Chalmers University of Technology Master’s thesis Binder jetting of aluminum alloys: Design for Additive Manufacturing and Material Characterization Master’s thesis in Product Development and Material Engineering Didarul Alam Omkar Rajan Dessai Department of Industrial and Material Science CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2024 www.chalmers.se www.chalmers.se Master’s thesis 2024 Binder jetting of Aluminum Alloys: Design for Additive Manufacturing and Material Characterization Didarul Alam Omkar Dessai Department of Industrial and Material Sciences Center for additive manufacture (Metal) Chalmers University of Technology Gothenburg, Sweden 2024 Binder jetting of aluminum alloys: Design for Additive Manufacturing and Material characterization © Didarul Alam, 2024. © Omkar Rajan Dessai, 2024. Supervisor at Volvo Car Corporation: Torbjörn Larsson Supervisor at Chalmers: Anok Babu Nagaram and Eduard Hryha Examiner at Chalmers: Eduard Hryha Master’s Thesis 2024 Department of Industrial and Material Science Center for Additive Manufacturing (Metal) - CAM2 Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 The thesis work was performed in the framework of the Centre for Additive Manufacturing- Metal (CAM2) in collaboration with Volvo Car Corporation, Gothenburg, Sweden and RICOH Company LTD. Printed by Chalmers Digital Printing Gothenburg, Sweden 2024 iv Abstract Binder jetting technology (BJT) is an additive manufacturing (AM) process wherein a liquid binder is deposited on the power bed to join the particles together layer by layer, thus creating the desired geometry. This process can print various metals and polymers with higher productivity. Even so, among other metals, aluminum alloys are not commercially available to be produced by BJT. However, due to aluminum’s intrinsic characteristics and properties, the passenger and commercial vehicle indus- tries extensively utilize this metal. For this reason, it is a highly desired material for BJT/M in the automotive industry due to the possibility of printing complex shapes and having higher productivity than other metal additive manufacturing technologies. A Japanese company, RICOH Company Ltd, is currently working on developing technology for printing aluminum alloys in BJT. In this master thesis, design for AM and material characterization studies were conducted in collabora- tion with Volvo Car Corporation (VCC) and RICOH to study a heat sink for the headlamp for the XC90 production series car against two different aluminum alloys. As-produced aluminum alloy was 4000 series aluminum alloy, which had a Si con- tent of less than 10 wt.%, and high pressure die casted aluminum alloy had a Si content of 12 wt.%. A study of design guidelines was carried out by printing design artefacts, and the geometry evaluation was done using scanning and superimposing CAD data. This provided knowledge about critical design features that can be pro- duced using the technology and the limit. 10mm3 produced cubes were analyzed for porosity using Light Optical Microscopy (LOM). Density verification on the cubes using LOM, Geometrical, and Archimedes showed a relatively high density of 98% with a standard deviation of 0.91. The primary function of the heatsink studied here is dissipating heat away from other components. Thus, thermal conductivity and convection performance were measured. Thermal conductivity for as-sintered material was found to be 154.41 W/mK at room temperature (20 °C), and it was increased to 166.68 W/mK after Hot Isostatic pressing (HIP), compared to 97.94 W/mK for as-casted aluminum alloy which is currently being used. To improve the thermal convection, the new design features, namely Gyroid with varying wall thicknesses, were produced as test pieces and tested in the experimental setup to determine the design with the best performance. Heat dissipation increased by 99% in free convection and 82% in forced convection compared to the as-casted heatsink; this was demonstrated during the experiment. A new heatsink design pro- duced using BJT Al alloys also resulted in a weight reduction of 42% to the existing design. Regarding the mechanical properties, HV5 was found to be 51.91% less for as-sintered aluminum alloy compared to as-casted aluminum alloy and 48.98% less for as-sintered aluminum alloy when HK1 was considered. The microstructure analysis showed that the reason for the thermal conductivity increase and hardness decrease in the aluminum alloy produced by BJT/M is connected to the lower Si content. Hence, the lower fraction of AlSi-precipitates was produced. Keywords: Additive manufacturing, Binder Jetting for Metals, Gyroid Structures, Material Characterisation, Hardness, Thermal conductivity, Thermal convection, Hot Isostatic Pressing v Acknowledgements We would like to extend our heartfelt gratitude to all the individuals who have sup- ported and guided us throughout our journey to completing this master’s thesis. The master’s thesis was carried out at the pilot plant & concept center at Volvo Car Corporation, Gothenburg, Sweden, and the Center for additive manufacturing (CAM2) at Chalmers. This was done as part of a Final degree project for the Master of Science in Product Development and Material Engineering at Chalmers Univer- sity of Technology, Sweden in collaboration with Volvo Car Corporation. First and foremost, we would like to express our sincere appreciation to our indus- trial supervisor, Torbjörn Larsson, for allowing us to work on this research project in a real-world setting. His valuable insights, constant encouragement, and construc- tive feedback were instrumental in shaping the direction of this thesis. We are extremely grateful to our examiner and academic supervisor Eduard Hryha for providing us with all the shared expertise and mentorship in the field of ad- ditive manufacturing and Binder jetting technology for metals in particular. His academic expertise has been pivotal in strengthening the theoretical foundation of this research. We are also deeply indebted to our academic co-supervisor, Anok Babu Nagaram, for his scholarly guidance, expert advice, and unwavering support. His academic expertise has been pivotal in strengthening the theoretical foundation of this research. We would like to take this opportunity to thank Markus Enmark and Alberto Cabo Rio for sharing their experience and practical knowledge that was helpful during the laboratory work and also helped us structure the thesis. We are also deeply indebted to Erika Steyn, Daniel Gren, and Dr. Antonio Mulone for supporting us with the needed training. We would like to express our gratitude to Miyazaki Yoshiyuki and his team at RI- COH as the technology provider company in this project for allowing us to work with their technology in printing various parts. We would also like to extend our appreciation to the faculty members, research peers, and staff who have contributed in various ways to this thesis, including their thoughtful discussions and technical assistance. We acknowledge that this thesis would not have been possible without the collective efforts and support of all these individuals, and we are truly grateful for their contributions. January 29, 2024 Gothenburg, Sweden. Didarul Alam & Omkar Dessai vii viii List of Acronyms Below is the list of acronyms that have been used throughout this thesis listed below: AM Additive Manufacturing PBF/LB Powder Bed Fusion - Laser Beam PBF/EB Powder Bed Fusion - Electron Beam BJT Binder Jetting Technology HPDC High Pressure Die Casting AlSi Aluminum Silicate SS Solid State Sintering LPS Liquid Phase Sintering SLPS Supersolidus Liquid Phase Sintering MG Microgrid PV Photovoltaic RES Renewable-based Energy Sources LOM Light Optical Microscopy HIP Hot Isostatic Pressing ix Contents Contents xi List of Figures xiii List of Tables 1 1 Introduction 3 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Aim and research questions . . . . . . . . . . . . . . . . . . . . . . . 4 2 Literature Study 5 2.1 Binder jetting for metal . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Debinding process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.3 Sintering process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.3.1 Liquid phase sintering . . . . . . . . . . . . . . . . . . . . . . 8 2.4 Sintering of aluminum alloys . . . . . . . . . . . . . . . . . . . . . . . 10 2.5 Effects of silicon on thermal conductivity of Al-Si alloys . . . . . . . . 11 2.6 Heat exchange performance . . . . . . . . . . . . . . . . . . . . . . . 13 2.7 Design guidelines for Binder jetting for metal . . . . . . . . . . . . . . 14 3 Methods 19 3.1 Design for Binder jetting technology for metal . . . . . . . . . . . . . 19 3.2 Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.3 Density measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.4 Dimensional and geometrical inspections . . . . . . . . . . . . . . . . 21 3.4.1 Metallography . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.5 Mechanical properties testing . . . . . . . . . . . . . . . . . . . . . . 23 3.5.1 Hardness test . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.5.2 Thermal conductivity test . . . . . . . . . . . . . . . . . . . . 24 3.5.3 Thermal convection test . . . . . . . . . . . . . . . . . . . . . 28 3.6 Materials characterisation . . . . . . . . . . . . . . . . . . . . . . . . 32 3.6.1 Light optical microscopy . . . . . . . . . . . . . . . . . . . . . 32 3.7 Post-Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 3.7.1 Hot Isostatic Pressing - HIP . . . . . . . . . . . . . . . . . . . 34 4 Results and Discussion 37 4.1 Verifying the design guidelines . . . . . . . . . . . . . . . . . . . . . . 37 xi 4.1.1 Artefact designs for verifying the guidelines . . . . . . . . . . 37 4.1.2 Results from verifying the guidelines . . . . . . . . . . . . . . 42 4.2 Porosity analysis using CT scanning . . . . . . . . . . . . . . . . . . . 51 4.3 Evaluation of the shrinkage and deflection after sintering . . . . . . . 52 4.3.1 Shrinkage in 10mm3 cubes . . . . . . . . . . . . . . . . . . . . 52 4.3.2 Shrinkage and deflection in gyroids shapes . . . . . . . . . . . 53 4.4 Density measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.5 Comparative porosity analysis using CT scanning and light optical microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 4.6 LOM analysis for as-cast, as-sintered, and HIPed samples . . . . . . . 58 4.6.1 LOM analysis for as-cast Al-12Si-Cu1 . . . . . . . . . . . . . . 58 4.6.2 LOM analysis for as-sintered aluminum 4000 series alloy . . . 59 4.6.3 LOM analysis for HIPed aluminum 4000 series alloy . . . . . . 61 4.7 Mechanical performance . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.7.1 Hardness test . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.7.2 Thermal conductivity test . . . . . . . . . . . . . . . . . . . . 63 4.7.3 Heat exchange performance test . . . . . . . . . . . . . . . . . 68 4.8 Printing final design . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4.8.1 Evaluating of the heatsink produced by BJT/M . . . . . . . . 71 5 Conclusions 73 5.1 Answering the research question . . . . . . . . . . . . . . . . . . . . . 73 5.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 5.3 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Bibliography 77 A Reports: Design Guideline Evaluations i A.1 Artefact 2A: Pins with conventional sintering Process . . . . . . . . . i A.2 Artefact 2B: Pins with ceramic-supported sintering Process . . . . . . v A.3 Artefact 3A: Bridge with no compensation . . . . . . . . . . . . . . . xi A.4 Artefact 3B: Bridge with two spans compensated . . . . . . . . . . . xvii A.5 Artefact 3C: Bridge compensated using Simufact . . . . . . . . . . . . xxiii A.6 Artefact 4A: Bridge with base convectional sintering process . . . . . xxix A.7 Artefact 4B: Bridge with base ceramic-supported sintering process . . xxxviii A.8 Artefact 5: Tree with ceramic-supported sintering . . . . . . . . . . . xlvii A.9 Gyroid having 1mm thickness . . . . . . . . . . . . . . . . . . . . . . liii A.10 Gyroid having 1.5mm thickness . . . . . . . . . . . . . . . . . . . . . lxii A.11 Gyroid having 2mm thickness . . . . . . . . . . . . . . . . . . . . . . lxxi A.12 Gyroid having 2 to 1mm thickness . . . . . . . . . . . . . . . . . . . . lxxx B Reports: Final Heat Sink evaluation lxxxix xii List of Figures 2.1 Typical schematic process diagram of BJT/M [1] . . . . . . . . . . . 5 2.2 Major factors that determine the characteristics of BJT Green Body [1] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.3 Simple schematic diagram of LPS [2] . . . . . . . . . . . . . . . . . . 9 2.4 Schematic diagram of collision of free electrons with eutectic Si con- taining particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.5 Thermal conductivity of binary Al-Si, Al-Cu, Al-Fe, and Al-Mg alloys [3] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.6 Recommended Wall Thickness Printable by MBJ . . . . . . . . . . . 15 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.7 Overhang length and thickness for BJT/M . . . . . . . . . . . . . . . 15 2.8 Powder Removal holes for Shell parts in BJT/M . . . . . . . . . . . . 16 2.9 Radius at corners and sharp edges for BJT/M . . . . . . . . . . . . . 16 2.10 Minimum recommended printable feature by BJT/M . . . . . . . . . 17 2.11 Vertical Pins printable by BJT/M . . . . . . . . . . . . . . . . . . . . 18 3.1 Cubes (green body cube and sintered cube) . . . . . . . . . . . . . . 20 3.2 Build orientation of cubes . . . . . . . . . . . . . . . . . . . . . . . . 20 3.3 Metallography-Sectioning and Mounting . . . . . . . . . . . . . . . . 22 3.4 Metallography-Polishing and Grinding . . . . . . . . . . . . . . . . . 23 3.5 Struers DuraScan for Hardness Test Measrurement . . . . . . . . . . 24 3.6 Laser Flash Technique . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.7 LFA 447 Instrument . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.8 Signal to noise performance between graphite coated and without graphite-coated as-sintered aluminum 4000 series alloy sample . . . . 26 3.9 Visual comparison between different coating techniques . . . . . . . . 27 3.10 Placement and schematic of in-plane diffusivity measurement . . . . . 27 3.11 Test Rig developed to measure the heat exchange performance of the artefact designs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.12 Environmental test chamber housing the assembled test rig with the data acquisition system . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.13 Schemetic representation of calculation for power lost through the device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.14 Schematic illustration of all three convection setups . . . . . . . . . . 32 3.15 Zeiss Axioscope 7 light optical microscope . . . . . . . . . . . . . . . 33 xiii 3.16 Hot Isostatic Pressing schematic . . . . . . . . . . . . . . . . . . . . . 34 3.17 QIH 15L from Quintus Technologies AB . . . . . . . . . . . . . . . . 35 3.18 HIPing parameters utilized in this work. The Y-axis on the left show the temperature and the Y-axis on the right show the pressure. The X-axis shows time in seconds . . . . . . . . . . . . . . . . . . . . . . . 36 4.1 Artefact 1 screw Threads in metric standard(all dimensions in mm) . 38 4.2 Artefact 2 Pins with various heights and diameters (all dimensions in mm) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.3 CAD-designed artefact for bridges(all dimensions in mm) . . . . . . . 39 4.4 CAD-designed artefact for Bridge without base(all dimensions in mm) 39 4.5 CAD design of the artefact for trees(all dimensions in mm) . . . . . . 40 4.6 Image showing the building, recoating, and binder dispensing direc- tion for all the artefact designs . . . . . . . . . . . . . . . . . . . . . . 42 4.7 Shows the threads after bolting. 4.7a refers to conventionally sintered threads and 4.7b refers to ceramic sintered thread . . . . . . . . . . . 42 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.8 Dimension evaluation on the laser-scanned data for pins sintered in unsupported sintering process (Artefact 2A) . . . . . . . . . . . . . . 44 4.9 Dimension evaluation on the laser scanned data for pins sintered in ceramic media (Artefact 2B) . . . . . . . . . . . . . . . . . . . . . . . 45 4.10 Shows the evaluation of the non-compensated vs compensated bridge geometry for gravity using the scanned data . . . . . . . . . . . . . . 46 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.11 Artefact 3C Bridge using Simufact data: 4.11a Shows the deforma- tion predicted by Simufact, 4.11b Shows the compensation geometry provided by Simufact and 4.11c shows the scanned file for the final print using the compensated CAD data from Simufact . . . . . . . . 47 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.12 Image showing evaluation using scanned data for the bridge with base: without and with ceramic support respectively . . . . . . . . . . . . . 48 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 (d) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.13 The outcome using conventional sintering process 4.13aand The out- come using the ceramic-supported sintering process 4.13b . . . . . . . 49 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 xiv 4.14 Shows the superimposed nominal & scanned geometry 4.14a, shows the measurement of deviation in branches of the tree 4.14b, shows the measurement of the holes on branches 4.14c and shows the centre of densification of the tree 4.14d . . . . . . . . . . . . . . . . . . . . . . 50 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 (d) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.15 Shows the internal porosity in two views i.e. 4.15a & 4.15b. 4.15c shows the result of the measurement of various holes on the tree branch. 4.15d shows the high-resolution scan performed at 10µm focusing on the specific session of the artefact . . . . . . . . . . . . . 51 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 (d) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.16 Shrinkage percentage of the pre-sintered cubes S1, S2,S3, and S4 . . . 53 4.17 X,Y & Z Percentage Deformation for Gyroids G1, G1.5, G2 and G2to1 54 4.18 Angular deviation in X & Y for Gyroids G1, G1.5, G2 and G2to1 . . 54 4.19 Relative densities for studied samples. Here are density values derived using several techniques: Based on geometry, LOM, and Archimedes . 56 4.20 shows the Light optical microscopy for bridge with base(artefact 4A and 4B). 4.20a refers to with ceramic and 4.20b refers to without ceramic support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 4.21 X-ray CT scanning for threads(artefact 1A and 1B). 4.21a refers to without ceramic support and 4.21b refers to with ceramic support . . 58 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 4.22 As-cast AlSi12CU1 Optical Micrograph . . . . . . . . . . . . . . . . . 59 4.23 CAD cross-section . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4.24 Optical micrographs for as-sintered aluminum 4000 series alloy for XY, XZ, and YZ orientation . . . . . . . . . . . . . . . . . . . . . . . 60 4.25 Optical micrographs for HIPed aluminum 4000 series alloy for XY, XZ, and YZ oritentation . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.26 Vickers Hardenss and Knoops Hardness of studied materials . . . . . 62 4.27 Comparison of thermal diffusivity between as-cast Al-Si12-Cu1, as- sintered aluminum 4000 series and HIPed aluminum 4000 series alloy 64 4.28 Comparison of thermal conductivity between as-cast Al-Si12-Cu1, as- sintered aluminum 4000 series and HIPed aluminum 4000 series alloy 65 4.29 Optical micrographs between as-cast Al-Si12-Cu1, as-sintered alu- minum 4000 series alloy and HIPed aluminum 4000 series alloy . . . . 67 4.30 Performance(power/mass) graph for all Artefact in two self convec- tion setups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 xv 4.31 Performance(power/mass) graph for all artefact in forced convection setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 4.32 performance increase in percentage from free to forced convection . . 69 4.33 Shows the various views of the heatsink produced by BJT/M . . . . . 70 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 (c) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 (d) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 4.34 Evaluation of the new heatsink.4.34a: Shows the surface deviations using the heatmap. 4.34b: Shows the heatsink assembled on headlamp 71 (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 xvi List of Tables 3.1 The Grinding and Polishing methodology followed for sample prepa- ration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2 Experiment setting for LFA analyzer . . . . . . . . . . . . . . . . . . 28 3.3 Details of the Test Prints . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.4 Parameters used during Hot Isostatic Pressing . . . . . . . . . . . . . 35 4.1 Table showing the summary of all the artefacts along with the scaling factors designed by RICOH . . . . . . . . . . . . . . . . . . . . . . . 41 1 2 1 Introduction 1.1 Background Heat removal from mechanical and electrical devices is becoming increasingly impor- tant as society’s reliance on electronics and computational power grows. Thermal management equipment, such as heat sinks, heat exchangers, and heat spreaders, is frequently used to remove heat. Like electronics, automobiles also use efficient heat dissipation from their working parts to function correctly. It is known that to reduce weight and price, automotives are mainly made out of aluminum alloy. Hence it is crucial that the aluminum alloy can perform as per requirements. Currently, to increase the longevity and efficiency of the car’s headlamp, high-pressure die-casted aluminum alloy is used. This is mainly due to its high production rate and lower manufacturing cost than other methods. However, there is still room to improve, possibly by introducing additive manufacturing (AM), which can enable the design of more complex shapes and improve performance & properties while getting close to similar productivity. Through a layer-by-layer process, additive manufacturing (AM) generates accurate models or things directly from computer-aided design (CAD) models with minimal geometric limitations. Using several processes, AM may be applied to practically any material [4]. Each approach has specific applications and constraints [5]. In powder bed fusion, a heat source is used as a Laser beam (PBF/LB) or electron beam (PBF/EB) to fuse raw materials together and print the parts. On the other hand, Binder jetting technology is a multistep AM process "in which a liquid bond- ing agent is selectively deposited to join powder materials," according to ISO/ASTM 52900 [6]. Due to the higher production rate compared to PBF/LB, one of the most sought-after AM technologies for the automotive industry is Binder jetting tech- nology (BJT). It glues the particles together to produce the shape layer by layer, eliminating the need for heat during fabrication. Furthermore, unbound powder temporarily supports unconnected regions of the component, allowing interior vol- umes to develop and reducing material waste [7]. This procedure is usually referred to as the printing stage in BJT and must be followed by de-powdering, curing, sin- tering, and further post-processing to achieve the final material properties[8]. The quality of these final products determines the process’s application space. These final qualities depend on various process parameters, including the raw material, powder bed formation, construction parameters, sintering, as mentioned before, and post-processing techniques such as surface smoothening and heat treatment. Binder jetting technology for metals (BJT/M) has numerous potential advantages 3 as an AM process. It may, for example, deal with almost any powdered feedstock and integrate functionally-graded components. Because it only has to print the binder—a portion of the overall part volume—BJT systems have a comparatively high build rate when compared to other printing methods like Powder Bed Fusion – Laser Beam [9]. A 100-nozzle print head, for example, can produce parts at speeds of up to 200 cm3 per minute [4]. Numerous studies have been conducted and continue to be driven to grasp the full potential of this method [10]. The flexibility of conceptualizing and creating intricate patterns with BJT/M comes with its own difficulties. Discussions are cur- rently taking place in process parameter optimization, surface-roughness manage- ment, support structure removal, post-processing needs, compatible raw materials, and cost-competitiveness compared to traditional methods for bulk production. In this master thesis, design for additive manufacturing is explored for this particular BJT/M technology provided by RICOH Company Ltd. by printing various artefacts related to the heat sink for the Volvo XC90 model production car. Hardness, thermal conductivity, and thermal convection tests are measured to understand performance compared to the as-casted heat sink currently used in the XC90 production series car model. 1.2 Aim and research questions Additive Manufacturing components including Binder jetting technology for Metals (BJT/M) do not always fulfill the product requirements. In the case of BJT/M, it is mostly due to the presence of distortions, porosity, and shrinkage that influence the final component tolerances and properties. Hence, in this project, it was decided to evaluate these aspects with respect to the existing component. For this purpose, one of the heat sinks in the headlamp of the Volvo XC90 production series car was taken for re-design and printing using BJT/M for aluminum alloys. Based on the issue under investigation following research ques- tions were formulated to be answered in this project. • How to design for selected application using the BJT/M technology for alu- minum alloys? • What are the thermal and mechanical properties of Al-alloys produced by the BJT/M process? • What does the design methodology look like in the different process steps compared to the existing technology in BJT/M? • Does the produced application add any new features, improvements, perfor- mance, or other values to consider? These questions were answered at the end of the Thesis based on the knowledge gained from the project. 4 2 Literature Study 2.1 Binder jetting for metal Binder jetting technology is an AM method defined as a “process in which a liquid bonding agent is selectively deposited to join powder particles” [6]. During the process, the spreading of a powder layer in a build-box is followed by the deposition of a binder, which selectively glues the particles together. This procedure is repeated layer-by-layer to join the powder within the designed three-dimensional geometry in the resulting powder bed [8]. The typical BJT printer system can be divided into several main parts: a binder deposition system, powder dispensing, and recoating system build platform, and excess powder collecting chamber, as illustrated in 2.1. Figure 2.1: Typical schematic process diagram of BJT/M [1] BJT printer systems come in a variety of configurations, with the primary differences being the size of the powder bed, the features of the printhead, and the operation of the powder dispensing and recoating mechanism [1]. These systems’ performance can be adjusted by altering several factors that have an impact on the printing process, green components, and eventually the finished sintered components. Three categories can be used to classify these variables: 5 • Features of Powder Feedstock • Spreading and Deposition of Powder • Binder Placement The influence of each of these parameters throughout the BJT printing process determines the printing quality and characteristics of the green component, which represents the initial condition for the subsequent de-binding and sintering processes. The binder deposition system and the powder spreading and re-coating system can be separated into two major systems that must cooperate to print the desired part. The arrangement of the particles within the produced part will be significantly influenced by the operating principles of these mechanisms and the settings chosen for their operation. These fundamental elements, which define the characteristics of the BJT green components, are shown in 2.2. Figure 2.2: Major factors that determine the characteristics of BJT Green Body [1] For powder-bed-based AM techniques, spherical-shaped particles are typically pre- ferred due to the excellent flowability [8, 11, 12], but the recent studies [8, 13] employed using irregular shaped particles produced by water atomization. The at- omization process naturally results in a normal (Gaussian) PSD distribution, which is known as the particle size distribution, or PSD. PSD is commonly represented by the cumulative percentile values D10, D50, and D90 and can be measured using various methods (such as XRD method, Malvern Particle Size Analyzer) [1]. Generally, the mean size of BJT powders can range from sub-micron sizes to over 100µm. Although agglomeration problems could arise during the printing process, smaller powder (5 µm) has better sintering activity, which leads to faster densifica- tion [1]. Smaller particles are favored during sintering for faster densification and better final characteristics, but larger particles are favored during printing due to greater flowability and higher packing density. Hence, a compromise solution must be found for each unique situation. One of the properties of the powder is the pack- ing density. Whereas tap density is the density of powder that has been tapped to settle its contents in a container under specific conditions, apparent density is measured on loosely packed particles [1]. The powder bed packing density often resides between the apparent and tap densities 6 of the powder being used because there is no external pressure applied during powder deposition or printing [1]. Each powder layer is applied to the previous print layer using the powder spreading mechanism. As it affects how the particles are arranged while printing, it also has a significant influence on the quality of the powder bed that is created. The design of the mechanism (such as the type of blade or roller) and the characteristics of the particles will have the most effects on how the powder material interacts with the printing parameters. All of these characteristics, along with the printing parameters, will affect how the particles are arranged within the layer after spreading. To bind the chosen volume of particles together, the liquid binder is deposited on each layer. The printhead, the system that moves it, and the binding solution are the three essential parts. The distribution of binder droplets would depend on the cartridge movement path and binder deposition parameters since the impact of binder droplets on the powder bed may perturb it [14]. For the sintering process, where the evolution of the green components will be greatly influenced by the initial porous green structure, the control of these possible flaws is essential [1]. The geometrical precision and ultimate surface roughness of the parts are also impacted by the interaction between the powder and the binder. The BJT process, which must adhere to certain requirements, also relies heavily on the binder solution utilized. A few specifications to give the green components the required strength without impairing the sintering procedure and final properties. It is necessary to cure the binder in order to improve the mechanical strength of the bound metal part for handling. Next, during depowerding, green components are removed. The green components are then sintered to achieve consolidation. 2.2 Debinding process The binder within the green bodies must burn off before going forward with the sintering process [1]. The debinding process creates the pyrolysis of the binder component at a relatively low temperature, breaking down and evaporating the binder out of the green component [15]. Generally, the debinding process can be done in the same sintering cycle in the same furnace or prior to the sintering stage in a separate oven. A component’s microstructure and phase transitions during the sintering process may be impacted by the presence of carbon residues from the binder in the compo- nent [16, 17]. The impact of carbon residues during the sintering of stainless steel 316L has been demonstrated in earlier research [18]. However, the impact of binders during the sintering of AlSi-alloys hasn’t been published yet. Yet, when proper debinding is done, the amount of these residues is minimal, thus not adversely affecting the sintering process and providing desirable properties of the component. Generally, thermal analysis methods are used to determine the debinding process [1]. For example, thermogravimetry analysis (TGA) enables high-precision assessment of the binder removal, by mass loss during debinding when in contact with the process atmosphere, and illustrates the kinetics of the binder decomposition. RICOH’s propitiatory binder is used in the current thesis study, and the suggested debinding 7 procedures from the technology provided were utilized. 2.3 Sintering process Sintering is a thermal process that uses mass transport mechanism to bind par- ticles into solid structures [1]. This bonding contributes to the densification and strengthening of the components during sintering. The relative density is a popular way to express the degree of sintering at the macroscale and is a measure of the ratio between measured density to the theoretical density (in g/cm3). The internal microstructure evolves as a result of pore volume reduction and temperature distri- bution, whereas the geometry experiences volumetric/dimension changes as a result of the sintering process. Sintering is categorized in different classes such as solid-state sintering and liquid state sintering. When particles are heated during sintering, solid linkages between them are created. By eliminating free surfaces, the linkages lower surface energy, and grain boundary area is subsequently eliminated by grain growth [19]. This sintering process is known as solid-state sintering. Although dimensional change is undesirable in many sintering systems, it is possible to reduce the pore volume with prolonged heating, which results in compact shrinkage. Hence, there are two types of industrial sintering: types that concentrate on densification and types that concentrate on strengthening without necessarily causing dimensional change. Through sintering at relatively high temperatures, structural materials such as silicon nitride, alumina, cemented carbides, steels, and silicon carbide are treated to full density. Contrarily, densification is decreased when structures like capacitors, cores, and mechanical spacers are sintered [19]. The adhesion of solid particles created by liquid formation, on the other hand, is what propels LPS [19]. Several powder materials are used in this procedure, one of which has a significantly lower melting point than the others, such as W-Ni or Fe-Cu alloys. AlSi alloy used in this study is typically sintered by Liquid Phase Sintering (LPS) due to its ability of pore fillings [20]. Since aluminum alloy has a lower melting point, it is also possible to achieve a higher density in a shorter time [21] . On the other hand, stable oxide film acts as an effective diffusion barrier for the sintering to continue during SS [22], which resulted in the selecting of LPS for densification. 2.3.1 Liquid phase sintering The reduction in total interfacial energy is what propels sintering. By (a) replacing high surface energy interfaces with lower surface energy interfaces, such as replacing solid-vapour interfaces with grain boundaries, or (b) decreasing the total surface area per unit volume, the total interfacial energy can be reduced [21]. According to Kingery’s classical theory [21], liquid phase sintering happens as a result of three concurrent mechanisms, as depicted in figure 2.3. These mechanisms are (1) particle rearrangement, which results in rapid densification; (2) contact flattening by solution re-precipitation, which reduces the distance between the particles’ centers; and (3) the coalescence of solid particles, where solid-state sintering mechanisms constrain densification. As the sintering process continues, the dominating mechanism evolves. 8 Although it has been criticized, Kingslay’s theory is still considered the foundation for understanding LPS. Figure 2.3: Simple schematic diagram of LPS [2] The typical temperature range for sintering aluminum alloy is 570°C to 620°C. The low-melting additive powders can then melt and interfuse to form liquid phases. High liquid fraction typically accelerates densification [19] but also makes dimen- sional control more challenging [22]. Densification happens during sintering via various overlapping but balanced mechanisms [21], all of which are influenced by temperature, time, and ambient conditions. In a typical circumstance, solid particles are soluble in liquid [2]. Its solubility makes the liquid penetrate the solid, creating a capillary force that draws the grains together. In addition, the high temperature softens the solid, which aids the densifi- cation process. Liquids are related to high diffusion rates, resulting in rapid sintering or lower sintering temperatures [2]. Based on Kingery’s classical theory on LPS, [2] also explains the LPS phenomena which are easy to visualize. Based on this literature , During heating, the solid grains undergo solid-state sintering. Microstructure evolution processes are conceivable depending on the solid–liquid solubility interactions. The surface energy associated with pores causes their annihilation, whereas microstructure coarsening increases the material’s tensile strength in the as-sintered state compared to the as-printed state. [19]. Hence, the liquid phase’s surface tension is a crucial influence in determining the sintering rate of LPS. The key factors to control the LPS consist of, interfacial energies, wetting, spreading, penetration, segregation, contact angle, capillarity, contact force, and solubility [19]. In the case of contact angle, the lower the contact angle, the better the densification, meaning better wettability of liquid in the solid-liquid interface. In the case of wetting, a high liquid content guarantees quick densification [19]. However, if there is an excessive amount of liquid, distortion will develop. The wetting of the liquid in the solid-liquid interface can also be aided by chemical reactions. A large contact angle implies insufficient wetting, therefore, the liquid withdraws from the solid [2]. 9 This causes compact swelling and liquid exudation from pores. Hence, depending on the contact angle, liquid transformation results in either densification or expansion. The extent of the capillary effect is determined by liquid volume, particle size, and contact angle [2]. When the dihedral angle is considered, lower is better [19]. 2.4 Sintering of aluminum alloys Aluminum is always protected by an oxide due to its extremely high affinity for oxy- gen [22]. Aluminum powders are made mainly using the air atomization technique and, hence, typically have an oxygen content of around 0.3-0.6 wt.%. resulting in the formation of a thermodynamically stable oxide film (Al2O3 film) of a few nm (around 2 nm according to [23]) forms on the surface. The oxide coating effectively prevents the inter-particle diffusion of aluminum atoms and the diffusion of solute atoms into the aluminum powder particles [22]. To facilitate successful sintering, the oxide film must be chemically reduced or physically disrupted. In the case of physically disrupting the oxide layer, high pulverization of powder is needed dur- ing solid-state sintering. However, there will still be the presence of oxide particles which will continuously hinder the sintering process [22]. As a result, mechanical disruption frequently falls short of allowing for substantial sintering. As the oxide layer (Al2O3 film) in the aluminum alloy has a higher melting point, it is not anticipated that liquid aluminum will wet alumina near the melting point of the metal. However, the solid-liquid interface must have a smaller contact angle to be wettable [2]. Magnesium is therefore added as a minor alloying element to break the oxide layer as a workaround. Magnesium can decrease aluminum oxide via reaction, therefore rupturing the oxide coating sufficiently to enable diffusion, wetting, and sintering via the production of spinel (MgAl2O4) [24]. Depending on the particle size of the aluminum, the highest effect occurs at 0.1 wt% Mg. The powder being aluminum 4000 series, it can be said that the trace amount of Mg is essential for this sintering to occur effectively. For the sintering of aluminum alloy, the sintering atmosphere also plays a significant role. Nitrogen is more effective than Argon for sintering aluminum, while magnesium is an active component. Based on [25], The self-gettering theory is formed to explain this phenomenon. Deep inside the pore network of P / M aluminum alloy, the gas flow rate (either Ar or N2) is substantially lower than at the surface, resulting in the consumption of O2 by the aluminum quicker than it can be replaced by incoming gas. The local oxygen partial pressure will then decrease. Aluminum essentially functions as a self-getter. Thermal expansion occurs during the sintering process, but because the pore network lacks a large concentration of O2, the oxide layer is not regenerated. As a result, fresh metal is exposed to the sintering atmosphere. In the presence of N2, exposed metal will rapidly undergo the reaction described by the equation. 2Al + N2 = 2AlN which is exothermic. This will result in a local rise in temperature, which may promote local melting and facilitate sintering. 10 2.5 Effects of silicon on thermal conductivity of Al-Si alloys The rate at which heat moves through a material is measured by its thermal conduc- tivity. It is primarily controlled by thermal consumption processes (Specific heat), elastic vibrations of the lattice (phonons), and electrical conductivity. If the influ- ence of phonons is minimal (as it is for pure metals), the mobility of electrons, or electric conductivity, σe is what primarily affects thermal conductivity [26]. Elec- trons are scattered when they come into contact with moving, vibrating atoms, as depicted in the figure 2.4 [27]. Electrical resistance and thermal resistance are produced by flaws and phonon scattering acting on free electrons. The mean free range of the free electrons thus serves as the primary determinant of their electrical and thermal conductivity. Furthermore, according to [28], the ability of an alloy to transfer heat increases with the length of the average free route of phonons. Nev- ertheless, crystal defects such as stacking faults and twins, as well as precipitates formed because of secondary alloying elements, would scatter phonons, reducing the average free path and ultimately having a negative impact on thermal conduction. Figure 2.4: Schematic diagram of collision of free electrons with eutectic Si con- taining particles While the heat conductivity of pure aluminum (237 Wm/K) is high, it is typically poor in alloys made of aluminum and silicon. For instance, the A380 alloy has a heat conductivity of just 96 Wm/K. Thermal conductivity is impacted by alloying elements in various ways [3]. They disintegrate in the matrix phase and impact the phonon and electron pathways, as mentioned before. The solubility limit is reached when the concentration levels of alloying elements are above it, resulting in variable morphologies and amounts of precipitates, which alter the thermal conductivity. It is also mentioned in this paper, [3] that, through altering the routes for electron migration and phonon contributions, microstructures linked to a variety of thermo- mechanical properties also have an impact on thermal conductivity. The strength of aluminum alloys can be significantly increased by adding Si [29]. The thermal con- ductivity in Al alloys is primarily controlled by electronic thermal conductivity and secondarily by lattice thermal conductivity [30]. As mentioned previously, adding alloying elements significantly reduce electrical conductivity and thermal conductiv- ity, with elements in the solid solution having lower values than the same number of 11 elements that form intermetallic phases. These latter often exhibit a proportionate decrease in thermal conductivity with increasing volume fraction [31]. According to [27], the primary intermetallic phase of Al-Si binary alloys, eutectic Si, serves as the thermal energy carrier’s scattering center. For low Si element contents, the α Al grain size in binary Al-Si alloys decreases as the Si percentage rises. The number of grain boundaries grows, which slows down the flow of free electrons and lowers thermal conductivity. figure 2.5 shows a summary of the thermal conductivity of different alloying elements in the aluminum matrix. Figure 2.5: Thermal conductivity of binary Al-Si, Al-Cu, Al-Fe, and Al-Mg alloys [3] According to [3], the thermal conductivity of Al-Si alloys reduces as the Si concen- tration rises up to 6 wt%. The thermal conductivity essentially stays constant at 157 W/mK when the Si concentration surpasses 6 wt% [3]. This is because, accord- ing to [3], the eutectic components and Si proeutectics did not significantly alter the free routes for electron migration when Si was present as an equiaxed phase in the hypereutectic alloys, raising the Si concentration had no impact on the thermal conductivity of the composites. Up until the larger volume percentage of proeutectic Si became the dominant phase in the hypereutectic alloys, the almost flat thermal conductivity could be maintained. For foundry alloys, the electrical conductivity in the Al-Si binary alloy falls from 29.26 to 25.33 mS/m and the thermal conduc- tivity from 182.38 to 164.16 Wm-1 K with increasing Si concentration [27]. The addition of Si, Fe, Mg, Mn, and Cu also reduces thermal conductivity. Moreover, it was discovered that the Mn element significantly reduced thermal conductivity 12 [32]. The morphology (shape and size) of eutectic Si, in addition to the impacts of the alloying elements, has a significant impact on the thermal conductivity of Al-Si alloys [33, 27, 34]. It is discussed that the size of the α-Al grain varies with the change in Si content [27]. Moreover, as the Si content rises up to 9 wt%, nucleation undercooling, and recalescence increase. The constitutional undercooling of different kinds of solutes in the Al solution can be layered on one another, according to [35], The presence of trace amounts of Fe (6 ppm), Cu (8 ppm), Mg (4 ppm), and Ti (1 ppm) makes it difficult to obtain a high-purity Al-Si binary alloy. As a result, the Si does not combine with these impurity elements to produce intermetallic compounds when the Si level is low. The compositional supercooling of the melt, on the other hand, increases with increasing Si concentration and encourages grain refinement. Moreover, as Si concentration rises, the growth restriction factor rises as well, which prevents the formation of a-Al grains. In contrast, if the Si concentration is in- creased further, the size of the a-Al grains grows. This is because as Si is reacted with other impurity elements in the melt to produce intermetallic compounds, the effect of composition supercooling is diminished. Although the thermal conductiv- ity of single-phase materials is well understood, heterogeneous alloys such as Al-Si comprise the majority of those utilized in industry. [26]. 2.6 Heat exchange performance Heat sinks are the devices or objects which are generally used to take away heat from the electronic and other devices producing heat. They act as heat reservoirs absorbing excess amounts of heat away from the heat generating device [36]. Nor- mally heat is carried away from the heat source to the heat sink via conduction and dissipation of the heat from the heat sink to the environment takes place by convection [37]. Conduction is directly proportional to the heat transfer area and the difference of temperature across the layers. Conductive heat transfer across the plane wall of thickness ∆X and area A, having temperature difference across the layers ∆T is given by: Qcond = kA ∆T ∆X where k is the thermal conductivity of the material[38]. Hence, Heat carried away by conduction also depends upon the material property. In this study, the focus was not on change of parameters ∆T or ∆X , since the input heat from the headlamp is predefined while the length of the heat sink cannot be increased due to space constants inside the car. Thus, the thermal conductivity of the produced AlSi is of great significance. While the convective heat transfer can be given by Newtons law of cooling Qconv = hA(Ts − T∞), where h is the convective heat transfer coefficient, while the A is the surface area. It is not practical in many cases to increase the heat transfer coefficient. Hence increase in the surface area is the known method for increasing the convective heat transfer, as can be seen in the case of heatsinks with fins. While increasing the surface area alone does not guarantee an increase in performance 13 since packing the fins too closely reduces the heat transfer coefficient due to extra resistance in the added fins [39]. It also increases the mass of the heatsink which is undesired property in this application. By disturbing the flow channel in the fluid, the heat and mass transfer by the fluid also increases, thus resulting in higher heat transfer performance. This disturbance to the flow is better achieved in triply periodic minimum surface(TPMS) lattice structures like gyroid compared to the conventional fin-based heat sink fluid gets flown into non-flow channels due to the complex shape of the gyroids and thus more secondary flow in the fluid causing increased convection [40]. 2.7 Design guidelines for Binder jetting for metal DfAM is a methodology of creating, optimizing, and adapting the components, assemblies or products in order to utilize the benefit offered by AM [41]. DfAM in general offers a great deal of freedom to design the parts without the limitations put on by conventional manufacturing processes. DfAM designers have fewer limitations when it comes to the geometry of the part. Thus resulting in the creation of complex parts by combining one or more components into one component [42]. While there is a lot of freedom while designing the parts for AM, there are cer- tain limitations as well. These need to be taken care of to ensure that parts are manufactured successfully. This includes both pre and post-processing along with the printing of the parts [43]. These limitations are taken care of for all the AM processes by the defined set of rules called as design guidelines. They can be de- fined as the information that helps in understanding the particular AM process, the procedures involved, and the best practices for designing the part [44], due to the differences in AM processes, their procedures and geometrical properties. Each of the AM processes has its own set of design guidelines that must be followed to achieve the best end results. To fabricate components using BJT/M, the following design guidelines have to be followed: • Maximum Component Size: Although theoretically, the maximum size that can be produced is approximately equivalent to the size of the build vol- ume after compensating for clearances on all the sides. But as the size of the component increases it becomes unstable due to increased anisotropic shrink- age. This means that many iterations and design changes with each iteration are required to successfully obtain the part. Basic knowledge in BJT/M is an optimum set for printing is considered to 50mm in any direction[45], but depending on geometry it could sometimes become larger. Larger parts need to be handled with more care since the brittleness in the green component increases with size. • Wall Thickness: Standing walls which are non-supported should at least have a minimum thickness of 2mm. While walls with a thickness of 1mm have support on one side or both sides. While the height-to-thickness ratio should be 8:1 or less at all times. 14 (a) (b) Figure 2.6: Recommended Wall Thickness Printable by MBJ • Overhang Angles: The angle between the vertical surface/face of the com- ponent and the surface of the build plate should be 60 degrees or more. • Overhangs: These features can be subdivided into two types based on the thickness and length of the overhang. The maximum length which can be produced accurately is 10mm. While the minimum thickness of 2mm should be maintained for the overhang feature. Refer to the figure 2.7. Figure 2.7: Overhang length and thickness for BJT/M • Powder removal holes: It depends significantly on the design. For parts designed with shells, this feature is critical. While solid parts may also require escape holes to remove the powder from critical areas. It is recommended to have a minimum of two escape holes of 5mm in diameter[43]. Refer to the figure 2.8. 15 Figure 2.8: Powder Removal holes for Shell parts in BJT/M • Radius: It is recommended to provide a fillet of 1mm minimum at corners, knife Edges formed between two surfaces. These sharp features can result in a stress concentration, especially during the sintering [43]. They can also help with the removal of the powder without damaging the geometry during de-powdering. Refer to the figure 2.9. Figure 2.9: Radius at corners and sharp edges for BJT/M • Base Thickness: The surface of the component parallel and in contact with the build surface can be produced with as little thickness as 1mm[46]. • Font Size depth and Width: Samples engraved with features such as by providing identification numbers or fonts need to have the minimum depth and the width of 1 mm to be achievable [47]. • Minimum Feature Size: Although BJT/M has the highest resolution when it comes to metal powder AM. The main concern is damaging the components during the handling. Therefore it is recommended to have the feature size no less than 2mm [48]. Refer to figure 2.11. 16 Figure 2.10: Minimum recommended printable feature by BJT/M • Tolerances: Tolerance in the close-fitting parts are not often discussed when it comes to guidelines but it can be considered as a good practice, to keep tolerances of 1mm between close-fit components. • Lattice: One of the major advantages of AM is the ability to print lattice structures. Lattice can be formed using the network of struts or by using the walls. They provide lightweight high specific strength, and higher stiffness[49]. In the case of MBJ it is recommended to print the lattice with a minimum 1 mm thickness of the walls or the diameters of the struts. • Horizontal Holes: Since BJT/M has a very high resolution, very small hole diameters can be produced. However, due to restrictions on powder removals larger diameters are recommended for higher depths. Diameters of 0.5 mm can be successfully produced up to the depth of 1mm. This number increases exponentially to 5mm for a depth of 30mm and above [46]. • Vertical Pins: vertical non-supported pins should have a minimum diameter of 1mm. For the length of the pins 5mm or higher the diameter must be 2mm minimum for successful printing. 17 Figure 2.11: Vertical Pins printable by BJT/M Most of the design guidelines studied and summarised are from previous studies. Also, some public once’s by machine providers on their website, but most of them where derived from previous work with stainless steel 316L. Since aluminum alloys are new in the BJ Technique for metals there are no existing design guideline studies to be best of the author’s knowledge. The above guidelines only act as the first steps in understanding where the criticality lies in the BJT for Metals. Guidelines specific to the application have been produced and evaluated before being incorporated into the final design. This is further explained in section 4.1. 18 3 Methods 3.1 Design for Binder jetting technology for metal In order to print the final part it needs to be designed in a way that complies with the rules of AM and more specifically particular technology. These are the design guidelines for the Binder jetting technique for metals discussed in session 2.7. This is done using a combination of design software. It generally involves using 3D parametric CAD software for an initial form of the design, followed by software for topology optimization. or software that can design complex features like lattice structures i.e. gyroids. Then the design files are sent to build preparation software where things like part print orientation are set, necessary support structures are given, and optimized staking of the part is worked out based on the build box volume. There are few build preparation software available on the market currently. Among them, Materialise Magics by Hexagon is one of the most commonly used software in the industry [50]. In this project, the existing design for the heatsink of Volvo XC90 was used as the reference. The file was in the parent CAD format of software Catia V5 [51]. Modification to the design was done in this software and then the file was taken into N-topology software [52]. It helps to design complex gyroid shapes and other intricate features of the components. It was used to design gyroid for heat dissipation on the heat sink. The final component design was exported as mesh and the build preparation was done using Materialise Magics [50]. Finally, printing was done in RICOH’s in-house prototype printer for Binder jetting machine for aluminum alloys [53]. 3.2 Material To understand the sintering anisotropy and microstructural evolution during sinter- ing, sintered cubes were used. As-sintered cubes have a volume of 10x10x10 mm3, and the theoretical full density of aluminum 4000 series alloy is approximately 2.69 g/cm3. 19 Figure 3.1: Cubes (green body cube and sintered cube) Cubic samples were oriented in the build space with cubes planes orthogonal to the printing directions: building direction (Z axis), powder spreading - recoater direction (Y axis), and binder deposition direction (X Axis). Each sample was designed with an identification spot produced on the top surface which was used for axis identification. Figure 3.2: Build orientation of cubes After the printing is done, the samples are extracted through a powder removal process. In this process, at first, the area where the binder is jetted is cured in a drying oven, which is followed by powder removal, namely taking it to a powder station to collect the excess powder. After the powder removal step, the debinding and sintering process was carried out in the same furnace. Debinding and sintering of the green bodies were conducted by continuous heating in a vacuum. Sintering was performed by heating at approximately 600°C for several hours. 3.3 Density measurements Three approaches were utilized to quantify the relative density of these sintered sam- ples. Geometry-based density was measured using calipers and weight-measuring de- 20 vices. Archimedes’ density follows ASTM standard B328 and uses a high-precision weight-measuring machine. LOM-based density is measured using ImageJ. The geometry-based density of cubical samples is calculated from weight (m) and volume (V) from dimensional measurements where ρ =2.69 g/cm3 was used. Us- ing the following equation 3.5.2 relative density was calculated. The measurements were taken with a single-point automatic caliper with a resolution of 0.0001 mm and high-precision balance with a resolution of 0.0001 g as well as with high precision 3D scanning instrument at the Geometry and verification center in Volvo Car Cor- poration. GOM Inspect was used to measure the x, y & z dimensions of the cubes. The working methodology of GOM inspect is further explained in section 3.4. ρ(%) = m V × ρfull × 100 ASTM B328 standard was used to measure Archimedes’ density. The following equation was used to calculate the relative density when Archimedes’ density is considered. ρ(%) = W1 W2−W3 ρfull × 100 Where, • W1 refers to the weight of the dry sample • W2 refers to the weight of the sample submerged in water • W3 refers to the weight of the sample when pores are filled by water The LOM-based density was calculated by splitting the cross-section into 30x30 areas of interest (ROI) and measuring area fractional density using image analysis. FIJI ImageJ particle analysis plug-in was used to analyze porosity morphology [1]. 3.4 Dimensional and geometrical inspections As discussed previously in BJT/M, there is a large deviation from the nominal geometry due to deformation and distortion caused during the sintering. This is further increased with an increase in the size of the part. The component targeted in this project is relatively large from a BJT/M perspective. Hence dimensions and geometrical features of the design needed to be evaluated for deviation and distortion from the nominal CAD geometry. Knowledge gained from the evaluation could be then incorporated into the new heat sink design. GOM Inspect by Zeiss[54] is a software that was used for this purpose. It has features that allow for CAD data (nominal geometry) to be compared with the printed components. To get the comparison printed components are required to be converted to Standard Triangle language(STL) files. This was done by 3D scanning the parts using a guided laser beam from Model Maker H120 by Nikon [55]. This system was available at Volvo 21 Cars Corporation. 3D scanned parts are then saved as STL CAD files which were used for comparison with the nominal CAD. The inspection methodology and the results are summarized in the section 4.1. 3.4.1 Metallography For metallographic preparation, sintered cubic samples were sectioned along three possible cross-section planes specified by the samples’ axes (i.e. XZ, YZ, or XY), whereas, casted samples were sectioned into one orientation. The Metallography of the alloy was divided into several sections. During sectioning for optical microscopy, a nonferrous cutting blade is used whereas, to section for thermal conductivity using laser flash analysis, a diamond wafering blade is used. Sectioning was carried out using a Buehler IsoMet High-Speed Pro machine. (a) Buehler IsoMet High Speed Pro (b) Struers Citopress-20 for mounting Figure 3.3: Metallography-Sectioning and Mounting During the mounting, x, y, and z directions were marked, in order to keep track of the recoater, binder deposition, and building height direction. Polyfast polymer was used for the mounting procedure. Hot pressing was done and it took approximately 6 minutes to complete both the heating and cooling cycle. Mounting is done using the Struers-Citopress-20 machine. 22 Table 3.1: The Grinding and Polishing methodology followed for sample prepara- tion Speed (rpm) Time (min) Force (N) Rotation Lubricant Grit/Polishing Cloth 150/150 3:30 15 ⟳ ⟲ Water 220 SiC Foil 150/150 2 15 ⟳ ⟲ Water 320 SiC Foil 150/150 ∼3 25 ⟳ ⟲ Water 500 SiC Foil 150/150 3 25 ⟳ ⟲ Water 800 SiC Foil 150/150 ∼3 25 ⟳ ⟲ Water 1000 SiC Foil 150/150 3:30 25 ⟳ ⟲ Water 1200 SiC Foil 150/150 ∼3 25 ⟳ ⟲ Water 2000 SiC Foil 150/150 ∼3 25 ⟳ ⟲ Water 4000 SiC Foil 150/150 6 25 ⟳ ⟲ DiaPro Dac 3 micron 3 micron Cloth 150/150 6 25 ⟳ ⟲ DiaPro Nap 1 micron 1 micron Cloth 150/150 1:30 15 ⟳ ⟲ OP-S OP-S Cloth (a) Struers labopol-21 for cham- fering the edge (b) Struers Tegrapol-31 for grind- ing and polishing Figure 3.4: Metallography-Polishing and Grinding Grinding and polishing were done in 2 phases. During the 1st phase, a manual grinding machine, Struers laboplo -21 was used. It was used mainly to chamfer the edge of the polymer mount to prevent it from tearing the grit papers. After the initial manual grinding, during the 2nd phase, Struer Tegrapol-31 was used and several grinding and polishing parameters were followed, mentioned in table 3.1. 3.5 Mechanical properties testing This section outlines the procedures and principles utilized for the mechanical prop- erties testing of the produced aluminum alloy in the research study whose results are summarized in this report. 23 3.5.1 Hardness test Rockwell, Vickers, Brinell, and Knoop are the most common hardness tests. Hard- ness testing values are arbitrary, meaning that there is no linear relationship between hardness testing values from one load/force to another. Depending on the material, hardness test methods can vary. One more thing that should be kept in mind when it comes to using higher loads, is that higher loads result in lower hardness. Hardness test measurement was done using Struers DuraScan as depicted in figure 3.5, which supports Vickers hardness and Knoop hardness. Generally, Vickers hardness is pre- ferred for ductile materials such as aluminum alloy. In contrast, Knoop hardness is used to determine cracking in brittle materials and to facilitate the hardness testing of thin coatings such as surface coating. The time for indentation has been set to 15 seconds. The ASTM standard E384 for hardness was adopted. The macro hardness test was chosen to measure the material’s hardness. The surface was smoothed and ground. The time for indentation has been specified as 10 seconds. 5 measurements were done for each sample and for each orientation. Since as-sintered material is used, HV3, HV 5, or HV 10 is recommended [56]. High loading is advantageous when as-sintered material is used as ’bulk’ information is required. In addition, by employing high loads, values can be more consistent because they are less susceptible to micron/pixel variations. Figure 3.5: Struers DuraScan for Hardness Test Measrurement 3.5.2 Thermal conductivity test Thermal conductivity (TC) measurements were carried out on a Netzsch LFA 457 MicroFlash® table-top instrument (Netzsch Gmbh, Germany), which employs the Laser Flash Method (LFM) to calculate thermal diffusivity and specific heat values. 24 The measurement was done by following the ASTM E1461 standard, shown in figure 3.6. The LFA 447 HyperFlash is a vertical system, with the flash source at the base, the sample in the middle, and the detector at the top. Figure 3.6: Laser Flash Technique Figure 3.7: LFA 447 Instrument 25 12.7 mm in diameter and 1 mm thick samples were used for this experiment. The samples were machined using a diamond wafering blade to have better precision and less material removal from a produced or cast bulk sample given by RICOH or from the heat sink of the production series car, model XC90. Due to the machining, the outer surface was smooth. In order to obtain a good detector signal and to increase the emission ratio, the sample must not reflect light and must feature good emission and absorption [57]. Due to this, samples that have higher reflectivity such as aluminum should be coated. As coating improves the sample’s emission/absorption properties, optimizing the signal-to-noise ratio. Graphite coating is done in this experiment as depicted in figure 3.8. (a) Shot detection without graphite coat- ing (b) Shot detection with graphite coating Figure 3.8: Signal to noise performance between graphite coated and without graphite-coated as-sintered aluminum 4000 series alloy sample Graphite can have higher thermal conductivity than both copper and aluminum. However, when measuring high thermal conductivity materials with short half time the graphite coating will have a larger influence on the half time compared to coating low thermal conductivity samples like polymers. For this reason, dusting is done instead of coating the surface of aluminum samples with graphite in order to have a better signal-to-noise ratio while maintaining high thermal conductivity as depicted in figure 3.9. 26 (a) Sample with no graphite coating (b) Sample dusted with graphite (c) Sample coated with graphite Figure 3.9: Visual comparison between different coating techniques Dusting of graphite on the surface was followed by, placing the samples on the sample holder as depicted in figure 3.10. A disk-shaped mask with a big orifice is placed on the top of the sample to control the flash path. (a) Sample holder (b) Schematic of radial or in-plane diffusivity mea- surement. Figure 3.10: Placement and schematic of in-plane diffusivity measurement For the specific heat capacity measurement, Cp the samples were not "dusted" but coated with more graphite. The reference POCO graphite was measured "as is." We followed the coating procedure recommended in Netzsch documentation 27 Using the laser flash technique, thermal diffusivity, and specific heat capacity are measured. Combining these thermophysical properties with the density value per- mits the following thermal conductivity calculation: λ(T ) = a(T ) × Cp(T ) × ρ(T ) Where, • λ refers to thermal conductivity [W/(mK)] • cp refers to specific heat [J/(gK)] • a refers to thermal diffusivity [mm2/s] • ρ refers to bulk density [g/cm3]. A total of 6 temperature points was used to calculate the thermal conductivity. The initial temperature was set to 20 °C, and the final temperature was set to 145 °C. Details for the experiment parameters can be found in Table 3.2. LFA analyzer software was used to put the values during the experiment. Co coefficient of linear thermal expansion was assumed to be 0. Table 3.2: Experiment setting for LFA analyzer Parameters Values Starting Temperature °C 20 Shots 3 Increments °C 25 temperature data points 6 Delay (ms) 0 Main Amp Changes with temperature It should be noted that, although the specific heat capacity Cp of as-sintered alu- minum 4000 series alloy was measured, it was 0.82 J/(gK). It was characterized by the temperature span used for the measurements. However, in the case of the cast aluminum alloy sample, the same specific heat capacity was used, as mentioned before, due to time restriction as an approximation. In the case of pure aluminum 1050, the specific heat capacity was 0.89 J/(gK) in all temperatures. 3.5.3 Thermal convection test To understand the heat exchange performance of the 3D-printed heatsink. It was decided to perform the experimental study. For this reason, a literature study was carried out. It was learned that to perform the test main elements apart from the test piece are the heating element, insulation for the heating element, thermocouples for recording the temperature, DC power regulator, and the data acquisition system. As well to get good-quality results it is essential to conduct the test in a stable ambient temperature. 28 One of the critical aspects was the loss of heat and thus, good insulation to the back side of the heating element is essential. In the study conducted by Haghighi et.[37] they used additional heater on the back side of the primary heater to control the heat loss. A lot of inspiration was taken from the study carried out by Feng et. [58] called natural convection on metal foams heat sinks with open slots. Based on the studies and research done, test rig design was worked out which can be seen in figure 3.11. The heating element of 2”X2” was used and purchased online from the website Omega [59]. Kypton heater with a maximum power of 40 watts was chosen. Since the test pieces had the substrate area of 40mmX40mm heater bigger than this was chosen so as to get the uniform initial distribution of heat throughout the test piece substrate. The heater was operated by connecting it to a power supply through a DC power regulator. This allowed the user to control the input power at any given amount of time. Figure 3.11: Test Rig developed to measure the heat exchange performance of the artefact designs For the purpose of insulation polyurethane foam having a very low thermal conduc- tivity of 0.0036Wm/k was chosen [60]. For the back lid. PU foam of 91mm x 91mm having a thickness of 50mm was taken. The slot was cut out into this to house the heater and the electrical connection to the heater. Whereas front lids had the di- mension of 91mm x 91mm x 3mm, with the cutout in the center matching the shape of the testing devices. The front lid was used to prevent the loss of heat from the sides of the substrate. While the back lid was to prevent the loss of heat from the 29 back as well as the sides of the heating element. A thermocouple was connected at the interface of the heating element and the testing devices to get the temperature at the heater. Data from the thermocouple and the power supply unit was gathered using the data acquisition system called the IPTRONIK’s M-Thermo2 u which was loaned from the measuring department at Volvo Car Corporation. To replicate the conditions of the force convection a blower was placed below the testing device, thus allowing the air to blow through the test pieces. The setup was mounted on the frame which is made using the aluminum pipes which are procured from the company called AluFlex AB [61]. These are aluminium pipes that are designed such that they allow for easy and modular fabrication. The top of the frame was made up of horizontal slab aluminium covered with foam used inside the hood of the car to get closer to the air circulation conditions in the Car where the hood of the car blocks the free passage of the hot air upwards. The test rig was placed inside the environmental Chamber to maintain the constant ambient temperature throughout the test which was maintained at 25◦C throughout. This can be seen in the figure 3.12. Figure 3.12: Environmental test chamber housing the assembled test rig with the data acquisition system A simple experimental procedure was worked out to calculate the heat loss as well as the performance of the testing devices, as can be seen in Fig3.13. For the first case, 30 power was increased gradually such that the temperature at the heater was raised to 35◦C which resulted in a ∆T = 10. Power reading(P1) was recorded at a constant temperature. The temperature at the heater was considered constant if the change in temperature was less than 0.3◦C in 5 minutes. Since the boundary for insulation was symmetrical. Loss of power through insulation for specific temperature differences was considered as half of P1. The front insulation lid was replaced with a lid housing a specific design of the heatsink. Then the same procedure was repeated until similar ∆T = 10 was achieved. The power (P2) required to achieve this same temperature difference was measured and logged. Based on this performance of the heating device(P3) was calculated at specific boundary conditions, which is nothing but the power lost through the heating device that is given by:- P3 = P2 − 0.5P1 Figure 3.13: Schemetic representation of calculation for power lost through the device This was repeated for ∆T of 15, 20, 25, 30, and 40 for each of the test devices whose performance was to be tested. All the readings were taken for three different test setups • Self Convection with vertical test setup was carried at ∆T10, 20, 30&40. • Self Convection with the horizontal test setup was carried at ∆T10, 20, 30&40. • Forced Convection with the vertical test setup was carried at ∆T10, 15, 20&25. All three setups can be seen in Fig 3.14. 31 Figure 3.14: Schematic illustration of all three convection setups This resulted in a total of 63 data points of power lost through various test pieces in 3 different setups. This data was then analyzed by calculating the performance factor as well as computing the increase in performance from free to forced convection. Results were plotted on the bar graph for evaluation. Results can be seen in 4.7.3 To understand which gyroid design performs the best before they could be incorpo- rated in the new heat sink design, artefact gyroid designs mimicking certain features of the heat sink were produced. These shapes had the same overall dimension of 40X40X33 mm including a 3mm thick base plate. Their shapes had thicknesses of 1mm, 1.5mm, and 2mm respectively. While the fourth shape had a varying thickness from the base to the top between 2 to 1 mm respectively. Furthermore, the fin design very similar in box volume to the gyroid lattice was produced to be tested in the experimental test setup. This was done to evaluate the gyroid designs on more level ground that has similar material properties. Fur- thermore, the existing heatsink was machined to match the box volume of all the produced shapes approximately. This allowed us to understand the combined effect of the material and the shape produced on thermal conductivity. Details for the test samples can be found in Table 3.3. 3.6 Materials characterisation This section outlines the procedures and principles utilized for characterization in the research study, whose results are summarized in this report. 3.6.1 Light optical microscopy A Zeiss Axioscope 7 light optical microscope (LOM) fitted with a Zeiss Axiocam 105 color camera was used to capture images of the polished samples. It is possible to obtain photographs of the whole cross-section by piecing together high-magnification 32 Table 3.3: Details of the Test Prints Si no Geometric Parameters Gyroid 1mm Gyroid 1.5 mm Gyroid 2mm 1 Box Volume=LXBXH(mm3) 40X40X33 40X40X33 40X40X33 2 Surface Area (mm3) 34272 34221 33678 3 Material AlSi AlSi AlSi 4 Material Density(kg/m3) 2800 2800 2800 5 Actual Volume(mm3) 9296 13961 18679 6 Mass (g) 26.3 39.5 52.86 7 Ratio=SA/VA 3.69 2.45 1.80 Si no Geometric Parameters Gyroid 2to1 mm As-sintered Fins As-Cast Fins 1 Box Volume=LXBXH(mm3) 40X40X33 40X40X34 40X40X34.5 2 Surface Area (mm3) 33907 16526 16803 3 Material AlSi AlSi AlSiCu 4 Material Density(kg/m3) 2800 2800 2700 5 Actual Volume(mm3) 13974 24166 25007 6 Mass (g) 39.5 68.4 67.1 7 Ratio=SA/VA 2.43 0.68 0.67 images. Although the wavelength of visible light limits the resolution, it is enough for identifying the particles and pores in the sample produced by BJT/M [1]. Figure 3.15: Zeiss Axioscope 7 light optical microscope 33 3.7 Post-Processing 3.7.1 Hot Isostatic Pressing - HIP Hot isostatic pressing (HIP) is a method of fabricating materials in which an initial powder or preformed shape is simultaneously subjected to high temperatures and high isostatic pressures, utilizing a gas transfer medium as depicted in figure 3.16. Principally distinguishing HIP from other processing techniques is using gas as the pressure-transmitting medium to induce equivalent three-dimensional changes in the surrounding materials. Until recently, argon was used almost exclusively; however, presses that can process powders and bulk parts in oxygen-rich or other reactive atmospheres are now available [29]. Figure 3.16: Hot Isostatic Pressing schematic During this thesis, QIH 15L hot Isostatic machine from Quintus Technologies AB was used, depicted in figure 3.17. It has a furnace chamber of 170 mm x 290 mm, at a maximum pressure of 200 MPa and a maximum temperature of 1400 °C. It is possible to utilize either uniform rapid cooling (URC) or Uniform Rapid quench- ing (URQ) furnace using this machine. However, the selection of either furnace depends on the requirements and productivity. Due to microstructural differences, it appears that the optimal process parameters for hot isostatic pressing additively manufactured materials differ from those established for cast materials [62]. In spite of this, hot isostatic pressing can be used to reduce the porosity of various additively manufactured materials. 34 Figure 3.17: QIH 15L from Quintus Technologies AB HIP was done at the Application Center at RISE in Mölndal. As mentioned before, the machine was a QIH15L, as depicted in figure 3.17. The sample was handled with specialized gloves to avoid any carbon residue. Any pencil marking was also removed using acetone. Table 3.4: Parameters used during Hot Isostatic Pressing Parameters Values Dwell Pressure (MPa) 150 Heating rate °C /min 7 Dwell Temperature 520 Upper temperature tolerance (°C) 10 Lower Temperature tolerance (°C) 10 Dwell time (min) 120 Cooling rate (°C /s) 15 Hot isostatic pressing was performed at 520 °C for 2 hours at 150 MPa in an Argon atmosphere [63]. Uniform rapid cooling is used instead of uniform rapid quenching, to avoid cracking of the sample. The heating rate was 7°C /s while after the heating, the sample was gas-cooled at a rate of 15 °C /s. The temperature, pressure, and holding time for the sintered sample are given in table 3.4 and figure 3.18. 35 Figure 3.18: HIPing parameters utilized in this work. The Y-axis on the left show the temperature and the Y-axis on the right show the pressure. The X-axis shows time in seconds 36 4 Results and Discussion 4.1 Verifying the design guidelines Design guidelines mentioned in the Theory Chapter needed to be optimized before they could be incorporated into the final design. One of the reasons for this was that they were mainly referring to the studies of stainless steel and not aluminium alloys being completely different metals belonging to another family, and behaving differently. This effect is amplified due to changes in the process during the sintering phase with the use of ceramic powder during sintering as support. This meant that parts behaved differently when it came to deformation under gravity as well as shrinkage. 4.1.1 Artefact designs for verifying the guidelines The first step in verifying the design guidelines involved designing and producing the artefacts that will embody the required guidelines. Since the scope of the project is not just about understanding the design guidelines, but also evaluating the technol- ogy from the application point of view. It was decided to print the guidelines which are specific to the application that is heatsink. This was done using two sinter- ing processes which RICOH has under development. In one process they carry the sintering with ceramic powder used as support media this is referred to as ceramic- supported sintering /ceramic support henceforth. The second process is similar to the existing process and hence is called unsupported/convectional sintering hence- forth. After careful evaluation of the heatsink, it was decided to print Artefact 1: consisting of threaded screw holes with conventional sintering(1A) and with ceramic- supported sintering(1B), Artefact 2: vertical pins with convectional sintering(2A) and with ceramic-supported sintering(2B), Artefact 3: a bridge of varying spans without baseplate, Artefact 4: a bridge of varying spans with the baseplate. Arte- fact 5: A tree design consisting of a horizontal overhang, circular holes, and an angular overhang. For Artefact 1, it was decided to print the threads corresponding to the ISO metric. A thick pitch metric standard was used. All the printed screw-holes had a depth of 10mm. 6 printed threads were M2, M2.5, M3, M3.5, M4, and M4.5. This can be seen in the figure 4.1. 37 Figure 4.1: Artefact 1 screw Threads in metric standard(all dimensions in mm) For Artefact 2 a total of 12 vertical pins were printed in four different heights (H1=3mm, H2 =5 mm, H3=8mm, & H4=10mm) on a base plate that was 2mm thick. Each height consisted of 3 pins each having the same height. Pins of H1 had diameters of 1mm, 1.5mm, and 2mm respectively. Pins H2 had diameters of 1mm, 1.5mm, and 2mm respectively. Pins H3 diameters of 1mm, 2mm, and 3mm respectively. Pins H4 had diameters of 1mm, 2mm, and 3mm respectively. Figure 4.2: Artefact 2 Pins with various heights and diameters (all dimensions in mm) 38 For the bridges, it was decided to print two artefacts without base-plate (Artefact 3) and with base-plate (Artefact 4) having similar parameters. The thickness of the horizontal wall was kept at 3mm while the bridge legs were kept at 2mm thick constant. It was decided to have the minimum value of 4mm for the bridge span with constant increments of 2mm for every other span. Thus, resulting in a bridge with 10mm as the largest span. Knowledge learned from this was useful from the final design point of view. Details of the bridge artefacts can be seen in figure 4.4. Figure 4.3: CAD-designed artefact for bridges(all dimensions in mm) Figure 4.4: CAD-designed artefact for Bridge without base(all dimensions in mm) A tree design was created to evaluate the overhang thickness, length, and angle in a part. This can be seen in figure 4.5. The feature to evaluate the overhang thickness and length was designed as horizontal branches with various lengths and thicknesses. Horizontal branches were as follows 20mmX4mm, 15X3mm, and 10X2mm. The sym- metrical design was placed on the opposite side but a radius of 1mm was provided at the intersection of the branch and the main stem. This was done to understand how deformation varies after providing the radius. While the angles branches were 39 used to evaluate the overhang angle. It was decided to print 3 branches per side with an overhang angle of 60 degrees. The length and thickness of the branches varied similarly to the horizontal branches, with a radius of 1mm on one side incorporated to non on the other side. Figure 4.5: CAD design of the artefact for trees(all dimensions in mm) 40 The below table 4.1 shows the summary of the artefacts with details of evaluation, the scaling factor designed by RICOH for each artefact, and the method that was used for evaluation. Table 4.1: Table showing the summary of all the artefacts along with the scaling factors designed by RICOH various design features in each artefact were decided based on the current design of the heatsink to evaluate how feasible is it to print the critical features of the existing part without compromising the function and also how they can be further optimized. 41 4.1.2 Results from verifying the guidelines Printed artefacts were then evaluated to understand the printing feasibility limits of the various design features. To begin with the evaluation it is essential to understand the various directions that are building, re-coating, and binder dispensation. figure 4.6 shows the various directions for all the printed artefacts. Figure 4.6: Image showing the building, recoating, and binder dispensing direction for all the artefact designs Artefact 1:Threads were divided into two artefacts 1A: convectionally sintered and 1B: sintered with ceramic support. Both the artefacts were scaled similarly with 15% in X and Y and 17% in Z direction. (a) (b) Figure 4.7: Shows the threads after bolting. 4.7a refers to conventionally sintered threads and 4.7b refers to ceramic sintered thread In the case of threads, it was not possible to evaluate the design using laser scans since internal features on the thread cylinders were hidden from the laser beam. 42 Thus the best way to evaluate the thread was by screwing them with bolts corre- sponding to the ISO metric specifications of the thread. Bolts for M4.5 were not available. During the evaluation of 1B, it was found that all the bolts fit perfectly in the threads. In the smallest thread M2 the bolt didn’t go all the way to the bottom of the thread. This can be accounted to the trapping of excess powder internally. Whereas in the case of the 1A: bolts fitted in one specification lower than they were designed for. For example, the M4 bolt is fitted in an M4.5 thread hole. While the ease of screwing was the same as 1B. From this evaluation, it was clear that this technology can print intricate features like thread in both sintering processes. While the shrinkage due to the absence of ceramic support is higher than in the presence of ceramic support and thus needs to be compensated differently. The size difference between the two parts raised con- cerns about different densification in different parts. To check the densification of both the artefacts, these parts were later CT scanned and compared for porosities. This is found in section 4.5. Artefact 2:Pins were divided into two artefacts 2A: which was sintered using the convectional sintering and 2B: pins that was sintered in the pool of ceramic powder. Both the artefacts were scaled similarly with 15% in X and Y and 17% in Z direction. Artefact 2A among the pins with 1mm diameter only one which was processed properly was 3mm in height with a very slight angular deviation of 0.13 degrees in the YZ direction and 1.96-degree deviation in the XZ direction. Both pins with 1.5mm diameter printed with less than 1-degree deflection which was acceptable. All the pins in the last row consisting of D3X10mm, D3X8mm, D2X5mm, and D2X3mm printed successfully with an acceptable deflection as it can be seen in figure 4.8. It was also noticed that the diameters of all the pins were nearly the same as the nominal CAD dimensions with tolerances of positive 0.03mm which is deemed acceptable. The detailed report for this can be found in Appendix A.1. 43 Figure 4.8: Dimension evaluation on the laser-scanned data for pins sintered in unsupported sintering process (Artefact 2A) Artefact 2B among the pins with ceramic support none of all pins of diameter 1mm produced with a significant deflection in axis and were considered unacceptable. From the ones that were evaluated the best result was a 10X2mm pin which produced nearly perpendicular with a deflection as less as 1.69 degrees in XZ and 0.49 degrees in YZ, while 10X3mm had the second least deflection among all. The diameter of all the pins was larger than the nominal dimension by 0.10 to 0.15. Although all the pins were scaled using the scaling factor mentioned in the table 4.1, higher dimensions than the nominal suggest a lack of densification in part similar to the thread design, a detailed report of the pins Artefact 2B can be found in the appendixA.2. 44 Figure 4.9: Dimension evaluation on the laser scanned data for pins sintered in ceramic media (Artefact 2B) Artefact 3:Bridge which is varying bridge spans without base was divided into 3A: no compensation apart from the scaling, 3B: two largest spans compensated for gravity, and 3C: geometry was compensated by using Simufact (Hexagon) BJT/M module with optimized parameters for AlSi. In 3A and 3B manual scaling of 10% was applied before printing. All of them were sintered using the non-supported sintering process. Artefact 3B the bending of the legs internally was observed due to frictional de- formation which was highest on the outermost legs with higher tilting on the leg which was in connection with the largest span of 10mm. Although the compensa- tion was done for gravity there was still an overall sagging in the horizontal span of the bridge. Deflection as high as 0.33mm can be seen. This can be accounted to frictional deforamtion which causes the horizontal span to flex due to compressive force exerted on it along with the gravity effect combined causing the bending in the downward direction. The scaling factor used clearly underestimates the shrinkage by certain values. In the case of artefact 3A, no additional compensation was done apart from the normal scaling mentioned above. The results were very similar to artefact 3B apart from the local sagging which was seen in 10 mm and 8mm span. A maximum sagging of 0.51mm was seen in the Artefact. This can be seen in the figure 4.10a. Detailed reports of the comparisons can be seen in Appendices A.3 and A.4. 45 (a) (b) Figure 4.10: Shows the evaluation of the non-compensated vs compensated bridge geometry for gravity using the scanned data Artefact 3C was the Simufact compensated bridge geometry by RICOH. The defor- mation model, compensated model and Final produced & scanned geometry can be seen in figure 4.11. 46 (a) (b) (c) Figure 4.11: Artefact 3C Bridge using Simufact data: 4.11a Shows the deforma- tion predicted by Simufact, 4.11b Shows the compensation geometry provided by Simufact and 4.11c shows the scanned file for the final print using the compensated CAD data from Simufact It can be seen that Simufact predicted frictional deformation and deformation due to gravity along with shrinkage. However, the compensation model only accounted for the frictional deformation and shrinkage. This was investigated with RICOH who are in the developing parameters for aluminum in the Binder jetting module of the software and they said that the parameters which produce the compensated geome- try are in the early development stage and cannot compensate geometry against the gravitational deformations. The final produced and sintered part was then scanned and evaluated to check the accuracy of the compensations. It showed that there was an improvement in negating the frictional deformation, especially in the outermost legs but the inner legs looked overcompensated this is due to the fact that the effect is not uniform throughout the geometry and very much depends on the design feature and its position in the geometry for example outermost features experience more frictional deformation compared to the inner. This is not accounted for properly by Simufact. The assumption in overall shrinkage was more accurate than both 3A and 3B artefacts which can be seen in AppendixA.5. Due to confidentiality, this scaling factor could not be revealed. Since the horizontal spans were not compensated for gravity in the compensated CAD by software it resulted in very high local sagging of about 0.52mm as maximum. Since the compensation was done for friction there was no overall sagging. Artefact 4: Bridge with baseplate consisted of two versions 4A: which was sintered 47 using a conventional sintering process and 4B: wherein ceramic support was used in sintering. Both were scaled by 10% before printing. Artefact 4A has shrunk and deformed in a similar way to the bridges without base- plate i.e. Artefact 3 in one with conventional process. The only major difference that was observed was the frictional deformation of the legs was negated. This is due to the fact that there was a base-plate in the design which took care of this. This also meant that the deflection of the horizontal span was restricted to two long spans of 8mm and 10mm due to the gravitational effect on the two unsupported spans. This can be seen in figure 4.12a and 4.12b. The detailed report on the bridge with baseplate can be found in Appendix A.6. (a) (b) (c) (d) Figure 4.12: Image showing evaluation using scanned data for the bridge with base: without and with ceramic support respectively Artefact 4B sintered in ceramic powder compared to unsupported one behaved very differently. It can be seen from the figure 4.12c and figure 4.12d, that shrinkage in this component was not uniform. In the continuous columns such as vertical legs and the horizontal span, it shrunk at a higher rate while the hollow spaces between the legs and under the horizontal span were not allowed to reduce in volume by the ceramic powder occupying this space. Hence, the geometrical shape of the compo- nent changed to accommodate these changes. Due to lower shrinkage final size of the part is bigger than the nominal CAD, thus densification needs to be cross-verified. Detailed report of evaluation can be found in appendix A.7. For Artefact 5: Tree design first trial was done by sintering the artefact in this conventional sintering process. The result of this was that it failed and broke into pieces during the sintering as seen in figure 4.13a. It was learned that such a complex 48 geometry with a high aspect ratio was difficult to achieve using the normal process. In the second trial RICOH process for sintering was used, during this the tree was placed in the sintering box and the box was filled with the ceramic powder. The result from the sintering can be seen in figure 4.13b. (a) (b) Figure 4.13: The outcome using conventional sintering process 4.13aand The out- come using the ceramic-supported sintering process 4.13b From the evaluation following things were noted • The ceramic-supported sintering process allows for the printing of complex geometrics in the first attempt. • The deformation factor for this process varies significantly from the normal process • Deformation along the direction of gravity is neglected • Deformation in the continuous features(tree stem) in the part is different from the protruding features(branches) • Unlike the normal process where the origin of deformation is the centre of the bottom plan on the tree, in the case of the ceramic process the origin/centre of deformation is concentrated towards the portion in the body that acts as the clamping feature(longest branches) The tree was also evaluated by laser scanning and superimposing the geometry on the nominal CAD data. This revealed that there were a lot of deviations, especially in the branches. The top branch deviated as much as 2mm upwards fr