**1. Introduction**

Metal additive manufacturing (MAM), commonly known as metal 3D-printing, is a process by which complex multifunctional metal parts are produced in a layer by layer fashion using 3D computer-aided design (CAD) models [1–6]. Several MAM techniques are available. They can be separated into two main groups: direct energy deposition (DED) methods and powder bed fusion (PBF) technology [7]. During direct energy deposition, focused thermal energy is used to fuse materials by melting as they are being deposited; while during powder bed fusion, thermal energy selectively fuses regions of a powder bed [7]. DED processes such as wire arc additive manufacturing (WAAM) and laser metal deposition (LMD) can generally be used on existing parts of arbitrary geometry with a relatively high deposition rate; however, the shape complexity is limited. This makes DED processes the preferred methodology for repairing or improving existing parts [8]. On the other hand, on PBF methods such as selective laser melting (SLM), selective laser sintering (SLS), and electron beam melting (EBM), the dimension of the pieces is limited and the starting substrate has to be a flat surface. However, they generally allow the fabrication of pieces with extremely high structural complexity at a

relatively high level of precision. Among the several MAM processes, those utilizing a metal powder feedstock and a laser source to achieve the metal fusion are the most widely used [1–6]. From those, SLM is regarded as the most used and studied MAM method. This is not only because it allows a higher level of precision compared to other MAM techniques, but also because (in contrast to SLS) it allows the full melting of the material, and therefore the production of solid and dense metal parts in a single process (without the need to use binders and/or other post-process furnace operations).

Additive manufacturing is considered one of the enabling technologies for Industry 4.0 [9]. In particular, MAM presented a global market valued at € 2.02 billion in 2019 [10]. This included systems, materials, and services. MAM allows the near-net shape manufacturing of geometrically complex parts such as lattice structures and 3D structures with undercuts or cavities, which is why this technology has found numerous applications in industries such as medical implants, energy, aerospace, and automotive. As an example in aerospace applications, MAM has made possible the re-design of complex fuel injector nozzles (commonly requiring the assembly of more than 20 parts) in a single operation [11,12], as well as the re-design of several other complex engineered parts, resulting in considerable cost and weight reduction. In aerospace, as well as in the automotive industry, MAM is also actively used in prototyping and the fabrication of custom tooling.

Nowadays, a grea<sup>t</sup> number of metals and alloys can be processed by additive manufacturing techniques, depending mainly on the availability of the raw materials as metal powders or metal wires [5]. Amongst these, aluminium alloys are of grea<sup>t</sup> interest for applications requiring high strength and sti ffness to weight ratio, good wear and corrosion resistance, and recycling potential, which is why they are attracting increasing attention of the automotive and aerospace industries [12,13]. The most common Al-based alloys processed by additive manufacturing (AM) either for commercial use or for research purposes are [5,14,15]: AlSi12, AlSi10Mg, AlSi7Mg0.6, AlSi9Cu3, AlSi5Cu3Mg, AA1050, AA2017, AA2024, AA2219, AA6061, AA7020, AA7050, AA7075, and AA5083; next to proprietary industrial alloys like Scalmalloy. From those, Al-Si alloys, and more specifically AlSi10Mg (followed by AlSi12), are undoubtedly the most investigated and commercially used additively manufactured Al-based alloys. These materials, particularly relevant for light-weight and high-strength applications, are widely used for aluminium casting due to the proximity to the eutectic composition (~12.5% Si) [16]. Therefore, they are relatively easy to process by laser applications, which are known to lead to a small solidification range [17]. Additionally, minor additions of magnesium (0.3–0.5 wt.% Mg) are known to induce hardening of the alloy by forming Mg2Si precipitates upon natural or artificial ageing treatments [18]. However, the actual formation of these precipitates on additively manufactured Al-Si parts is still a topic of discussion.

Due to the special conditions of the MAM processes (namely that the metal powder used is already pre-alloyed, and the melting occurs in small pools that rapidly solidify), a very fine microstructure with unique directional growth features far from equilibrium is achieved [1]. This distinctive microstructure, together with other special features and microstructural defects originating from the additive manufacturing process is known to greatly influence the corrosion behaviour of these materials. Sander et al. [19] and Kong et al. [20] reviewed the impact of these special features and defects on the corrosion performance of additively manufactured metals. These works also consider the main corrosion issues of several additively manufactured materials, including some studies on Al alloys. Zhang et al. [17] presented a review of Al-based alloys summarizing the microstructural characteristics and mechanical properties. Chen et al. [21] reviewed the main research studying the corrosion behaviour of selective laser melted Al alloys, classified/structured by the used technique.

The present work reviews the main studies to date investigating the corrosion aspects of additively manufactured aluminium alloys from a general perspective. Some of the defects and intrinsic issues from the additive manufacturing process that can influence the corrosion performance of these materials are discussed. Additionally, the most relevant studies and results concerning the e ffect of microstructure, heat treatments, Si content, surface roughness, and surface treatments on the corrosion behaviour of Al-based alloys are identified and discussed. Moreover, the main aspects concerning the behaviour of these materials to corrosion protection surface treatments such as anodizing are summarized and discussed for the first time. A summary and outlook of relevant directions to be explored in future works are also presented. The majority of the works investigating additively manufactured Al alloys was conducted on SLM materials, and a few studies were realized using SLS specimens. The microstructural characteristics and corrosion behaviours of the materials fabricated using those two methods are remarkably similar. Therefore, the use of "AM Al alloys" in this work refers, in general, to SLM and/or SLS Al alloys.

#### **2. Influence of Defects in AM Specimens on Corrosion Resistance**

Metal additive manufacturing presents, without a doubt, grea<sup>t</sup> potential to become a key manufacturing technology in several industries and daily life applications. One of the conditions still needed to achieve this goal is better control and understanding of the numerous macro- and micro-structural defects introduced by the special manufacturing conditions of these processes. These defects can greatly influence the materials' performance against corrosion [22]. Further investigations are needed to better understand the e ffect of features such as porosity, surface roughness, and residual stresses, among others, on the corrosion resistance of these materials.

Metal additive manufacturing is, in general, a very complex process with a grea<sup>t</sup> number of operation parameters involved (for instance the laser/beam power, scanning speed, the spot size of the power source, hatch distance, layer thickness, powder flow speed in the case of DED processes, etc.). The e ffect of each process parameter in the final printed part cannot be seen independently of the rest. Instead, the energy density (E), which indicates the energy input into the material (see Equation (1)) and is equal to the laser/beam power (P) divided by the scanning speed (v), the layer thickness (d), and the hatch distance (h), was the quantity considered to evaluate the e ffect of process parameters on the properties of the final parts [23]. Previous studies have shown that the energy input is the determining parameter for defining and optimizing the "process window" during additive manufacturing [23]. A low energy input per unit length results in droplet formation and a bad wetting to the previous layer. On the other hand, a relatively high energy input per unit length causes distortions and irregularities due to big melt pool volumes and the balling e ffect. The specific process parameters and, therefore, the energy density used, are usually optimized per alloy type in each AM system to obtain high-density parts with the lowest possible level of porosity. Nevertheless, confirming the fact that MAM is a complex process, other researchers believe that the energy density cannot be the sole parameter analysed during the process optimization [24]. Other parameters such as the scanning strategy, laser spot size, and materials' properties (i.e., thermal conductivity and reflectivity) should also be considered. Moreover, the researchers sugges<sup>t</sup> that the laser power has a greater influence than the rest of the parameters, and therefore should also be considered independently [24]. Furthermore, Leung et al. [25] also demonstrated that the oxidation state of the metal powder could have a grea<sup>t</sup> influence on the formation of defects in the printed parts.

$$\mathbf{E} = \frac{\mathbf{P}}{\mathbf{v} \times \mathbf{h} \times \mathbf{d}} \tag{1}$$

An incorrect choice of process parameters could lead to high levels of porosity within additively manufactured metal parts. Pores in additive manufacturing can be classified into two groups: trapped-gas pores and lack-of-fusion pores. Trapped-gas pores, as its name says, are the result of gas trapped inside the powder particles during gas atomization, or inside the printed part during the actual manufacturing. These pores have a rather spherical shape as can be seen in Figure 1a. Lack-of-fusion pores have, on the other hand, an irregular shape (see Figure 1b) and are much larger than trapped-gas pores. Lack-of-fusion pores appear when there is no complete adherence of the current melt to the surrounding part (when powder particles are only partially molten) [26]. These two types of pores possess intrinsically di fferent characteristics, and therefore, they could a ffect/impact corrosion in di fferent ways and/or to a di fferent extent.

**Figure 1.** Secondary electron image representing defects that can be present in additively manufactured metal parts: (**a**) trapped-gas pores; (**b**) lack-of-fusion pores; (**c**) unmolten powder.

In general, close-to-the-surface pores can compromise the passivation properties of the native surface oxides, and therefore the pitting resistance of the material. This was demonstrated in previous studies for stainless steel [27,28], in which the researchers have shown that the level of porosity and the size of the pores play an important role in the resistance to pitting corrosion. However, no work has been conducted to date to systematically study the effect of porosity (dimension, type, and extent of porosity) on the corrosion behaviour of AM Al-based alloys. Therefore, more focus should be given to this issue in future studies. Moreover, pores might also play an important role in the susceptibility of the printed parts to stress corrosion cracking (SCC), since pores are intrinsically stress concentrators during mechanical loading. Once corrosive species reach the pores, occluded cell conditions could rapidly built-up and promote a premature failure of the printed part. This is a topic that needs more attention in the future.

As in the case of porosity, unmolten, or partially molten powder (see Figure 1c) results from incorrect choices of process parameters. Unmolten powder on the surface of the printed part can compromise the passive behaviour of the material by introducing defects in the native oxide film and greatly increases the surface roughness. There are currently several processes available (such as abrasive blast, shoot peening, electrochemical polishing, etc.) that are used to improve the surface finishing and reduce the roughness of AM metal parts [29,30]. These methods greatly affect the surface state and therefore, the surface energy, which influences the nucleation and growth of the native oxide film. This can, consequently, affect the barrier properties of the native oxide, and hence the passivity of the material.

Another intrinsic feature of additively manufactured metal parts that could compromise their physical integrity is the existence of residual stresses. Residual stress formation in metal additive manufacturing is caused by the high thermal gradients and cooling rates associated with these processes [31]. Particularly high tensile residual stresses are very typical near the surface [31]. These residual stresses can affect the materials' susceptibility to stress corrosion cracking. Previous studies have shown the formation of micro-cracks in as-built AM Al-Si alloys after exposure to corrosive media (see Figure 2) [32,33]. The researchers associated this cracking with the combined effect of the local disruption of the Si network, the selective dissolution of the Al matrix around the disrupted

zones, and the existence of residual stresses from the MAM process [33]. Additionally, surface residual stresses could, in general, influence the surface energy and therefore, the nucleation and growth of the native oxide film, affecting as a consequence the passivity and pitting resistance of the material. Nevertheless, no relevant work has been published so far on the study of the effect of residual stresses on the corrosion behaviour of additively manufactured metals.

**Figure 2.** Secondary electron microscopy image of the surface of an as-built additive manufacturing (AM) Al-Si (AlSi10Mg) after immersion for 48 h in 0.1 M NaCl. The formation of micro-cracks in the heat affected zone next to the MPB can be seen. MPB—melt pool border.

## **3. Al-Si Alloys**

#### *3.1. E*ff*ect of Microstructure on Corrosion Behavior*

Several studies characterizing the microstructure of as-built additively manufactured Al-Si alloys have been conducted in recent years [33–43]. As-built specimens exhibit a fine distribution of Si, forming a three-dimensional network that encloses the primary face-centred cubic α-Al in very small cells (see Figure 3). The size of these cells varies over the melt pool due to the thermal gradient created by the moving heat source. Finer cells are formed towards the middle of the melt pools (MP), while coarser cells are present in the melt pool borders (MPBs) [36]. A marked anisotropy has been described in past studies concerning the shape of these cells. These cells are known to present an approximately round shape in the plane parallel to the building platform (XY), whereas in the plane perpendicular to the building platform (XZ) more elongated cells have been observed [36,42]. A heat-affected zone (HAZ) located right outside the borders of the melt pools has also been identified (see Figure 3). This HAZ, in which the silicon network is partly broken and discontinuous, has been associated with overheating of the underlying layer during the scanning of a newly deposited layer [36].

**Figure 3.** (**a**) Secondary electron image representing the microstructure of additively manufactured Al-Si alloys (AlSi10Mg). The surface parallel to the building platform is represented here. Similar features can be observed in the surface perpendicular to the building platform, but the shape of the cells is more elongated in that case. (**b**) Higher magnification image of a zone in the melt pool border. (**c**) Higher magnification image of a zone within the melt pool. MPB—melt pool border; HAZ—heat-affected zone. (Adapted from reference [43]).

A grea<sup>t</sup> number of works have already shown that the corrosion behaviour of additively manufactured Al-Si alloys is greatly influenced by this special and unique microstructure created during the additive manufacturing process.

#### 3.1.1. Influence of the Melt Pool Borders on Corrosion

Among the different details of the microstructural features of as-built AM Al-Si alloys, the borders of the melt pools have been pointed out to be a key element in the corrosion of these materials. Several studies have shown that after potentiodynamic polarization tests and/or immersion in NaCl solution a particularly pronounced selective corrosion of the α-Al cells in the melt pool borders is observed (see Figure 4) [32,33,44–57]. This selective corrosion attack along the borders of the melt pools has been reported for polished and ground specimens [32,33,44–56], but also for as-built materials presenting a low surface roughness [57], for which the microstructural features are believed to control the electrochemical performance of the material.

**Figure 4.** Secondary electron microscopy image of the surface of an as-built AM Al-Si (AlSi10Mg) corroded in NaCl solution. (Adapted from reference [47]).

The initiation and further propagation of the corrosion attack in the Al cells along the MPBs are caused by the higher driving force for micro-galvanic corrosion between the α-Al and the Si phase in these regions compared to other areas within the melt pools [32,45,54]. Due to the presence of a relatively larger microstructure at the MPBs (see Figure 5), a greater Volta-potential difference has been reported between the Al and the Si phase compared to regions within the MPs (Figure 5c). Nevertheless, Kubacki et al. [51] do not support the assumption of a galvanic couple between the Al and Si phase accelerating corrosion in simulated atmospheric conditions at the melt pool borders. The researchers believe that under these conditions the cathodic kinetics on the Si phase is not fast enough to support the active dissolution of Al. Instead, the selective/pronounced corrosion attack at the borders of the melt pools is attributed to the high discontinuities of the Si network around the MPBs (in the heat-affected zone).

**Figure 5.** (**a**) Topography and (**b**) surface potential map (obtained by scanning Kelvin probe force microscopy—SKPFM) of an area on the surface of a polished AM AlSi10Mg specimen in which a melt pool border is visible (delimited by the discontinuous line). Scan size: 20 × 20 μm2. Colour bar: (**a**) 30 nm range, (**b**) 210 mV range. (**c**) Surface potential profile of the line represented in (**b**). (Adapted from reference [32]).

The morphology of the corrosion attack in as-built AM Al-Si alloys has been described by several researchers to be rather superficial due to the existence of a considerably connected Si network that holds the corrosion from penetrating deeper into the material [32,33,51,54,58]. However, several works have found a large penetration of the corrosion attack along the melt pool borders [46,50–53,55]. Moreover, the formation of micro-cracks during corrosion has been reported in regions next to the MPBs, where heat-affected zones exist [32,33,54,55,59]. As mentioned above, Revilla et al. [33] associated the formation of these micro-cracks to the synergistic effect of the selective dissolution of the α-Al cells along the MPBs, next to which the disrupted heat-affected zones are found, and the existence of internal and superficial residual stresses from the MAM process. Rafiezad et al. [49] also confirmed through intergranular corrosion test an accelerated preferential attack combined with the formation of micro-cracks along the melt pool borders for specimens fabricated using recycled powder. Moreover, while conducting electrochemical tests in NaCl solution, Girelli et al. [59] found that the MPBs/HAZ represent preferential paths for exfoliation-like corrosion to occur.

The schematic in Figure 6 shows the proposed evolution of the corrosion attack in as-built AM Al-Si alloys [33]. Because of the greater Volta-potential difference between the α-Al and the Si phase at the melt pool borders (as measured by scanning Kelvin probe force microscopy SKPFM [32,45,54]—see Figure 5), a higher driving force for galvanic corrosion provokes the initiation of the corrosion attack in these regions. The corrosion spreads superficially to adjacent zones, including the neighbouring heat-affected zones. This corrosion is partially contained by the connected portion of the Si network. However, due to the disruption in the heat-affected zone, the corrosion can propagate further into the material along the MPBs.

**Figure 6.** Schematic diagram representing the process of corrosion initiation and corrosion propagation for as-built additively manufactured Al-Si alloys. The illustrations portray the corrosion process seen from a cross-sectional perspective, with the top side of the images representing the surface exposed to the corrosive medium. The Al phase is represented with green, while black portrays the Si phase. A melt pool border (MPB), where a coarser microstructure is seen, and a heat-affected zone (HAZ), characterized by discontinuities in the Si network, can be observed in the images (**a**). The corrosion attack initiates at the MPB, where there is a larger potential difference between the Al and the Si phase, and therefore, a larger driving force for galvanic corrosion (**b**). Because of the partial containment of the corrosion by the Si network, the attack spreads superficially to adjacent zones, reaching the HAZ (**c**). Due to the disruption of the Si network in the HAZ, combined with the existence of internal stresses from the MAM process, micro-cracks are formed along the HAZ (**d**). In a later stage of the corrosion, additional lateral spreading of the attack occurs, accompanied by further propagation of the crack and corrosive medium through the HAZ (**e**) [33].
