• No results found

Consolidation phenomena in laser and powder-bed based layered manufacturing

N/A
N/A
Protected

Academic year: 2021

Share "Consolidation phenomena in laser and powder-bed based layered manufacturing"

Copied!
30
0
0

Loading.... (view fulltext now)

Full text

(1)

Consolidation phenomena in laser and powder-bed based layered manufacturing

J.-P. Kruth1(1), G. Levy2(1), F. Klocke3(1), T.H.C. Childs 4(1) 1K.U.Leuven, Division PMA, Leuven, Belgium 2 FHS - University of Applied Sciences St. Gallen, Switzerland 3 Fraunhofer, Institute for Production Technology IPT, Aachen, Germany

4 University of Leeds, School of Mechanical Engineering, UK

Abstract

Layered manufacturing (LM) is gaining ground for manufacturing prototypes (RP), tools (RT) and functional end products (RM). Laser and powder bed based manufacturing (i.e. selective laser sintering/melting or its variants) holds a special place within the variety of LM processes: no other LM techniques allow processing polymers, metals, ceramics as well as many types of composites. To do so, however, quite some different powder consolidation mechanisms are invoked: solid state sintering, liquid phase sintering, partial melting, full melting, chemical binding, etc. The paper describes which type of laser-induced consolidation can be applied to what type of material. It tries to understand the underlying physical mechanisms and the interaction with the material properties. The paper demonstrates that, although SLS/SLM can process polymers, metals, ceramics and composites, quite some limitations and problems cause the palette of applicable materials still to be limited. There is still a long way to go in tuning the processes and materials in order to enlarge the applicability of LM. This is not surprising if one compares it to the decades of R&D work devoted to tuning processes and materials for hot or cold forming, metal cutting (e.g. development of free machining steels), casting and injection moulding (including powder injection moulding: MIM, CIM, etc.).

Keywords

Rapid Prototyping and Manufacturing, Selective Laser Sintering (SLS), Selective Laser Melting (SLM) 1 INTRODUCTION

Layered manufacturing (LM) goes back to the late 1980’s, early 1990’s [95, 98] with a clear breakthrough in 1994 at which time machine sales took off exponentially: see Figure 1 [216]. Today, distinction is made between Rapid Prototyping (RP) and Rapid Manufacturing (RM) [49, 121, 164].

x RP means the production of prototypes, visual design aids, touch, feel, fit and assembly test parts, etc., that are used in the product development phase and are not meant to be equivalent to real production parts at all levels.

x RM means the production of functional parts that are meant to be used as real production parts (end products) and should meet the various basic requirements put to such production parts. Rapid Tooling (RT) may be considered in this context as a sub-category of RM, i.e. production of functional tool components produced by layered manufacturing. [41, 153, 181, 208]

Many layered manufacturing techniques exist today, the most popular being: photo-polymerisation (Stereolithography (SLA) and its derivates), ink-jet printing (IJP), 3D printing (3DP), Fused Deposition Modelling (FDM), Selective Laser Sintering or Melting (SLS/SLM and EBM1) and to a lesser extend Laminated Object Manufacturing (LOM and similar sheet stacking processes) and laser cladding (LC) processes [98]. The importance of these technologies is confirmed by a recent study of NACFAM (National Council for Advanced Manufacturing, USA) that has identified Rapid Manufacturing as the most innovative and potentially disruptive manufacturing technology to emerge within the next 3 to 5 years [139]. Many of those techniques are however limited to Rapid Prototyping as they do not allow common engineering materials to be processed with sufficient mechanical properties (polymers, metals, ceramics, and composites thereof).

In the prospect of Rapid Manufacturing, SLS/SLM seems to be the most versatile process, capable of

1

Selective Electron Beam Melting (EBM) may be considered as a variant of this category.

processing engineering polymers, metals, ceramics and a wide range of composites. In order to cover this wide range of materials, the laser processing of powder materials in SLS/SLM calls on various consolidation mechanisms: the binding of polymers, metals, ceramics and their mixtures or composites varies substantially with initial powder composition, the final aimed material composition and the aimed material structure (aimed porosity or density, microstructure, properties, etc.). Some process variants may apply post-processing after SLS/SLM: in such case, the selected layer consolidation mechanism should account for the post-process as well: e.g. need for open porosity if post-infiltration is applied.

Figure 1 : Yearly additive system unit sales worldwide [216]

This paper gives a survey of powder consolidation

mechanisms applicable in SLS and SLM for various types of materials.

2 CLASSIFICATION

2.1 Types of powder consolidation mechanism (binding mechanism)

Laser-based consolidation of 3D parts from layers of powder material pre-deposited on a ‘build platform’ is commonly referred to as Selective Laser Sintering (SLS) or Selective Laser Melting (SLM). The distinction between SLS and SLM is rough, vague and does not cover all

(2)

types of consolidation. Various papers describe possible ways to consolidate material with a laser [18, 126]. A more detailed classification of consolidation mechanisms is given in Figure 2.

Solid State Sintering (SSS) is a consolidation process

occurring below the material’s melting temperature: diffusion of atoms in solid state (volume diffusion, grain boundary diffusion or surface diffusion [82, 156]) will create necks between adjacent powder particles that will grow with time (Figure 3) [66, 204]. This binding mechanism is only rarely applied in layer manufacturing (LM) as diffusion of atoms in solid state is slow and not very compatible with the desired high laser scan speed that should yield process productivity and economic feasibility. However, it is one of the consolidation mechanisms suited for consolidating ceramic powders [121]: see §2.2. SSS diffusion sintering is sometimes applied in post-processing of metallic or ceramic parts after initial (partial) laser consolidation using solid state, liquid phase or chemical binding [121]: see §4.

Figure 3 : Typical neck formed between two stainless steel powder grains obtained with Nd:YAG laser [204] Liquid Phase sintering (LPS) and partial melting include a number of binding mechanisms in which part of the powder material is melted while other parts remain solid. The liquefied material will spread between the solid particles almost instantaneously as it is driven by intense capillary forces. This allows much higher laser scan velocities than for SSS. The material that melts may be different from the one remaining solid: the former ‘low melting point’ material is then often called the binder, while the ‘high melting point’ material is generally called the structural material. Different ways exist to bring those distinct binder and structural materials together:

x mixture of two-component powder (i.e. separate binder and structural powder particles),

x using composite powder particles that have a micro composite structure containing the structural and binder material, or

x using coated particles where the binder material is applied as a coating around the structural material. The binder material (coating) will preferentially absorb the incident laser radiation. This enforces the intended melting of the binder.

When a different binder and structural material is applied, the two materials may differ substantially: e.g. polymer binder versus metallic structural material, metallic binder vs. ceramic structural material, low-melting metal (e.g. Cu) vs. high melting metal (e.g. steel or Fe). Examples and references are given in next sections. In all cases, the binder may be permanent (i.e. remain as part of the final product) or may be sacrificial (i.e. removed during a de-binding cycle). In both cases, the ‘green part’ coming out of the SLS machine might be porous (and become even more porous after de-binding) and might require a post-densification process: the latter being generally either a furnace infiltration process or a furnace post-sintering process (thermal sintering, HIP or other). In most cases, thermal post-sintering without applying pressure is not feasible: the porosity of the ‘green’ SLS part is still too high (25-50%) to initiate further densification by thermal heat only. Alternatively, this process may be used to produce porous structures [122, 168, 185].

Partial melting is also possible even in cases where no

clearly distinct binder and structural materials are used. SLS parameters are then adjusted to only partially melt the powder particles, which may be a single phase material or a mixture of different powders but without distinct binder powder material (particles):

1. When the heat supplied to a powder particle is insufficient to melt the whole particle, only a shell at the grain border is melted. The core of the grain remains solid. This way the molten material will form necks between the particles and act as a binder between the nonmolten particle cores. This binding mechanism can arise as well with metals as with polymers, although the consolidation of polymer powders may also result from other mechanisms (consolidation at the glass transition temperature, which is lower than the melting temperature, polymer chain rearrangement and cross-linking).

Such partial melting phenomenon and the resulting liquid neck formation were modeled at EPFL, Lausanne, [53]. Using a simple thermal model, skin and core temperatures of the powder particles were Figure 2 : Laser-based powder consolidation mechanisms

(3)

calculated. This way the minimal pulse energy to fully melt the particle can be calculated. Below this value, the core temperature never exceeds the melting temperature and only partial melting is obtained.

2. Powders consisting of multiple phases or a mixture of different powder particles can be classified as partial melting when they are only partially molten. For example, the University of Leuven experimented with a Fe-Fe3P-Ni-Cu powder mixture, aiming at the production of fully dense parts [93]. The resulting part clearly contains unmolten Fe particles (see Figure 37). None of the 4 powders can be identified as a distinct binder material. The addition however of a melting point lowering additive like Fe3P or Cu3P is favorable in making the process more energy efficient: see details in §3.2.2, section B2.

3. Partial melting may also occur when using a single material powder having a bi-modal distribution: small particles are melted, while larger ones remains solid. Liquid phase sintering and partial melting may have advantage over full melting (see below), even when aiming at full dense parts. The former allows the laser consolidation speed (scan speed, scan spacing…) to be increased drastically. The resulting economic benefit has to be considered against the disadvantage of the need for post-densification. Comparative studies demonstrated that molds made from PA-coated steel that are post-infiltrated with copper may not only be more cost effective [103], but also technically may perform better due to the beneficial properties of copper [106, 208].

Full melting is a third major consolidation mechanism,

often applied to achieve fully dense parts without need for any post-process densification. Major progress has been achieved towards full melting of metal powders up to densities of 99.9% by applying modern laser sources and optics yielding high energy densities in the spot [105]. Dedicated SLM machines are now offered by 4 German vendors [58, 103]. As compared to early day SLS machines equipped with CO2 lasers, these machines apply solid state lasers: diode pumped Nd:YAG, fibre or disc lasers [150].

Full melting has the main advantage to produce almost full dense products in one step, but also has drawbacks that require careful process control:

x The high temperature gradients and densification ratio during the process yield high internal stresses or part distortion (from 50% powder porosity to 100% density in one step) [6, 134].

x The risk of balling and dross formation in the melt pool2 may result in bad surface finish.

Chemical induced binding is a fourth main consolidation

mechanism. Today it is not commonly used in commercial LM equipment, but it turns out to be a feasible consolidation mechanism for polymers, metals and ceramics. IPT-Aachen has applied chemical binding for laser sintering SiC ceramic powder [88]. Chemical induced binding of metals could e.g. be invoked for sintering Al powder by making the Al react with the N2 atmosphere used commonly in SLS machines, thus creating an AlN binder phase holding the Al particles together. Such mechanism has been used by the University of Queensland, be it during post-sintering in a N2 furnace, rather than during the initial SLS phase itself, at which stage a sacrificial polymer binder was used to consolidate

2 Those melt pool phenomena have been studied within the CIRP STC-E Working Group on “Melt pool phenomena in SLS/SLM/LC” that met in Leuven (Jan. 2004), Aachen (Oct. 2004) and Leeds (Sept. 2005). They are dealt with in §2.4 and 3.5.

the green Al part [172]. Other (potential) examples of “Selective Laser Reactive Sintering” were investigated at the University of Texas at Austin [11]. Some investigators tried to apply self-propagating high-temperature synthesis (SHS) in which a low laser power induces a self-propagating exothermal reaction capable of self-heating and consolidating the powder in a controllable way [107,177,182]. This allows reducing the laser power significantly. The metallic liquid phase stays longer and the accuracy of the produced piece will be higher. Shi et. al. applied a SHS process to laser sinter metallic materials (Ni-Al, Al-Ti, Ni-Ti) [176]. The Osaka Sangyo University tried to produce cermets (Al2O3-Cu), ceramics (MoSi2), and intermetallics (TiAl) by SLS induced chemical reaction [80]. This will be discussed in §3.4.7.

2.2 Type of consolidation mechanisms versus materials

The applicable type of consolidation is basically dependent on the material being processed. Table 1 gives an overview of what material can be basically consolidated with the various binding mechanisms.

2.3 Influence of laser type

Different types of lasers are used for powder-bed based RM: CO2, Nd:YAG (lamp pumped, diode pumped, or Q-swithed) [60], fibre lasers, disc lasers, Cu-vapour lasers [76]. The type of laser has a large influence on the consolidation of powder particles, because:

x laser absorption of various materials greatly depends on the laser wavelength: e.g. high absorption of polymers and oxy-ceramics at 10.6 μm wavelength of CO2 laser; high absorption of metals and carbo-ceramics at ±1 μm wavelength of Nd:YAG, fibre and disc lasers [105, 196];

x the possible consolidation mechanism highly depends on features like energy density: e.g. SLM of metals and ceramics requires a high energy density that is more easily achieved with fibre and disc lasers (high beam quality, i.e. high focusability);

x laser mode (continous, pulsed, Q-switched,…) also has a large influence on consolidation (see end of §2.4).

Several researchers investigated the influence of different laser types on the absorption and consolidation of powders in SLS/SLM. Some did it experimentally [100, 114, 156], others by simulation [114].

2.4 Problems in consolidation

There are two main general problems in consolidation. One, applicable to all process variants other than the full melting of metals, stems from the short material heating times caused by the scanning laser beam, relative to the time required for consolidation, and the fact that it is only temperature effects and gravity and capillary forces that can provide the driving force (there is no mechanical pressure, as in moulding processes). It can lead, as has already been written, to porosity in parts. Post-processing is then required if a pore-free material is needed. For polymer processing, it means that melt viscosity must be kept less than some critical value. For example Nylon-12 (PA 12) with an in-process viscosity K | 100 Pas can be fully densified under surface tension driving forces J | 30 mN/m (J/K | 0.3 10-3 m/s), but polycarbonate (viscosity |

5000 Pas, and similar surface tension) can not [28]. The other problem, important for the full melting of metals, is quite the opposite. Here the viscosity K is about 1 to 5 mPas [16], the surface tension J about 1 to 2 N/m [12] and consequently J/K | 100 to 1000 m/s. The problem is not time for flow, but controlling the flow [87]. The pool of

(4)

molten metal must wet the previously processed metal below it (except on the rare and sometimes problematic occasions that it is an overhang). And when it solidifies, its upper surface must be flat enough to enable a next layer of powder to be spread over it. Both materials properties and processing variables can influence these things. As far as wetting the material below is concerned, contamination, usually by oxygen, and in that case either as oxide, impurities within the powder or as trapped gas in the powder bed, is a main materials cause of consolidation problems. For example, the surface tension of pure liquid iron near its melting point is almost 2 N/m, but 0.02 wt.% of oxygen reduces this towards 1 N/m [12, 110]. Liquid iron will not spread out over an iron oxide covered surface. Even in the absence of such contamination, there may be a problem of a liquid metal wetting its solid form, if the solid has almost the same temperature. In this homologous wetting case, there is no driving force for wetting. It has been suggested on the basis of experimental studies on a stainless steel [35] that it is necessary for the previously processed material to be re-melted by the currently melted material. The excess energy (and hence the higher peak temperatures in the melt pool) required for this will have consequences for the quality of the upper surface of the melt pool.

When temperature gradients are created in a melt pool, there is the potential for convective motions in the pool to reduce those gradients. Temperature gradients through the depth of the pool, causing density gradients, and hence buoyancy forces, are probably of no importance to the millimeter or less melt pool sizes of SLM (compared to the much larger sizes that can make buoyancy effects relevant to welding [137]), although see [142]. But temperature gradients in the surface, coupled with a temperature dependent surface tension, can cause rapid motions (tens of cm/s or more), known as thermo-capillary or Marangoni flow. The equation below defines the dimensionless Marangoni number Ma.

a

d

w

d

dT

w

M

dx

w

dT

dx

w

J

N

J

N

K

§

ª

º

ª º

·

§ ·

{

§

ª

º ª

º

ª º

·

§

¨

«

¬

¼

»

¬ ¼

« »

¸

¨ ¸

© ¹

¨

¬

«

¼ ¬

» «

»

¼

« »

¬ ¼

¸

¨ ¸

©

©

¹

©

K

¹

·

¹

(1) It is the ratio of the speed by which a surface temperature gradient dT/dx may be reduced by convection or conduction (w is the linear size of the pool, N is its thermal diffusivity) [137]. For typical ferrous material property values, also supposing an initial temperature gradient dT/dx to be due to the difference between liquidus and solidus temperatures over the half-width of the melt pool, and w = 1 mm, Ma can be calculated to be of the order 1000. Marangoni flow dominates.

When dJ/dT is negative, as it naturally is with decreasing surface energy from the liquidus to the vapourisation temperature, the Marangoni flow is from high to low temperature, that is to say from the centre to the edge of the melt pool, and a flattened or dished surface is created. 3

HM = hardmetals

But with oxygen contamination of steels, and dependent on sulphur and nitrogen content too, dJ/dT can be positive [83, 159]. Then the melt pool becomes humped. Again, such effects are clearly seen in welding [127, 128], as well as in SLM [156, 158].

In the previous paragraphs, different effects at the under and top side of the melt pool have been considered. But in SLM there is a leading edge to the melt pool too, where the surface, subject to the surface tension induced flow, is advancing into powder that at some distance ahead of the melt front will certainly still be below the solidus temperature. The wetting or not of this solid powder and, if it is wetted, its dragging or not into the melt pool, are further complications of the consolidation process that have not been systematically studied. It may be that the compromises needed to balance good wetting or remelting of previously melted material, with the top surface of the melt pool freezing sufficiently flat, and the powder bed not being disturbed too much ahead of and around a currently melting track, will put a large constraint on metals and powder preparations that are suitable for processing by SLM. Much basic work remains to be done in this area.

The consolidation flows and the materials dependent phenomena accompanying them and discussed above differ from flow instabilities in a pool that can arise because of its shape. Long thin melt pools are known to break up into balls, called “balling” and commonly described as due to Rayleigh instabilities [3, 19, 74, 87, 135, 136, 157]. For a free melt pool, as would describe a single melted track in a deep powder bed, balling is predicted for pools (imagined to be cylindrical) of length to diameter ratio greater than S. Balling as this ratio is approached is reported by [27, 156, 157]. That work also showed length to diameter ratio increasing with laser scan speed. Thus in this case, instability limits processing to a low scan speed. A melt pool on a solid substrate can also be unstable, depending on its contact behaviour with the solid [38, 169]. A limiting length to diameter ratio of 2.1 has been reported [96], also creating an upper limit to the laser scan speed.

Most balling studies have been performed in conditions of single line or raster scanning powder beds when the laser-caused temperature rise has essentially been due to the current line scan. At higher scanning velocities and/or shorter vector lengths, however, previously scanned tracks are not yet solidified when a new track is being scanned. Thus, the actual width of the melt pool is enlarged, since it now spans multiple scan tracks [135, 136]. Therefore, the length-to-width ratio may actually be reduced (even if the melt pool length enlarges), resulting in a more stable behaviour compared with a single track scanned with the same parameters. Mercelis [135] showed that melt pool dimensions may depend much more on the geometry of the 3D part and the 2D scanning path, than on the laser power and scanning velocity. Therefore, also the scan track instability cannot be Solid State Sintering Liquid Phase Sintering Partial melting Full melting Chemical

Polymers No Yes (e.g. PS) Yes (e.g. PA) Seldom (e.g. partial cross linking PMMA)

Metals Seldom Yes (many kinds) Yes (steel, Ti...) Yes (e.g. Alu)

Cermets/HM3 No Yes (e.g. WC-CO) No Yes (e.g. Al2O3 – Cu)

Ceramics Yes (Phenix) Yes (e.g. SiC) Yes (e.g. ZrO2) Yes (e.g. SiC)

Other composites No Yes No Yes

(5)

examined without taking into account the scanned geometry.

That brings this section finally to the influence of process variables on problems in consolidation. First is the influence of laser type (see also §2.3). The higher the laser absorption into the powder bed, the less scope there is for variation in absorption, and the easier is the process to control. Pulsed laser beams and smaller beam diameters may lead to smaller, more stable melt pools, and in some cases pulsing may result in plasma formation and recoil forces adding a mechanical component to consolidation phenomena [97, 144, 157]. Smaller layer thickness reduces the need for over-heat if re-melting the previous layer is important. Thinner layers means finer powder sizes. These in turn are easier to melt due to their higher surface-to-volume ratio. Processing changes that lead to smaller melt pools also seem to lead to easier consolidation. It may be that a final problem in consolidation is that it cannot be scaled up to faster processing speeds.

3 DISCUSSION OF MATERIAL CLASSES

In this section the consolidation phenomena are discussed per material class:

x Polymers: i.e. substance with a high molecular weight

which is made of many repeating smaller chemical units or molecules.

x Metals: i.e. substance whose atoms are connected by

metal bonds.

x Ceramics: inorganic and non-metallic materials.

x Cermets and hard metals: i.e. composites in which

ceramic particles are bound in a metal matrix (binder).

x Composites: mixtures of different materials which

result in an inhomogeneous compound. Cermets and hard metals may be considered a special class of composites.

3.1 Polymers

Even though polymer is the most processed type of material in SLS/SLM, the consolidation phenomena invoked for polymers are probably still amongst the least understood or at least the least described in literature [81]. This explains why commercial applications of SLS today are limited to a small number of polymers: mainly polyamide (PA 12 and PA 11), and some polycarbonate (PC) [48], polystyrene (PS) and variants of those [175].

Figure 4: Thermoplastic polymers (Red = material used in SLS)

In literature there is no agreement on the nomenclature of the physical state of parts after laser powder processing. Laser consolidation of polymers normally involves melting

of thermoplastics (partial-SLS or full-SLM). A clear

distinction should be made between (Figure 4):

x (semi-)crystalline thermoplastics

x amorphous thermoplastics.

When heated up from very low to very high temperatures, all those thermoplastic materials will change from a hard (solid and glassy) structure to a softer (tough leathery or rubbery, solid or non-pourable) structure and finally turn into a viscous flowing melt. Materials however differ in the way and the temperatures at which those transitions occur. Semi-crystalline polymers have a glass transition temperature Tg that is below or around room temperature (-100° to 50°C) and a distinct melting temperature Tm that is above 100°C (between 100° and 400°C) at which a significant volume change happens. Amorphous polymers do not depict a clear melting temperature range. They have a glass transition temperature Tg that lies around 100°C and above which the material will gradually evolve to a leathery, rubbery and finally liquid state as temperature increases, without clear transitions. Figure 5 indicates the glass transition range ('Tg) and melting range ('Tm) for three important semi-crystalline polymers (PE, PP, PA 6) and indicates for three amorphous polymers (PS, PC, PMMA) the glass transition temperature ranges ('Tg) and the temperature range in which the transition to leathery material goes quickly ('Tf). This last range ('Tf) can physically not be defined exactly.

Figure 5 : Phases and transition temperatures of some polymers

Notice also that the Tg and Tm values largely depend on the molecular weight (MW) of the polymer. This explains why there might be a significant difference in SLS processability between a low and a high molecular weight PA or PE.

The temperature transitions and melting range of polymers can be observed by recording a DSC plot by Differential Scanning Calorimetry analysis. It measures the difference in the amount of heat required to increase the temperature of a sample and a reference as a function of temperature. Both the sample and reference are maintained at very nearly the same temperature throughout the experiment. Figure 6 gives the DSC plots of Duraform PA 12 and PA 6 [166]. The plots illustrate the much smaller melting range (small width of peak around 187°C) and the smaller melting temperature (Tm=187°C) of PA 12 as compared to PA 6 (Tm=223°C). This explains, besides viscosity differences (see §2.4 and §3.1.1), why PA 12 is more prone to SLS than PA 6 (or PA 66 having Tm= 262°C and a wider peak than PA 12) [212].

The laser consolidation of (semi-)crystalline polymer

powder will happen by heating above their melting

temperature Tm. Semi-crystalline materials have a highly ordered molecular structure with sharp melt points. They do not gradually soften with a temperature increase but rather remain hard until a given quantity of heat is absorbed and then rapidly change into a viscous liquid.

(6)

The molten polymer flows in between the powder particles, forming sintering necks. With enough heat, the complete layer is fully molten and overlaps to the previous layer. As the molten polymer cools down below Tm, polymer crystals nucleate and grow, recreating regions of ordered molecular chains (crystallites) mixed up with disordered amorphous regions. Above Tm, the polymer depicts a relatively low viscosity, as compared to amorphous polymers, which favours the rate and amount of consolidation. The densities obtained will be close to full density and the mechanical properties will be close to those of moulded polymers. However, the freezing of the polymer at Tm coincides with an important shrinkage (phenomenon not occurring with amorphous polymers), that may induce geometrical inaccuracies and distortion of the part (Figure 13). A good way to prevent this is to preheat the polymer powder to a temperature slightly below its melting temperature and keep it there for a certain time after consolidation [26,197, 212].

Figure 6 : DSC plot of PA 12 (Duraform) and PA 6 The consolidation of amorphous polymer powder

r semi-crystalline polymers we have rather

thermal behaviour of polymers is amplified occurs by laser heating above the glass transition temperature, at which the polymer is in a much more viscous state than semi-crystalline polymers at similar temperature. Amorphous polymers have a randomly ordered molecular structure. They do not have a sharp melt point but instead soften gradually as the temperature rises. The viscosity of these materials changes when heated, but they seldom are as easy flowing as semi-crystalline materials. The flow and sintering rate will be less, resulting in a lower degree of consolidation, higher porosity, less strength, but also a lower shrinkage that is favourable in cases of SLS of patterns for producing mould for moulding or casting (i.e. Indirect Rapid Tooling, investment casting patterns, etc.). In several applications post-infiltration of the pores is applied to consolidate the part (see § 4).

In conclusion fo

full melting consolidations whereas for amorphous polymers we rather see a partial melting binding mechanism.

This complex

in SLS processing due to several factors. The costly pulverization either by cryogenic milling or precipitation-process (Figure 7) limits the availability of polymer powders. The pulverization can influence also the

consolidation process: the powder layer’s thermal characteristic differs.

The material is in powder form with a certain particle size distribution (Figure 8) and the induced energy, heating the fine powder particles, may lead to evaporation or disintegration which in turn disturbs the process. It shields off the laser window and the IR heating elements. In short, this has a negative influence on the process consistency and the economy of the process. Further the fine particles may create either a dusty atmosphere or coagulations clustering preventing the homogeneous deposition of layers, thus preventing good layer consolidation.

a

b

Figure 7 : Powder particle shapes REM, a) precipitation, b) Cryogenic milling

A variety of commercially available technical polymers with excellent properties for SLS have been tuned by blending or by special design of the chain length (analogy to molecular weight ranges). Often they result in a multi-peak DSC diagram. SLS can handle only one fixed controlled consolidation temperature or several very close peaks (typically 5-10 °C difference), which means that polymers with more than one tight melting range are not processable with SLS. Figure 9 demonstrates the DSC curve of a commercial pulverized blend with no real full melting SLS processability, due to two different melting points at 178°C and 185°C.

We can associate seven main SLS polymer powder materials with different application fields. These material families can also be attributed to a specific consolidation mechanism as discussed in §2.

The four main thermoplastic SLS applications and the related thermoplastic material are:

a. SLS of polymer parts: semi-crystalline polymers (§3.1.1)

b. SLS of investment casting patterns: amorphous polymers (§3.1.2)

c. SLS of metal or ceramic parts using a sacrificial polymer binders: thermally degradable amorphous polymers (§3.1.3)

(7)

d. SLS of reinforced or filled semi-crystalline

eric material parts (§3.1.5)

in the racing mosetting materials (§3.1.7)

T

polymers for highly loaded parts (§3.1.4)

SLS today also allows production of parts using: e. Flexible elastom

f. Polymer-polymer blends (used industry) (§3.1.6)

g. Ther

hose different material categories are discussed in more detail below.

Figure 8 : Typical particle size distribution of Duraform PA 12 (Source FHSG St. Gallen)

Figure 9 : Commercial pulverized blend with no real ltcurve (right)

[151,

on is then obtained by keeping the powder bed

ay be raised to compensate for that, but only in a narrow range and with limited effectiveness (see also § 3.1.8).

a

SLS processability: (left) me recrystallisation curve 3.1.1 Semi-crystalline polymers

Today, the production of prototypes (RP) and functional parts (RM) by SLS is basically limited to nylon, i.e. polyamide (PA) [7, 22, 24, 26, 84, 102, 218]. Quite some research is going on using other (semi-)crystalline thermoplastics: polyethylene (PE) [155, 165], PEEK 152], PCL [148, 203, 214]. and PEEK are of special interest as biocompatible polymers [151, 192, 214].

The fundamental reason why PA sinters well, while more difficulties are experienced with other polymers is not totally clear yet. Some semi-crystalline polymers are able to crystallize much faster than others, as a result of their chain structure. The rate of crystallization is minimum near Tg and Tm, and reaches a maximum between those two temperatures. Crystallization rate should however be kept relatively slow (relative to the vertical build rate) in order to avoid part distortion due to freezing shrinkage. Good crystallizati

at high temperature for a sufficient long time after sintering.

To obtain fully dense parts, the melt viscosity should be low enough to allow complete consolidation within the time scale of the process. Pure polymers with melt viscosity in the range of few tens to few thousands poise (e.g. nylon and waxes) can be processed successfully and reach near full density [11]. However, even for PA-12, the SLS part still contains small amounts of porosity and the

process is something between partial and full melting. The densities obtained in SLS of PA 12 are just slightly below those of compression moulded parts (0.95-1.00 versus 1.04 g/cm3), yielding very similar mechanical properties under compression, but somewhat lower tensile strength and notched Izod values that are sensitive to small voids. The melt viscosity is linearly related to the molecular weight (MW). For good interfusion of the polymer chains, sufficient to provide particle necking and layer to layer adhesion, a low melt viscosity is desirable. However, low viscosity results in high shrinkage and poor part accuracy. Thus an optimum MW range exists but is not easily controllable, neither at the polymer production nor during aging in process. The thermal degradation over time when re-using the same powder repeatedly, results in the decay of the MFI (Mold Flow Index) hence a drop in powder flowability and a rise in melt viscosity preventing proper sintering with reasonable quality: the surface gets an unsmooth “orange peel” texture (Figure 10b) after a certain number of runs (Figure 10a). A drop in mechanical properties can be observed as well. The consolidation temperature m

b

Figure 10 : a) The rise in viscosity of re-used PA 12 material, b) resulting orange peel texture on SLS part To achieve functionally strong prototypes, all powder bed based processes need to work with polymers that have molecular weights as high as possible [155]. High molecular weight can lead to difficulty in placing and forming the material because of the high viscosities involved. Increasing the molecular weight of the polymer

oxidation resulting in de-polymerization). Typically powder can reduce shrinkage in the sintered part and thus improve the dimensional accuracy.

PA and other semi-crystalline polymer powders for SLS are normally supplied without additives. Some additives may be added to favour the powder fluidity and powder spreading, but those have no influence on viscosity and sinterability. Atmospheric control in the SLS chamber is very critical in terms of temperature (control of sintering and shrinkage), humidity (powder fluidity) and oxygen content (to avoid thermal degradation of the polymer by Cooling curve

Heating curve

Orange peel surface texture

(8)

polymers like PE and PP are more prone to rapid oxidation than aromatic backbone polymers like PET

e to the purity differences compared to the first powder.

a

(semi-crystalline) or PC (amorphous).

Several researchers investigated SLS of biocompatible

semi-crystalline thermoplastics like PEEK and PCL

(Poly-ε-caprolactone) [9, 31, 148, 152, 170, 192, 214]. PCL is a biodegradable linear polyester, while PEEK is not biodegradable. A polymer has never one specific molecular weight, but consists of a mixture of molecules with different chain lengths, and so, different molecular weights. The mean molecular weight and the molecular weight distribution are important properties of a polymer. Typically SLS of PCL with MW of 50.000 g/mol demonstrated good laser sinterability [214], while PCL with a MW of 40.000 g/mol could not be laser sintered well [203] due to big distortions caused by shrinkage. This shrinkage is due to the lower molecular weight of the second PCL powder, and du

b

Figure 11 : a) PA 12 Powder, b) tensile break cross section showing some air voids (Source FHSG-St.

Gallen)

break cross section oulded parts concerning

makes them well suited for investment casting patterns.

a

Powder particle morphology also influences the laser consolidation. Figure 11a shows powder (PA 12) with a near to spherical shape whereas Figure 12a shows irregular shaped PS powder. When no pressure is applied during consolidation, voids and air traps remain in the solid part. This reduces the effective cross section and the tensile strength compared to moulded parts from the same polymer (10-20% reduction). Figure 11b (PA 12) and Figure 12b (PS) depict the tensile

demonstrating partial melting regime. Comparison between SLS and m strength is discussed in §3.1.8 .

3.1.2 Amorphous polymers

Amorphous polymers with strongly temperature-dependent viscosities (or high activation energy for viscous flow) have been readily processed using SLS [11, 14, 75, 141]. Typical examples are polycarbonate (PC) and polystyrene (PS). Depending on the powder size and MW, they are generally pre-heated to a temperature near or even above the glass transition temperature Tg and will

melt as the temperature further rises above Tg. They mainly have the advantage that their volumetric shrinkage, when cooling down, does not show a jump, as it does for semi-crystalline polymers: see Figure 13. This fact provides high accuracy with little distortion. However, amorphous polymers show a lower level of consolidation (higher residual porosity) and hence do not reach the strength of their moulded equivalent. The higher degree of porosity, low thermal shrinkage/expansion and high accuracy

b

gure 12 : a) PS Powder (CastForm), b) Tensile brea

Fi k

cross section demonstrating partial melting regime us polymers for

3.1.3 Thermally degradable amorpho

SLS of metal or ceramic parts

Various SLS applications make use of a powder combining a degradable polymer binder and a structural material (metal or ceramic) that produces a “green part” in which the structural particles are bound into a degradable polymer matrix [77]. These ‘green parts’ are further processed in a furnace to burn out the polymer (“de-binding”) and to consolidate the remaining polymer-free metal or ceramic part by post-sintering, infiltration, or HIPing: see § 4. The polymers used as the sacrificial binder in those applications are different from the polymers described in the previous and subsequent sections. They should allow de-binding, i.e. thermal depolymerization. Polymers used for this purpose are PMMA and copolymers MMA-BMA that can be de-binded in a furnace at temperatures around 350-450°C. They should not be water soluble [11, p. 112] and should have sufficient strength, even when heated up in the de-binding furnace, in order to avoid that the ‘green part’ would collapse before some solid state sintering can consolidate the metallic or ceramic particles together. The polymer must also be fully de-bindable (i.e. depolymerised and sublimated) and should not yield excessive contamination of the remaining part with carbon residuals (important e.g. for steel parts). If the strength of the polymer in the green stage is insufficient, the green part can be consolidated by cross-linking the binder. This can be done by impregnating the green part with a water soluble thermosetting acrylic emulsion and drying it in an oven at 50°C, before it goes to the de-binding furnace [11, p. 139]. Alternatively, a cross-linker can be added into the degradable polymer

(9)

coating of the powder [11, p. 139]. This will eliminate the step of impregnating the green part.

When using sacrificial binders, care should be taken because depolymerization could also occur during SLS [11 p. 108, 82, 88]. Indeed, these polymers are deliberately designed to depolymerise as an aid to their thermal removal in post-SLS processing. Polymers based on copolymers of n-butyl methacrylate (BMA) and methyl methacrylate (MMA) have been observed to also depolymerise during SLS processing [200, 201].

Figure 13 : Comparison of relative volume between amorphous and semicrystalline polymers [11]

a

b

c

Figure 14 : SLS of AW glass ceramic with sacrificial binder (a) Powder mixture (b) Green part (c) Brown part The polymer binder may be added to the structural material in two ways: mixing [34, 123] or coating [171, 172, 199, 217]. Typical volume of binder in the powder is 5% for mixed powders [34] as well as coated ones [11,

rates three stages in the SLS

e just enough to give the

still is not totally dense and may need further densification to convert the brown part into a final one. p136]. The latter typically corresponds to a 5 μm polymer coating on a 55 μm 1080 carbon steel grain [200].

Figure 14 [108] demonst

process of Apatite-Wollastonite (A-W) glass ceramic with 5 wt% MMA-BMA binder:

a) powder mixture before processing (white particles are A-W glass ceramic, grey particles are MMA-BMA), b) highly porous part after SLS processing (The few

polymer connections ar

'green' part sufficient strength to be transferred to the post-treatment furnace)

c) 'brown' part after polymer debinding and 1 hour firing at 1150°C. During firing, A-W is partly melted. The part

Figure 15 : Glass-nylon powder mixture (Source: K.U.Leuven)

Figure 16 : SEM micrograph of HA particle distribution in HDPE matrix [165]

Figure 17: PA-12 GF One of the most popular glass bea reinforced HTD materials (Source FHSG St. Gallen)ds

binders described in 3.1.4 Semi-crystalline reinforced polymer for metal,

ceramic or glass reinforced plastic parts

Today there exist several SLS powders aimed at production of reinforced polymer parts. Applications include parts made from glass reinforced PA [28, 30], Cu filled PA [20], Al filled PA [1, 46], SiC-PA [59, 76], HA-HDPE, i.e. hydroxyapatite reinforced high density polyethylene [165], HA-PA [69, 166], Polysiloxane-SiC [54]. Unlike the sacrificial polymer

(10)

§3.1.3, the polymers used here should withstand thermal degradation and should be durable.

The initial powder may be a mixture of polymer particles and reinforcement beads. Figure 15 shows such a mixture of spherical PA and glass particles: on the right side of the picture, the glass particles are colored blue with Prussic ink for better visibility. Alternatively, the single powder particles may already be a composite consisting of a polymer matrix (PA or PE) containing filler particles [165,

166] (see particles

[20].

Figure 18 : New PA powder with a high-aspect ratio filler

a

Figure 16) or a polymer coated filler

(Source FGHS St.-Gallen)

b

Figure 19 : a) PA 12 and 30% Al blend; b) a tensile break

turned out to be better

heat concentration on the filler has a negative influence on

only (for PA 6-8 times depending on judgment

s, strength

ls for Shore A hardness and elongation at break.

a

cross section with defects due to inhomogeneity (Source FHSG St. Gallen)

Figure 17 shows the fracture surface of a part made with a popular commercial glass filled nylon powder. One can see that the connection between the glass beads and PA binder was not perfect, that many glass beads have been pulled out and that the PA was stretched during breakage, resulting in a rough spongy surface. Figure 18 shows a new reinforced SLS powder having elongated filler beads (patent pending). This new powder

that the traditional GF-PA powder: tensile strength x 1.8; E-modulus x 1.3; elongation x 3.3.

Aluminium filled PA 12 often shows coagulation, since mixing the aluminium and PA powder particles isn’t always very succesfull. This causes segregation during layering due to the differences in size and specific weight (Figure 19). This fact is even more striking in PA 12-Cu. The inhomogeneous solid with voids creates mechanical defects and deteriorates the tensile strength. The local

the binder polymer causing a very short recycle ability of 2-3 times

criteria).

3.1.5 Elastomeric materials

Elastomeric polymers have a structure with long chains and only few cross-links between them. Below the glass-transition temperature they are brittle, above they are very elastic. When the density of cross-links enlarge

and stiffness also enlarge, and ductility lowers.

A new elastomer powder material for the SLS process was recently released [120]. Figure 20 demonstrates this polyester based elastomer (patent pending). The most significant properties in the thermoplastic material selection were considered and compared with reference to elastomeric materials as Neoprene, EPDM and natural rubber. Figure 21 gives a comparison for the different elastomeric materia

b

gure 20 : Polyester based elastomer a) Green par r sintering at low hardness b) Infiltrated w

Fi t

afte ith

polyurethane (Source FHSG St. Gallen)

relation to commonly HSG

St. Gallen)

Figure 21 : Developed elastomeric material for SLS in used elastomers (Source F

0 25 50 75 100 0 1 2 3 4 5 6 7 8 Material Sh or e A Ha rd ne ss 0 250 500 750 1000 El o n ga ti on %

(11)

Sintaflex not infiltrated 15W 13W 11W 5W 9W 7W 0 20 40 60 80 100 100 150 200 250 300 350 Elongation (%) Sh o re A H ard n es s Sintaflex infiltrated 5W 7W 9W 11W 13W 15W 0 20 40 60 80 100 100 150 200 250 300 350 Elongation (%) S h o re A Har d n e s s

Figure 22 : Hardness Shore A vs. Elongation [%] for infiltrated and not infiltrated parts processed with different laser powers (Source FHSG St. Gallen)

a

Sintaflex: E module vs. LS Laser Power

0 5 10 15 0 5 10 15 20 Laser Power [W] E M o d u lu s [ M Pa ] not infiltrated infiltrated b

Sintaflex: Rupture at break Rb vs. E Module

0 2 4 6 0 2 4 6 8 10 12 E Modulus [MPa] Rb [ M p a ] Rb Rb (inf.)

Figure 23 : (a) The elastic modulus [MPa] against laser power for infiltrated and not infiltrated parts (b) The tensile strength Rb [MPa] at ruptures against

elastic modulus (Source FHSG St. Gallen) The influence of laser processing parameters on Shore A hardness and ductility is shown by Figure 22. Clearly the parts processed with higher laser power become more ductile and have a higher Shore A hardness.

The influence of laser processing parameters on elastic modulus E and the tensile strength Rb at rupture is shown by Figure 23.

3.1.6 Polymer blends

Polymeric blends offer an alternative mean to obtain SLS parts with specific structure and properties, permitting the development of new applications [174]. Most polymeric

blends are multiphase systems and, therefore, their properties largely depend on their microstructure [162, 215]. Salmoria, for instance, used blends of PA and HDPE (blend ratio’s of 80/20, 50/50 and 20/80 wt%) to achieve dedicated properties. Depending on the blend ratio, different phases and micro-structures were observed using SEM, EDX and XRD analysis [162].

FH St.Gallen experimented with powder blends (PA12 with additions of PA11 or Polyester) mainly to enhance the elongation at break. Recently, a company from Austin started commercializing a mixture of PA12 and PA11 that demonstrates two DSC melt peaks at 185°C and 187°C respectively. The process is set to only melt the lowest peak. This clearly suggests a liquid phase sintering consolidation process.

In a US patent [40] the following is claimed:

A particle for use in selective laser sintering (SLS), including a core (1) formed from at least one first material, and at least partial coating (2) of the core (1) with a second material (further components are optional), the second material having a lower softening point than the first material, wherein the softening point of the second material is lower than approximately 70°C.

The coating (2) generally contains a polymer, preferably a thermoplastic polymer, e.g. a polyvinyl acetal, preferably a polyvinyl butyral. It may consist of alloys with a low softening point which are used e.g. in fuses. Moreover saturated linear carboxylic acids with a chain length of ≥16 (e.g. heptadecanoic acid, melting point 60-63°C.) or polymers in the broadest sense may also be suitable. The softening point of the 2nd material of approximately 70°C or below, allows laser sintering to be carried out at significantly lower temperatures compared to particles which have been used hitherto, and therefore also allows a significantly lower temperature difference between irradiated particles and standard room temperature. Tests have shown that the lower maximum temperature difference also improves the temperature homogeneity of the building space as a whole.

3.1.7 Thermosetting materials

Thermosetting polymers can be used in different steps of the SLS process. These materials can be used as an infiltrant [47], because it is a relatively inexpensive path to a fully dense, stiff, net shape, polymer matrix composite part. In addition this is a very useful intermediate step for fully functional materials, as critical surfaces and tolerances may be achieved easily at this stage. An example of infiltration with thermosetting polymers is the production of metal-epoxy molds via SLS indirect processing [11, p143]. In this process, SLS is used to form green mold cavity inserts from metal powder that is coated with fusible thermoplastic binder. In subsequent steps, the binder is thermally removed and the metal powder is oxidized to form a porous metal/ceramic cavity that shows little shrinkage and generally excellent retention of geometry, relative to the green part. The cavity is then strengthened and sealed by infiltration and cure of an epoxy tooling resin.

It is also possible to produce metal parts by laser sintering a mixture of metals with thermosetting polymers [123]. When a thermosetting material is exposed by a laser source, the thermosetting material turns into a viscous liquid instantly. With for instance SLS of epoxy resin mixture with iron powder, the polar groups (e.g. epoxy group) in the molecule of the resin are activated simultaneously. The liquid flows penetrate the ‘pipes’ made from pores and wet metal particles. As such, they make bridges from one particle to another. The binding effect, depicted in Figure 24, is dominated by the interfacial characteristic between the resin and the iron.

(12)

Iron surfaces commonly attach some active hydrogen atoms because of its attraction of some molecule such as H2O and HCl due to its high polarity. Hydrogen bonds occur between the electronegative oxygen atoms in polar groups in resin molecules and the electropositive active hydrogen ones on iron surfaces. Iron particles are strongly bonded because hydrogen bond attraction is more intensive than that of inter-molecular action existing on the interfaces between iron surfaces and other nonpolar polymers. The resin viscosity is lowered at higher temperature (i.e., laser energy) and its viscous liquid can spread easily. Thus many more iron particles surfaces can be attached by resin. However, degradation of resin, induced by excessive laser energy, might occur and reduce the bonding capability.

Figure 24 : Binding mechanism for epoxy resin to iron particles Injection vs. SLS Materials 0 4000 8000 12000 16000 0,0 100,0 200,0 300,0 400,0 Elongation [%] T e ns ile mo dulu s [ M P a ] Injection SLS c 2 3 4 5 6 8 9 a b d e f 1 7 0 50 100 150 200 250 0,0 50,0 100,0 150,0 200,0 250,0 300,0 350,0 400,0 Elongation [%] T e n si le st re n g th [ M P a] 1 Injection SLS 2 3 4 10 6 7 8 9 a c d e f

Figure 25 : Commercial available SLS materials (yellow) in comparison to some injection polymer materials (in blue) 3.1.8 Properties of molded and SLS polymers

Although great efforts have been made and some results were achieved, the reasons for the relative few material options are many. Some are surely of interest to this paper about consolidation, however when dealing with polymers other phenomena have to be considered as to enhance the binding models.

By the end of the day we want to have nearly the choice as e.g. in injection moulding. The obvious way is to compare the differences and compensate for deficits. Figure 25 shows all the commercial available SLS materials (in yellow) in comparison to some injection polymer materials (in blue): the comparison is shown for tensile strength, tensile modulus and elongation. We find some deficits. While some SLS materials have comparable properties to molded ones, none of them is able to reach the highest values of tensile modulus and strength, nor elongation. The low elongation capability can be explained by the many short binding necks creating high stiffness with little elongation.

A further interesting finding indicated by Zarringhalam [218] is the boundary between the loose powder and the solid part (see Figure 28, which shows the microstructure of a part cross section). Under certain conditions unconsolidated particles may remain in the solid part. This can be overcome easily by proper energy intake settings, correct scan strategies like cross scanning (in X, Y, in X&Y and so on) or multi times scanning of the same layers if required. The boundary conditions may be partially improved by the so called out-line scanning imposing a sharper border line between loose powder and consolidated part.

A powder with slow recrystallization rate, as depicted by non-overlapping or slightly overlapping endothermic and exothermic peaks during the heating and respective cooling phases of the DSC analysis (Figure 26), might result in nearly fully dense parts with minimal distortion, while highly overlapping peaks will yield bad results (Figure 27) [212].

Figure 26 : DSC plots (heating and cooling) of new reinforced PA 12 powder containing elongated rather

than spherical beads for better strength

Figure 27 : DSC plots (heating and cooling) of IP60 powder showing overlap between melting and

recrystallization peaks [US patent 5,648,450] In contrast to metals, one has to be aware about the thermal deterioration of the powder material. The recyled material has a limited usability as demonstrated for PA 12. The long exposure to heat leads to chain growth, a rise in

(13)

MW and so a rise in viscosity. This causes nonconstant consolidation conditions and a shift in the melting temperature. The empirical melting temperature ramps up as a function of build height and age of the powder. In the extreme this can lead to creation of patterns on the surface known as « Orange Peel » because of its texture (see Figure 10b). This can be improved by mixing with virgin material (about 70% virgin – 30% reused).

Melt Flow Index (MFI) is flow in grammes that occurs in 10 minutes through a standard die of 2.095 ± 0.005 mm diameter and 8.000 ± 0.025mm in length when a fixed pressure is applied to the melt via a piston and a load of total mass of 2.16 kg at a temperature of 190°C (some polymers are measured at a higher temperature, some use different weights and some even different orifice sizes). MFI is an assessment of average molecular mass and is an inverse measure of the melt viscosity; in other words, the higher a MFI, the more polymer flows under given test conditions. Hence the MFI of a polymer is vital to anticipating and controlling its processing. Generally, higher MFI polymers are used in injection moulding, and lower MFI polymers are used with blow moulding or extrusion processes.

When polymer powder is reused in SLS, the MFI is already about three times lower than the virgin powder (Figure 29). This leads to higher viscosity, and sometimes the quality of the product becomes unacceptable.

Figure 28 : Boundary between loose powder and solid part -20 0 20 40 60 80 1 6 11 16 21 26 31 36 41 46 51 56 61 66 Build (run) MF I QA MFI Linear (MFI) Virgin Interior quality Recycled

Figure 29 : Melt flow index of powder as a function of number of builds in which powder is used (Source

FHSG St. Gallen)

Figure 30 : Tensile strength and elongation of PA-GF polymers as function of number of build in which

non-refreshed powder is used

With PA 12, process stability and repeatability in the consolidation process can be obtained by keeping the MFI within a constant range and never working with virgin powder [119]. This phenomenon is amplified for the glass filled PA 12 material, leading to a rapid decline in mechanical properties, a clear evidence of different consolidation conditions (Figure 30).

The aging and the resulting change in consolidation phenomena and final component properties are very minor with amorphous materials like PS and elastomeric polyester based material.

To conclude on polymers, Table 2 gives an overview of some commercial available SLS polymers with their most important properties concerning the SLS process.

Material Tg (Glass transition temperature) (°C) Tm (Melt temperature) (°C) MW (molecular weight) (g/mol) PA 6 50 230 PA 12 (41) 184 9.600 PCL -65 … -60 58 … 60 40.000 … 80.000 PEEK 143 340

Table 2 : Overview of commercial available polymers with most important properties for SLS

3.2 Metals

Today metals are the second most used material type in SLS/SLM. Almost all consolidation mechanisms listed in Figure 2 can be applied to consolidate metallic parts [164]. 3.2.1 Solid state sintering (SSS)

Theoretically SSS can be applied for consolidating metal parts by SLS [204]. Its application is however limited for two reasons:

x The process is difficult to control: the process transits between no consolidation (at lower energy levels) and partial/full melting (at higher energy levels) within a very narrow process window (range), making it difficult to adjust the process parameters for real SSS.

x SSS requires a long interaction time between laser beam and powder particles. This calls for slow scanning velocities and makes the process not Un-molten particle core Un-molten complete particle stuck to edge Fully molten particle (no core)

(14)

economically viable. Tests performed at K.U.Leuven with steel powders indicated that scanning speed in solid state SLS needs to be limited [204], where those scanning speed may reach about 1500 mm/s in SLM and about 10000 mm/s in liquid phase SLS with Laserform [103].

Early SSS laser sintering tests with steel were performed by Van der Schueren [95, 204]. Gusarov has applied and modelled laser SSS of Ti powder [66, 67]. The laser interaction time was of the order of magnitude of 5 seconds, as compared to 0.1- 0.3 ms with other binding mechanisms. Tolochko sintered Ti teeth with a static defocused Nd:YAG laser beam [194]. The Ti parts obtained were dense in the core, i.e. in the laser spot (high energy density yielding SLM), but had a porous shell (lower energy yielding SSS). No post-infiltration was aplied to remove this porosity, as it favoured osteo-integration of bone into the Ti tooth.

Lanzetta has experimented with gold powders [111]. He succeeded to thermally sinter pure Au spherical powder ranging between 5 and 40 μm in size and then to infiltrate the obtained skeleton with binary gold eutectics based on silicon, germanium and tin, with a melting point as low as 278° C. The initial thermal sintering is not yet laser-based, but this is the logical next step. This laser sintering could be done by solid state sintering or by liquid phase sintering (partial melting) as described further on (see §3.2.2, subsection B1).

3.2.2 Liquid phase sintering and partial melting

This category unites many different kinds of technologies. Most of these techniques combine a structural material remaining solid throughout the process and a binder material being liquefied. In some cases, however, the solid and the liquid phases result from the same material. The

first group of technologies is characterized by a clear distinction between the binder and structural materials.

A. Different binder and structural materials

Technologies in this category can be further divided into three groups according to the type of powder grains that are used (Figure 2).

A1. Separate particles

These technologies use different binder and structural particles [4]. The structural material (in this case, a metal, alternatively a ceramic as discussed below) should generally have a higher melting point than the (metallic) binder material (Use of polymer binders has been discussed in §3.1). The binder particles are usually smaller than the structural ones, in order to facilitate their preferential melting. However, preferential binder melting may be counteracted by the higher reflectivity or lower laser absorption of the metallic binder material (typically Cu or Co) as compared to the structural material (metal or ceramic). In some cases, this might even lead to a reverse melting process, in which the structural particles will melt prior to the binder grains [105, 204].

The combination of small binder particles and larger structural particles has the additional benefit of better packing with small pores, favoring fast spreading of the molten binder by capillary forces and fast rearrangement of the particles.

Generally a ‘green’ part is produced which is still porous and brittle. Therefore, a post treatment consisting of a furnace post-sintering, Hot Isostatic Pressing (HIP) or an infiltration with a low melting point material (metal, alternatively epoxy or other) is usually necessary: see §4. To obtain sufficient mechanical properties, most commonly infiltration is applied because of its effectiveness and relative ease.

a) WC and Co powder mixture b) mech. alloyed WC-Co powder

c) sintered powder mixture d) sintered mech. alloyed WC-Co powder Figure 33 : WC-Co powder mixture and sintered parts

(source: University of Leuven)

Many different material combinations have been tested in the past. At the University of Leuven, metal-metal composites were tested as well as metal-ceramic composites [64, 100, 101, 109, 112]. Some examples of metal-metal composites are Fe-Cu and Stainless Steel-Cu. As for the metal-ceramics combinations (see also §3.3), WC-Cu, WC-Co, WC-CuFeCo, TiC-Ni/Co/Mo, ZrB2 -Cu and TiB2-Ni were tested. Figure 31 and Figure 32 show cross sections of Stainless Steel-Cu and WC-Co green parts (see also Figure 33 a and c).

A2. Composite particles

Composite powder particles contain both the binder and the structural material within each individual powder grain. The powder may be obtained by mechanically alloying a mixture of two different powders, causing powder particles to be repeatedly milled, fractured and welded together. Figure 31 : LPS of Stainless Steel-Cu powder mixture

(a: nonmolten steel particle, b: molten Cu, c: porosity)]

100 mP

Figure 32 : LPS of WC-Co powder mixture; top: before infiltration (a: nonmolten WC particle, b: molten Co, c:

(15)

Figure 33 a shows an initial powder mixture of WC and Co particles. Mechanical alloying results in WC-Co powder particles with a micro grain composite structure in which the two different phases (WC and Co) can still be identified (see Figure 33 b). Such composite particles yield a higher SLS green density and a better surface roughness than a mixture of separate WC and Co powders (compare Figure 33 c and d) [112]. Figure 34 shows an example of a WC-Co injection-moulding insert.

Figure 34 : Bronze infiltrated WC-Co injection moulding insert (source: University of Leuven)

A3. Coated grains

A third possibility to combine a binder and a structural material is to coat the structural material with the binder phase. This ensures that the laser radiation hitting the powder particles is preferentially absorbed by the binder material that is to be melted. Moreover a more effective bonding of the structural particles is realized since the binder material already surrounds all structural particles. Coated powders exist both with metal and polymer binder coatings.

The University of Leuven experimented with Cu coated steel powder which turned out to be successful despite the fact that the powder coating process was hard to control [204].

Figure 35 shows a cross section of an SLS part made from commercially available polymer coated stainless steel powder (LaserForm), after polymer debinding and infiltration with bronze.

Figure 35 : Bronze infiltrated Laserform ST 100 part At the University of Queensland, an aluminium LPS process has been developed in collaboration with 3D Systems [171, 172, 173]. Since the strong oxide layer on the grains prohibits direct sintering of Al powder, an indirect procedure was developed. Nylon coated Al particles are fused together using the LPS mechanism. Next, the nylon binder is de-binded in a furnace with N2 atmosphere. A small amount of magnesium added to the powder mixture makes it possible to form aluminium nitride instead of aluminium oxide when heating up to 540 ºC. The aluminium nitride acts as a rigid skeleton, which is necessary during the infiltration phase. The infiltrant being used is a ternary eutectic aluminium alloy (Al-13.8Si-4.7Mg), having a lower melting point than the basic aluminum alloy. Relative part densities up to 95 % were obtained. Figure 36 shows the different processing steps.

B. No distinct binder and structural material

Technologies in this group do not exhibit a clear distinction between binder and structural phases (even in cases where different powders are mixed). Rather than the distinction between binder and structural material, there is a distinction between molten and nonmolten

material areas. Therefore ‘Partial Melting’ is a better name for these technologies than ‘Liquid Phase Sintering’. B1. Single phase, partially molten

When the heat supplied is insufficient to completely melt powder particles, only a shell or small particles will melt. This way the molten material acts as a binder between the nonmolten particle cores [184, 195].

The partial melting phenomenon was modeled by Karapatis [51, 52, 53, 82]. Using a simple thermal model, skin and core temperatures of the powder particles were calculated. This way the minimal pulse energy to fully melt the particle can be calculated. Below this value, the core temperature never exceeds the melting temperature and only partial melting is obtained.

B2. Fusing a powder mixture

Powders consisting of multiple kinds of powder particles can be classified in this group when they are only partially molten [64, 220]. For example at the University of Leuven experiments were done with a Fe-Fe3P-Ni-Cu powder mixture, aiming at the production of full density parts [93]. The addition of a melting point lowering additive like Fe3P is favorable in making the process more energy efficient. For instance, alloying pure Fe with a small amount of P lowers the melting point of pure Fe (1538°C) to the eutectic temperature of the Fe-P system (1048°C, see Figure 38). Moreover, the dissolution of P in Fe also has the benefit of a lower surface tension of the melt, resulting in a better wetting behavior. Ni is added for its marked strengthening effect. A micrograph of the structure obtained after partial melting of this powder is given in Figure 37. A closer look at the microstructure reveals that not all the powder particles are molten. The final part consists of a low melting point P-rich phase (no. 3), a high melting point phase with no significant amount of P (no. 2), some remaining porosities (no. 4) and some remaining unmolten Fe powder particles (no. 1). Therefore, the process cannot be called ‘full melting’ and the name ‘partial melting’ is preferred. Notice that, even though no full melting occurs, almost full density is achieved.

50 mP 50 mP

20 mP

a b

c

Figure 36 : Steps in Aluminium SLS: a) green part (nylon binder in black, not visible) b) aluminium nitride skeleton surrounding the aluminium grains, c) infiltrated

part

Figure 37 : Micrograph of multiphase steel powder: 1) unmolten Fe particle, 2) high melting P-poor phase, 3)

References

Related documents

• Contract your glutes and hamstrings and slowly lower yourself down until your body is flat over the ball. • Contract your glutes and hamstrings to raise back up…and use your

The Ardennes Forest Cup your hands to scoop up sleep as you would draw a grain of water and the forest will come: a green cloud a birch trunk like a chord of light and a

variables mediate the effects of human capital, time in the United States, and religious minority. status on naturalization, Figure 4 shows a path model including just these measures

The advantage of the flat rate pricing scheme is its simplicity. However, under such a scheme, the provider will not have incentives to provide the service if the

Scientific Study This is most otien characterized by the artist's attempting to include every aspect of a plant and its structure all on one page. There may

Containers may bear the statement: “This material meets the requirements of ANSI/AWWA B605, Stan- dard for Reactivated Granular Activated Carbon,” provided the requirements of

In this paper we have developed representer theorems for convex regular- ized inverse problems (1), based on fundamental properties of the geome- try of convex sets: the solution