1. Introduction
Additive manufacturing (AM) is a highly developed fabrication technique, which fundamentally creates a three-dimensional object using a layer-by-layer build process. The first recognized patent for AM technology was by Charles Hull for what became known as stereo lithography. He is also credited for the development of the standardized interfacing model format, STL, which is also named after this manufacturing process. As these early days, AM has been gaining in popularity because of the inherent advantages when compared with conventional manufacturing methods such as casting and machining. As designs are made additively rather than by subtractive means, there can be substantial saving on raw material consumption and wastage. Equally, for prototyping purposes, there are almost a limitless number of possibilities to examine alternative design iterations. However, the true power of AM is the symbiotic relationship to modern computer aided design (CAD) and the power to incorporate design complexities. By contrast, traditional manufacturing processes require considerable cost, both in terms of time and expense, when complexity is increased (Gibson et al., 2010). The typical phases associated with the AM production cycle are illustrated in Figure 1. The first stage is the digital modeling using some form of CAD. The digital model is then typically converted into an STL file, which has become the current AM standard. This is then converted into a machine code format, containing information relating to the build parameters for a given printer, before manufacturing commences. Depending on the specific AM technology either a reusable print bed can be used, as with many polymer-based technologies, or a sacrificial bed is used, as with metal printing techniques. With reusable beds, printed parts can simply be removed for post-processing. However, with metal printing, wire cutting is normally required to remove parts from the sacrificial bed. Following part removal, post-processing may be required to remove excess build material used to support overhanging structures or to improve the surface finish and integrity of the part, before it is ready for use (Gibson et al., 2010).
Several mainstream industries have taken advantage of AM technology owing to its uniqueness both regarding prototyping and production.
Historically, the automotive industry were the primary users of AM to increase the speed of new product development and rapidly send parts to market. Equally, the ability to create highly complex, high-performance parts has seen several aerospace companies embrace the use this technology for production of parts within airplane cabins and engine components (Joshi and Sheikh, 2015). Additionally, medical practitioners and device companies have leveraged the benefits of AM technology, initially creating medical models for preclinical planning (Stoker et al., 1992). Following from this early use, the sector has matured rapidly and now allows for the fabrication of implants (Mohammed et al., 2016; Osman et al., 2017), orthotics (Chen et al., 2016; Fitzpatrick et al., 2017) and prosthetics (Mohammed et al., 2017). It is believed that as AM matures over the coming years, usage of this technology is likely to become more widespread and likely capture a greater portion of the global manufacturing market (Campbell et al., 2012).
The majority of commercial applications use polymer or metallic printing, with metal printing currently demanding most attention (Gibson et al., 2010). There are three common metal AM systems according to American Society for testing and materials (ASTM) standard; powder bed fusion (PBF), directed energy deposition (DED) and sheet lamination (Terminology, 2015). The most common is PBF with processes such as electron beam melting (EBM) (Murr et al., 2012), selective laser melting (SLM) (Frazier, 2014) and laser engineered net shaping (LENS) as the most well-known DED technologies. PBF and LENS share similarities in that they both perform melting of metallic powders using a directed thermal energy source, such as a laser or an electron beam, to complete metallurgical bonding for each successive layer (Frazier, 2014; Syed et al., 2005).
PBF systems normally comprise a confined thermal beam, powder bed, movable table/fabrication piston, powder delivery system and a roller, as can be seen in Figure 2(a) (Thompson et al., 2015). Powder is melted selectively by a laser or electron beam with the scanning or melting pattern comprising pulses or continuous lines. Once one layer is completed, the table moves down and a roller or scraper distributes a new layer of powder before the process repeats to selectively melt the subsequent layer. The primary advantages of this process are the ability to produce high-resolution features, alongside the control of a part’s dimensional accuracy (Frazier, 2014; Vandenbroucke and Kruth, 2007; Yadroitsev et al., 2007).
However, a typical DED system generally comprises a laser, material delivery nozzles, fixed substrate and adjustable deposition head, as can be seen in Figure 2(b) (Thompson et al., 2015). By contrast to PBF systems, energy delivery and material deposition are focused on the same region, allowing for greater control of material feed. Additionally, the print platform remains fixed and the laser beam is moved in the Z direction. As with PBF, DED processes require the use of inert gas to reduce oxidation rates (Thompson et al., 2015). The most common DED process is LENS, where powder delivery is controlled using up to four delivery nozzles (Palcic et al., 2009), as illustrated in Figure 2(c). Variations of this technology exist and are largely centered on the coaxial beam and powder delivery (Wen et al., 2009; Pinkerton, 2007; Syed et al., 2006; Keicher and Miller, 2001). It is believed that the differing directional powder delivery systems could result in different mechanical properties and dimensional accuracies. However, no specific strategy has proven to be superior. With the DED process, there are three types of lasers, which are typically used; CO2, Nd, namely, YAG and fiber lasers. The choice of laser significantly influences the powder consolidation during the build process, which, in turn, influences the micro-structure and mechanical properties. This is primarily attributed to laser absorptivity of a respective material, and which, in turn, depends on the laser wavelength. Equally, the operative metallurgical mechanism for powder densification is determined by the input laser energy density (Gu et al., 2012; Kobryn and Semiatin, 2001).
With respect to DED and powder feed processes, there are current a number of excellent reviews that have been conducted. Frazier (2014) explored the material science, processes and business consideration of metal AM. Lewandowski and Seifi (2016) provided a review of mechanical properties of all metal AM process. In addition, Herzog et al. (2016) reviewed several materials microstructure of samples deposited in SLM, EBM and DED. Despite the insights gained all reviews have neglected an in-depth discussion of build process parameters. Some insights on DED processes were presented by Thompson et al. (2015), but were limited to discussions primarily on different types of DED and their build attributes such as solidification, morphology, etc. and not focused on the effects of varying specific LENS process parameters. In this review, we shall focus primarily on fiber laser-based LENS technology and discuss in detail facets relating to the processing parameters and their potential optimization. The review will have the following structure. The LENS process will be briefly outlined. Hence, the work on the controllable process parameters is then reviewed. This is followed by a survey of the parameters that have an effect on the LENS process but are not controlled by the process. This leads to a discourse on the outputs of the LENS process and a discussion of the future directions of the LENS process.
2. Laser engineered net shaping
Prior to any in-depth discussion on process parameters, it is useful to explore the evolution, advantages, disadvantages, post processing and application of LENS to provide a clearer picture of this technology’s capability.
2.1 Laser engineered net shaping evolution
To appreciate the evolution leading to the LENS technology, we should give credit to laser cladding. Laser-based material processing was a significant area of development in the 1970s when power and efficiency of commercial lasers grew rapidly. Laser cladding was first used by Gnanamuthu at Rockwell International Corporation in California. Steen and Weerasinghe at Imperial College, UK also had a notable impact on laser cladding technology, presenting laser cladding using powder injection (Toyserkani et al., 2004). Powell (1983) also investigated this powder laser cladding technique and in 1988, he and Steen, developed a theoretical model and analysis technique for pre-placed laser cladding (Powell et al., 1988).
In the late 1990s, laser cladding for rapid prototyping was introduced. At the University of Illinois, Urbana, the Mazumder research group developed a method for rapid prototyping, which was later named direct metal deposition and later called DED. Their research evaluated building parts in one and two dimensions. Many other research and development groups investigated methods for prototyping metallic parts based on laser cladding by powder injection. In the late 1990s, a development group at Sandia National Laboratory, operated by Lockheed Martin Corporation and funded by US Department of Energy, initiated research toward the development of a laser near-net-shape fabrication method. Afterward on the technology was called LENS (Toyserkani et al., 2004). LENS has since been commercialized by Optomec in 1998 and is still under continued development.
2.2 Advantages of laser engineered net shaping
Salient features of LENS technology are the unconstrained powder feed, which in some LENS models can realize five-axis control. Additionally, the typical build volume is comparatively large and can be 900 × 1,500 × 900 mm (LENS 1500, 2018). Compared to most commonly used PBF technologies (Salmi et al., 2018), this allows for the creation of much larger parts while simultaneously reducing relative powder demands (Dutta and Froes, 2015).
LENS, machines generate smaller melt-pools, allowing for higher solidification/cooling rates as compared with conventional manufacturing processes like casting (Balla et al., 2010a; Balla et al., 2010b). However, a truly defining feature of the LENS process is the ability to fabricate with multiple materials and create functionally graded materials, owing to the multiple powder feed lines (Espana et al., 2010).
LENS can also be used to fabricate products using materials with high melting temperatures, which have a greater tendency to oxidize using conventional manufacturing methods. It can build such elements in the absence of oxygen to create solid or non-uniform porous structures (Balla et al., 2010a; Balla et al., 2010b). In addition, manufactured parts are found to be more robust, less brittle and less prone to cracking at low stress (Bandyopadhyay et al., 2010). This is believed to be due to the LENS process realizing a more defect-free interfacial bond between the layers for optimization within a given process window (Balla and Bandyopadhyay, 2010). Another possibility with the LENS process is the creation of complex porous structures. Porosity can be introduced not only through design architecture but also inherently within the bulk of the processed material itself.
Furthermore, the LENS build process is not constrained by the requirement to commence building on a level platform, as with PBF, providing opportunity to build upon pre-existing parts. This can be useful when considering repair of worn or damaged components (Frazier, 2014).
2.3 Disadvantages of laser engineered net shaping
All metal AM systems exhibit uneven heating and cooling due to the large temperature gradients between the ambient environment, laser focal point, part layers/build plate and the melt-pool. This can result in difficult to predict thermal control. The situation for the LENS is exacerbated by layers closer to the substrate having a higher cooling rate in comparison with higher layers. Unlike PBF, there is no unmelted powder around the melt-pool to help create a more even thermal gradient, which would consequently decrease residual stresses. Hence, residual stress is significantly higher with LENS, which could possibly affect part precision or even lead to part collapse during deposition (Wong and Hernandez, 2012). In addition, PBF can use the unmelted powder as a support structure while LENS cannot realize this feature.
Another effect of the uneven heating and cooling rate in LENS can be non-uniformity in microstructure and macrostructure with respect to the height of build. Furthermore, complex thermal wave effects have been observed during deposition and when combined with build height irregularities, have been found to significantly affect mechanical properties of a manufactured part (Balla et al., 2010a; Balla et al., 2010b, 2010b; Hofmeister and Griffith, 2001; Yadollahi et al., 2015). It is noted that this further influences the mechanical properties differently for different materials (Amine et al., 2014; Yan et al., 2016). Indeed, the dimensional accuracy of a LENS manufactured product has always been a challenge, with no clear general protocol to attain satisfactory build outcomes particularly when fabricating complex geometry parts. As with all additive based build processes, each successive deposition highly depends on the previous layer geometry and the height. In LENS, if the previous layer is thinner or thicker than designed, then the dimension of melt-pool may vary due to an incorrect laser focal point (Pi et al., 2011; Lu et al., 2010; Shamsaei et al., 2015). Unsurprisingly, the build accuracy is entirely related to the optimization of build process parameters for a given metallic powder to produce a respective model at a desired layer height. Consequently, to resolve build irregularities more time is generally needed for post processing as compared to PBF technologies.
In addition, based on the nature of AM there is a staircase effect that effects surface roughness as a result of manufacture layer by layer (Kaji and Barari, 2015). On DED in particular, there could be many partial melted particles, which are associated with the surface. This could increase the surface roughness for deposited parts (Dinda et al., 2008). Low surface quality can negatively affect the mechanical properties of the finished part. For instance, by increasing the surface roughness the fatigue life decreases on EBM fabricated Ti-6Al-4V parts (Frazier, 2014). On samples produced by LENS, the surface roughness is sometimes higher than EBM (Zhai et al., 2016), therefore, potentially leading to lower fatigue life when using the LENS process.
2.4 Laser engineered net shaping post processing
Annealing and heat treatment are the most common forms of post-processing researchers currently conduct on LENS fabricated parts (Yadollahi et al., 2015; Sterling et al., 2016; Durejko et al., 2016; Zhai et al., 2016). Yadollahi et al. (2015) found a protocol of heat treatment for 2 h at 1,150°C followed by air cooling of stainless steel 316 results in a homogeneous microstructure, removing laser track footprints and deposition layer artefacts. Other studies found heat treatment of Ti-6Al-4V for an hour at 760°C within a vacuum followed by air cooling led to stress relief and a decomposition of martensitic phase, resulting in a significant increase in tensile strength but decrease in ductility (Zhai et al., 2016).
In addition to heat treatment, hot isostatic pressing (HIP) is another post processing technique that has been studied. Joseph et al. (2018) using LENS deposited three high entropy alloys based on AlxCoCrFeNi with aluminium molar fractions (x) of 0.3, 0.6 and 0.85. They studied the effect of HIP on the microstructure and mechanical properties. Comparing as-deposited and HIP-processed Al0.3, HIP noticeably improved the tensile ductility and work hardening related to dissolution of grain boundary particles that were evident in the as-deposited condition. On the other hand, for HIP-processed Al0.6 the tensile ductility was reduced compared with as-deposited sample. This suggested the development of coarse B2 precipitates in the inter-phase and grain boundaries. Finally, the HIP-processed Al0.85 lost ductility significantly compared with as-deposited samples. This is due to HIP-processing resulting in coarsening of the microstructure and the formation of σ-phase precipitates.
There are other post process treatments such as ageing (Meneghetti et al., 2017; Shun and Du, 2009; Zadi-Maad and Basuki, 2018), homogenizing (Kao et al., 2009) and thermomechanical processing (Stepanov et al., 2015; Kuznetsov et al., 2012). However, to the best knowledge of the authors these have not been used on LENS deposited samples.
2.5 Laser engineered net shaping application
LENS technology can be used to fabricate parts for applications in key industries such as medical and aerospace (Guo and Leu, 2013; Gurrappa, 2003; Camacho et al., 2017; Dutta and Froes, 2017). Compared to many traditional metal AM build processes, LENS has many unique characteristics, which allow for a range of interesting opportunities for part production. The first such attribute stems from the ability of LENS to create functionally graded porous structures, which has been exploited for orthopaedic implant applications. Figure 3(a) shows various LENS fabricated hip stem implants, each with graded levels of intrinsic porosity resulting from the build process to allow for the process of bone osseointegration. Another major advantage of the LENS process is that as the material is delivered to a build area, manufacturing can be performed on objects of uneven surfaces and without the need for a powder bed. This opens up a range of interesting possibilities with respect to using the build process to build additional components or to repair complex, geometry parts. LENS is also highly suited to such applications owing to the high level of environmental control, which eliminates issues relating to part oxidation during manufacturing as parts are processed in an inert environment. Figure 3(b) shows one such example, where LENS was used to repair a bladed rotor. Beyond these standalone advantages, LENS is also very effective at creating high value parts. Figure 3(c) shows a 1/6 scale mixing nozzle for a gas turbine exhaust for Bell helicopter made by LENS.
3. Controllable laser engineered net shaping input parameters
This section discusses more closely the parameters that can be controlled by the machine operator either prior to or during the build process. LENS provides flexibility to use non-proprietary materials in the build process, opening up endless possibilities with respect to existing and future metal alloy combinations. To realize this potential requires considerable concerted effort to refine build parameters based on the intrinsic characteristics of materials and their responses to achieve respective layer heights and resolutions. Critical build parameters discussed will include laser power, scan speed, powder feed rate, hatch distance, laser focal distance and deposition pattern. Ultimately, these parameters and their interplay dictate the overall quality of the final build and require optimization to ensure build integrity.
3.1 Laser power
In the LENS process, the laser is primarily used to liquefy metallic powder to form a melt-pool, which solidifies upon cooling to form the build voxel of a respective layer. This form of building process is not unique to LENS and has widely been exploited in welding (Benyounis et al., 2005; Akman et al., 2009; Sathiya et al., 2012), cladding (Gedda et al., 2002; Smurov, 2008; Yang, 2009; Yevko et al., 1998) and more recently in PBF AM (Khairallah et al., 2016; King et al., 2015; Yadroitsev et al., 2010). Therefore, the influence of laser power has been widely explored and shown to affect mechanical properties of a metal part. The majority of current studies have focused on non-laser welding. Although these studies could be useful for the understanding of laser based processes, there is limited insight due to differing heat distribution characteristics. Therefore, examination of laser power characteristics is an emerging area of LENS research.
Ultimately, laser power is possibly the most important process parameter in LENS, where variation of this parameter can have the most significant effect on a part’s mechanical properties. For instance, when a dense structure is required, the laser power could be increased to avoid partial melting zones (porous structures) but doing so can also lead to undesired microstructural changes (Balla et al., 2009). It has been observed in some cases that grain size grows by increasing laser power, consequently reducing hardness and potentially affecting other mechanical characteristics (Balla et al., 2010a; Kobryn et al., 2000; Wang et al., 2008). For instance, it has been demonstrated for titanium samples that if laser power increases from 250-300 W while other parameters are kept constant, part stiffness increases by more than three times (Krishna et al., 2007). Increasing laser power could also increase the residual stress between layers (Pratt et al., 2008). In some cases, the residual stress can affect material deposition, particularly for immiscible alloys such as when depositing brass on an AISI 410 substrate (Vandenbroucke and Kruth, 2007). Vandenbroucke et al. demonstrated that for powers more than 450 W, brass layer delaminated occurred during the deposition process. However, this was most pronounced for the initial rather than subsequent layers.
Furthermore, laser power can affect surface roughness. Zhai et al. (2016) conducted an experiment on the LENS process using Ti-6Al-4V to study the surface roughness. They reduced laser power from 780-330 W, which resulted in a reduction of surface roughness from 63.9 to 30.6 µm.
3.2 Scan speed
Scan speed seems to be the second most critical process parameter and refers to the relative translational speed of the laser focal point. Increasing the scan speed decreases the energy density imparted on the build surface for a constant laser power and frequency. Similar to laser power, this parameter affects melt-pool size and so potentially influences build integrity and resulting part mechanical properties. By increasing scan speed, the width and depth of melt-pool decreases due to a reduction in the imparted energy from the laser (Ferguson et al., 2015; Xiong et al., 2012), but somewhat counter-intuitively the deposited line has been observed to become wider and thicker (Bandyopadhyay et al., 2010; Kobryn et al., 2000; Krishna et al., 2007; Lewis and Schlienger, 2000).
In theory, increasing scan speed can decrease build time. However, physical limits exist where the resulting decrease in imparted energy on the build surface results in partially melted regions. This may be important particularly for layers consisting of two different materials with respect to the forming of a metallurgical bond (Espana et al., 2010; Balla et al., 2009). Furthermore, another effect of changing scan speed is the change in cooling rate, which can affect the resulting microstructure of the processed material. It has been found that increased scan speeds lead to higher cooling rates, resulting in smaller grain size, as grains have less time to grow (Kobryn et al., 2000). Moreover, it seems increasing scan speed beyond experimentally determined thresholds may lead to increased porosity in the bulk of the material, a lack of fusion porosity and decreased ductility (Pegues et al., 2017). Finally, it seems when scan speed increases, true fracture strain decreases, resulting in less ductility for Ti-6Al-4V. Reduction in ductility is most likely a consequence of a greater fraction of α’ phase due to more rapid heating and cooling along the beam path (Pegues et al., 2017).
3.3 Powder feed rate
Powder feed rate also plays an important role in LENS and relates to the speed and volume of metallic powder introduced to the melt-pool. Similar to the previous parameters, changing this can result in different mechanical properties (Bandyopadhyay et al., 2010; Krishna et al., 2007). For instance, when feed rate decreases, there may be an insufficient powder to fill the gap between two parallel lines, hence creating porous zones. Conversely, high feed rate may result in a thicker layer, which compromises the integrity of the subsequent build layers or introduces part non-uniformities (Kummailil et al., 2005). Another possible effect of excessive powder could be the creation of a masking region surrounding the melt-pool, which can lead to reduced absorption of laser energy (Paul et al., 2006).
Common powder sizes used for LENS range between 50 to 150 µm. Using bimodal powder distribution has the potential to decrease defects, particularly unwanted porosity (German, 1992). The effect of powder size distribution and density still need to be examined for most powder AM processes (Hebert, 2016). It should be noted that powder delivery for PBF processes are fundamentally simpler than DED, and a more uniform powder density distribution can be expected in PBF, as described in Section 1. In the LENS process a rotor takes the powder from feeders and then gas carries it to nozzles. When optimizing the feed rate for a given source powder, the LENS operator runs the feed system to ascertain the average flow rate in grams per min. This value is used in predicting the melt-pool characteristics. This method can realize stable average powder delivery densities. However, there is likely to be minor variances in the instantaneous powder density due to fluctuations in the average particle size and group velocity. The consequence of this is a reduction in build accuracy resulting from minor variances in the melt-pool size. Therefore, there is a need for a better understanding of the governing process to optimize build consistency. Furthermore, this must be related to a new metric we call the “powder capture rate,” which relates to the interplay and synchronicity of this parameter with others in governing the overall melt-pool consistency.
To study the impression of powder capture rate, Haley et al. (2018) investigated the entrance of powder particles to a melt-pool using four high-speed cameras (≥10,000 fps). This resulted in a mathematical relationship to demonstrate how particle self-shielding ultimately limits the powder capture efficiency of LENS.
3.4 Hatch distance
Hatch distance can be defined as a cross-sectional distance between two scan lines (Balla et al., 2010b) and governs the attainable smallest feature size during the AM process. Figure 4(a) illustrates the relationship between scan line, hatch distance and melt-pool size. Effectively the hatch distance must be optimized based on the perceived size of the melt-pool to ensure the consistency of the formed x-y layer within a part. Should the hatch distance be too small relative to the melt-pool, an excess of material will be formed on a layer resulting in build inconsistencies. Equally, when the hatch distance is too large relative to the size of the melt-pool, a lack of material will be formed on a layer leading to localized porosities. Some studies examined this effect to create a controlled porous structure, as can be seen in Figure 4(b) (Bandyopadhyay et al., 2010). Although a porous structure can arise by varying other parameters such as scan speed and laser power, the influence of changing hatch distance results in more profound variations. This effect greatly decreases the part stiffness (Krishna et al., 2007). Additionally, it has been found that using a hatch distance variation could result in greater uniformity in shape and pore distribution within the structure (Krishna et al., 2007; Kummailil et al., 2005).
3.5 Focal distance
With respect to LENS, focal distance refers to the optimal convergence of both the maximum laser intensity and the localized spatial concentration of the powder beam, as illustrated in Figure 5(a). Failure to achieve an optimal focal distance can not only affect the perceived laser power at the build surface but also distribution of the powder, which can impact the geometry of the melt-pool. Maximum laser intensity can be achieved when the substrate/previous layer is located at or near the focal point of the laser. If the working distance is greater than optimal focal distance, the laser power becomes less localized and distributed over a larger area, while the powder spreads over a larger than desired working area. Combined, this results in the melt-pool diameter becoming larger and height smaller, resulting in unpredictable variations in the scan line created (Xiong et al., 2009).
Conversely, if the work distance is shorter than the optimal focal distance, the laser power again becomes less localized and distributed over a larger area. Now, the powder progressively moves away from the desired focal point, until no powder is delivered to the area encompassing the laser. In this instance, the melt-pool diameter becomes progressively larger until a critical point where the powder cannot reach the laser area and the melt-pool is no longer created (Xiong et al., 2009), as illustrated in Figure 5(b). Ultimately, to maintain build consistency the focal distance is critical, however interestingly there exist possibilities to explore the build properties close to the focal point and leveraging these into the build process. For instance, Sharman et al. (2018) reported that the manipulation of focal distance for parts fabricated from titanium aluminide can result in a crack-free deposition. The researchers compared the quality of deposition with different laser power, scan speed and focal distance. Adjusting the focus position of laser above focal distance could control the cooling rate by preheating the powder particles just before they enter the molten-pool. This could result in reduction of cooling rate below the critical level that causes cracking (Sharman et al., 2018).
3.6 Deposition pattern
As with PBF, in LENS the deposition pattern refers to the pattern traced out by the laser during the build process. Some pattern examples are illustrated in Figure 6(A). Deposition pattern appears to affect residual stress, deflection, substrate distortion, macro-structure and mechanical properties such as tensile strength and elongation (Nickel et al., 2001; Yu et al., 2011; Dai and Shaw, 2002). LENS has a default deposition pattern, specifically for bulky items, which can be seen in Figure 6(B). Firstly, the outer contour is deposited (Figure 6B-1), before the laser reverse rasters to form hatch lines. The subsequent layer is then formed by deposition of the outer contour (Figure 6B-2), followed by depostion of another outer contour (Figure 6B-3). Finally, it deposits perpendicular laser raster to form the hatch lines (Figure 6B-4). It is believed that there exist opportunities to explore alternative deposition patterns.
In addition to the investigation of large area planar deposition, Wang et al. (2018) studied the effect on deposition patterns on an angled, overhanging thin wall of 17-4PH stainless steel thoughout the Z build direction. They concluded that in unidirectional deposition the wall structure suffers an uneven surface and cavity at the end of the deposition path. Using a reciprocating deposition method, both ends were more even and symmetical.
4. Fixed input parameters
Beyond the controllable parameters for LENS, there are a number of parameters that are intrinsic and so remain fixed during a respective build process. Generally, such parameters relate to the intrinsic material properties of the powder (type, morphology and purity), the nature of the laser (wavelength), build atmosphere and thermal relaxation properties. Feasibly between build processes these parameters can be controlled and adjusted to varying degrees and so we shall discuss their significance in the overall LENS build process.
4.1 Powder morphology
This parameter is related to the overall geometry of the powder used in the build process. It has been found for both LENS and PBF processes that powder size can affect the quality of deposition and the smallest attainable build feature. Using finer powders potentially reduces the required heat to complete melting. As the fabrication of amorphous structures requires a relatively high cooling rate and complete melting, using a finer powder would be more logical instead of applying higher power (Balla and Bandyopadhyay, 2010). It is recommended to use spherical powder with 50 to 150 μm diameter (Xiong et al., 2009; Zheng et al., 2010). Moreover, there may exist scope to explore other powder morphologies in future studies.
4.2 Powder purity
This can be defined as a level of impurities and porosity in the powders. Impurities can cause nucleation, oxidization (Balla and Bandyopadhyay, 2010) and irregularities in deposited structure (Ferguson et al., 2015). Powder storage environment can also affect purity, particularly during the time when the LENS hopper needs to be filled. Care should be taken to prevent moisture absorption and oxide layer formation, which can impact the predictability of powder melting point (Hebert, 2016). Moreover, after deposition, recover and reuse of unmelted powder may be problematic and the particle size, distribution and oxidation state of the powder cannot be guaranteed and can compromise subsequent builds. Finally, should several different materials be used on the LENS machine, potential cross contamination of powders may become problematic.
4.3 Laser absorption
Laser absorption coefficient of a material is directly proportional to electrical resistance and inversely proportional to the wavelength of laser (Balla et al., 2010a; Espana et al., Balla et al., 2009). The implication of this is that alloys with higher electrical resistance require increased laser power to ensure complete melting during the build process. Additionally, lasers with a shorter wavelength require proportionately more energy to melt a respective powder. The relationship between laser absorption coefficient and electrical resistance can result in partial melting regions during the build process when multiple materials have significantly different electrical resistance. In the multi-material samples of Ta and Ti for instance, Ta absorbs less energy compared to Ti, which can lead to unmelted regions during deposition (Balla et al., 2010a). Also, high electrically conducting materials such as Al and Cu are difficult to fabricate with this method due to their low laser energy absorption coefficients (Balla et al., 2009).
4.4 Atmosphere
The LENS process is generally performed in an inert argon environment to avoid issues relating to oxidation of the fabricated parts. Impurities such as oxygen in the build environment during deposition can result in nucleation. This prevents amorphous structure creation and unwanted pores due to the reaction of molten metals with oxygen (Balla and Bandyopadhyay, 2010). There have been suggestions to use an alternative gas like nitrogen as preliminary findings during deposition of titanium-molybdenum and pure titanium have resulted in increases in micro-hardness compared with when argon gas is used (Borkar et al., 2012; Shishkovsky and Smurov, 2012; Mohseni et al., 2015). We speculate that there will be an increased interest by research groups to investigate alternative gaseous build environments and to examine the resulting microstructural changes.
4.5 Thermal build properties
In LENS the heat dissipation occurs mostly through the substrate or previously deposited layers unlike PBF processes, which also realizes heat dissipation through the surrounding build powder. The consequence of this is that thermal control can be more difficult to manage and is inherent to the respective thermal characteristics of the selected build platform and processed materials. It has been found in previous studies that the cooling rate can be increased by decreasing the temperature of either the substrate or previous build layer, as can be seen in Figure 7 (Zheng et al., 2008). With respect to LENS there are two feasible methodologies that can allow for control of this parameter. Firstly, the build process could be started with a specified substrate temperature. In the study by Baek et al. (2017) examining the high-speed deposition of tool steel M4 on an AISI D2 substrate, it was found that by increasing the preheating temperature, the hardness increased while the strength and toughness decreases. Moreover, the tensile and impact properties deteriorated rapidly at excessively high preheating temperatures (greater than 500°C). However, in many LENS setups there is no accommodation for pre-heating of the substrate beyond the ambient temperature. Furthermore, we can define a delay interval between layer depositions to allow thermal dissipation until an ideal temperature is attained. For instance, in the study by Yadollahi et al. (2015), it was found that by increasing the time interval by a factor of nine, a more uniform microstructure and increased microhardness is attainable for stainless steel 316 L. Finally, Mazumder et al. (1999) studied the effect of gas flow rate on cooling rate. They indicated there is no significant influence of changing gas flow rate on cooling rate.
5. Laser engineered net shaping additive manufacturing outputs
Despite the significant potential as a metal AM process, LENS has been found to create parts with varying mechanical properties. This has been true even when using the same material, thereby limiting its mainstream popularity as a robust and repeatable build process. In an attempt to shed further light on this situation, we will examine more closely the typical LENS build parameters and outputs, specifically the melt pool, scan geometry and resulting build porosity.
5.1 Mechanical properties
As illustrated in previous sections, input parameters have a significant influence on mechanical properties. In fact, there is a complex relationship between the input/output parameters and resulting mechanical properties. Several input parameters can have similar effects on mechanical properties, but some are more influential than others.
One of the main mechanical properties that is very important to understand is residual stress. The main causes of residual stress after DED are the dynamic temperature distribution and cooling/heating rate within the component. These contain high thermal gradients and repetitious/rapid local heat transfer (Saboori et al., 2017). Residual stresses, whose magnitude can exceed the yield strength of the alloy, affect corrosion resistance, fracture toughness, crack growth behavior and fatigue performance (Mukherjee et al., 2017). Increasing laser power could increase the residual stress between layers significantly (Pratt et al., 2008). On the other hand, powder feed rate has an insignificant effect on residual stress (Mukherjee et al., 2017). It is seen that, in Ti-6Al-4V alloy there is a correlation between residual stresses within the building plane and the laser scanning direction in which there is a compressive residual stress at the center and tensile stress at the edges of the component (Saboori et al., 2017). In some cases, the residual stress can affect material deposition, particularly for immiscible alloys such as when depositing brass on an AISI 410 substrate (Vandenbroucke and Kruth, 2007). Table 1 illustrates the relationship between input/outputs parameters and mechanical properties.
5.2 Melt-pool
The melt-pool is an intermediate output that can be examined during the build process and appears to be the most significant measurable output to determine final LENS part quality. The size of melt-pool itself significantly influences solidification and the resulting layer height consistency. Therefore, it is considered one of the most influential factors relating to the build consistency and ultimately dictates the mechanical properties of the built part. Too large a melt-pool causes deep re-melting of the previously deposited material, affecting product shape and even may cause the top portion of the product to collapse. Conversely, too small a melt-pool can lead to porosity in the structure. For the optimal build, a uniform melt-pool size is required relative to the controllable build parameters, to ensure geometrical repeatability and formation of a high-strength bond between the manufactured layers (Beuth and Klingbeil, 2001).
To better understand melt-pool dimensional effects and relations to input parameters, several thermal transformation simulations have been carried out. Wang et al. (2008) simulated a heat distribution model using SYSWELD software on a single wall deposition of stainless steel 410 to obtain a regular melt-pool size through optimization of laser power. Due to residual temperature during the LENS build process, the melt-pool dimensions were shown to increase. They determined a nominal laser power for each pass during ten layers of deposition to obtain an approximate two-dimensional melt-pool size, as illustrated in Figure 8. However, it seems the authors ignored the effect of mass transformation on the melt-pool.
Neela and De (2009) studied thermal behavior of LENS to obtain a constant melt-pool dimension when processing 316 stainless steel by simulating the combined effect of laser power, scan speed and powder flow rate. Figure 9 illustrates the results of their finite element simulation assuming a constant 60 per cent powder capture rate into the melt-pool. As mentioned previously, if the melt-pool dimensions change, the powder capture rate will change accordingly. The authors concluded from this study that the synchronicity of these parameters are vital to avoid over-melting. Additionally, parameters are required to be adjusted dynamically as the build progresses to maintain a uniform layer height. Combining these findings with those in the study by Wang et al., we speculate that the time-correlated function of the laser power is arguably the most influential parameter for melt-pool consistency.
In another study, Qi et al. (2006) worked on a mathematical model to predict length and width of the melt-pool. The model predicted that the length and width are directly proportional to the laser power. However, the model’s calculated length and width were greater than the deposited samples by 22 per cent. The authors assumed that the difference between the simulated material properties of the model and the process were the cause of mismatch between the calculated and actual values.
As a part of a comprehensive mathematical model, done by Han et al. (2004) the effects of laser power, scan speed and powder feed rate were studied on the melt-pool length and the peak temperature. It was found that while the melt-pool dimensions and peak temperature reduce by increasing the powder feed rate and scan speed, they grow by increasing the laser power. The effect of scan speed was explained by the interaction time between the laser and the powder, that is, the material decreases when the scan speed is increased. In terms of the powder feed rate, it was believed that the increasing feed rate can intensify the power attenuation and eventually reduce the melt-pool temperatures and lengths.
Han et al. (2005) studied a simulation to predict the melt-pool length and peak temperature during the deposition of a line. The results showed that while the actual melt-pool lengths fluctuate more than the calculated ones, the error between the mean theoretical and experimental length is only 6-7 per cent. The fluctuation was explained by the delivery of the powder feed during the process. Regarding the peak temperature, the simulated values were greater by 100-250 K, as the material absorption coefficient is assumed to be constant. The material absorption coefficient may be poorly estimated due to the differences in the temperature of the melt-pool.
Another approach has been to predict trends in the solidification microstructure using melt-pool behavior. Bontha et al. (2006) developed dimensionless process maps for thermal gradient and solidification cooling rate for a thin-wall Ti-6Al-4V structure using a two-dimensional moving heat source Rosenthal solution, as illustrated in Figure 10. The authors analytically extracted thermal gradient and cooling rates throughout the deposition process variables and computed resulting melt-pool depth. However, to do so several assumptions were required. Firstly, wall thickness (b) is much less than the height (h) and length (L), hence, heat conduction is restricted to the X-Z plane. Secondly, it is assumed that L and h are sufficiently large such that the steady-state two-dimensional Rosenthal solution for a point heat source traversing an infinite half-space applies. Thirdly, the relative coordinates (X0, Z0) in Figure 10 are related to fixed spatial coordinates (X, Z) at any time (t) as (X0, Z0) = (X-vT, Z), where v is the laser velocity (
They came up with several expressions for dimensionless cooling rate, dimensionless thermal gradient and melting point temperature. The expression for the dimensionless cooling rate
where
and
is the dimensionless normalized melting temperature and is determined by:
Here T0 is the reference temperature, αQ the absorbed laser power, V the velocity, b the wall thickness, L the length, (X0, Z0) the relative coordinates, t the time, K0 is the modified Bessel function, T is the temperature at a location, (X0, Z0) the thermophysical properties and ρ, c and k are the density, specific heat and thermal conductivity of the material, respectively.
By using these expressions, they plotted dimensionless process maps to anticipate microstructure. These process maps could be used for any material system by modifying input thermophysical properties ρ, c, k and Tm.
Figure 11(a) provides useful information about the relationship between the depth of melt-pool, normalized melting temperature and cooling rate. It shows cooling rate reduces significantly from the top of the surface
Studies were also performed to examine the effect of thermal gradient and cooling rate on solidification. Where G = [∇T] and the expression of solidification R, determined as;
In the same study, they examined the effect of scan speed and laser power on grain morphology with respect to the parameters G and R. Solidification maps for Ti-Al6-V4 were determined for variations in laser power and scan speed, as illustrated in Figure 12. It was found that by increasing the laser power, while maintaining a constant scan speed, the thermal gradient decreases and the microstructure changes from fully columnar to mixed columnar and equiaxed. Furthermore, for increased build height, the solidification decreases in most cases. Conversely, it was found that for increased scan speed, while maintaining a constant laser power, the thermal gradient and solidification increases, while the microstructure changes from fully columnar to mixed columnar and equiaxed (Bontha et al., 2006). These findings once again highlight the critical nature of the build parameter synchronicity in controlling the melt-pool dynamics.
In recent years, Optomec (Albuquerque and US) has introduced a new melt-pool sensor for their LENS machines. The sensor monitors the deposition process and control software automatically adjusts laser power to maintain a constant size of melt-pool. This level of melt-pool process control vastly improves uniformity in micro-structure and geometry (Grylls, 2013). In addition, there is a dual-wavelength pyrometer that can measure temperature during the process. Marshall et al. (2016) provided a pyrometer data article showing temperature response of Ti-6Al-4V of a thin wall structure during deposition. The authors stated this information could be used for understanding heat transfer, cooling rate, melt-pool stability and other influential dynamics. However, they did not perform any analysis on the data.
5.3 Scan geometry
Scan geometry is found to influence the part layer thickness in LENS, making it a critical build parameter. Kobryn et al. (2000) attempted to find a relationship between the build height of a scan line, scan speed and laser power for a thin and thick build substrate. It was observed that thin substrates have a greater initial build layer height compared to thick substrates due to heat dissipation factors within the build plate. Conversely, increasing scan speed leads to a thinner scan line where it seems, there was no clear relationship between substrate and layer thicknesses.
The effects of laser power, powder feed rate, scan speed and hatch distance have also been assessed by measuring the scan line height (Figures 13(A) and 13(B)) and a mathematical equation for prediction of the height was offered by Kummailil et al. (2005). It was found that while the built height is inversely proportional to scan speed and hatch distance, it is directly proportional to laser power and powder feed rate. It was concluded that the feed rate and scan speed have a higher impact on the build height as opposed to the laser power and hatch distance.
Kummailil et al. also proposed a mathematical model [Equation (5)] for the effect of mentioned parameters on the height:
where H is the height, P is laser power, A is absorptivity of powder, ṁ is feed rate, HD is hatch distance, u is scan speed and a is the correlation coefficient.
Manvatkar et al. (2011) attempted to estimate the dimension of the scan line by calculating the input heat and also heat transfer for the deposited layers. It was found that the geometry of the scan line could be estimated by finding the solidification profile. Equation (6) was used to calculate the height for x, y ≤
where qv is the volumetric heat input, η is the absorption coefficient, P is the laser power, d is the beam distribution,
In addition to the build dynamics of a single scan line, Izadi et al. (2017) studied the effect of process parameters on the construction of cylindrical structures of stainless steel 316 L. Interrogation parameters were based on Taguchi design of experiment and validated against previously derived equations for the energy density in powder based systems (Vandenbroucke and Kruth, 2007; Di et al., 2012; Gu et al., 2013; Jia and Gu, 2014). It was discovered that the geometry of the printed samples varied from one another, resulting in dramatic fluctuations of the observed sample diameter and height (Figure 14). In several instances samples were found to contain a concave form at the upper surface of the cylinder, resulting in several depth variations from what was anticipated. Ultimately, these defects resulted from under deposition of material, which resulted in the part porosity changing from bottom to top. The authors concluded that current mathematical models for the energy density were not sufficient to accurately model DED based build processes and new models require development.
The build height of a deposited line was predicted by a mathematical model and compared with the experimental samples by Qi et al. (2006). The model could prove that, as shown in Figure 15, the height is increased by laser power. This increase is linear except at low power, less than 400 W. The difference between the actual sample height and calculated value was up to 32 per cent, the authors’ assumed this was due to the difference between the theoretical and actual powder concentration distribution and powder capture.
Similarly, He and Mazumder (2007) worked on a three-dimentional model to calculate the build height of a deposited single crack when the laser power, scan speed and powder feed rate are changed. The model could successfully predict that the build height is increased by increasing the laser power and decreasing the scan speed. However, the accuracy of the model decreased at higher power as the specific heat and thermal conductivity were assumed to be constant and the actual powder distribution may be different. The models used in both studies were used for only a single deposited line. More work is needed to analyze multiple layer samples.
Katinas et al. (2018) worked on a simulation predicting the powder capture rate and eventually the dimensions of the deposited line and melt-pool geometry. The results showed a reasonable correlation between the simulated values and two deposited lines done in the experiment. The track width and height errors were 2 and 3 per cent, respectively.
5.4 Porosity
Structural porosity is another important consequence of LENS that can significantly influence both the mechanical properties and the quality of products. In the majority of cases, porosity is perceived as a defect due to a reduction in mechanical properties. For instance, Razavi et al. (2018) performed a comparison study between a porous and non-porous, part built using LENS with different process parameters. He found the fatigue strengths of non-porous samples to be almost four times higher than porous samples. However, they did not study the exact effect of each process parameter on fatigue strength. In some instances engineers can take advantage of the inherent porosity, specifically in medical applications (Sun et al., 2015; Das et al., 2013; Parthasarathy et al., 2011), where there are two important benefits to porous structures. Firstly, the porous structure can decrease the stiffness mismatch between implants and bone tissue, minimizing stress shielding effects. Secondly, porous implants can result in osseointegration, with living cells growing into open pores to provide mechanical locking and assist long-term fixation (Xue et al., 2007).
There are several demonstrated examples of porous structures using LENS, including:
Partial fusion porosity: This results from insufficient/incomplete melting between or along layer boundaries. As this process is chaotic, there is no well-defined pore shape (Krishna et al., 2007).
Gas porosity: This results from gas entrapment in the feed powder, which expands during laser melting. This leads to very low porosity, characterized by spherical bubbles (Kobryn et al., 2000).
Hatch distance porosity: Resulting from pore connectivity and can be controlled by altering the hatch distance; with increased distances resulting in higher porosity (Krishna et al., 2007; Kummailil et al., 2005).
Despite these methods of introducing porosity within the build process, production of accurate porous structures still remains challenging due to the complexity of thermal control in LENS. Given the limited number of studies in this area, particularly for industrial applications, LENS based porous structures warrant further investigation.
6. Discussion and conclusion
LENS AM is an emerging technology with significant potential for the fabrication of metallic parts, with several potential advantages over more common powder bed technologies, particularly with regards to the formation of functionally graded and larger components. It has been discovered that process parameters, their synchronicity and optimization, play a vital part in the effectiveness of LENS, and significant work remains to create a robust framework for production with a high degree of repeatability.
With respect to critical parameters, there are several controllable and inherent variables, all of which impact the final build quality. Of these, laser power, powder feed rate, scan speed and hatch distance appear to be the most influential and have been the primary focus of the limited number of current studies of this technology. Drawing parallels with PBF processes, researchers have found a relationship between energy density, macro-structure and mechanical properties (Vandenbroucke and Kruth, 2007; Di et al., 2012; Gu et al., 2013; Jia and Gu, 2014), where the energy input is proportional to the laser power, scan speed, hatch distance and layer thickness. Researchers of the LENS build process have tried to derive similar examples of the energy density equation. However, they have been largely unsuccessful due to the instabilities in the layer thickness of manufactured parts. This is believed to be because the layer thickness depends highly on powder capture rate, which, in turn, dictates the melt-pool’s dimensions. As the melt-pool dimensions have a strong relationship with process temperature, thermal control and its associated parameters, it is critical to establish a similar mathematical relationship for energy density in the LENS process.
Ultimately our findings reveal there is significant scope for research to improve the build process and address issues such as, deposited part precision, micro-structure quality and part porosity, which hinder LENS becoming an AM standard. We conclude future studies should focus upon:
The effect of the parameters on final geometrical accuracy.
Examination of a wider range of build parameters, as current studies have only examined laser powers between 150-400W, scan speeds of 6-12 mm/min and powder feed rates of 5-10 gram/min.
Development of robust mathematical models reflecting the build process.
Investigating multiple material depositions in a single build.
Investigation of alternative scan patterns.
More precisely understanding the interrelationship between the various process parameters with and across to a wider variety of material.
Developing an algorithm to control parameters such as powder feed rate and scan speed in real-time.
Studying the effect of post-processing on the deposited part to improve mechanical properties and eliminate residual stress.
Investigating the uniformity of microstructure and macrostructure in a deposited part from bottom to top.
Additive manufacturing eight stages
(a) Powder bed fusion schematically, (b) direct energy deposition (LENS). Reproduced from Thompson et al. (2015) with the permission from Elsevier and (c) photograph of a LENS MR-7
(a) Hips fabricated by LENS (Espana et al., 2010); (b) a repaired blisks using LENS (LENS Blisk Repair Solution, 2014) and (c) 1/6 scale mixing nozzle for gas turbine exhaust for Bell helicopter produced by LENS (Guo and Leu, 2013)
(a) Figure illustrating hatch distance relative to the scan line and melt-pool and (b) various samples of designed porous structure
(a) A schematic of working distance illustrating the balance of focusing both the laser and powder beams and (b) illustration of the resulting scan line over different working distances
(A) Deposition patterns for (a) raster, (b) offset out, (c) offset in and (d) fractal and (B) default deposition pattern on LENS process
A graph illustrating the relationship between cooling rate and substrate/deposited temperature for Ti-6Al-4V deposition
(a) Nominal laser powers for each layer to have a uniform size of melt-pool and (b) melt-pool shapes and sizes when the laser beam is in the center of a thin wall for different layers. The average size of the melt-pool is 2.0 mm
A graph illustrating the combined relationship between laser power, scan speed and powder mass flow rate based on a steady melt-pool size for SS316
Schematic of thin wall
Process maps of (a) cooling rate and (b) thermal gradient (
G vs R plots, (a) effect of laser power and (b) effect of scan speed
(A) Images illustrating the influence of different laser power and scan speed on the build height of a single track at (a) 200 W, 6 mm/s, 6.3 g/min; (b) 250 W, 8 mm/s, 6.3 g/min and (c) 300 W, 4 mm/s, 6.3 g/min and (B) images illustrating the influence of different powder feed rate on the build height of a single track at (a) 300 W, 6 mm/s, 7.8 g/min; (b) 200 W, 4 mm/s, 8.8 g/min and (c) 300 W, 6 mm/s, 9.8 g/min
L9 orthogonal matrix of Taguchi design of experiment
Predicted and experimental samples build height comparison
Effect of inputs parameters on outputs parameter and mechanical properties
| Input parameters | Output parameters | Mechanical properties |
|---|---|---|
| Increasing laser power | Decrease partial melting (Balla et al., 2009) | Reduction on hardness (Wang et al., 2008) |
| Increasing scan speed | Width and depth of melt-pool decreases (Bandyopadhyay et al., 2010; Kobryn et al., 2000; Krishna et al., 2007; Lewis and Schlienger, 2000) |
Less ductility (Pegues et al., 2017) |
| Feed rate decreases | Thinner layer (Kummailil et al., 2005) |
Reduction in part stiffness (Kummailil et al., 2005) |
| Hatch distance increases | Porosity volume increases (Krishna et al., 2007) | Reduction in part stiffness (Krishna et al., 2007) |
| Positive focal distance | Preheated powder particles (Sharman et al., 2018) |
May result in crack free deposition (Sharman et al., 2018) |
| Negative focal distance | Affect built geometry (Xiong et al., 2009) | May result in a cracked deposition (Sharman et al., 2018) |
| Deposition pattern | Affect build geometry (Wang et al., 2018) |
Affect residual stress (Nickel et al., 2001) |
| Finer powder | Less heat required (Balla and Bandyopadhyay, 2010) | No studies found |
| Powder purity | Irregularities in build structure (Ferguson et al., 2015) |
No studies found |
| Laser absorption | Partial melting regions (Balla et al., 2010a) | No studies found |
| Atmosphere | Nucleation (Balla and Bandyopadhyay, 2010) | Affect micro-hardness (Borkar et al., 2012; Shishkovsky and Smurov, 2012; Mohseni et al., 2015) |
© Mojtaba Izadi, Aidin Farzaneh, Mazher Mohammed, Ian Gibson and Bernard Rolfe. This work is published under https://creativecommons.org/licenses/by-nc/3.0/legalcode (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.