key: cord-0059689-941ihuf4
authors: DebRoy, T.; Bhadeshia, H. K. D. H.
title: Picture to Parts, One Thin Metal Layer at a Time
date: 2021-01-05
journal: Innovations in Everyday Engineering Materials
DOI: 10.1007/978-3-030-57612-7_4
sha: cd04fabd1e3ac2a252b9c50343c1ef3839938a46
doc_id: 59689
cord_uid: 941ihuf4

Additive manufacturing is a process of manufacturing that creates a three-dimensional object by progressively depositing thin layers of material guided by a digital drawing. The creation of metallic objects using this technology is one of the fastest growing implementations, although other materials such as concrete, ceramics and polymers are also amenable to this manufacturing process, enabling applications that might not otherwise have been possible. Stainless steels, aluminium, titanium and nickel alloys in the form of powders or wires are melted by heating with a high-energy source such as a laser beam, electron beam or an electric arc. Metal printing is now used in aerospace, consumer products, health care, energy, automotive, marine and other industries because in cases where it has advantages over conventional methods. Here we examine the key processes and principles for printing metallic parts, their unique features, and review how their microstructure and properties develop. There are major challenges that need to be addressed for its continued expansion, requiring imagination and ingenuity to drive the future of metal printing.

. 1 An abstract metal object that is made depositing thin layers on top of each other in a sequence that reproduces the desired shape (a) (b) Fig. 4.2 The ceramic chess piece that is a castle, with a spiral staircase leading to the battlement, accompanied by a spiral hand-rail at the centre An additive manufacturing facility can in principle switch at a moments notice to the production of a different part once a digital design is available. This can have life-changing consequences. The COVID-19 pandemic led to a massive shortage of components for ventilators and equipment for the protection of medical staff. AM facilities in many parts of the world switched to the production of these vital goods, sometimes with innovative design features. In fact, the rate limiting factor was not the ability to manufacture but to get approval for the use of equipment in medical scenarios. 3 Bruce Wilmore, commander of the international space station showing a ratchet wrench made using a 3D printer [1] The International Space Station is resupplied at regular though long intervals compared with the normal shopping practices on Earth. When a ratcheting socket wrench was required, a digital file was sent electronically by NASA to a desktop AM machine located in the orbiting space station. The machine then built a wrench including the movable parts, all in one piece, layer by layer in 104 layers from polymers. On completion, an astronaut simply retrieved the wrench and used it with good effect, Fig. 4.3 . This was the first time an object designed on Earth was manufactured in space.

Metallic foams are used in sound damping, to provide rigidity in structures and have an advantage over polymeric foams in that they resist fire. They can form substrates for gaseous reactions. Their density can be less than that of water. Some applications require specific pore-geometries which present ideal opportunities in additive manufacturing. Figure 4.4 shows an open-pore metallic foam created in this manner. A metallic hollow-sphere is also illustrated-such spheres can have enormous specific strength (strength divided by density) so they can be used in lightweight construction with the advantage that additive manufacturing permits the design of such spheres as opposed to the acceptance of geometries limited by conventional manufacturing processes. Figure 4 .5 shows the distribution of revenues from consumer products, automotive, health care, aerospace, marine and other industries. The viability of these products in the context of AM depends on the ability to manufacture more easily than using conventional methods. Nevertheless, if the component is sufficiently sophisticated, then it becomes possible to manufacture it more rapidly than conventional methods given the avoidance of excessive machining, assembly and inspection. i.e., consumer products, automobile, health care and other industries. The pictures represent a part in an industrial sector. For example, GE's fuel nozzle is a part used in aerospace industry. The revenue from all printed parts in the aerospace industry is 13% of the total revenue of all metal printed parts from all industrial sectors. Currently, the revenue generated by metal printing is growing but small compared with the total manufacturing industry [2] 

Some of the aspects of the additive manufacturing process have been described above, but it is useful to present the big picture before dwelling into detail, Disadvantages Complex shapes that are difficult to produce by conventional manufacturing processes can be printed Productivity is limited because components are created in a stepwise manner. Resolution is limited by this step size (defined by powder size, layer thickness), both in the raster pattern and the thickness of the two-dimensional slice Economical use of material given that final finishing of the product is all that is required Surface roughness can present challengesfor example, the cooling channels in jet engine turbine blades can be less than 1 mm in diameter Relatively small capital investment in manufacturing equipment High productivity requires banks of printing machines Ability to rapidly switch to production of different components Residual stresses and distortion can compromise metallic components The print-on-demand model can be implemented Limited availability and limited variety of metallic powders or wires at affordable cost. In contrast, the alloys available for conventional manufacture have many orders of magnitude greater variety and lower cost Customised parts can be produced without new tooling or equipment Functionally graded components possible by using different "inks" during the creation of a single component Structural integrity of critical metallic components is difficult to achieve.

Metallic components usually have less than 100% density

The heat sources required for metal deposition include continuous-wave carbon dioxide lasers, solid-state Nd:YAG lasers, Yb fibre lasers, electron beams and electric arc. Alloy powders or wires are commonly used as feedstock. A moving heat source melts an alloy layer by layer which subsequently solidifies. The trajectory of the heat source is determined from the algorithm for the part to be manufactured. In the powder bed fusion process, shown schematically in Fig. 4 .6a, the bed is lowered by a small distance after each layer is deposited, a roller spreads a thin layer of powder over the part and its surrounding area before another layer of the alloy is deposited with the help of a laser or electron beam [3] . In the directedenergy deposition system shown in Fig. 4 .6b, the powder is supplied by a powder feeder co-axial with a laser beam. Both the laser beam and the powder feeder move relative to the part. The powder particles are heated during their flight and after they impinge on the part. After each layer of metal is deposited, the substrate is lowered slightly so that the distance between the heat source and the deposition surface does not change. A laser beam, an electron beam, or an electric arc can be used as a heat source for the directed-energy deposition process. Either a stream of powder or a wire feed can be used to build components. Table 4 .2 shows the main features of the common metal printing processes which can help select the optimum method for the intended component. They all take considerably longer times to build parts than casting or injection moulding. A larger powder particle-size or wire diameter limits the resolution possible, making it more difficult to achieve fine features. The powder bed process uses finer powders to deposit intricate features; greater scanning speeds are also possible compared with the direct energy deposition method. Consequently, the powder bed fusion method is associated with less heat input per unit length of the deposit, the layers are much thinner, and they cool much faster. Smaller laser spot diameters and finer powders allow better control of geometric features in the parts. Slower deposition rates produce improved surface quality using thinner layers at the expense of productivity.

When large parts are made they tend to retain heat for a longer duration and the cooling rates are relatively slow. Often, items larger than 30 × 30 × 30 cm are produced in near net-shape by melting of a wire followed by machining. High deposition rates are achieved by simultaneously using two wire electrodes. Wire based techniques that use welding power sources are gaining popularity, partly because of the ready availability of wires that would normally be used in welding technologies.

The printing of metals is attractive because it can produce components that cannot be easily and economically produced by conventional manufacturing. Jet engine fuel nozzles are now made routinely in a custom-designed factory [4]. In the past, each fuel nozzle was an assembly of about twenty individual parts. They are now partially manufactured by laser melting of alloy powders, layer by layer, in 20 µm thin layers (one-fifth of the thickness of a human hair). The fuel nozzle, shown in Fig. 4 .7a, reduces the number of assembly steps required and it is claimed to be five times as durable as the conventional nozzles.

Metal printing offers additional opportunities to create components with site specific chemical composition and properties. The nuclear industry often uses joints between steel and nickel alloys, the latter being more resistant to elevated temperature exposure. An abrupt change from one to the other causes huge changes in the vicinity of the joints, in particular, the partitioning of carbon on to the steel side where it locally embrittles the structure. This abrupt change can be mitigated by designing a joint where the chemical composition varies gently from the steel to nickel alloy. This can easily be achieved using directed energy deposition to create a compositionally graded joint as illustrated in Fig. 4 .7b. All that is needed is to feed different proportions of the component powders during deposition. Compositional profile of a functionally graded joint between a nickel alloy 800H and a Cr-Mn steel [5] . The points are measured values of concentrations in layers directly deposited using a laser as the energy source. Image adapted from [5] with the permission of Elsevier

Metal printing now routinely makes customised products such as patient-specific implants and legacy products where the supply chain no longer exists.

The physics associated with additive manufacturing process affects the microstructure, properties, and the ability to service the manufactured components. Metal printing involves heating and melting of the feedstock, followed by solidification of the liquid and then cooling in the solid state. In the direct energy processes, the feedstock receives heat even before reaching the build surface and may absorb heat during flight before impinging on the deposition surface, which also receives heat directly from the source. The powder or wire therefore melts quickly and the molten pool is propagated along the predetermined track.

The highest temperature on the melt pool surface is attained directly below the heat source and then decreases with distance from this location. Surface tension is a function of temperature so its variation with position creates the so-called Marangoni stress on the surface of the molten pool, which makes the liquid move from regions of low to high surface tension.

The three-dimensional flow of liquid metal in the pool is important because it affects both the dissipation of heat and the mixing of the feedstock with the molten metal from the pre-existing layers. However, it is difficult to determine the motion experimentally since liquid metals are opaque and the pool is small and moves rapidly. A recourse is to simulate metal flow by numerically solving the equations of conservation of mass, momentum and energy with initial and boundary conditions [5, 6] . This approach of simulating liquid metal velocities in a computer rather than by direct experimental measurement is widely adapted in engineering practice. Most of our current knowledge of the flow of liquid metal in AM originated from numerical modelling. Figure 4 .8a shows the computed flow pattern inside a molten pool during powder bed fusion. The liquid metal moves away from under the heat source to the periphery of the liquid pool, turns around and recirculates. The speed and orientation of circulation determine the extent of convective heat transfer and the mixing of the hot and the cold fluids in the pool.

The circulation pattern obviously influences the temperature distribution in the liquid alloy, its heating and cooling rates, solidification pattern, and the evolution of various solid phases that make up the microstructure. The solidification morphology, grain structure and the phases that form define the microstructure and the mechanical properties of the printed component.

An important consequence of building parts layer by layer is the temperature excursion that each location of the part experiences. Unlike most other materials processing operation, in AM each location of the part experiences multiple temperature peaks. For example, Fig. 4 .8b shows the computed thermal cycles at various monitoring locations inside a part. Temperatures at the mid-height and midlength of several layers are shown as a function of time. The first temperature peak The subsequent peaks occur during the deposition of the upper layers. So, at each location, the microstructure and the grain structure of the alloy that forms after the first thermal cycle are often changed by the subsequent thermal cycles depending on the specific temperatures and times. These thermal cycles affect the evolution of microstructure and the eventual mechanical properties of the part. The substrates used in AM are effective heat sinks. As a result, the peak temperatures attained in the lower layers close to the substrate are somewhat lower than those in the upper layers [6] . In the upper layers the distance from the heat sink increases and the peak temperature rises because of the reduced heat loss. So, the thermal cycles are inherently spatially dependent. An important consequence of this result is the microstructural asymmetry of the part. Since the structure affects properties, the properties may also be inherently different at different locations. Each location in the component experiences multiple thermal cycles and phase transformations, grain growth, residual stresses, distortion that affect the mechanical properties of the component.

The structure and properties of alloys depend on their thermal histories that are affected by many variables such as the scanning speed, power, power density, scanning pattern, part geometry and the thermo-physical properties of the alloy. All these variables affect heat transfer within the part, which control temperature profiles and cooling rates. Given the many causative variables and their wide range of values, it is no surprise that the cooling rates reported in the literature for the Fig. 4.9 Reported cooling rates for the printing of stainless steel for a wide variety of processing conditions [2] . (a) Powder bed fusion using laser. (b) Directed energy deposition using laser. (c) Direct energy deposition using electric arc printing of a stainless steel in Fig. 4.9 show a wide range as a function of processing variables [2] . The heat input, i.e., the energy deposited per unit length, has a seminal influence of the cooling rate. The wide range at the same time provides an opportunity to customise the cooling rate for specific properties. The cooling rate may not of course be uniform throughout the component, which will lead to corresponding variations in properties that must be taken into account in the design process.

The morphology of the solidification structure is affected by the temperature field, which in turn depends on the scanning pattern used during deposition. Consider the two scanning patterns, one where the heat source travels along the same direction, i.e., always scanning from left to right and another where the direction alternates between left to right followed by right to left, in the context of the deposition of a nickel base alloy, Fig. 4 .10 [7] . Solidification occurs by the epitaxial growth from the substrate of columnar dendrites, with growth direction influenced by that in which heat flows, which in turn depends on the scan pattern. The orientation of the dendrites is identical in all layers when the scan direction is maintained constant. In contrast, alternating the scan direction causes corresponding changes in the dendrite orientations between adjacent layers [8] . The orientations of crystals are important in determining properties because if they are all similarly oriented, they may not, for example, have the same fracture properties as when they are differently oriented (i.e., differently textures). Therefore, the deposition sequence matters as illustrated in Fig. 4.10 . There are other properties, such as hardness, that are influenced by the chemical composition of the deposit. If the composition of a particular class of nickel alloys is expressed empirically in terms of a single parameter φ: φ = w Ni + 0.65w Cr + 0.98w Mo + 1.05w Mn + 0.35w Si + 12.6w C − 6.36w Al + 3.80w B + 0.01w Co + 0.26w Fe + 7.06w Hf + 1.20w Nb + 4.95w Ta + 5.78w Ti + 2.88w W 4.11 The dependence of Vickers hardness on chemical composition of nickel alloys. Various processing conditions were used. For details and the limitations of the analysis, see [7] where w i is the weight percent of solute i, then φ is found to correlate strongly with the hardness of the deposit, Fig. 4. 11. An exciting opportunity will arise when sufficient data on additive manufacturing are openly available so that they can be subjected to machine learning techniques to reveal quantitative patterns that can be used to make further advances. Such work has been developed using neural networks, for the powder bed fusion based on electron beams as the energy sources, to estimate the strength of the deposits [9] .

Porosity, lack of fusion, solidification cracking, residual stresses and distortion are issues that need to be improved upon for the qualification of printed parts. When alloys contain volatile elements, some might selectively evaporate leading to uncontrolled changes in the intended chemical composition.

Several mechanisms are responsible for the porosity formation. The feedstock may contain dissolved gases that are evolved during solidification to form small, spherical pores. In the powder bed fusion process, gases present in the inter-particle spaces may become entrapped. The very high-temperature vapour zone beneath an electron or laser beam may collapse due to instability of the power density or local changes in powder packing, leading to porosity when solidification occurs after the beam has left the locality.

Adjacent layers of deposit may not fuse together if the fused region does not penetrate the surface to a sufficient depth. Insufficient overlap between adjacent tracks in the scan or between layers may leave unmelted regions in between. So the appropriate choice of deposition conditions is vital to ensure a dense solid, and the conditions may differ with the types of material deposited. Figure 4 .12a shows macroscopic defects that arise during the powder bed process due to a lack of fusion during deposition. In Fig. 4 .12b, successive layers separate because the thermal contraction stresses that are not homogeneously distributed exceed the strength of the interface between the layers.

Solidification cracking occurs when the tensile stress due to volume shrinkage associated with the liquid→solid transformation, is such that during cooling, the shrinkage stresses exceed the strength of the solidifying region the elevated temperatures involved. The composition of the alloy, geometry of the deposited bead and scanning speed can all affect this type of cracking. Irregular cracks, up to a few millimetres in size can be generated. This type of cracking depends on the chemical composition of the alloy, particularly when it contains impurities such that the temperature range over which a mixture of solid and liquid is extended. Residual stresses are those that exist in a body even at equilibrium. They evolve during spatially non-uniform heating and cooling of the metal, thermal expansion and contraction, phase transformations and uneven distribution of plastic strains. Mitigation strategies include control of the substrate preheat temperature, shorter deposition length or scanning in smaller bits, more rapid scanning and thinner layers. A large preheat temperature can also be helpful in preventing solidification cracking. Table 4 .3 lists a few methods to reduce the effects of residual stresses.

Most engineering alloys typically contain one or more volatile alloying elements. Manganese and chromium in stainless steels, magnesium and zinc in aluminium alloys and aluminium in titanium alloys are examples. Their selective evaporation changes the chemical composition, which may or may not be an issue. A reduction in peak temperature by selecting an appropriate power distribution pattern, higher heat source power that results in smaller surface-to-volume ratio of the molten pool, and faster scanning speed will minimise the problem.

A distinguishing feature of additive manufacturing is its ability to create internal features such as cooling channels within an otherwise solid part. If deposition leads to distortion and surface roughness, then these internal features may not be as intended in the original design. A smaller liquid pool size with lower heat input can often help, but this must inevitably reduce productivity and increase cost.

Additive manufacturing is obviously a layer-by-layer process, so curved surfaces are approximated in a stepwise profile. A smaller step size is needed for a smoother macroscopic curvature, but it may ultimately be necessary to smooth the surface mechanically. Tiny metal drops and unmelted or partially melted powder particles ejected from or near the fusion zone (i.e., spatter) by high-speed metal vapours and gases, often land on the build surfaces and contribute to roughness. Large unmelted powder particles or "balls" are often found at the edge of the molten pool. High heat input may reduce the severity of the problem by melting large particles, but small powder particles also result in a smoother surface. Powders as fine as 20 µm have been used to ensure a better surface finish.

The scientific challenges include the interrelation between processing, microstructure, properties and performance, microstructure control, minimisation of defects, and poor solidification and grain structure. A better understanding can eliminate some of the trial-and-error used in fixing AM parameters. Similarly, rapid qualification of parts, overcoming geometric limitations, scaling-up, printing sequence, and the health and safety concerns of handling fine metal particles, are examples of technological problems. Commercial challenges include cost competitiveness, availability of feedstock and a need for standards.

To facilitate more rapid printing of large parts, multiple heat sources and wire feed mechanisms are explored, together with hybrid methods that use a combination of manufacturing technologies.

Alloying elements Elements in an alloy added to enhance its properties.

Columns of tree like solid structures that form during solidification of liquid alloys.

Mechanism of heat transfer in a stationary solid or liquid due to temperature difference.

Movement of liquids or gases. If hot gases and liquids are in a motion, they can carry significant amounts of heat with their motion. Electron beam A stream of energetic electrons capable of heating and melting alloys in a focused area. Epitaxial growth Atomic arrangements of a new growth layer conforming to the structure of the existing layer. Laser beam A device that can emit an intense beam of light through stimulated emission of radiation. A focused laser beam can melt and vaporise alloys.

Flow of liquids from low to high surface tension regions.

The magnified pattern of a surface observed using a microscope. Solidification morphology Shape of the solids that form from the liquid alloys. Surface tension A measure of how closely molecules on the surface stick to each other.

3D printed wrench

Scientific, technological and economic issues in metal printing and their solutions

Additive manufacturing of metallic components-process, structure, and properties

Additive manufacturing of functionally graded transition joints between ferritic and austenitic alloys

Melt pool geometry, peak temperature, and solidification parameters during laser additive manufacturing

Evolution of solidification texture during additive manufacturing

Porosity evolution and its thermodynamic mechanism of randomly packed powder-bed during selective laser melting of Inconel 718 alloy

Progress toward an integration of process-structure-property-performance models for "threedimensional (3-D) printing" of titanium alloys

Producing crackfree, high density M2 HSS parts by selective laser melting: pre-heating the baseplate