1523430527968ggi copy-128-165.pdf

5 Powder Bed Fusion Processes 5.1 Introduction Powder bed fusion (PBF) processes were among the first commercialized AM processes. Developed at the University of Texas at Austin, USA, selective laser sintering (SLS) was the first commercialized PBF process. Its basic method of operation is schem...

5 Powder Bed Fusion Processes 5.1 Introduction Powder bed fusion (PBF) processes were among the first commercialized AM processes. Developed at the University of Texas at Austin, USA, selective laser sintering (SLS) was the first commercialized PBF process. Its basic method of operation is schematically shown in Fig. 5.1, and all other PBF processes modify this basic approach in one or more ways to enhance machine productivity, enable different materials to be processed, and/or to avoid specific patented features. All PBF processes share a basic set of characteristics. These include one or more thermal sources for inducing fusion between powder particles, a method for controlling powder fusion to a prescribed region of each layer, and mechanisms for adding and smoothing powder layers. The most common thermal sources for PBF are lasers. PBF processes which utilize lasers are known as laser sintering (LS) machines. Since polymer laser sintering (pLS) machines and metal laser sintering (mLS) machines are significantly different from each other, we will address each separately. In addition, as electron beam and other thermal sources require significantly different machine architectures than laser sintering machines, non-laser thermal sources will be addressed separately from laser sources at the end of the chapter. LS processes were originally developed to produce plastic prototypes using a point-wise laser scanning technique. This approach was subsequently extended to metal and ceramic powders; additional thermal sources are now utilized; and variants for layer-wise fusion of powdered materials are being commercially introduced. As a result, PBF processes are widely used worldwide, have a broad range of materials (including polymers, metals, ceramics, and composites) which can be utilized, and are increasingly being used for direct manufacturing of end-use products, as the material properties are comparable to many engineering-grade polymers, metals, and ceramics. In order to provide a baseline description of powder fusion processes, pLS will be described as the paradigm approach to which the other PBF processes will be # Springer Science+Business Media New York 2015 I. Gibson et al., Additive Manufacturing Technologies, DOI 10.1007/978-1-4939-2113-3_5 107 108 5 Powder Bed Fusion Processes Fig. 5.1 Schematic of the selective laser sintering process compared. As shown in Fig. 5.1, pLS fuses thin layers of powder (typically 0.075– 0.1 mm thick) which have been spread across the build area using a counter-rotating powder leveling roller. The part building process takes place inside an enclosed chamber filled with nitrogen gas to minimize oxidation and degradation of the powdered material. The powder in the build platform is maintained at an elevated temperature just below the melting point and/or glass transition temperature of the powdered material. Infrared heaters are placed above the build platform to maintain an elevated temperature around the part being formed, as well as above the feed cartridges to preheat the powder prior to spreading over the build area. In some cases, the build platform is also heated using resistive heaters around the build platform. This preheating of powder and maintenance of an elevated, uniform temperature within the build platform is necessary to minimize the laser power requirements of the process (with preheating, less laser energy is required for fusion) and to prevent warping of the part during the build due to nonuniform thermal expansion and contraction (resulting in curling). Once an appropriate powder layer has been formed and preheated, a focused CO2 laser beam is directed onto the powder bed and is moved using galvanometers in such a way that it thermally fuses the material to form the slice cross section. Surrounding powder remains loose and serves as support for subsequent layers, thus eliminating the need for the secondary supports which are necessary for vat photopolymerization processes. After completing a layer, the build platform is lowered by one layer thickness and a new layer of powder is laid and leveled using the counter-rotating roller. The beam scans the subsequent slice cross section. This process repeats until the complete part is built. A cool-down period is typically 5.2 Materials 109 required to allow the parts to uniformly come to a low-enough temperature that they can be handled and exposed to ambient temperature and atmosphere. If the parts and/or powder bed are prematurely exposed to ambient temperature and atmosphere, the powders may degrade in the presence of oxygen and parts may warp due to uneven thermal contraction. Finally, the parts are removed from the powder bed, loose powder is cleaned off the parts, and further finishing operations, if necessary, are performed. 5.2 Materials In principle, all materials that can be melted and resolidified can be used in PBF processes. A brief survey of materials processed using PBF processes will be given here. More details can be found in subsequent sections. 5.2.1 Polymers and Composites Thermoplastic materials are well-suited for powder bed processing because of their relatively low melting temperatures, low thermal conductivities, and low tendency for balling. Polymers in general can be classified as either a thermoplastic or a thermoset polymer. Thermoset polymers are typically not processed using PBF into parts, since PBF typically operates by melting particles to fabricate part cross sections, but thermosets degrade, but do not melt, as their temperature is increased. Thermoplastics can be classified further in terms of their crystallinity. Amorphous polymers have a random molecular structure, with polymer chains randomly intertwined. In contrast, crystalline polymers have a regular molecular structure, but this is uncommon. Much more common are semi-crystalline polymers which have regions of regular structure, called crystallites. Amorphous polymers melt over a fairly wide range of temperatures. As the crystallinity of a polymer increases, however, its melting characteristics tend to become more centered around a well defined melting point. At present, the most common material used in PBF is polyamide, a thermoplastic polymer, commonly known in the US as nylon. Most polyamides have fairly high crystallinity and are classified as semi-crystalline materials. They have distinct melting points that enable them to be processed reliably. A given amount of laser energy will melt a certain amount of powder; the melted powder fuses and cools quickly, forming part of a cross section. In contrast, amorphous polymers tend to soften and melt over a broad temperature range and not form well defined solidified features. In pLS, amorphous polymers tend to sinter into highly porous shapes, whereas crystalline polymers are typically processed using full melting, which result in higher densities. Polyamide 11 and polyamide 12 are commercially available, where the number designates the number of carbon atoms that are provided by one of the monomers that is reacted to produce polyamide. However, crystalline polymers exhibit greater shrinkage compared to amorphous materials 110 5 Powder Bed Fusion Processes and are more susceptible to curling and distortion and thus require more uniform temperature control. Mechanical properties of pLS parts produced using polyamide powders approach those of injection molded thermoplastic parts, but with significantly reduced elongation and unique microstructures. Polystyrene-based materials with low residual ash content are particularly suitable for making sacrificial patterns for investment casting using pLS. Interestingly, polystyrene is an amorphous polymer, but is a successful example material due to its intended application. Porosity in an investment casting pattern aids in melting out the pattern after the ceramic shell is created. Polystyrene parts intended for precision investment casting applications should be sealed to prevent ceramic material seeping in and to achieve a smooth surface finish. Elastomeric thermoplastic polymers are available for producing highly flexible parts with rubber-like characteristics. These elastomers have good resistance to degradation at elevated temperatures and are resistant to chemicals like gasoline and automotive coolants. Elastomeric materials can be used to produce gaskets, industrial seals, shoe soles, and other components. Additional polymers that are commercially available include flame-retardant polyamide and polyaryletherketone (known as PAEK or PEEK). Both 3D Systems and EOS GmbH offer most of the materials listed in this section. Researchers have investigated quite a few polymers for biomedical applications. Several types of biocompatible and biodegradable polymers have been processed using pLS, including polycaprolactone (PCL), polylactide (PLA), and poly-Llactide (PLLA). Composite materials consisting of PCL and ceramic particles, including hydroxyapatite and calcium silicate, have also been investigated for the fabrication of bone replacement tissue scaffolds. In addition to neat polymers, polymers in PBF can have fillers that enhance their mechanical properties. For example, the Duraform material from 3D Systems is offered as Duraform PA, which is polyamide 12, as well as Duraform GF, which is polyamide 12 filled with small glass beads. The glass additive enhances the material’s stiffness significantly, but also causes its ductility to be reduced, compared to polyamide materials without fillers. Additionally, EOS GmbH offers aluminum particle, carbon fiber, and their own glass bead filled polyamide materials. 5.2.2 Metals and Composites A wide range of metals has been processed using PBF. Generally, any metal that can be welded is considered to be a good candidate for PBF processing. Several types of steels, typically stainless and tool steels, titanium and its alloys, nickel-base alloys, some aluminum alloys, and cobalt-chrome have been processed and are commercially available in some form. Additionally, some companies now offer PBF of precious metals, such as silver and gold. Historically, a number of proprietary metal powders (either thermoplastic binder-coated or binder mixed) were developed before modern mLS machines 5.2 Materials 111 were available. RapidSteel was one of the first metal/binder systems, developed by DTM Corp. The first version of RapidSteel was available in 1996 and consisted of a thermoplastic binder coated 1080 carbon steel powder with copper as the infiltrant. Parts produced using RapidSteel were debinded (350–450
C), sintered (around 1,000
C), and finally infiltrated with Cu (1,120
C) to produce a final part with approximately 60 % low carbon steel and 40 % Cu. This is an example of liquid phase sintering which will be described in the next section. Subsequently, RapidSteel 2.0 powder was introduced in 1998 for producing functional tooling, parts, and mold inserts for injection molding. It was a dry blend of 316 stainless steel powder impact milled with thermoplastic and thermoset organic binders with an average particle size of 33 μm. After green part fabrication, the part was debinded and sintered in a hydrogen-rich atmosphere. The bronze infiltrant was introduced in a separate furnace run to produce a 50 % steel and 50 % bronze composite. RapidSteel 2.0 was structurally more stable than the original RapidSteel material because the bronze infiltration temperature was less than the sintering temperature of the stainless steel powder. A subsequent material development was LaserForm ST-100, which had a broader particle size range, with fine particles not being screened out. These fine particles allowed ST-100 particles to be furnace sintered at a lower temperature than RapidSteel 2.0, making it possible to carry out sintering and infiltration in a single furnace run. In addition to the above, H13 and A6 tool steel powders with a polymer binder can also be used for tooling applications. The furnace processing operations (sintering and infiltration) must be carefully designed with appropriate choices of temperature, heating and cooling rates, furnace atmosphere pressure, amount of infiltrant, and other factors, to prevent excessive part distortion. After infiltration, the part is finish machined as needed. These issues are further explored in the post-processing chapter. Several proprietary metal powders were marketed by EOS for their M250 Xtended metal platforms, prior to the introduction of modern mLS machines. These included liquid-phase sintered bronze-based powders, and steel-based powders and other proprietary alloys (all without polymer binders). These were suitable for producing tools and inserts for injection molding of plastics. Parts made from these powders were often infiltrated with epoxy to improve the surface finish and seal porosity in the parts. Proprietary nickel-based powders for direct tooling applications and Cu-based powders for parts requiring high thermal and electrical conductivities were also available. All of these materials have been successfully used by many organizations; however, the more recent introduction of mLS and electron beam melting (EBM) technology has made these alloys obsolete, as engineering-grade alloys are now able to be processed using a number of manufacturers’ machines. As mentioned, titanium alloys, numerous steel alloys, nickel-based super alloys, CoCrMo, and more are widely available from numerous manufacturers. It should be noted that alloys that crack under high solidification rates are not good candidates for mLS. Due to the high solidification rates in mLS, the crystal structures produced and mechanical properties are different than those for other manufacturing processes. These structures may be metastable, and the heat treatment recipes needed 112 5 Powder Bed Fusion Processes to produce standard microstructures may be different. As mLS and EBM processes advance, the types of metal alloys which are commonly utilized will grow and new alloys specifically tailored for PBF production will be developed. 5.2.3 Ceramics and Ceramic Composites Ceramic materials are generally described as compounds that consist of metaloxides, carbides, and nitrides and their combinations. Several ceramic materials are available commercially including aluminum oxide and titanium oxide. Commercial machines were developed by a company called Phenix Systems in France, which was acquired by 3D Systems in 2013. 3D Systems also says it offers cermets, which are metal-ceramic composites. Ceramics and metal-ceramic composites have been demonstrated in research. Typically, ceramic precipitates form through reactions occurring during the sintering process. One example is the processing of aluminum in a nitrogen atmosphere, which forms an aluminum matrix with small regions of AlN interspersed throughout. This process is called chemically induced sintering and is described further in the next section. Biocompatible materials have been developed for specific applications. For example, calcium hydroxyapatite, a material very similar to human bone, has been processed using pLS for medical applications. 5.3 Powder Fusion Mechanisms Since the introduction of LS, each new PBF technology developer has introduced competing terminology to describe the mechanism by which fusion occurs, with variants of “sintering” and “melting” being the most popular. However, the use of a single word to describe the powder fusion mechanism is inherently problematic as multiple mechanisms are possible. There are four different fusion mechanisms which are present in PBF processes [1]. These include solid-state sintering, chemically induced binding, liquid-phase sintering (LPS), and full melting. Most commercial processes utilize primarily LPS and melting. A brief description of each of these mechanisms and their relevance to AM follows. 5.3.1 Solid-State Sintering The use of the word sintering to describe powder fusion as a result of thermal processing predates the advent of AM. Sintering, in its classical sense, indicates the fusion of powder particles without melting (i.e., in their “solid state”) at elevated temperatures. This occurs at temperatures between one half of the absolute melting temperature and the melting temperature. The driving force for solid-state sintering 5.3 Powder Fusion Mechanisms a Unsintered particle 113 b Neck c Increased Necking Pore Decreased porosity Fig. 5.2 Solid-state sintering. (a) Closely packed particles prior to sintering. (b) Particles agglomerate at temperatures above one half of the absolute melting temperature, as they seek to minimize free energy by decreasing surface area. (c) As sintering progresses, neck size increases and pore size decreases is the minimization of total free energy, Es, of the powder particles. The mechanism for sintering is primarily diffusion between powder particles. Surface energy Es is proportional to total particle surface area SA, through the equation Es ¼ γ s  SA (where γ s is the surface energy per unit area for a particular material, atmosphere, and temperature). When particles fuse at elevated temperatures (see Fig. 5.2), the total surface area decreases, and thus surface energy decreases. As the total surface area of the powder bed decreases, the rate of sintering slows. To achieve very low porosity levels, long sintering times or high sintering temperatures are required. The use of external pressure, as is done with hot isostatic pressing, increases the rate of sintering. As total surface area in a powder bed is a function of particle size, the driving force for sintering is directly related to the surface area to volume ratio for a set of particles. The larger the surface area to volume ratio, the greater the free energy driving force. Thus, smaller particles experience a greater driving force for necking and consolidation, and hence, smaller particles sinter more rapidly and initiate sintering at lower temperature than larger particles. As diffusion rates exponentially increase with temperature, sintering becomes increasingly rapid as temperatures approach the melting temperature, which can be modeled using a form of the Arrhenius equation. However, even at temperatures approaching the melting temperature, diffusion-induced solid-state sintering is the slowest mechanism for selectively fusing regions of powder within a PBF process. For AM, the shorter the time it takes to form a layer, the more economically competitive the process becomes. Thus, the heat source which induces fusion should move rapidly and/or induce fusion quickly to increase build rates. Since the time it takes for fusion by sintering is typically much longer than for fusion by melting, few AM processes use sintering as a primary fusion mechanism. 114 5 Powder Bed Fusion Processes Sintering, however, is still important in most thermal powder processes, even if sintering is not the primary fusion mechanism. There are three secondary ways in which sintering affects a build. 1. If the loose powder within the build platform is held at an elevated temperature, the powder bed particles will begin to sinter to one another. This is typically considered a negative effect, as agglomeration of powder particles means that each time the powder is recycled the average particle size increases. This changes the spreading and melting characteristics of the powder each time it is recycled. One positive effect of loose powder sintering, however, is that the powder bed will gain a degree of tensile and compressive strength, thus helping to minimize part curling. 2. As a part is being formed in the build platform, thermally induced fusing of the desired cross-sectional geometry causes that region of the powder bed to become much hotter than the surrounding loose powder. If melting is the dominant fusion mechanism (as is typically the case) then the just-formed part cross section will be quite hot. As a result, the loose powder bed immediately surrounding the fused region heats up considerably, due to conduction from the part being formed. This region of powder may remain at an elevated temperature for a long time (many hours) depending upon the size of the part being built, the heater and temperature settings in the process, and the thermal conductivity of the powder bed. Thus, there is sufficient time and energy for the powder immediately next to the part being built to fuse significantly due to solidstate sintering, both to itself and to the part. This results in “part growth,” where the originally scanned part grows a “skin” of increasing thickness the longer the powder bed is maintained at an elevated temperature. This phenomenon can be seen in Fig. 5.3 as unmolten particles fused to the edge of a part. For many materials, the skin formed on the part goes from high density, low porosity near the originally scanned region to lower density, higher porosity further from the part. This part growth can be compensated in the build planning stage by offsetting the laser beam to compensate for part growth or by offsetting the surface of the STL model. In addition, different post-processing methods will remove this skin to a different degree. Thus, the dimensional repeatability of the final part is highly dependent upon effectively compensating for and controlling this part growth. Performing repeatable post-processing to remove the same amount of the skin for every part is thus quite important. 3. Rapid fusion of a powder bed using a laser or other heat source makes it difficult to achieve 100 % dense, porosity-free parts. Thus, a feature of many parts built using PBF techniques (especially for polymers) is distributed porosity throughout the part. This is typically detrimental to the intended part properties. However, if the part is held at an elevated temperature after scanning, solid-state sintering combined with other high-temperature phenomena (such as grain growth in metals) causes the % porosity in the part to decrease. Since lower layers are maintained at an elevated temperature while additional layers are added, this can result in lower regions of a part being denser than upper regions 5.3 Powder Fusion Mechanisms 115 Unmolten particle fused to edge Spherulite from fully melted & crystallised particle Unmolten particle core 100 µm Spherulite from melted & crystallised region Fig. 5.3 Typical pLS microstructure for nylon polyamide (Materials Science & Engineering. A. Structural Materials: Properties, Microstructure and Processing by Zarringhalam, H., Hopkinson, N., Kamperman, N.F., de Vlieger, J.J. Copyright 2006 by Elsevier Science & Technology Journals. Reproduced with permission of Elsevier Science & Technology Journals in the format Textbook via Copyright Clearance Center) [5] of a part. This uneven porosity can be controlled, to some extent, by carefully controlling the part bed temperature, cooling rate and other parameters. EBM, in particular, often makes use of the positive aspects of elevated-temperature solidstate sintering and grain growth by purposefully maintaining the metal parts that are being built at a high enough temperature that diffusion and grain growth cause the parts being built to reach 100 % density. 5.3.2 Chemically Induced Sintering Chemically induced sintering involves the use of thermally activated chemical reactions between two types of powders or between powders and atmospheric gases to form a by-product which binds the powders together. This fusion mechanism is primarily utilized for ceramic materials. Examples of reactions between powders and atmospheric gases include: laser processing of SiC in the presence of oxygen, whereby SiO2 forms and binds together a composite of SiC and SiO2; laser processing of ZrB2 in the presence of oxygen, whereby ZrO2 forms and binds together a composite of ZrB2 and ZrO2; and laser processing of Al in the presence of N2, whereby AlN forms and binds together the Al and AlN particles. For chemically induced sintering between powders, various research groups have demonstrated that mixtures of high-temperature structural ceramic and/or intermetallic precursor materials can be made to react using a laser. In this case, raw materials which exothermically react to form the desired by-product are 116 5 Powder Bed Fusion Processes pre-mixed and heated using a laser. By adding chemical reaction energy to the laser energy, high-melting-temperature structures can be created at relatively low laser energies. One common characteristic of chemically induced sintering is part porosity. As a result, post-process infiltration or high-temperature furnace sintering to higher densities is often needed to achieve properties that are useful for most applications. This post-process infiltration may involve other reactive elements, forming new chemical compounds after infiltration. The cost and time associated with postprocessing have limited the adoption of chemically induced sintering in commercial machines. 5.3.3 LPS and Partial Melting LPS is arguably the most versatile mechanism for PBF. LPS is a term used extensively in the powder processing industry to refer to the fusion of powder particles when a portion of constituents within a collection of powder particles become molten, while other portions remain solid. In LPS, the molten constituents act as the glue which binds the solid particles together. As a result, hightemperature particles can be bound together without needing to melt or sinter those particles directly. LPS is used in traditional powder metallurgy to form, for instance, cemented carbide cutting tools where Co is used as the lower-meltingpoint constituent to glue together particles of WC. There are many ways in which LPS can be utilized as a fusion mechanism in AM processes. For purposes of clarity, the classification proposed by Kruth et al. [1] has formed the basis for the distinctions discussed in the following section and shown in Fig. 5.4. 5.3.3.1 Distinct Binder and Structural Materials In many LPS situations, there is a clear distinction between the binding material and the structural material. The binding and structural material can be combined in three different ways: as separate particles, as composite particles, or as coated particles. Separate Particles A simple, well-mixed combination of binder and structural powder particles is sufficient in many cases for LPS. In cases where the structural material has the dominant properties desired in the final structure, it is advantageous for the binder material to be smaller in particle size than the structural material. This enables more efficient packing in the powder bed and less shrinkage and lower porosity after binding. The dispersion of smaller-particle-size binder particles around structural particles also helps the binder flow into the gaps between the structural particles more effectively, thus resulting in better binding of the structural particles. This is often true when, for instance, LS is used to process steel powder with a polymer binder (as discussed more fully in Sect. 5.3.5). This is also true when metal-metal 5.3 Powder Fusion Mechanisms 117 a c b Cross-section of coated particles Binder Structural material d Fig. 5.4 Liquid phase sintering variations used in powder bed fusion processing: (a) separate particles, (b) composite particles, (c) coated particles, and (d) indistinct mixtures. Darker regions represent the lower-melting-temperature binder material. Lighter regions represent the highmelting-temperature structural material. For indistinct mixtures, microstructural alloying eliminates distinct binder and structural regions mixtures and metal-ceramic mixtures are directly processed without the use of a polymer binder. In the case of LPS of separate particles, the heat source passes by quickly, and there is typically insufficient time for the molten binder to flow and surface tension to draw the particles together prior to resolidification of the binder unless the binder has a particularly low viscosity. Thus, composite structures formed from separate particles typically are quite porous. This is often the intent for parts made from separate particles, which are then post-processed in a furnace to achieve the final 118 5 Powder Bed Fusion Processes part properties. Parts held together by polymer binders which require further postprocessing (e.g., to lower or fill the porosity) are termed as “green” parts. In some cases, the density of the binder and structural material are quite different. As a result, the binder and structural material may separate during handling. In addition, some powdered materials are most economically manufactured at particle sizes that are too small for effective powder dispensing and leveling (see Sect. 5.5). In either case, it may be beneficial for the structural and/or binder particles to be bound together into larger particle agglomerates. By doing so, composite powder particles made up of both binder and structural material are formed. Composite Particles Composite particles contain both the binder and structural material within each powder particle. Mechanical alloying of binder and structural particles or grinding of cast, extruded or molded mixtures into a powder results in powder particles that are made up of binder and structural materials agglomerated together. The benefits of composite particles are that they typically form higher density green parts and typically have better surface finish after processing than separate particles [1]. Composite particles can consist of mixtures of polymer binders with higher melting point polymer, metal, or ceramic structural materials; or metal binders with higher melting point metal or ceramic structural materials. In all cases, the binder and structural portions of each particle, if viewed under a microscope, are distinct from each other and clearly discernable. The most common commercially available composite particle used in PBF processes is glass-filled nylon. In this case, the structural material (glass beads) is used to enhance the properties of the binding material (nylon) rather than the typical use of LPS where the binder is simply a necessary glue to help hold the structural material together in a useful geometric form. Coated Particles In some cases, a composite formed by coating structural particles with a binder material is more effective than random agglomerations of binder and structural materials. These coated particles can have several advantages; including better absorption of laser energy, more effective binding of the structural particles, and better flow properties. When composite particles or separate particles are processed, the random distribution of the constituents means that impinging heat energy, such as laser radiation, will be absorbed by whichever constituent has the highest absorptivity and/or most direct “line-of-sight” to the impinging energy. If the structural materials have a higher absorptivity, a greater amount of energy will be absorbed in the structural particles. If the rate of heating of the structural particles significantly exceeds the rate of conduction to the binder particles, the higher-melting-temperature structural materials may melt prior to the lower-melting-temperature binder materials. As a result, the anticipated microstructure of the processed material will differ significantly from one where the binder had melted and the structural material had 5.3 Powder Fusion Mechanisms 119 remained solid. This may, in some instances, be desirable, but is typically not the intent when formulating a binder/structural material combination. Coated particles can help overcome the structural material heating problem associated with random constituent mixtures and agglomerates. If a structural particle is coated with the binder material then the impinging energy must first pass through the coating before affecting the structural material. As melting of the binder and not the structural material is the objective of LPS, this helps ensure that the proper constituent melts. Other benefits of coated particles exist. Since there is a direct correlation between the speed of the impinging energy in AM processing and the build rate, it is desirable for the binder to be molten for only a very short period of time. If the binder is present at the surfaces of the structural material, this is the most effective location for gluing adjacent particles together. If the binder is randomly mixed with the structural materials, and/or the binder’s viscosity is too high to flow to the contact points during the short time it is molten, then the binder will not be as effective. As a result, the binder % content required for effective fusion of coated particles is usually less than the binder content required for effective fusion of randomly mixed particles. Many structural metal powders are spherical. Spherical powders are easier to deposit and smooth using powder spreading techniques. Coated particles retain the spherical nature of the underlying particle shape, and thus can be easier to handle and spread. 5.3.3.2 Indistinct Binder and Structural Materials In polymers, due to their low thermal conductivity, it is possible to melt smaller powder particles and the outer regions of larger powder particles without melting the entire structure (see Fig. 5.3). Whether to more properly label this phenomenon LPS or just “partial melting” is a matter of debate. Also with polymers, fusion can occur between polymer particles above their glass transition temperature, but below their melting temperature. Similarly, amorphous polymers have no distinct melting point, becoming less viscous the higher the temperature goes above the glass transition temperature. As a result, in each of these cases, there can be fusion between polymer powder particles in cases where there is partial but not full melting, which falls within the historical scope of the term “liquid phase sintering.” In metals, LPS can occur between particles where no distinct binder or structural materials are present. This is possible during partial melting of a single particle type, or when an alloyed structure has lower-melting-temperature constituents. For noneutectic alloy compositions, melting occurs between the liquidus and solidus temperature of the alloy, where only a portion of the alloy will melt when the temperature is maintained in this range. Regions of the alloy with higher concentrations of the lower-melting-temperature constituent(s) will melt first. As a result, it is commonly observed that many metal alloys can be processed in such a way that only a portion of the alloy melts when an appropriate energy level is applied. This type of LPS of metal alloys was the method used in the early EOS M250 direct metal laser sintering (DMLS) machines. Subsequent mLS 120 5 Powder Bed Fusion Processes commercialized processes are all designed to fully melt the metal alloys they process. 5.3.4 Full Melting Full melting is the mechanism most commonly associated with PBF processing of engineering metal alloys and semi-crystalline polymers. In these materials, the entire region of material subjected to impinging heat energy is melted to a depth exceeding the layer thickness. Thermal energy of subsequent scans of a laser or electron beam (next to or above the just-scanned area) is typically sufficient to re-melt a portion of the previously solidified solid structure; and thus, this type of full melting is very effective at creating well-bonded, high-density structures from engineering metals and polymers. The most common material used in PBF processing is nylon polyamide. As a semi-crystalline material, it has a distinct melting point. In order to produce parts with the highest possible strength, these materials should be fully melted during processing. However, elevated temperatures associated with full melting result in part growth and thus, for practical purposes, many accuracy versus strength optimization studies result in parameters which are at the threshold between full melting and LPS, as can be seen from Fig. 5.3. For metal PBF processes, the engineering alloys that are utilized in these machines (Ti, Stainless Steel, CoCr, etc.) are typically fully melted. The rapid melting and solidification of these metal alloys results in unique properties that are distinct from, and can sometime be more desirable than, cast or wrought parts made from identical alloys. Figure 5.5 summarizes the various binding mechanisms which are utilized in PBF processes. Regardless of whether a technology is known as “Selective Laser Sintering,” “Selective Laser Melting,” “Direct Metal Laser Sintering,” “Laser Cusing,” “Electron Beam Melting,” or some other name, it is possible for any of Primary Binding Mechanisms in Powder Bed Fusion Processes 1. Solid State Sintering 2. Chemically Induced Binding 3. Liquid Phase Sintering (Partial Melting) 3.1 Distinct binder and structural materials 4. Full Melting 3.2 Indistinct binder and structural materials 3.1.1 Separate particles 3.1.2 Composite particles 3.1.3 Coated particles Fig. 5.5 Primary binding mechanisms in powder bed fusion processes (adapted from [1]) 5.3 Powder Fusion Mechanisms 121 these mechanisms to be utilized (and, in fact, often more than one is present) depending upon the powder particle combinations, and energy input utilized to form a part. 5.3.5 Part Fabrication 5.3.5.1 Metal Parts There are four common approaches for using PBF processes in the creation of complex metal components: full melting, LPS, indirect processing, and pattern methods. In the full melting and LPS (with metal powders) approaches, a metal part is typically usable in the state in which it comes out of the machine, after separation from a build plate. In indirect processing, a polymer coated metallic powder or a mixture of metallic and polymer powders are used for part construction. Figure 5.6 shows the steps involved in indirect processing of metal powders. During indirect processing, the polymer binder is melted and binds the particles together, and the metal powder remains solid. The metallic powder particles remain largely unaffected by the heat of the laser. The parts produced are generally porous (sometimes exceeding 50 vol. % porosity). The polymer-bound green parts are subsequently furnace processed. Furnace processing occurs in two stages: (1) debinding and (2) infiltration or consolidation. During debinding, the polymer binder is vaporized to remove it from the green part. Typically, the temperature is also raised to the extent that a small degree of necking (sintering) occurs between the metal particles. Subsequently, the remaining porosity is either filled by infiltration of a lower melting point metal to produce a fully dense metallic part, or by further sintering and densification to reduce the part porosity. Infiltration is easier to control, dimensionally, as the overall shrinkage is much less than during consolidation. However, infiltrated structures are always composite in nature whereas consolidated structures can be made up of a single material type. Loose Powder Green Part Powder bed fusion processing Metal or ceramic particles mixed with polymer binders. Brown Part Furnace Processing Melting and resolid fication of polymer binders enable complex parts to be formed without thermally affecting the metal or ceramic powders. Finished Part Furnace Processing Polymer vaporization and particle sintering at elevated temperatures results in a porous, sintered component. Fig. 5.6 Indirect processing of metal and ceramic powders using PBF Infiltration with a lowermelting-temperature metal results in a dense, finished component. 122 5 Powder Bed Fusion Processes The last approach to metal part creation using PBF is the pattern approach. For the previous three approaches, metal powder is utilized in the PBF process; but in this final approach, the part created in the PBF process is a pattern used to create the metal part. The two most common ways PBF-created parts are utilized as patterns for metal part creation are as investment casting patterns or as sand-casting molds. In the case of investment casting, polystyrene or wax-based powders are used in the machine, and subsequently invested in ceramic during post-processing, and melted out during casting. In the case of sand-casting molds, mixtures of sand and a thermosetting binder are directly processed in the machine to form a sand-casting core, cavity or insert. These molds are then assembled and molten metal is cast into the mold, creating a metal part. Both indirect and pattern-based processes are further discussed in Chap. 18. 5.3.5.2 Ceramic Parts Similar to metal parts, there are a number of ways that PBF processes are utilized to create ceramic parts. These include direct sintering, chemically induced sintering, indirect processing, and pattern methods. In direct sintering, a high-temperature is maintained in the powder bed and a laser is utilized to accelerate sintering of the powder bed in the prescribed location of each layer. The resultant ceramic parts will be quite porous and thus are often post-processed in a furnace to achieve higher density. This high porosity is also seen in chemically induced sintering of ceramics, as described earlier. Indirect processing of ceramic powders is identical to indirect processing of metal powders (Fig. 5.6). After debinding, the ceramic brown part is consolidated to reduce porosity or is infiltrated. In the case of infiltration, when metal powders are used as the infiltrant then a ceramic/metal composite structure can be formed. In some cases, such as when creating SiC structures, a polymer binder can be selected, which leaves behind a significant amount of carbon residue within the brown part. Infiltration with molten Si will result in a reaction between the molten Si and the remaining carbon to produce more SiC, thus increasing the overall SiC content and reducing the fraction of metal Si in the final part. These and related approaches have been used to form interesting ceramic-matrix composites and ceramic-metal structures for a number of different applications. 5.4 Process Parameters and Modeling Use of optimum process parameters is extremely important for producing satisfactory parts using PBF processes. In this section, we will discuss “laser” processing and parameters, but by analogy the parameters and models discussed below could also be applied to other thermal energy sources, such as electron beams or infrared heaters. 5.4 Process Parameters and Modeling 5.4.1 123 Process Parameters In PBF, process parameters can be lumped into four categories: (1) laser-related parameters (laser power, spot size, pulse duration, pulse frequency, etc.), (2) scanrelated parameters (scan speed, scan spacing, and scan pattern), (3) powder-related parameters (particle shape, size, and distribution, powder bed density, layer thickness, material properties, etc.), and (4) temperature-related parameters (powder bed temperature, powder feeder temperature, temperature uniformity, etc.). It should be noted that most of these parameters are strongly interdependent and are mutually interacting. The required laser power, for instance, typically increases with melting point of the material and lower powder bed temperature, and also varies depending upon the absorptivity characteristics of the powder bed, which is influenced by material type and powder shape, size, and packing density. A typical PBF machine includes two galvanometers (one for the x-axis and one for the y-axis motion). Similar to stereolilthography, scanning often occurs in two modes, contour mode and fill mode, as shown in Fig. 5.7. In contour mode, the outline of the part cross section for a particular layer is scanned. This is typically done for accuracy and surface finish reasons around the perimeter. The rest of the cross section is then scanned using a fill pattern. A common fill pattern is a rastering technique whereby one axis is incrementally moved a laser scan width, and the other axis is continuously swept back and forth across the part being formed. In some cases the fill section is subdivided into strips (where each strip is scanned sequentially and the strip angle is rotated every layer) or squares (with each square being processed separately and randomly). Randomized scanning is sometimes utilized so that there is no preferential direction for residual stresses induced by the scanning. The use of strips or a square-based strategy is primarily for metal parts, whereas a simple raster pattern for the entire part (without subdividing into strips or squares) is typically used for polymers and other low-temperature processing. In addition to melt pool characteristics, scan pattern and scan strategy can have a profound impact on residual stress accumulation within a part. For instance, if a part Fig. 5.7 Scan strategies employed in PBF techniques 124 5 Powder Bed Fusion Processes is moved from one location to another within a machine, the exact laser paths to build the part may change. These laser path changes may cause the part to distort more in one location than another. Thus it is possible for a part to build successfully in one location but not in another location in the same machine due simply to how the scan strategy is applied in different locations. Powder shape, size, and distribution strongly influence laser absorption characteristics as well as powder bed density, powder bed thermal conductivity, and powder spreading. Finer particles provide greater surface area and absorb laser energy more efficiently than coarser particles. Powder bed temperature, laser power, scan speed, and scan spacing must be balanced to provide the best tradeoff between melt pool size, dimensional accuracy, surface finish, build rate, and mechanical properties. The powder bed temperature should be kept uniform and constant to achieve repeatable results. Generally, high-laser-power/high-bed-temperature combinations produce dense parts, but can result in part growth, poor recyclability, and difficulty cleaning parts. On the other hand, low-laser-power/lowbed-temperature combinations produce better dimensional accuracy, but result in lower density parts and a higher tendency for layer delamination. High-laser-power combined with low-part-bed-temperatures result in an increased tendency for nonuniform shrinkage and the build-up of residual stresses, leading to curling of parts. Laser power, spot size and scan speed, and bed temperature together determine the energy input needed to fuse the powder into a useable part. The longer the laser dwells in a particular location, the deeper the fusion depth and the larger the melt pool diameter. Typical layer thicknesses range from 0.02 to 0.15 mm. Operating at lower laser powers requires the use of lower scan speeds in order to ensure proper particle fusion. Melt pool size is highly dependent upon settings of laser power, scan speed, spot size, and bed temperature. Scan spacing should be selected to ensure a sufficient degree of melt pool overlap between adjacent lines of fused material to ensure robust mechanical properties. The powder bed density, as governed by powder shape, size, distribution, and spreading mechanism, can strongly influence the part quality. Powder bed densities typically range between 50 and 60 % for most commercially available powders, but may be as low as 30 % for irregular ceramic powders. Generally the higher the powder packing density, the higher the bed thermal conductivity and the better the part mechanical properties. Most commercialized PBF processes use continuous-wave (CW) lasers. Laserprocessing research with pulsed lasers, however, has demonstrated a number of potential benefits over CW lasers. In particular, the tendency of molten metal to form disconnected balls of molten metal, rather than a flat molten region on a powder bed surface, can be partially overcome by pulsed energy. Thus, it is likely that future PBF machines will be commercialized with both CW and pulsed lasers. 5.4 Process Parameters and Modeling 5.4.2 125 Applied Energy Correlations and Scan Patterns Many common physics, thermodynamics, and heat transfer models are relevant to PBF techniques. In particular, solutions for stationary and moving point-heatsources in an infinite media and homogenization equations (to estimate, for instance, powder bed thermo-physical properties based upon powder morphology, packing density, etc.) are commonly utilized. The solidification modeling discussed in the directed energy deposition (DED) chapter (Chap. 10) can also be applied to PBF processes. For the purposes of this chapter, a highly simplified model which estimates the energy-input characteristics of PBF processes is introduced and discussed with respect to process optimization for PBF processes. Melt pool formation and characteristics are fundamentally determined by the total amount of applied energy which is absorbed by the powder bed as the laser beam passes. Both the melt pool size and melt pool depth are a function of absorbed energy density. A simplified energy density equation has been used by numerous investigators as a simple method for correlating input process parameters to the density and strength of produced parts [2]. In their simplified model, applied energy density EA (also known as the Andrews number) can be found using (5.1): EA ¼ P=ðU  SPÞ ð5:1Þ where P is laser power, U is scan velocity, and SP is the scan spacing between parallel scan lines. In this simplified model, applied energy increases with increasing laser power and decreases with increasing velocity and scan spacing. For pLS, typical scan spacing values are ~100 μm, whereas typical laser spot sizes are ~300 μm. Thus, typically every point is scanned by multiple passes of the laser beam. Although (5.1) does not include powder absorptivity, heat of fusion, laser spot size, bed temperature, or other important characteristics, it provides the simplest analytical approach for optimizing machine performance for a material. For a given material, laser spot size and machine configuration, a series of experiments can be run to determine the minimum applied energy necessary to achieve adequate material fusion for the desired material properties. Subsequently, build speed can be maximized by utilizing the fastest combination of laser power, scan rate, and scan spacing for a particular machine architecture based upon (5.1). Optimization of build speed using applied energy is reasonably effective for PBF of polymer materials. However, when a molten pool of metal is present on a powder bed, a phenomenon called balling often occurs. When surface tension forces overcome a combination of dynamic fluid, gravitational and adhesion forces, the molten metal will form a ball. The surface energy driving force for metal powders to limit their surface area to volume ratio (which is minimized as a sphere) is much greater than the driving force for polymers, and thus this phenomenon is unimportant for polymers but critically important for metals. An example of balling tendency at various power, P, and scan speed, U, combinations is shown in 126 5 Powder Bed Fusion Processes Fig. 5.8 [3]. This figure illustrates five typical types of tracks which are formed at various process parameter combinations. A process map showing regions of power and scan speed combinations which result in each of these track types is shown in Fig. 5.9. As described by Childs et al. tracks of type A were continuous and flat topped or slightly concave. At slightly higher speeds, type B tracks became rounded and sank into the bed. As the speed increased, type C tracks became occasionally broken, although not with the regularity of type D tracks at higher speeds, whose regularly and frequently broken tracks are perfect examples of the balling effect. At even higher speeds, fragile tracks were formed (type E) where the maximum temperatures exceed the solidus temperature but do not reach the liquidus temperature (i.e., partially melted or Track type Fig. 5.8 Five examples of test tracks made in 150/ +75 μm M2 steel powder in an argon atmosphere with a CO2 laser beam of 1.1 mm spot size, at similar magnifications (# Professional Engineering Publishing, reproduced from T H C Childs, C Hauser, and M Badrossamay, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 219 (4), 2005) Fig. 5.9 Process map for track types shown in Fig. 5.8 (# Professional Engineering Publishing, reproduced from T H C Childs, C Hauser, and M Badrossamay, Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 219 (4), 2005) b c d e Direction of scan a P(W) 110 U(mm/s) 0.5 110 2 110 15 110 25 77 40 5.5 Powder Handling 127 liquid phase sintered tracks). In region F, at the highest speed, lowest power combinations, no melting occurred. When considering these results, it is clear that build speed optimization for metals is complex, as a simple maximization of scan speed for a particular power and scan spacing based on (5.1) is not possible. However, within process map regions A and B, (5.1) could still be used as a guide for process optimization. Numerous researchers have investigated residual stresses and distortion in laser PBF processes using analytical and finite element methods. These studies have shown that residual stresses and subsequent part deflection increase with increase in track length. Based on these observations, dividing the scan area into small squares or strips and then scanning each segment with short tracks is highly beneficial. Thus, there are multiple reasons for subdividing the layer cross section into small regions for metals. Randomization of square scanning (rather than scanning contiguous squares one after the other) and changing the primary scan direction between squares helps alleviate preferential build-up of residual stresses, as shown in Fig. 5.7. In addition, scanning of strips whereby the angle of the strip changes each layer has a positive effect on the build-up of residual stress. As a result, strips and square scan patterns are extensively utilized in PBF processes for metals. 5.5 Powder Handling 5.5.1 Powder Handling Challenges Several different systems for powder delivery in PBF processes have been developed. The lack of a single solution for powder delivery goes beyond simply avoiding patented embodiments of the counter-rotating roller. The development of other approaches has resulted in a broader range of powder types and morphologies which can be delivered. Any powder delivery system for PBF must meet at least four characteristics. 1. It must have a powder reservoir of sufficient volume to enable the process to build to the maximum build height without a need to pause the machine to refill the powder reservoir. 2. The correct volume of powder must be transported from the powder reservoir to the build platform that is sufficient to cover the previous layer but without wasteful excess material. 3. The powder must be spread to form a smooth, thin, repeatable layer of powder. 4. The powder spreading must not create excessive shear forces that disturb the previously processed layers. In addition, any powder delivery system must be able to deal with these universal characteristics of powder feeding. 128 5 Powder Bed Fusion Processes 1. As particle size decreases, interparticle friction and electrostatic forces increase. These result in a situation where powder can lose its flowability. (To illustrate this loss of flowability, compare the flow characteristics of a spoon full of granulated sugar to a spoon full of fine flour. The larger particle size sugar will flow out of the spoon at a relatively shallow angle, whereas the flour will stay in the spoon until the spoon is tipped at a large angle, at which point the flour will fall out as a large clump unless some perturbation (vibration, tapping, etc.) causes it to come out a small amount at a time. Thus, any effective powder delivery system must make the powder flowable for effective delivery to occur. 2. When the surface area to volume ratio of a particle increases, its surface energy increases and becomes more reactive. For certain materials, this means that the powder becomes explosive in the presence of oxygen; or it will burn if there is a spark. As a result, certain powders must be kept in an inert atmosphere while being processed, and powder handling should not result in the generation of sparks. 3. When handled, small particles have a tendency to become airborne and float as a cloud of particles. In PBF machines, airborne particles will settle on surrounding surfaces, which may cloud optics, reduce the sensitivity of sensors, deflect laser beams, and damage moving parts. In addition, airborne particles have an effective surface area greater than packed powders, increasing their tendency to explode or burn. As a result, the powder delivery system should be designed in such a way that it minimizes the creation of airborne particles. 4. Smaller powder particle sizes enable better surface finish, higher accuracy, and thinner layers. However, smaller powder particle sizes exacerbate all the problems just mentioned. As a result, each design for a powder delivery system is inherently a different approach to effectively feed the smallest possible powder particle sizes while minimizing the negative effects of these small powder particles. 5.5.2 Powder Handling Systems The earliest commercialized LS powder delivery system, illustrated in Fig. 5.1, is one approach to optimizing these powder handling issues. The two feed cartridges represent the powder reservoir with sufficient material to completely fill the build platform to its greatest build height. The correct amount of powder for each layer is provided by accurately incrementing the feed cartridge up a prescribed amount and the build platform down by the layer thickness. The raised powder is then pushed by the counter-rotating roller over the build platform, depositing the powder. As long as the height of the roller remains constant, layers will be created at the thickness with which the build platform moves. The counter-rotating action of the roller creates a “wave” of powder flowing in front of the cylinder. The counter-rotation pushes the powder up, fluidizing the powder being pushed, making it more flowable for a particular particle size and shape. The shear forces on the previously processed 5.5 Powder Handling a 129 Hopper b Hopper Feed Direction Roller Doctor blade Feed Direction Platform Newly deposited layer Powder Platform Newly deposited layer Fig. 5.10 Examples of hopper-based powder delivery systems [6] layers created by this counter-rotating roller are small, and thus the previously processed layers are relatively undisturbed. Another commonly utilized solution for powder spreading is a doctor blade. A doctor blade is simply a thin piece of metal that is used to scrape material across the surface of a powder bed. When a doctor blade is used, the powder is not fluidized. Thus, the shear forces applied to the previously deposited layer are greater than for a counter-rotating roller. This increased shear can be reduced if the doctor blade is ultrasonically vibrated, thus partly fluidizing the powder being pushed. An alternative approach to using a feed cartridge as a powder reservoir is to use a hopper feedi