Metal Additive Manufacturing
Current State of the Art in Laser Powder Bed Systems
Additive manufacturing (AM) – the process of building up solid layers of material to form a finished solid part — is an emerging and exciting technical discipline. Also referred to as “3D printing,” many misconceptions exist about the capabilities and promises of the technology.
Yes, AM has the potential to create customized parts. Yes, AM is able to create parts that cannot be practically produced by other traditional manufacturing methods. Yes, AM offers tremendous opportunities for part weight reduction and parts consolidation. And yes, AM will be and has been used to produce some large volume production parts. But AM does not offer infinite freedom in shape; complexity is not “free”; parts do not pass through the printer at the touch of a button or the speed of printing a page of text; and one must carefully design both preprocessing and post processing activities to achieve a finished metallic part from the AM process.
Although many material systems have been printed via AM (including polymers, metals, ceramics and biological tissue), this article will focus on metal AM and its benefits, limitations and applications.
Laser Powder Bed AM Process
A number of metal AM processes have been developed over the past 30 years. Table 1 highlights those recognized by ASTM International  . A thorough description of each of these processes may be found in the annual report produced by Wohlers  . Nearly all of these methods rely on a computer-controlled heat source (typically laser or electron beam) to melt or fuse the metal feedstock and deposit it point-by-point at precise 3D locations. Both wire and powder are commonly used for feedstock. (A notable exception is gaseous deposition  .) Here, we will only focus on the laser powder bed fusion (L-PBF) process, by far the most commonly available method for production of metal AM parts. Electron beam powder bed fusion systems are also available; in principle, they operate in much the same way as laser-based systems, except that the build is completed in a vacuum rather than in an inert gas environment.
Figure 1 highlights the laser powder bed process where a laser is used to melt selected regions of a thin layer (typically 20–50 microns) of metal powder. To keep the molten metal from absorbing oxygen or nitrogen from the atmosphere, the system is enclosed in a chamber pressurized with a flowing inert gas – typically argon. Builds are completed on a build plate to anchor the solid material, thereby keeping it accurately positioned during the build process.
Considering this method of production, several steps must be completed prior to fusing metal powder. These steps are highlighted in Figure 2.
One should consider several geometric factors when establishing the part shape and build orientation. First, overhanging features – like the bottom of a bridge – greater than a few millimeters in length require support structure. These features allow material to be anchored to a solid structure so that when the first few layers of fused metal are generated in the overhang they don’t sag or get displaced during construction of subsequent layers. Examples of inadequately supported overhangs are shown in Figure 3. To avoid these situations, parts are oriented in the build process to minimize overhangs.
When orientation selection is insufficient to eliminate overhangs, part geometry may be altered to ensure such overhanging surface is at least 45° from the horizontal. Shallower angles have resulted in overhanging features that consistently yield defects like those in Figure 3. When shallow overhang angles cannot be avoided, lattice support structures are typically added. These structures provide thin truss-like features onto which the overhanging material will be firmly supported to avoid delamination-like defects shown in Figure 3. However, these lattice support structures typically require removal as part of the post-processing of completed builds. Horizontally positioned holes, for example must either use support structure or the top of the hole must have flat sloping sides that do not exceed 45° from the horizontal. Therefore, beyond designing a part for form, fit and function, one must consider the manufacturing process in part design, not unlike designing parts to account for limitations of any other manufacturing process.
After defining the solid mass (i.e., the as-built shape) of the part, including lattice structures, one must determine the process conditions to be used to produce the part. Although some laser powder bed fusion manufacturers fix process conditions for a given alloy, other manufacturers offer open architectures that allow for user-defined process definition including build plate preheat temperature; laser power, focal point and travel speed; powder layer thickness; laser rastering pattern and angle increment; hatch spacing; and many other potential process details.
Optimum process conditions depend on the selected alloy as well as the powder size distribution and production process used to produce the powder. The powder layer thickness is used by slicing software to establish the region to be melted and fused for each build layer. This is commonly completed by software using the computer solid model of the as-built shape and orientation.
Also completed automatically is the laser rastering pattern, that is, the path the laser will take as it locally sinters small sections of the powder in each build layer. Multiple rastering patterns are available, as highlighted in Figure 4. At this point, the detailed build plan is completed.
Machine preparations include sieving and loading the metal powder. Sieved metal powders typically range in size from 15 to 55 microns. Several processes are used to produce metal powders. Argon atomized is the most commonly specified powder. Air-atomized powder is lower in cost, but may result in undesirable defects in finished parts. Plasma atomized powder has the highest quality resulting in minimal porosity in finished parts.
In essence, one must consider more than powder chemistry when selecting a metal powder for production of AM parts. Powder sizes outside this range are undesirable as small particles typically lead to clogged filters, easily become airborne and are extremely flammable due to surface area. Larger powder particles may not completely melt or may interfere with good powder spreading.
While the powders typically arrive from the powder producers in this size range, unfused powder that is reused from previous builds requires onsite sieving to eliminate satellites/irregularly shaped particles to ensure the sieved powder remains within the acceptable size range. Once ready, the powders are introduced into the powder handling system, which must also be kept under inert gas to avoid oxygen or nitrogen contamination from the atmosphere. After backfilling the system with an inert gas, the build plate is heated and made ready for the build.
The actual building process is characterized by four steps that repeat until the build is complete.
From a moving hopper, deposit a layer of metal powder along the entire surface of the build plate.
Level the powder layer to a uniform thickness by using either a solid recoater blade made of a metallic or elastomeric material, or a roller that traversed from one side of the build plate to the other and slightly compacts the powder in addition to leveling it.
Fuse the powder via one or more lasers (each typically between 200 and 700 watts) that follow the pattern established in preprocessing; the unaffected powder remains in the build chamber and is often recycled for use in subsequent builds.
Lower the build plate by an amount equal to the powder layer thickness.
To minimize directional property variations, the raster angle is modified from one layer to the next as illustrated in Figure 4. The part borders in each layer are typically scanned multiple times to improve geometric accuracy and to minimize surface roughness. As the molten metal solidifies and cools, it shrinks, resulting in residual stresses that if not properly accounted for may lead to severe part distortion or even cracking, as shown in Figure 5. In this case, the sharp corners were later replaced with ones having a small radius of curvature to eliminate the associated stress concentration at this location, and yield a crack-free build.
Post processing of the builds includes loose powder removal, separating components from the build plate, removal of support structure, thermal processing, final machining and surfacing (as needed). Operators typically wear personal protective equipment during this phase of the build to avoid breathing in fine metal particles. In addition, care must be exercised when removing loose powder (i.e., that powder that remains unfused during the build process) as the fine particles could spark a spontaneous, localized fire if a sufficient quantity of particles becomes airborne in the presence of an ignition source. Static-free mats are used in the vicinity of AM machines to minimize the probability of such an event. Typically, the loose powder is recycled after sieving. However, premium-quality components may always be made from virgin powder to avoid any contamination from recycled powder.
Once the loose powder has been removed, the build plate is removed from the machine and the builds taken off the build plates via wire electrical discharge machining (EDM) or band saw. The associated metal removed by these machining operations must be accounted for in the preprocess design phase mentioned above. In a few instances, parts can be pried off through use of a flathead screwdriver; however, this method is not recommended since it may damage the parts or build plate. After removal of the parts, the build plate is redressed, typically by machining a thin layer of material from its upper surface and ensuring the top surface is flat and parallel to the underside of the build plate. This is needed to ensure the next build has a predictable, uniformly thick initial powder layer.
The built parts require several additional processing steps. Additional powder removal may be required at this point. Powder from readily accessible pockets generally can be easily removed with a soft brush; powder remaining in small crevices and deep, small-diameter holes, however, may require use of ultrasonic vibration or pressurized air to remove loose powder. Typical as-built surface finish for metallic components is 100–250 microinch Ra. While this finish is often sufficient for exterior surfaces, machining is typically required to obtain the necessary finish for mating surfaces. Remaining build structure must also be removed by machining.
Thermal treatments typically include a stress relief while the components are still attached to the build plate. Additional post-build thermal treatments may include hot isostatic pressing (HIP) to consolidate minor porosity in the build and/or thermal treatments designed to tune mechanical properties. (Note that in some cases, all thermal treatments are completed prior to removal of parts from the build plate.) All thermal treatments are similar to those used for other metalworking processes such as casting or forging. Finally, any additional surfacing operations, such as tumbling (for surface finish and/or cleaning), painting, etc. may also be required. Inspection for dimensions, material flaws or other features is also typically completed at one or more stages of the process.
Common Metallic Materials
Although some research is underway to develop a series of alloys specifically designed for additive processes, the most common metallic materials are those currently used and qualified for casting and/or welding. Though the list of alloys used to produce metallic materials is ever expanding, the most commonly used alloys are shown in Table 2. Optimum process conditions vary widely among the various alloys. In some instances, minor modifications are made to the starting alloy composition to account for evaporation of light alloying elements.
Benefits and Applications of AM
Until recently the most popular applications of AM was for prototyping, low-rate production, and complex dies including those with conformal cooling passages. However, over the last several years, large volumes of production parts have been realized. By taking advantage of the AM process, manufacturers have been able to justify the higher production cost (relative to other established metalworking processes) of AM primarily as a result of lower overall system production costs, lower weight and/or improved performance of the parts in service.
AM offers the opportunity to consolidate parts that would otherwise be produced by other means. For example, bolted, riveted and/or welded structures may be produced as a single-piece part, thereby eliminating assembly costs. This has been true in several well-publicized GE developments  .
A fuel nozzle for the LEAP jet engine consolidated 20 parts into a single AM component, which also was 25% lighter and had a projected life five times that of the legacy design.
An advanced turboprop engine consolidated 855 components into 12 AM parts, which reduced fuel consumption by as much as 20% while achieving a 10% boost in power
Much of the in-service benefits from these GE parts have been realized due to lower weight. Other improvements in performance have been realized as a result of part geometries that would not be practical to produce via traditional metalworking processes. In another instance, conformal cooling channels have been used in electronic chassis to improve its cooling characteristics  . The complex three-dimensional path that includes the cooling channels could not be manufactured by any other means than through AM. Conformal cooling channels have also been produced during AM of dies/molds for die and permanent mold castings, metal forming, plastic injection molding and other processes. Conformal cooling allows for more rapid production of components and reduced stresses as a result of better thermal management throughout the die/mold.
Another benefit of AM is weight reduction in parts. Given the practical limitations associated with traditional metal-working processes, components often have excess material. In these cases, components often take on a blocky geometry. Through advanced design methodologies, including topology optimization and generative design, engineers have developed complex geometries that do not include excess metal mass. The resulting structures often include features looking more like a labyrinth of ligaments rather than a solid blocky structure. In other instances, structural lattice structures are used to lighten components.
Reentrant features are generally challenging and costly to achieve in practice. However, with the design flexibility offered by AM, such features may be included without the need for additional follow-up process steps. Figure 6 shows an example of a reentrant lip on the top of a L-PBF build. The part is shown as it was removed from the build plate. Note that the threads in the middle knob are as built. Though not close tolerance, they functioned well enough to allow for a tight connection to a mating part of the assembly.
This stainless steel component took approximately 35 hours to build and less than one hour to remove from the build plate. Lead time for the prototype component by other manufacturing methods, including machining and casting, exceeded the required delivery time for the component. Furthermore, the additive design permitted a significantly smaller axial length, which allowed for use in pipe repairs having 90° elbows, whereas the legacy design could not be used in such applications.
Current Hurdles to More Widespread Use of AM
Several hurdles exist to more widespread use of AM in metallic system components. Perhaps the most significant hurdle is cost. Parts produced by AM nearly always cost more to produce than shaped, mass produced parts. Given the processing steps required, AM feedstock is more expensive per unit weight (by a factor of 2 to 4) than wrought material of the same composition.
However, when powder recycling is used, L-PBF nearly always has a higher part-weight-to-feedstock usage rate. Furthermore, this argument assumes the geometry of the AM part is identical to that produced by alternative processes. Direct substitution of identically shaped parts may be suitable for replacement of unavailable parts; however, as a general rule, production of a geometrically identical AM substitute for parts produced by alternative processes is not economically viable. Parts designed for the AM process should take advantage of the capabilities offered by the AM process. Therefore, the best parts to consider for AM are those whose shape cannot be economically produced by well-established alternative processes such as machining, casting, or forging.
Production rates have been cited as an issue with some AM parts. Equipment manufacturers have recently introduced machines with multiple lasers to simultaneously sinter multiple areas within each build layer. Heat management of these builds does limit the number of lasers that can practically be applied in any given AM machine.
Build chamber size has excluded the use of powder bed machines to produce some large components as a single piece. The majority of commercially available production L-PBF machines have a build volume limit of approximately 12 in × 12 in × 12 in. Machines that can produce builds of approximately 24 in × 12 in × 12 in are available from equipment suppliers. Metal powder bed machines with build volumes of 1 m × 1 m × 1 m (39 in × 39 in × 39 in) have been announced, but such machines would only be used for niche applications by today’s demand for AM builds. Furthermore, machines of this size require additional automated powder metal handling systems and a significant investment in machine costs.
The most common defect in AM production is porosity. When operating at a low heat input, lack of fusion can occur among the powder particles; on the other hand, operating at a high heat input may lead to vaporization of low-melting-point alloy constituents, which may be trapped in solidified metal. Even when operating in the middle ground, some gas porosity is typically present in the as-built state. Typical levels of porosity are less than 1% by volume. This porosity generally does not hamper mechanical strength of metallic builds, but will detrimentally impact elongation and fatigue properties. Porosity is often reduced or eliminated through HIP.
Residual stresses and distortion may occur in builds as a result of the thermal response of the build associated with solidification and the transient temperature gradients throughout the build. These effects may manifest themselves in the part detaching from the build plate or lifting the top surface of the build above the top of the next powder layer. Cracks or slightly misshapen parts may also result from the transient temperature gradients. Guidance on how to avoid these conditions is an area of active research.
Part detachment and solid protruding beyond the next build layer can result in recoater blade damage, which will manifest itself in ridges throughout the unfused powder and potentially pieces of the blade breaking off into the powder bed. Cracks are rarely repairable. Misshapen parts are often corrected during final machining. These problems are overcome by reprinting the parts with improved part orientation, different support structures, part shape changes or different AM process conditions.
Properties of AM builds are not isotropic. Strength and elongation in the build direction are consistently lower than those in the build plane. While the differences in these directions are normally not more than 10–15%, engineers must be aware of – and account for – these effects during part design and development.
While surface finish of the builds is often acceptable, fatigue-critical regions of parts may need additional attention. A finer surface finish may be obtained by altering the process design of border contour passes during building and/or smoothing the surface of these fatigue-critical areas during post processing.
Industrial standards for AM have been developed by ASTM International, SAE International, ISO and others. However, many standards have yet to be developed to the point of application by a broad range of industries. These standards are useful, but they remain somewhat in flux as the industry continues in its current rapid development, expansion and evolution. As these standards are recognized and implemented by more industries, the application of AM will to continue to expand.
Inspection of parts produced by AM often pose a challenge. Many low-cost, high-volume inspection processes have limited application for the broad range of geometrically complex shapes produced by AM. Exterior part geometry is easy to confirm with laser scanning techniques. However, the most demanding applications require assurance of internal quality of AM builds.
Computed tomography offers a method of determining internal flaws in complex 3D components. However, the process is relatively expensive and slow. Furthermore, the information provided by the process is often difficult to relate to existing quality standards.
To combat this issue, a number of organizations are pursuing part certification and process/producer qualification through in-situ process monitoring and control. When coupled with software to predict the resulting microstructure and properties, these systems offer an alternative to post-process inspection. They also offer the potential of either stopping a defective build before its completion or repairing certain defects as they are detected. The need for in-situ process monitoring and control is also justified since each L-PBF system responds somewhat differently.
Performance differences also occur due to degradation of equipment, potential soot buildup on equipment, lot-to-lot variability in powder characteristics and other factors. Therefore, without in-situ process monitoring and control each machine has to be periodically tuned to achieve optimum part performance. In-situ monitoring and control will also correct for any differences brought on by dissimilarities in powder lots.
The future of AM is very bright according to published reports  . Double digit annual growth rates are predicted for the next five or more years. Current research and development will reduce or eliminate many of the hurdles identified above.
This article was written by Michael L. Tims, Advisor Engineer, and Kenneth Sabo, Senior Director of Manufacturing, Concurrent Technologies Corporation (CTC) (Johnstown, PA). For more information, visit here .