstructures produced by AM. Shang, et al.,⁴³  reported that Type 316L manufactured by SLM was comprised of a dendritic and cellular microstructure. The primary dendritic spacing and the average diameter of the cellular structure were both found to decrease with increasing scanning speeds.⁴³  Such cellular growth was reported to result in solute segregation at the cell boundaries;^21,43,62-63  however, the extent of segregation upon Type 316L stainless steel was shown in one study, via careful atom probe tomography testing, to be near-negligible.⁶⁴  Of the studies to date, there is insufficient information available from nanoscale characterization to permit a more general understanding to emerge thus far. The mechanisms of dendritic/cellular growth, solute segregation, and their correlation with the laser parameters are yet to be elucidated in published works.

The uneven heat distribution and the rapid cooling rates during AM commonly results in the generation of what may be large residual stresses in the build. Liu, et al.,⁶⁵  used x-ray methods to determine the residual stresses in Type 316L parts manufactured using SLM. Overall, it was observed that the residual stress is mainly compressive at the lower portions of the build and tensile in nature at the top of the build. This explanation is plausible, as during SLM, when the molten layers (at the top) solidify, they tend to shrink due to thermal contraction. This deformation is restricted by the layers below, resulting overall in a tensile stress at the top and compressive stresses in the lower/bottom layers.⁶⁶  Liu, et al.,⁶⁵  also observed that the residual stress is much larger in the direction parallel to the laser scanning direction, compared to the direction perpendicular to the laser scanning direction. In that study, the residual stress was determined to decrease with lower ω.

The rapid heating and rapid cooling rates in AM, particularly in SLM, tend to result in the formation of nonequilibrium microstructures; in particular, the high-temperature phases formed during melting are retained after rapid solidification. Chao, et al.,⁶⁴  observed that the manganese sulfide (MnS) typically present in Type 316L is annihilated during SLM, and instead manganese (Mn) or silicon (Si) oxide based inclusions are preferentially formed at the elevated temperatures. It is known that MnO and SiO have higher precipitation temperatures than MnS and they form in the liquid melt. These MnO/SiO inclusions are retained in the microstructure due to rapid solidification. Alam, et al.,⁶⁷  investigated the effects of varying the laser power/scanning speed on the resultant properties of Type 420 (UNS S42000), manufactured by DLD. They observed that higher laser power/scanning speed (2.5 kW, 10 mm/s) presented a higher percentage of nonequilibrium secondary phases (i.e., eutectic ferrite, retained austenite [γ-Fe], and metastable metallic carbide phases) formed compared to specimens produced with a lower laser power/scanning speed (2 kW, 7.5 mm/s). The Type 420 specimen manufactured with the higher laser power/scanning speed also exhibited higher hardness (743 Hv) and higher residual stress (486 MPa tensile stress and 1,002 MPa compressive stress), compared to the specimen manufactured with lower laser power/scanning speed. It is noted that such residual stresses are quite significant. It is routine in many production processes to perform a stress relief heat treatment immediately following AM component production.

A schematic depicting the effect of different process variables on the macro- and microstructural features of the build is shown in Figure 2.^{21,58,61,64,68-69} 

It is important to highlight that the aforementioned defects and microstructural features are not confined to a single length scale, but cover a scale of >10 orders of magnitude as seen in Figure 3.

As indicated in the preceding section, the AM process enables the production of unique and microstructurally complex systems.^19,36,70  In fact, it allows (1) the formation of a significantly finer grain structure—i.e., 10 μm is the approximate average grain size for SLM 316L stainless steel, while 30 μm to 60 μm is common for the wrought counterpart;²¹  (2) unique second phase precipitation—i.e., ultrafine (α+β) lamellar structures (200 nm to 300 nm) in SLM Ti6Al4V;⁷¹  and (3) solute segregation—i.e., niobium (Nb) distribution in DLD 718.⁷²  Such changes not only significantly alter the composition, distribution, and size of second phase particles, but also have the potential to modify the electrochemical properties of AM-produced alloys. In order to consolidate a large number of studies and address the authors’ assessments of their respective conclusions, this section will list the different processing variables encountered in AM and discuss their influence on corrosion susceptibility. Herein, only the key variables reported in the literature are covered (Table 2).^61,64,73-85 

Several researchers have, often with little to no experimental support, reported that porosity may significantly impact the pitting characteristics of alloys produced by AM.^61,73-77  However, a mechanistic description of how pores (particularly unconnected pores) may assist in pit initiation/propagation is yet to be clarified. Similarly, solute segregation driven by the formation of cellular structures has also been proposed to be detrimental toward pitting of alloys manufactured by AM.⁸⁶  However, advanced compositional characterization of the cellular structures formed in SLM 316L has shown that neither Cr nor Mo content are significantly depleted at the cell boundaries.²¹  No post-mortem microstructural examination of corrosion upon SLM prepared Type 316L exists to date, which would reveal if corrosion preferentially occurs at sites of apparent solute segregation. The influence of the cellular structures and the associated solute segregation on the corrosion characteristics of alloys manufactured by AM is yet to be clarified for any alloy system. Similarly, the effects of residual stress, surface roughness, and ultra-refined grain structures on the corrosion characteristics of alloys manufactured by AM are topics that needs further investigation. Undoubtedly, the isolation of such microstructural variables may be a challenge; however, there exist a number of studies on model systems and nonAM systems that may be used as references.⁸⁷ 

Studies related to the corrosion of AM-prepared Al alloys have been essentially confined to the Al-Si alloy class, to date. The findings of such studies are summarized in Table 3.^79,88-94 

Some of the pertinent results summarized in Table 3 are discussed in more detail below, allowing for mechanistic aspects to be examined, but also illuminating that in many cases there are contradictory results from different studies. In cases where only a very small number of studies exist, contradictory conclusions between studies are the present hallmark of the status in the understanding of corrosion for many AM-prepared alloys.

The effect of the use of recycled powder, laser scan passes, and layer thickness on the corrosion of the AlSi10Mg produced by SLM was studied by Revilla, et al.⁸⁸  In the study, the corrosion potential (E_corr) values derived from potentiodynamic polarization of the cast and SLM AlSi10Mg specimens were similar, with values ranging between −0.7 V_Ag/AgCl and −0.6 V_Ag/AgCl (Figure 4). The corrosion characteristics observed for SLM and cast AlSi10Mg were correlated by Revilla, et al., who observed similar microstructure and composition. However, silicon-magnesium (Si-Mg) segregation was detected in the cast sample, whereas the same type of segregation was not observed in AM-prepared sample designated S2 (Figure 5). The presence of Si-Mg segregation increased the number of galvanic couples present on the sample surface, which was reported to directly enhance the localized corrosion of cast AlSi10Mg. The ability of SLM to refine such second phase regions was apparently beneficial in terms of corrosion.

The weight loss of AlSi10Mg produced by SLM immersed in a 3.5 wt% NaCl was investigated by Leon, et al.⁸⁹  This study showed that after 45 d of immersion, the SLM AlSi10Mg presented slightly lower weight loss than cast AlSi10Mg. Leon, et al., hypothesized that the corrosion is more accentuated in the cast alloy due to its extensive segregation of Fe- and Mn-containing phases, irregular porosity, and a dendritic microstructure. However, statistical analysis was not reported for the results, and empirical evidence was presented to support the suggested hypothesis. In a similar study, Leon, et al.,⁹⁰  investigated the effect of surface finish on the corrosion behavior of SLM AlSi10Mg. Corrosion tests (potentiodynamic polarization, immersion, electrochemical impedance spectroscopy [EIS], and fatigue) were conducted in 3.5 wt% NaCl for specimens denoted as a “polished” SLM AlSi10Mg sample (prepared by 1200 grit paper, unspecified abrasive, which should strictly be termed grinding as opposed to polishing) and for “unpolished” (i.e., as manufactured by SLM). The corrosion rate of the “polished” sample was lower than the “unpolished” in all tests performed. The presence of cavities and irregular surface morphology commonly observed for AM-prepared alloys was hypothesized by Leon, et al., to be the determining factor for the inferior corrosion resistance of the “unpolished” SLM AlSi10Mg. However, no mechanism nor detailed explanation was presented to support this hypothesis. Likewise, the corrosion of SLM AlSi10Mg was investigated by Fathi, et al.,⁹¹  as a function of surface finish in 3.5 wt% NaCl solution. Ground and as-produced (AP) SLM AlSi10Mg were compared with cast counterparts under the same experimental conditions. The SLM AlSi10Mg specimens presented an apparently better corrosion resistance than their cast counterparts (albeit only through evaluation by corrosion potential values, which are not a measure of corrosion rate). If the results are to be considered reliable, then they may be in agreement with Leon, et al.,⁸⁹  though they contradict the results obtained by Revilla, et al.⁸⁸  In contrast, Fathi, et al.,⁹¹  suggested that grinding the SLM AlSi10Mg specimens was detrimental to the continuous oxide layer formed on the AP specimen surface, inconsistent with the findings of Leon, et al.⁹⁰  Therefore, the role of surface finish (i.e., roughness) on the corrosion of SLM AlSi10Mg remains unclear. This lack of clarity is exacerbated by a lack of standardized testing by many authors, and also by the use of an arguably aggressive electrolyte for such an alloy, which might obscure surface finish effects.

The corrosion of SLM AlSi10Mg in aerated Harrison’s solution (3.5 g/L (NH₄)₂SO₄ + 0.5 g/L NaCl) was investigated by Cabrini, et al.^79,92-93  The effect of heat treatment and surface finish on the corrosion of SLM AlSi10Mg⁹²  was investigated for three specimens: untreated (UT), stress relieved (SR) at 300°C for 2 h, and water quenched (WQ) after solution annealing (SA) at 550°C for 4 h. The UT, SR, and WQ specimens were subject to potentiodynamic polarization testing for AP and so-called polished (P) surface finishes (Figure 6). The AP samples showed active dissolution without passivity. The highest E_corr was measured for the UT condition, while the WQ specimen presented the highest dissolution rate over a range of potentials (Figure 6[a]). The surface polishing allowed the development of a passive window during the potentiodynamic polarization of the UT and SR samples, indicating an enhancement in corrosion resistance (Figure 6[b]). The WQ specimen did not exhibit improvement in decreasing electrochemical dissolution following polishing, and a passive range was not observed. Cabrini, et al., suggested that the AP designated samples had higher surface roughness, which increased the area susceptible to corrosion. In addition, it was suggested that thermally produced oxide layers formed during SLM and heat treatments were less effective than the oxide layers naturally formed in contact with the natural atmosphere. The active behavior in earlier stages of the polarized AP specimens was purported to be related to the surface morphology and oxide layer quality rather than their microstructures. The polished samples were suggested to have had their thermally produced oxide layers removed by mechanical grinding. It was also proposed that the oxide layer formed inside pores by the SLM process are sites for the initiation of localized attack, as observed in the case of UT and SR samples. The heat treatment of WQ samples was suggested to homogenize the α-Al phase, which was deemed to be susceptible to chemical dissolution when exposed to Harrison’s solution. The scanning electron microscope (SEM) images of the UT and WQ cross sections after 164 h and 150 h of immersion, respectively, are shown in Figure 7. The pitting of WQ samples occurred as a large uniform cavity on the surface (Figure 7[a]), whereas for the UT samples it appears that pitting occurred with penetration along melting pool boundaries (Figure 7[b]).

The polished SLM AlSi10Mg specimens in general presented better corrosion resistance than the AP SLM AlSi10Mg. The heat treatment of SLM AlSi10Mg did not appear to alter corrosion performance when performed at low temperatures (at 300°C for stress relief of this alloy); however, heat treatment was detrimental when performed at higher temperatures (550°C) due to Si particle precipitation and coalescence. The results presented by Cabrini, et al., were analyzed statistically to validate their conclusions. However, in order to further understand the role of Si phases in the microstructure of SLM AlSi10Mg, characterization at the nanometer scales (e.g., transmission electron microscopy [TEM]) would aid the development of mechanistic models for the corrosion of SLM AlSi10Mg.

The effect of build orientation during SLM of the AlSi10Mg alloy was also investigated by Cabrini, et al.⁹³  The samples were built along the XY and XZ reference planes (Figure 8[a]) and compared electrochemically after being subjected to different surface treatments in an aerated Harrison’s solution. The samples were analyzed as-received, after shot peening and 0.1 μm alumina polishing. The E_corr and E_pit of the specimens are shown in Figures 8(b) and (c). The E_corr was measured for different exposure times in Harrison’s solution, where the values were similar for all of the specimens tested up to exposures of 2,000 s. However, the pitting potentials increased following a surface treatment of shot peening and polishing. The corrosion performance of AlSi10Mg in Harrison’s solution was enhanced by abrasively treating the surface.

Furthermore, the effect of post-build surface treatment on the corrosion of the SLM AlSi10Mg was investigated⁷⁹  using hot bright dip (BD), Ce(III) conversion coatings and was compared to an untreated SLM specimen. Both polished and as-produced specimens were analyzed for two build directions, XZ and XY (Figure 8[a]), and polarized in an aerated Harrison’s solution. The E_pit values determined from electrochemical tests are shown in Figure 9. It can be observed that a more noble E_pit was determined for the specimens tested in the XZ plane when compared with the specimens tested in the XY plane. The BD followed by the conversion coating using Ce(III) in the polished specimens presented improved resistance to pitting compared to the other tested specimens in all of the investigated planes. Furthermore, the surface polishing was beneficial to most of the specimens, except for the BD treated sample, as it removed the protective surface treatment applied on the surface. Therefore, the corrosion of the SLM AlSi10Mg could be improved by post-build surface treatments such as Ce(III) conversion coating. However, in this work, the mechanistic and microstructural scale phenomena that led to changes in corrosion with build direction were not thoroughly investigated.

The corrosion properties of SLM produced Al-12Si were investigated by Prashanth, et al.⁹⁴  In that study, the SLM Al-12Si specimens were heat treated in temperatures ranging from 200°C to 550°C, immersed in a 1 M HNO₃ (pH = 0) solution for 14 d, and tested using mass loss. Cast and SLM AP Al-12Si were subjected to the same immersion test conditions. The SLM and cast Al-12Si specimens presented a linear increase in weight loss with an increase of heat treatment temperature. The SLM (AP) and the cast Al-12Si had a similar weight loss, and both were inferior to the other heat-treated SLM Al-12Si. It was proposed that this behavior was correlated to the morphology of the Si phase within the alloy microstructure, with preferential dissolution of the Al phase occurring in the presence of HNO₃. The cast Al-12Si had Si plates, and the SLM AP Al-12Si had a Si network in their respective microstructures that prevented the detachment of Si phase from the surface. Spherical and disconnected Si particles created on the surface of the heat-treated SLM specimens were detached from the surface as Al was consumed. The so-called “vacancies” left by Si particles became exposed and the bulk Al became susceptible to further dissolution. However, the experiments were conducted in a highly corrosive, oxidizing acid environment, so investigations in milder solutions (e.g., dilute NaCl) may not fit in the proposed corrosion mechanism.

To date, the reported corrosion studies involving AM aluminum alloys remain essentially limited to Al-Si systems. The impact of the microstructures formed after AM on the corrosion of Al alloys are still not adequately understood or well reported. In order to understand corrosion mechanisms, further microstructural characterization, including methods that probe the nanoscale (e.g., TEM), should be combined with electrochemical characterization. Surface finish has been shown to influence the localized corrosion of SLM AlSi10Mg. The procedures adopted in the investigations reviewed herein (e.g., electrolyte, surface finish, heat treatments, etc.) did not follow common standards to produce comparable trends on the corrosion of SLM Al alloys.

The corrosion properties of AM-prepared Fe alloys have principally been reported for Type 316L stainless steel. However, a smaller number of accompanying studies regarding the corrosion of other steels have also been recently investigated. The tabulation of such studies are presented in Table 4.^{61,64,73-77,80,82-83,86,95-104} 

Environments containing chloride ions are well known to be detrimental to the durability of stainless steels, with a number of studies tabulated above appropriately focusing on such electrolytes.¹⁰⁵  In a study by Ganesh, et al.,⁹⁵  two batches of specimens (termed “A” and “B”) were produced via DLD, by utilizing similar parameters to produce Type 316L specimens. The DLD 316L was analyzed by electrochemical polarization in 0.6 M NaCl, and the results were compared with the electrochemical polarization of both a solution annealed, wrought 316L, and a plasma-transferred arc weld (PTAW) 316L. The corrosion of the DLD 316L specimens was intended to be analyzed in three different sections: parallel to the top surface (S), transverse cross-section (T), and longitudinal cross-section (L).

The DLD 316L specimens were investigated AP and after SA, and were polished down to a 0.5 μm finish before electrochemical testing. The E_pit values measured by Ganesh, et al.,⁹⁵  after potentiodynamic polarization of the DLD 316L specimens are presented in Table 5. The results showed that the wrought 316L after SA is the most resistant to pit initiation, as judged by the pitting potential, E_pit. The PTAW 316L presented a large scatter of results, and comparison was deemed to not be reliable without further data acquisition and statistical validation. The overall resistance to pitting was inferior for the DLD 316L “set A,” although the matrix presented was not completed.

A study performed by Zietala, et al.,⁸⁰  attempted to explain the corrosion DLD 316L in a 0.9 wt% NaCl solution. In their work, they concluded that AM samples present inferior corrosion performance; however, only limited electrochemical evidence to support their conclusions was provided.

The work performed by Chao, et al.,⁶⁴  investigated the pitting corrosion characteristics of SLM prepared Type 316L stainless steel. The experiments were conducted for three different Type 316L specimens: SLM “as-printed,” SLM samples in the annealed condition, and a wrought (annealed) counterpart. The samples were potentiodynamically polarized in 0.6 M NaCl at 25°C (Figure 10). The polarization curves reveal that the E_pit values were notably higher in the as-printed (0.75 V_SCE) and the annealed (0.70 V_SCE) SLM 316L, when compared to the wrought alloy (0.30 V_SCE).

Chao, et al., suggest that there was more resistance to pit initiation in the SLM 316L specimen due to the nature of the inclusions formed in the microstructure during the SLM process. The TEM-energy dispersive spectroscopy (EDS) maps (Figure 11) show the absence of MnS particles in the microstructure of SLM 316L, with Mn co-segregated or associated with what is principally oxygen and silicon. The presence of MnS particles in austenitic stainless steels (including Type 316L) has been reported to be associated with the initiation of pitting corrosion.^106-109  However, in the work of Chao, et al., it was revealed that MnS particles are annihilated during SLM, and do not reform upon rapid solidification. As a consequence, the authors have revealed a mechanistic rationalization for highly enhanced resistance to pit initiation.

The effect of porosity on pitting of the SLM 316L was investigated by Sun, et al.,⁷⁴  and Sander, et al.,⁶¹  in NaCl solution with similar concentration. Prior to electrochemical testing, the Type 316L specimens (SLM and wrought) were ground to a 1200⁷⁴  and a 2000⁶¹  grit finish using SiC paper. Herein, a composite figure has been prepared that reveals the parameters determined from electrochemical testing using potentiodynamic polarization performed by Sun, et al.,⁷⁴  and Sander, et al.,⁶¹  as a function of porosity shown in Figure 12.

Generally, the reported E_corr and corrosion rate (i_corr) values of SLM 316L do not significantly vary with porosity (Figure 12), while the repassivation potential (E_rep) reveals a decrease with increasing porosity. Sander, et al., hypothesize that the lower E_rep could be related to pores being detrimental to repassivation (i.e., following the initiation of pitting, the presence of pores requires that a lower potential be reached before pits can be repassivated). Sun, et al., state that repassivation of pits formed on the SLM 316L is more difficult than in wrought 316L, although no rationale was presented to explain this finding. Given the relevance of the repassivation characteristics to the utilization and durability of stainless steel, a clear understanding of the origins of reduced repassivation potentials remains a research priority.

The trend in E_pit obtained for the two studies (Figure 12) reveals contradictory results in comparison with the wrought 316L. The increase of porosity was found to adversely influence the E_pit of the Type 316L in the work presented by Sun, et al.⁷⁴  In addition, the SLM 316L presented lower E_pit values than the analyzed wrought counterpart. Sun, et al., suggest that the SLM 316L specimens were more prone to pitting due to an ion diffusion barrier formed around on top of pores. Such a barrier is postulated to lead to the concentration of aggressive agents (e.g., chloride ions) inside the pores, hence increasing the probability of oxide layer breakdown, although evidence for neither hypothesis was presented and would be difficult to experimentally validate. The SLM 316L analyzed by Sander, et al.,⁶¹  shows E_pit values considerably more noble than 0.3 V_SCE (the E_pit of the wrought 316L), whereas no trend in E_pit was observed with the increase of porosity. The presence of a finer distribution of inclusions formed during the SLM of the 316L is hypothesized by Sander, et al.,⁶¹  to reduce the number of pit nucleation sites on the metal surface and consequently decrease the susceptibility to pitting. The porosity values analyzed for the SLM 316L specimens are very distinct, varying between 1% and 7% in the work performed by Sun, et al.,⁷⁴  and between 0.01% and 0.4% in the work presented by Sander, et al. The antagonistic results create a performance window on the pitting of SLM 316L that requires further investigation to clarify how porosity could affect corrosion, or whether there is a true reversal of the porosity effect at some critical porosity between the two studies documented in Figure 12. The porosity of the SLM 316L specimens was determined to play a decisive role in the pitting of these alloys, evidenced by the reduction E_pit for porosities above 1%. However, for porosities lower than 0.4%, it was proven possible to produce SLM 316L specimens more resistant to pitting corrosion than the conventionally produced Type 316L (i.e., wrought). The specimens analyzed by Sun and Sander were produced by different machines using dissimilar powder batches, which could also affect the final product quality and properties. The corrosion of SLM 316L as a function of E_corr and i_corr was analyzed by Zhang, et al.⁹⁶  Despite showing decreased E_corr for more porous SLM 316L, the conclusions were made using a semiqualitative approach, and further characterization (i.e., quantitative porosity, microstructural imaging) should be performed to either support or refute the outcomes presented by Zhang, et al.⁹⁶  The effects of pore size and morphology on the corrosion of SLM 316L were not discussed in the studies performed by Sun, et al.,⁷⁴  Sander, et al.,⁶¹  and Zhang, et al.⁹⁶ 

Trelewicz, et al.,⁸⁶  investigated the corrosion of SLM 316L specimens in 0.1 M HCl. The SLM 316L was determined to be less resistant to corrosion than the wrought 316L in that environment, based on the passive current density measured during potentiodynamic polarization (135 μA/cm² and 22.5 μA/cm², respectively). This difference in the behavior is attributed to “inhomogeneous solute distribution and nonequilibrium microstructures” formed during the SLM of Type 316L. However, the explanations provided are insufficient in explaining the results, and further characterization would be required to better understand and validate the outcomes presented.

The environmentally assisted cracking (EAC) behavior of SLM 316L has been investigated by in the context of corrosion fatigue (CF),⁸³  different build directions,⁷⁶  and the role of inclusions on stress corrosion cracking (SCC).⁷⁷  In order to simulate the corrosive environment commonly present in nuclear reactors, hot pressurized water at 288°C with different contents of dissolved oxygen (DO) and dissolved hydrogen (DH) was utilized as an electrolyte. The CF was assessed for the SLM 316L and tested in two different directions: X-Z (load applied on the X plane and crack grow in the Z direction) and Z-X (load applied on the Z plane and crack growth in the X direction), as shown in Figure 13(a). The loading plane and direction were varied to observe the fatigue crack growth as a function load orientation. The SLM 316L specimens were also heat treated prior CF testing: (i) stress relieved at 650°C for 2 h and (ii) hot isostatically pressed (HIP’ed) at 1,150°C and 1,000 bar for 4 h in an inert argon atmosphere, followed by SA. The effect of mechanical deformation on CF was investigated after cold working (CW) the heat-treated SLM 316L specimens. The CF crack growth rate (da/dN) as a function of the loading frequency for the specimens after heat treatment and CW in high-temperature water with 2 ppm DO is shown in Figure 13. The SLM 316L specimens tested for the X-Z plane and cold-worked along the X direction presented higher crack growth rates for loading frequencies above 0.1 Hz (Figure 13[b]), and the cracks were associated with directional grain growth along during SLM.

The fatigue crack growth rate was reduced from 2 to 4 × 10⁻⁴ to ∼10⁻⁵ mm/cycle with the increase of loading frequency from 10⁻³ to 0.5 Hz for all of the analyzed Type 316L specimens. For loading frequency values close to 1 Hz, the crack growth rate was similar for the investigated specimens (∼0.1 mm/cycle). The SLM 316L specimens after CW and HIP showed increased fatigue crack growth rate with the decrease of loading frequency (Figure 13[c]). The performance of the wrought 316L analyzed in the work presented by Lou, et al.,⁸³  was superior to the SLM 316L. However, the stress intensity range (ΔK) and load ratio (R) utilized to test wrought 316L (ΔK = 8.3 MPa√m, R = 0.7) were not the same as SLM 316L specimens (ΔK = 11 MPa√m, R = 0.6), as shown in Figure 13(c). The American Society of Mechanical Engineers, Section XI (ASME XI) fatigue crack growth rate in air of Type 316L stainless steel is included in Figures 13(b) and (c) for performance reference.

SCC of SLM 316L was analyzed in another study by Lou, et al.,⁷⁶  utilizing similar specimens as those shown in Figure 13(a). The SCC experiments were performed in pure water at 288°C with 2 ppm DO (normal oxidizing water chemistry), and in pure water at 288°C with 63 ppb DH (reducing hydrogen water chemistry). The main results arising from SCC testing of the SLM 316L are presented in Figure 14.

The 63 ppb DH solution successfully inhibited the growth of cracks for most of the analyzed SLM 316L samples. The crack growth rate was, however, found to be higher for the SLM 316L specimens in the 2 ppm DO electrolyte prior to CW (1.2 × 10⁻⁷ mm/s), as shown in Figure 14(a). Moreover, longer and branched cracks grew over time for the X-Z sample, indicating that cracks propagate more easily along this direction of build.⁷⁶  After CW parallel to the Z direction, the SLM 316L specimens presented superior performance against SCC than the CW wrought 316L (Figure 14[b]). However, SCC rates were enhanced when 20%CW was performed on the X direction of the X-Z sample. CW further elongated the grains in the Z direction, which produced an elongated grain boundary path for cracks to propagate. In addition, the 63 ppb DH solution was not able to mitigate the crack growth of the X-Z sample 20%CW along the X direction as observed for the other specimens. The combination of HIP and CW was detrimental for the SCC of the SLM and wrought 316L, as shown in Figure 14(c). The HIP induces the formation of equiaxed grains which reduce the susceptibility toward cracks growth. The CW of the Type 316L specimens following HIP elongated the grains along the forging direction and facilitated crack growth. Low porosity was found to be beneficial toward reducing the crack growth rate in the SLM 316L, as revealed in Figure 14(d). The low porosity sample (0.08% mean porosity and 2.8 μm mean pore size) had a superior performance (i.e., 40% lower crack growth rate) when compared to the high porosity sample (0.3% mean porosity and 16 μm). The authors suggest that higher porosities on the SLM 316L enhanced the crack growth rate in hot pressurized water. They also note that more research on the effect of pores on the SCC of these alloys must be performed to provide more precise conclusions.

The influence of inclusions on the SCC of SLM 316L was also investigated by Lou, et al.,⁷⁷  in high-temperature water (288°C) and compared with their wrought counterpart. The authors found that Si-rich inclusions were detrimental to the SCC of SLM 316L due to preferential dissolution in hot water, thus increasing susceptibility to cracking. Therefore, controlling the formation of Si-rich phase could be essential to increasing the resistance of SLM 316L to SCC in hot water for suitable use of this alloy in nuclear applications.

Tribocorrosion of SLM 316L was investigated by Stendal, et al.,⁹⁷  in both 0.9 wt% NaCl and simulated body fluid (SBF) solution, and the performance compared to wrought 316L. The SLM 316L specimens showed a greater tendency to passivate and were less susceptible to metastable pitting than their wrought counterpart. After wear damage was imposed upon the samples, corrosion caused close to 50% lower volumetric material loss in SBF than in 0.9 wt% NaCl, for both SLM and wrought 316L. However, the effect of porosity and other microstructural features was not explored.

The so-called “oxidization” of SLM 316L was evaluated by Harun, et al.,⁸²  to examine the effect on the corrosion characteristics of SLM 316L in Ringer’s solution (8.6 g/L NaCl + 0.48 g/L CaCl₂ + 0.30 g/L KCl). Thermal oxidation was applied to thicken the oxide layer on the SLM 316L specimen surfaces. Treatments at 700°C for 150 h were found to improve the corrosion of SLM 316L by presenting lower corrosion rates (3.34 × 10⁻⁶ mm/y) in comparison to the untreated SLM 316L (9.21 × 10⁻⁶ mm/y). Exposure for longer periods of 200 h and 250 h decreased the corrosion resistance of SLM 316L in Ringer’s solution, with corrosion rates higher than 1.53 × 10⁻⁵ mm/y. Porosity of SLM 316L was correlated to corrosion characteristics in 0.5 M H₂SO₄ by Geenen, et al.⁷⁵  The results were compared with SLM 316L after HIP and a cast counterpart. Corrosion susceptibility was increased by HIP of SLM 316L due to oxide shape transformation from lamellae into spheroid particles which led to passive current densities 10 times higher than the other analyzed specimens. Geenen, et al., also suggest that the oxides present inside pores or in residual cracks can coalesce during HIP¹¹⁰  and apparently increasing the number of corrosion nucleation sites; however, the characterization of pores and residual cracks was not performed to confirm such a mechanism.

Heat treatments were performed in SLM 316L specimens to investigate SCC in boiling MgCl₂ (ASTM G36) in the work presented by Bruycker, et al.⁹⁸  The increase of heat-treatment temperature from 450°C to 950°C increased the resistance to SCC of SLM 316L based on visual inspection of cracks formed on the analyzed specimen. The SLM 316L heat treated at 950°C did not present visual cracks after SCC testing. However, the influence of other features, such as inclusions, surface finish, and grain size, on the SCC of SLM 316L were not investigated.

The corrosion characteristics of Type 316L produced by WAAM were investigated by Chen, et al.,⁹⁹  for AP and heat-treated specimens. Interestingly, in contrast with other AM methods capable of producing Type 316L specimens, WAAM of Type 316L promoted δ and σ phases in the AP microstructure. These phases played an adverse role in the corrosion of WAAM 316L, acting as nucleation sites for pitting. The heat treatment at 1,200°C for 4 h eliminated the residual delta and sigma phases and homogenized the microstructure into the common austenitic monophasic Type 316L stainless steels. Thus, the heat treatment of WAAM 316L increased the resistance to pitting in 3.5 wt% NaCl solution by increasing E_pit from 0.4 V_SCE in the AP specimen to values above 1.0 V_SCE after heat treatments at temperatures >1,200°C for 1 h.

Zhang, et al.,¹⁰⁰  studied novel stainless steel + x wt% Ni produced by SLM with nickel contents varying as x = (0, 3, 6, and 9), in order to assess the effect of Ni on the corrosion of stainless steels. The increase of Ni wt% in iron alloys has been reported to enhance their corrosion performance.^110-112  As predicted, the SLM stainless steel specimen containing 9 wt% Ni presented better resistance to corrosion with higher E_corr and lower i_corr in 3.5 wt% NaCl than the lower Ni content alloys investigated by Zhang, et al.

The corrosion of precipitation hardened 17-4 PH (UNS S17400) stainless steels produced by SLM was investigated by Schaller, et al.,⁷³  and Stoudt, et al.,¹⁰¹  in similar concentrations of NaCl solution (0.5 M and 0.6 M, respectively). Schaller, et al., analyzed the corrosion properties of SLM 17-4 PH as a function of porosity and compared the results with their wrought counterpart. The wrought 17-4 PH with average i_corr of 2.30 × 10⁻⁸ A/cm² showed better resistance to corrosion than the SLM 17-4 PH (i_corr: 2.41 × 10⁻⁷ A/cm²), measured by potentiodynamic polarization after 1 h of open-circuit potential (OCP) exposure. The i_corr measured by linear polarization for both the wrought and SLM 17-4 PH decreased by one order of magnitude after 24 h OCP exposure values were determined to be 2.73 × 10⁻⁹ A/cm² and 7.28 × 10⁻⁸ A/cm², respectively. The effect of porosity was evaluated based on the average pore size of different specimens in comparison to their wrought counterpart (with no porosity). Specimens with pores larger than 50 μm were found to be more susceptible to pitting (E_pit: ∼0.75 V_SHE) compared to specimens with pores smaller than ∼10 μm (E_pit: ∼0.5 V_SHE) and the wrought samples with no pores (E_pit: ∼0.30 V_SHE). Therefore, the presence of pores in the SLM 17-4 PH was deemed detrimental to their resistance to corrosion. Although evidence reveals that pores were responsible for the inferior corrosion performance of SLM 17-4 PH, microstructure characterization by TEM and EDS techniques could unveil other factors (e.g., nano-inclusions, martensite vs. austenite or ferrite) affecting the corrosion characteristics of these alloys. The effect of post-build heat treatments and exposed plane on the pitting of SLM 17-4 PH specimens was analyzed by Stoudt, et al., in 0.5 M NaCl. The heat treatments increased the average E_pit of the SLM 17-4 PH (>−130 mV_SCE) when compared to the wrought counterpart (−135 mV_SCE). Three planes (XY, XZ, and YZ) of the analyzed SLM 17-4 PH specimen showed dissimilar average values for E_pit (−130 mV_SCE, −104 mV_SCE, and −93 mV_SCE, respectively) and evidence of anisotropic pitting behavior. Within the scatter of values recorded, the lowest E_pit of the SLM 17-4 PH is similar to the average E_pit detected for the wrought counterpart. The results presented by Schaller, et al., and by Stoudt, et al., are contradictory regarding the corrosion of SLM 17-4 PH. However, while Schaller, et al., did not heat treat the SLM 17-4 PH specimens, Stoudt, et al., did not investigate specimen porosities. Therefore, further investigations should more systematically consider analyzing both porosity and heat treatments of SLM 17-4 PH in order to clarify the results presented to date.

The corrosion in artificial seawater of Type 4340 (UNS G43400) stainless steel produced by SLM (ASTM D1141-98) was investigated by Schmidt, et al.¹⁰²  The work reveals that there is no significant difference in the corrosion rates of the SLM (4.1 mpy to 4.9 mpy) and wrought 4340 (3.6 mpy) calculated by polarization resistance testing. The work performed by Schmidt, et al., provides a preliminary insight regarding the corrosion of SLM 4340 stainless steel; however, the corrosion of such alloys requires further investigation.

In related work, Lefky, et al.,¹⁰³  worked on the development of low chromium steel support structures that were supposed to be electrochemically removed after the DLD of martensitic Type 431 (UNS S43100) stainless steels. DLD readily permits the deposition of various material/alloy types, making this possible. Such low chromium steel supports were sought to be removed utilizing nitric acid as the etchant. However, the etching process was not selective and the martensitic Type 431 was also dissolved. Although the initial purpose was not successfully achieved, Lefky, et al.,¹⁰³  suggest that the etching of low chromium steels can be controlled to improve the surface finish of steels produced by DLD. It is noted, however, that the correlation between surface finish of AM steels and corrosion characteristics is still unknown.

The corrosion of SLM FeMn-Ag was investigated by Wiesener, et al.,¹⁰⁴  in SBF. The interaction of Ag precipitates and FeMn matrix created a galvanic couple in the early stages of immersion in SBF. However, as corrosion progressed, both phases reacted with the SBF and deposition of oxides and phosphates on the surface reduced corrosion rates, elucidated by increased impedance from 600 Ω·cm² to 1,300 Ω·cm² at low frequency (0.1 Hz) and from 15 Ω·cm² to 25 Ω·cm² at high frequency (5 kHz) over 38 h. The work presented by Wiesener, et al., represents an example of utilizing AM to produce alloys under nonequilibrium conditions and with atypical corrosion behavior.

Steels produced by AM still require considerable characterization of corrosion performance and related mechanisms. There were no trends nor general conclusions on the corrosion of steels according to the results presented in the literature to date. However, it is evident that the microstructural changes arising from AM processing can differ from the microstructures of conventionally manufactured steels. These microstructural changes directly affect the corrosion characteristics of AM steels. Thus, the variability of manufacturing strategies to produce steels via AM provides the tools necessary to manipulate potential microstructural features (e.g., porosity, inclusions, surface finish, etc.) in order to enhance corrosion.

Titanium alloys find use in two key industries, namely aerospace and biomedical, owing to their good mechanical strength, corrosion resistance, and biocompatibility. The investigations related to the corrosion of AM-prepared Ti alloys reviewed in this work are summarized in Table 6.^{68-69,84,113-132} 

The work performed by Dai, et al.,^84,113  analyzes the corrosion characteristics of Ti6Al4V alloys produced by SLM in different chloride-containing media. The authors evaluated the corrosion of SLM Ti6Al4V by altering exposed planes to different electrolytes⁸⁴  and by post-build heat treatments.¹¹³  The corrosion of SLM Ti6Al4V was not significantly affected by the specimen exposed plane in a 3.5 wt% NaCl. However, the same behavior was not observed in 1 M HCl whereby, according to Dai, et al., the plane parallel to the build substrate showed slightly better resistance to corrosion than the plane perpendicular to the substrate. Weight loss after 15 d immersion testing in 1 M HCl was measured to be 0.7 mg/cm² for the parallel plane to the substrate compared to 0.9 mg/cm² for the perpendicular plane. Potentiodynamic polarization revealed that the passive current densities of the parallel and perpendicular planes (to the substrate) were 2.8 μA/cm² and 2.5 μA/cm², respectively. This proposed corrosion rate difference between planes was attributed to the formation of “weaker” passive layer on the planes perpendicular to the substrate. In addition, the higher proportion of acicular martensitic phase (α′) along the perpendicular planes to the substrate was also correlated with lower corrosion rates.⁸⁴  Residual stresses were measured to be similar for both planes parallel and perpendicular to the substrate. Therefore, there was no evidence or suggestion in that work that the presence of residual stress in the SLM Ti6Al4V could influence corrosion in 3.5 wt% NaCl and 1 M HCl.⁸⁴  The corrosion of SLM Ti6Al4V and residual stresses from the build process still require further investigation in order to determine any causal or conclusive correlations. The SLM Ti6Al4V specimens investigated by Dai, et al.,¹¹³  were heat treated for 2 h at 500°C, 850°C, and 1,000°C. The heat treatment of SLM Ti6Al4V was found to be detrimental to passivation in 3.5 wt% NaCl with passive current densities of 0.9 μA/cm² (no heat treatment), 1.3 μA/cm² (500°C), 1.5 μA/cm² (850°C), and no fixed value (although higher than 1.5 μA/cm²) for heat treatments at 1,000°C.¹¹³  The i_corr increased as a function of heat-treatment temperature, with magnitudes of 13.1 nA/cm², 14.6 nA/cm², 55.8 nA/cm², and 76.1 nA/cm² for no heat treatment, 500°C, 850°C, and 1,000°C, respectively. Although the heat treatments performed reduced the α′ phase volume fraction, there was an increase of the volume fraction of beta phase (β) which further reduced the corrosion resistance of SLM Ti6Al4V, which is usually unexpected for these alloys. The authors hypothesized that grain refinement during heat treatment could be responsible for the atypical correlation between corrosion resistance and β phase volume fraction, and further investigation should be performed to assess this hypothesis.

Chandramohan, et al.,¹¹⁴  also investigated the effect of post-build heat treatments of SLM Ti6Al4V in 3.5 wt% NaCl. In accordance to the results obtained by Dai, et al.,¹¹³  there was an increased corrosion rate of SLM Ti6Al4V in 3.5 wt% when post-build heat treatments at 1,100°C and 900°C were performed for 1 h. The corrosion rates were measured and found to be 1.97 × 10⁻⁴ (mm/y) for SLM Ti6Al4V with no heat treatment, 3.4 × 10⁻⁴ (mm/y) for SLM Ti6Al4V heat treated at 1,100°C, and 5.9 × 10⁻⁴ (mm/y) for SLM Ti6Al4V heat treated at 900°C. In that work, it was suggested that the increase of β phase against the reduction of α′ after heat treatment of SLM Ti6Al4V was responsible for the increase of corrosion susceptibility.¹¹⁴  However, microstructural characterization was not presented to support such conclusions. The SLM Ti6Al4V specimens were also tested for corrosion in 1 M HCl and 1 M H₂SO₄ electrolytes for two different build directions: horizontal and vertical to the substrate. The authors concluded that build direction did not significantly affect the corrosion in all of the electrolytes studied¹¹⁴  when compared to the effect of post-build heat treatment.¹¹⁴  However, there was no trend on the susceptibility to corrosion of SLM Ti6Al4V immersed in 3.5 wt%, 1 M HCl, and 1 M H₂SO₄ built vertically and horizontally. In addition, the authors concluded that SLM Ti6Al4V is more resistant to corrosion in alkaline media, whereas the pH of the solutions analyzed was not presented and no comparison was made with wrought or cast counterparts.

SLM Ti6Al4V was exposed to Hank’s solution on different planes for corrosion testing in the work performed by Chen, et al.¹¹⁵  The corrosion resistance of the SLM Ti6Al4V on all of the exposed planes was found to be superior to wrought Ti6Al4V tested in the same electrolyte. Chen, et al., suggest that the defects generated during the rolling of wrought Ti6Al4V enhance the susceptibility to corrosion. However, the microstructural characterization provided by the authors was not sufficient to support this conclusion—not to mention that many Ti6Al4V products are actually never subjected to rolling. The SLM laser scan pattern was hypothesized to influence porosity and was associated to an anisotropic corrosion of SLM Ti6Al4V on differently exposed planes (X, Y, and Z). The X plane (perpendicular to the substrate and to the laser scan track) presented an E_corr of −0.48 V_Ag/AgCl, an i_corr of 0.03 μA/cm², and polarization resistance of 0.5 MΩ·cm². The Y (perpendicular to the substrate and parallel to the laser scan track) and Z (parallel to the substrate) planes showed similar (between each other) and superior E_corr, i_corr, and polarization resistance (−0.33 V_Ag/AgCl, 0.023 μA/cm², and 1.3 MΩ·cm², respectively) in comparison with the X plane. However, porosity was not quantitatively measured in their study to support the correlation that pores affect the corrosion of SLM Ti6Al4V.

Damborenea, et al.,¹¹⁶  compared the corrosion of SLM Ti6Al4V with a wrought counterpart in phosphate buffered saline (PBS) at 37°C. The polarization curves reveal that at 2.4 V_Ag/AgCl, there was a sharp increase of current density for the SLM Ti6Al4V from ∼5 × 10⁻⁶ A/cm² to ∼5 × 10⁻² A/cm², which was not observed for the wrought counterpart. The authors concluded that the SLM Ti6Al4V was more susceptible to the formation of hydrated titanium oxides on their surface at 2.4 V_Ag/AgCl, being associated with the current density increase during the polarization test. However, the validity of this assertion was not further tested in the work, and it was noted that the potential range being studied in such work was above the oxygen evolution potential and of no relevance to biomedical uses (for which PBS is often used as a test electrolyte).

SBF was utilized to investigate the corrosion of SLM Ti6Al4V as function of heat treatment and laser energy density by Xu, et al.¹¹⁷  The SLM Ti6Al4V specimens were annealed by Xu, et al., at 700°C, 750°C, and 799°C for 4 h, and electrochemically corrosion tested following heat treatment. The results for the annealed SLM Ti6Al4V were compared to the AP SLM Ti6Al4V and their wrought counterpart. The heat-treated SLM Ti6Al4V (i_corr = 0.305, 0.295, and 0.202 [× 10⁻² A/m²] for 700°C, 750°C, and 799°C, respectively) were less susceptible to corrosion in SBF than the AP specimen (i_corr = 0.384 × 10⁻² A/cm²). Although the wrought Ti6Al4V presented i_corr = 0.270 × 10⁻² A/cm², their anodic current density increased two orders of magnitude from i_corr during potentiodynamic polarization after breakdown at 2.4 V_Ag/AgCl. Passivity breakdown was not observed for the SLM Ti6Al4V specimens before and after heat treatment. The annealing process in the SLM Ti6Al4V produced acicular microstructures purported to generate more stable oxides, while the equiaxed wrought Ti6Al4V allowed alloying element segregation. The fast cooling inherent of the SLM process enabled metastable α′ phase to form, which increased corrosion susceptibility, as observed for the AP Ti6Al4V. Therefore, according to Xu, et al., the annealing of SLM Ti6Al4V at higher temperatures allowed for the reduction of the α′ volume fraction, decrease of α/β interface area, and agglomeration of the β phase. These microstructural changes deriving from the annealing of SLM Ti6Al4V were attributed in the improvement of their performance against corrosion through reduction of their i_corr.

The effect of heat treatment on the corrosion of SLM Ti6Al4V in 3.5 wt% NaCl was investigated by Yang, et al.¹¹⁸  Electrochemical characteristics of SLM Ti6Al4V before and after heat treatments were compared to their wrought counterpart. Heat treatment at 750°C was shown to decrease the corrosion rate of SLM Ti6Al4V to ∼2 × 10⁻³ mm/y in comparison to wrought (4 × 10⁻³ mm/y) by controlling microstructural transformations (i.e., altering grain size, type, and morphology of the constituent phases). Volume fractions of metastable α′ phase were hypothesized by Yang, et al.,¹¹⁸  to be reduced with the heat treatment of SLM Ti6Al4V, hence improving their corrosion resistance. This hypothesis is in accordance with the findings presented by Xu, et al.¹¹⁷ 

The tribocorrosion of SLM Ti6Al4V was investigated by Chiu, et al.,¹¹⁹  in SBF by altering SLM parameters. The laser scan speed, laser power, and hatch distance were altered to change the surface energy density (i.e., laser energy input per area) to produce the SLM Ti6Al4V specimens. Because none of the SLM parameters were fixed, any comparison of trends regarding the effect of energy density on corrosion was rather difficult, although significant change in corrosion resistance between the analyzed SLM Ti6Al4V and wrought counterpart was not observed. Wear loads of 1 N, 2 N, and 5 N were applied on both the SLM and wrought Ti6Al4V. The tribocorrosion was evaluated by the corrosion/wear ratio (C/W) and was found to be lower for wrought Ti6Al4V (< ∼10%) than for SLM Ti6Al4V (> ∼10%) for all of the loads applied.

Tribocorrosion of DLD Ti6Al4V in PBS solution in comparison with their hot pressed (HP) and cast counterparts was reported by Buciumeanu, et al.¹²⁰  Corrosion was assessed by the electrochemical potential before and simultaneous to wearing the surface. The OCP values measured for the undamaged samples were −0.03 V_SCE for the DLD, −0.25 V_SCE for the HP Ti6Al4V, and −0.36 V_SCE for the cast Ti6Al4V. The damage on the surface of the analyzed samples caused the OCP to drop to values around −0.61 V_SCE, −0.72 V_SCE, and −0.91 V_SCE, respectively. After wear, the OCP returned to the initial values of the undamaged surfaces for all specimens. The DLD Ti6Al4V was deemed less susceptible to corrosion than the HP Ti6Al4V and cast Ti6Al4V before, during, and after surface wear. The correlations between the phases present in the microstructure of DLD Ti6Al4V and tribocorrosion characteristics were not provided to enable a better explanation for enhanced wear corrosion.

The corrosion of EBM Ti6Al4V in 3.5 wt% NaCl with pH ranging from acid to alkaline was evaluated by Abdeen and Palmer.¹²¹  The critical pitting temperature (CPT) of EBM Ti6Al4V in neutral 3.5 wt% NaCl was also investigated and compared to its wrought counterpart. The CPT was found to be around 30°C for both EBM and wrought Ti6Al4V specimens in 3.5 wt% NaCl. The pH levels did not seem to significantly affect the corrosion of the EBM Ti6Al4V. The current densities during potentiostatic polarization of the specimens were similar and did not exceed 30 μA/cm². However, electrochemical characterization was not performed multiple times to ensure reproducibility of results, and the microstructure of the EBM Ti6Al4V was not investigated to provide explanatory conclusions for the observed corrosion behavior. Gong, et al.,⁶⁹  investigated the corrosion of EBM Ti6Al4V in chloride-containing electrolyte (1 M HCl) as a function of different build directions (i.e., angles relative to the substrate). The grain boundary length and proportion of α and β phases in the grain boundary of each specimen, built in different directions, were found to directly influence corrosion. The corrosion performance of the EBM Ti6Al4V specimens is not compared to the wrought or cast counterpart in the work presented by Gong, et al.⁶⁹  Almanza, et al.,¹²²  compares the corrosion performance of EBM Ti6Al4V with the ASTM F75 (Co-Cr) alloy in Hank’s solution. Aside from comparing two distinct alloys, the work reported by Almanza, et al., does not evaluate the corrosion of the studied alloys in comparison with their respective (conventional) counterparts. Bai, et al.,¹²³  reported that the resistance to corrosion of Ti6Al4V specimens in PBS solution is enhanced when produced by EBM in comparison to the wrought counterpart. The presence of higher volume fraction of β phase and refined lamellar α/β phase are suggested to be the factors leading to the observed superior corrosion resistance of EBM Ti6Al4V.

Devika, et al.,¹²⁴  conclude that the EBM Ti6Al4V corrosion is superior to their wrought counterpart in SBF. However, the data presented are insufficient to assess the corrosion performance of the studied alloys. The corrosion performance of SLM and EBM Ti6Al4V are compared to their wrought counterpart by Zhao, et al.,¹²⁵  in SBF. The SLM Ti6Al4V presents superior overall resistance to corrosion when compared to the EBM and wrought counterparts. However, the EBM Ti6Al4V is less prone to crevice corrosion than the SLM and wrought counterparts, which is suggested to be related to grain boundary density. Based on the above works, however, the corrosion performance of EBM Ti6Al4V, particularly relative to the body of knowledge on wrought Ti6Al4V, is inconclusive.

The effect of Cu content on the corrosion performance of SLM Ti6Al4V was investigated by Liu, et al.,¹²⁶  in 0.9 wt% NaCl. The increase of Cu content up to 6 wt% was found to be beneficial for the oxide layer formed on the surface of the alloys, which decreased the corrosion rate of the SLM Ti6Al4V by 67% to 82%. The authors¹²⁶  assume that the presence of Cu influenced the oxide layer formation, although the oxide layer characteristics were not investigated. Mahamood, et al.,¹²⁷  investigated the corrosion of DLD Ti6Al4V/TiC composite in 3.5 wt% NaCl. The laser scan speed utilized to manufacture the DLD Ti6Al4V/TiC specimens was varied from 0.005 m/s to 0.05 m/s. The increase of laser scan speed caused the reduction of corrosion rates from 0.323 mm/y to 0.008 mm/y. This reduction in corrosion rate was associated to the morphology and melting efficiency of the carbides present in the alloy microstructure.

The effect of porosity⁶⁸  and Mo content¹²⁸  on the corrosion of SLM Ti-Mo alloys in 0.9 wt% NaCl was investigated by Xie, et al. The results are similar for both studies, whereby the increase of Mo content led to higher porosity (as opposed to a chemically dictated influence) which was associated with reduced corrosion resistance. The SLM Ti-TiB composite was compared to commercially pure Ti for corrosion performance in Hank’s solution by Chen, et al.¹²⁹  The presence of TiB phase facilitated the oxide layer deposition on the SLM Ti-TiB surface, which reduced the average i_corr from 0.75 μA/cm² to 0.22 μA/cm² and increased the E_pit from 0.70 V_SCE to 1.23 V_SCE. Hot corrosion of SLM Ti-Re was investigated by Majchrowicz, et al.,¹³⁰  as function of rhenium (Re) content in a 75% Na₂SO₄/25% NaCl salt mix. Corrosion of SLM Ti-Re did not follow any trends with the increase of Re content. However, the greatest corrosion resistance was observed for the SLM Ti-Re specimen with the highest Re content analyzed (6%). Mohammad, et al.,¹³¹  investigated the corrosion of EBM γ-TiAl in a mixture of Hank’s solution with 10% of fetal bovine serum. The results obtained by Mohammad, et al., were compared to HIP γ-TiAl and commercially pure Ti. The AP EBM γ-TiAl (i_corr = 0.002 μA) presents what is apparently better corrosion performance than the HIP γ-TiAl (i_corr = 0.103 μA) and the commercially pure Ti (i_corr = 0.330 μA). However, the corrosion current values reported are not area corrected current densities, so it is difficult to know if this ranking is valid or meaningful.

Corrosion of a varied group of alloys—SLM Ti6Al4V, DLD 50Ti-35Nb-15Zr, DLD 67Ti-25Nb-8Zr, and DLD pure Ti—was investigated by Wang, et al.,¹³²  in MEM. The AM Ti alloys analyzed by Wang, et al., were compared to wrought Ti6Al4V, commercially produced pure Ti, and a custom cast Ti-29Nb-13Ta-4.5Zr alloy. The passivation of all of the Ti alloys studied by Wang, et al., was tested by scratching the specimens in situ during potentiostatic polarization in MEM. The wrought/cast Ti alloys showed slower repassivation (132 s to 167 s) after scratching during potentiostatic polarization than the tested AM Ti alloys (22 s to 50 s).

Titanium alloys clearly present anisotropic corrosion for dissimilar planes on the AM specimens. Surfaces near planes parallel to the AM build direction tend to show comparatively poorer corrosion resistance. This behavior is often associated with the commonly observed columnar shaped grains on the mentioned planes.^9,19,21,70  Therefore, the AM build direction could play a decisive role to control grain growth and consequently reduce the anisotropic corrosion behavior. Also, the distribution, shape, and morphology of constituent phases in the AM Ti alloys, especially α′ martensite, was found to influence corrosion characteristics. The corrosion of AM-prepared Ti alloys reported to date are either comparable or superior to their wrought/cast counterparts for most studies, although some exceptions were reported; in particular, the corrosion performance of EBM prepared Ti alloys would benefit from more research.

Limited investigations have been performed on the corrosion behavior of AM-produced alloys other than aluminum, iron, and titanium, including cobalt-chromium (CoCr), Ni, Mg, Cu, and high entropy alloys (HEAs). The findings from such studies are summarized in Table 7.^133-148 

In the case of CoCr, Xin, et al.,¹³³  investigated the corrosion characteristics of SLM CoCr in artificial saliva solution—due to the applicability of CoCr-based alloys in odontology—and compared the results with a cast counterpart. The SLM CoCr (i_corr = 0.02 μA/cm²) exhibited similar corrosion characteristics to cast (i_corr = 0.07 μA/cm²) from potentiodynamic polarization in the analyzed environment. In a similar study, Xin, et al.,¹³⁴  analyzed the effect of the artificial saliva solution pH on the corrosion of SLM CoCr. The corrosion resistance of SLM CoCr was similar to the cast counterpart in artificial saliva solution with pH = 5, with charge transfer resistance measured by EIS of 4.25 MΩ·cm² and 3.82 MΩ·cm², respectively. However, the SLM CoCr was less susceptible to corrosion in artificial saliva solutions at pH = 2.5 (charge transfer resistance = 5.83 MΩ·cm²). The surface of SLM CoCr was treated via porcelain firing by Xin, et al.,¹³⁵  and tested for corrosion at different pH levels in artificial saliva solution. Porcelain firing is a common practice in dental prostheses for aesthetic reasons.¹³⁵  Xin, et al., reported that this surface treatment reduced the SLM and cast CoCr charge transfer resistance by approximately 50%, in comparison with the untreated specimens in both 2.5 pH and 5 pH artificial saliva. The porcelain-fired SLM CoCr showed higher charge transfer resistance (2.88 MΩ·cm²) than the porcelain fired cast counterpart (0.78 MΩ·cm²) in pH 2.5. In pH 5 artificial saliva, cast and SLM CoCr presented similar charge transfer resistance of 2.90 MΩ·cm² and 2.88 MΩ·cm², respectively. The total ion release (Co, Cr, and Mo) was measured to evaluate the corrosion susceptibility of the heat treated and non-heat treated SLM CoCr in artificial saliva by Alifui-Segbaya, et al.¹³⁶  The SLM CoCr was heat treated in two steps: 450°C for 30 min, followed by 45 min exposure at 750°C. The heat-treated sample had a total ion release of 16.47 μg/L over 42 d exposure, while the nonheat-treated sample presented a total ion release of 0.63 μg/L over the same period. The release of those ions from potential SLM CoCr dental prostheses could be a hazard for biological applications.¹³⁶  Guoqing, et al.,¹³⁹  investigated the biocompatibility of SLM CoCrMo in artificial saliva and 0.9 wt% NaCl solution in comparison to their cast counterpart. The ion concentration of the electrolyte after 4 weeks was measured to be ∼175 μg/L (Co) and ∼80 μg/L (Cr) for the cast CoCrMo and ∼125 μg/L (Co) and ∼5 μg/L (Cr) for the SLM CoCrMo in artificial saliva. The Co ion concentration was similar for both cast and SLM CoCrMo samples in 0.9 wt% NaCl after 4 week immersion. However, the Cr ion concentration measured over the same period was significantly inferior in 0.9 wt% NaCl, being ∼1 μg/L and 3 μg/L for the SLM and cast CoCrMo, respectively.

Preliminary investigation on the corrosion characteristics of SLM CoCrW and DLD CoCrMo alloys were reported by Lu, et al.,¹³⁷  and Mantrala, et al.,¹³⁸  respectively. Both studies intended to correlate the laser parameters utilized in each manufacturing process with corrosion performance. Lu, et al.,¹³⁷  found that SLM CoCrW resistance to corrosion was enhanced in 0.9 wt% NaCl when the alloy was produced with laser scan speeds between 700 mm/s and 900 mm/s. The laser scan speed of 700 mm/s showed lowest i_corr (7.9 nA/cm²), with a small increase at 900 mm/s (9.8 nA/cm²). The SLM CoCrW sample produced with laser scan speed of 300 mm/s showed an i_corr of 17.4 nA/cm², and a significant increase of i_corr was measured for the sample produced with laser scan speed of 1,500 mm/s (70.8 nA/cm²). Mantrala, et al.,¹³⁸  reported that 200 W laser power with a 10 mm/s laser scan speed at 5 g/min powder feed rate was optimal to produce DLD CoCrMo resistant to corrosion in 3.5 wt% NaCl with i_corr of 0.013 μA/cm².

In vitro corrosion of SLM CoCrMo was investigated by Hedberg, et al.,¹⁴⁰  in PBS and PBS + bovine serum albumin, and compared to their cast counterpart. The SLM parameters were altered in order to optimize the corrosion resistance of the SLM CoCrMo. The metal release during exposure over 2 h and 168 h was utilized to evaluate corrosion susceptibility. The total amount of metal released by the SLM CoCrMo (∼0.2 μg/cm²) was lower than the cast counterpart (∼0.5 μg/cm²) when immersed in PBS for 2 h. After 168 h, the total metal release in PBS increased to approximately 0.5 μg/cm² and 1.2 μg/cm² for the SLM and cast CoCrMo, respectively. The SLM CoCrMo specimens immersed in PBS + bovine serum albumin showed metal release around 0.6 μg/cm² after 2 h and 168 h, whereas the cast counterpart presented metal release of ∼1.2 μg/cm² after 2 h and approximately 0.8 μg/cm² after 168 h. Hedberg, et al., suggested that the improved corrosion resistance of the SLM CoCrMo could be attributed to the segregation of Mo to the boundaries of cellular structures, which suppresses the formation of phases that preferentially dissolve in this alloy system (Cr-depleted regions and carbide-rich boundaries).

Corrosion of SLM and DLD NiTi was investigated by Marattukalam, et al.,¹⁴¹  and Ibrahim, et al.,¹⁴²  respectively. Marattukalam, et al.,¹⁴¹  aimed to further improve the corrosion resistance of SLM NiTi by altering the laser power and scan speed utilized during the manufacturing process. The increase of laser power from 200 W to 400 W did not affect the i_corr of SLM NiTi in Ringer’s physiological solution for scan speeds of 10 mm/s and 20 mm/s. The E_corr values moved toward nobler values by approximately 100 mV_SCE for the SLM NiTi produced with scan speed of 20 mm/s, in comparison with the specimens produced with scan speed of 10 mm/s. The increase of laser power also increased the E_corr of the SLM NiTi specimens by ∼30% for both laser scan speeds. Although the corrosion resistance could be improved by the laser scan strategy, no comparison was performed with conventionally produced NiTi alloys. Ibrahim, et al.,¹⁴²  analyzed the effect of porosity on the corrosion of DLD NiTi by measuring the Ni ion release after immersion in SBF for 3 d. The conventionally fabricated NiTi and dense DLD NiTi (porosity not provided) presented similar Ni ion release of 59.45 ppb and 64 ppb, respectively. The porous DLD NiTi was produced with porosities of 15, 25, 35, and 50% and presented Ni ion release of 127.94, 176.35, 194.3, and 196.8 ppb, respectively. Therefore, the increase of porosity enhanced the susceptibility of DLD NiTi to corrode in SBF.

Sarmiento Klapper, et al.,¹⁴³  reported the corrosion performance of SLM 718 in a highly corrosive environment containing 13 wt% NaCl solution at 80°C. The SLM 718 was also produced with different build orientations to analyze their effect on corrosion. The resistance to pitting of SLM 718 was found to be inferior to the conventionally produced counterpart (not specified if wrought or cast). The E_pit of the conventionally produced Alloy 718 presented an average value of 416 mV_SHE, whereas a large amount of scatter was detected for the E_pit of the SLM 718 with potentials varying approximately from 250 mV_SHE to 450 mV_SHE. The build direction did not appear to affect the overall corrosion characteristics of the SLM 718 in the studied environment.

Shuai, et al.,¹⁴⁴  reported successful production of a SLM Mg-Zn-Zr (ZK60 [UNS M16600]) alloy, and the corrosion performance was subsequently investigated in Hank’s solution. The laser input energy density was varied in order to optimize the corrosion performance of the SLM ZK60. The hydrogen evolution rate reflecting on the alloy dissolution was optimal for the SLM Mg-Zn-Zr specimens produced with energy densities of 600 J/mm³ (0.006 mL·cm⁻²·h⁻¹). The cast ZK60 presented significant higher hydrogen evolution rate of 0.154 mL·cm⁻²·h⁻¹, hence lower resistance to corrosion than the SLM ZK60. The corrosion of Mg WE43 (UNS M18430) produced by SLM was investigated by Li, et al.,¹⁴⁵  in SBF + 5% fetal bovine serum. The specimens were assessed by measuring the hydrogen (H₂) release over 28 d immersion and by potentiodynamic polarization at different exposure intervals over 14 d. The temporal evolution of hydrogen release presents a linear slope over the analysis period, suggesting a constant corrosion rate during the exposure test. The potentiodynamic polarization of the SLM Mg WE43 showed a maximum i_corr of ∼0.6 μA/cm² after 1 d of exposure, which was reduced to approximately 0.2 μA/cm² after 14 d. Li, et al., suggest that the formation of calcium (Ca), Mg, and phosphorus (P) products were responsible for the reduction of corrosion current density over time during the electrochemical testing, which is not atypical in such an SBF electrolyte.

The corrosion of DLD Cu-30Ni was investigated by Bhattacharya, et al.,¹⁴⁶  following ASTM Standard G31. The blown powder composition was gradually altered from the substrate composition of C71500 to Cu-30Ni. Bhattacharya, et al.,¹⁴⁶  report that the C71500 presented the lowest corrosion rate (0.08 mpy), whereas the DLD Cu-30Ni (i.e., 100% Cu-30Ni) presented the highest corrosion rate (0.36 mpy). The lower corrosion resistance was associated to the high porosity levels on the DLD Cu-30Ni.

There are limited studies on the corrosion of AM-prepared HEAs; however, the corrosion of HEAs produced by EBM and DLD were reported by Fujieda, et al.,¹⁴⁷  and Wang, et al.,¹⁴⁸  respectively. The EBM CoCrFeNiTi was analyzed for the effect of heat treatment on corrosion performance in 3.5 wt% NaCl at 80°C.¹⁴⁷  Water quenching post air exposure at 1,120°C for 3 h was observed to improve the resistance to pitting of the EBM CoCrFeNiTi from the as-built condition. However, a corrosion comparison between the EBM CoCrFeNiTi with their cast or wrought counterparts was not performed. Wang, et al.,¹⁴⁸  investigated the corrosion of DLD AlCoCrFeNi in 0.6 M NaCl solution for different heat treatments and also compared the results with the corrosion performance of Type 304L stainless steel for reference. The DLD AlCoCrFeNi resistance to pitting was enhanced for aging at 1,200°C; however, the overall corrosion performance was inferior to Type 304L stainless steels.

Given the great interest in AM at present, the variety of alloys produced by AM is, almost daily, extending beyond the number of investigations reviewed herein. Such emerging studies have been identified as necessary, as the review herein has identified that there exist numerous knowledge gaps in the context of all of the alloy systems and variables reviewed.

The review herein has indicated that corrosion has been considered in a number of studies that are concerned with AM metallic alloys. In many cases, the studies have been quite specific and not systematic, or of insufficient depth (in terms of the subject of corrosion) to provide general findings. As a consequence, the following points may be made.

• •

  Corrosion studies of AM-prepared alloys to date have investigated corrosion in a vast range of electrolytes, with different compositions and different pH, making cross-study comparisons (even for the same alloys) difficult.

• •

  Similarly, studies to date have investigated corrosion using a range of different corrosion test methods (such as polarization testing, impedance spectroscopy, exposure followed by electron microscopy, etc.), making cross-study comparisons (even for the same alloys) difficult.

• •

  Of the studies reviewed herein, there is a vast range of AM process parameters utilized. Various laser powers, scan speeds, etc., were used even for the same alloy on the same AM instrument, and indeed the variations were even more vast factoring in the variety of AM instruments utilized in the works reported in this review

• •

  Following on the preceding point, there is an obvious lack of standards for which baseline or standardized tests are executed. This situation is not only the case for the corrosion testing itself, but for the AM process.

• •

  A number of studies present results for alloys that have a wrought (or cast) counterpart, while others do not compare the AM sample with the fully dense wrought (or cast) counterpart. The latter means that the rationalization of conclusions is weakened. Furthermore, there are instances where comparison of alloys has been performed when the AM-prepared alloy is not from the same alloy or powder.

• •

  There are, in some cases, studies where feedstock powders have been either argon- or nitrogen-atomized, and build chambers either argon- or nitrogen-purged, making the matrix of variables that may influence the final consolidated component rather vast (even for the same alloy produced).

• •

  There exist very limited data on the EAC of AM-prepared alloys. Similarly, there are very limited data relating to the exposure of AM-prepared alloys to nuclear conditions.

• •

  Similarly, there is a distinct lack of information on hydrogen effects for AM-prepared metals and alloys.

• •

  To date, there are essentially no reports of “very long-term” exposure, such as the data that are available from exposure site testing, or from extended test periods (i.e., beyond many weeks and into the relevant exposure timescale of months and years).

• •

  In essentially all studies, the results (and conclusions) are “pigeon-holed,” in the sense that there is little or no attempt to generalize findings from critical appraisal of accompanying literature, or by including supplementary testing to compare with the conditions in like studies.

• •

  The study of corrosion of AM-prepared Al alloys has been limited to the Al-Si alloy system to date.

• •

  There have been no studies that have studied corrosion on the range of length scales that correspond to the length scale of AM microstructures, or defects as outlined in Figure 4. Not all studies have presented results with “control materials” or relationships to model materials, meaning a fundamental understanding of key factors in AM alloy corrosion are yet to be determined.

• •

  The literature has a lack of separation of variables; i.e., it is complex to generate trends from changes in porosity and chemistry simultaneously.

• •

  The understanding and control of surface finish is essential for a wide range of applications. However, the surface finish of AM-produced alloys has yet to be optimized during production, while post processing also remains a nascent field with little progress to date.

Based on the above points, it is clear that there exists significant future work to be executed in the broader understanding related to the corrosion of AM-prepared metals and alloys. One means by which this could be accelerated is by standardized testing procedures (which may be aided by professional bodies, such as ASTM in the United States or any equivalent), and by relating the corrosion observed to key microstructural (i.e., deterministic) features where possible. Separation of variables and systematic studies are also recommended. A coherent understanding of the corrosion of AM-prepared metals and alloys will be essential in the assessment of novel or emerging “AM-only” alloys, for which there is no benefit of a cast or wrought counterpart to compare.

• •

  It is clear from the numerous works reviewed herein, in the context of corrosion but also more generally in the context of AM, that a range of unique microstructural defects can form in metals/alloys. Such microstructural defects are distinct in comparison with those of cast or wrought alloys, and span many orders of magnitude in length scale. Defects such as solute segregation or dislocation networks are on the order of atoms/nanometers, while porosity or cracking can be micrometer to millimeter scale, and residual stress can be on the scale of meters.

• •

  It was revealed that many alloys are capable of being successfully produced by AM processes. It is also possible to make the very general—and in the context of this review, extremely important—statement that the vast majority of studies revealed that corrosion performance was affected for AM-prepared alloys and is not identical to performance in wrought counterparts. As a consequence, AM plays a significant role in corrosion performance. There are in some cases beneficial outcomes, such as increased pitting resistance in austenitic stainless steels, while in the case of Ti alloys, the influence of AM was overall minimal. Conversely, there are numerous examples of detrimental influences arising from AM, and high variability in findings (such as in the case of Al-Si alloys reviewed herein).

• •

  Overall, there are significant prospects for AM to contribute to the development of durable alloys, and the prospect of new and superior alloys. In general, the metallurgy of AM-prepared alloys is also in need of deeper fundamental understanding, with evidence of metastable phases in AM alloys, and rapid solidification microstructures (and associated phases) revealed.

• •

  The number of unique variables in the AM process that may influence corrosion is high with many degrees of freedom, ranging from laser power to build orientation to name but a few, that challenge straightforward identification of structure property relationships—and at this stage a unified (or holistically meaningful) understanding is difficult to obtain. In order to approach a unified understanding, the need for standards—not only standardized testing, but also standards in specimen preparation—is critical.

ST and NB gratefully acknowledge financial support from the Office of Naval Research (ONR) and Office of Naval Research Global (with Dr. Airan Perez and Dr. Pae Wu as Scientific Officers). JRS and RGK also acknowledge support from the ONR (with Dr. Airan Perez and contract monitor). The work was also supported by Woodside Energy and the Pump Priming Scheme (Advanced Engineering Platform, Monash University Malaysia).