POLYAMIDE (NYLON) 12 POWDER DEGRADATION DURING THE SELECTIVE LASER SINTERING PROCESS A Quantification for Recycling Optimization

By

LUKAS JL DUDDLESTON A thesis submitted in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE (MECHANICAL ENGINEERING) at the

UNIVERSITY OF WISCONSIN – MADISON 2015

Final Oral Examination: August 4th, 2015

Committee: Tim A Osswald, Professor, Mechanical Engineering Lih-Sheng Turng, Professor, Mechanical Engineering Natalie Rudolph, Assistant Professor, Mechanical Engineering

APPROVAL PAGE The following thesis, Polyamide (Nylon) 12 Powder Degradation during the Selective Laser Sintering Process: A Quantification for Recycling Optimization, has been approved by:

Professor Tim Andreas Osswald KK & CF Wang Professor Department of Mechanical Engineering College of Engineering University of Wisconsin-Madison

Date

© Copyright by Lukas JL Duddleston, 2015 All Rights Reserved

i For my best friend, Ashtyn.

ii

ACKNOWLEDGEMENTS Before all else, I must thank Professor Tim Osswald for giving a chemist the opportunity to join the Polymer Engineering Center and for all of the support over the next sixteen months. I must also thank my loving mother, Priska, and Diane Osswald for their serendipitous meeting at the Luzern train station, which initiated this journey. To my father, William, thank you for wholeheartedly supporting my all of academic endeavors, past, present and future.

A thank you must be given to my committee members as well. Professor Lih-Sheng Turng, thank you for giving me access to your lab and your instrumentation in the Wisconsin Institutes for Discovery building. Professor Natalie Rudolph, thank you for joining the PEC last fall and for always having a minute to talk and then discussing ideas with me for the next hour.

Andrew Puck, Alex Harris and Elio de Stephanis, you three have been superb undergraduate assistants, without all of the time you three spent in the lab running countless experiments, I would still be toiling over experiments. Neil Doll, your wealth of knowledge regarding the SLS process was invaluable. Patrick Mabry, thank you for the great tête-à-têtes during our epic ping-pong matches. Lastly, Dr. David C Mazur and Christopher T Ruska, cheers to spring break 2014.

Finally, to my dearest Maddie Kay, the hours spent at playgrounds provided me with an escape from the real world and helped me stay young at heart.

iii

What is the most resilient, a parasite? Bacteria? A virus? An intestinal worm? An idea. Resilient... highly contagious. Once an idea has taken hold of the brain it's almost impossible to eradicate. An idea that is fully formed - fully understood - that sticks; right in there somewhere. - Cobb

iv

ABSTRACT Selective Laser Sintering, a 3-dimensional printing technique, converts powdered thermoplastic resins, e.g. polyamide 12 (nylon), into end-use parts using a laser to melt and fuse the particles. In this layer-by-layer additive manufacturing process the powder is both the raw material and the mold material. Therefore unsintered powder can be recovered and recycled in subsequent builds to significantly decrease net costs. However, unsintered powder is thermally degraded, which results in inferior parts unless blended (refreshed) with virgin powder. To improve blending protocols, the powder quality was quantified using differential scanning calorimetry (DSC), thermogravimetric analysis, and the melt flow index (MFI) for molecular degradation and using flowability measurements to measure changes in bulk properties. The results suggested that the sensitivity of DSC to small changes in molecular weight could reproducibility measure small changes in artificially aged (degraded) powder. Angle of repose, a flowability measurement, and MFI were sensitive to bulk and molecular degradation, respectively; however, both techniques lacked reproducibility. In conclusion, DSC could be a powerful tool to help optimize recycling of SLS powder.

v

TABLE OF CONTENTS 1

INTRODUCTION ................................................................................................ 1

2

THEORETICAL BACKGROUND ....................................................................... 4 2.1. POLYMERS ........................................................................................................ 4 2.2. POLYAMIDE 12 SYNTHESIS ................................................................................. 8 2.3. POLYMER POWDERS ........................................................................................ 11 2.3.1. Powder Flowability and Packing Efficiency ............................................. 13 2.4. MELTING OF SEMI-CRYSTALLINE POLYMERS ....................................................... 13 2.5. ADDITIVE MANUFACTURING ............................................................................... 16 2.5.1. Selective Laser Sintering ........................................................................ 20 2.5.1.1. SLS Machine .................................................................................... 21 2.5.1.2. Pre-Build .......................................................................................... 23 2.5.1.3. The Build and Consolidation ............................................................ 24 2.5.1.4. Post-Build ......................................................................................... 25 2.5.1.5. Advantages ...................................................................................... 27 2.5.1.6. Disadvantages ................................................................................. 28 2.6. AGING AND DEGRADATION ................................................................................ 28 2.6.1. Thermal................................................................................................... 31 2.6.2. Thermal-Oxidative .................................................................................. 31 2.7. QUANTIFICATION TECHNIQUES .......................................................................... 32 2.7.1. Differential Scanning Calorimetry ........................................................... 32 2.7.2. Thermogravimetric Analysis.................................................................... 36

vi 2.7.3. Powder Flowability and Packing Efficiency ............................................. 36 3

LITERATURE REVIEW .................................................................................... 38 3.1. POWDER FLOWABILITY ..................................................................................... 38 3.2. DEGRADATION PATHWAYS ................................................................................ 39 3.3. THE EFFECTS OF THERMAL DEGRADATION ......................................................... 42 3.3.1. Molecular Weight .................................................................................... 42 3.3.2. Surface Finish Issues.............................................................................. 43 3.3.3. Inferior Mechanical Properties ................................................................ 44 3.4. RECYCLING LITERATURE................................................................................... 47 3.4.1. Choren et al - 2001 ................................................................................. 47 3.4.2. Gornet et al - 2002 .................................................................................. 48 3.4.3. Dotchev, Pham and Yussoff – 2008 to 2009........................................... 49 3.4.4. Mielicki – 2014 ........................................................................................ 52 3.4.5. Wudy et al – Present............................................................................... 52 3.4.6. Summary ................................................................................................ 53

4

PATENT REVIEW ............................................................................................ 54 4.1. DIAMINE- AND DIACID-REGULATED PA12 ........................................................... 54 4.2. HYDROLYSIS .................................................................................................... 56

5

COST-ANALYSIS ............................................................................................ 58 5.1. COST VERSUS INJECTION MOLDING ................................................................... 58 5.2. REFRESH RATE THOUGHT EXPERIMENT ............................................................. 60

vii 6

METHODOLOGY ............................................................................................. 64 6.1. INTRODUCTION................................................................................................. 64 6.2. MATERIALS ...................................................................................................... 65 6.3. CONTROLLED AGING ........................................................................................ 65 6.4. RECYCLING SCHEME – BLENDING ..................................................................... 66 6.5. THERMAL ANALYSIS ......................................................................................... 67 6.5.1. Differential Scanning Calorimetry ........................................................... 67 6.5.1.1. Degradation Analysis ....................................................................... 67 6.5.1.2. Recycling Scheme – Blending .......................................................... 68 6.5.2. Thermogravimetric Analysis.................................................................... 68 6.6. MELT FLOW INDEX (MFI) .................................................................................. 68 6.7. POWDER FLOW ANALYSIS ................................................................................. 69 6.7.1. Pourability ............................................................................................... 69 6.7.2. Apparent Density .................................................................................... 70 6.7.3. Angle of Repose ..................................................................................... 71

7

RESULTS ......................................................................................................... 72 7.1. THERMAL ANALYSIS ......................................................................................... 72 7.1.1. DSC – Aged Powder ............................................................................... 72 7.1.2. TGA – Aged Powder ............................................................................... 77 7.1.3. Particle Size Dependent Degradation ..................................................... 79 7.2. RECYCLING ..................................................................................................... 80 7.2.1. Recycling Scheme – Blending ................................................................ 80

viii 7.3. MELT FLOW INDEX ........................................................................................... 84 7.4. POWDER FLOWABILITY ..................................................................................... 87 8

CONCLUSIONS AND FUTURE WORK ........................................................... 91 8.1. CONCLUSIONS ................................................................................................. 91 8.2. FUTURE WORK ................................................................................................ 93

9

REFERENCES ................................................................................................. 95

ix

LIST OF FIGURES Figure 2.1: Synthesis of a polymer, polyamide-12, with 2 function end groups from a monomer. ................................................................................................................... 4 Figure 2.2: The effect of molecular weight on the glass transition temperature, the melting temperature and the degradation temperature. ............................................. 5 Figure 2.3: A hypothetical distribution of polymer molecular weight........................... 7 Figure 2.4: ROM Polymerization of PA12 from Lauryl Lactam ................................... 9 Figure 2.5: A schematic of particle coalescence ...................................................... 11 Figure 2.6: Crystalline Structures, Courtesty of Osswald48. ..................................... 14 Figure 2.7: A representative FDM printer (Top) A close-up of the print head and the part (Bottom) ............................................................................................................ 17 Figure 2.8: Two FDM printed parts with build orientation depedent support material ................................................................................................................................. 18 Figure 2.9: SLA printed parts ................................................................................... 19 Figure 2.10: A schematic of the Sinterstation 2500.................................................. 22 Figure 2.11: The general AM build sequence from CAD to end-use part................. 23 Figure 2.12: The SLS Build Cycle ............................................................................ 26 Figure 2.13: A schematic of a DSC based on the Netzsch Polyma DSC61. ............. 33 Figure 2.14: A sketch of a DSC thermogram showing the melting endotherm ......... 35 Figure 2.15: A cross-sectional schematic of the dynamic angle of repose apparatus ................................................................................................................................. 37

x Figure 3.1: The polymerization of PA12 from either laurolactam or 12-aminolauric acid via a ROP or a condenstation polymerization, respectively. Note that ROP yields PA12, while the condenstation reaction yields PA12 and water. ............................. 40 Figure 3.2: GPC results of degraded PA12 from Pham et al6 .................................. 43 Figure 3.3: The mechanical properties of parts printed with recycled powder, from Wudy38. .................................................................................................................... 44 Figure 3.4: The mechanical properties of parts printed with recycled powder, from Gornet35. .................................................................................................................. 45 Figure 3.5: The mechanical properties of parts printed with recycled powder, from Zarringhalam27. ........................................................................................................ 46 Figure 3.6: Surface quality of parts printed with SLS powders with decreasing MFR, from the work of Dotchev7. ....................................................................................... 50 Figure 3.7: The Dotchev Blending Curves ............................................................... 51 Figure 4.1: Diamine- & Diacid-Regulated PA12 Particles ........................................ 55 Figure 5.1: Cost comparison of SLS versus Injection Molding ................................. 59 Figure 5.2: Effects of Refresh Rate for the 5th build ................................................ 62 Figure 6.1: The experimental workflow .................................................................... 64 Figure 6.2: A schematic of the funnel for ASTM D1895-96, the standard for measuring powder pourability and angle of repose ................................................................... 70 Figure 7.1: Tm,onset for the aged powder. Error Bars: +/- 1 Standard Deviation (SD) 72 Figure 7.2: The Tm,peak for the aged powder. Error Bars: +/- 1 SD ........................... 73 Figure 7.3: The Tm,end for the aged powder. Error Bars: +/- 1 SD............................ 73 Figure 7.4: The ΔHm for the aged samples. Error Bars: +/- 1 SD ............................. 74 Figure 7.5: The ΔTm for samples aged at 170 °C. Error Bars: +/- 1 SD ................... 75

xi Figure 7.6: TGA results, the decomposition onset temperature of aged powder ..... 78 Figure 7.7: TGA results, the decomposition inflection temperature of aged powder 78 Figure 7.8: The Tm,peak and Tm,end of degraded powder as a function of particle size. ................................................................................................................................. 79 Figure 7.9: The Tm,end for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by powder age. ......................................................................................... 81 Figure 7.10: The Tm,end for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by refresh rate. ......................................................................................... 81 Figure 7.11: The Tm,peak for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by powder age. ......................................................................................... 82 Figure 7.12: The Tm,peak for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by refresh rate. ......................................................................................... 82 Figure 7.13: The Melt Flow Rate (MFR) of the aged powder, clustered by aging temperature. Error Bars: +/- 1 SD. ........................................................................... 84 Figure 7.14: The Melt Volume Rate (MVR) of the aged powder, clusted by aging temperature. Error Bars: +/- 1 SD ............................................................................ 84 Figure 7.15: The MFR results presented by Dotchev in 20097................................. 86 Figure 7.16: The Angle of Repose as of the aged powder, clustered by aging temperature. Error Bars: +/-1 SD ............................................................................. 88 Figure 7.17: Apparent density of the aged powder, clustered by aging temperature. Error Bars: +/- 1 SD. ................................................................................................ 89

xii

NOMENCLATURE 2D

2-Dimensional

3D

3-Dimensional

ABS

Acrylonitrile-Butadiene-Styrene Copolymer

ALM

Advanced Laser Materials, LLC

AM

Additive Manufacturing

ASTM

American Society for Testing & Materials

CAD

Computer-Aided Design

CDT

Cyclododecatriene

Cp

Isobaric Heat Capacity

DSC

Differential Scanning Calorimetry

FDM®

Fused Deposition Model (A registered trademark of Stratasys)

FFF

Fused Filament Fabrication

HDPE

High Density Polyethylene

ΔHf

Enthalpy of Fusion (Solidification)

ΔHm

Enthalpy of Melting

LS

Laser Sintering

MFI

Melt Flow Index

MFR

Melt Flow Rate

̅𝑛 𝑀

Molecular Number Average

̅𝑤 𝑀

Molecular Weight Average

MW

Molecular Weight

xiii MWD

Molecular Weight Distribution

MVR

Melt Volume Rate

PA

Polyamide (Nylon)

PA11

Polyamide 11

PA12

Polyamide 12

PEEK

Poly Ether Ether Ketone

PDI

Polydispersity Index

PLA

Polylactic Acid

PMMA

Polymethylmethacrylate (Acrylic)

POM

Polyoxymethylene (Polyacetal)

PS

Polystyrene

RM

Rapid Manufacturing

RP

Rapid Prototyping

ROP

Ring-Opening Polymerization

SLA

Stereolithography

SLM

Selective Laser Melting

SLS®

Selective Laser Sintering (A registered trademark of 3D Systems)

Tg

Glass Transition Temperature

Tm

Melting Temperature

Tm,end

Temperature at the end of the melting endotherm

Tm,onset

Temperature at the onset of the melting endotherm

Tm,peak

Temperature at the peak of the melting endotherm

xiv ΔTm

Melting endotherm range (Tm,end – Tm,onset)

TGA

Thermogravimetric Analysis

UHP

Ultra-High Purity

1

1 INTRODUCTION The first industrial revolution at the close of the 19th century brought about an unprecedented increase in worldwide production, which was quickly followed by the second industrial revolution in the early 20th century that was started by Henry Ford’s assembly line and the petroleum fueled economy1. In the second half of the century, the inclusion of computer technology improved efficiency, but the majority of consumer and industrial products were still manufactured on assembly lines1. Now in the 21st century, prominent economic theorists, such as Jeremy Rifkin, have suggested that 3-dimensional (3D) printing will usher in the third industrial revolution and dramatically shift the paradigm away from the centralized assembly-line factories to localized and custom production that meets the specific demands of the consumer1–3.

Some of the strongest candidates to drive this revolution are the additive manufacturing (AM) techniques, which are defined by ASTM International as:

A process of joining materials to make objects from 3D model data, usually layer upon layer, as opposed to subtractive manufacturing methodologies4.

Therefore, AM meets both requirements set forth by Rifkin because one machine eliminates the need for an assembly line of subtractive techniques, and because enduse parts are highly customizable and built to order by simply modifying the computeraided design (CAD) model.

2 While 3D printing has the potential to be the keystone of the third industrial revolution, there are still many challenges that must be overcome before this emerging technology can challenge the status quo of assembly-line based manufacturing. In particular, quality must be improved and cost must be driven down; however, all emerging technologies face this barrier.

Selective laser sintering (SLS®) is one area that holds great potential for cost efficiency improvements; the process converts powdered thermoplastic resins, such as polyamide 12 (PA12), into end-use parts by melting and consolidating the powder with a laser. The process currently requires powdered thermoplastics that cost approximately $100 per kilogram for PA12 and have a build conversion of powder of only 15 %5,6. While the unsintered (unused) powder is recoverable, it is thermally degraded during the process and its properties are altered. Therefore, virgin powder must be blended in to refresh the recovered powder’s properties. Manufacturers recommend 30 to 70 % virgin powder7. Not only is this blending recommendation arbitrary, it also is very costly given that refreshing at 30 % is $40 per kilogram cheaper than refreshing at 70 % virgin powder. In addition, the current literature has not explored the topic in great detail, and the results have improved the recycling process from an arbitrary process to a slightly quantitative process that focuses on part quality rather than understanding the fundamental phenomena that necessitate powder refreshing. If the cost of SLS is to decrease and break down the high cost barrier, the reasons for recycling must be understood.

3 The goal of this work is to further explore and quantify the degradation of powder by measuring powder properties as a function of the degree of degradation (time spent at a specific temperature) and to assess the influence of refreshing the aged powder with virgin material. While the long-term goal is to decrease the cost of the SLS process, a small but significant step must first be taken in understanding the fundamental science of recycling powder.

Chapter 2 provides a theoretical background. A literature review and a patent review of the recycling of PA12 powder is given a Chapter 3 and Chapter 4, respectively. Chapter 5 explores the costs of the SLS process and recycling. Chapter 6 describes the methodology. The results are presented and discussed in Chapter 7. A summary and outlook is outlined in Chapter 8.

4

2 THEORETICAL BACKGROUND

2.1. Polymers

Synthesized from monomer units, polymers are macromolecules with extremely large average molecular weight (MW) compared to the monomer and other organic molecules.

The simplest polymers are assembled from a single monomer. For

example, 12-aminododecanoic acid is linked successively to form long chains to form PA12. When a monomer reacts with a second monomer it is transformed into the repeat unit of the polymer.

Figure 2.1: Synthesis of a polymer, polyamide-12, with 2 function end groups from a monomer.

5 In the case of PA12, the carboxylic acid and the amine functional groups rearrange to form the amide function group, which now links the two original molecules. This occurs thousands to tens of thousands of times in the synthesis to create long chains of repeat units, which ultimately yields the macromolecule referred to as a polymer.

Figure 2.2: The effect of molecular weight on the glass transition temperature, the melting temperature and the degradation temperature.

In addition to the repeat unit, there are some other important features of the polymer chain. While the vast majority of the polymer consists of the repeat unit, the ends of the chain are typically capped by a different moiety. In the case of polyamide, one end is usually an amine functional group while the other terminus is a carboxylic acid

6 group. This has great implications as these end groups behave differently than the repeat unit. For example, these end groups may be more reactive than the rest of the polymer chain, allowing for additional chemistry to be performed at this location. One could imagine modifications to improve stability. End groups also influence crystallinity, as they are an impurity, which in turn affects the melting of the polymer.

When characterizing the polymer, the size of the polymer chain is one of the most commonly reported values. However, there are many ways to report the size since most—especially synthetic polymers—are not monodisperse (i.e. the chains do not all have the same number of repeat units). Therefore, after synthesis, the polymer has a distribution of chain sizes. From this distribution, the number average and weight average can be calculated by taking first and second moment of the molecular weight function, respectively. A polymer with a degree of polymerization i, will be defined as an i-mer. In other words, an i-mer has an i number of repeat units. The number average (

) is defined by

(2.1)

where mi is the weight of all i-mers, ni is the number of i-mers , Mi is the molecular weight of an i-mer and Mo is the weight of one repeat unit. Therefore ni can be related to mi by

7 (2.2)

The molecular weight average (

) is defined by

(2.3)

The

is therefore always a larger number than the

since more value is given to

i-mers of higher molecular weight for the weight average.

Figure 2.3: A hypothetical distribution of polymer molecular weight

From

and

the dispersity of the polymer can be calculated. The polydispersity

(PDI) is defined by

8

(2.4)

The PDI is always greater than one since

. As the dispersity of the polymer

decreases, the PDI approaches 1 as the weight of the i-mers converges. For monodisperse polymers, such as protein that have an exact number of repeat units, the PDI equals 1.

2.2. Polyamide 12 Synthesis

Both the polymerization of dodecane-12-lactam (Lauryl Lactam) and 12aminododecanoic acid (12-aminolauric acid, ω-aminolauric acid) yield PA12; however, due to improved stability, especially at elevated temperatures, the ringopening polymerization (ROP) of the lactam is the preferred method of Evonik Degussa GmbH, the primary supplier of PA12 for SLS powder8–10. Catalyzed by water and either cationic or anionic initiated, the ROP yields a bifunctional terminated PA12 with an amine and carboxylic acid end group.

Of the two synthetic pathways that yield PA12, the ROP of lauryl lactam is preferred in industry because it yields a more stable product. The degree of polymerization for ROP of laurel lactam is not dependent on the concentration of water and the initiators are less detrimental to polymer stability. When the condensation polymerization is

9 used, water must be removed to drive the reaction due to Le Châtelier's principle. Therefore, the presence of water would drive the polymerization in reverse.

It is also important to remember that the polymer is a bifunctional terminated molecule with one amine and one carboxylic acid end group. These two functional groups can react, just like the bifunctional 12-aminododecanoic acid monomers. Therefore, postpolymerization can occur with significant implications. If two end groups on two different polymer chains were to react, the average MW would double for a two chain system. Statistically it is rare for chain ends to find each other, as the chain-end-torepeat-unit ratio is extremely small; however, if this does occur—for example at elevated temperatures—the average MW would increase, possibly significantly.

Figure 2.4: ROM Polymerization of PA12 from Lauryl Lactam

10

If PA12 is synthesized from lauryl lactam, the source of the feedstock is typically butadiene, which is a petroleum product. The first step is to react three equivalents of butadiene in the presence of catalysts to yield one equivalent of cyclododecatriene (CDT). CDT gained notoriety in 2012 when a fire at the Marl Chemicals Park plant owned by Evonik GmbH severely limited PA12 feed stocks, which greatly affect the automotive and the 3D printing industry. There are many synthetic routes to convert CDT to Lauryl Lactam. One such route first epoxidizes CDT with peracetic acid and then hydrogenates it to yield a saturated epoxide11. The reactive epoxide is treated with magnesium iodide to yield an oxime, which is reacted with hydroxylamine to yield the lauryl lactam. The exact synthetic route is not important but it is important to remember that currently PA12 is a petroleum based product.

Since the 2012 fire that had significant implications in the world economy, especially for PA12 automotive parts, there has been a push to find alternative feed stocks— preferably renewable ones9. In 2013 Evonik started to explore the possibility of using 12-aminododecanoic acid derived from renewable sources; however, this project is only at the pilot plant stage12. Evonik also built a second CDT plant in Singapore signaling that the petroleum based synthetic route has long-term importance12.

11

2.3. Polymer Powders

Polymer powders hold an important niche market for rotational molding, spray coating of metal substrates and additive manufacturing13,14. As the market is relatively small, especially compared to plastic pellets, there are very few suppliers of commercially available powders. For example there are approximately 30 commercially available LS powders compared to the nearly unlimited number of resins available as pellets for high shear applications such as extrusion or injection molding14.

Figure 2.5: A schematic of particle coalescence

The most common techniques for manufacturing commercially available polymer powders employ cryogenic grinding, suspensions, gas phases or emulsion to produce

12 the particles13. Evonik GmbH, the supplier of powder to ALM uses a physical process, i.e. cryogenic grinding to produce powdered PA1215. Even though the market share is limited, powders, typically between 10 µm and 1000 µm in diameter, have distinct advantages over conventional pellets, which are a few millimeters in diameter. As powders are manufactured from resins with low melt viscosities, particle coalescence will occur due to surface tension even under no-shear13,16. This is in stark contrast to pellets, which require shear forces for pellet coalescence, e.g. extrusion or injection molding.

Particle coalescence is a multistep process that results in the strong fusion of the particles if the polymer chains from adjacent particles become intertwined. First the particles must melt, which results in a flow field due to the low zero-shear viscosity of the polymer and the high surface tension. This results in the formation of “necks” between the particles. Finally, if the particles remain molten for long enough, molecular motion will result in intertwining of the polymer chains. If the polymer does not reach the melt state or for only for a very short period of time, the molecules from adjacent particles cannot interact. The particles will be fused but the interface will be very weak, analogous to a weak weld line.

13 2.3.1. Powder Flowability and Packing Efficiency

The flowability and packing efficiency of polymer powders are very important for the SLS process and are determined by factors such as particle size distribution, moisture content, morphology, surface texture, density, electrostatic charges, the temperature, etc17–20.The four mechanisms that resist particle movement (flow) are: friction, mechanical interlocking, inter-particle forces and liquid bridging17. Friction between particles is a material property that is highly dependent on the surface roughness of the particles. If the particles are not spherical and have organic shapes, it is possible for two particles to interlock like two puzzle pieces. Interlocking of particles will resist flow. Inter-particle forces can result in cohesion that could be due to an electrostatic potential. Finally, if there are liquids between the particles, for example water, capillary forces can create a bridge across the interface of two particles resulting in a diminution of independent particle movement.

2.4. Melting of Semi-crystalline Polymers

Melting of polymeric materials is a complex process that is highly dependent on numerous material properties and is of great importance in the SLS process as the laser must completely melt the 2D cross-section of the given layer21–23. For this reason the melting process and the changes in the melting process as function of the degree of aging must be well understood for successful builds.

14 In the simplest terms, melting is an increase of free volume when temperature increases, which results in the complete dissociation of the lamella24,25. In semicrystalline polymers, such as PA12, the polymer chains regularly align during cooling, similar to crystal lattice structures of low molecular weight molecules. The chains first order at the molecular level and then at the nanoscale as lamella and finally form spherulites at the micrometer scale26, as seen in Figure 2.6. As crystallization is a molecular phenomenon, it is not surprising that small changes in the polymer’s structure can influence the dissociation of the lamella and in turn the melting process.

Similar to employing freezing point depression to melt ice by adding salt, the melting point of a polymer can be decreased with the addition of an impurity. While plasticizers are the quintessential additive for freezing point depression of polymers, other factors

Figure 2.6: Crystalline Structures, Courtesty of Osswald48.

can have the same effect. While often neglected, the chain-ends do act as impurities as they differ significantly, especially in PA12, from the repeat unit and therefore

15 depress the melting temperature25. The depression of the melting temperature can be described by

(2.5) where Tm infinity is the melting temperature of a polymer of infinite molecular weight, or in other terms, a polymer with no chain ends.

Additional emphasis must be placed on melting because it is well know that in order to achieve successful sintering the powder must be quickly melted by the laser. The polymer melt must then consolidate to ultimately form the solid part, otherwise there are issues with porosity16. Therefore it is important that the melting temperature range be as narrow as possible. More in regards to this topic will be discussed in future sections.

The melting process is so critical that significant work has been completed on the effects of poor sintering due to incomplete melting. Zarringhalam et al has shown that incomplete sintering can occur if the laser powder is not sufficient or the particle size is too great21,22,27. Both optical microscopy and DSC were used to determine that there were regions of PA12 that never melted. The authors suggested that a different crystal structure forms when the PA12 crystalizes during the SLS processes compared to the powder production because the melting endotherm of an incompletely sintered part shows two distinct peaks.

16

2.5. Additive Manufacturing

Before discussing a specific 3D printing technique, it is imperative to frame the topic with an overview of the state of the art of AM. AM and the closely related field of rapid manufacturing (RM) or rapid prototyping (RP) produces end-use parts by joining materials, layer by layer, to produce 3D parts directly from a CAD file. This is in contrast to traditional manufacturing that is subtractive, e.g. milling or lathing. The field has also been referred to as solid free-form fabrication as the AM techniques do not require a mold to produce parts.

Without the need for costly and time-consuming mold design and manufacturing, AM has been instrumental in the field of RP and RM for production of highly customizable parts, including prototypes. Parts can be designed and manufactured in a fraction of the time compared to traditional prototyping methods. Without the restrictions of a mold, part geometries can have nearly infinite complexity. It has been estimated that RP reduces development costs 40 to 70 % and reduces development time by 90 % 28

. For example, Saab Avitronics released a case study in 2008 that showed RM can

speed development, reduce part weight and decrease manufacturing costs for low volume production of unmanned aerial vehicle components29. It should be noted that rapid does not refer to the speed of manufacturing, since AM techniques are relatively slow compared to tradition polymer processing machines such as extrusion or injection molding. The savings arise since purchasing a mold can include a multi-week

17 lead time and be very costly, thousands to hundreds of thousands of dollars depending on complexity.

Figure 2.7: A representative FDM printer (Top) A close-up of the print head and the part (Bottom)

18 While all of the AM techniques uses the layer by layer manufacturing approach, the materials used vary greatly and can be divided into three primary categories: filaments, liquids and powders. Fused deposition modeling (FDM®) or fused filament fabrication (FFF) is the highly publicized technique due to the low-cost of machines and materials. FFF feeds a filament into the print nozzle, where it is heated to above the Tm of the polymer and extruded from the nozzle. The nozzle moves in 3D space in respect to the build platform to lay down a bead of molten polymer. One major disadvantage of the FFF technique is the requirement for support structures if overhangs or undercuts exist. Upon build completion, these support structures must either be physically removed or chemically dissolved if a second, water or organic solvent soluble material is used for the support material, it can be dissolved. An

Figure 2.8: Two FDM printed parts with build orientation depedent support material

example of an FMD part with the required support material is shown in Figure 2.8.

19 Note that for overhangs and undercuts support material is required and can be very dependent on the orientation of the part.

Figure 2.9: SLA printed parts

The most prominent liquid based technique is stereolithography (SLA), which in 1987 was the first commercially available 3D printing machine [SOURCE]. 3D Systems was the manufacturer of the SLA-1 and has continued to be an industry leader for both SLA and SLS after the acquisition of DTM Corp, the first company to sell a commercially available SLS machine. SLA uses a laser beam to selectively polymerize a liquid photopolymer near the surface of a vat of the liquid. The laser initiates the polymerization, which converts the liquid in to a solid polymer. The build surface is then lowered into the vat and the subsequent layer is built upon the previous layer. SLA can be extremely accurate and yields an excellent surface finish; however,

20 the materials are extremely limited and post-build curing is required to full polymerize (increase the strength) the parts. Just like FFF, SLA also requires support structures for overhangs and undercuts.

There are two major powder based techniques, binder jetting and SLS®. Both techniques layer down a layer of powder, which must be fused. For binder jetting, a print head sprays a liquid binder material onto the powder layer to fuse the particle. One advantage of this technique is that the binder material can be colored, which allows for multicolored parts. Unfortunately the binder material does not create strong parts, especially compared to SLS, which will be discussed in more detail in the next section.

2.5.1. Selective Laser Sintering

For uniformity and simplicity this thesis will be solely based on the DTMa Sinterstation 2500 SLS machine and Advanced Laser Materials, LLC (ALM) 650 PA12 SLS grade powder; however, the results should be applicable to all PA12 powders and all SLS machines as the majority of the differences are superficial.

a

DTM Corporation was acquired by 3D Systems Corporation in 2001

21 Laser Sintering (LS), the generic term, or SLS®, the registered trademark of 3D Systems Corp., is a powder bed fusion process, which builds the end-use part, layer by layer, by melting and consolidating a powdered material, e.g. PA12. The powder is melted by a CO2 laser, which traces 2D cross-section of the current layer. These layers are sequentially consolidated within the part cylinder to produce the final part4,30. Before describing the process in more detail, the nomenclature should be clarified. Sintering is a historical misnomer as consolidation, see §2.5.1.3, of powder particles requires fully melting the powder. The term sintering is inappropriate since sintering has traditionally been reserved to describe particle fusion processes below Tm4,13. However, due to ubiquitous use of the term sintering, the term SLS® will still be used in this thesis. As SLS® is a registered trademark of 3D Systems Corporation, Laser Sintering (LS) has been adopted as the generic term. While the term Selective Laser Melting (SLM) would be more appropriate, it has been reserved for a metal technique related to metal SLS that fully melts the metal versus metal SLS, which does not fully melt the metal powder. For these reasons, SLS will be used with the proviso that the reader should remember that SLS refers to the complete melting of the polymer by the laser to achieve coalescence.

2.5.1.1. SLS Machine

The SLS machine can be split into three major components: the powder system, the heating elements and the laser. The powder system includes the part cylinder,

22 commonly referred to as the part bed, the powder feed, the overflow cartridges and the roller system.

Figure 2.10: A schematic of the Sinterstation 2500

The heating system includes the heaters above the feed cylinders and the heating bulbs located above the surface of the part bed. There are also heaters located around

23 part and feed cylinders, include on the pistons. The pistons move powder out of the feed cylinders, which is moved the now lowered part cylinder by the roller system. The overflow cartridges collect any excess powder that the roller did not deposit in the part cylinder. The laser is a 25, 50 or 100 watt continuous-wave CO2 laser that operates at a wavelength of 10.6 µm, or 28.3 THz and can travel at 508 cm/s (200 inches/s).

2.5.1.2. Pre-Build

Figure 2.11: The general AM build sequence from CAD to end-use part

First the CAD file must be converted to a .STL (the de facto file accepted by 3D printers) and sliced into 2D cross-sections. The parts are then distributed by the software within the build cylinder to optimize powder conversion. Next, the powder feed bins are loaded with either virgin or recycled powder and the build chamber is

24 sealed. The build chamber is flushed with nitrogen to decrease the oxygen concentration to less than 5 %. This is an important step as polymers, especially PA12, are sensitive to thermal-oxidative degradation31. Finally the part cylinder is loaded with an approximately 2.5 cm layer of powder using the roller system. The feed bins are heated to 80 °C and the part cylinder to between 166 and 170 °C.

2.5.1.3. The Build and Consolidation

Once the system is at the temperature set-points, the machine can start printing the parts; this step is commonly referred to as the build. After preheating, the roller pushes the powder from the feeder bin to the build cylinder. At this point the surface of the build surface is heated to 170 °C. Once at temperature, the laser traces the 2dimensional (2D) cross-section for the first layer of the build based on the slice generated from the .stl file. These steps are shown in Figure 2.12. The energy from the laser is absorbed by the powder, which is converted into thermal energy and thus increases its temperature. The powder is at 170 °C and must be raised to just above the Tm, which includes the phase change to molten polymer. When the powder melts, it coalesces as described in §2.3. It should also be noted that only gravity and capillary force drive consolidation as there is no mechanical pressure, such as in the injection molding process16. To minimize warpage the build cylinder is kept near Tm to minimize part warpage32. Therefore, unsintered powder is maintained at this elevated

25 temperature until the completion of the build. While necessary to ensure part quality, this is the reason the powder degrades during the build.

Upon completion of the layer, the build cylinder piston is then lowered, typically 0.1 mm and the next lay of powder is deposited by the roller. Therefore, to print parts, hundreds of layers of powder must be sintered to yield the final part. As each layer can take between 30 to 60 seconds, which is highly dependent on many factors, the full build takes many hours.

2.5.1.4. Post-Build

Upon build completion the build volume is cooled very slowly, typically between 24 and 48 hours in total, to minimize thermal gradients within the build cylinder that would result in warpage and to provide sufficient time for stress relaxation to prevent warpage32,33. The “part cake”, the term given to the unsintered powder in which the parts are contained, is removed from the machine and parts are pulled from the powder. The dense consolidated parts are approximately twice the density of the powder16. At this point, the unsintered part cake powder can be recovered if it is to be recycled and used in subsequent builds. The powder is typically sieved through a #70 mesh, 200 µm to remove large agglomerated particles. Interestingly virgin powder is between 10 and 100 µM, which is significantly small than the mesh. Therefore, some agglomerated particles could pass through the sieve.

26

Figure 2.12: The SLS Build Cycle

27 2.5.1.5. Advantages

The advantages of AM and the related field of RP are well documented. The ability to build a physical model without tooling and tooling restrictions has ushered in an era of unlimited possibilities for part design in combination with the ability to produce the part nearly instantaneously. All AM techniques, such as the filament based FFF, the photopolymer liquid based SLA or the powder based SLS have these advantages However, due to the stark differences between the different AM techniques, there are significant pros and cons for each method.

There are three major advantages of the SLS process over other AM techniques, such as FFF. First, the unsintered powder acts as support material, which allows for parts with undercuts and overhangs5. Second, the mechanical properties of the parts are comparable to that of injection molded parts, allowing for end-use applications5. Lastly, the breadth of material available for the SLS process is greater compared to other processes. For example, FFF primarily uses poly-lactic acid (PLA) and arcrylonitrile-butadiene-styrene (ABS) copolymer. In contrast, polyamide 11 (PA11), PA12, polystyrene (PS), polyetheretherketone (PEEK), polyoxymethylene (POM), polymethylmethylarcylate (PMMA) and high-density polyethylene (HDPE) have all been sintered successful, however, PA12 accounts for 95 % of the industry usage16.

28 2.5.1.6. Disadvantages

The greatest disadvantage of the SLS process is the economics of the process because of the long build times and the high-cost of the powder and the machine. The cost of SLS will be discussed in more detail in Chapter 5 The complexity of the process could be considered a disadvantage compared to other techniques such as FFF or SLA. The complexity primarily originates from the requirement to melt only the 2D cross section with the laser and no more and no less. The vast number of variables, including the degree of degradation of the powder, makes successful melting a daunting task, especially compared to a method such as FDM.

2.6. Aging and Degradation

In order to properly discuss the aging of PA12 powder, the wide and varied mechanisms of aging must be defined and discussed. Aging is any change in the molecular, supermolecular, or phase structure of a polymer that alters physiochemical properties31. Aging will influence all aspects of the polymer’s life cycle from manufacturing, storage and service. While all three are important, emphasis will be placed on aging during the manufacturing period.

The cause of aging can be from one of four categories: Internal, External, Chemical or Physical31. Chemical and physical processes are negligible in the SLS process but internal process arising from thermodynamic instabilities could be important and

29 external processes, such as thermal degradation are known issues5,31,34–38. To ensure a complete investigation, all of the factors known to cause polymer degradation as presented the textbook, Resistance and Stability of Polymers, by Ehrenstein and Pongratz, will be considered31. The factors are listed in Table 2.1.

Table 2.1: External and Internal Factors that contribute to polymer degradation.

External Factors Temperature

Internal Factors Chain Structure

Oxygen Concentration

Molecular Weight and Distribution

UV Light

Structure of the Molecular Chains

Humidity

Radical Concentration

Contaminations

Crystal Structure

Radiation

Crystallinity

Mechanical Loads

Morphology Free Volume Orientation Residual Stresses Stabilizers Residual Catalyst



Temperature is of significant importance as the build chamber is between 80 and 170 °C.



Oxygen Concentration is a variable that is controlled in the SLS build chamber and should be as low as possible as polyamides, including PA12 are sensitive to oxidation.

30 

UV Light is not a concern as there is no source of UV radiation within the build chamber.



Humidity is not a concern as the build chamber is flushed with dry nitrogen gas.



Radiation is of possible concern as heating bulbs are used to raise the surface temperature of the build area and IR radiation is used to heat the powder.



Mechanical Load is not a concern.



Chain Structure is of interest but there is no control over this factor since powder can only be purchased from select manufacturers.



Molecular Weight and Distribution are of interest but just as the chain structure cannot be controlled due to external factors, there is no control over these attributes of the powder. However, it could be monitored during the aging of the PA12.



Radical Concentration is of no concern as polyamides are a product of a condensation polymerization or ROP and not a free-radical polymerization.



Crystal Structure and Crystallinity are both concerns and will be monitored as a function of aging.



Free Volume is a material property that cannot be controlled and should remain relatively constant.



Orientation and Residual Stresses are not of concern.

31 In summary, temperature, oxygen concentration, molecular weight, radiation, crystal structure and crystallinity will be considered in detail.

2.6.1. Thermal

Thermal degradation of polymers is any change to the molecular or supra-molecular structure, i.e. the crystallinity, due to increasing the temperature of the material but not due to any reactions with oxygen31. The increase of temperature increases the amount of energy in the system, which increases the rate of degradation. For example, a rise in temperature intensifies molecular vibration, which increases the probability of chain scission. If the scission occurs near the chain end, this would slightly decrease the MWD; however, random scission can dramatically decrease the MWD of the polymer. The rise in temperature also increases the probability of chain-end reactions, as the PA12 is bifunctional and the carboxylic acid and amine groups can react. This would result in increasing the MWD.

2.6.2. Thermal-Oxidative

If the polymer is heated under an oxygen rich atmosphere, there are additional degradation pathways as diatomic oxygen will readily react with intermediates of the thermal degradation pathways or with PA12. For example, thermal-oxidation of PA12 results in a yellowing of the polymer due to the generation of impurities. It has also been hypothesized that oxygen can initiate chain-scission pathways38.

32

2.7. Quantification Techniques

To observe the degradation of the powder both at the molecular and macroscopic level, the powder properties will be measured using quantification techniques that are dependent on either the molecular structure, i.e. the MWD, or on the particles, i.e. the flowability of the powder. A wide variety of tests were conducted as some properties are more susceptible to changes due to degradation and techniques vary in their level of sensitivity to detect these changes.

2.7.1. Differential Scanning Calorimetry

As the melting of the powder by the laser is the crux of the SLS process, the maximum amount of information regarding the melting of the PA12 is critical for understanding changes in the powder. Thermal analysis, in particular differential scanning calorimetry (DSC) is a powerful tool for measuring the heat flux during melting.

A DSC measures the energy required to heat a sample at a constant temperature ramp, typically between 1 and 20 K/min under a constantly purged nitrogen atmosphere. The temperature range should include the glass transition (Tg) and the melting temperature (Tm) for semi-crystalline materials. The sample is placed within a crucible, commonly made of aluminum or other inert materials depending on application. The second crucible is empty and acts as the reference. The two crucibles are placed upon heat flux sensors within the DSC oven, and the difference between

33 the sample and the reference is the energy required to increase the sample temperature. From this data the following can be obtained:



Enthalpy of Melting (ΔHm)



Enthalpy of Fusion or Crystallization (ΔHf)



Melting Temperature (Tm)



Glass Transition Temperature (Tg)



Specific Heat Capacity at Constant Pressure (Cp)

Figure 2.13: A schematic of a DSC based on the Netzsch Polyma DSC61.

34 Figure 2.14 is a sketch of a DSC thermograms that shows the five main values of interest when melting a polymer. When a polymer melts, it does not do so at a specific temperature but rather over a temperature range, since the melting of a polymer is more complex than a simple molecule (see §2.3). Therefore, the melting endotherm of a polymer occurs over a temperature range in the DSC thermograms. From this endotherm the following values can be measured: Tm,peak is the temperature at which the greatest endothermic heat flux occurs; Tm,onset and Tm,end are determined by drawing a tangent to the baseline of the thermogram and a tangent to the thermograms just before and after Tm,peak; the ΔTm is difference between Tm,end and Tm,onset; the total heat flux required to melt the polymer, ΔHmelting, is determined by interpolating the baseline across the melting endotherm and calculating the area under the curve. Of the five values, ΔHmelting typically is the least reproducible because baseline must be interpolated, which can introduce error. There are also multiple methods of baseline interpolation. For example Netzsch Proteus 7.0 offers these possibilities: linear, tangential, sigmoidal, horizontal left or right started, tangential left or right started and Bezier. The user must choose an appropriate method for baseline interpolation and then chose proper bounds. Fortunately, the four Tm values do not require user inputs and are significantly more reliable.

35 In the thermal analysis literature, ΔHfusion, is often used instead of ΔHmelting, as for many materials and some polymers the melting endotherm and crystallization (solidification) exotherm are quite similar. However, for polymers, including PA12, the endotherm and exotherm are very different both in shape and the temperature at which it occurs. Therefore, ΔHmelting will be used to emphasize this difference.

Figure 2.14: A sketch of a DSC thermogram showing the melting endotherm

More importantly DSC data can be used to elucidate information about the molecular structure, including the average molecular weight and the polydispersity polymer. Based on polymer melting theory, § 2.4, an increase of Tm could indicate an increase of average MW.

36 2.7.2. Thermogravimetric Analysis

Thermal stability is a critical attribute of any polymer, especially the SLS grade PA12 as it must be stable at build-chamber temperatures over extended periods of time. Thermogravimetric analysis (TGA) is a thermal analysis technique that measures mass changes as a function of temperature and time. Using a precision oven and microbalance (resolution of 0.0001 mg) small changes in mass can be detected, from which moisture content, plasticizer content, decomposition temperature, etc. can be quantified. There are two forms of TGA: dynamic and isothermal. Dynamic TGA ramps the oven temperature at a constant rate, which can determine the decomposition temperature of the polymer and any additives. Isothermal TGA quickly ramps the temperature to a set point and then maintains this temperature to yield stability information at the given temperature.

2.7.3. Powder Flowability and Packing Efficiency

There are numerous techniques to determine the flowability and packing efficiency of powders, including polymer powders. For flowability, the techniques can be subdivided into two general categories. First, a method that measure the elapsed time required for a powder to pass through an orifice, e.g. a funnel. This is the premise behind the hourglass. Second, a method that measures the angle bulk powder can maintain without sliding (avalanching). The simplest of which is the angle of repose

37 test. A pile of powder is created and the angle of the pile relative to the surface is measured. More advanced techniques include sprinkling powder on a flat surface and then rotating the surface until the powder avalanches. To observe this phenomena for dynamic powders, powder can be placed in a rotating cylinder and the powder shape can be measured, see Figure 2.15.In this apparatus, the angle of repose is both dependent on the powder and the angular velocity of the cylinder. This dynamic angle of repose could be valuable, as spreading the powder is a dynamic process.

Figure 2.15: A cross-sectional schematic of the dynamic angle of repose apparatus

38

3 LITERATURE REVIEW

3.1. Powder Flowability

As the powder is transferred from the feeder bins to the part bed by a roller system, the flowability of the powder is of great importance as the quality of powder deposition affects the final part quality18,19,39–41. There are numerous factors that influence how well particles flow, including shape, surface roughness, interparticle friction, etc., and there are numerous techniques for powder flow quantification39,42. Krantz et al suggest a suite of techniques to determine particle size distribution, the bed expansion ratio, the angle of repose and the avalanche angle and the cohesion results from powder rheology would sufficiently describe powders used in industrial applications, include SLS. Building off this work, Ziegelmeier et al suggested specific tests, including those of Krantz et al that would be applicable to packing efficiency within the build chamber and the flowability during the layering the powder.

The work of Ziegelmeier et al and Krantz et al did not focus on recycled powders but the results of their analysis of virgin powder suggested that degradation could influence flowability significantly. These two studies showed that flowability varies between grades of SLS powder and is a function of the temperature at which the powder was tested. If merely increasing the temperature changes the flowability of a powder, then degradation of that powder surely could be significant.

39

3.2. Degradation Pathways

Before the effects of thermal degradation of the powder can be explored, the degradation pathways should be understood. Unfortunately, only poorly supported hypotheses have been suggested, primarily in the publications by Wudy et al, at the University of Erlangen. While underwhelming, it is still necessary to review these hypotheses and critically examine the implications on the SLS process, especially from the recycling and cost perspective.

A recent hypothesis from Wudy et al is that a post-condensation polymerization would increase the average molecular weight and would be classified as thermal degradation31,43,44. While plausible, it assumes that the polyamide powder was produced via a condensation polymerization of 12-aminolauric acid. Interestingly, the schematic of this “post-condensation” polymerization presented in the article actually shows monomeric laurolactam reacting with terminal carboxylic acid of PA12, which is a ROP and not a condensation reaction, Figure 3.1. This is critically flawed, as ROP would not occur at the conditions within the SLS build chamber as catalysts, typically cationic, are required for polymerization. These catalysts are removed from the polymer during purification as they are typically very expensive and toxic to humans.

40 In addition, this polymerization would not be a condensation polymerization but a ROP.

Figure 3.1: The polymerization of PA12 from either laurolactam or 12-aminolauric acid via a ROP or a condenstation polymerization, respectively. Note that ROP yields PA12, while the condenstation reaction yields PA12 and water.

For thermal-oxidative degradation, Wudy et al have suggested that PA12 radical species are formed at elevated temperatures (170 °C), which initiates a series of possible chemical reactions that would result in cross-linking and/or chain-scission.

41 However, there is no evidence that this is the reaction pathway that actually occurs, let alone the cause of the observed increase of average molecular weight. First, this mechanism has only been observed in PA6,6 and one cannot assume a similar reaction will occur with PA1245. Second, the authors provided no data supporting the hypothesis that this pathway is the cause for the observed changes. Assuming hydrogen abstraction at the α-carbon due to stabilization of the secondary carbon on the adjacent amide can be justified46,47. However, this rationale nor any other supporting data was provided by the authors. Work by Brodski et al used electron spin resonance spectroscopy to observe radical formation of PA 6, 6,6; 6,8; 6,10; and 11 irradiated by γ radiation47. It was found that radical formation occurred predominantly at the α-carbon and could lead to their chain-scission or cross-linking. Before jumping to the conclusion that this occurs in the SLS powder, electron spin resonance spectroscopy could be used to observe if this phenomena occurs for PA12 as well.

It is unfortunate that the current literature has audaciously included statements regarding the complex chemistry of polymer degradation; however, this topic must be addressed if the recycling process is to be fully understood. For the time being, until the true degradation pathway occurring during the SLS process can be elucidated, these assumptions must be disregarded to ensure that the data is analyzed in good faith. Therefore, the assumption must be that changes are occurring at the molecular level that could result in increased or decreased average MW.

42

3.3. The Effects of Thermal Degradation

It has been well documented that printing with thermally degraded powder yields inferior parts, and, in extreme cases, the print job can completely fail. The major observable issues of these inferior parts are a poor surface finish and a decrease in the mechanical properties. However, the causes for these issues are not well understood.

3.3.1. Molecular Weight

One of the most important properties of any polymer is its molecular weight distribution (MWD) as countless other properties, such as Tm, are highly dependent on it25.

In the scope of degradation, molecular weight (or, more specifically, the average molecular weight) must be considered as degradation at the molecular level which, in turn, can include chain-scission, additional polymerization or cross-linking. All of which will increase or decrease the average molecular weight. As a result, bulk properties such as Tm or melt viscosity will be altered, which ultimately will affect the sintering process.

Pham et al reported a significant increase in molecular weight as a function of the number of times the powder was recycled. Using gel permeation chromatography, it was shown that

and the molecular number average

both increased. In

43 addition there was a significant broadening of the PDI from approximately 2 to 8, Figure 3.2. The increase of PDI is noteworthy since this increase indicates the presence of extremely long chains. Chains of this length are mostly formed when the end groups of two or more of the original polymer chains reacted to form massive chains. This is a major issue because extremely long chains will increase the zeroshear viscosity, which will in turn limit coalescence16,48,49.

Average Molecular Weight

Average Molecular Number

1,000,000

10

750,000

7.5

120,000

10

7.5 80,000

500,000

5

250,000

2.5

5 40,000

0

0 Virgin 1 Build 2 Builds 3 Builds Mw

PDI

2.5

0

0 Virgin 1 Build 2 Builds 3 Builds Mn

PDI

Figure 3.2: GPC results of degraded PA12 from Pham et al6

3.3.2. Surface Finish Issues

Surface finish quality is another major issue encountered when printing with recycled material and has been well documented in the literature. The most common phenomena observed is the “orange peel” effect, as the surface of the final part has a

44 texture similar to that of an orange6,7. While a known issue, surface quality was first quantified in 2015 by Petzold et al and a significant increase in surface roughness was observed in samples printed with aged powder50.

3.3.3. Inferior Mechanical Properties

A popular method for determining the effect of thermal degradation of the powder is to print parts, such as tensile bars and measure the mechanical properties such as tensile strength, fracture strain, the elastic modulus, etc. The literature has clearly

Figure 3.3: The mechanical properties of parts printed with recycled powder, from Wudy38.

45 shown that mechanical properties are highly dependent on the degree of degradation. For the purposes of this review, it is not practical nor necessary to investigate all trends but a representative sample will be shown to highlight that it is a significant issue.

Figure 3.4: The mechanical properties of parts printed with recycled powder, from Gornet35.

Interestingly the reasons for the decrease in mechanical properties are not well understood and many hypothesis exists for why there is a significant change in

46 mechanical properties. For example, one hypothesis is that the recycled powder does not consolidate (sinter) as completely as virgin powder22.

Figure 3.5: The mechanical properties of parts printed with recycled powder, from Zarringhalam27.

These results demonstrate the necessity for powder characterization before the build for several reasons. If minimum strength properties are to be expected, these results suggest the reuse of recovered power can be dangerous, since degradation of the powder significantly weakens the parts. If recycling of the recovered powder is to be successful, powder characterization must be predictive of final part quality. To assure quality control, one must be able to determine the minimum refresh rate of virgin powder to improve the properties of the parts printed form recycled powder.

47

3.4. Recycling Literature

3.4.1. Choren et al - 2001

Even with the high cost of powder and the known issues with recycling recovered powder, it is surprising that the first scholarly article on the subject was published in 2001 a full 15 years after the invention of the SLS process at the University of Texas at Austin by Carl Deckard

34,51

. The original study on recycling by Choren et al from

the Milwaukee School of Engineering was a cursory exploration of the mechanical properties of parts as a function of powder “age” and the laser power. Laser power was of interest as it was and still is a common practice to increase laser power when recycled powder is used. This unwritten rule has been based on empirical observations, but there have not been any rigorous analyses to support the general rule of thumb.

The results confirmed that recycled powder yields parts of poorer quality but the study did not yield any meaningful results as the design of experiments was unfortunate. First, the powder “age” was based on the number of build cycles, which is not a fixed value for each build and powder used in the build experiences had varied thermal histories depending on the location within the SLS machine. The mechanical properties, such as tensile strength, were measured and plotted while simultaneously varying laser powder and powder age. Therefore it is impossible to determine the

48 effects of either the powder aged or the laser power as both independent variables were changed at the same time.

3.4.2. Gornet et al - 2002

The 15 years thereafter has even more surprisingly yielded very few investigations of the thermal degradation of the PA12 that causes the difficulties when trying to recycle the SLS powder. The next chronological publication on the subject matter came a year later from Gornet et al from the University of Kentucky. Like Choren et al, the goal of the study was to improve the final part quality without first investigating the root of the problem.

The notable improvement was measuring powder properties rather than solely focusing on the final part properties. As the powder properties must certainly influence the final part quality and properties, it is only natural and necessary to measure powder properties as a function of thermal history. The powder’s melting temperature was determined via DSC and the melt flow rate (MFR) was determined using a Melt Flow Indexer (MFI). Two interesting trends were noted. The Tm increased and the MFR decreased as a function of “builds”. Like Choren et al, Gornet et al aged powder by using it in subsequent builds. The Tm increased approximately 1.5 K, while the MFR decrease from over 50 g/10 minutes to less than 20 g/10 minutes. Surprisingly the 1.5 K increase of Tm was disregarded by the authors as insignificant and was not

49 discussed. Rather, the authors focused heavily on the MFR, which was touted as a potential measurement for predicting final part quality when using recycled powder even though no explanation for the increasing Tm or decreasing MFR was ever offered. This could have been attributed to the fact that the authors’ goal was to ensure quality SLS builds at the University of Kentucky Rapid Prototyping Center rather than the fundamental science behind the thermal degradation of PA12.

3.4.3. Dotchev, Pham and Yussoff – 2008 to 2009

The next major contribution to the recycling literature was in 2008 and 2009 in two papers by Dotchev, Pham and Yussoff6,7. The research was a collaboration between Cardiff University in the United Kingdom and the International Islamic University in Malaysia.

This study may be one of the most comprehensive studies of the degradation as the powder was aged between 12 and 200 hours at 100, 120, 140, 150, 160 and 180 °C and analyzed by MFI, DSC, SEM and GPC. The work also included data from builds using aged and recycled powders. The culmination of the work was a workflow for testing and refreshing the recovered based on the MFI data.

First, from the results of printing with powder with a known MFR, it was shown that if the MFR was less than 25 g/10 minutes, the surface quality decreased stepwise until

50 the MFR was 18 g/10 minutes. At this point the surface quality was described as very bad. From this information, the recycling workflow was developed and published.

Figure 3.6: Surface quality of parts printed with SLS powders with decreasing MFR, from the work of Dotchev7.

After recovering the powder, the first step would be to measure the MFR of the powder and classify it according to Table 3.1. If the powder has a MFR of 24 or less, the powder would have to be refreshed with virgin powder to increase the MFR to above 25 g/10 minutes. To estimate the ratio of recovered powder to virgin powder, a blending curve was generated, Figure 3.7. The blending curve indicates that if the MFR of the powder is ~21 g/10 minutes, then 30 % virgin powder would be required and 40 % if the MFR is ~18 g/10 minutes.

51 Table 3.1: The Dotchev Classifcation of Recovered Powder

MFR (g/10 min) > 50 45 – 49 40 – 44 35 – 39 30 – 34 25 – 29 18 – 24 < 18

Classification A (virgin powder) B1 B2 B3 B4 B5 B6 (Surface Quality Issues) C (Unusable Powder)

The work of Dotchev greatly improved the recycling of PA12 by providing the first workflow for recovering powder, quantifying the powder and determining the required

Figure 3.7: The Dotchev Blending Curves

refresh rate to ensure surface finish quality of the final part. This work flow could be even more powerful if part strength and not only surface quality were a metric for determining the refresh rate. Even with acceptable surface finish, the parts will be of

52 inferior strength compared to virgin powder, as shown by Gornet and Choren. Lastly, the blending curve only includes data for powders with an MFR of between 18 and 48 g/10 minutes. The only drawback of the workflow is that MFI is time consuming and required 10-20 grams of powder. If multiple tests are conducted during refreshing, technician time and powder consumption would become significant, especially compared to a method such as DSC that only requires a few milligrams and sample prep takes only a minute or two.

3.4.4. Mielicki – 2014

It has been observed that powders recovered from the SLS part cylinder have an increased particle size distribution due to particle agglomeration. It has been hypothesized that this agglomeration is occurring at the interface of the powder acting as the mold and the melted powder. The work of Mieklicki et al further reinforced this hypothesis as agglomeration was not observed in oven aged powder5,37,43. Interestingly, it was shown that the particle size distribution does not change when powder is aged in the oven at 174 °C.

3.4.5. Wudy et al – Present

Since the work of Dotchev et al in 2009, the recycling of powder literature has for the most part come from universities in Germany with the majority from the LKT at the University of Erlangen. Wudy et al has published a series of papers on the effects of

53 powder degradation, but the work has primarily been a repeat of work done by Choren et al, Gornet et al and Dotchev et al6,7,34,35,38,43,44. The only significant contribution to the field has been the verification of previous results and no new techniques to improve recycling efficiencies have been provided.

3.4.6. Summary

In conclusion, the literature has thoroughly documented that printing with recovered powder yields inferior parts with decreased mechanical properties and/or surface finish issues. The decrease of the MFI has been well documented in numerous publications, which in conjunction with GPC data from Pham et al has confirmed an increase of molecular weight as the powder is thermally degraded. Lastly, DSC has been disregarded as a method to quantify degradation, since authors have questioned if the technique is sensitive enough.

The vast majority of publications have focused on demonstrating the effects of degradation on end-use parts, Dotchev et al were the only authors to develop a workflow for recovering and blending powder to ensure an acceptable surface finish. However, no authors have created an analogous protocol for ensuring mechanical properties.

54

4 PATENT REVIEW The patent literature is a rich source for possible solutions for preventing thermal degradation and/or improving the recyclability of recovered powder. While there are countless patents, primarily from Evonik Degussa GmbH and Arkema, Inc., the two most relevant or promising will be discussed.

4.1. Diamine- and Diacid-regulated PA12

As post condensation is an issue with many commercially available PA12 powders due to the high temperatures and low moisture content within the SLS machine, several patents aim to prevent these reactions38,43,52,53. The method of choice to prevent condensation is to use diamine- and/or diacid-regulated PA12. The reason post-condensation is undesirable is that is increases the MWD significantly and in turn the melt viscosity, which can decrease the consolidation of the powder. By introducing diacids and/or diamine, the stoichiometric ratio of carboxylic acid to amine end groups is shifted away from 1:1. This decreases the probability of end group condensation reactions. The addition of these additives was successful in preventing condensation when the powder was exposed to build chamber conditions since the solution viscosity did not increase, as unregulated PA12 would. Unfortunately the addition of diacids increase the brittleness of the part while the diamines cause gelation issues52.

55

Figure 4.1: Diamine- and Diacid-Regulated PA12 Particles

The promising viscometry results lead to further work by Evonik Desugga GmbH that included mixing of diacid- and diamine-regulated PA12 powders53. The rationale for this idea was not provided; however, the strength of parts printed with this mixture of powders did not exhibit the issues that arose when just diacid- or diamine-regulated PA12 powders were used. Additionally, since the amine and carboxycylic acid groups were isolated in different powder particles, condensation reactions could not occur, which prevented an increase of the MWD. This patent should solve some if not most

56 of the issues related to recycling of the recovered powder; however, there does not appear to have been widespread adoption of the technique. Implementation has either not occurred or other unknown factors have limited the impact of this invention.

4.2. Hydrolysis

Instead of attempting to prevent condensation reaction, which increase the average MW, US2009/0291308 suggests that the recovered powder could be treated to initiate chain scission of the PA12 by hydrolysis54. The patented process required the user to recover the powder and transfer the material to a container in which the material can either be water or steam treated. This promoted chain scission as the amide bond is a product of a condensation reaction and the addition of water would drive this reaction in reverse according to Le Châtelier's principle.

The results shown in the patent strongly suggest that this technique can reverse the effects of degradation of observed printed parts; however, the process is complex and true cost savings were not demonstrated. There are several reason that this process is neither practical nor economical. First, the approximate average MW of the recovered powder must be determined. As the goal of this treatment is to decrease the average MW to that of virgin powder, the degree of water or steam treatment must be controlled. Once treated, the powder must be completely dried to remove free water and bound water must be reduced to an acceptable level. This can time

57 consuming as PA12 is a hydrophilic polymer. At this point the approximate MW should be determined again to confirm successful treatment.

This patent may successfully solve the degradation issues; however, as widespread implementation of this process has not occurred, a cost-benefit analysis might explain why. Therefore it is imperative to continue to search for cost-effective solutions to the problem of increasing molecular weight.

58

5 COST-ANALYSIS Traditional

plastic processing

techniques use commodity

and engineering

thermoplastic pellets. These resins cost generally less than $5 per kilogram and account for very little of the total cost of a final product. In unmistakable contrast, the $100 per kilogram cost of the SLS grade thermoplastic powder contributes significantly to the cost of the final part5. Therefore, it is essential to consider the conversion rate of powder into final parts.

5.1. Cost versus Injection Molding

Prior to discussing the implications of recycling and the refresh rate, the cost of SLS should be compared to that of a traditional manufacturing method such as injection molding. One of the most comprehensive investigations into the costs of SLS in comparison to injection molding was conducted at Loughborough University in Leicestershire, United Kingdom in 2006 by Ruffo et al55. While nearly a decade old, it still provides an excellent comparison. The Euro (€) was chosen by the authors, and these values were converted to US Dollars at a rate of $1.20 to €1, as the exchange rate hovered around that value for the fourth quarter of 2005 and the first quarter of 2006.

59

Figure 5.1: Cost comparison of SLS versus Injection Molding

The authors accounted for dozens of factors that influence the cost of SLS including powder costs, machine costs, building overhead, production overhead, maintenance and a technician’s salary. The machine was depreciated over 8 years and it was assumed that the machine was operating at 60 %. The powder was priced at $70 per kilogram for the powder and the machine at $435,000. The cost of injection molding a

60 generic part was compared to producing the same part using SLS. For the cost of LS, two models were investigated when virgin powder was used for all builds. In addition, a 50 % recycling scenario was included. Depending on the model used, the cost per unit breakeven point was 8,000 and 11,000 parts, and when recycling powder was utilized at 50% the breakeven point grew to approximately 15,000 parts. This is because powder is being reused and not thrown away, so the cost per part does not include as much waste powder that is not converted into parts.

5.2. Refresh Rate Thought Experiment

In most plastics processing, scrap material is a trifling nuisance that is dealt with by collecting scrap and grinding the scrap for reuse. For example, in the injection molding process, the sprue and runner system can be collected and ground. The regrind, can be mixed with virgin pellets with no significant decrease in the properties of the final product56. In stark contrast, the “waste” material in the SLS process cannot be simply recovered and reused. The manufacturers of SLS powder recommend various refreshing protocols for recycling powder that involve mixing a percent of virgin powder with recovered powder. However, with a range of 30 to 70 % virgin powder, the cost and long term powder quality can vary greatly.

61 Table 5.1: The suggested refresh rates for four SLS powders manufactured by EOS GmbH or 3D systems. The range is from 30 to 70 virgin powder.

Manufacturer EOS GmbH 3D Systems

Product Name

Refresh Rate

PA2200 PA3200 DuraForm GF DuraForm

30 to 50 50 to 70 30 + 30 (overflow) 50 +

A thought experiment can highlight the dramatic effects of different blending regimes. For this experiment, it is assumed that SLS grade powder costs $100 per kilo, 15 % of the powder is consumed during a build, 85 % of the powder is recovered, the build chamber is 30 x 30 x 30 cm3 and the density of the powder is 0.46 g/cm3. Therefore one build chamber holds approximately 12.5 kg and a build would consume 1.875 kg or 15 % of 12.5 kg. With a build chamber of this size, it would cost $1250 for sufficient powder for the initial build with virgin powder.

For the thought experiment, refresh rates of 15 % to 65 % will be considered at intervals of 10 % given that 15 % was the minimum refresh rate as this amount of powder is required for a subsequent build to compensate for the 15 % consumed in the prior build. When looking at the powder used in subsequent builds, it has be categorized into one of three groups. The first group considered high quality powder will be virgin powder and powder used in 2 or fewer prior builds based on the work of Dotchev et al. Powder that was virgin at the start of the prior 3 or 4 builds will be the

62 midgrade material. The final group will be the virgin powder that has been used in the previous 5 or more builds.

100%

$1,000

75%

$750

50%

$500

25%

$250

0%

$0 15%

25% 2 or less

35% 3 to 4

45% 5 or more

55%

65% Cost per Build

Figure 5.2: Effects of Refresh Rate for the 5th build

When looking at the 6th build, Figure 5.2, the amount of high quality powder is low if refreshing at 15 or 25 %, while when refreshing at a rate of 35 or 45 %, over half of the powder is relatively new. Lastly, when refreshing at 55 or 65 %, there is almost no powder that has been through the build cycle more than 5 times.

For the subsequent builds, i.e. builds using recovered powder, the price per build ranges from $187.50 to $812.50, compared to the $1250 for the initial build with virgin

63 powder. While the increase is significant, more importantly the ratio of high quality to medium to low quality powder changes dramatically.

Looking more closely at the 45 % to 65 % refresh rate where greater than 75 % of the powder is of higher quality, it is important to note that there is a significant change in cost per build. For the 45 % refresh rate, the cost of each subsequent build is $562.50 while for 65 % the cost is $812.50. This an increase of $250 dollars or 44 %. This is a nontrivial price difference yet powder manufacturers suggest that the refresh rate should fall somewhere in that range. This adds significant cost if unnecessary high ratio virgin powder is blended with recovered powder.

In conclusion, if the cost of SLS is to decrease, one major area that must be improved upon is the science behind the recycling of recovered powder. This simple thought experiment has demonstrated that staying more precisely within manufacturer’s suggestions can cut the raw material cost by more than half.

64

6 METHODOLOGY

6.1. Introduction

The experimental work can be subdivided into three major areas, which are the:

1. Controlled thermal degradation, aging, of the PA12 under conditions that emulate the DTM Sinterstation 2500 SLS machine. 2. Refreshing aged powder with virgin powder 3. Quantification of the powder quality

Figure 6.1: The experimental workflow

The characterization can be split into two subgroups, based on if the results are dependent on molecular scale (DSC, MFI and TGA) or macroscopic scale properties

65 (Repose Angle, Pourability and Apparent Density) of the powder. As previous work has focused heavily on MFI, see §3.4, this was one of the critical measurements to compare this work with the literature. However, due to recent advances in DSC sensitivity, this technique was used extensively. Powder flowability and packing were explored in a cursory manner as the literature contained limited information.

6.2. Materials

All aging experiments were conducted with commercially available PA12, PA650 Nylon 12 Laser Sintering Material manufactured by Advanced Laser Systems (ALM), LLC (Temple, Texas). The powder has a bulk density of 0.46 g/cm3, average particle size of 55 µm, a particle size range (D10 to D90) of 30 to 100 µm and a sintered density of 1.02 g/cm3. The listed Tm is 181 °C and a melt flow rate (MFR) of 50 g/10 minutes at 235 °C, 5.0kg and pre-heated for 3 minutes (the die size was not specified)57.

6.3. Controlled Aging

As recovering powder from builds introduces a multitude of variable into the thermal history of the powder due to significant temperature gradients and varied build times, it is not practical, economical or beneficial to age the powder within the SLS machine when studying the aging process. To emulate the build chamber of a DTM Sinterstation® 2500 a Lindberg/Blue M™ Vacuum Oven manufactured by Thermo

66 Fisher Scientific (Waltham, Massachusetts) with digital temperature controller and a nitrogen purge system was used. Industrial grade nitrogen supplied by Airgas (Radnor, Pennsylvania) was used to purge the oven since the Sinterstation® 2500 purges the build chamber with nitrogen and will not operate if the oxygen concentration is greater than 5 %. No more than 100 g of powder was aged in a 1750 mL Pyrex® drying trays at 80, 100 and 170 °C for either 2,4,6,8,12, 24, 48, 96 or 192 hours (1/12, 1/6, 1/4, 1/3, 1/2, 1, 2, 4, 8 days). After aging, the powder was stored under a nitrogen atmosphere at less than 5 °C in HDPE Nalgene® bottles to minimize postaging degradation.

6.4. Recycling Scheme – Blending

Given that both the manufactures and the literature has strongly emphasized the importance of refreshing degraded powder with virgin powder the effects of this addition will be investigated. Specifically by adding 20 %, 40 % or 60 % virgin powder to the powder aged at 170 °C for 0 to 96 hours. Preliminary results showed that the effects of aging were significantly greater at 170 °C, so to observe the trends due to blending this temperature subset was selected. To ensure mixing, a minimum of 5 g of refreshed powder was prepared by vigorously shaking the powder in a small Nalgene bottle for 60 seconds. Mixing was not a concern as the virgin and aged powder should mix well according to the work of Venables et al on the mixing of powders58.

67 As recycling is a repetitive process that occurs at the end of each build, the blended powder was also aged an additional 12, 24 or 48 hours at 170 °C to emulate the degradation during a print using refreshed powder

6.5. Thermal Analysis

The crux of the SLS process is melting the PA12 with the laser to sinter the crosssections that ultimately will combine to create the part but at the same time, not melt any additional powder. Both DSC and TGA were used to fully characterize the thermal properties.

6.5.1. Differential Scanning Calorimetry

6.5.1.1. Degradation Analysis

It is well know that the melting temperature of the PA12 powder increases due to degradation over time but significantly more information can be gleaned from the DSC data. All samples were analyzed using the NETZSCH DSC 214 Polyma® with sealed Concavus® crucibles. Samples were heated from 30 °C to 200 °C at 5 K/minute under a continuously purged (20 mL/min) nitrogen (UHP Nitrogen, Air Gas) atmosphere. The NETZSCH Proteus® 7.0.1 analysis software was used to determine Tm,onset, Tm,peak, Tm,end, ΔHmelting. From Tm,onset and Tm,end the melting temperature range was determined. All samples were tested in triplicate.

68 6.5.1.2. Recycling Scheme – Blending

To determine the effect of adding virgin powder to the aged powder, sufficient material was mixed to yield at least 5 grams of either 20, 40 or 60 % virgin to aged powder mixtures. The DSC analysis as described in §6.5.1.1 was repeated for these samples.

6.5.2. Thermogravimetric Analysis

All samples were analyzed using the NETZSCH TG 209 F1 Libra®. Samples of approximately 5 mg were heated under a nitrogen atmosphere (Purge Gas: 20 mL/min of Ultra High Purity (UHP) N2; Protective Gas: 20 mL/min of UHP N2) in an alumina crucible to 600 °C at 20 K/min. Mass loss as a function of temperature was recorded by the internal microbalance with a resolution of 0.0001 mg. All samples were tested in triplicate. The data was analyzed using the NETZSCH Proteus® 6.1.0B.

6.6. Melt Flow Index (MFI)

As the recycling and consolidation literature has suggested that the increase viscosity of the melt can be indicative of degraded powder and result in poor consolidation, the MFR of the powder was determined6,7,16,34,35,38. The MFR was measured in triplicate for all aged samples in accordance with ASTM D1238-13, Procedure C59. In short, the powder is poured into a vertical barrel, which is at 235 °C. The polymer melts and then equilibrates at this temperature for 3 minutes. The polymer melt is then forced through

69 a die with a diameter of 2.096 +/- 0.005 mm by a 5.0 kg weight placed upon a rod, which is slightly smaller than the barrel. The mass of the extrudate, the time to extruder that mass, and the displacement of the rod are then used to calculate the MFR and MVR.

6.7. Powder Flow Analysis

There are two ASTM standards used to quantify the followability of a polymer powder. Standard D1895-96 for the pourability and the apparent density, which determines the rate at which the powder flows through a funnel and the amount of powder required to fill a fixed volume, respectively. Standard C1444 measures the angle of repose for a 1.5 inch tall powder pile created by pouring powder through the same funnel used in D1895-96.

6.7.1. Pourability

Comparable to the hourglass, which measures time by particle flow through a funnel, the ASTM standard determines the pourability of a powder by measuring the elapsed time for powder to flow through a standardized funnel. The funnel was manufactured by Plastic Consulting, Inc (Palm City, Florida) was used in accordance with the ASTM D1895-9660. The dimensions of the funnel are provided in Figure 6.2.

70

Figure 6.2: A schematic of the funnel for ASTM D1895-96, the standard for measuring powder pourability and angle of repose

6.7.2. Apparent Density

The apparent density was determined per ASTM D1895-96, Test Method A. In this test, the powder flows through the funnel, which is placed above a 100 mL beaker (3.99 cm in diameter, 7.98cm tall) until full. At which point mass of the powder within the beaker was weighed, from which the apparent density was calculated.

71 6.7.3. Angle of Repose

The powder is allowed to fall from the funnel onto a surface normal to the funnel flow direction. The angle of the powder pile relative to this surface is then measured to determine the angle of repose. The greater the angle the poorer the powder flows. The tests were conducted in accordance with ASTM C 1444b, except instead of using a glassy paper for the surface, this sheet was replaced with a sheet of printer paper60. The angle was determined in accordance to the standard by passing sufficient powder through the funnel until the powder cone was 1.5 inches (3.81 cm) tall. The radius of the base of the cone was determined at 4 equispaced predetermined points using concentric circles printed on the sheet of paper.

b

ASTM C 1444 was withdrawn from the standards but due to a lack of application in industry even though the test yields valuable information.

72

7 RESULTS

7.1. Thermal Analysis

A wealth of information was gleaned from the thermal analysis, in particular the DSC data. The Netzsch DSC 214 Polyma® was sensitive enough to reproducibly detect changes in the thermal properties, specifically the melting temperature, to differentiate powders based on the degree of aging.

7.1.1. DSC – Aged Powder

The DSC thermograms for the aged powder were analyzed to determine the Tm,onset, Tm,peak, Tm,end and ΔHf for the melting endotherm. The results shown in Figure 7.1, 180 °C

175 °C

170 °C 80 °C Virgin

100 °C 12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.1: Tm,onset for the aged powder. Error Bars: +/- 1 Standard Deviation (SD)

73 Figure 7.2, Figure 7.3 suggest that the Tm,onset does not change when the powder is aged. While Tm,peak and Tm,end are both dependent on the time at 170 °C but not at 80 °C and 100°C. 190 °C

185 °C

180 °C 80 °C Virgin

100 °C 12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.2: The Tm,peak for the aged powder. Error Bars: +/- 1 SD

190 °C

185 °C

180 °C 80 °C Virgin

100 °C 12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.3: The Tm,end for the aged powder. Error Bars: +/- 1 SD

74 105 J/g

100 J/g

95 J/g

90 J/g

85 J/g 80 °C

100 °C Virgin

12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.4: The ΔHm for the aged samples. Error Bars: +/- 1 SD

The most important trend is the increasing Tm,peak and Tm,end for powder aged at 170 °C for 0 to 96 hours but also the decrease for powder aged for 192 hours. Given the unchanged Tm,onset and the increasing Tm,end the metling range, ΔTm increased as a function of aging for the powder stored at 170 °C, Figure 7.5.

These results are of significant importance for the SLS process since narrow and known melting temperature range is necessary for successful consolidation of the powder as the laser power is finely tuned16. With this deviation in the end of the melting endotherm, it could be possible that the aged powder would not melt completely. If the powder does not melt, consolidation would be incomplete. This could have serious

75 consequence as Zarringhalam has shown that incomplete sintering is a known issue that decrease part quality21,22,27.

15 K

10 K

5K

0K Virgin

12 h

24 h

48 h

96 h

192 h

170 °C

Figure 7.5: The ΔTm for samples aged at 170 °C. Error Bars: +/- 1 SD

This novel result agrees with the results of previous work by Choren, Gornet and others; however, only the increase of peak melting temperature was observed, which in and of itself is important. However, it is much more significant that the melting temperature range onset remains constant yet the end increases as a function of powder aged. In combination with the small increase in the ΔHmelting, the energy required to melt the powder increases because more energy is required melt the powder and the final temperature that the powder must reach is higher.

76 Traditionally this requirement for more energy has either been neglected, which would result in partial sintering, or the laser power was increased arbitrarily when recycled powder was used. From these results, a model to predict laser power based on the melting endotherm would be the next logical step in improving the recycling of PA12 powder.

A very simple equation could be developed to account for the increase in necessary energy. For example

𝑃 = 𝛼 ∙ 𝑃0

𝑇𝑚,𝑒𝑛𝑑 𝑇𝑚,𝑜𝑛𝑠𝑒𝑡

(7.1)

where P is the laser power in watts, P0 is the default laser power, 12.0 W, and α is an empirically determined correction factor that would intrinsically account for the increase of ΔTm. As α would have to be determined empirically, there is significant work that is required; however, once this correction factor is known, the laser power input would no longer be a parameter typically determined by operator hunches. In addition, with modern DSC instrumentation with autosamplers and automatic data analysis, minimal time would be required to obtain the necessary data. Even with industrial DSC time costing between $50 and $100 per hour, the test would cost $50. This cost is trifling compared to the thousands of dollars of wasted material and opportunity cost lost due to repeating a SLS print job because the laser power was

77 incorrect. Even the capital investment of purchasing a DSC could be justified for manufacturers running the multiple SLS machines.

7.1.2. TGA – Aged Powder

Unlike the DSC results that showed an obvious trends for all three of the aging temperatures, the TGA results showed no explicit trends as a function of time at elevated temperatures. As the powder is neat PA12, there was a single mass loss event, for which the onset temperature and the inflection temperature was determined. The inflection temperature is defined as the temperature at which the first derivative of the mass is the greatest—or in other words, the temperature at which the material lost mass at the highest rate.

The decomposition onset temperature for the aged powder showed no significant change for the powders aged at 80 or 100 °C, Figure 7.6. There is a possible decreasing trend for the powder aged at 170 °C, however, the TGA analysis did exhibit great reproducibility, which leads to uncertainty in this trend. Similarly the decomposition inflection temperature showed no trends either, Figure 7.7. Therefore TGA is not sensitive enough to detect the small changes in the molecular structure and is not a technique suitable for quantifying the degree of degradation for PA12 powders.

Decomposition Onset Temperature

78 420 °C

410 °C

400 °C 80 °C

100 °C Virgin

12 h

24 h

48 h

170 °C 96 h

192 h

Decomposition Infection Temperature

Figure 7.6: TGA results, the decomposition onset temperature of aged powder

445 °C

435 °C

425 °C 80 °C

100 °C Virgin

12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.7: TGA results, the decomposition inflection temperature of aged powder

79 7.1.3. Particle Size Dependent Degradation

Since thermal analysis showed a strong trend as a function of degradation, degraded powder was analyzed after fractionating the powder by particle size using mesh sieves 195 °C

190 °C

185 °C

180 °C Tm,peak 0-53 µm

Tm,end 54-75 µm

76-125 µm

125-212 µm

Figure 7.8: The Tm,peak and Tm,end of degraded powder as a function of particle size.

(53, 75, 125 and 212 µm). Since particle size does not increase significantly during oven aging (see §3.4.4), powder was recovered from the part cake of a random build. As expected, the peak melting temperature of the fractions was greater than virgin powder; however, Tm,peak was also dependent on particle size. The results showed degradation was greatest for the smallest particles and the least for the largest particles.

These results suggest that when recovering powder there may be value in removing the smaller particles and not just the largest particles, which is the current paradigm. This would remove the most degraded powder, unfortunately the small particles are

80 important for producing dense parts. One solution would be to demand that manufacturers produce commercially available powder of 10 µm to 30 µm so that blending in this virgin powder would decrease the particle size distribution.

The study did not include further tests to determine the cause for this trend; however, the small particles have a higher surface to volume ratio. This could suggest that observed degradation is no occurring the bulk but rather at a higher rate on the surface of the particles. On untested hypothesis is that thermal-oxidative degradation is greater for the small particles, as oxygen present at higher concentrations at the surface. Polymers have a low thermal conductivity compared to other materials, e.g. metals; however, particles, even the ones at the upper range of the size distribution, are small and should reach temperature equilibrium quickly.

7.2. Recycling

7.2.1. Recycling Scheme – Blending

The same thermal analysis was completed for powder aged at 170 °C for between 0 and 96 hours when blended with virgin powder at loading rates of 20 %, 40 % and 60 %. As expected from the results in §7.1.1, the addition of virgin powder decreased the Tm,peak and Tm,end of the recovered powder but left the Tm,onset unchanged. The Tm,end results are shown in Figure 7.9, Figure 7.10, and Tm,peak are shown in Figure 7.11 and Figure 7.12.

81 189 °C

187 °C

185 °C 12 h

24 h 0%

48 h 20 %

40 %

96 h

60 %

Figure 7.9: The Tm,end for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by powder age. 189 °C

187 °C

185 °C 0%

20 % 12 h

40 % 24 h

48 h

60 %

96 h

Figure 7.10: The Tm,end for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by refresh rate.

82 188 °C

186 °C

184 °C

182 °C 12 h

24 h 0%

20 %

48 h 40 %

96 h

60 %

Figure 7.11: The Tm,peak for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by powder age.

188 °C

186 °C

184 °C

182 °C 0%

20 % 12 h

24 h

40 % 48 h

60 %

96 h

Figure 7.12: The Tm,peak for powder aged from 0 to 96 hours, refreshed at 0 to 60 %, clustered by refresh rate.

83 These trends could be one of the reasons that the addition of virgin powder to recovered material improves the final part quality when using recycled powder. This is significant as the DSC was able to reproducibly distinguish the quality of the powder.

The lower Tm,end should improve consolidation at the same laser power and part bed surface temperature and therefore issues caused by incomplete sintering would be mitigated. The relationship between fresh rate and the decrease of Tm,peak and Tm,end could also be a potential avenue for determining the amount of virgin powder required to refresh the recovered powder. One could imagine measuring the Tm,peak and Tm,end of the recovered powder, from which the required percent of virgin powder could be determined. Equation 7.1. A dimensionless refresh rate could be described by:

(7.1) where R is the dimensionless refresh rate in percent, β is an empirically determined scaling factor.

The use of DSC would also be very practical in an industrial setting as the amount of material and the time to complete the test are minimal. Only a few milligrams of powder is required and the test can be completed in less than 30 minutes. Additionally, samples could be sent out for testing and shipping costs would be trifling. In conclusion, the DSC analysis is a possible method to reproducibly quantify degradation and classify the quality of the powder.

84

7.3. Melt Flow Index

Before exploring the MFI data, it is important to note that there is a discrepancy between the collected data and the specification sheet provided by ALM for PA650.

3/10 minutes cm minutes g/10

240 200

180 150

120 100

60 50

0 0

80 80 °C °C

100 100 °C °C Virgin Virgin

12 12 h h

24 24 h h

48 48 h h

170 170 °C °C 96 96 h h

192 192 h h

Figure Rate (MVR) aged powder, clusted by by aging aging temperature. temperature. Error Error Figure7.14: 7.13:The TheMelt MeltVolume Flow Rate (MFR) of of thethe aged powder, clustered Bars: +/1 SD Bars: +/- 1 SD.

The data sheet reports an MFI of 50 g/10 minutes for the same test conditions, 3 minute preheat at 235 °C with a 5.0 kg mass, which is approximately half of MFR. One explanation for this could be that the ASTM Standard, D1238, specifies two different diameters for the die 2.095 mm and 1.048 mm for the “half” die. It could be possible that ALM used the half-die prescribed by method C (a method for high melt flow rate polyolefins) since SLS grade PA12 has a high melt flow rate.

The MFR and MVR data suggests that PA650 does not change when aged at 80 or 100 °C and dramatically decreases when aged at 170 °C. The most noteworthy result

85 was the MFR/MVR data for 170 °C, as the MFR/MVR decreases by half after 12 hours of aging but did not change significantly from 12 hours to 96 hours. In addition, these results do not agree well with the data presented by Dotchev et al in 2009 that showed a stepwise decrease in MFR as a function of time and temperature. This is very important as the blending workflow developed from this work and presented in §3.4.3 was solely based on the observed decrease in MFR.

There are numerous possible explanations for why the MFR decreases. Ockam’s razor would suggest that this difference is due to the fact Dotchev was using PA2200 manufactured by EOS Manufacturing Solutions. In addition, Dotchev published this data in 2009 and SLS powder technology has surely advanced, even if the manufactured have not disclosed these advances. The 50 % drop of MFR shown for the PA2200 aged at 100 °C could suggest that PA650 is modified to improve thermal stability or PA2200 is less stable. Improved stability could be due to end group modification or the additives, while decrease stability could be due to residual catalyst. These are obviously only rash hypothesis and by no means the answer; however,

86 these differences should raise questions about what exactly the manufacturers of SLS powder are doing to improve thermal stability of the PA12.

In light of these results, there is even greater need to consider DSC as a possible replacement for measuring powder quality as it is a more sensitive technique. The MFI results showed no difference between powder aged at 170 °C for between 12 and 96 hours. In contrast, the DSC data showed a stepwise increase for the same powder for each time step.

Figure 7.15: The MFR results presented by Dotchev in 20097.

87

7.4. Powder Flowability

Of the three flowability tests (angle of repose, pourability and bulk density) the only results that hinted a trend as a function of the degree of aging came from the angle of repose data, while pourability and bulk density could only distinguish between virgin and powder aged at 170 °C.

The angle of repose results showed an interesting trend for all three test temperatures. For this data, an increase of the angle indicates a decrease in flowability and a decrease in angle indicates improved flowability.

For 80 °C, there appears to be a slight increasing trend but it is not significant. For 100 °C the angle first increases and then decreases to almost the original value, while for 170 °C there is a clear increasing trend. On frustrating aspect of the data is the large standard deviation, which leads to great uncertainty in the results. However, the results strongly suggest that angle of repose could be a valuable measure as all 15 aged samples had a greater angle of repose than virgin powder. This suggests thermal degradation is decreasing the flowability of the powder.

88

65

55

Angle of Repose (Degrees)

38°

36°

34°

32° 80 °C

45

100 °C

80 °C Virgin Virgin

12 h 12 h

100 °C 24 h 48 h 24 h 48 h

170 °C 170 °C 96 h 96 h

192 h 192 h

Figure 7.16: The Angle of Repose as of the aged powder, clustered by aging temperature. Error Bars: +/-1 SD

Unfortunately the literature has not explored the effects of powder flowability in great detail, so it is not known if decreased flowability is the cause for inferior parts when recycled powder is used; however, the results indicate further could yield valuable information. This only further reinforces the notation that much can still be learned about the fundamentals of powder degradation.

Unlike the angle of repose, the pourability shows no trend for 80 and 100 °C. The pourability was sensitive to the thermal degradation at 170 °C, unfortunately the results showed a decrease in pourability for 12 to 48 hours and then significant decrease that resulted in no flow through the funnel. While promising, pourability

89 cannot distinguish between 12 h and 48 h at 170 °C, which limits the usefulness of this test.

Apparent Density (g/cm3)

0.50

0.40

0.30 80 °C

100 °C Virgin

12 h

24 h

48 h

170 °C 96 h

192 h

Figure 7.17: Apparent density of the aged powder, clustered by aging temperature. Error Bars: +/- 1 SD.

The least useful of the 3 tests, apparent density showed no change until 96 h at 170 °C. It was reassuring that the apparent density agreed with the specification sheet provided by ALM for PA650. The only practical data the apparent density could provide would be that the powder is significantly degraded but no information between virgin and that point.

The decrease in apparent density and the lack of pourability for the samples aged for 96 and 192 hours could be related to thermal-oxidative degradation as these two samples exhibited yellowing, commonly associated with oxidation of polymers, especially polyamides. This is important to note, as these samples were both

90 degraded by thermal and thermal-oxidative mechanisms. This suggests that apparent density and/or pourability could be valuable test if thermal-oxidation is suspected.

In conclusion the results suggest that angle of repose should be explored in more depth, especially since the apparatus used could be modified to improve the results. For the current apparatus, the powder falls upon a flat sheet of paper, which resulted in repose angles of between 32° and 38°. It may be possible to redesign this surface to better elucidate out slight changes in the powder flowability. In addition, the work of Schmid et al that used the rotating cylinder to measure the dynamic angle of repose, see §2.7.3, should be explored for aged powders.

91

8 CONCLUSIONS AND FUTURE WORK

8.1. Conclusions

The work in this study built off of and elaborated upon the work of other research groups from multiple university. The results both verified and expanded the results of previous work on the degradation and recycling of SLS grade PA12 powder. The goal was to explore at the problem from a different perspective, primarily thermal analysis to question to the widely held belief that DSC was not sensitive enough to detect degradation. In addition to thermal analysis, the effect of refreshing powder was explored from the thermal analysis perspective. The results showed that refreshing at rates of 20, 40 and 60 % changed the properties of the aged powder to varying degrees.

One of the most promising results was the ability to detect minor changes in the thermal properties with the Netzsch DSC 214 Polyma®, even though the literature review suggested DSC was not sensitive enough to detect degradation. In stark contrast, the results showed that DSC is capable of quantifying degradation and is even sensitive enough to distinguish between powder that has been refreshed at either 20, 40 or 60 %. This could hold significant importance, as unlike other methods such as MFI or the pourability, which require 10-50 g of powder, DSC only requires a few milligrams. In addition, MFI, which has been touted in the literature as great

92 method to determine degradation is a destructive technique. Assuming 50 g of powder is requited for MFI, per the ASTM standard, the test would require $5 of powder. In contrast, flowability testing is nondestructive and the 5-10 mg of powder destroyed in a DSC sample has a value of essentially zero, $0.001. Yes, DSC instrumentation is more expensive but also less labor intensive, so testing costs should be comparable.

The results of the MFI were disappointing, as it does not appear that the method can differentiate powder aged for 12 hours at 170 °C and 96 hours at 170 °C. This is significant because the thermal analysis strongly suggests a change in powder quality.

TGA should be eliminated as a possible technique to quantify degradation of PA12 powder as there was no evident change in the decomposition temperature or rate of decomposition as a function of powder degradation.

Considering that only DSC and angle of repose showed any trends as a function of aging temperature and time, the results from these two could be considered for inclusion. Building off of Equation 7.1, the inclusion of angle of repose could yield this equation

(8.1)

93 where D is the dimensionless number, A is the angle of repose converted into radians, and γ is an empirically determined scaling factor. The dimensionless number, D, would be in percent and be the percent virgin powder required to refresh the powder.

8.2. Future Work

The results strongly suggest that future work is necessary if recycling efficiency of PA12 powder is to be maximized, especially given the novel results. The most salient work would be to build samples with oven aged powder to correlate the quality of parts to the values provided by both the flowability and thermal analysis. Given the sensitivity of the DSC, it should be possible to predict the quality of parts depending on the Tm,peak, Tm,end and the ΔTm. In addition, other quantification techniques should be investigated to determine the sensitive for detection of degradation. For example, the patent literature is a proponent of solution viscometry to detect changes in average MW. However, all of these techniques, DSC, MFI, solution viscometry, indirectly measure changes in the MWD. To truly investigate the subtle changes that these techniques pick up, gel permeation chromatograph, which measures the MWD, of aged powder is a vital next study.

The use of dimensionless numbers could also be a valuable addition to the recycling literature. Equations 7.1 and 8.1 are two examples of possible dimensionless numbers that would build off of the work presented in this thesis. However, both of these

94 dimensionless numbers would require a scaling factor that could only be determined empirically. Therefore, it would be necessary to build off of the novel results in this work. If parts could be built using powder aged in a controlled environment, it would be possible to determine if there is a correlation between the quality and strength of printed parts to the values such as Tm,peak, Tm,end, angle of repose, etc. The scaling factors could be determine by plotting a property such as tensile strength versus Tm,peak.

In conclusion, DSC was shown as a novel and very sensitive technique to quantify changes in the SLS powder when the powder was thermally degraded. MFI did not show useful results in this study but did in the work of others. Angle of repose showed promising results and could be the basis of future techniques to quantify the macroscopic quality of the powder.

95

9 REFERENCES (1)

Rifkin, J. The Third Industrial Revolution: How Lateral Power Is Transforming Energy, the Economy, and the World; St. Martin’s Press, 2011.

(2)

Rifkin, J. 2011, p 270.

(3)

Berman, B. Bus. Horiz. 2012, 55 (2), 155–162.

(4)

ASTM International. 2013, F2792-12a, 10–12.

(5)

Goodridge, R. D.; Tuck, C. J.; Hague, R. J. M. Prog. Mater. Sci. 2012, 57 (2), 229–267.

(6)

Pham, D. T.; Dotchev, K.; Yusof, W. a. Y. 2008.

(7)

Dotchev, K.; Yusoff, W. Rapid Prototyp. J. 2009, 15 (3), 192–203.

(8)

Evans, S. Fire in small German town could curb world car production http://www.bbc.com/news/business-17769466 (accessed Jan 1, 2015).

(9)

Smock,

D.

Lessons

from

the

Nylon

12

Supply

Disruption

http://www.mypurchasingcenter.com/commodities/commoditiesarticles/lessons-learned-from-nylon-12-supply-disruption/ (accessed Jan 1, 2015).

96 (10) Ullmann’s Fine Chemicals, 3 Volume Set; Wiley, 2014.

(11) Frisch, K.C., Lie, G., Axelrood, S.L., & Mich, T. 1965, No. 2, 3–7.

(12) Evonik Industries. An Altern. raw Mater. Polyam. 12 acid Evonik is Oper. a Pilot plant bio-based ω -amino lauric 2013.

(13) Rosenzweig, N.; Narkis, M. Polymer Powder Technology; J. Wiley, 1995.

(14) Aquite, W. A Novel Polymer Extrusion Micropelletization Process, Univesity of Wisconsin-Madison, 2015.

(15) Evonik Industries. VESTOSINT ® Polyam. 12 Coat. Powders.

(16) Kruth, J. P.; Levy, G.; Klocke, F.; Childs, T. H. C. CIRP Ann. - Manuf. Technol. 2007, 56 (2), 730–759.

(17) Freeman

Technologies.

Why

are

http://www.freemantech.co.uk/_powders/?pt=Why

Powders are

Powders

Complex? Complex

Page&p=NA==.

(18) Amado, A.; Schmid, M.; Wegener, K. Rapid.Tech 2014, 1–7.

(19) Schmid, M.; Amado, F.; Levy, G.; Wegener, K. Proc. 6th Int. Res. Virtual Rapid Prototyp. 2014.

97 (20) Davis, J. R.; Committee, A. S. M. I. T. S. S. T. Handbook of Thermal Spray Technology; ASM International, 2004.

(21) Majewski, C. E.; Zarringhalam, H.; Hopkinson, N. Proc. SFF Symp. 2008, 45– 54.

(22) Zarringhalam, H.; Majewski, C.; Hopkinson, N. Rapid Prototyp. J. 2009, 15 (2), 126–132.

(23) Shi, Y.; Li, Z.; Sun, H.; Huang, S.; Zeng, F. Proc. I MECH E Part L J. Mater. Appl. 2004, 218 (3), 247–252.

(24) Flory, P. J. J. Chem. Phys. 1947, 15 (6), 397–408.

(25) Hiemenz, P. C.; Lodge, T. P. Polymer Chemistry, Second Edition; Taylor & Francis, 2007.

(26) Osswald, T. A.; Menges, G. Materials Science of Polymers for Engineers; Hanser Publications, 2012.

(27) Zarringhalam, H.; Hopkinson, N.; Kamperman, N. F.; de Vlieger, J. J. Mater. Sci. Eng. A 2006, 435-436, 172–180.

(28) Wohler Associates. Wohlers Report 2009; 2009.

98 (29) Du Plessis, P. In Proceedings of the TCT Conference; CRP Group: Coventry, 2008.

(30) Deckard, C. R. Selective laser sintering, University of Texas - Austin, 1988.

(31) Ehrenstein, G. W.; Pongratz, S. Resistance and Stability of Polymers; Carl Hanser Verlag GmbH & Co. KG: München, 2013.

(32) Beaman, J. J.; Barlow, J. W.; Bourell, D. L.; Crawford, R. H.; Marcus, H. L.; McAlea, K. P. Solid Freeform Fabrication: A New Direction in Manufacturing: with Research and Applications in Thermal Laser Processing; Springer US, 2013.

(33) Forderhase, P.; McAlea, K. Proc. SFF Symp. 1994, 102–109.

(34) Choren, J.; Gervasi, V.; Herman, T.; Kamara, S.; Mitchell, J. Solid Free. Fabr. Symp. 2001, 39–45.

(35) Gornet, T. J.; Davis, K. R.; Starr, T. L.; Mulloy, K. M. Solid Free. Fabr. Symp. 2002, 546–553.

(36) Katalinic, E. B.; Harmony, M.; Technology, B. 2010, 21 (1).

(37) Mielicki, C.; Gronhoff, B.; Wortberg, J. 2014, 728, 728–731.

99 (38) Wudy, K.; Drummer, D.; Kühnlein, F.; Drexler, M. AIP Conf. Proc. 2014, 1593 (May), 691–695.

(39) Ziegelmeier, S.; Wollecke, F.; Tuck, C.; Goodridge, R.; Hague, R. Solid Free. Fabr. Symp. 2013, 354–367.

(40) Amado, A.; Schmid, M.; Levy, G.; Wegener, K. Proc. SFF Symp. 12.

(41) Rietzel, D.; Aquite, W.; Drummer, D.; Osswald, T. a. 2–7.

(42) Krantz, M.; Zhang, H.; Zhu, J. Powder Technol. 2009, 194 (3), 239–245.

(43) Wudy, K.; Drummer, D. Proc. PPS-29 2014.

(44) Drummer, D.; Wudy, K.; Drexler, M. Proc. PPS-30 2015.

(45) Allen, N. S. Polym. Degrad. Stab. 1984, 8, 55–62.

(46) Graves, C. T.; Ormerod, M. G. Polymer (Guildf). 1963, 4, 81–91. (47) Brodskii, A. I.; Fomenko, A. S.; Abramova, T. M.; Furman, E. G.; Dar’eva, E. P.; Kukhtenko, I. I. In Dokl. Akad. Nauk SSSR; 1964; Vol. 156, pp 1147–1149.

(48) Osswald, T. A.; Menges, G. Material Science of Polymers for Engineers; Hanser Publications: Cincinatti, Ohio, 2012.

100 (49) Osswald, T.; Rudolph, N. Polymer Rheology, 1st ed.; Strohm, C., Ed.; Hanser Publications: Cincinatti, Ohio, 2014.

(50) Petzold, S.; Smet, J.; Osswald, T. A. SPE ANTEC 2015 2015.

(51) Selective

Laser

Sintering,

Birth

of

an

Industry

http://www.me.utexas.edu/news/2012/0712_sls_history.php (accessed Nov 6, 2015).

(52) Monsheimer, S.; Simon, U.; Grebe, M.; Christoph, W.; Altkemper, S. Powder with Improved Recycling Properties, Process for Its Production, and use of the Powder in a Process for Producing Three-Dimensional Objects. US 2006/0071359 A1, 2006.

(53) Monsheimer, S.; Grebe, M.; Altkemper, S.; Baumann, F.-E.; Hessel-Geldmann, S.; Warnke, K.; Stemmer, H.; Muegge, J.; Tuerke, M.; Simon, U. Copolyamide Powder and Its Preperation, Use of Copolyamide Powder in a Shaping Process and Mouldings Produced from this Copolyamide Powder. US 8,591,797 B2, 2013.

(54) Pfister, A.; Gersch, M. Method and Device of Layerwise Manufacturing A ThreeDimensional Object of a Powdery Material. US2009/0291305 A1, 2009.

101 (55) Ruffo, M.; Tuck, C.; Hague, R. Proc. Inst. Mech. Eng. Part B J. Eng. Manuf. 2006, 220 (9), 1417–1427.

(56) Osswald, T.; Baur, E.; Brinkmann, S.; Oberbach, K.; Schmachtenberg, E. International Plastics Handbook: The Resource for Plastics Engineers; Hanser, 2006.

(57) Advanced Laser Materials, L. Technical Data Sheet (PA 650); 2011.

(58) Venables, H. J.; Wells, J. I. Drug Dev. Ind. Pharm. 2001, 27 (7), 599–612.

(59) ASTM International. 2014, D1238, 1–16.

(60) ASTM International. 2000, C1444 (1), 1–2.

(61) NETZSCH-Geraetebau GmbH. DSC 214 Polyma; 2014.