STUDY OF THIN-WALL INJECTION MOLDING

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By Guojun Xu, M.E. ***** The Ohio State University 2004

Dissertation Committee: Professor Kurt W. Koelling, Adviser Professor L. James Lee Professor Jose M. Castro

Approved by

Adviser Department of Chemical Engineering

ABSTRACT

Thin-wall injection molding has received increasing attention over the past few years due to economic and environmental concerns. However, due to the difficulties encountered in the thin-wall molding process, systematic investigation is lacking in machine performance, mold design/manufacture requirement, molding characteristics, computer aided engineering (CAE) simulation, part quality and part design criteria. Furthermore, the combination of viscoelastic materials, complex molding geometry and cyclic processing conditions has generated some problems, such as flow marks, polymer degradation, sink marks and warpage, under high-speed and high-pressure injection molding. So it is very important to design, operate, and control thin-wall molding optimally to guarantee part quality as well as reduce cost. In this study, alternate and synchronous dull and glossy flow marks, two surface quality problems, were studied. For the alternate flow marks, the effect of polymer rheology, mold geometry, operating variables, and mold surface coatings on the appearance of the flow marks was studied. The flow marks occurred above a critical wall shear stress, but disappeared at high injection speeds. Mold geometry and mold temperature were found to affect the wavelength and the width of the flow marks, while melt temperature did not have much effect. Slip was not the cause of the generation of ii

the alternate flow marks. For synchronous dull and glossy flow marks, the effect of operating parameters, mold geometry, and mold surface coatings on the flow marks was studied. The flow marks occurred above a certain flow front velocity, but were less visible as the mold temperature was increased. It was also found that mold surface coatings did not eliminate the flow marks. The generation of these flow marks was explained by an entry viscoelastic flow instability. Furthermore, thin-wall injection molding with micro-features was investigated. The filling length in microchannels was measured and compared with simulation. The heat transfer coefficient was found to be very sensitive to the filling length prediction. In order to investigate the effect of input properties on the simulation output, mold cavity pressure was studied. The goal was to understand the effect of pressure-dependent viscosity, heat capacity, heat transfer coefficient, juncture pressure loss and pvT-data on cavity pressure and pressure drop prediction, and evaluate the importance of each parameter. The cavity pressure and pressure drop were measured experimentally and compared. Furthermore, the method to improve the prediction accuracy was discussed to help design and predict. As the increasing use of plastics, the plastics waste has become a main concern. The final part of the research focuses on the mechanical and rheological properties of virgin and recycled high impact polystyrene materials. In this study, we describe our progress in evaluating the viability of reusing post-consumer and virgin polymer blends of HIPS from electronics equipment housings. Plastics reuse challenges are briefly reviewed, and experimental results, such as rheological properties, mechanical iii

properties, molecular weight and morphology of different blends, are summarized and discussed for reuse of HIPS. Finally, the study introduces a new approach to determine initial processing parameters for thin-wall injection molding of post-consumer resin.

iv

This dissertation is dedicated to my family.

ACKNOWLEDGMENTS

I would like to express sincere gratitude to my adviser, Dr. Kurt W. Koelling, for his priceless guidance, encouragement, and support throughout this work. I also would like to thank Dr. Julie Ann Stuart and Dr. Blaine Lilly for their instruction, encouragement and support. Special thanks go to Dr. L. James Lee for his considerable advice and help. I wish to thank Dr. David Tomasko, Dr. Jose Castro, and Dr. Robert Brodkey for their valuable suggestions and comments. I would like to thank Dr. Paula Stevenson for her proofreading and many helps during the past five years. Thanks also go to everyone who helped me in various ways, Paul Green, Leigh Evrard and Carl Scott. I would like to thank previous and current colleagues in the polymer research group. In addition, Micro Metallics Corporation and Nova Chemical, Inc. donated postconsumer and virgin polymers, Eastman Kodak Company loaned two molds, Dow Chemical donated polypropylene, 3M Company donated Dynamar 9613, and GenCorp Research donated a blender. The authors thank Professor Terry Gustafson and research assistants Tony Frost and Kristin Frost of the Chemistry Department at The Ohio State University for measuring the infrared and Raman vibrational spectra. The author thanks Dr. John Clay for the measurement of the molecular weight, and Michael Ferry, Tu Tran, vi

Sadu Prabowo, Andy Divine and Eric Mosser for help in measuring some physical properties. Finally, I would like to thank my parents for their continuing support through the years of my study and my wife, Xia Cao, for her understanding, support, and encouragement.

vii

VITA

September 25, 1967..Born - Cixi, Zhejiang, P. R. China September 1985 - July 1989.B.S. Chemical Engineering Zhejiang University Hangzhou, Zhejiang, P. R. China September 1989 - March 1992.M.S. Chemical Engineering Zhejiang University Hangzhou, Zhejiang, P. R. China September 1998 present....Graduate Research Associate The Ohio State University Columbus, OH

PUBLICATIONS 1. Guojun Xu and Kurt Koelling, "Flow Marks/Tiger striping during Thin-Wall Injection Molding of Polypropylene", J. Injection Molding Technology (Submitted). 2. Jose L. Garcia, Kurt W. Koelling, Guojun Xu, and James W. Summers, PVC Degradation During Injection Molding: Experimental Evaluation, Journal of Vinyl & Additive Technology (In press). 3. Christiana Kuswanti, Guojun Xu, Jianhong Qiao, Julie Ann Stuart, Kurt Koelling, and Blaine Lilly, "An Engineering Approach to Plastics Reuse", Journal of Industrial Ecology, 6, 125-35, 2003. 4. Guojun Xu and Kurt Koelling, "Flow Marks during Injection Molding", ANTEC, Nashville, TN, 566-70, 2003. 5. Guojun Xu, Jianhong Qiao, Christiana Kuswanti, Kurt Koelling, Julie Ann Stuart, and Blaine Lilly, "Characterization of Virgin/Post-consumer Blended High Impact viii

Polystyrene Resins for Injection Molding", J. of Applied Polymer Science, 84, 1-8, 2002. 6. Guojun Xu and Kurt Koelling, "Flow Marks during Injection Molding", ANTEC, San Francisco, CA, 521-5, 2002. 7. Guojun Xu and Kurt Koelling, "Study of Flow Marks during Thin-Wall Injection Molding", ANTEC, Dallas, TX, 604-7, 2001. 8. Guojun Xu, Jianhong Qiao, Christiana Kuswanti, Molly Simenz, Kurt Koelling, Julie Ann Stuart, and Blaine Lilly, "Insight into Reuse of High Impact Polystyrene from Post-Consumer Electronics Equipment Housing", IEEE International Symposium on Electronics and the Environment, San Francisco, CA, 348-53, 2000. 9. G. J. Xu, Y. M. Li, Z. Z. Hou, L. F. Feng and K. Wang, "Gas-Liquid Dispersion, Mixing and Heat Transfer in a Stirred Vessel", Can. J. of Chem. Eng., 75, 299-306, 1997. 10. Y. Li, G. Xu, M. Chen and K. Wang, "Slow Pelleting Coagulation of MBS Latex", Chem. Eng. Res. & Des., 75, 81-6, 1997. 11. Xu Guojun, Lianfang Feng, Yunming Li and Wang Kai, 'Pressure Drop of Pseuoplastic Fluids in Static Mixers', Chinese J. of Chem. Eng. (English), 5(1), 93, 1997. 12. Y. M. Li, M. W. Chen, G. J. Xu, and K. Wang, "Continuous Slow Coagulation of Polymer Latex in Series Agitated Vessels", 36th IUPAC International Symposium on Macromolecules, IUPAC MACRO SEOUL'1996, Korea, 6-p01-01, 597, 1996. 13. Y. M. Li, G. J. Xu, M. W. Chen, S. H. Ou and K. Wang, "Slow Pelleting Coagulation of Polymer Latex Emulsion", 36th IUPAC International Symposium on Macromolecules, IUPAC MACRO SEOUL'1996, Korea, 6-p01-02, 598, 1996. 14. G. J. Xu, Y. M. Li and K. Wang, "Particle Growth Kinetics for Seed Coagulation of Polymer Latex", 36th IUPAC International Symposium on Macromolecules, IUPAC MACRO SEOUL'1996, Korea, 6-p01-03, 599, 1996. 15. Hou Zhizhong, Feng Lianfang, Li Yunming, Xu Guojun, Wang Kai and Pan Zuren, "Power Consumption of Agitation in a Gas-liquid System" (Chinese), 7th National Conference on Chemical Engineering, Beijing, China, B54, 424, 1994. 16. Hou zhizhong, Li Yunming, Feng Lianfang, Xu Guojun, Wang Kai and Pan Zuren, "Study on Heat Transfer of Gas-liquid System in an Agitated Vessel" (Chinese), 7th National Conference on Chemical Engineering, Beijing, China, B53, 420, 1994. 17. Hou Zhizhong, Wang Kai, Feng Lianfang, Li Yunming, Xu Guojun and Pan Zuren, "Fluid/Wall Heat Transfer in an Agitated Gas-Liquid Reactor" (English), International Workshop on the Advances in Chemical Engineering, Hangzhou, China, 1994.

ix

18. Guojun Xu, Lianfang Feng and Kai Wang, "Pressure Drop and Friction Factor for non-Newtonian Fluids in Static Mixers" (English), International Workshop on the Advances in Chemical Engineering, Hangzhou, China, 1994. 19. Hou Zhizhong, Feng Lianfang, Li Yunming, Xu Guojun and Wang Kai, "Gas-liquid Dispersion and Mixing Properties of Different Impellers in an Agitated Vessel", China Synthetic Rubber Industry (Chinese), 18(3), 147-50, 1995. 20. Hou Zhizhong, Li Yunming, Feng Lianfang, Xu Guojun and Wang Kai, "Properties of Gas-liquid Dispersion in a Baffle-gassed Multistage Agitated Vessel", China Synthetic Rubber Industry (Chinese), 18(4), 218-20, 1995. 21. Guojun Xu, Zhangmao Wang and Gantang Chen, "Study of Axial Diffusion Coefficients and Distinguish of Particulate/Aggregative Fluidization", Chemical Reaction Engineering and Technology (Chinese), 10(3), 306-10, 1994. 22. Guojun Xu, Zhangmao Wang and Gantang Chen, "A Model of Fluid Flow and Particle Circulation in a L/S Fluidized Bed", Chemical Reaction Engineering and Technology (Chinese), 11(3), 277-83, 1995. 23. Guojun Xu, "Fluidized Polymerization Reactors", China Synthetic Rubber Industry (Chinese), 18(1), 40-2, 1995. 24. Li Yunming, Xu Guojun, Ou Shuhui, Chen Miwen and Wang Kai, "Slow Coagulation of Polymer Latex" (Chinese), Annual Conference on Polymers, Guangzhou, 1179-80, 1995. 25. Zhizhong Hou, Lianfang Feng, Yunming Li, Guojun Xu and Kai Wang, "Heat Transfer Properties in Aerated Agitated Reactor", China Synthetic Rubber Industry (Chinese), 18(6), 338-40, 1995. 26. Yunming Li, Guojun Xu and Jingjing Xu, "A Study of Particle Growth in Seed Coagulation of Polymer Latex" (Chinese), Annual Conference on Polymer, Guangzhou, 1175-6, 1995. 27. Guojun Xu, Yunming Li and Jingjing Xu, "Methods of Seed Coagulation of Polymer Latex" (Chinese), Annual Conference on Polymer, Guangzhou, 1177-8, 1995. 28. Zhangmao Wang and Xu Guojun, "A Study Expansion and Axial Diffusion in a Liquid/Solid Spouted Fluidization Bed", Chemical Reaction Engineering and Technology (Chinese), 12(2), 184-8, 1996. 29. Guojun Xu, Lianfang Feng, Yuming Li, and Kai Wang, "A Study of Pressure drop for Pseudo-plastics Fluids in Kenics Mixers", China Synthetic Rubber Industry (Chinese), 19(2), 97-9, 1996.

30. Deming Mao, Lianfang Feng, Guojun Xu and Kai Wang, "Effect on Control Volume and measured Points When the Beams Pass through Circular Media", Journal of Experimental Mechanics (Chinese), 11(1), 13-7, 1996. 31. Deming Mao, Lianfang Feng, Guojun Xu and Kai Wang, "Experimental Study on Agitator by LDA", Chem. Eng. J of Chinese University (Chinese), 10(3), 258-63, 1996. 32. Deming Mao, Lianfang Feng, Guojun Xu and Kai Wang, "Study of Spectral Analyses and Scales of Turbulence in Rushton Turbine", Chem. Eng. J of Chinese University (Chinese), 1996. 33. Yuming Li, Miwen Chen, Guojun Xu and Kai Wang, "Slow Pelleting Coagulation of Polymer Latex Emulsion", Chinese Chemical Letter (English), 7(3), 297-8, 1996.

FIELDS OF STUDY

Major Field: Chemical Engineering Minor Field: Polymer Processing Rheology Chemical Reaction Engineering

xi

TABLE OF CONTENTS

Page Abstract..ii Dedication...v Acknowledgments.vi Vita..viii List of Tables....xv List of Figures....xvii Chapters 1. Introduction....1 2. Literature review..12 2.1 Flow marks.12 2.1.1 2.1.2 2.2 Alternate flow marks.......13 Synchronous flow marks.....19

Experiment with micro-features and improvement of simulation accuracy during thin-wall injection molding.....21 2.2.1 2.2.2 Thin-wall injection molding with micro-features....21 Cavity pressure and its prediction....34

2.3 3.1

Reuse of HIPS.....37 Alternate dull and glossy flow marks.....45 3.1.1 Introduction..45 xii

3. Flow marks during thin-wall injection molding...45

3.1.2 3.1.3

Experimental.47 Results and discussion.50 3.1.3.1 Rheological characterization....50 3.1.3.2 3.1.3.3 3.1.3.4 3.1.3.5 Injection molding results...51 Morphology and crystallinity....56 Extrusion...57 Simulation.....57

3.1.3.6 Mechanism...59 3.1.4 3.2 3.2.1 3.2.2 3.2.3 Conclusion.......64 Introduction...65 Experimental67 Results and discussion.69 3.2.3.1 Rheological characterization.....69 3.2.3.2 3.2.3.3 3.2.3.4 3.2.3.5 3.2.4 Injection molding results..69 Morphology and crystallinity....71 Extrusion...72 Simulation.....72 Synchronous dull and glossy flow marks65

3.2.3.6 Mechanism....73 Conclusion...76

4. Experiment with micro-features and simulation accuracy improvement during thinwall injection molding......126 4.1 Thin-wall injection molding with micro-features.....126 4.1.1 4.1.2 4.1.3 4.1.4 4.1.5 4.2 4.2.1 Introduction...........126 Experimental.....127 Experimental results..129 Simulation results...132 Conclusions...........134 Introduction.......135 xiii

Cavity pressure and its prediction during thin wall injection molding135

4.2.2 4.2.3 4.2.4

Simulation.....137 Results and discussion...140 Conclusions...........145

5. Characterization of virgin/post-consumer blended high impact polystyrene resins for injection molding...190 5.1 5.2 Introduction......190 Experimental.193 5.2.1 5.2.2 5.2.3 5.2.4 5.2.5 5.2.6 5.3 5.3.1 5.3.2 5.3.3 5.3.4 5.3.5 5.3.6 5.4 Characterization of materials.193 Measurement of molecular weight193 Microscopy and spectroscopy194 Processing parameters for ASTM specimens....195 Physical properties of ASTM specimens...196 Application.197 Characterization of materials.198 Molecular weight.......199 Microscopy and spectroscopy199 Processing parameters for ASTM specimens200 Physical properties of ASTM specimens...200 Application203

Results and discussions....198

Conclusions......204

6. Conclusions and future work.....225 6.1 6.2 6.3 Flow marks.......225 Experiment with micro-features and simulation accuracy improvement.226 Reuse of HIPS..230

References..231

xiv

LIST OF TABLES

Table

Page

3.1 3.2 3.3 3.4 4.1 4.2 4.3

Relaxation time and zero viscosity at 200C....78 Viscosity-molecular weight......79 Average roughness of the dull and shiny regions.....80 Average roughness of the dull and shiny regions.....81 Orthogonal array of the simulation.147 Coefficients of Cross-WLF equation..148 Relative influence of each factor on peak cavity pressure at different injection speeds at 230C..149 Relative influence of each factor on peak cavity pressure at different injection speeds at 250C..150 Relative influence of each factor on peak cavity pressure at different melt temperatures for HDPE at 0.5/s151 Relative influence of each factor on maximum pressure drop at different injection speeds at 230C..152 Relative influence of each factor on maximum pressure drop at different injection speeds at 250C..153 Relative influence of each factor on maximum cavity pressure drop at different melt temperatures for HDPE at 0.5/s.154 xv

4.4

4.5 4.6

4.7

4.8

5.1 5.2 5.3 5.4 5.5 5.6

Molecular weight and polydispersity..206 Weight percentage blends..207 Mold design characteristics208 Processing parameters from C-MOLD...209 CMOLD parameters for film canister210 Tensile strength of film canisters.........211

xvi

LIST OF FIGURES

Figure 1.1 1.2 1.3 1.4 1.5 2.1 2.2 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8

Page
Difference between thin-wall and conventional injection molding....7

Typical molding problems (1).........8 Typical molding problems (2).........9 Typical molding problems (3)...........10 Environmentally conscious engineering system perspective........11 Alternate and synchronous dull and glossy flow marks.43 Thin-wall plate with microstructures.44 Alternate dull and glossy regions..........82 Comparison of viscosity vs. frequency at 200C......83 Comparison of complex viscosity of PP-C at 180, 200, and 220C.84 Comparison of elastic and viscous modulus at 200C..85 First normal stress difference vs. shear rate at 200C...86 The first normal stress difference of PP-C vs. shear rate at 180, 200, and 220C87 Transient extensional viscosity at 130C..88 Determination of relation time by one-mode Giesekus model.....89 xvii

3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 3.20 3.21 3.22 3.23 3.24 3.25 3.26 3.27 3.28 3.29 3.30 3.31

Flow marks of PP-C at different injection speeds....90 A typical example of the alternate dull and shiny flow marks.91 Effect of melt temperature on the wavelength ..92 Effect of mold temperature on the wavelength .93 The effect of mold thickness on the wavelength ..94 Effect of melt temperature on the width of the flow marks.95 Effect of mold temperature on the width of the flow marks96 The effect of mold thickness on the width of the flow marks..97 The starting of the flow marks, Vcri vs. melt temperature98 Effect of melt temperature on the transition velocity, Vtrans.99 Flow mark zone of PP-C100 Morphology of surfaces of dull and shiny regions.101 Gross melt fracture of the PP in extrusion.102 The wall shear stress versus apparent shear rate in the extrusion..103 Wall shear stress vs. percentage filled in the thin spiral mold...104 The critical wall shear stress at the middle of the gate at different melt temperatures...105 The critical wall shear stress at the middle of the gate at different mold temperatures...106 The similarity between extrusion and injection molding processes...107 Oscillating flow generates alternate flow marks....108 Frequency of the flow marks versus flow front velocity109 Synchronous dull and glossy regions.110 Comparison of viscosity vs. frequency at 180C...111 Comparison of Elastic and viscous modulus at 180C..112 xviii

3.32 3.33 3.34 3.35 3.36 3.37 3.38 3.39 3.40 3.41 3.42 3.43 3.44 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11

First normal stress difference vs. shear rate at 180C113 Extensional viscosity vs. time at 100C.114 Synchronous dull and shiny flow marks of HDPE2...115 Effect of melt temperature on wavelength.116 Effect of mold temperature on wavelength117 Effect of melt temperature on Vcri.118 Morphology of dull and shiny region of HDPE2...119 Flow curve of HDPE2 in extrusion120 Different extrudate irregularities at different wall shear stresses...121 Critical wall shear stress vs. percentage filled at different melt temperatures...122 Critical wall shear stress vs. percentage filled at different mold temperatures..123 Pulsating flow generates synchronous flow marks.124 Frequency of flow marks vs. Flow front velocity125 The long rectangular mold base with a disk-like insert...155 The rectangular mold bases with a disk-like insert156 The disk-like mold insert which contains microchannels..157 SEM picture of the a microchannel158 Dynamic viscosity of polypropylene..159 Dynamic viscosity of PMMA.160 SEM of a micro-channel.161 Measured filling lengths in microchannels for PMMA in the long mold..162 Measured filling lengths in microchannels for PP in the long mold..163 Measured filling lengths in microchannels for PMMA in the long mold..164 Measured filling lengths in microchannels for PP in the long mold..165 xix

4.12 4.13 4.14 4.15 4.16 4.17

Measured filling lengths in microchannels for PP in the short mold..166 The cavity pressure profile in the long mold and the short mold...167 The filling length vs. Fourier number....168 The effect of packing stage on filling lengths........169 The effect of holding pressure on filling lengths........170 Comparison of the filling lengths between the simulation and experiment with constant heat transfer coefficients. Main flow heat transfer coefficient=25000 W/m2.K...171 Comparison of the filling lengths between the simulation and experiment with constant heat transfer coefficients. Main flow heat transfer coefficient=2000 W/m2.K...172 Comparison of the filling lengths between the simulation and experiment with variable heat transfer coefficient173 Schematic of the mold with thickness of 1 mm.174 Heat capacity of HDPE and PS..175 Specific volume of HDPE..176 Specific volume of PS177 Experimental and fit viscosity vs. shear rate/ frequency for PS.178 Experimental and fit viscosity vs. shear rate/ frequency for HDPE...179 Comparison of cavity pressure with/without the effect of pressure on specific volume180 Comparison of cavity pressure with/without the effect of pressure on viscosity..181 Comparison of cavity pressure with different heat transfer coefficients....182 Comparison of cavity pressure with constant Cp and temperature-dependent Cp...183 Comparison of cavity pressure with/without juncture loss.184

4.18

4.19 4.20 4.21 4.22 4.23 4.24 4.25 4.26 4.27 4.28 4.29 4.30

xx

4.31

Pressure profiles right after the gate and at the end of the cavity at the injection speed of 76.2 mm/s and the melt temperature of 230 and 250C...185 Pressure profiles right after the gate at the melt temperature of 230C with different injection speeds....186 Pressure profiles at the end of the cavity at the melt temperature of 230C with different injection speeds....187 Comparison of experimental and predicted pressure drop at the injection speed of 12.7 mm/s...188 Comparison of experimental and predicted pressure drop at the injection speed of 508 mm/s..189 Film canister........212 Comparison of the viscosity curves for post-consumer HIPS and virgin HIPS at 220C..213 Viscosity of Huntsman PS 702 blends with different percentages of postconsumer resin at about 200C...214 Viscosity of Nova PS 3350 blends with different percentages of post-consumer resin at about 200C...215 The images of different blends from ESEM (The length of the scales in the figures are 2 m)......216 Raman spectroscopy of injection-molded post-consumer and Huntsman PS 702......217 Infrared vibrational spectra of injection-molded post-consumer and Huntsman PS 702..218 Average Ra for six blends of Huntsman PS 702.219 Average Wa for six blends of Huntsman PS 702....220 Tensile strength and tensile modulus vs. weight percentage of virgin resin...221 Flexural strength and flexural modulus vs. weight percentage of virgin resin...222 Impact strength and tensile modulus vs. weight percentage of virgin resin...223 Meshing model of the film canister224 xxi

4.32

4.33 4.34 4.35 5.1 5.2

5.3

5.4

5.5

5.6 5.7 5.8 5.9 5.10 5.11 5.12 5.13

CHAPTER 1

INTRODUCTION

Among the large number of polymer processing operations, injection molding has found the widest application for making articles which could be put to direct use. Because of the superior manufacturability and the high degree of freedom of the form of plastics products, injection molding is one of the most widely used processes for processing plastics. In injection molding process, the polymer melt flows through a runner system and gates to fill the mold cavity. When the filling is completed, more melt is packed into the mold to compensate for volume shrinkage. The cooling stage follows until the melt solidifies. Finally the part is ejected from the mold. Thin-wall injection molding (TWIM) is conventionally defined as molding parts that have a nominal wall thickness of 1 mm or less and a surface area of at least 50 cm2 [Whetten and Belcher, 1994; Fasset, 1995]. Thin wall is relative, however. It also can be named thin-wall as the flow length/thickness ratio is above 100 or 150 [Mahishi, 1998; Maloney and Poslinski, 1998]. TWIM has been paid more and more attention, 1

especially in computer, communication and consumer electronic (3C) industries, due to economic and environmental concerns. The reason is that thin-wall molded parts could be made lighter, more compact, less expensive, and quicker because of fast cooling [Smialek and Simpson, 1998]. New environmental regulations require less plastic to be used at the source or in the initial stage of manufacturing [Miller, 1995]. Thus, TWIM is a viable option for reducing the weight and size of plastic components. The difference between conventional injection molding and TWIM is shown in Fig. 1.1. The solidified skin layers are about 0.25 mm regardless of part thickness [Fasset, 1995]. It means that the flow channel is very narrow and thus flow resistance is very high in TWIM. Reducing flow resistance can be reached by increasing the melt or mold temperature, reducing melt viscosity (increasing melt index), increasing injection pressure, or injection speed [Fasset, 1995; Belcher and Hoenig, 1991]. However, high melt temperature may cause degradation and increases cooling time which are unacceptable. A rise in melt index shows a decrease in physical properties [Belcher and Hoenig, 1991]. Therefore, high injection speed is preferred, and extremely high

injection pressure, 200-250 MPa (30,000-40,000 psi), is required [Colangelo and Tremblay 1997]. Due to the thin part, cooling is fast. Thus the combination of the fast cooling and high melt velocity (short fill time) significantly reduces the cycle time. The typical cycle time of TWIM is 6-20 seconds while the cycle time for conventional injection molding is 40-60 seconds [Selden, 2000]. The shrinkage is also low because of the reduced part thickness [Delbarre, et al., 1991]. TWIM is characterized as high flow rate, high pressure, high shear rate, high viscous heating, fast cooling and fast shrinkage. 2

However, TWIM has some disadvantages. Due to the rapid cooling of the polymer melt, the operating window becomes narrower as the part becomes thinner [Bozzelli, et al., 1997; Coxe, et al., 2000]. Specialized material is also required to balance the trade-off between processability and physical properties [Cha and Lai, 2000], which means material should both flow easily (high melt index) and retain good physical properties. TWIM also makes design and process control more complicated. It is a big challenge to fill the mold with a high flow length/thickness ratio at a high speed under high pressure. For example, an additional accumulator is needed to maintain high pressure at a short fill time. However, the operation of the accumulator affects the molding stability [Chen, et al., 2000]. More robust control systems are required to control the molding precisely and with a short response time [Selden, 2000; Hatch, et al., 2001]. High injection pressure also needs high clamp tonnage which increases the capital investment of equipment. Processing, material, tooling, and machine interact with each other and greatly affect the end results. For TWIM, systematic investigation about machine performance, mold design/manufacture requirement, molding characteristics, computer aided engineering (CAE) simulation, part quality and part design criteria is required [Chen, et al., 2000]. However, the study is lacking due to the difficulty of thin-wall molding process. Furthermore, the combination of viscoelastic materials, complex molding

geometry and cyclic processing conditions has generated some problems [Schmidt, 1998], such as flow marks, polymer degradation, high residual stress, sink marks and warpage, under high-speed, high-pressure injection molding. So it is very important to 3

design, operate and control thin-wall molding optimally to guarantee part quality as well as reduce cost. In this study, some issues, such as surface flow marks, thin-wall

injection molding with micro-features, mold cavity pressure and its prediction, and reuse of post-consumer resin, are investigated. Part appearance is one important criterion for assessing part quality because it can be quickly evaluated. There are many aesthetic indictors that include warpage, surface finish or gloss level, flash, sink marks, short shot, color, burns, bubbles, transparency, pecks, scratching, stress marks, splay, drag, streaks, etc. [Salamon, et al., 1998]. Some typical surface problems are shown in Figs. 1.2-1.4 [C-Mold design guide, 1998]. Flow marks are one of these problems created during injection molding. They exhibit different levels of gloss on the surface of molded parts. These surface defects are related to the melt flow and are thus called flow marks. They are also referred to as tiger stripes, striping, halos, slip-stick, haze patterns, webs, chatter marks, blush or rings [Salamon, et al., 1998; Dharia, 1999]. These flow marks occur especially on automotive exterior parts and are very difficult to mask with paint. The defects limit the use of many polymers in unpainted applications. In this study, two types of flow marks, alternate dull and glossy and synchronous dull and glossy flow marks, are studied. In recent years, the fabrication of polymer-based micro-components for optical and biomedical applications has been paid more and more attention. The polymer material is favored because of its low cost, good bio-compatibility, high optical clarity, and high impact strength compared with silicon or glass. Micro-injection molding has the potential for economical mass-production. It usually combines various lithography 4

techniques and injection molding [Weber and Ehrfeld, 1999]. Two types of micro-parts are available: micro-sized parts whose delivery system including the runner and sprue is much larger than the parts themselves and regular-sized parts with micro-features. Micro-injection molding (MIM) is the injection molding of plastic parts with structure dimensions in the micron or sub-micron range. The replication of the micro-features is an important issue and it depends greatly on the size, aspect ratio and covered area [Weber and Ehrfeld, 1999]. This study focuses on the thin-wall injection molding with micro-features by experiment and numerical simulation. The filling lengths in

microchannels are simulated and compared with experimental results. Because the predicted filling lengths in microchannels are very sensitive to the heat transfer coefficients selected, it is necessary to study the effect of input property models on the simulation outputs. We further study how the input properties affect the simulation output in thin-wall injection molding. The output we choose is mold cavity pressure. Injection mold cavity pressure is an important injection molding parameter. It is

regarded as a good indicator of molded part quality and injection machine control performance. Cavity pressure not only indicates the material condition in the mold but also affects the microstructure and part quality. Computer Aided Engineering (CAE) is a common practice nowadays to help design, process, optimize, and troubleshoot thin-wall injection molding processes. However, almost all users prefer better accuracy of CAE simulation because large discrepancy between simulation and experiment may occur. The difference may result from simplifying some important physical, thermal or other properties, such as the pressure-dependent viscosity, variable heat transfer coefficient, 5

and variable material properties. The goal of the study is to understand how pressuredependent viscosity, heat capacity, heat transfer coefficient, juncture pressure loss and pvT-data affect pressure prediction, and the importance of each parameter. methods to improve the prediction accuracy will also be discussed. As the increasing use of plastics, the plastics waste has become a main concern. Environmentally conscious design and manufacturing is a strategic and competitive practice. The reduction of material consumption is a big challenge for industrial Then

ecology. The attention paid to polymer recycling has increased in the past decade. However, the life cycle trade-offs between collection, disposal, use of recycled plastics, recyclability, reduction of process wastes, energy consumption, yields, and product performance are complex, as shown in Fig. 1.5 [Stuart, 1999]. Life cycle assessment and life cycle production planning models are very useful tools to analyze these tradeoffs. However, resin degradation characteristics and potential design details are required in order to apply these tools. Plastics recycling is important because more efficient re-use of materials will reduce the quantities of plastics sent to landfills as well as reduce raw material extraction. Waste prevention practices are increasingly

significant and are increasingly encouraged with the advent of take-back legislation [Gamalski, 1996; Meffert and Kirchgeorg, 1997; Hubschman, et al., 1995]. In this study, characterization and reuse of post-consumer resin (PCR) in a thin-wall application is addressed.

Thin-Wall Part (1 mm)

Conventional Part (3 mm)

Flowing Core 0.5 mm

Flowing Core 2.5 mm

Solid Skin 0.25 mm

Solid Skin 0.25 mm

Fig. 1.1. Difference between thin-wall and conventional injection molding.

Black specks

black streaks

Brittleness

Burn marks

Delamination

Dimensional variation

Flash

Flow marks

Fig. 1.2. Typical molding problems (1) [C-Mold design guide, 1998].

Hesitation

Jetting

Ripples

Silver streaks

Fig. 1.3. Typical molding problems (2) [C-Mold design guide, 1998].

Fish eyes

Sink marks

Weld lines or Meld lines Fig. 1.4. Typical molding problems (3) [C-Mold design guide, 1998].

10

Virgin Materials

Degradation Reusable Material Model Content Model

Raw Materials
Energy

Concurrent Product, Process & System Design for Assembly/Reuse/End-of-Use Manufacturing/Services

Rework

Distribution Consumer Repair Recycling Downcycling Disposal

Fig. 1.5. Environmentally conscious engineering system perspective [Stuart, 1999].

11

CHAPTER 2

LITERATURE REVIEW

2.1 FLOW MARKS The application of injection molding has greatly increased in recent years. However, there is a conflict between the high quality of exterior appearance and short cycle time. Injection molding sometimes creates several kinds of surface defects during processing, differing levels of gloss on the surface of molded parts. The surface defects are related to the melt flow and are thus called flow marks. They are also referred to as tiger stripes, striping, halos, slip-stick, haze patterns, webs, chatter marks, blush or rings [Salamon, et al., 1998; Dharia, 1999]. These matte areas occur on one or both sides of parts. When they occur on both sides, those on one side of the parts are in phase or out of phase with those on the other side of the parts. According to Yokoi [1994a], three kinds of flow marks are classified according to the surface conditions of flow marks. They are (1) micro-grooved zones like LP records, (2) synchronous dull and glossy surfaces and (3) alternate dull and glossy surfaces. The kind of flow mark with microgrooved zones like LP records is also well known as wave-like flow marks. Flow marks

12

may occur on center-gated parts made with multi-phase polymer systems, including rubber modified polymers, thermoplatic olefin (TPO), blends (HIPS, PC/ABS), copolymer (ASA, ABS, EP), and semi-crystalline polymers (LDPE, HDPE, PP/Talc). They may also occur on edge-gated parts [Salamon, et al., 1998]. Here we discuss two types of these surface defects: those associated with the flow instability and those associated with the change of flow front velocity [Salamon, et al., 1998]. One type of flow mark is characterized as alternate dull and glossy surfaces where flow marks on one side are out of phase with those on the other side of the part, as shown in Fig. 2.1(a). Another type of flow mark is characterized as repeated dull and glossy regions where a dull/glossy zone on one side corresponds to a dull/glossy zone on the other side, as shown in Fig. 2.1(b). These flow marks cause surface defects that occur especially on automotive exterior parts. Flow marks can be very difficult to mask with paint due to the change in porous structure of dull and shiny regions. The defects also limit the use of many polymers in unpainted applications.

2.1.1 Alternate Flow Marks (AFM) One type of flow mark is characterized as alternate dull and glossy surfaces [Yokoi, 1994a]. Flow marks were observed as early as in 1961 and some work has been done to explain and eliminate them [Yokoi, 1994b; Chang, 1996a; Chang, 1996b; Hobbs, 1996; Heuzey, et al., 1997; Hamada and Tsunasawa, 1996; Dharia, 1999; Bulters and Schepens, 2000; Grillet, et al., 2000; Charmeau, et al., 2000; Xu and Koelling, 2001; Jayaraman, et al., 2002]. In the literature, the effect of operating variables, physical 13

properties of plastics, and mold geometry has been discussed. Operating variables include injection speed, injection pressure, melt temperature, and mold temperature. The physical properties cover different kinds of polymers, the rheology of polymer melt and molecular weight distribution, while the mold geometry includes gates, mold thickness and different molds. Moreover, several kinds of mechanisms have been proposed to explain the generation of flow marks. However, little has truly been understood on why flow marks occur and how to predict and eliminate them. Furthermore, the results in literature sometimes conflict with one another. Alternate flow marks were typically observed in LDPE and PP/talc and were caused by a cyclic unsymmetrical flow front [Yokoi, 1994b]. There seemed to be a clear correlation between the shear stress level on the cavity wall and the flow mark generation region, and thus the flow marks were thought to occur due to the melt fracture on the cavity surface. The flow marks may involve wall slip, though the author did not state this. Chang used a slip mechanism to explain the surface defects [Chang, 1996a; Chang, 1996b]. When ASA polymer melt flow meets a thickness change from gate to cavity, slip can occur if the melt has low adhesion (friction) to the mold surface due to a low die swell. The slip can initiate a melt flow instability. Thus, this kind of flow instability causes flow marks with alternate dull and glossy regions. It was shown that recoverable shear strain, shear stress and the coefficient of friction between the melt and mold were key controlling factors for the generation of flow marks. It was also found that the higher the ratio of step size, the more severe the degree of flow marks.

14

In a study of several blends of BPA polycarbonate and ABS resins, Hobbs found that at higher injection rates, the flow marks were more continuous and pronounced [Hobbs, 1996]. The study indicated that stick/slip flow at high wall shear stresses created the flow marks and wall slip was worsened by lower friction coefficients. It was found that wall slip first occurred on one face of the mold. When the melt front flow chattered across the surface, high frequency ripples developed. This kind of slip

generated a distortion in the velocity gradient across the flow channel and caused the flow front to oscillate back and forth. The flow marks were formed by dragging the partially solidified melt across the mold surface. It was hypothesized that wall slip is associated with some kinds of flow marks. However, Heuzey, et al. found no obvious relationship between the wall slip and flow marks [Heuzey, et al., 1997]. Using linear polyethylene, they found that one of the resins did not slip in capillary flow experiments. Furthermore, coating on a mold wall did not affect the occurrence of flow marks. They concluded that wall slip does not affect the occurrence of the flow marks. They believed that three main factors were involved in the occurrence of alternate dull and glossy flow marks: the surface cohesive strength of the semi-solidified polymer, the adhesion between the solid layer and the mold, and the high shear stress in the melt near the wall. The generation of flow marks was due to the filamentation and stretching of semi-solidified materials. In the thin-wall injection molding experiment, flow marks with alternate dull and glossy regions were studied using PC and ABS blend [Hamada and Tsunasawa, 1996]. It was found that when no flow marks occurred, PC and ABS flowed in steady, laminar 15

motion with a normal fountain flow and in layers due to its low viscosity. However, when flow marks occurred, the PC and ABS flowed with oscillation. So, the center of the flow moved in the direction of the mold thickness. This kind of abnormal fountain flow happened because of the high viscosity of resin flow. Under this condition, high shear stress is applied on the PC and ABS. The result was that the PC phase at the tip of the flow front might be broken and then PC and ABS coexisted, causing cloudy (dull) regions. In a study of binary blends of polypropylene and ethylene co or ter-polymers [Dharia, 1999], it was concluded that the increase in built-in stress between the skin and core at the melt front increased surface defects. The study showed that the tendency was the combined effect of rapid stress build-up and slow recovery. It was also found that flow marks were caused by a melt flow instability and the inability of melt to recover from the stress change at the flow front. More recently, Bulters and Schepens visualized the mold filling process by a layered block of PP with contrasted colors. It was found that the flow front was unstable and the black layer broke through at the surface. They claimed that flow marks resulted from a flow front instability [Bulters and Schepens, 2000]. Furthermore, Grillet, et al. conducted a finite element simulation for a very strain hardening fluid and a very strain softening fluid. After the steady numerical

calculations, a linear stability analysis was performed and it was found that the most unstable eigenvector was an oscillatory, swirling flow near the stagnation point at the free surface [Grillet, et al., 2000]. Alternate flow marks of TPO blends were studied by Jayaraman, et al. [2002]. The disperse phase morphology was analyzed in detail. It was 16

found that the rubber particles in the out of the flow mark region were highly stretched and had a high aspect ratio, while the rubber particles in the flow mark region were less stretched and had a low aspect ratio. It was concluded that the flow marks occurred in the long spans of the unbalanced flow front. The effect of mold geometry and processing variables was studied by Chang [Chang, 1996b]. It was found that the mold thickness and mold surface temperature were the controlling factors on flow marks. It was also found that a larger thickness ratio caused more severe flow marks, and an increase in mold temperature decreased the rank of defect severity. Though the increase of injection speed and pressure worsened the flow marks, their effects were not as important as the above two factors. When rubber levels decreased, surface appearance improved because of the combination of the effect of recoverable shear strain and shear stress. As for the effect of carbon black loading, surface defect severity increased with the increase of carbon black levels. Chang concluded that the higher the melt elasticity, the better the surface appearance. However, the surface appearance was improved by increasing the coefficient of friction or decreasing the lubricant level. Hobbs found that with the increase of the injection speed, the flow marks were more continuous and pronounced [Hobbs, 1996]. The trend of the results is consistent with Changs work [1996b]. High surface roughness values of compound lowered the coefficient of sliding friction, increased stick/slip flow, and thus reduced gloss value. In the injection molding for processing LLDPE and HDPE, flow marks with alternate dull and glossy regions were formed [Heuzey, et al., 1997]. It was found that 17

the flow marks were affected by mold surface finish. In their experiment, injection rate was the most important factor affecting the flow marks. The severity of flow marks increased with the increase of the injection rate. It is consistent with Changs [1996a; 1996b) and Hobbs [1996] results. However, wall slip did not lead to in the generation of flow marks. It was found that coating had no great effect on flow marks. However, surface defects were amplified by adding silicone oil. It was also found that mold thickness had a small effect on flow marks, but the observation was different from that of Chang's work [Chang, 1996b]. In the thin-wall injection mold, flow marks were generated when processing a PC/ABS blend [Hamada and Tsunasawa, 1996]. It was found that low cylinder

temperature, low mold cavity temperature and high injection speed were the factors generating flow marks. The gate design was also found to be important. Dharia found that the most important factor producing flow marks was lower mold temperature [Dharia, 1999]. The second important factor was injection pressure, while the third important factor was injection speed. It was found that the high injection speed generated more pronounced flow marks. In all cases, even moderate backpressure could reduce flow marks. Although experimental evidences show that oscillating instability occurred in mold filling [Chang, 1996a; Hamada and Tsunasawa, 1996; Bulters and Schepens, 2000], there is disagreement on why the instabilities occur and where it originates. Furthermore, it is widely known that elastic instabilities occur upstream of a contraction, such as a capillary die or slit die, in extrusion or spinning. [Piau, et al., 1988; White, et 18

al., 1987; Boger and Walters, 1993; Koelling and Prudhomme, 1991]. However, the researchers studying the flow marks have not studied the entry instability and its relationship to the flow marks, although Dharia already noticed the similarity between extrusion and injection molding processes [Dharia, 1999]. In this study, we investigate the effect of operating parameters and different polymer melts on the wavelength, width, and gloss variation of the flow marks. Then several methods to reduce the flow marks are discussed. We study the correlation between gross melt fracture in extrusion and alternate dull and glossy flow marks in injection molding. Then, an entry viscoelastic flow instability mechanism is proposed to explain the alternate flow marks.

2.1.2 Synchronous Flow Marks (AFM) Although alternate dull and shiny flow marks and wavelike flow marks have been studied in the literature [Tredoux and Satoh, 1999; Tredoux, et al., 2000; Yokoi, et al., 1994a; Yoshii, et al., 1993; Yoshii, et al., 1996; Lee and Mills, 1994], little attention has been given to synchronous flow marks [Yokoi, et al., 1994c; Salamon, et al., 1998]. The synchronous dull and glossy flow marks usually occur in high viscosity PP with a narrow gate [Yokoi, et al., 1994c]. A glass-inserted mold was used to visualize the flow front during the mold filling process, and homopolymer PP was used in their experiment. It was observed that synchronous dull and glossy flow marks were generated, but no flow marks were observed on the glass surface or polished cavity surface. It was found that at high injection rates, many thin and narrow flow marks occurred [Yokoi, et al., 19

1994c]. The gate shapes greatly affected the generation of flow marks and the flow marks varied dramatically with the front flow velocity. At higher melt temperature, flow marks became thinner. At the melt temperature of 240C, flow marks did not show up where the variation of gate pressure and flow front velocity was small. They explained that near the gate, the flow resistance was high causing the flow quantity to decrease. The gate pressure was increased during the filling, while the melt velocity was decreased at the flow front and the melt was cooled down. On the other hand, the gate was frozen and the flow resistance was increased. Therefore, the melt at the flow front was

accelerated with the release of the high gate pressure. At that moment, the resin near the flow front underwent cooling, so the transcription precision decreased in the subsequent fountain flow process. Then the flow marks were formed. In the injection molding of HIPS in a rectangular mold with a center-gate, halos similar to flow marks with synchronous dull and glossy regions were formed [Salamon, et al., 1998]. It was found that the temperature gradient between any two zones was the cause of the formation of the halos. It was also proposed that the temperature gradient must decrease in the direction of the flow in order to form the halo and the halo could be reduced by a design that minimizes the heat losses to reduce the temperature gradient. It was also shown that the mold temperature was significant in controlling the halo. When the mold temperature was high enough, the halo did not occur because the surface stresses relaxed and recovered. However, the halo always appeared when the mold temperature was set below the annealing temperature. With the increase of injection rates, the halos became more prominent and the diameter of the halos became larger 20

because the thickness of the skin layer was reduced. It was also found that the part surface was rougher in the halo region, and the valleys in the halo region were aligned with the flow direction. They proposed that the stress was rapidly decreased for the polymer melt from a colder zone to a hotter zone, causing rapid reduction of strain rate. This reduction of strain rate therefore increased the size of the melt and the melt must wrinkle or fold to respond to this increase because the melt was confined in the mold cavity. The wrinkles or folds were aligned with the flow direction and quickly solidified without relaxation when contacting the cold mold surface, causing flow marks to form. This study considers the effect of operating parameters and mold surface coatings on flow marks of different polymer melts. A possible correlation between gross melt fracture in extrusion and synchronous dull and glossy flow marks in injection molding is discussed. Finally, a possible mechanism is proposed.

2.2 EXPERIMENTS WITH MICRO-FEATURES AND IMPROVEMENT OF SIMULATION ACCURACY DURING THIN-WALL INJECTION MOLDING 2.2.1 Thin-Wall Injection Molding with Micro-Features In recent years, the fabrication of polymer-based micro-components for optical and biomedical applications has been paid more and more attention. The polymer material is favored because of its low cost, good bio-compatibility, high optical clarity, and high impact strength compared with silicon or glass. Micro-injection molding has the potential for economical mass-production. It usually combines various lithography techniques and injection molding [Weber and Ehrfeld, 1999]. Two types of micro-parts 21

are available: micro-sized parts and regular-sized parts with micro-features. Microinjection molding (MIM) is the injection molding of plastic parts with structure dimensions in the micron or sub-micron range. Micro-injection molding process meets the requirement of cost-effective replication in large scale series. Different small or micro components with the following specifications can be injection molded [Hanemann, et al., 1997a]: Plate-shaped microparts with microstructures of any lateral form. Volume of the standard substrate base plate: 20602 mm3 (widthlengthheight). Microstructure height up to 1.6 mm. Smallest wall thickness down to 30 m. Smallest structure detail 0.2 m. Aspect ratio up to 30:1. Suitable materials: PMMA, PC, PSU, POM, PA12, PEEK, etc. Usually, micro structured mold inserts are made by special processes and then attached to standard molds [Piotter, et al., 1997]. The critical dimensions which can be produced by micro-injection molding in good shape are mainly determined by aspect ratios. Common microstructured products such as CDs and DVDs could not be

compared with LIGA microstructures with aspect ratios of ten to 600 [Piotter, 1997]. Modification of the molding machinery, the tools construction, and the molding operation is demanded to injection molding of microstructures with high aspect ratios.

22

The main difference between thin-wall injection molding and micro-injection molding is described in detail as follows: a. Mold Technology For thin-wall injection molding, high speeds and high pressures can make mold plates flex. It may cause flash or thicker wall sections of molded parts. Thus thick and strong molds are required in thin-wall injection molding to resist high pressure. TWIM also requires relatively large and/or multi gates for easier mold filling. More ejection pins are needed because parts are tightly packed. Larger ejection pins are used to avoid part distortion. Sometimes vacuum evacuation is recommended to minimize weld lines and possible burning of compressed gas [Fasset, 1995]. However, usually venting along the parting line combining the venting of ejectors and core pins can solve this problem. In injection molding of microstructures, micro structured mold inserts are made and then attached to standard molds. The mold cavity can be prepared by LIGA process or more traditional processes such as micro-turning, micro-sparking and laser-erosion [Piotter, et al., 1997]. LIGA process is a relatively new process to produce molds or cavities. Typically, micro mold inserts have high aspect ratios, especially from LIGA process. Parallel plane walls and lacking of injector slope make demolding difficult. However, multi-stepped master structures can be produced by inclined x-ray exposure, two-stepped resist structures, or the combination of several microstructuring techniques [Piotter, et al., 1997]. These techniques generate pretty smooth surface (roughness is smaller than 10 nm). Molds must meet high demand, such as accuracy requirement.

23

Because conventional venting through parting planes or gaps is impossible for microstructures due to the blind holes in microstructures [Hanemann, et al., 1997a], venting is a problem. Compressed hot air may burn the polymer, so compressed air must be evacuated by a vacuum pump. The mold inserts should be carried out certain number of shots in practice. To avoid damage, it is wise to reduce stress on the mold insert. The variothermal heating is a good choice in this point because high temperature lowers viscosity and makes the mold inserts easy to be filled. Reducing injection pressure or holding pressure is a choice. Wear is another problem. It is reported that wear did not occur after 1000 shots for LIGA mood inserts made from nickel or nickel-cobalt [Piotter, et al., 1997]. b. Machine Technology For thin-wall injection molding, high injection speed, 500 mm/s, is preferred and extremely high injection pressure, 200-250 MPa, is required [Colangelo and Tremblay, 1997]. The purpose is to reduce flow resistance caused by narrow flow channels. High clamping force is also required because of high pressure. High clamp tonnage means high capital investment of equipment. Due to the thin part, cooling is fast. Thus the combination of the fast cooling and high melt velocity significantly reduces the cycle time. Precise control is required to get good surface finishing. TWIM also makes design and process control more complicated. It is a big challenge to fill the mold with a high flow length/thickness ratio at a high speed under high pressure. For example, an accumulator is needed to maintain a high pressure at a short fill time. More robust

24

control system is required to control the molding precisely within a short response time [Selden, 2000]. Development of micro-injection technology started in early 1980s. No

appropriate injection molding machine was available at that time and people had to modify the commercial hydraulic driven units with a low clamping force. To mold microstructures, people usually use a small screw in a conventional screw-injection molding machine. However, the screw is easy to be broken under shear. To reduce the shot size suitable for microstructures, people adopt properly sized runner systems or directly inject melt into cavities using a hot runner nozzle without runner systems [Rogalla and Michaeli, 1997]. Brand new injection molding machine for

microstructures was under development in middle 1990s. The machine developed at FZK can inject very small amount of resin, for example 0.025g, with a stable process [Piotter, et al., 2001]. The machine for micro-injection molding includes venting and variothermal heating systems. In contrast to thin-wall injection, high injection pressure and speed are not essential. Of course, injection pressure and speed as well as other parameters influence the part quality and dimension stability, which is also true for thinwall injection molding. Incomplete filling is a main concern in micro-injection molding. People use a variety of reaction injection methods to reduce viscosity. The common method is using photoinduced polymerization of MMA/PMMA based resins. The molding can be conducted at ambient temperature using a machine with a small sized powerful UV light

25

source. The polymerization time of 2.5 minutes could be obtained [Hanemann, et al., 1997a]. Thermally initiated RIM is another technique. A relatively simply and reliable mold filling and good accuracy could be obtained because of the low viscosity of the cast resin. Resins based on acrylates, methacrylates, amides and silicones are thermally curable. However, this technique needs elevated temperature to start polymerization. The process is relatively slow and also needs mixing and metering units. c. Material For thin-wall injection molding, the material should flow easily, have enough impact strength and high stiffness and resist polymer degradation due to shear heating. However, good flowing ability (high melt index) usually means low physical properties. So, specialized material is required to balance the trade-off between processability and physical properties. Some suitable materials especially for thin-wall injection have been developed. Appropriate materials for micro-injection molding must have low viscosity but satisfactory mechanical properties. Common materials used are PMMA, PC, PSU, POM, PA12, PEEK, etc. d. Operating In thin-wall injection molding, the pressure and velocity are very high. TWIM prefers low viscosity and it mainly relies on high shear rate instead of high temperature. The mold temperature usually is low in order to accelerate heat transfer and reduce cooling time. Due to the rapid cooling of polymer melt, the operating window is narrower as the part becomes thinner. 26

Common injection molding parameters such as relatively low mold temperature and injection pressure will cause incomplete filling of mold inserts. In order to fill the mold inserts completely, the temperature on the surface of mold inserts is usually heated up to melt temperature. This is so-called variothermal heating. Usually the temperature is above glass transition point for semi-crystalline polymers and near melting point for crystalline polymers to reduce flow resistance. The mold is completely filled just before the ejection because conventional venting method is impossible for microstructures. However, it will inevitably increase the cycle time if using a variothermal process. The shortest cycle time reported is 70 s with aspect ratio of only 2.5 and microstructures with high aspect ratios needs more than 6 minutes [Piotter, et al., 1997], which is much longer than thin wall injection molding. It should be noted that for most microstructures with low aspect ratios (<2), usually a constant temperature is used but the temperature is higher than that in conventional injection molding. e. Simulation The modeling of micro-injection molding is different from conventional or thinwall injection molding process. Several models have been developed during last

decades to simulate the filling of conventional injection molding. Most injection molded parts are complex, the filling process is non-isothermal, and polymer fluids demonstrate non-Newtonian behavior. So, it is very difficult to simulate the filling process without simplification. The pioneering work focused on pressure and temperature prediction of

27

simple geometries. Usually, a generalized Hele-Shaw flow model is used to simplify the governing equations for non-isothermal, non-Newtonian melt flows. In most cases, the simplification successfully predicts the modability (pressure and velocity fields, air entrapment, temperature distribution and stress concentration regions) [Hetu, et al., 1998]. The Hele-Shaw approximation neglects flow in the gapwise direction [Garcia , et al., 1991]. So, the velocity in the thickness direction w=0. Continuity equation:

u = 0 Momentum equation:
P + (u ) = 0

i ( ) = 2& ( ) = (u + (u ) T )
The Hele-Shaw approximation can be written as
SP = 0 i.e.
H

P P (S ) + (S ) = 0 x x y y

where S = (
0

z2

)dz

Because the heat conduction in flow direction can be neglected, energy equation can expressed as:

C p (

T T T T +u + v ) = (k ( )) + t x y z z

where is viscous heating.

28

Boundary conditions: Injection gates: Q=Q(t) or P=P(t) T=Tmelt Moving flow fronts: P=0 T=Tcore Mold wall: un = 0 T=Tmold or q=h(Tw-T) It should be noted that the no-slip does not hold anymore after the Hele-Shaw approximation. However, these models are limited in the scope of the information that they can generate. Furthermore, the Hele-Shaw approximation can not accurately predict the fluid behavior at flow front and the flow behavior near or at solid walls, the phenomenon occurring at the merging of two or more streams (weld lines), and the kinematics in gates, ribs, or sudden thickness change, the areas where shear and extensional deformations contribute significantly to the stress field [Gao, et al., 1998]. A threedimensional simulation could provide complementary and more detailed information. However, because of its intensive computation nature, 3D simulation started only several years ago [Han and Gupta, 1999]. The difficulties met in simulating 3-D filling are [Gao, et al., 1998]: 29

 The computational domain is usually a 3D volume having a complex shape. The free surface is subject to large deformations and multiple interfaces may come in contact with each other. The prediction of the flow boundary layers requires no-slip boundary condition. The equations of continuity, momentum and energy can be expressed as [Chang and Yang, 2001]:
+ u = 0 t ( u ) + ( u u ) = P + g t

= (u + (u ) T )
C p (
T + u T ) = (kT ) + & 2 t

The boundary conditions [Gao, et al., 1998]: u = u 0 ; T = Tmelt at inlet

(u ) n P n = t on tractions
u = 0; T = Tmold or q=h(Tw-T) on wall The tracking of the evolution of melt front is usually modeled by pseudo-concentration method:
f + (u f ) = 0 t

30

where f=0 is defined air phase, and f=1 as melt phase. The inertia term and body force term in momentum equation and viscous heating in energy equation can be neglected sometimes [Gao, et al., 1998]. For thin-wall injection molding, the Hele-Shaw approximation is usually used and it generally provides good results. However, the results are not perfect and error is large in some cases, as discussed above. Furthermore, due to the characteristic of thinwall injection molding, the main reason of unsatisfied results is due to the inaccurate description of polymer physical properties in the unstable process. Because of extremely high pressure, the effect of pressure on compressibility and viscosity should be considered. Because the typical filling time is 0.2 s in thin-wall injection molding, the temperature changes dramatically. The isothermal condition combining with high

pressure make it very difficulty to describe the heat conductivity, heat capacity, especially heat transfer coefficient. Study showed that neglecting the effect of pressure on viscosity may cause large error in predicting cavity pressure. The heat conductivity and heat capacity used are often measured at constant temperature and low pressure. So, the main effort to improve the prediction accuracy focuses on the improvement of the property description, which will be discussed in detail in Section 2.2.2. Simulation of micro-injection molding is a new area and very little work has
1 been done. For flat and thin parts, so-called standard injection molding parts, 2 D 2

codes usually provide good results. However, difficulty occurs when simulating the filling of microstructures with high aspect ratios [Piotter, et al., 1997; Hanemann, et al.,

31

1997b; Yu, et al., 2001]. Although the simplification by assuming thin and flat parts makes calculation easy and fast, the dimensional character of microstructures is not thin and flat anymore. The examples are micro parts like micro gearwheels, micro sensors, etc. It is expected that the simplification does not hold anymore and proper 3D

simulation is necessary. For the parts with microstructures, such as LabCD, they show thin-wall plates with microstructures, as shown in Fig. 2.2. Hele-Shaw approximation gives the average information in gapwise direction in the large thin plate. Obviously, local information of T, P and v at inlet of the micro-channel is crucial for the simulation of the flow in the micro-channel. So, it will cause big discrepancy in simulating the microstructures. Furthermore, viscosity and surface tension are even more important for microstructures. The surface roughness may also play a significant role in microstructures. Moreover, the material data, especially rheology data for macroscopic application should be re-examed when applied to micro scale. Previous study showed
1 that 2 D simulation such as C-MOLD is not sufficient to describe all molding effects 2

anymore for extremely small structures of microparts [Hanemann, et al., 1997b]. Modifications to most conventional programs such as MOLDFLOW and

CADMOOULD-SD or new 3-D transient codes are required in order to simulate the filling of micro-injection molding [Hanemann, et al., 1997b]. For the thin part with microstructures such as LabCD, the modification can be as follow: Using 2D codes to simulate the flow in the large domain while using 3D to simulate the flow in the microstructures [Hanemann, et al., 1997b].

32

In this study of the thin-wall molding of base plate with microchannels, the velocity variation in the width direction y is negligible, a 2D x-z plane simulation is used. The momentum equations, the continuity equation and the energy equation are written as follows at a quasi-steady state [Yu, et al., 2004a]:

( v ) = 0 ( vu ) (u ) = ( vv) (v) = p x p z

( C p T ) + ( C p vT ) = (kT ) + H t

where

In this study, thin-wall injection molding with micro-features was studied experimentally and numerically. The filling lengths in microchannels are simulated and compared with experimental results. Because the predicted degree of filling in

microchannels are very sensitive to the heat transfer coefficients selected, it is important to study the effect of selection of property models on simulation outputs. We then further study how the input properties generally affect the output in thin-wall simulation. The output we choose is mold cavity pressure.

33

u H = 2 x

v +2 z

u v + + z x

2.2.2 Cavity Pressure and its Prediction

Injection mold cavity pressure is one of the most important parameters in thinwall injection molding. It is regarded as a good indicator of molded part quality and injection machine control performance [Angstadt, 2001; Dubay, 2001]. It not only indicates the material condition in the mold but also affects the microstructure and part quality [Macfarlane and Dubay, 2000; Gao, et al., 1996; Gao, et al., 1996]. Cavity pressure can affect part weight, dimensions, cosmetics, gloss, warpage, shrinkage, etc. [Bozzelli and Cardinal, 1996]. It is therefore very useful to study the effect of injection operating variables and material properties on cavity pressure (gradient). Usually, low cavity pressure is preferred because low pressure demands low injection capacity that reduces equipment cost, reduces shear orientation, and produces low shear stress which is essential to avoid quality problems such as warpage and low mechanical properties. Low stress is even more important in stereolithography [DellArciprete, et al., 1999; Palmer and Colton, 2000] or micro-injection molding [Yu, et al., 2001] where mold wear and durability are main concerns. Today it is common to use computer aided engineering (CAE) programs to successfully design a part. CAE can be used to troubleshoot and solve problems

concerning filling time, injection pressure, gate location and dimension, warpage, coolant efficiency, etc. [Kalnin and Zluhan, 1999]. The application of CAE has the potential to reduce overall production cost and improve part quality. Using CAE to

34

analyze part quality has given encouraging results and it is possible to design a good mold without any tool tryouts [Kansal, 2000]. However, almost all users would prefer better accuracy of CAE simulation [Ainoya and Amono, 2001]. During thin-wall injection molding (TWIM), the error of the prediction of cavity pressure from CAE simulation can vary from 50% to more than 100%, and the error increases as the parts become thinner [Chen, et al., 2000]. The discrepancy may result from neglecting some important factors during simulation. For example, the effect of pressure on viscosity is important because of very high pressure which occurs in TWIM [Chen, et al., 2000; Amano and Ainoya, 2000; Fasset, 1995; Mahishi, 1998]. However, accurate pressure-dependent data are rare and not available commercially. The actual testing is time consuming, expensive, and test equipment is not commonly available [Ainoya and Amono, 2001]. The heat transfer coefficient between the part and mold wall changes with time and operating variables. It affects the cooling time and melt pressure. However, it is usually a constant in commercial CAE packages. For example, both C-MOLD and MoldFlow set a default value of 25,000 W/m2K, which result in higher predicted cavity pressure. Chen, et al. [2000] noticed that material properties might be the reason for the prediction discrepancy. Ainoya and Amono [2001] found that pvT-data affected fill time and cavity pressure. They also found that the heat transfer coefficient and pressure-dependent viscosity had a great effect on pressure prediction. Cavity pressure drop was extremely overpredicted when the effect of pressure on viscosity and juncture loss were not considered. Slightly lower filling pressure was predicted when the tabulated heat capacity was used instead of 35

constant heat capacity. However, the thermal conductivity had little influence on filling pressure. Sherbelis and Friedl found that neglecting the effect of pressure on viscosity led to overprediction of cavity pressure, while neglecting the juncture loss led to underprediction of nozzle pressure [Sherbelis and Friedl, 1996]. Sridhar and Narh [1999] found that the heat capacity and thermal conductivity had almost no effect on cavity pressure, but they could affect cooling time and part shrinkage and warpage. Another cause for the discrepancy between simulation and experiment is the lack of a high quality database for the polymer, such as heat conductivity and pvT data [Chen, et al., 2000]. For example, viscosity and thermal properties are usually measured under equilibrium conditions, and they greatly affect simulation accuracy when these properties, such as pvT data, are used in non-equilibrium injection molding process [Chen, et al., 2000]. Furthermore, the difference may result from the difference between the actual molding conditions and set conditions. For example, the actual melt

temperature and injection velocity may be greatly different from the set parameters [Ainoya and Amono, 2001]. Thus the set parameters do not reflect the actual melt conditions and using these parameters for simulation results in the difference between the simulation results and experimental results. Some work has been conducted to study the effect of material properties on pressure. However, a systematic study of the effect of these parameters is rare. In this study, the effect of pressure-dependent viscosity, heat capacity, heat transfer coefficient, juncture pressure loss and pvT-data on cavity pressure and pressure drop prediction will be considered, and the importance of each parameter will be evaluated. Then the 36

simulation results and measured data will be compared. Finally the method to improve the prediction accuracy will also be discussed. The study aims to help in understanding which material property is important and needs rigorous testing, in order to improve simulation accuracy and reduce time and cost for expensive property testing.

2.3 REUSE of HIPS

As the demand for plastics is increasing, the disposal of plastics is also increasing. World thermoplastic consumption was over 100 million kilograms in 2000 [Society of the Plastics Industry, 2001]. However, only approximately 5.4% of postconsumer plastics was recovered in the US [U.S. Environmental Protection Agency, 2002]. Plastics recycling is important because more efficient re-use of materials will reduce the quantities of plastics sent to landfills as well as reduce raw material extraction. Waste prevention practices are increasingly significant and are increasingly encouraged with the advent of take-back legislation [Gamalski, 1996; Meffert and Kirchgeorg, 1997; Hubschman, et al., 1995]. It is accepted that direct use of postconsumer polymer is the most efficient and reliable way to treat plastic waste [Kartalis, et al., 1999]. However, how to characterize the post-consumer resin and how to increase the percentage of the post-consumer resin are two of the problems in recycling plastic. High impact polystyrene (HIPS) occupies a large market share in computer, business machines, and other electronics [Arola and Legarth, 1999]. Furthermore,

monitor housings and printers are two of the largest applications. However, less than

37

1% is recovered from the total 19% market share of HIPS [Dillon and Aqua, 2000]. Therefore, it is important to evaluate and develop viable options for discarded polymer products. However, the analysis of life cycle trade-offs between use of recycled plastics, recyclability, reduction of process wastes, energy consumption, yields, and product performance are complex [Allenby and Laudise, 1995; Stuart, et al., 1999; Szekely and Laudise, 1995]. To date, many companies process either 100% virgin material or virgin material with small percentages of regrind. Sources of post-industrial regrind may be internal or from another industrial processor(s). Companies embracing product

stewardship are struggling to develop viable approaches to process and recycle returned products economically. Post-consumer polymers may be contaminated by other

materials [Langerak, 1997]; post-consumer products may contain polymer blends as well as additives such as reinforcements, paint, or flame retardants [Dillon, 1999]. Thus, post-consumer plastics introduce additional raw material uncertainties into the manufacturing process. In addition, incompatible polymer blends may be present in a product, requiring expensive disassembly procedures or less valuable mixtures. As a result, many plastics recyclers currently select between options such as incineration or downcycling, the formation of lower grade polymer materials. Another complication is that returned polymers have been exposed to various thermal and mechanical conditions and degradation could happen. One important challenge in post consumer resin recycling is the contamination from other materials. If the contaminants are not removed, then the mixed materials may be down-cycled for use in simpler applications than their original products. 38

Contaminants may be metal, stickers, or other polymers. The level of contamination in post-consumer resins depends on the product design and the separation techniques used. After hazardous materials such as batteries are removed manually, disassembly or automatic size reduction (shredding) follows. Separation techniques may include

manual labor, magnets, air separation, or float sink approaches [Hendrix, et al., 1996]. Langerak [1997] compared there different separation methods for television housings: (1) complete manual disassembly; (2) manual removal of front and back casings, shredding, magnetic ferrous metal removal, and float-sink separation of nonferrous metals and plastics; and (3) complete shredding, magnetic ferrous metal removal, and float-sink separation of nonferrous metals and plastics. Langerak concluded that option 2 was the most cost effective materials recovery method. Because many variations of plastics have similar densities, plastics identification based on density may not be sufficient. separation. Furthermore, identification marks usually appear at the most general level, such as low-density polyethylene, and do not include information such as the manufacturer product code or manufacturer. Even recyclers with close relationships with So, contamination is still a concern because option 2 relies on density

manufacturers and suppliers and sophisticated information links have faced challenges to identify materials from the model number [Grenchus, et al., 1998]. Moreover, the source of plastics collected is correlated to the age and diversity of the materials. For example, the plastics from municipal solid waste collection will be older and contains

39

more diverse assortment than the plastics from post-industrial regrind. Thus, further characterization of the actual post-consumer resins is required through testing. Another big challenge is the material degradation. During polymer processing, such as injection molding, materials undergo severe thermal histories and mechanical loadings (shear and extensional flows) that produce molecular degradation. Molecular degradation of plastics usually decreases the polymer chain length and leads to a decrease in melt viscosity and mechanical properties of the final product. The degree of degradation varies significantly depending on the type and amount of polymer and additives used in each commercial resin. Blending polymers for recycling has been studied in the context of improving the properties of PCR [Liu and Bertilsson, 1999]. Several researchers have studied property degradation during thermoplastic recycling processes. Ries and Menges [1988] studied the degradation of polypropylene and found a decrease in impact strength due to a decrease in molecular weight. They believed that the melt index could be useful in predicting the molecular weight, which may allow off-line monitoring of impact strength degradation. Zahavich, et al. [1992] showed that viscosity and swell ratio are the best indicators for degradation of an HDPE resin. This is not surprising, since swell ratio is a strong function of the high molecular weight fraction. Pagel [1989] showed that reground ABS exhibits very stable physical properties over successive regrind generations, but the resins became yellow. Dzeskiewicz, et al. [1993] studied the decrease in mechanical and rheological properties of glass-filled nylon with successive generations, but showed that more than 50% regrind could be blended with virgin resin to exceed specifications. 40

The degradation of polymer chains due to mechanical stresses (shear and extensional flows) has been an object of scientific interest for at least 50 years, starting with the Frenkels work [Frenkel, 1944]. Using a bead-spring model, Frenkel predicted that the polymer chains would align with the flow, that the stress would be maximum in the middle of the chain, and that as a result the middle of the chain would be the site of fracture. Scientists have observed mid chain fracture for a variety of polymers under various flow conditions. The flow birefringence experiment on polystyrene showed that the degradation products formed a narrow distribution around half the molecular weight of the initial polymer molecules [Odell and Keller, 1986]. Nguyen and Kausch [1986] studied the degradation of polystyrene in a different flow device, and found that the polymer degraded to a broad molecular weight distribution. These studies point to the importance of understanding how thermal history and mechanical stresses impact polymer degradation. Studies have been conducted on determining the number of cycles a polymer can be molded [Shriver, et al., 1994; Bernardo, et al., 1996]. Bernardo, et al. [1996]

developed a model to predict the properties of virgin/recycled polymer mixes based on the number of previous processing cycles. This information is useful in determining the materials recycling threshold for polymer components. In isolated cases where housings are returned to the original manufacturer, the number of cycles may be tracked [Timmons, 1998]. However, in most cases, it is not viable to track the number of processing cycles.

41

To use PCR, it is important to decide the processing parameters quickly in injection molding application. However, the challenge is how to characterize the PCR in order to injection mold it. Suppliers of virgin materials provide ranges of typical

property values for tensile strength, tensile modulus, impact, mold shrinkage, and other characteristics [GE Plastics, 1992]. Some companies seek used polymers with

certification of their mechanical properties [Jones, 1996]. Furthermore, manufacturers often use mold filling simulation software with virgin resin databases to reduce the time to determine initial processing parameters. A major gap is that current databases do not contain entries for used resins. Although Narh, et al. [1999] investigated the viscosity and injection molding processing parameters for post-consumer ABS, PC, and nylon 5.5, with mold-filling simulation and design software, they did not specify how they obtain their inputs for the PCR in a mold-filling simulation. Without material characteristic data of the PCR, molders cannot easily determine whether a PCR may be a candidate for use alone or in blends with virgin resins depending on the material characteristics and the complexity of the application. Designers are hesitant to include post-consumer recycled material content. Further complicating the inclusion of recycled materials is the uncertainty of material content, contaminants, and degradation [Eriksson, et al., 1998]. In this study, we describe our progress in evaluating the viability of reusing postconsumer and virgin polymer blends of high impact polystyrene from electronics equipment housings. The study also introduces a new approach to determine initial processing parameters for injection molding of post-consumer resin.

42

Dull region

Dull regions are on the phase on the top and the bottom
a. Alternate dull and glossy regions.

Dull region

Dull regions are out of phase on the top and the bottom

b. Synchronous dull and glossy regions.

Fig. 2.1. Alternate and synchronous dull and glossy flow marks.

43

Microchannels Main flow direction

Fig. 2.2. Thin-Wall plat with microstructures.

44

CHAPTER 3

FLOW MARKS DURING THIN-WALL INJECTION MOLDING

3.1 ALTERNATE DULL AND GLOSSY FLOW MARKS 3.1.1 Introduction

Several kinds of surface defects may occur during injection molding. One type of surface defect is characterized as alternate dull and glossy surfaces in which flow marks on one side of the part are out of phase with those on the other side [Yokoi, 1994a], as shown in Fig. 3.1. This is often referred to as tiger striping. This surface defect occurs especially on automotive exterior parts. Some work has been done to explain and eliminate flow marks [Yokoi, 1994b; Chang, 1996a; Chang, 1996b; Hobbs, 1996; Heuzey, et al., 1997; Hamada and Tsunasawa, 1996; Dharia, 1999; Bulters and Schepens, 2000; Grillet, et al., 2000; Charmeau, et al., 2000; Xu and Koelling, 2001; Jayaraman, 2002]. Several mechanisms have been proposed to explain the generation of flow marks. However, little has been truly understood on why flow marks occur and how to predict and eliminate them. A slip mechanism was proposed by Chang [Chang, 1996a; Chang, 1996b]: When the melt has low adhesion to the mold surface, slip can occur and initiate a melt flow instability. Thus, this kind of flow instability can cause 45

flow marks with alternate dull and glossy regions. Hobbs believed that stick/slip flow at high wall shear stresses caused flow marks [Hobbs, 1996]. Conversely, Heuzey, et al. [1997] found that there was no obvious relationship between the wall slip and flow marks. Furthermore, coating on a mold wall did not affect the occurrence of flow marks. They believed that the generation of flow marks was due to the filamentation and stretching of semi-solidified materials. Hamada and Tsunasawa [1996] found that in cases where flow marks occurred, the PC and ABS flow oscillated, while in cases where no flow marks occurred, the PC and ABS flowed in steady laminar motion with a normal fountain flow. In a study of binary blends of polypropylene and ethylene co or terpolymers, Dharia [1999] proposed that the flow marks were generated by melt flow instability and the inability of melt to recover from the stress changes at the flow front. More recently, Bulters and Schepens claimed that flow marks resulted from a flow front instability [Bulters and Schepens, 2000]. Furthermore, Grillet, et al. conducted a linear stability analysis and found that the most unstable eigenvector was an oscillatory, swirling flow near the stagnation point at the free surface [Grillet, et al., 2000]. The effect of operating variables, physical properties of plastics, and mold geometry has been discussed. Although experimental evidences show that oscillating instabilities occurred in mold filling [Chang, 1996a; Hamada and Tsunasawa, 1996; Bulters and Schepens, 2000], there is disagreement on why the instability occurs and where it originates. In this section, we study the effect of operating parameters and different polymer melts on the wavelength, width, and gloss variation of the flow marks. Then several methods to reduce the flow marks are discussed. Furthermore, it is widely 46

known that elastic instability occurs upstream of a contraction, such as a capillary die or slit die, in extrusion or spinning [Piau, et al., 1988; White, et al., 1987; Boger and Walters, 1993; Koelling and Prudhomme, 1991]. However, the researchers studying the flow marks have not studied the entry instability and its relationship to the flow marks, although Dharia already noticed the similarity between extrusion and injection molding processes [Dharia, 1999]. We study the correlation between gross melt fracture in extrusion and alternate dull and glossy flow marks in injection molding, then propose an entry viscoelastic flow instability mechanism to explain the alternate flow marks.

3.1.2 Experimental

Molding experiments were conducted on a Sumitomo SG M-HP 180-ton injection molding machine. The materials used were four types of polypropylenes, namely PP-A, PP-B, PP-C and PP-D. Two spiral molds were employed with different thicknesses (1.58 and 3.17 mm). The width of the mold channel was 1". The total flow length was 16". The melt temperature was 204.4, 223.9 and 232.2C. The mold temperature for most experiments was set at a constant value of 29.4C. The mold temperature was changed to 79.4, 51.7, and later to 18.3C. Two rectangular molds were also employed with a thickness of 5.08 and 1mm. The length and width of the mold channels was 150 and 51 mm, respectively. The edge gates used were 2.54 and 0.5 mm in thickness for the thick and thin mold, respectively. The melt temperature was 190, 225 and 260C. The mold temperature ranged from 22 to 85C. The effect of

47

holding pressure and injection pressure was studied at melt temperature 190C, mold temperature 22C, and injection speed 0.4 m/s where the flow marks were pronounced. For the parts that exhibited tiger striping marks, the wavelength was measured. The wavelength was the distance from one shiny region to another on one side. We also measured stripe width, which was the width of one shiny region. The rheological properties were measured by a Rheometrics RMS 800. The complex viscosity, storage and loss modulus, and first normal stress difference of each polypropylene sample were measured at 180, 200, and 220C, respectively. The

extensional viscosity was measured at 130C by a tensile tester, Instron 8511, based on the standard ASTM test. The samples were standard tensile bars with 13 mm in narrowsection width, 57 mm in total length, and 3.2 mm in thickness. To obtain a constant strain rate, one needs to program the Instron machine to follow the exponential-type increase of sample length. The viscosity-molecular weight was measured based on ASTM D445-97. To check the slip effect, Dynamar 9613 (a 3M product), a fluorocarbon elastomer, was used as a coating agent. It is a slip promoting product. Its dilute acetone solution, ca. 1%, was coated on the hot mold surfaces to allow evaporation of the solvent. We also studied the disappearance of the flow marks by directly adding a small amount of Dynamar, 0.2%, into the PP-C pellets. The resin was well mixed before tests. Samples were collected after 100 shots to stablize. A Differential Scanning Calorimeter (DSC) from TA Instruments, DSC 2920, was used to measure the crystallinity of dull and shiny regions. The sample was scanned 48

from 30 to 200C at the rate of 10C/min. A Scanning Electron Microscope (SEM), Philips XL 30, was employed to observe the morphology of dull and shiny regions. An optical profilometer, Wyko NT330, was used to measure the roughness of the dull and shiny regions. A two-stage single-screw extruder (Rheomex 252p) from Haake was applied to exam the melt fracture phenomenon that usually shows up with PP. The screw had a diameter of 3/4 inch and a length to diameter ratio (L/D) of 25. The diameter of the capillary die used was 1.2 mm and its length was 12. Because it is very difficult to estimate the wall shear stress during mold filling, C-MOLD 2000 was used to simulate filling our spiral molds. C-MOLD is a set of integrated computer aided engineering (CAE) simulations for plastics molding processes, including injection mold filling, post-filling and cooling, part shrinkage and warpage. CAE provides an easy-to-use data visualizer for viewing mesh information and analysis results. First, the geometry was built, and then a mesh with 672 elements was set up for C-MOLD simulation. In the simulation, all four types of polypropylenes were adopted in both thin and thick molds. Moldflow Plastics Insight (MPI) 3.0, a software integrating C-MOLD 2000 and Moldflow Plastics Insight 2.0, was used to simulate filling our rectangular molds to estimate the wall shear stress at the gate. The geometry and the mesh with 754 elements built in C-MOLD was imported and then the simulation was run on MPI 3.0. In the simulation, the values of the processing

parameters, such as shot size, injection pressure, holding pressure, holding time, mold

49

temperature and cooling time, were the same as those in the real injection molding processes.

3.1.3 Results and Discussion 3.1.3.1 Rheological Characterization

The complex viscosity of PP-A, PP-B, PP-C, and PP-D was measured at 180, 200 and 220oC, respectively. Fig. 3.2 shows the complex viscosity of four polymer melts at 200oC. It was found that the complex viscosity decreased with an increase in frequency, and PP-C had the largest viscosity at the same frequency. Fig. 3.3 shows the complex viscosity of PP-C at different temperatures. Fig. 3.4 shows the storage modulus and loss modulus of different PPs at 200oC. It was found that the storage modulus and loss modulus of PP-C were the largest. Also we can see that the storage modulus of PPC was very close to its loss modulus, as compared to other PPs. Fig. 3.5 shows the first normal stress difference N1 versus shear rate at 200oC. PP-C had the largest N1 at the same shear rate. Fig. 3.6 shows the first normal stress difference N1 versus shear rate at 180, 200 and 220oC, respectively. Fig. 3.7 shows the transient extensional viscosity at 130oC when the strain rate was 0.01 s-1. It was found that PP-C had the largest

extensional viscosity at the same moment. It was also found that in the range of tested shear rate and time, the ratio of N1/extensional stress was PP-C > PP-D > PP-A or PP-B at the same conditions. To determine relaxation time, we shifted all dynamic viscosity ' and elastic material function 2"/ data to the master curve at 200oC using time-temperature 50

superposition and fit the data with the 1-mode Giesekus model [Bird, et al., 1987], as follows:
= S + P
P + 1 P1
. 1 { P . P } = P P
.

S = S

An example is shown in Fig. 4.8. The relaxation time determined from the model fit for different materials is shown in Table 3.1. We found that PP-C had the longest relaxation time, and therefore the largest Deborah number (De= ) at the same shear rate. Thus PP-C is the easiest material to develop a viscoelastic flow instability. The viscosity-molecular weight of all four polymers measured is listed in Table 3.2.

3.1.3.2 Injection Molding Results

Experimental trials were conducted in the spiral molds at first. It was found that PP-A and PP-D did not generate flow marks. However, flow marks usually occurred for PP-C and PP-B under a certain range of processing conditions. The flow marks are generally out of phase between the top and the bottom of the part. However, the shiny region on one surface is not exactly in the center of two neighboring shiny regions on the opposite surface. The flow marks that occurred were characterized as alternate dull and shiny regions. Flow marks of PP-C occurred when the flow front velocity was as low as 0.01 m/s for the thick mold and 0.1 m/s for the thin mold. When the injection speed was 51

very low, the flow marks occurred at the end of the flow length. They did not occur immediately after the distance of about one wavelength from the gate. With the increase of the injection speed, flow marks became more pronounced. When the injection speed was higher, flow marks occurred immediately after the polymer melt entered the mold cavity. Also, the flow marks were very pronounced. When the injection speed was increased further, flow marks became dimmer. When the injection speed was above a critical value the flow marks disappeared. The width and discernible level of shiny regions changed along the flow length. Usually, flow marks at the end of the flow length were pronounced. Even for the same shiny region, the width changed in the width direction. Usually, the shape of the shiny region also changed along the flow length, and in some cases irregular shapes appeared. Fig. 3.9 shows a typical set of samples of flow marks for PP-C. For PP-B, the flow marks are similar to those of PP-C. However, because of the filler, the color of the molded parts was yellowish and the flow marks were less pronounced, making it very difficult to distinguish the neighboring dull and shiny regions. Although we can notice the flow marks, it is very difficult to measure the wavelength and width of the flow marks except in a narrow range of flow front velocity at which the flow marks are apparent. It was observed that the wavelength and width increased with the increase of flow front velocity. Above a critical flow front velocity, the flow marks disappeared. Although the wavelength is close to the wavelength of PPC at the same conditions, the width of flow marks of PP-B is narrower compared to that of PP-C. 52

In the thick rectangular mold, the alternate dull and shiny flow marks of PP-C only occurred above a certain injection speed. At first the flow marks gradually became more pronounced as the injection speed was increased; however, with the further increase of injection speeds, the flow marks became less visible and finally disappeared. This phenomenon is somewhat different from other researchers observations that the flow marks became more severe as the injection speed increased [Chang, 1996b; Hobbs, 1996; Heuzey, et al., 1997]. A typical example of alternate dull and glossy flow marks is shown in Fig. 3.10. In the thin rectangular mold, alternate dull and shiny flow marks of PP-C occurred once the mold was filled. The wavelength and stripe width of the flow marks of PP-C were measured. The wavelength is defined as the distance from one shiny region to another on one side, while the stripe width is the width of a single shiny region. Fig. 3.11 shows the effect of flow front velocity on the wavelength in the thin mold. It was found that for the thin mold the wavelength increased with an increase in flow front velocity, and then remained relatively constant. Moreover, the wavelength was almost the same at the same flow front velocity. However, the higher the melt temperature, the longer the final wavelength. Fig. 3.12 shows that the mold temperature has little effect at low flow front velocities; however, at higher flow front velocities, the higher the mold temperature, the longer the wavelength. Fig. 3.13 shows that the thicker mold exhibits a longer

wavelength. Thus a longer wavelength can be contributed to a higher melt temperature, a higher mold temperature, and a thick mold. Fig. 3.14 shows the effect of melt temperature on the width of the shiny stripes in the thin mold. It was found that the 53

width of the shiny stripes increased with an increase in the flow front velocity. At a low flow front velocity, the melt temperature had little effect on the width of the shiny stripes; however, at a higher flow front velocity, the higher the melt temperature, the wider the shiny stripes. Fig. 3.15 shows that the mold temperature does not have much effect when the flow front velocity is small, yet the width of the shiny stripes increases as the mold temperature is increased at a high flow front velocity. The trend of the change of the width is very similar to the trend of the change of the wavelength. Fig. 3.16 illustrates that the width of the shiny stripes increases with the increase of mold thickness. Compared to previous work [Xu and Koelling, 2001], the effect of mold temperature and melt temperature was clearly observed in the rectangular molds. For the gloss variation of the flow marks of PP-C in the thin rectangular mold, it was found that increasing either melt or mold temperature made the flow marks less visible. The observed effect of mold temperature and thickness is in agreement with Changs work [Chang, 1996a; Chang, 1996b]. The effect of melt temperature in our experiment agrees with Hamada and Tsunasawas result [Hamada and Tsunasawa, 1996]. It was found for the first time that the flow marks were less visible as the holding pressure was increased. However, injection pressure had almost no effect on the

visibility of the flow marks, which was different from other researchers observations [Chang, 1996b; Dharia, 1999]. Furthermore, the effect of the molecular weight of PP-C was studied. Adding 20% PP-D into PP-C greatly alleviated the flow marks, and the flow marks were scarcely

54

visible compared to pure PP-C at the same operating conditions. However, adding 20% PP-A had little effect. From the above discussion, the flow marks could be reduced by one or more of the following factors: high injection speed, high melt or mold temperature, mold surface coatings, and/or changing molecular weight or its distribution. The flow marks occurred only above a certain flow front velocity in the thick mold, Vcri. It was further found that the mold temperature almost had no effect on Vcri. Vcri scarcely changed at different mold temperatures. However, Vcri increased as the melt temperature was increased, as shown in Fig. 3.17. The effect of the coating on the surfaces of the mold or gate was also studied. It was found that a coating on the mold surfaces could not prevent the occurrence of the flow marks, although it could alleviate the flow marks and make them less pronounced. One interesting phenomenon is that coating on the mold surfaces did not change the Vcri, implying that slip is not the cause of the alternate flow marks. The reason is that coating on the mold surface reduces the critical shear stress where the slip occurs, thus decreasing the Vcri where the slip is triggered. Another interesting phenomenon observed was that the flow marks disappeared at high injection speeds, which has not been reported previously. We define it as the transition velocity, above which the flow marks disappear. Fig. 3.18 shows the

transition flow front velocity vs. melt temperature. It was found that the transition velocity increased as the melt temperature was increased. However, the mold

55

temperature almost had no effect on the Vtrans. The zone of the flow marks in which the flow marks may occur in operation is shown in Fig. 3.19. It was found through our experiment that mold surface coatings increased the Vtran. It was further found that adding 0.2% Dynamar into PP-C also increased the Vtrans. It is well known that a slippery surface or adding a small amount of fluorelastomer into the polymer reduces the wall shear stress at which slip occurs [Yang, et al., 1998; Kazatchkov, et al., 1995]. This implies that the slip is not the cause for the

disappearance of the flow marks at high injection speeds. The reason is that the slippery surface or the addition of Dynamar decreases the wall shear stress where slip occurs, thus decreasing the Vtrans at the same operating variables. Therefore, slip does not cause the flow marks to disappear. Their disappearance may be due to the higher melt

temperature induced by high flow front velocity, and thus greater shear heating.

3.1.3.3 Morphology and Crystallinity

The Differential Scanning Calorimeter (DSC) experiment showed that no difference in crystallinity was observed between the dull and shiny regions. The sample thickness was about 100 micrometers. From the scanning electron micrograph (SEM), it was found that polymer molecules were highly oriented in the shiny region, but the polymer molecules were only slightly oriented in the dull region, as shown in Fig. 3.20. This is in agreement with other researchers results, except that either high orientation or no orientation was observed for shiny regions [Charmeau, et al., 2000]. The measured average roughness by optical profilometer was smaller in shiny regions than in dull 56

regions, as shown in Table 3.3. The reported average roughness is the average value of 5 randomly selected positions.

3.1.3.4 Extrusion

The PP was extruded at the die temperature 170C. It was found that when the wall shear stress was low the extrudate was smooth, but gross melt fracture occurred at higher wall shear stresses. The extrudate irregularity was wavy, as shown in Fig. 3.21. The wall shear stress was estimated by
P without the Bagley correction, where P 4L / D

is the pressure drop in the die, L is the die length, and D is the diameter of the die. The apparent shear rate was calculated by

32Q , where Q is volumetric flow rate [Macosko, D 3

1994]. The experiment showed that the critical wall shear stress for the onset of the gross melt fracture was 0.13 MPa. This is in agreement with other researchers' results [Kazatchkov, et al., 1995]. The flow curve is shown in Fig. 3.22.

3.1.3.5 Simulation

It is very difficult to obtain wall shear stress and temperature profiles during filling the spiral molds. By the C-MOLD simulation, the wall shear stress at the center of flow front was obtained. Fig. 3.23 shows the wall shear stress of different polymer melts vs. filling percentage at the injection speed of 0.1 inch/s in the thick mold. It was found that generally, the wall shear stress of PP-C > PP-D > PP-B > PP-A at the same

57

filling percentage. At the same operating conditions with an injection speed of 0.1 inch/s, melt temperature of 204.4C, and mold temperature of 29.4C, the wall shear stress of PP-C was about 0.13 MPa, while PP-B was only about 0.05 MPa. 0.05 MPa is much smaller than 0.13 MPa above which macroscopic slip usually occurs for PP [Kazatchkov, et al., 1995; Hatzikiriakos, 1991]. However, our experiment showed that PP-B usually had flow marks. PP-D showed no flow marks, although its wall shear stress was larger than that of PP-B. It was concluded that slip is unlikely the reason for the generation of the flow marks in our case. Furthermore, slip could not explain why the flow marks disappear at even higher injection speeds. MPI 3.0 was used to obtain the wall shear stress during filling of the rectangular mold. The critical wall shear stresses of PP-C at the middle of the gate were obtained at Vcri where the flow marks began to form. The processing parameters and the Vcri used had been determined from the injection molding experiment. Fig. 3.24 shows the critical wall shear stress at the middle of the gate at different melt temperatures. It was found that the wall shear stresses were in a narrow range. That means the flow marks start to form at a wall shear stress of around 0.24 MPa at the gate at different melt temperatures. Fig. 3.25 shows the critical wall shear stress at the middle of the gate vs. the filling percentage at different mold temperatures. The figure shows that the wall shear stress generally was very close at different mold temperatures. From the simulation, it was found that flow marks of PP-C started at the same critical wall shear stress almost independent of melt or mold temperature.

58

3.1.3.6 Mechanism

In injection molding, the process that polymer melts experience is similar to that in extrusion. The general picture of these processes is that the polymer melt meets a contraction and experiences high shear at the die or gate, then the polymer melt leaves the die or gate and the polymer molecular chains relax, as shown in the following Fig. 3.26. Therefore, the melt fracture in extrusion processes and the flow marks in injection molding are related, although a difference in the movement of polymer melts exists between extrusion and injection. In extrusion processes, the shear rate at the die is on the order of 1000 1/s [Schramm, 1994]. After the melt leaves the die, the melt swells and moves like a plug flow with a free boundary. The shear rate is zero and the shear stress at the surface is zero if we neglect the small extension near the surface. For injection molding processes, the shear rate in the gate is higher than in the die. After the melt leaves the gate, the melt moves in the mold, but the swell of the melt is restricted by the mold wall because of the rigid boundary. Also the wall shear rate and wall shear stress are still large, although they are much smaller than in the gate. The shear rate is on the order of 10,000 1/s in the nozzle. In extrusion processes, when the shear stress is low, the surface of extrudates is smooth. However, flow instabilities occur when the stresses are sufficiently high. The extrusion stability is associated with the appearance of distortion on the extrudate surface, sometimes accompanied by oscillatory flow. Usually, melt fracture is a general term used to describe different irregularities and instabilities that generate distortions and non-smooth surfaces. Denn proposed a set of instabilities of LLDPE [Denn, 1990; 59

Denn, 2001]. When the shear stress reaches a critical value, typically about 0.1 MPa, the surface becomes rough and wavy, which is commonly called sharkskin or surface melt fracture. This type of irregularity with wavelengths is much smaller than the capillary radius, and is about 1/10-1/5 of the overall specimen diameter [Loenov and Prokunin, 1994]. At a higher stress, the alternate smooth and sharkskin occurs. It is known as slipstick, or spurt flow. At a still higher stress, a transition region occurs where the surface is relatively smooth with long-wavelength distortion. At a much higher stress, gross or wavy distortion occurs. The wavelength is about the specimen diameter [Loenov and Prokunin, 1994]. This set of phenomena is common for linear polymer melt, such as HDPE, LLDPE, and PBD. However, most branched polymers do not show sharkskin or slip-stick regions [Denn, 2001]. They only exhibit gross distortion. Sometimes

extrudates exhibit smooth surfaces again when the stress is much higher than the stress where gross melt fracture occurs. It is commonly known as superextrusion and was well reviewed by Leonov, et al. [Loenov and Prokunin, 1994]. It is believed to result from the uniform slip along the die. The gross melt fracture has been studied for more than 50 years. However, controversy still exists about the melt fracture phenomenon [Piau and Agassant, 1996; Piau, et al., 1990a]. There are two common mechanisms in the literature to explain the melt fracture [Piau and Agassant, 1996]. Some researchers believe that slip at the die wall is the origin of the melt fracture. However, Den Otter clearly showed that wall slip at the die could not explain the melt fracture [Den Otter, 1970]. Most researchers agree that an entry instability causes gross melt fracture [Piau and Agassant, 1996; Larson, 60

1992]. The instability is also affected by various properties, such as polymer structure, geometry of die entry, melt temperature, and die temperature [Piau and Agassant, 1996]. For viscoelastic fluids, a Newtonian fluid-like corner vortex may occur at a low flow rate. The streamlines are the same as those of Newtonian creeping flow with a small corner vortex, named "Moffatt eddy". The corner vortex zone is a dead zone that does not interact with the fluid outside. The formation of vortices may be due to the increasing extensional viscosity with the deformation rate or also due to the shearthinning effect [Den Otter, 1970; Cogswell, 1972]. Two different pathways are possible for the development and growth of vortices as the flow rate is increased [Rothstein and McKinley, 1999; Rothstein and McKinley, 2001]. For some viscoelastic fluids, the corner vortex grows in strength as the flow rate is increased [Yesilata, et al., 1990]. At a very high flow rate, the corner vortex grows upstream, fluctuates, and makes the flow field entirely unstable. For some other viscoelastic fluids, two types of vortices coexist. One is the corner vortex, and the other is the lip vortex [Yesilata, et al., 1999]. As the flow rate is increased, corner vortex and lip vortex recirculating areas expand; then lip vortex gradually develops and invades the corner vortex, and finally generates a single area of recirculation [Piau and Agassant, 1996]. At a still higher flow rate, the flow becomes unstable, and the vortex pulsates or rotates, causing a global change of flow structure [Rothstein and McKinley, 2001; Boger, et al., 1986]. It is yet not clear whether lip vortices occur for a given viscoelastic fluid. However, researchers found that its occurrence depends on fluids and contraction ratio, i.e. the ratio of diameter upstream to the diameter of the contraction [White, et al., 1987; Rothstein and McKinley, 1999]. 61

It was found that the development of upstream instabilities governed the appearance of the extrudates and the helix pitch. Above a critical stress, the flow instability occurs independent of downstream flow conditions [Piau, et al., 1990b]. The melt fracture occurs when the vortex is unstable. The amplitude and frequency of pulsation increases with the pressure [Piau and Agassant, 1996]. Usually the frequency of surface distortion and the vortex pulsation is identical. As a result, we expect that vortices may form and an instability may happen at the entry in injection molding when the flow rate is high enough, since extrusion processes and injection molding processes are similar although differences exist between extrusion and injection molding. The oscillating entry instability can propagate and affect the downstream flow. Thus, the symmetrical oscillating flow may occur in the mold, and the different thermal and shearing history of the melt causes alternate flow marks. In fact, the oscillating flow has been found in other researchers experiments [Hamada and Tsunasawa, 1996; Bulters and Schepens, 2000; Yokoi, 1994b]. Our experiment showed that PP-C exhibited gross melt fracture at a high flow rate, i.e., the entry instability occurred before a contraction, although the surface is smooth at a low flow rate. It is therefore not surprising that the PP-C exhibits alternate dull and glossy flow marks. As discussed above, slip is not the reason for the generation of the flow marks. Some researchers [Xu and Koelling, 2002; Heuzey, 1997] already excluded slip as the possible reason. It is also well known that PP behaves like a branched polymer. This means that its molecular chains are not highly entangled and so the slip-stick phenomenon does not occur. 62

One interesting question is why so many polymers exhibit gross melt fracture, yet fewer polymers are reported as apparently showing flow marks. The reason may be due to the different geometry of contraction used in extrusion and injection mold processes. It was found that the symmetrical contraction generates vortex more easily than a nonsymmetrical contraction, such as planar die [White, et al, 1987]. Many researchers used capillary die in extrusion, while most gates in injection are usually not symmetrical. Viscoelastic melts therefore show vortices and thus gross melt fracture in extrusion, but are less likely to exhibit vortices in injection molding (and thus flow marks). This mechanism could explain our experimental results of flow marks. At a low injection speed, the flow is stable and no flow marks occur. At a high injection speed that reaches a critical wall shear stress, an entry flow instability occurs, resulting in symmetrical oscillating flow that generates alternate flow marks, as shown in Fig. 3.27. As the injection speed increases, the wall shear stress increases. The melt becomes more severely sheared and the flow marks become more pronounced. However, with a further increase of the injection speed, shear heating becomes more important and the melt temperature is increased. This may make molecular chains easier to relax and thus less sheared, so the flow marks are less visible. Finally, the flow marks disappear.

Increasing the melt temperature or mold temperature decreases the wall shear stress and thus makes the flow marks less visible. The effect of holding pressure is complicated. It may be explained as follows: Although increasing the holding pressure increases the wall shear stress, the high viscous heating increases melt temperature and thus makes the flow marks less visible. The effect of injection pressure was very unusual, although it was probably overshadowed by the effect of holding pressure in our case. So no clear 63

effect of injection pressure on the visibility of the flow marks was observed in our experiment. People found that the frequency and amplitude of the vortex increased with the increase of flow rate [Den Otter, 1970]. So the wavelength and width of the flow marks increased as the flow front velocity was increased. Furthermore, the frequency increased with the increase of flow front velocity, as shown in Fig. 3.28. The trend of frequency is similar to that of gross melt fracture in extrusion [Den Otter, 1970]. Increasing the melt or mold temperature may increase both the wavelength and the width of the flow marks. This may be explained as follows: The increase of melt or mold temperature decreases wall shear stress, causing a more stable entry flow. The

molecules relax and move more easily in the mold cavity. As for the effect of the mold thickness, the oscillating flow has more space to move before it hits the mold wall and bounces back in the thick mold although the gate wall shear stress is smaller. So the wavelength and width would be larger. It is well known that the gross melt fracture (entry instability) happens at a critical wall shear stress, independent of temperature [Kazatchkov, et al., 1995]. So, as the melt temperature is increased, flow marks occur at higher flow front flow velocities to reach the same critical wall shear stress, i.e., Vcri increases with the increase of the melt temperature. However, mold temperature had little effect on the wall shear stress in our case, so the flow marks happened almost at the same flow front velocities at different mold temperatures.

3.1.4 Conclusion

For the alternate flow marks, the effect of polymer rheology, injection speed, mold geometry, melt temperature, mold temperature, holding pressure, injection 64

pressure, and mold surface coatings on the appearance of the alternate flow marks was studied. It was found that a polymer with the highest dynamic viscosity, elastic

modulus, first normal stress difference, transient extensional viscosity, and the longest relaxation time exhibited the alternate flow marks. For the alternate dull and shiny flow marks, flow front velocity is a very important variable. The flow marks occurred above a critical wall shear stress, but disappeared at high injection speeds. For the wavelength and the width of the flow marks, mold geometry or mold temperature had an effect. However, melt temperature did not have much effect. The flow marks could be reduced by one or more of the following factors: high injection speed, high melt or mold temperature, mold surface coatings, and/or changing molecular weight or its distribution. It was found that there was no difference between the crystallinity of dull regions and shiny regions. The melt in dull regions was slightly oriented while the melt in shiny regions was highly oriented. It was also found that coating these surfaces could not prevent the occurrence of the flow marks, although it could alleviate them. Slip was not the cause of the generation and disappearance of the alternate flow marks. The

generation of the flow marks could be explained by an entry viscoelastic flow instability.

3.2 SYNCHRONOUS DULL AND GLOSSY FLOW MARKS 3.2.1 Introduction

Several types of flow marks may occur during polymer melt injection molding processes, such as alternate dull and glossy flow marks and synchronous dull and glossy

65

flow marks. Flow marks cause aesthetic defects on the surface of molded parts, and are very difficult to cover with paints. Because they are not well understood, much attention has been paid to the flow marks in recent years. Although alternate dull and shiny flow marks [Yokoi, 1994b; Chang, 1996a; Chang, 1996b; Hobbs, 1996; Heuzey, et al., 1997; Hamada and Tsunasawa, 1996; Dharia, 1999; Bulters and Schepens, 2000; Grillet, et al., 2000; Charmeau, et al., 2000; Xu and Koelling, 2001; Jayaraman, 2002] and wavelike flow marks [Tredoux and Satoh, 1999; Tredoux, et al., 2000; Yokoi, et al., 1994a; Yoshii, et al., 1993; Yoshii, et al., 1996; Lee and Mills, 1994] have been studied in literature, little work has been undertaken to synchronous flow marks [Yokoi, et al., 1994c; Salamon, et al., 1998]. This type of flow marks is characterized as repeated dull and glossy regions where a dull/glossy zone on one side corresponds to a dull/glossy zone on the other side, as shown in Fig. 3.29. Yohoi [Yokoi, et al., 1994c] found that the gate shapes and mold surface quality had a great effect on the generation of flow marks and the flow marks varied dramatically with the front flow velocity. At a higher melt temperature, flow marks become thinner. At the melt temperature of 240C, flow marks as well as pressure variation did not occur. They explained that during the filling process, the gate pressure was increased, while the melt velocity was decreased at the flow front and the melt was cooled down. On the other hand, the gate was frozen and the flow resistance was increased. Therefore, the melt at the flow front was accelerated with the release of the high gate pressure, and thus the resin near the flow front underwent cooling. Consequently, the transcription precision decreased in the subsequent fountain flow 66

process. Then the flow marks were formed. In injection molding with a center-gate, halos similar to flow marks with synchronous dull and glossy regions were formed [Salamon, et al., 1998]. It was found that the temperature gradient was the cause of the formation of the halos. It was also shown that with the increase of injection rates, the halos became more prominent and the diameter of the halos became larger. In this paper, we study the effect of operating parameters and mold surface coatings on flow marks of different polymer melts. Possible correlation between gross melt fracture in extrusion and synchronous dull and glossy flow marks in injection molding is discussed, and a possible mechanism is proposed.

3.2.2 Experimental

Molding experiments were conducted on a Sumitomo SG M-HP 180-ton injection molding machine. The materials used were two types of high-density

polyethylene, named HDPE1 and HDPE2. Two rectangular molds were employed with different thicknesses (1 and 5.1 mm). The edge gates were used with a thickness of 0.5 mm for the thin mold and 2.5 mm for the thick mold. The length and width of the mold channel was 150 and 51 mm, respectively. The melt temperature was 180, 210 and 240C, while the mold temperature varied from 20 to 70C. The complex viscosity, storage and loss modulus, and first normal stress difference were measured by a Rheometrics RMS 800. A tensile tester, Instron 8511, was used to measure the

extensional viscosity at 100C, based on the standard ASTM test. The samples were standard tensile bars with 13 mm in the narrow-section width, 57 mm in total length, and 67

3.2 mm in thickness. To obtain the constant strain rate, one needs to program the Instron machine to follow the exponential type increase of sample length. Dynamar 9613 (a 3M product), a fluorocarbon elastomer, was used as a coating agent. Its dilute acetone solution, ca. 1%, was coated on hot surfaces and then the solvent was allowed to evaporate. A Differential Scanning Calorimeter from TA Instruments, DSC 2920, was used to measure the crystallinity of dull and shiny regions. The sample was scanned from 30 to 200C at the rate of 10C/min. A Scanning Electron Microscopy, Philips XL 30, was employed to observe the morphology of dull and shiny regions. An optical profilometer, Wyko NT330, was used to measure the roughness of the dull and shiny regions. To exam the melt fracture phenomena, a two-stage single-screw extruder (Rheomex 252p) from Haake was applied. The screw has a diameter of 3/4 inches and a length to diameter ratio (L/D) of 25. A capillary die with 1.2 mm in diameter and 12 mm in length was used. The temperature profile from front zone to the die was 100C /125C /145C /145C. The flow rate was calculated by dividing the measured sample weight collected by the time duration. Because it is very difficult to estimate the wall shear stress in molding filling, CMOLD 2000 was used to simulate filling molds. First the geometry was built, then the mesh was generated. In the simulation, processing parameters, such as shot size, V/P switch pressure, holding pressure, holding time, cooling temperature and cooling time, were the same as those in the real injection molding processes. After the simulation, the wall shear stresses at the center of flow front and gates were read. 68

3.2.3 Results and Discussion 3.2.3.1 Rheological Characterization

The complex viscosity of HDPEs was measured at 180, 200 and 220C. Fig. 3.30 shows the complex viscosity at 180C. It was found that the complex viscosity decreased with an increase in frequency. Fig. 3.31 shows the storage modulus and loss modulus at 200C. Fig. 3.32 shows the first normal stress difference N1 versus shear rate at 180C. It was found that HDPE1 had a larger complex viscosity, elastic modulus, viscous modulus, and first normal stress difference than HDPE2. Fig. 4.33 shows that the extensional viscosity vs. time at the strain rate of 0.001. It was found that HDPE1 had a larger extensional viscosity at the same strain rate.

3.2.3.2 Injection Molding Experiment

It was found through the experiment trials that both HDPEs exhibited synchronous flow marks at certain processing conditions. generally in the phase between the top and the bottom. The flow marks were

Flow mark description

The flow marks occurring were characterized as synchronous dull and shiny regions. Flow marks occurred in the thin mold only when the flow front velocity was large. The flow marks did not occur in the thick mold. The flow marks did not occur immediately after the gate. With an increase in the injection speed, flow marks became 69

more pronounced and continuous. When the injection speed was further increased, flow marks became more continuous and it was difficult to distinguish different regions. The width of the dull regions changed with the velocity of the flow front. Fig. 3.34 shows a typical sample of flow marks for HDPE2.

Effect of operating conditions on flow marks

It was found that flow mark patterns changed as the injection speed increased. Fig. 3.35 shows the effect of flow front velocity on the wavelength for HDPE2. It was found that for the thin mold, the wavelength decreased with the increase of the flow front velocity. Furthermore, the wavelength is shorter at a lower melt temperature; nevertheless, at a higher melt temperature, the wavelength is not affected by the melt temperature. Fig. 3.36 shows that for HDPE2 the lower the mold temperature, the longer the wavelength. It was also found that with an increase of the mold temperature, the flow marks were dimmer. For the synchronous flow marks of HDPE2, the width of the dull regions was usually very narrow, around 1-3 mm at the flow front velocity ranging from 0.4-0.9 m/s. At a higher flow front velocity, the dull and shiny regions became irregular and mixed together, making it very difficult to distinguish different regions. However, it was clearly observed that the width increased with an increase of the flow front velocity. It was found that flow marks occurred above a certain flow front velocity, Vcri. It was further found that the mold temperature almost had no effect on Vcri. However, Vcri increased as the melt temperature increased, as shown in Fig. 3.37. It was also found in 70

the experiment that the flow marks were less visible as the mold temperature increased. The flow marks were almost invisible when the mold temperature was larger than 85C.

Mold surface coating

The effect of the coating on the surfaces of the mold or gate was studied. It was found that the coating on these surfaces could not prevent the occurrence of flow marks, although it could alleviate the flow marks and make them dimmer. Another interesting phenomenon is that coating on the mold surfaces did not change the Vcri, implying that slip is not the cause of the synchronous flow marks. The reason is that coating on the mold surface reduces the critical shear stress where the slip occurs, thus decreasing the Vcri where the slip is triggered. From the above discussion, we can see that for the wavelength and the width of flow marks, injection speed is the most important factor and mold thickness plays a role. Changing the melt temperature and mold temperature has an effect on the flow marks.

3.2.3.3 Morphology and Crystallinity

The DSC experiment showed that there was no difference observed in the cystallinity between the dull and shiny regions for both HDPEs. From the SEM, it was found that the polymer in the shiny region was highly oriented, but the polymer was slightly oriented in the dull region, as shown in Fig. 3.38. This is in agreement with our previous results for alternate flow marks and other researchers results [Salamon, et al., 1998]. The measured average roughness by optical profilometer was smaller in shiny 71

regions than in dull regions, as shown in Table 3.4. The reported average roughness is the average value of 5 randomly selected positions.

3.2.3.4 Extrusion

The HDPEs were extruded at the die temperature 145C. It was found that above a certain wall shear stress, sharkskin melt fracture occurred for both HDPEs. At a higher wall shear stress, spurt flow instability occurred. At a still higher wall shear stress, gross melt fracture occurred. The extrudate irregularity was helical. The flow curve of HDPE2 is shown in Fig. 3.39. The average pressure was used to calculate the wall shear stress when pressure oscillation occurred. The wall shear stress was estimated by

P without the Bagley correction, where P is the pressure drop in the die, L is the 4L / D

die length, and D is the diameter of the die. The apparent shear rate was calculated by 32Q , where Q is volumetric flow rate. Th experiment showed that the critical wall D 3 shear stress for the onset of sharkskin is about 0.24 MPa, and the critical wall shear stress for the onset of the helical irregularity is about 0.51 MPa. The typical examples of smooth surface, sharkskin, spurt flow, and helical melt fracture are shown in Fig. 3.40.

3.2.3.5 Simulation

To obtain wall shear stress during filling of the spiral molds, a simulation was run on C-MOLD 2000. The critical wall shear stresses at the middle of the gate were

72

obtained for HDPE2 where the flow marks began to form. The processing parameters and Vcri were determined from the injection molding experiments. Fig. 3.41 shows the critical wall shear stress at the middle of the gate at different melt temperatures. It was found that the wall shear stresses were very close before the F/P switch. That means the flow marks start to form at the same wall shear stress 0.84 MPa at the gate at different melt temperatures. Fig. 3.42 shows the critical wall shear stress at the middle of the gate vs. the filling percentage at different mold temperatures. It shows that for the same resin, the wall shear stress generally did not change much at different mold temperatures. The decreasing sections of the curves in Figs. 3.41 and 3.42 were the pressure holding stages. Short shots happened for the samples. From the simulation, it was found that flow marks started at the same critical wall shear stress independent of melt temperature and mold temperature.

3.2.3.6 Mechanism

The extrusion instability is associated with the appearance of distortion on the extrudate surface, sometimes accompanied by oscillatory flow. For linear polymer melts such as LLDPE and HDPE, when the shear stress reaches a critical value, the surface becomes rough and wavy, and sharkskin occurs. At a higher stress, slip-stick or spurt flow occurs. At a still higher stress, a transition region may occur where the surface is relatively smooth with long-wavelength distortion. At much higher stress, gross or wavy distortion occurs [Denn, 2001]. This is in agreement with our extrusion experimental results. Although there is a disagreement about the cause of the origin of gross melt 73

fracture, most researchers agree that entry instability causes the melt fracture [Piau and Agassant, 1996; Piau, et al., 1990; Den Otter, 1970]. The instability is also affected by various properties, such as polymer structure, geometry of die entry, melt temperature, and die temperature [Piau and Agassant, 1996]. For the generation of vortices and gross melt fracture, a detailed introduction can be found in Section 3.1.3. For viscoelastic fluids, a corner vortex may occur at a low flow rate before a contraction. The corner vortex zone is a dead zone and does not interact with the outside fluid. The formation of vortices may be due to the increasing extensional viscosity with the deformation rate and/or the shear-thinning effect [Denn Otter, 1970; Cpgswell, 1972]. For some viscoelastic fluids, as a flow rate is increased, it grows inward toward lip [Yesilata, et al., 1990]. The upstream flow is steady in this stage. At a very high flow rate, it grows upstream, and the corner vortex fluctuates and makes the flow field entirely unstable. For some other viscoelastic fluids, as the flow rate is increased, two types of vortices, corner vortex and lip vortex, coexist. As the flow rate is increased, the corner vortex and lip vortex recirculating areas expand [Piau and Agassant, 1996; Yesilata, et al., 1999]. Then the lip vortex gradually develops and invades the corner vortex, and finally generates a single area of recirculation. At a still higher flow rate, the flow becomes unstable and the vortex pulsates or rotates, causing a global change of flow structure [Boger, et al., 1986; Rothstein and McKinley, 2001]. It was found that the development of upstream instabilities governed the appearance of the extrudates and the helix pitch. The melt fracture occurs when the vortex is unstable. The amplitude and frequency of pulsation increases with the pressure [Piau and 74

Agassant, 1996].

Usually the surface distortion and the frequency of the vortex

pulsation are identical. In short, vortices may form before a contraction, and unstable vortices and gross melt fracture are closely related. In fact, the process that polymer melts experience in injection molding is similar to that of extrusion, as shown in Fig. 3.18. That is, the polymer melt meets a contraction and experiences high shear stress at the die or gate, then the polymer melt leaves the die or gate and the polymer chains relax. Thus, for polymer melt experienced injection molding, we could logically expect that a vortex may form and an instability may happen at the entry when the flow rate is high. The oscillating entry instability can propagate and affect the downstream flow. Thus

oscillating or pulsating flow may also occur in the mold, and the different history of heating and shearing which the melt experiences generates flow marks. The different flow marks, alternate dull and glossy flow marks and synchronous dull and glossy flow marks, may be due to two different types of vortex instabilities, oscillating or pulsating instability. Furthermore, the pulsating instability probably results in synchronous flow marks. The pulsating flow makes the flow front velocity change periodically from low to high, and it also changes the flow front melt temperature and shear stress periodically, as shown in Fig. 3.43. Thus, this periodical changing generates synchronous flow marks because the top and bottom are identical, i.e. in phase. Our experiment showed that both HDPEs exhibited helical gross melt fracture at a high flow rate, so it is not surprising that HDPEs exhibited flow marks. In fact, synchronous flow marks of HDPE2 started to occur at the same wall shear stress independent of melt and mold temperature, implying 75

that they have the same characteristic as gross melt fracture and may have the same origin of abnormal appearance the entry instability. The reason is that entry

instability for the generation of gross melt fracture in extrusion occurs at the same critical wall shear stress level and does not changed with the die temperature [Kazatchkov, et al., 1995]. This mechanism could explain our experimental results of flow marks. It was found that the frequency of the flow marks increased with the increase of flow front velocity, as shown in Fig. 3.44. The frequency was calculated by dividing the flow front velocity by the wavelength. The trend of frequency is in a reasonable range compared to the frequency of vortices of gross melt fracture in extrusion [Denn Otter, 1970]. We believe that entry viscoelastic instability accounts for the synchronous dull and glossy flow marks.

3.2.4 Conclusion

For synchronous dull and glossy flow marks, the effect of operating parameters, mold geometry, and mold surface coatings on the flow marks was studied. Synchronous dull and glossy flow marks occurred above a certain flow front velocity. It was also found in the experiment that the flow marks were less pronounced as the mold temperature increased. It was found that there was no difference between the

crystallinity of dull and shiny regions. However, the polymer was highly oriented in shiny region while it was slightly oriented in dull regions. It was also found that mold surface coatings did not eliminate the flow marks. Mold surface coatings scarcely 76

changed the Vcri, meaning that slip was not the cause of the generation of the flow marks. Extrusion experiments showed that helical gross melt fracture occurred for both HDPEs. Finally, it was proposed that entry viscoelastic instability was the reason for the generation of the synchronous flow marks.

77

PP-A Relaxation time (s) Zero viscosity o (Pa.s)

s (Pa.s)

PP-B 0.447 5446 619

PP-C 1.382 30166 1500

PP-D 0.548 7445 787

0.223 915 182

Table 3.1 Relaxation time and zero viscosity at 200C

78

Polymer M

PP-A 128,000

PP-B 142,000

PP-C 121,000

PP-D 164,000

Table 3.2 Viscosity-molecular weight

79

Sample 1 (nm) Shiny 495.9 Dull 692.4

Sample 2 (nm) Shiny 387.1 Dull 571.1

Sample 3 (nm) Shiny 435.8 Dull 504.4

Sample 4 (nm) Shiny 530.4 Dull 571.2

Sample 5 (nm) Shiny 336.9 Dull 344.7

Table 3.3 Average roughness of the dull and shiny regions

80

Sample 1 (nm) Shiny 453.8 Dull 514.1

Sample 2 (nm) Shiny 422.0 Dull 480.6

Sample 3 (nm) Shiny 398.8 Dull 484.8

Sample 4 (nm) Shiny 390.8 Dull 467.1

Sample 5 (nm) Shiny 356.3 Dull 486.3

Table 3.4 Average roughness of the dull and shiny regions

81

Dull region

Dull regions are out of phase on the top and the bottom

Fig. 3.1. Alternate dull and glossy regions.

82

1.E+05

PP-A PP-B PP-C PP-D

Viscosity (Pa.s)

1.E+04

1.E+03

1.E+02 1.E-01

1.E+00

1.E+01

1.E+02

Frequency (1/s)

Fig. 3.2. Comparison of viscosity vs. frequency at 200C.

83

1.E+05
180C 200C 220C

Viscosity (Pa.s)

1.E+04

1.E+03

1.E+02 1.E-01

1.E+00 1.E+01 Frequency (1/s)

1.E+02

Fig. 3.3. Comparison of complex viscosity of PP-C at 180, 200, and 220C.

84

1.E+05

Modulus G' and G" (Pa)

1.E+04

1.E+03
PP-A G' PP-A G" PP-B G' PP-B G" PP-C G' PP-C G" PP-D G' PP-D G"

1.E+02

1.E+01

1.E+00 1.E-01

1.E+00

1.E+01

1.E+02

Frequency (1/s)

Fig. 3.4. Comparison of elastic and viscous modulus at 200C.

85

1.E+04

1.E+03

N1 (PA)

1.E+02
PP-A 200C

1.E+01

PP-B 200C PP-C 200C PP-D 200C

1.E+00 0.001

0.010

0.100

1.000

10.000

Shear rate (1/s)

Fig. 3.5. First normal stress difference vs. shear rate at 200C.

86

10000 180C 200C 220C 1000

N1 (Pa)
100 10 0.001 0.010 S hear rate (1/s) 0.100

Fig. 3.6. The first normal stress difference of PP-C vs. shear rate at 180, 200, and 220C.

87

1.E+09

Extensional viscosity (Pa)

1.E+08
PPA PPB PPC PPD

1.E+07 0.1 1.0

Time (s)

10.0

100.0

Fig. 3.7. Transient extensional viscosity at 130C.

88

1.E+04 1.E+03 1.E+02 1.E+01 1.E+00 1.E-01 1.E-02 1.00E-02

n' ' 2n"/w 2"/

', 2 "/

' Giesekus
2"/ Giesekus

1.00E-01

1.00E+00

1.00E+01

1.00E+02

1.00E+03

Reduced frequency T (1/s)

Fig. 3.8. Determination of relation time by one-mode Giesekus model.

89

a: 0.02"/s

b: 0.1"/s

c: 0.5"/s

d: 2"/s

e: 4"/s

f: 6"/s

Fig. 3.9. Flow marks of PP-C at different injection speeds. 90

Fig. 3.10. A typical example of the alternate dull and shiny flow marks.

91

18 16
(mm)

14 12 10 8 6 4 2 0 0 2 4 Flow front ve locity (m/s) 6

Tmold=22C M old thickness: 1 mm Tmelt=190C Tmelt=225C Tmelt=260C

Wavele nth

Fig. 3.11. Effect of melt temperature on the wavelength .

92

16 14
(mm) Wavele ngth

12 10 8 6 4 2 0 0 0.5 1 Flow front velocity (m/s) 1.5

Tmelt=190C M old thickness: 1 mm Tmold=22C Tmold=50C Tmold=80C

Fig. 3.12. Effect of mold temperature on the wavelength .

93

40 35
(mm)

30 25 20 15 10 5 0 0 0.5

Wavele ngth

Thickness 1 mm Thickness 5.1 mm T melt=190C T mold=22C

1.5

Flow front ve locity (m/s)

Fig. 3.13. The effect of mold thickness on the wavelength .

94

Width of flow marks (mm)

8 7 6 5 4 3 2 1 0 0 2 4 6 Flow front velocity (mm/s)

Tmold=22C M old thickness: 1 mm Tmelt=190C Tmelt=225C Tmelt=260C

Fig. 3.14. Effect of melt temperature on the width of the flow marks.

95

Width of flow marks (mm)

8 7 6 5 4 3 2 1 0 0 0.5 1 Flow front velocity (mm/s) 1.5

Tmelt=190C M old thickness: 1 mm Tmold=22C Tmold=50C Tmold=85C

Fig. 3.15. Effect of mold temperature on the width of the flow marks.

96

35 30
Width (mm)

25 20 15 10 5 0 0 0.5 1 Flow front ve locity (m/s) 1.5

Tmelt=190C Tmold=22C Thickness 1 mm Thickness 5.1 mm

Fig. 3.16. The effect of mold thickness on the width of the flow marks.

97

0.25 0.2 T mold =22 o C

Vcri (m/s)

0.15 0.1 0.05 0 150

200

250

300

Melt tempe rature ( o C)

Fig. 3.17. The starting of the flow marks, Vcri vs. melt temperature.

98

10 9
Transition velocity (m/s)

8 7 6 5 4 3 2 1 0 150

T mold =22 o C

200 250 Me lt temperature (o C)

300

Fig. 3.18. Effect of melt temperature on the transition velocity, Vtrans.

99

Transition & critical velocity (m/s)

1.E+01

Vtran
1.E+00

Flow mark zone

1.E-01

Vcri Tmold=22C
1.E-02 150 200 250
o

300

Melt temperature ( C)

Fig. 3.19. Flow mark zone of PP-C.

100

(a) Shiny region

(b) Dull region

Fig. 3.20. Morphology of surfaces of dull and shiny regions.

101

(a) Low wall shear stress

(b) High wall shear stress

Fig. 3.21. Gross melt fracture of the PP in extrusion. 102

0.25
Wall she ar stre ss (MPa)

0.2 0.15 0.1 0.05 0 1.E+01

1.E+02

1.E+03

1.E+04

Appare nt she ar rate (1/s)

Fig. 3.22. The wall shear stress versus apparent shear rate in the extrusion.

103

0.25

Shea r stress(Mpa )

0.2 0.15

PP-A PP-B PP-C PP-D

0.1 0.05 0 0 20 40 60 80 100

Fill%

Fig. 3.23. Wall shear stress vs. percentage filled in the thin spiral mold.

104

0.4

0.3
Wall Shear Stress (MPa)

0.2

Tmelt=190C
0.1

Tmelt=225C Tmelt=260C

0 20%

30%

40%

50%

60%

70%

80%

90%

100%

Filling Pe rce ntage

Fig. 3.24. The critical wall shear stress at the middle of the gate at different melt temperatures.

105

0.3

Wall Shear Stress (MPa)

0.2

Tmold=22C
0.1

Tmold=55C Tmold=80C

0 20%

30%

40%

50%

60%

70%

80%

90%

100%

Filling Pe rce ntage

Fig. 3.25. The critical wall shear stress at the middle of the gate at different mold temperatures.

106

die barrel

extrudate

runner

gate mold

Fig. 3.26. The similarity between extrusion and injection molding processes.

107

gloss Cavity thickness

normal

dull Flow front

Flow direction

dull

normal

gloss

Fig. 3.27. Oscillating flow generates alternate flow marks.

108

70 60

Frequency (1/s)

50 40 30 20 10 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Flow front velocity (m/s)

Fig. 3.28. Frequency of the flow marks versus flow front velocity.

109

Dull region

Dull regions are on the phase on the top and the bottom

Fig. 3.29. Synchronous dull and glossy regions.

110

1.E+04

(Pa.s)

1.E+03
HDPE1 HDPE2

1.E+02 0 1 Frequency (1/s) 10 100

Fig. 3.30. Comparison of viscosity vs. frequency at 180C.

111

1.E+05

G',G" (Pa.s)

1.E+04

1.E+03

HDPE1 G' HDPE1 G" HDPE2 G' HDPE2 G"

1.E+02 0.1 1.0 10.0 Frequency (1/S) 100.0

Fig.3.31. Comparison of Elastic and viscous modulus at 180C.

112

1.E+05

1.E+04

N 1 (Pa)

1.E+03

1.E+02

HDPE1 HDPE2

1.E+01

1.E+00 0.1 1 S hear rate (1/s) 10 100

Fig. 3.32. First normal stress difference vs. shear rate at 180C.

113

1.E+10

Extensional viscosity (Pa)

1.E+09

1.E+08 HDPE1 HDPE2 1.E+07 0.1 1.0 10.0 Time (s) 100.0 1000.0

Fig. 3.33. Extensional viscosity vs. time at 100C.

114

Fig. 3.34. Synchronous dull and shiny flow marks of HDPE2.

115

10 9 8 7 6 5 4 3 2 1 0
0.2 Tmold=20C Mold thickness: 1 mm Tmelt=180C Tmelt=210C Tmelt=240C 0.4 0.6 0.8 Flow front velocity (m/s) 1

Wavelength

(mm)

Fig. 3.35. Effect of melt temperature on wavelength.

116

10 (mm) Wavelength 9 8 7 6 5 4 3 2 1 0 0.2 T melt =210C Mold t hickness: 1 mm

Tmold=20C Tmold=50C
0.4 0.6 0.8 Fl ow fron t ve l oci ty (m /s) 1

Fig. 3.36. Effect of mold temperature on wavelength.

117

0.6

0.4 V cri (m/s)

0.2

0 150

200 Me lt te m pe ratu re ( o C )

250

Fig. 3.37. Effect of melt temperature on Vcri.

118

(a) Dull region

(b) Shiny region Fig. 3.38. Morphology of dull and shiny region of HDPE2.

119

0.7

Wall shear stress (MPa)

0.6 0.5 0.4 0.3 0.2 0.1 0 1.E+01

1.E+02

1.E+03

1.E+04

1.E+05

Apparent shear rate (1/s)

Fig. 3.39. Flow curve of HDPE2 in extrusion.

120

Fig. 3.40. Different extrudate irregularities at different wall shear stresses.

121

1.2

1
Wall Shear Stress (MPa)

0.8

0.6

Tmelt=180C Tmelt=210C

0.4

Tmelt=240C
0.2

0 50%

60%

70% Filling Perce ntage

80%

90%

Fig. 3.41. Critical wall shear stress vs. percentage filled at different melt temperatures.

122

1.2

1
Wall Shear Stress (MPa)

0.8

0.6

Tmold=20C Tmold=50C

0.4

Tmold=70C
0.2

0 50%

55%

60%

65%

70%

75%

80%

85%

Filling Pe rce ntage

Fig. 3.42. Critical wall shear stress vs. percentage filled at different mold temperatures.

123

gloss Cavity thickness

normal

dull Flow front

fast

normal

slow Flow direction

gloss

normal

dull

Fig. 3.43. Pulsating flow generates synchronous flow marks.

124

180 160 140 120 100 80 60 40 20 0 0 0.2 0.4 0.6 0.8 1 T mold=20C Mold thickness: 1 mm

Frequency (1/s))

Tmelt=180C Tmelt=210C Tmelt=240C

Flow front velocity (m/s)

Fig. 3.44. Frequency of flow marks vs. Flow front velocity.

125

CHAPTER 4

EXPERIMENT WITH MICRO-FEATURES AND IMPROVEMENT OF SIMULATION ACCURACY DURING THIN-WALL INJECTION MOLDING

4.1 THIN-WALL INJECTION MOLDING WITH MICRO-FEATURES 4.1.1 Introduction

Injection molding of thermoplastics with micro-features is a new field in thinwall applications. In recent years, the fabrication of polymer-based micro-components for optical and biomedical applications has been given increasing attention in industry and academia [Yu, et al., 2004b]. Polymer materials are favored because of their low cost, good biocompatibility, high optical clarity, and high impact strength compared with silicon or glass. Micro-injection molding has the potential for economical mass-

production. It usually combines various lithography techniques and injection molding [Weber and Ehrfeld, 1999]. Two types of micro-parts are available: micro-sized parts and regular-sized parts with micro-features. Micro-injection molding (MIM) is the injection molding of plastic parts with structure dimensions in the micron or sub-micron range. The micro-features can be considered very thin-wall parts. The replication of

126

micro-features is an important issue and depends greatly on the size, aspect ratio and covered area [Weber and Ehrfeld, 1999]. Furthermore, it is a challenge to simulate micro-injection molding processes. It has been shown that standard injection molding packages cannot describe all of the effects in micro-injection molding [Kenmann, et al., 20002; Yu, et al., 2002]. This study focuses on thin-wall injection molding with microfeatures by experiment and numerical simulation. The filling lengths in microchannels are simulated and compared with experimental results.

4.1.2 Experimental

A high-speed and high-pressure injection-molding machine, Sumitomo SG 180 M-HP, was used in our experiment. Two rectangular molds were employed. The mold cavity contained not only the base plate but also the microchannels. The main lengths of the two molds were 203 mm and 72 mm, respectively. The long rectangular thin-wall mold was 2 mm in thickness, 203 mm in length and 50.8 mm in width, as shown in Fig. 4.1. The distance from the last channel (channel B) to the end of the main flow is 135 mm for the long mold, as shown in Fig. 4.2. The corresponding distance for the short mold is 4 mm, as shown in Fig. 4.2. An edge gate with 3 mm in width was used for the main flow. A disk-like mold insert with a diameter of 508 mm was installed in the mold base. Two micro mold inserts were tested. One mold insert, including top view and side view, is shown in Fig. 4.3. The mold insert includes six microchannels made with wireEDM. There are three separate microchannels and the other three are next to each other

127

with a distance of 100 micrometers. All channels are 500 m deep and 100 m wide, giving an aspect ratio of 5. The detailed structure of a microchannel from SEM is shown in Fig. 4.4. Another mold insert has similar geometry but smaller width of

microchannels. The channels are 250 m deep and 50 m wide, giving an aspect ratio of 5 too. Two pressure sensors made by Kistler were mounted at the position right before and after the insert, respectively. The data acquisition system was built based on the data acquisition board from Keithley. The materials used were a semi-crystalline polymer, polypropylene (PP, Inspire C703-35U, Dow Chemical), and an amorphous polymer, poly (methyl methacrylate) (PMMA, PL 150, Plaskolite). The melt temperature was 240C for PMMA and

240/260C for PP. The mold temperature was 25 and 80C for both materials. The holding pressures were set at 0 psi or 500 psi. The injection speed varied from 0.2 to 5 inch/s. The complex viscosity, storage and loss modulus were measured by a Rheometrics RMS 800. A Scanning Electron Microscope (SEM), the Philips XL 30, was employed to observe the microchannels. An optical profilometer, Wyko NT3300, was used to measure the filling length of the microchannels. The vertical-scanning interferometry (VSI) mode was used to measure step heights by multiple wavelengths of light. The profilometer basically uses the interference of light to determine the surface shape and transmission properties. First a light source is split into two beams, then a pattern of interferences or fringes is formed when the two beams are reflected from a test

128

surface and a reference surface and join together.

A series of fringe patterns are

generated when the test surface is scanned. The recorded fringe patterns can be mapped [NDSU Center for Nanoscale Science and Engineering, 2003]. The non-contact Wyko NT3300 features outstanding software and advanced automation for highly accurate, three-dimensional surface topography measurements. Height can be measured from Angstroms to millimeters at a resolution of 0.1 nm. The profile of an object is

determined using interferometry instead of a stylus. Hence, the instrument is ideal for measuring micro-structure profiles because they can be measured without destroying their structure.

4.1.3 Experimental Results

The dynamic viscosity of PP and PMMA was measured by a Rheometrics RMS 800, as shown in Figs. 4.5 and 4.6. It was found that PP and PMMA were typical shear thinning thermoplastics. The parts were molded for PP and PMMA under different processing conditions. It was found that demolding was easy when the filling length in the microchannels was small. However, demolding was difficult when the filling in the micro channels with the width of 100 m was deeper than 150 m for PMMA and 300 m for PP. In the long mold, deeper filling could be observed in the adjacent channels than in the separated ones, and the difference might be significant. This is due to a reduced heat loss for the melt in adjacent microchannels. However, the filling lengths

129

were almost the same in 3 separated channels. Fig. 4.7 shows a SEM picture of a molded microstructure from PP. The molded parts with 100 m microchannels were examined after injection molding with the optical profilometer, to determine the replication accuracy of microfeatures. The measured filling lengths in the separated microchannels in the long mold against the main flow velocity for PMMA and PP are shown in Figs. 4.8 and 4.9, respectively. The filling lengths are longer at a higher main flow velocity for both PP and PMMA. The molding of the microchannels is similar to thin-wall injection

molding. A high injection speed is the most efficient way to increase the flow length. A higher mold temperature results in a longer filling length due to less flow resistance. Furthermore, higher holding pressure generates a somewhat longer filling length, although the effect is not significant. PP can achieve longer filling lengths than PMMA. A complete filling can be observed for PP at a high injection speed. It should be pointed out that at the mold temperature of 25C, the cavity volume was not totally filled and no packing stage occurred. However, the filling length for PP reached 500 m. It implies that the microchannels may be completely filled simply in the filling stage by PP. The filling lengths in 50 m micro mold are shown in Figs. 10 and 11. It was found that PP can completely fill the micro channels but PMMA can fill only less than about 50 m. That is, it is much more difficult to fill the microchannels when the size is scaled down. PP is much easier to fill the microchannels than PMMA. The effect of channel location was studied, as shown in Fig. 4.12. It was observed that the filling lengths in channels A and B in the short mold are much longer 130

than those in the long mold. Furthermore, the filling length in channel B is much longer than in channel A in the short mold. However, there is not much difference between the filling lengths in channels A and B in the long mold. The difference is that channel B is much closer to the end of the mold cavity in the short mold than in the long mold, as shown in Fig. 4.2. The recorded cavity pressure vs. filling time in the short mold and the long mold is shown in Fig. 4.13. If we define the filling time at which the main flow reaches the mold cavity end as tc, the pressure profile is the same before tc, and then the pressure increases sharply in the short mold. In the short mold, more melt can be packed into the channels and the filling lengths in channels A and B are longer than in the long mold. Also in the short mold, the filling length in channel B is longer than in channel A because the melt in channel B experienced a shorter cooling time. However, in the long mold, the polymer melt needs a relatively much longer time to flow from channel B to the end of the mold (135 mm), and the melt in both microchannels may have frozen before the main flow reaches the mold cavity end (i.e. before the sharp pressure rise). Therefore, the cooling time does not affect the filling lengths at different channel locations. We define the time the polymer melt needs to advance from the channel position to the end of the mold cavity as t, representing the degree of melt freezing in the microchannels. If we plot the dimensionless filling lengths (filling length/depth) vs. Fourier Number t/L2, the dimensionless filling lengths decrease with the increase of Fo. All of the data merge on a single curve, covering both molds and a wide range of main flow velocities, as shown in Fig. 4.14.

131

The effect of packing stage on filling lengths is shown in Fig. 4.15. When the shot size is small ( 0.0305 m), the mold cavity is not totally filled and there is no packing stage. When the shot size is equal or larger than 0.0307 m, the packing occurs. Whether the packing stage occurred or not was judged by a steep increase of the recorded cavity pressure profile and visual observation of short shots. It was found that at the main flow velocity of 200 mm/s, the filling lengths are about 164 m without the packing stage at the melt temperature of 240C, mold temperature of 25C and zero holding pressure; however, the filling length increases with an increase in shot size and the filling length reaches 334 m finally at the shot size of 0.0508 m. At the main flow velocity of 37.5 mm/s, the filling length is about 70 m without packing stage, but the filling length is 200 m with packing stage (shot size 0.0508 m). It implies that the packing stage is very important in filling the microchannels. The effect of holding pressure on filling lengths is shown in Fig. 4.16. It shows that the holding pressure has some effect on filling lengths, but is not significant.

4.1.4 Simulation Results

Because the thickness of the base plate is very large compared to the microchannels, the conventional midplane simulation using the Hele-Shaw

approximation may result in a large inaccuracy. Therefore, a 2D x-z plane simulation is applied (x is the axial flow direction and z is the thickness direction). To save

computational cost, a hybrid model is selected and numerical codes are developed by Liyong Yu. The cavity is divided into three regions: the upstream, the middle, and the 132

downstream region. The 1D Hele-Shaw equation is used in upstream and downstream regions where the control volume/finite element method (CVFEM) is used to solve the Hele-Shaw equation and the finite difference method (FDM) is used to solve the energy equation. The 2D general momentum equation is used in the middle region where (CVFEM) is used to solve the momentum and energy equations numerically. The hybrid approach has 584 triangle elements and 40 1D elements. There are 21 layers in the thickness direction. Detailed information can be found elsewhere [Yu, 2004a; Yu, 2004b]. A series of simulation were run in the long mold by Liyong Yu. Because the heat transfer coefficient h is very difficult to determine, the melt/base plate wall interface heat transfer coefficient h is assumed constant, h=25000 W/m2K, and three different constant melt/microchannel wall interface heat transfer coefficients are tested. Fig. 4.17 shows that the filling lengths are greatly underpredicted for most injection speeds for PP at the mold temperature of 25C. This is because the heat transfers so quickly (h=25000 W/m2K) that the melt near the wall freezes before the melt can enter into the microchannel. If the main flow heat transfer coefficient is selected to be 2000 W/m2K, the polymer melt flows much deeper into the microchannel, as shown in Fig. 4.18. Furthermore, the value of the melt/microchannel wall heat transfer coefficient also plays an important role in predicting the filling length. It can be concluded that the heat transfer coefficient is critical in the filling simulation of a mold with micro-features. Next, a different boundary condition for heat transfer at the wall was further tested: q = hx (Tm Tw ) 133

where Tm is the gapwise mean temperature and hx is the variable local heat transfer coefficient. hx is a function of the flow field and expressed by the local Nuselt number Nux and the hydraulic diameter Dh [Shah and London, 1978]: hx = Nu x k / Dh
0 Nu x = Nu x ( b / w ) 0.25

1/3 + 0.4 for x* 0.001 0 = 1.233 (x*) Nu x 7.541 + 6.874 (1000x*) 0.488 e 245x* for x* > 0.001

x* = x /( Dh Re Pr) where b and w are the bulk viscosity and the viscosity at the wall. Using the variable heat transfer coefficient, the filling length of PP in the long mold was predicted very well at mold temperatures of 25C and 80C, as shown in Fig. 4.19.

4.1.5 Conclusion

Thin-wall injection molding with micro-features was studied experimentally and numerically. It was found that the filling lengths in microchannels are affected by injection speed, mold temperature, and channel location. A high injection speed or high mold temperature results in a longer filling length. Moreover, the filling lengths in the microchannels increase with a decrease in the filling time flowing from the microchannel to the mold cavity end. It can be concluded that the filling stage is important, the

134

packing stage is also important (especially in the short mold), and the holding stage is not important in filling the microchannels with PP. It is more difficult to fill the smaller microchannels. Furthermore, the filling lengths in the microchannels are simulated by a hybrid simulation code with a combination of the momentum equation and the HeleShaw model, and compared with experimental results. The code has fewer elements and requires less computation time. The simulation shows that the filling lengths in

microchannels are sensitive to the heat transfer coefficients in the main flow cavity and in the microchannel, and extra attention is needed to select the proper heat transfer coefficient. By using a variable heat transfer coefficient, the filling length in the long mold was predicted very well.

4.2 CAVITY PRESSURE AND ITS PREDICTION DURING THIN-WALL INJECTION MOLDING 4.2.1 Introduction

Injection mold cavity pressure is one of the most important parameters in the thin-wall injection molding process. It plays an important role in determining the

molded part quality and is a good indicator of injection machine control performance [Angstadt, 2001; Dubay, 2001]. It not only indicates the material condition in the mold but also affects the microstructure and part quality [Macfarlane and Dubay, 2000; Gao, et al., 1996; Gao, et al., 1996]. Cavity pressure can affect part weight, dimensions, cosmetics, gloss, warpage, shrinkage, etc. [Bozzelli and Cardinal, 1996]. Therefore, it is

135

very important to study the effect of injection operating variables and material properties on the cavity pressure (gradient). Computer aided engineering (CAE) programs are commonly used today to design a part successfully, optimize the process, and troubleshoot [Kalnin and Zluhan, 1999]. The application of CAE has the potential to reduce overall production cost and improve part quality. However, the injection molding process is very complicated and many operating variables and physical properties affect the mold cavity pressure. Almost all users would prefer better accuracy of CAE simulation [Ainoya and Amono, 2001]. During thin-wall injection molding (TWIM), the prediction error in cavity pressure from CAE simulation may reach from 50% to more than 100% and the error increases as the parts become thinner [Chen, et al., 2000]. This error may be due to certain assumptions and simplifications. For example, the effect of pressure on viscosity is neglected

although it is important in high pressure thin-wall processes [Chen, et al., 2000; Amano and Ainoya, 2000; Fasset, 1995; Mahishi, 1998]. It was found that neglecting the effect of pressure on viscosity led to overprediction of cavity pressure, while neglecting the juncture loss led to underprediction of nozzle pressure [Sherbelis and Friedl, 1996]. The heat transfer coefficient in CAE packages, such as in C-MOLD and MoldFlow, is usually taken to be a constant (default value 25,000 W/m2K), but it changes with time and operating variables. pvT-data also affect cavity pressure [Ainoya and Amono, 2001]. Cavity pressure drop was extremely overpredicted when the effect of pressure on viscosity and juncture loss were not considered. However, Sridhar and Narh [1999] found that the heat capacity and thermal conductivity had almost no effect on cavity 136

pressure. Another reason for the error of simulation is the lack of a high quality database for polymers, such as heat conductivity and pvT data [Chen, et al., 2000]. In this chapter, the effect of pressure-dependent viscosity, heat capacity, heat transfer coefficient, juncture pressure loss and pvT-data on cavity pressure and pressure drop prediction will be studied, and the importance of each parameter will be evaluated. Then the importance of each variable will be evaluated, and the method to improve the prediction accuracy will also be discussed. The cavity pressure and pressure drop are measured experimentally and compared. The study aims to improve simulation accuracy and offer the guidance to reduce time and cost for expensive property testing.

4.2.2 Simulation

A rectangular mold with a mold thickness of 1 mm was used in the experiment, as shown in Fig. 4.20. A representative amorphous polymer, polystyrene (PS), and a representative semi-crystalline polymer, high-density polyethylene (HDPE), were selected. The melt temperature was 230C and 250C, and the mold temperature was 60C for the PS. The melt temperature was 300C and 320C, and the mold temperature was 80C for the HDPE. Actual molding experiments were performed on a Sumitomo SG M-HP 180 ton injection molding machine. A data acquisition system and software from Keithley Instruments were used to measure cavity pressure. Pressure transducers were made by Kistler Instrument Co. The simulation was run on a commercial FEM software, Moldflow 3.0. To estimate the effect of each factor on cavity pressure, the Taguchi array of the simulation 137

is shown in Table 4.1. Five two-level factors were discussed: the heat capacity Cp, the pressure-dependent viscosity p, the juncture loss P, the heat transfer coefficient h, and the specific volume v. Level 1 of the heat capacity was chosen as a constant heat capacity, and level 2 of the heat capacity changed with temperature. Level 1 of the pressure-dependent viscosity p neglected the effect of the pressure while level 2 included the effect of pressure. Level 1 of the juncture loss P ignored the juncture loss while level 2 considered it. Level 1 of the heat transfer coefficient used Moldflows default value, 25000 W/m2C. Level 2 of the heat transfer coefficient used 1500 W/m2C according to our experience. Level 1 of the specific volume v did not include the effect of pressure, while level 2 included effects of both temperature and pressure. The temperature-dependent heat capacity was measured by a DSC, TA 2920, at the heating rate of 3.33C/s. The results are shown in Fig. 4.21. The pvT modeling was described by a double-domain Tait equation [Chiang, et al., 1991].
v(T,P) = v o [1 C ln( P B(T) )]

(1)

where vo(T)=b1m+b2m T if T>Tt vo(T)=b1s+b2s T if T<Tt (2) (3) (4) (5)

B(T)=b3mexp(-b4m T ) if T>Tt B(T)=b3sexp(-b4s T ) if T<Tt

138

Tt=b5+b6 P
T

(6) (7)

=T-b5

For semi-crystalline HDPE, the additional term vt (T, P) is added for v (T, P) which is well known as the modified Tait equation:
v(T,P) = v o [1 Cln( P B(T) )] + vt (T,P)

(8) (9) (10)

where vt (T, P)=0

if T>Tt

vt (T, P)=b7 exp (b8 T - b9P) if T<Tt

pvT data were given by the Moldflow database and the specific volumes are shown in Figs. 4.22 and 4.23, respectively. The pressure-dependent viscosity was modeled by the Cross-WLF equation [Hieber, 1987] over a wide range of shear rates, as shown below. The parameter D3 characterizes the effect of pressure on the glass temperature T* and thus the viscosity.
= o /[1 + (

& o 1-n ) ] *
A1(T T*)

where o = D1 exp (

T* = D2 + D3 P
A2 = A2 + D3 P

(11)

A2 + T T*

(12)

(13) (14)

139

The viscosity under high pressure was measured by a capillary rheometer, the Rheomex 252p. The measured pressure drops were corrected by the Bagley correction. Furthermore, the viscosity at low shear rates under the ambient pressure was measured by a Rheometrics RMS 800. The viscosity for PS and HDPE is shown in Figs. 4.24 and 4.25, respectively, and the fit model parameters are shown in Table 4.2. To consider juncture pressure loss, the Bagley correction constants C1 and C2 were chosen as Moldflow recommended. That is, C1=6.7910-2 Pa-0.399 and C2=1.399 for HDPE, and C1=3.310-5 Pa-1.108 and C2=2.108 for PS.

4.2.3 RESULTS AND DISCUSSION The simulation results were obtained for PS in the thin-wall injection molding process. The importance of each factor is evaluated by analysis of variance (ANOVA) [Roy, 2001]. The peak cavity pressure values and the percent influence at three different injection speeds, 0.5, 3 and 20 inch/s, are shown in Table 4.3 when the melt temperature is 230C. It was found that the specific volume is the most important factor affecting the peak cavity pressure, and its importance increases with an increase in injection speed. A similar conclusion can be drawn when the melt temperature is 250C, as shown in Table 4.4. It means that a large error may occur if the effect of pressure on the specific volume is neglected and the property should be strictly measured. The discrepancy between the cavity pressures without the effect of pressure on the specific volume and with the effect of both pressure and temperature on the specific volume can be seen in Fig. 4.26.

140

Without considering the effect of the pressure on the specific volume, the simulation predicts a higher peak cavity pressure and relatively lower holding pressure. Two other important factors affecting the peak cavity pressure are the pressuredependent viscosity and the heat transfer coefficient. ANOVA shows that both the heat transfer coefficient and the pressure-dependent viscosity are significant factors when the injection speed is low. It also shows that the percent influence of the heat transfer coefficient is higher than that of the pressure-dependent viscosity, as shown in Tables 4.3 and 4.4. However, our previous simulation [Xu and Koelling, 2003] showed that the effect of pressure on the viscosity is relatively more important than deciding the proper heat transfer coefficient when the injection speed is high at lower melt temperatures. It implies that the percent influence of the heat transfer coefficient and the pressure dependent viscosity depends on both melt temperature and injection speeds. The percent influence of the heat transfer coefficient and the viscosity decreases with the increase of injection speeds. By including the effect of pressure on the viscosity, a higher peak cavity pressure and a lower holding pressure are predicted, as shown in Fig. 4.27. Using h=1500 W/m2.C predicts a lower peak cavity pressure and a holding pressure that is lower at first and then higher, as shown in Fig. 4.28. However, these two parameters are very difficult to measure. Sherbelis and Friedl also predicted lower cavity pressure when the effect of pressure on viscosity was considered [Sherbelis and Friedl, 1996]. The least important factors considered here are the heat capacity and the juncture loss. The percent influence is very small, as shown in Tables 4.3 and 4.4. From Figs. 4.29 and 4.30, it can be seen that the heat capacity and the juncture loss have almost no 141

effect in this case. However, Table 4.3 shows that the contribution of the heat capacity to cavity pressure increases at high injection speeds, which is usually true in thin-wall injection molding. For heat capacity, it is very difficult to get the true value because the cooling rate is very fast in thin-wall injection processes, such that common instruments cannot scan samples fast enough. At a higher melt temperature of 250C, the percent influence of each factor, as shown in Table 4.4, is similar as the melt temperature of 230C. Sridhar and Narh [1999] also found that thermal conductivity and heat capacity had almost no effect on cavity pressure. Other researchers also found that a tabulated heat capacity led to a slightly higher cavity pressure drop [Ainoya and Amono, 2001]. The simulation results of peak cavity pressure for HDPE were also obtained. The percent influence of each factor at two different melt temperatures, 300C and 320C, is shown in Table 4.5. It was found that the specific volume and the heat transfer

coefficient are significant factors. The specific volume has the largest percent influence, but pressure-dependent viscosity also has a relatively large percent influence compared to the effect of the heat capacity and juncture loss. The simulation implies that a large error may occur if the effect of pressure on the specific volume and/or the viscosity is neglected, and/or the heat transfer coefficient is not properly determined. These values should be carefully determined before running simulation. The juncture loss and the heat capacity do not play a significant role in this case. Thus full attention needs to be given to the specific volume, the pressure

dependent viscosity, and the heat transfer coefficient when the material property model

142

is selected for simulation, and less effort should be given in determining the heat capacity and the juncture loss. The maximum cavity pressure drops were simulated for PS. As shown in Tables 4.6 and 4.7, the percent influence of the heat transfer coefficient is the largest among the five factors under study, so the most important factor affecting cavity pressure drop is the heat transfer coefficient. Generally speaking, the percent influence of the viscosity and the specific volume are high, and the juncture loss is not a significant factor. Furthermore, it is interesting to note that the heat capacity is not a significant factor when the injection speed is low, but its percent influence increases dramatically with an increase in the injection speed. However, it shows that the percent influence of the viscosity decreases with an increase in the injection speed. For the pressure drop of HDPE, it was found that the heat transfer coefficient and the viscosity have a high percent influence, as shown in Table 4.8. However, the heat capacity, the juncture pressure loss, and the specific volume have almost no effect. The influence of the material property model is apparently different for different polymer structures. This also implies that the specific volume, the pressure dependent viscosity and the heat transfer coefficient generally play important roles in the maximum cavity pressure drop and these properties need to be carefully tested before running simulations. Furthermore, the heat capacity may be important when a very high injection speed is applied. The actual cavity pressure and pressure drop were also measured. Fig. 4.31 shows the pressure profiles right after the gate and at the end of the cavity at an injection speed of 76.2 mm/s. It was found that higher peak pressures both after the gate and at 143

the end of the cavity were detected when the melt temperature was lower, due to a higher flow resistance. Moreover, the pressure drops more rapidly when the melt temperature is lower because the melt cools down more rapidly. Fig. 4.32 shows the pressure profiles right after the gate at the melt temperature of 230C under different injection speeds. It was found that a higher injection speed caused a lower peak cavity pressure. This agrees with other researchers observations in the thin-wall injection molding process [Chen, et al., 2000; Bozzelli, et al., 1997; Fierens and Mertes, 1998]. Moreover, the pressure drops more slowly when the injection speed is higher, probably because of the short cooling period at the filling stage and the high viscous heating. Fig. 4.33 shows the pressure profiles at the end of the cavity under the melt temperature of 230C under different injection speeds. It was found that the peak pressure increases with an increase in the injection speed, but the peak pressure drops a small amount when the injection speed is further increased from 76.2 to 508 mm/s. Fig. 4.34 shows the measured cavity pressure vs. time right after the gate at the low injection speed of 12.7 mm/s. The simulation results from different material property models are also shown in this figure. It was found that Moldflow predicts the filling stage fairly well, but there is a large difference in the holding stage. It can be seen that neglecting the effect of pressure on the specific volume caused a large difference in the peak cavity pressure. Furthermore, the predicted pressure curve including the effect of pressure on the specific volume and the viscosity (the solid triangles) is closest to the measured pressure curve at the holding stage compared to other predicted curves. At a high injection speed of 508 mm/s, Moldflow overpredicts the cavity pressure at both the filling stage and the holding stage, as shown in Fig. 4.35. Neglecting the effect of pressure on the specific volume leads the 144

largest peak cavity pressure difference between the simulation and measurement, with a difference of about 66%. The predicted pressure curve including the effect of pressure on the specific volume and the viscosity (the solid triangles) is closest to the measured pressure curve at the holding stage, compared to other predicted curves. Unlike the low injection speed, Moldflow predicts the trend of the cavity pressure well at the holding stage, but the predicted curve shifts upward from the experimental pressure curve. The measured maximum cavity pressure drop is 38.9 MPa; however, the predicted pressure drop is in the range of 46-50 MPa depending on the material property models used. It can be seen that to obtain good simulation results, the effect of pressure on the specific volume and the viscosity must be included, and the default value of the heat transfer coefficient must be re-evaluated. At high injection speeds, good agreement cannot be obtained regardless of the property models selected. The reasons are out of the range we are considering. The possible reasons are: a discrepancy between the set operating variables and the actual values the machine reached; an inefficiency in the property models (e.g., a property measured under equilibrium is used to simulate a nonequilibrium injection molding process); and/or the software itself due to simplifications.

4.2.4 CONCLUSION For the thin-wall injection molding processes, it is very important to use proper material property models when running simulations. It was found that the effect of pressure on the specific volume is the most important factor to predict the peak cavity pressure. The effect of pressure on the viscosity and the heat transfer coefficient is also significant. The heat capacity and the juncture loss are relatively less important 145

compared to other factors considered here. It was also shown that the significant factors are somewhat different to predict maximum cavity pressure drop. The heat transfer coefficient is the most important factor, but in general the specific volume and the viscosity are still important. At a high injection speed, the simulation overpredicted the peak cavity pressure and the maximum cavity pressure drop, and good prediction cannot be achieved. Further study is necessary to understand why this happens and how to improve the simulation accuracy at very high injection speeds. However, the differences between the measurements and the simulations are smaller at low injection speeds in our case.

146

No. 1 2 3 4 5 6 7 8

Cp 1 1 1 1 2 2 2 2

1 1 2 2 1 1 2 2

P 1 1 2 2 2 2 1 1

h 1 2 1 2 1 2 1 2

v 1 2 1 2 2 1 2 1

Symbol Cp Pjuncture h v

1 Constant D3=0 C1=0 C2=0 25,000 W/m2K No effect of pressure

2 Variable PS: D3=1.51E-7 PE: D3=9.21E-8 PS: C1=3.3E-5, C2=2.108 PE: C1=0.0679, C2=1.399 1,500 W/m2K Effect of pressure is considered

Table 4.1 Orthogonal array of the simulation

147

Material PS PE

(Pa) 38264 2791

N (-) 0.177 0.542

D1 (Pa.s) 2.72E13 8.86E13

D2 (K) 368 256

D3(K/Pa) 1.51E-7 9.21E-7

A1 (-) 31.0 26.0

A 2 (K)

51.6 51.6

Table 4.2 Coefficients of Cross-WLF equation

148

No. 1 2 3 4 5 6 7 8

Cp 1 1 1 1 2 2 2 2

p 1 1 2 2 1 1 2 2

P 1 1 2 2 2 2 1 1

h 1 2 1 2 1 2 1 2

v 1 2 1 2 2 1 2 1

Peak P (MPa) 0.5/s 3/s 20/s 84.07 72.28 72.95 56.64 48.66 91.78 75.98 60.35 51.11 66.9 54.71 51.2 75.21 51.52 54.09 73.44 54.35 74.2

74.78 68.65 72.91 55.51 78.8 70.61

Percent Influence (%) Injectio n speed 0.5/s 3/s 20/s Cp 0 0 0.32 p 5.64 1.18 0.12 P 0 0.13 0 h 25.77 5.79 0.48 v 67.28 92.60 98.71 Significant Factors , h, v , h, v v

Table 4.3 Relative influence of each factor on peak cavity pressure at different injection speeds at 230C

149

Percent Influence (%) Injectio n speed 0.5/s 3/s 20/s Cp 0 0 0.20 p 4.27 0.89 0.13 P 0 0.03 0.03 h 24.73 6.32 0.66 v 70.09 92.59 98.69 Significant Factors , h, v , h, v v

Table 4.4 Relative influence of each factor on peak cavity pressure at different injection speeds at 250C

150

Percent Influence (%) Temperatur e 300C 320C Cp 0 0 p 5.81 6.39 P 0 0 h 27.14 28.44 v 61.72 58.98 Significant Factors v, h v, h

Table 4.5 Relative influence of each factor on peak cavity pressure at different melt temperatures for HDPE at 0.5/s

151

Percent Influence (%) Injectio n speed 0.5/s 3/s 20/s Cp 0 0.07 21.56 p 11.49 7.96 2.22 P 0 0 0.10 h 82.98 88.94 69.64 v 2.82 0 6.20 Significant Factors , h h Cp, , h, v

Table 4.6 Relative influence of each factor on maximum pressure drop at different injection speeds at 230C

152

Percent Influence (%) Injectio n speed 0.5/s 3/s 20/s Cp 0.06 1.14 21.56 p 8.99 5.35 1.74 P 0 0 0.32 h 86.78 93.27 67.54 v 2.55 0 7.40 Significant Factors , h Cp, , h Cp, h, v

Table 4.7 Relative influence of each factor on maximum pressure drop at different injection speeds at 250C

153

Percent Influence (%) Temperature 300C 320C Cp 0 0 p 8.11 7.48 P 0 0 h 88.23 87.94 v 0 0 Significant Factors h h

Table 4.8 Relative influence of each factor on maximum cavity pressure drop at different melt temperatures for HDPE at 0.5/s

154

Fig. 4.1. The long rectangular mold base with a disk-like insert.

155

Long mold A

Short mold

135 mm
A B

4 mm

Fig. 4.2. The rectangular mold bases with a disk-like insert.

156

View A-A

Fig. 4.3. The disk-like mold insert which contains microchannels.

157

Fig. 4.4. SEM picture of the a microchannel.

158

1.E+04
180C 200C 220C

Viscosity (Pa.s)

1.E+03

1.E+02 1.E-01 1.E+00 1.E+01 Frequency (1/s) 1.E+02

Fig. 4.5. Dynamic viscosity of polypropylene.

159

1.0E+04 210C

Complex viscosity (Pa.s)

220C 230C

1.0E+03

1.0E+02 0.10 1.00 10.00 100.00

Frequency (1/s)

Fig. 4.6. Dynamic viscosity of polypropylene

160

a. Top view

b. Side view Fig. 4.7. SEM of a micro-channel.

161

300 80C; HP=500 PSI

Channel Height (Micormeter)

250

80C; HP=0 PSI

200

150

100

50

0 0.001

0.01

0.1

10

Main Flow Velocity (m/s)

Fig. 4.8. Measured filling lengths in microchannels for PMMA in the long mold.

162

600
80C; HP=500 PSI 80C; HP=0 PSI 25C; HP=0 PSI

500

Channel Height (Micormeter)

400

300

200

100

0 0.001

0.01

0.1

10

Main Flow Velocity (m/s)

Fig. 4.9. Measured filling lengths in microchannels for PP in the long mold.

163

50 45 40
Filling length ( m)
80C; HP=500 PSI 80C; HP=0 PSI

35 30 25 20 15 10 5 0 0.01 0.1 1 Main flow velocity (m/s) 10

Fig. 4.10. Measured filling lengths in microchannels for PMMA in the long mold.

164

300
80C; HP=0 PSI

250

25C; HP=0 PSI

Filling length ( m)

200 150 100 50 0 0.001 0.01 0.1 1 Main flow velocity (m/s) 10

Fig. 4.11. Measured filling lengths in microchannels for PP in the long mold.

165

700 600 Short Mold, Channel A Short Mold, Channel B Long Mold, Channel B

Channel Height (Micormeter)

500 400 300 200 100 0 0.001

0.01

0.1

10

Main Flow Velocity (m/s)

Fig. 4.12. Measured filling lengths in microchannels for PP in the short mold.

166

Fig. 4.13. The cavity pressure profile in the long mold and the short mold.

167

1.2

Filling length / Depth

0.8

0.6

0.4

100 microns, Short B 100 microns, Short A

0.2

100 microns, Long A, B

0.01

0.1

1
Fo

10

100

1000

Fig. 4.14. The filling length vs. Fourier number.

168

80 70
Cavity pressure (MPa)

60 50 40 30 20 10
0 025.40 Start of packing

27.94 4

30.48 8

30.73 12

30.99 16

34.29 20

50.80 24

400 350 300 Filling length ( m) 250 200 150 100 50 0 25.40 27.94 30.48 30.73 30.99 34.29 50.80 Shot size (mm) Start of packing

201 mm/s

Fig. 4.15 The effect of packing stage on filling lengths. (Melt temperature 240C, mold temperature 25C, main flow velocity 0.2 m/s)

169

450 400 350

Main flow velocity 201 mm/s

Filling length (m)

300 250 200 150 100 50 0 0 500 1000 1500 1900

Hold pressure (psi)

Fig. 4.16 The effect of holding pressure on filling lengths. (Melt temperature 240C, mold temperature 25C, main flow velocity 0.2 m/s)

170

600 Expr. 500 400 300 200 100 0 1 10 100 1000 10000 h=500 h=2000 h=2500

Filling length (micron)

Main flow velocity (mm/s)

Fig. 4.17. Comparison of the filling lengths between the simulation and experiment with constant heat transfer coefficients. Main flow heat transfer coefficient=25000 W/m2.K.

171

600

Expr.
500

Filling length (micron)

h=500 h=2000 h=25000

400 300 200 100 0 1

10

100

1000

10000

Main flow velocity

Fig. 4.18. Comparison of the filling lengths between the simulation and experiment with constant heat transfer coefficients. Main flow heat transfer coefficient=2000 W/m2.K.

172

600 500 Filling length (micron) 400 300 200 100 0 1 10 100 1000 10000 Main flow velocity (mm/s) Expr. Tm =25C Sim u. Tm =25C Expr. Tm =80C Sim u. Tm =80C

Fig. 4.19. Comparison of the filling lengths between the simulation and experiment with variable heat transfer coefficient.

173

P transducer

P transducer

P transducer

76.2 mm 152.4 mm

Fig. 4.20. Schematic of the mold with thickness of 1 mm.

174

50.8 mm

12000 10000 8000 6000 4000 2000 0 0 50 100 150 200 o T ( C) 250 300 350
PS HDPE

Cp (J/kg. C)

Fig. 4.21. Heat capacity of HDPE and PS.

175

1.5 0 M Pa 1.4 50 M Pa 100 M Pa 1.3 150 M Pa

Specific volume (cm/g)

1.2

1.1

1.0

0.9 273 323 373 423 473 523

Temperature (K)

Fig. 4.22. Specific volume of HDPE.

176

1.10 0 M Pa 50 M Pa 100 M Pa 150 M Pa

Specific volume (cm/g)

1.05

1.00

0.95

0.90 273 323 373 423 473 523

Temperature (K)

Fig. 4.23. Specific volume of PS.

177

1.E+05 180C 200C 1.E+04 220C 180C Fit 200C Fit 220C Fit 1.E+03

Viscosity (Pa.s)

1.E+02

1.E+01 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04

Shear Rate/Frequency (1/s)

Fig. 4.24. Experimental and fit viscosity vs. shear rate/ frequency for PS.

178

1.E+05 160C 180C 200C 160C Fit 180C Fit 200C Fit

1.E+04

Viscosity (Pa.s)

1.E+03

1.E+02

1.E+01 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04

Shear Rate/Frequency (1/s)

Fig. 4.25. Experimental and fit viscosity vs. shear rate/ frequency for HDPE.

179

90 80 Cavity Pressure (MPa) 70 60 50 40 30 20 10 0 0 2 4 Time (s) 6 8

Effect of pressure not considered Pressure-dependent v

Fig. 4.26. Comparison of cavity pressure with/without the effect of pressure on specific volume.

180

80 70 Cavity Pressure (MPa) 60 50 40 30 20 10 0 0 1 2 3 4 5 6 7 Time (s)

Effect of pressure not considered Pressure-dependent viscosity

Fig. 4.27. Comparison of cavity pressure with/without the effect of pressure on viscosity.

181

80 70 Cavity Pressure (MPa) 60 50 40 30 20 10 0 0 2 4 Time (s) 6 8

h=25000 h=1500

Fig. 4.28. Comparison of cavity pressure with different heat transfer coefficients.

182

80 70 Cavity Pressure (MPa) 60 50 40 30 20 10 0 0 2 4 Time (s) 6 8

Constant Cp Temperature-dependent Cp

Fig. 4.29. Comparison of cavity pressure with constant Cp and temperaturedependent Cp.

183

80 70 Cavity Pressure (MPa) 60 50 40 30 20 10 0 0 2 4 Time (s) 6 8

No juncture loss Juncture loss included

Fig. 4.30. Comparison of cavity pressure with/without juncture loss.

184

80 70 60

Pressure after Gate, 250C Pressure at End, 250C Pressure after Gate (230C) Pressure at End (230C)

Pressure (MPa)

50 40 30 20 10 0 0 2 4 6 8 10 12 14 16 18

Time (s)

Fig. 4.31. Pressure profiles right after the gate and at the end of the cavity at the injection speed of 76.2 mm/s and the melt temperature of 230 and 250C.

185

80 70 60 50 40 30 20 10 0 0 2 4 6 Time (s) 8 10 12 14

12.7 mm/s 76.2 mm/s 508 mm/s

Fig. 4.32. Pressure profiles right after the gate at the melt temperature of 230C with different injection speeds.

Pressure (MPa)

186

40 35 30 25 20 15 10 5 0 0 2 4 6 Time (s) 8 10 12 14

12.7 mm/s 76.2 mm/s 508 mm/s

Fig. 4.33. Pressure profiles at the end of the cavity at the melt temperature of 230C with different injection speeds.

Pressure (MPa)

187

90 80 70 Experimental pressure Effect of pressure on v not considered P-independent viscosity Pressure-dependent viscosity h=1500

Cavity Pressure (MPa)

60 50 40 30 20 10 0 0 2 4 Time (s) 6 8

Fig. 4.34. Comparison of experimental and predicted pressure drop at the injection speed of 12.7 mm/s.

188

90
Experimental pressure

80
Cavity Pressure (MPa)

Effect of P on v not considered P-independent viscosity P-dependent viscosity h=1500

70 60 50 40 30 20 10 0 0 2 4 6 8 Time (s)

10

12

14

Fig. 4.35. Comparison of experimental and predicted pressure drop at the injection speed of 508 mm/s.

189

CHAPTER 5

CHARACTERIZATION OF VIRGIN/POST-CONSUMER BLENDED HIGH IMPACT POLYSTYRENE RESINS FOR INJECTION MOLDING

5.1 INTRODUCTION
Plastics have become a common materials choice in many new products and millions of kilograms of plastics are used annually [Society of the Plastics Industry, 2001]. The attention paid to polymer recycling has increased in the past decade because more efficient re-use of materials will reduce the quantities of plastics sent to landfills as well as reduce raw material extraction. Furthermore, the advent of take-back

legislation accelerates waste prevention practices [Gamalski, 1996; Meffert and Kirchgeorg, 1997; Hubschman, et al., 1995]. However, only a small amount of plastics is reused as introduced in Section 2.1. Reducing virgin resin consumption can be achieved by reduction of material requirement or resin recycling. One strategy is to use thinner wall molding to reduce the quantity of material required. However, thin-wall molding requires high injection speed, high injection pressure with polymers that could

190

withstand high shear rates and possible molecular degradation. Another strategy is to recycle resin. In this study, we focus on resin cycling. However, how to characterize the post-consumer resin (PCR) and how to increase the percentage of the post-consumer resin are two of the problems in recycling plastic. Currently, only less than one percent of HIPS is recovered from the total 19% market share [Dillon and Aqua, 2000]. Two big challenges to reuse post-consumer resin are material contamination and degradation. Post-consumer polymers may be

contaminated from other materials [Langerak, 1997]; post-consumer products may contain polymer blends as well as additives such as reinforcements, paint, or flame retardants [Dillon, 1999]. Another challenge is the material degradation because

returned polymers have been exposed to various thermal and mechanical conditions. Thus, molders are reluctant to use recycled plastics because extensive experimental testing is required to identify plausible use and determine molding parameters. Recyclers currently select between options such as incineration or downcycling. The major problem to reuse PCR is that polymer databases do not contain information about PCR. Beside the material selection assistance, polymer databases are used in mold filling simulation to design, reduce experimental time to decide processing parameters, and predict possible problems. If molders must use trial and error to

determine PCR molding parameters, then a higher setup time is required for PCR than for a virgin resin that is included in the database. Manufacturers usually use virgin resin

191

databases to decide processing parameters to reduce time because molders cannot easily decide them without the material characterization of PCR. Therefore, our initial investigation began with characterization of the postconsumer resin. The viscosity is one of the basic properties for the reuse of the postconsumer resin. The melt viscosity of the post-consumer resin was measured and the virgin resins were identified with the same melt viscosity as the PCR. Next, the melt viscosities of post-consumer and virgin resin blends were measured. Then the

mechanical properties of blends were measured and the effect of different virgin resins and weight percentage of virgin resins were discussed. This investigation helped us evaluate the viability of reusing the PCR in new injection molded products. Our goal is to characterize the relationship between the ratio of recycled content to virgin content and the mechanical properties. The mechanical properties, including tensile properties, flexural properties, and impact properties, of the blends with different percentages of reuse resin were analyzed through experiments. Furthermore, we investigated the

molecular weight and morphology of molded parts to help explain and predict the properties of recycled blends for injection molding. Understanding the relationship between rheological and design characteristics will provide both suppliers (recyclers) and customers (molders) with valuable insights regarding viable uses for post-consumer resins. Meanwhile, we introduce a sequence of steps to obtain PCR input for mold filling simulation. The purpose is to reduce the amount of experimental time to

determine molding parameters. The method is tested by molding ASTM specimens and a thin-wall application in film canisters. 192

5.2 EXPERIMENTAL 5.2.1 Characterization of Material

It is important to identify the post-consumer polymer properties. In general, it is nearly impossible to identify the original resin manufacturer for post-consumer polymers in electronics equipment. In our case, we only knew the polymer was labeled HIPS (high impact polystyrene); we did not know the original resin manufacturer or product code. Therefore, we tested the rheological properties of the PCR in order to identify the most suitable "virgin resin for the blends. The PCR material we used consisted of ground pieces of printer and monitor housings. The size of the fragments was greater than 100 mm. The incoming fragments were inspected manually for metal contamination and then were shredded again to reduce their size before mixing with virgin resins. The maximum dimension of the shredded fragments was 1-10 mm, which was close to the size of the virgin resins. The rheometer used was a Rheometrics Mechanical Spectrometer (RMS 800). The rheological properties of the blends which consisted of different percentages of postconsumer HIPS and virgin resins were also studied at three temperatures: 180C, 200C, and 220C. Molded discs were used for the measurement of viscosity for the blends.

5.2.2 Measurement of Molecular Weight

The molecular weight was measured by GPC (Gel Permeation Chromotography). The samples used were molded blends with 0%, 50%, and 100% Huntsman PS 702, 193

molded blends with 50% and 100% Nova PS 3350, and never molded, 100% virgin Nova PS 3350. The solution was prepared by dissolving blends in THF (Tetrahydrofuran). Each sample was analyzed twice with a running time of 45 minutes and an injection volume of 200 l. We report the average of the two runs in Table 5.1.

5.2.3 Microscopy and Spectroscopy

For the morphological measurement, the aim is to observe the dispersion of rubber phase in polystyrene and the size of rubber domains because the rubber particles can affect the mechanical properties. A Philip XL-30 FEG environmental scanning electron microscope (ESEM) was used. The sample was stained by 1% OsO4 aqueous solution for 15 days and carbon-coated for morphological measurements. The fracture surfaces were observed with 15 KV power. The magnification in this study varied from 200x to 10,000x. The purpose of the Raman spectroscopy tests is to determine if there is a detectable difference in the absorption spectra for the PCR and the virgin high impact polystyrene. The virgin resin Huntsman PS 702 was used in the experiment. The infrared vibrational spectra were obtained using a Bruker Equinox 55 with IR Scope 1. The instrument was operated in reflectance mode using the 15x microscope objective and 4 cm-1 resolution. OPUS software, version 2.2, was used for instrument control and data handling. The Raman vibrational spectra were obtained using a Chromex Raman 2000 spectrometer upon illumination by a 785 nm diode laser DSL and imaged on a 194

Photometrics 1024 X 256 pixel red enhanced CCD detector. The spectra were taken at a 180 collection angle with a depth of focus of several mm. The laser power was typically 50 mW with a spot size of 80 m.

5.2.4 Processing Parameters for ASTM Specimens

To determine initial processing parameters, a mold filling simulation was run on C-MOLD 97.7. C-MOLD is a set of integrated computer aided engineering (CAE) simulations for plastics molding processes, including injection mold filling, post-filling, cooling, part shrinkage and warpage. C-MOLD provides recommendations for

processing parameters such as fill time, inlet and melt temperature. CAE provides an easy-to-use data visualizer for viewing mesh information and analysis results. C-MOLD 97.7 was used to simulate filling our mold with one of the virgin resins, Huntsman PS 702, which had the same viscosity versus shear rate as our PCR. The ASTM mold consisted of six cavities, including two tensile bars, two sheets and two discs. From the results of the mold simulation and several experimental trials, the

operating parameters, such as inlet melt temperature, melt temperature, and cooling time, were selected to injection mold ASTM specimens. The mold filling simulation required input for the resin properties. They can be obtained from commercial resin databases or resin suppliers. However, resin databases only contain virgin resin. So, we identified a virgin resin with similar viscosity to use as input. In our approach, recyclers do not need to know the original resin manufacturers or 195

product codes. We assume that the PCR has been processed and sorted by manual disassembly [Meacham, et al., 1999] or new bulk recycling methods [Arola and Biddle, 2000] so that it is not contaminated by other materials.

5.2.5 Physical Properties of ASTM Specimens

Six different weight percentages of blends were prepared, as shown in Table 5.2. Two selected virgin resins, Huntsman PS 702 and Nova PS 3350, were used. These virgin resins were selected because they had the close viscosity versus shear rate curve as the PCR. The blends were mixed for one minute in a Little Ford Lodige Precision Mixer. The ASTM specimens were prepared with a 50 ton Sumitomo injection molding machine. The virgin material and post-consumer resin were mixed completely and then were dried at 160F for 2 hours prior to injection molding. According to the results of the mold design, the barrel temperature was set from 380F to 440F from the rear zone to front zone. The mold temperature was kept at 77F. For blends with Huntsman PS 702, the physical properties tested include: tensile strength and modulus (ASTM D 638) at 23.3C and humidity 21%, flexural strength and modulus (ASTM D 790) at 18.4C and humidity 12%, and notched Izod impact strength (ASTM D 256) at 18.4C and humidity 21%. For blends with Nova PS 3350, all tests were performed at 27.3C and humidity 30%. Roughness and waviness measure the small-scale surface irregularities. Roughness represents the range of groove heights of the surfaces while waviness is the regression line (mean line) of the roughness profile. Two surface parameters, roughness 196

average (Ra) and waviness average (Wa) were measured. According to ISO, ANSI, and DIN standards, Ra is the arithmetic average deviation of the roughness profile from the roughness centerline, while Wa is the arithmetic average deviation of the waviness profile from the waviness centerline [Sander, 1991]. The test was performed using the

Perthometer with a Gaussian filter type. The tests were conducted with a straight line (entire trace) tilt correction and an evaluation range of 4.00 mm. One data point was collected at each of five locations on each of three ASTM impact discs per blend for a total of fifteen data points per blend.

5.2.6 Application
The ASTM test specimen mold is specially designed to minimize material damage during molding. So, we tested the recycled material under high shear stress condition to assess the ability of the material to withstand more realistic industrial use. The film canister mold, loaned by Eastman Kodak Company, was used to test the postconsumer/virgin polymer blends and is a thin-wall application compared to the original printer and monitor housings. The film canister is shown in Fig. 5.1. The canister base has variations in thickness as well as the recycling logo. In Table 5.3, the mold design characteristics are listed.

C-MOLD 97.7 was used to simulate filling the film canister mold with the virgin resin, Huntsman 702. From the results of simulation, combined with experimental trials, the operating parameters, such as inlet melt temperature, melt temperature, cooling time, etc., were selected for making film canister specimens, as reported in section 5.3.6. At 197

the same operating parameters, six types of blends of different percentage of PCR were used to mold canisters. In order to compare the two virgin resins, we processed blends under the same conditions. However, Nova PS 3350 has a lower melt index of 0.27 g/min compared to the Huntsman PS 702 melt index of 0.75 g/min [IDES, 1999]. Thus, we used a higher injection velocity for virgin Nova PS 3350 after several experimental runs. One of the quality indicators tested for the canisters was the tensile strength. The specimens consisted of strips of uniform width and thickness. According to ASTM standards, we chose 100 mm as the width with a thickness less than 0.8 mm. Since the thickness of the canister wall was less than 1 mm, the ASTM D-882-97 was adjusted slightly by shortening the length of the specimens from 101.6 to 98 mm and 46 mm. To ensure uniform width, calipers with 0.25 mm capability were used to check the specimen width. The utmost care was exercised in cutting specimens to prevent nicks and tears that may cause premature failure. To eliminate the anisotropic effect of the material, two sets of test specimens were prepared having their long axes parallel with and normal to the flow direction. The flow direction of the material in injection molding was from the bottom to top.

5.3 RESULTS AND DISCUSSION 5.3.1 Characterization of Material

The rheological properties of the ground post-consumer HIPS were studied at three temperatures: 180C, 200C, and 220C. Fig. 5.2 is the viscosity versus frequency

198

curve of the post-consumer material at 220C. We identified two virgin resin candidates, Huntsman PS 702 and Nova PS 3350, in the C-MOLD resin databases by comparing the viscosity curves. The viscosity of blends of different percentage of the recycled resins was also investigated at three temperatures: 180C, 200C and 220C. Figs. 5.3 and 5.4 show the viscosity of blends with Huntsman PS 702 or Nova PS 3350 versus frequency at approximately 200C, respectively. It is found that all blends are shear thinning. It is also shown that the viscosity increases with the increase of the percentage of the PCR.

5.3.2 Molecular Weight

The molecular weights are listed in Table 5.1. It is shown that molecular weight and polydispersity of blends with Huntsman PS 702 increase with the increase of the percentage of Huntsman PS 702. However, for blends with Nova PS 3350, the molecular weight decreases with the increase of the percentage of Nova PS 3350, though the polydisperisity increases as the percentage of Nova PS 3350 increases. We can see that all blends, including recycled resin, have similar molecular weight and polydispersity, which would lead us to predict similar mechanical properties.

5.3.3 Microscopy and Spectroscopy

Fig. 5.5 shows the environmental scanning electron microscope (ESEM) images of different blends. It is found that for virgin resins, 100% Huntsman PS 702 and Nova PS 3350, the outer surfaces are dotted with a broad range of rubber domain with many 199

large rubber particles. The diameter is about 2 m. However, for 50% Huntsman PS 702 and 50% Nova PS 3350, we only observed relatively smaller rubber particles. The diameter is about 1 m. The surface structures for the 50% blends are less regular compared to those of virgin resins. For the PCR, we did not observe well defined rubber domains, and the surface was seemingly covered with poorly defined dispersed rubberphase and some very small particles which may be contaminants. Figs. 5.6 and 5.7 show the Raman Spectroscopy and Infrared vibrational spectra of recycled resin and virgin resin Huntsman PS 702, respectively. It is shown that recycled resin and virgin resin consist of almost the same components. Combined with the results of the molecular weight measurements, we predict that it is possible to mix the recycled resin and virgin resin for potential synergistic improvement of their properties.

5.3.4 Processing Parameters for ASTM Specimens

At first, the geometry was evaluated and then the mesh for the C-COLD simulation was created. Processing parameters for the ASTM specimens of Huntsman PS 702 from the C-MOLD simulation are given in Table 5.4. To compare the mechanical properties of the blends of Huntsman PS 702 to the properties of blends of Nova PS 3350, the same injection molding parameters were used to prepare the specimens of blends of Nova PS 3350.

5.3.5 Physical Properties of ASTM Specimens

For the six blends of Huntsman PS 702, the minimum, maximum, and average for the Ra and the Wa are shown in Figs. 5.8 and 5.9 respectively. As shown in the figures, 200

the roughness average was fairly stable for the various blends but the waviness average was best for the 0% virgin material. Due to our sample size of fifteen data points per blend, further tests are being conducted with a larger sample size. For the blends of Huntsman PS 702, the results of the physical properties tested are shown in Figs. 5.10-5.12. For each physical property, six samples were tested. The data shown in Figs. 5.10-5.12 are the averages of each sample. Fig. 5.10 shows the tensile modulus and tensile strength of the blends for two different virgin resins versus weight percentage of virgin resin. It is found that generally, both the tensile strength and tensile modulus decrease slightly with the increase of the weight percentage of virgin resin for the blends with virgin resin Huntsman PS 702, while the tensile strength and tensile modulus increase slightly with the increase of the weight percentage of virgin resin for the blends with virgin resin Nova PS 3350. The standard deviation of 12 samples at each percentage was calculated for each physical property. The average of the standard deviations for the six blends of the tensile strength and the tensile modulus are 0.63 and 67 respectively. Fig. 5.11 illustrates the results of flexural modulus and flexural strength. It is shown that flexural strength, like the tensile strength for the blends of Huntsman PS 702, decreases slightly. For the blends of Nova PS 3350, flexural strength has the same trend as tensile strength and increases slightly. However, flexural modulus has no specific changing trend for the blends of both Huntsman PS 702 and Nova PS 3350. The average of the standard deviations over the six blends of the flexural strength and the flexural modulus are 0.60 and 46 respectively.

201

As shown in Fig. 5.12, the impact strength of the blends of Huntsman PS 702 increases with the increase of weight percentage of recycled HIPS when the percentage is small. At 75% and greater recycled HIPS, the strength reaches a stable value. For impact strength of the blends of Nova PS 3350, it decreases with the increase of weight percentage of virgin resin when the percentage is small. At 75% and greater virgin resin, the strength reaches a stable value. The average of the standard deviation over the six blends of the impact strength is 2.6. Though Raman Spectroscopy and Infrared vibrational spectra show that recycled resin and virgin resin consist of almost the same components, and the blends have similar molecular weight and polydispersity, ESEM shows that the different blends have very different microstructure and different rubber domain sizes. Thus, it is not surprising that the different blends have different mechanical properties because the mechanical properties of HIPS can be affected by the amount of rubber added, the type of rubber, rubber size distribution, phase volume, the degree of crosslinking, or the level of adhesion [Hobbs, 1986; Cook, et al., 1993]. The reason for the higher tensile modulus, tensile strength, flexural strength, and impact strength of PCR compared to Huntsman PS 702 probably results from the higher tensile modulus and tensile strength of the original material or the addition of reinforcements in pure resin when the printers and monitors were made. Also, the experiments demonstrated that the mechanical properties of

recycled HIPS were slightly lower than those of Nova PS 3350. It is interesting to note that the mechanical properties of blends with Huntsman PS 702 and recycled resin are slightly better than the properties of the selected virgin material Huntsman PS 702. Our 202

experiments demonstrate that it is possible to reuse the post-consumer resin. Relative to the selected virgin materials with the same viscosities as the post-consumer resin, reuse of the post-consumer resin is an attractive option. We compared our 100% PCR tensile and flexural properties with those published in a study comparing disassembled versus shredded HIPS from post-consumer television sets [Langerak, 1997]. It is found that the tensile modulus of our blends is lower than that of the disassembled or shredded HIPS in the published study [Langerak, 1997]; however the tensile strength at yield of our blends is larger. It is also shown that the flexural modulus of our blend is lower than that of disassembled or shredded HIPS in the other study [Langerak, 1997], but the flexural strength is almost the same. The differences in mechanical properties of the PCR in the two studies may result from the different brands of the original materials.

5.3.6 Application
Before making the real film cans, the injection molding process was simulated by C-MOLD 97.7. The mesh is shown in Fig. 5.13. The simulation results are listed in Table 5.5. Film canisters were made using post-consumer Huntsman HI/PS 702 virgin resin blends. To obtain initial machine settings, we used the simulation results from C-MOLD 97.7 and IDES's handbook of injection molding specs [IDEAS, 1999]. difference between these resources is the processing temperature. The main

The handbook

recommended a lower temperature, 221C, while C-MOLD recommended a higher 203

temperature, 243C. After experimental trials, we selected the machine settings as shown in Table 5.5. The tensile tests for the film canisters were performed on the Instron machine. The results are listed in Table 5.6. It is shown that the tensile strength of the blends with Huntsman PS 702 increases with the increase of the weight percentage of recycled resin. The reason is that Huntsman PS 702 has lower tensile strength than the PCR and thus the PCR increases the tensile strength of the blend. If the PCR is cheaper and has a higher tensile strength than a virgin resin with similar rheology, then the PCR can be selected to increase the mechanical property or properties.

5.4 CONCLUSIONS
To determine the initial processing conditions for injection molding virgin/postconsumer resin blends, a precharacterized resin must be designated for a C-MOLD simulation. To select a precharacterized resin for the C-MOLD simulation, virgin resin viscosity curves were matched with the PCR viscosity curve. Then the recommended CMOLD simulation processing parameters were further refined for the blends for the ASTM test standard specimens by running several experimental runs. In our proposed approach, we can characterize and represent the PCR in a mold filling simulation by the virgin resin in the database. Experimental testing to determine injection molding

parameters for various blends is greatly reduced by this approach. All blends have similar molecular weight and polydispersity. Furthermore, the recycled resin and virgin resin consist of almost the same components, as shown in their 204

Raman and infrared spectra. For the ASTM specimens molded with either set of blends, the mechanical properties are similar. The tensile modulus, tensile strength, and flexural strength increase slightly with the increase of the weight percentage of PCR for the blends of Huntsman PS 702. The impact strength increases with the increase of weight percentage of PCR when the percentage is small and finally the strength reaches a stable value. It is found that the physical properties of blends having recycled resin are better than the properties of virgin resin Huntsman PS 702. On the other hand, the mechanical properties of PCR with Nova PS 3350 are slightly lower when compared to the pure virgin Nova PS 3350 resin.

205

Materials 100% Huntsman 702 50% Huntsman 702 0% Huntsman 702 100% Nova 3350 50% Nova 3350 Virgin Nova 3350

Mn 58198 56730 55262 54129 55724 57577

Mw 180875 171486 162099 196963 181306 183095

Polydispersity 3.06 3.03 2.93 3.64 3.26 3.18

Table 5.1 Molecular weight (Number average Mn and weight average Mw) and polydispersity

206

No 1 2 3 4 5 6

Weight percentage Of virgin resin (%) 100 85 75 50 25 0

Weight percentage of Recycled material (%) 0 15 25 50 75 100

Table 5.2 Weight percentage blends

207

Description Maximum dimension Maximum flow length Volume Thickness Gate geometry

Film canister 49.40 mm 65 mm 560 mm 3 0.76 mm Rectangular, 0.130 in wide 0.075 in deep

Projected area

730 mm 2

Table 5.3 Mold design characteristics

208

Max machine clamp force Max machine injection volume Max machine injection pressure Max machine injection rate Fill time Post-fill time Mold-open time Ambient temperature Min/max melt temperature Transition temperature Inlet melt temperature Average coolant temperature

4.90E+007 N 0.02 m3 1.8E+008 Pa 0.006667 m3/s 2.00 s 12.08 s 2s 298 K 449.15/533.15 K 365.15 K 522.09 K 298 K

Table 5.4 Processing parameters from C-MOLD

209

Resin Coolant The maximum flow length Thickness Projected area Volume Coolant channel diameter Clamp force Mold open time Mold temperature Min. Processing temperature Max. Processing temperature Max. machine inj. Press. Melt temperature Fill time

Huntsman HI/PS 702 Pure water 65 mm 0.76 mm 7.3 cm 2 0.56 cm 3 7 mm 50 ton(m) 2s 34.5C 176C 260C 180 MPa 242.9C 0.49 s

Table 5.5 CMOLD parameters for film canister

210

wt% of virgin resin Tensile strength (MPa) of Hunstman PS 702 0% 25% 50% 75% 85% 100% 17.54 15.69 16.02 15.42 14.76 14.78

Tensile strength (MPa) of NOVA PS 3350 17.54 16.31 16.36 17.62 17.91 19.44

Table 5.6 Tensile strength of film canisters

211

Fig. 5.1. Film canister.

212

1.E+04

Viscosity (Pa.s)

1.E+03

PCR Huntsman PS 702 Nova PS 3350

1.E+02 1.E-02

1.E-01

1.E+00

1.E+01

1.E+02

Frequency (1/s)

Fig. 5.2. Comparison of the viscosity curves for post-consumer HIPS and virgin HIPS at 220C.

213

1.E+05

Viscosity (Pa.s)

1.E+04

1.E+03

0% at 198.0C 25% at 197.5C 50% at 197.6C 75% at 197.5C 85% at 197.7C 100% at 197.4C 1.E-01 1.E+00 1.E+01 1.E+02

1.E+02 1.E-02

Frequency (1/s)

Fig. 5.3. Viscosity of Huntsman PS 702 blends with different percentages of postconsumer resin at about 200C.

214

1.E+05

Viscosity (Pa.s)

1.E+04

1.E+03

0% at 198.0 C 25% at 197.3 C 50% at 197.5 C 75% at 197.7 C 85% at 197.1 C 100% at 197.7 C 1.E-01 1.E+00 1.E+01 1.E+02

1.E+02 1.E-02

Frequency (1/s)

Fig. 5.4. Viscosity of Nova PS 3350 blends with different percentages of post-consumer resin at about 200C.

215

(a) 100% Huntsman PS 702

(b) 100% Nova PS 702

(c) 50% Huntsman PS 702

(d) 50% Nova PS 702

(e) PCR

Fig. 5.5. The images of different blends from ESEM (The length of the scales in the figures are 2 m).

216

Huntsman PS

Fig. 5.6. Raman spectroscopy of injection-molded post-consumer and Huntsman PS 702.

217

PC

Huntsman PS 702

Fig. 5.7. Infrared vibrational spectra of injection-molded post-consumer and Huntsman PS 702.

218

0.45 0.40

R(a) [micrometers]

0.35 0.30 0.25 0.20 0.15 0.00 0.25 0.50 0.75 0.85 1.00

Percentage of the virgin material

Fig. 5.8. Average Ra for six blends of Huntsman PS 702.

219

0.40 0.35

W(a) [micrometers]

0.30 0.25 0.20 0.15 0.10 0.05 0.00 0% 25% 50% 75% 85% 100%

Percentage of the virgin material

Fig. 5.9. Average Wa for six blends of Huntsman PS 702.

220

30 25
Tensile strength (MPa)

2500

2000
Tensile mudulus(MPa)

20 1500 15 1000 10
Tensile strength of Huntsman PS 702

5 0 0%

Tensile strength of Nova PS 3350 Tensile modulus of Huntsman PS 702 Tensile modulus of Nova PS 3350

500

20%

40%

60%

80%

0 100%

Weight percentage of virgin resin

Fig. 5.10. Tensile strength and tensile modulus vs. weight percentage of virgin resin.

221

50 45 40 35 30 25 20 15 10 5 0 0%

2500 2000 1500

Flexural strength of Huntsman PS 702 Flexural strength of Nova PS 3350 Flexural modulus of Huntsman PS 702 Flexural modulus of Nova PS 3350

1000 500 0 100%

20%

40%

60%

80%

Weight percentage of virgin resin

Fig. 5.11. Flexural strength and flexural modulus vs. weight percentage of virgin resin.

222

Flexural modulus (MPa)

Flexural strength (MPa)

120 100
Impact strength (MPa)

80 60 40 20 0 0% 20% 40% 60% 80% 100%

Weight percentage of virgin resin

Huntsman PS 702 Nova PS 3350

Fig. 5.12. Impact strength and tensile modulus vs. weight percentage of virgin resin.

223

Fig. 5.13. Meshing model of the film canister.

224

CHAPTER 6

CONCLUSIONS AND RECOMMENDATIONS

6.1 FLOW MARKS

For alternate dull and glossy flow marks, the effect of polymer rheology, injection speed, mold geometry, melt temperature, mold temperature, holding pressure, injection pressure, and mold surface coatings on the appearance of the flow marks was studied. It was found that the most important factor affecting the flow marks was injection speed. The flow marks occurred above a critical wall shear stress, but

disappeared at high injection speeds. Mold geometry had an effect on the flow marks, but mold temperature and melt temperature did not have much effect on the flow marks. No difference was observed between the crystallinity of dull regions and shiny regions. However, it was found from Scanning Electron Microscopy that the melt in dull regions was only slightly oriented while the melt in shiny regions was highly oriented. It was also found that coating these surfaces did not prevent the occurrence of the flow marks, although it could alleviate them. It was also found that the polymer with the highest dynamic viscosity, elastic modulus, first normal stress difference, transient extensional viscosity, and longest relaxation time exhibited flow marks over a wide range of processing conditions. Slip was not the cause of the generation of the alternate flow marks. The generation of the flow marks was explained by an entry viscoelastic flow instability. 225

Synchronous dull and glossy flow marks were also studied.

The effect of

operating parameters, mold geometry, and mold surface coatings on the flow marks was investigated. The flow marks occurred above a certain flow front velocity. It was also found in the experiment that the flow marks were dimmer as the mold temperature was increased. No difference was observed between the crystallinity of dull and shiny

regions. However, polymer was highly oriented in shiny region while it was slightly oriented in dull regions. It was also found that mold surface coatings did not eliminate the flow marks. Extrusion experiments showed that helical gross melt fracture occurred for both HDPEs. Finally, it was proposed that an entry viscoelastic instability was the reason for the generation of the synchronous flow marks. For the future work, we will prove the mechanism of the entry flow instability. More evidence is favored for the proposed mechanism. For example, the possible pressure fluctuation relating to melt fracture (flow marks) will be monitored. Using the glass window mold in our lab, the flow before the gate and the flow front will be visualized and recorded by high-speed camcorder. Then the flow will be analyzed. Moreover, the possibility of slip will be analyzed. The extensional viscosity will be measured to describe the fountain flow more accurately, and its effect on the formation of vortices will be analyzed. Furthermore, fundamental mechanism for the formation of the flow marks will be studied. The detailed morphology, crystallinity and structure of crystalloids, and the effective thickness of the flow marks will be investigated.

6.2

EXPERIMENTS

WITH

MICRO-FEATURES

AND

SIMULATION

ACCURACY IMPROVEMENT
Thin-wall injection molding with micro-features was studied experimentally and numerically. The filling lengths in microchannels are affected by injection speed, mold temperature and channel location. It was found that high injection speed or high mold temperature results in longer filling length. 226 Moreover, the filling lengths in

microchannels increase with the decrease in the filling time flowing from the microchannels to the main flow end. Furthermore, the filling lengths in microchannels are simulated by a hybrid simulation code with a combination of the momentum equation and the Hele-Shaw model, and compared with experimental results. The code has fewer elements and requires less computation time. The simulation shows that the filling lengths in microchannels are sensitive to the heat transfer coefficients in the main flow cavity and in the microchannel and extra attention is needed to determine proper heat transfer coefficient. Using the variable heat transfer coefficient, the filling length in the long mold is predicted accurately. Our future work will study the thin-wall injection molding of smaller microchannels with the width of 50 m and the depth of 250 m. The morphology of the microchannels, demolding problem, filling, freezing pattern, repeatability, durability, and the deformation of the wall of the microchannels will be studied. Moreover, the filling lengths in microchannels with different main flow thicknesses will be compared to study which main flow thickness is beneficial to long filling lengths. The argument is that in the thick mold the melt temperature is high but the pressure drop is low; in the thin mold the temperature is low but the pressure drop is high. So it is difficult to decide which mold thickness is favorable to long filling lengths. Furthermore, the filling length will be measured and it will be compared with simulation results. The effect of heat transfer coefficients both in the main flow and in the microchannels will be paid full attention. For the cavity pressure, the simulation showed that the effect of pressure on the specific volume is the most important factor to predict the peak cavity pressure. The effect of pressure on the viscosity and the heat transfer coefficient are also significant. The heat capacity and the juncture loss are relatively less important compared to other factors considered here. Therefore, it is very important to use proper material property

227

models when running simulation of thin-wall injection molding. It was also shown that the significant factors are somewhat different to predict maximum cavity pressure drop. The effect of the pressure-dependent viscosity, the heat capacity, the heat transfer coefficient, the juncture pressure loss and the pvT-data on the cavity pressure and pressure drop were studied. Another important thermal property, thermal conductivity, would be included for future work. Furthermore, future work could study the effect of these properties on the filling length in main flow and even in microchannels. When the injection speed was high, the discrepancy between the simulation results and experimental data was large and no good agreement could be achieved no matter what property models were used. So, the reason for the discrepancy might not be included within the factors we considered. The possible reason may be the difference between the set operating values and the actual conditions the machine reached. For example, the actual injection speed is intrinsically slower than the speed one sets, especially at high injection speeds, as the machine needs response time to reach the desired constant injection speed. The actual temperature in the barrel may be different from the set temperature. The effect of these differences should be checked. Material property measurement and models will affect the simulation results and proper conclusions. The pressure dependent viscosity was measured under relatively low pressure and then extrapolated to high pressure. Future work should measure the viscosity under very high pressure to get a more accurate pressure dependent viscosity model. The heat capacity was measured at a low heating rate of 3.33C/s. It is very useful to get the true value because the cooling rate is very fast in thin-wall injection processes. The heat transfer coefficient has a large effect on the cavity pressure and the default value 25,000 W/m2K must be re-evaluated to obtain good simulation results because other researchers work and the current work showed that the default is too large.

228

Finally, the software itself may affect the final pressure prediction due to its simplification, such as the assumption of Hele-Shaw flow. Hele-Shaw flow neglects flow in the gapwise direction and gives the average information in the gapwise direction. It cannot accurately predict the fluid behavior at the flow front and the flow near or at solid walls, the phenomenon occurring at the merging of two or more streams (weld lines), and the kinematics in ribs, gates, or sudden contractions/enlargement. Moldflow 1 is a 2 D software and uses mid-plane mesh. So, developing the code with less 2 assumptions or 3-D mesh based on our groups previous work may provide more accurate pressure prediction.

6.3 REUSE OF HIPS

This part focuses on the mechanical and rheological properties of virgin and recycled high impact polystyrene materials. The study shows that all blends have similar molecular weight and polydispersity. Furthermore, the recycled resin and virgin resin consist of almost the same components, as shown in their Raman and infrared spectra. For the ASTM specimens molded with either set of blends, the mechanical properties are similar. The tensile modulus, tensile strength, and flexural strength increase slightly with the increase of the weight percentage of PCR for the blends of Huntsman PS 702. The impact strength increases with the increase of weight percentage of PCR when the percentage is small and finally the strength reaches a stable value. It is found that the physical properties of blends having recycled resin are better than the properties of virgin resin Huntsman PS 702. On the other hand, the mechanical properties of PCR with Nova PS 3350 are slightly lower when compared to the pure virgin Nova PS 3350 resin. Our experiments demonstrate that the PCR may have good material properties and may even be used in a more challenging application.

229

Moreover, the study introduces a new approach to determine initial processing parameters for injection molding of post-consumer resin. To determine the initial

processing conditions for injection molding virgin/post-consumer resin blends, we can characterize and represent the PCR in a mold filling simulation by the virgin resin in the database. This approach greatly reduces experimental testing to determine injection molding parameters for various blends. This approach for plastics recycling is novel because we started with an initial rheological investigation of PCR characteristics rather than tracking the original virgin resin. We also tested our new approach by molding the PCR in a thinner wall design application. There are several areas of this research requiring more study in future. In our study, two different virgin resins were identified by PCR characterization. Both

candidates had similar viscosity versus shear rate curves, but different melt flow indices. Because plastics have complex properties, further study is needed to identify the properties of the unknown PCR, and then find virgin resins that match additional characteristics, such as mechanical properties. Then, the mechanical properties of a specific design with different percentage of PCR will be predicted. It will further improve the decision tool to decide the threshold of recycling. Our study used HIPS from computer and monitor housing. More cases are needed to get more general conclusions. To investigate the sensitivity of our approach to grade mixtures is an interesting extension of this work. Because shredding different plastic parts may generate a reground mixture of HIPS PCR grades, it will be useful to determine whether the viscosity versus shear rates of the resin grade mixture could be used to identify a proxy virgin resin.

230

REFERENCES

Ainoya, K. and O., Amono. SPE ANTEC Tech. Papers, 2001, 47, 726-35. Allenby, B. R. and R. A., Laudise. AT&T Technical Journal, 1995, 74(6), 8-16. Amano, O. and K., Ainoya. SPE ANTEC Tech. Papers, 2000, 46, 504-11. Angstadt, D. C. and J. P., Coulter. SPE ANTEC Tech. Papers, 2001, 47, 537-42. Arola, D. F. and M. B., Biddle. Proceedings of the IEEE International Symposium on Electronics and the Environment. Institute of Electrical and Electronic Engineers (IEEE): Piscataway, NJ. 2000, (May 8-10): 75-80. Arola, L. and J. B., Legarth. IEEE International Symposium on Electronics and the Environment. Institute of Electrical and Electronic Engineers (IEEE): Danvers, Massachusetts, May 1999, 187-191. Belcher, D. K. and S. M., Hoenig. SPE ANTEC Tech. Papers, 1991, 37, 305-8. Bernardo, C. A., Cunha, A. M. and M. J., Oliveira. Polymer Engineering and Science, 1996, 36(4): 511-519. Bird, R. B., Armstrong, R. C. and O., Hasager. Dynamics of Polymeric Liquids, John Wiley & Sons, 1987. Boger, D. V., Hur, D. U. and R. J., Binnington. J. Non-Newtonian Fluid Mech, 1986, 20, 31-49. Boger, D. V. and K., Walters. Rheological Phenomena in Focus, Elsevier Science Publishers, B. V. The Netherlands, 1993. Bozzelli, J. W. and J., Cardinal. SPE ATTEC Tech. Papers, 1996, 42, 584-7. Bozzelli, J. W., Cardinal, J. and B., Fierens. SPE ANTEC Tech. Papers, 1997, 43, 425-8.

231

Bulters, M. A. and A., Schepens. PPS Sixteen Annual Meeting, June 18-23, Shanghai, China. 2000, 144-5. Cha, S. and F. S., Lai. SPE ANTEC Tech. Papers, 2000, 46, 739-43. Chang, M., C., O. SPE ANTEC Tech. Papers, 1996, 42, 360-7. Chang, M., C., O. Intern. Polymer Processing, 1996, XI, 76-81. Chang, R. and W., Yang. SPE ANTEC Tech. Papers, 2001, 47, 661. Charmeau, J. Y., Tixier, M., Roux, D., Barthez, J. M., Durand, V., Gazonnet, J. P., Monasse, B., Mathieu, L., Stockman, L., Vincent, M. and J. M., Haudin. Mater. Tech. (Paris), 2000, 88 (Num. Hors Se), 39-43. Chen, S. C., Peng, H. S., Wang, L. T., Chang, C. T., Wei, D. Y. and R. C., Rao. SPE ANTEC Tech. Papers, 2000, 46, 543-6. Chiang, H. H., Hieber, C. A. and K. K., Wang. Polymer Engineering and Science, 1991, 31(2), 116-24. C-Mold design guide, C-MOLD Documentation, http://www.scudc.scu.edu/cmdoc/dg_doc/develop/books/dg/index_3.htm Cogswell, F. N. Polymer Engineering and Science, 1972, 12, 64-73. Colangelo, T. and G. E., Tremblay. SPE ANTEC Tech. Papers, 1997, 43, 492-6. Cook, D. G., Rudin, A., and A., Plumtree. Journal of Applied Polymer Science, 1993, 48(1), 75-84. Coxe, M., Barry, C. M. F., Bank, D. and K., Nichols. SPE ANTEC Tech. Papers, 2000, 46, 648-52. Delbarre, P., Pabiot, J., Rietsch, F., Daurelle, J. F. and V., Lamblim. SPE ANTEC Tech. Papers, 1991, 37, 301-4. DellArciprete, J., Malloy, R. and S., McCarthy. SPE ATTEC Tech. Papers, 1999, 45, 467-71. Denn, M. M. Ann. Rev. Fluid Mech., 1990, 22, 13-34. Denn, M. M. Ann. Rev. Fluid Mech., 2001, 33, 265-87. Den Otter, J. L. ibid., 1970, 38, 155-168. Dharia, A. SPE ANTEC Tech. Papers, 1999, 45, 532-8. 232 1998.

Dillon, P. Stakeholder Dialogues, Tufts University, November 1999. Dillon P. S., and E. N., Aqua. Technical Report #20. Chelsea Center for Recycling and Economic Development, University of Massachusetts. March, 2000. Dubay, R. SPE ANTEC Tech. Papers, 2001, 47, 3351-5. Dzeskiewicz, L., Farrel, R. E. and J., Winkler. SPE ANTEC Tech. Papers, 1993, 39, 3208-3212. Eriksson, P. A., Albertsson, A. C., Boydell, P. and J. A. E., Manson. Polymer Engineering and Science, 1998, 38(5), 749-756. Fasset, J. Plast. Eng., 1995, 51(12), 35-7. Fierens, B. and S., Mertes. SPE ANTEC Tech. Papers, 1998, 44, 496-8. Frenkel, J. Acta Physicochim, URSS, 1944, 19, 51. Gamalski, J. Senior Director. Corporate Production and Logistics: Technologies and Materials Assembly, Siemens, 1996. Gao, D. M., Nguyen, K. T., Laroche, D.and M. A., Garcia. Adv. Performance Materials, 1998, 5, 43. Garcia, M. A., Macosko, C. W., Subbiah, S.and S. I., Guceri. Intern. Polymer Processing, 1991, VI, 1. GE Plastics. Properties Guide: GE Engineering Thermoplastics. Pittsfield, 1992. Gao, F., Patterson, W. I. and M. R., Kamal. Polymer Engineering and Science, 1996, 36, 1272-85. Gao, F., Patterson, W. I. and M. R., Kamal. Polymer Engineering and Science., 1996, 36, 2467-76. Grenchus, E., Keene, R., Nobs, C., Brinkley, A., Kirby, J. R., Pitts, D. and I., Wadehra. International Symposium on Electronics and the Environment. Institute of Electrical and Electronic Engineers (IEEE): Oakbrook, IL, 1998, 270-274. Grillet, A. M., Bogaerds, A. C., Peters, G. W. M., and F. P. T., Baaijens. PPS Sixteen Annual Meeting, June 18-23, Shanghai, China, 2000, 148-9. Hamada, H. and H., Tsunasawa. J. Appl. Polym. Sci., 1996, 60, 353-62. Han, R. and M., Gupta. SPE ANTEC Tech. Papers, 1999, 45, 599. 233

Hanemann, T., Piotter, V., Ruprecht, R. and J. H., Hausselt. Molding of Polymeric Microstructures, 1997a, 99. Hanemann, T., Ruprecht, R. and J., Hausselt. Microsystem Technology, 1997b, 129. Hatch, D., Kazmer, D. and B., Fan. SPE ANTEC Tech. Papers, 2001, 47, 428-32. Hendrix, J., Massey, K. A., Whitham, E. and B., Brass. International Journal of Environmentally Conscious Design & Manufacturing, 1996, 5 (1): 35-47. Hetu, J. F., Gao, D. M., Garcia-Rejon, A.and G., Salloum. Polym. Eng. Sci., 1998, 38, 223. Heuzey, M. C., Dealy, J. M., Gao, D. M. and A., Garcia-Rejon. SPE ANTEC Tech. Papers, 1997, 43, 532-7. Hieber, C. A. Injection Molding and Compression Molding Fundamentals, A. I. Isayev, ed. Marcel Dekker, New York (1987). Hobbs, S. Y. Polymer Engineering and Science, 1986, 26(1), 74-81. Hobbs, S., Y. Polymer Engineering and Science, 1996, 32(11), 1489-94. Hubschman, J., Johnson, K., Ketchum, J., Lundquist, B., Meier, B., Moore, E., Osdoba, T., Andre, S., and B., Barnum. Management of Waste Electronic Appliances, Minnesota Office of Environmental Assistance, St. Paul Minnesota, 1995. IDES. Pocket Specs for Injection Molding 1999-2000, 4th Ed., 1999, 478-479. Jayaraman, K., Papworth, P., Shu, C and M. D., Wolkowicz. SPE ANTEC Tech. Papers, 2002, 48, 2330-4. Jones, G. Proceedings of the 3rd SPE Annual Recycling Conference. Society of Plastics Engineers: Brookfield, CT. (7-8 November), 1996, 181-193. Kalnin, F. A. and G. P., Zluhan. SPE ANTEC Tech. Papers, 1999, 45, 675-9. Kansal, G. Polym. Plast. Technol. Eng. 2000, 39(1), 61-81. Kartalis, C. N., Papaspyrides, C. D., Pfaendner, R., Hoffmann, K. and H., Hherbst. Journal of Applied Polymer Science. 1999, 73, 1775-1785. Hatzikiriakos, S. G. PhD thesis, McGill University, Montreal, 1991. Kazatchkov, I. B., Hatzikiriakos, S. G. and C. W., Stewart. Polymer Engineering and Science, 1995, 35,1864-71. 234

Koelling, K. W. and R. K., Prudhomme. Rheol. Acta, 1991, 30, 511-22. Langerak, E. E. IEEE International Symposium on Electronics and the Environment. Institute of Electrical and Electronic Engineers (IEEE): San Francisco, CA, 63-68, 1997. Larson, R. G. Rheol. Acta, 1992, 31, 213-63. Lee, T. H. and N. J., Mills. J. Materials Science, 1994, 29, 2704-12. Liu, X., and H., Bertilsson. Journal of Applied Polymer Science, 1999, 74: 510-515. Loenov, A. I. and A. N., Prokunin. Non-linear Phenomena in Flow of Viscoelastic Polymer Fluids, Chapman & Hall, 2-6 Boundary Row, London SE1 8HN, UK. 1994. Macfarlane, S., and R., Dubay. SPE ANTEC Tech. Papers, 2000, 46, 653-7. Macosko, C. W. Rheology Principles, Measurements, and Applications, Wiley-VCH, Inc. 1994. Mahishi, M. SPE ANTEC Tech. Papers, 1998, 44, 547-51. Maloney, R. P. and A. J., Poslinski. SPE ANTEC Tech. Papers, 1998, 44, 542-6. Meacham, A., Uzsoy, R., and U., Venkatadri. 1999. Journal of Manufacturing System 18(5): 311-322. Meffert, H., and M., Kirchgeorg. Business and the Environment, 1997, 8(2), 5. Miller, T. A. SPE ANTEC Tech. Papers, 1995, 41, 4267-70. Narh, K. A., Guo, J., Xanthos, M., Yilmazer, U., and V., Tan. SPE ANTEC Tech. Papers, 1999, 45, 3294-3298. NDSU Center for Nanoscale Science and Engineering, http://www.ndsu.edu/cnse/facilities/dunbar_cleanroom/cleanroom_equipment.php#veec o, 2003. Nguyen, T. Q., and H. H., Kausch. Chimia, 1986, 40, 129. Odell, J. A., and A., Keller. Journal of Polymer Science, Part B: Polymer Physics, 1986, 24(9), 1889-916. Pagel, S. M. SPE ANTEC Tech. Papers, 1989, 35, 330-332. Palmer, A. E. and J. S., Colton, Polymer Engineering and Science, 2000, 40, 1395-404.

235

Piau, J. M., Kissi, N. E. and B., Tremblay. J. Non-Newtonian Fluid Mech., 1988, 30, 197-232. Piau, J. M., Kissi, E. E. and B., Tremblay. J. Non-Newtonian Fluid Mech. 1990, 34(2), 145-80. Piau, J. M., Kissi, E. E., Toussaint, F. and A., Mezghani. Rheol. Acta, 1990, 34, 40-57. Piau, J. M. and J. F., Agassant. Rheology for Polymer Met Processing, Elsevier Science B. V. 1996. Piotter, V., Hanemann, T., Ruprecht, R., and J., Hausselt. Microsystem Technology, 1997, 129. Piotter, V., Bauer, T., Benzler, T. and A., Emde. Microsystem Technology, 2001, 7, 99. Ries, H., and Menges, G. Kunststoffe German Plastics, 1988, 78(7), 32-34. Rogalla, A. and W., Michaeli. SPE ANTEC Tech. Papers, 1997, 43, 364. Rothstein, J. P. and G. H., McKinley. J. Non-Newtonian Fluid Mech. 1999, 86, 61-88. Rothstein, J. P. and G. H., McKinley. J. Non-Newtonian Fluid Mech. 2001, 86, 33-63. Roy, R. K. Design of Experiments by Taguchi Approach: 16 Steps to Product and Process Improvement, John Wiley & Sons, Inc., New York (US), 2001. Salamon, B. A., Koppi, K. A. and J., Little, SPE ANTEC Tech. Papers, 1998, 44, 515-9. Sander, M. A. Practical Guide to the Assessment of Surface Texture. Feinpruf Perthen GmbH. Gottingen, Postfach, Germany, 1991. Schramm, G. A. Practical Approach to Rheology and Rheometry, Gebrueder HAAKE Gmbh, Karlsruhe, Federal Republic of Germany. 1994. Selden, R. J. Injection Molding Technol., 2000, 4(4), 159-66. Shah, R K. and A. L. London. Advances in Heat Transfer, Acadamic Press, New York. 1978. Selden, R. J. Injection Molding Technol., 2000, 4(4), 159-66. Sherbelis, G., and C., Friedl. SPE ATTEC Tech. Papers, 1996, 42, 778-82. Shriver, M. E., Beiter, K. A., and K., Ishii. SPE ANTEC Tech. Papers, 1994, 40, 29102915. 236

Smialek, C. D. and C. L., Simpson. SPE ANTEC Tech. Papers, 1998, 44, 3373-7. Society of the Plastics Industry, Inc., http://www.plasticsdatasource.org/. Last updated July 26, 2002. Copyright 2001.

Sridhar, L. and K. A., Narh. SPE ANTEC Tech. Papers, 1999, 45, 496-9. Stuart, J. A. NSF Annual Report for Award #9727209, 1999. Stuart, J. A., Ammons, J. C. and L. J., Turbini. Operations Research, 1999, 47(2), 221234. Szekely, J., and R., Laudise. Journal of Materials Research, 1995, 10(3), 485-486. Timmons, D. Eastman Kodak Company. Personal Communication, 17 November, 1998. Tredoux, L. and I., Satoh. Polymer Engineering and Science, 1999, 39(11), 2233-41. Tredoux, L., Satoh, I. and I., Kurosaki. Polymer Engineering and Science, 2000, 40(10), 2161-74. U.S. Environmental Protection Agency. Municipal Solid Waste in the United States: Facts and Figures for 2000. EPA530-R-02-001. Office of Solid Waste and Emergency Response. June, 2002. Weber, L. and W., Ehrfeld. Kunststoffe, 1999, 89 (10), 192. Whetten, A. and D., Belcher. SPE ANTEC Tech. Papers, 1994, 40, 593-600. White, S. A., Gotsis, A. D. and D. G., Baird. J. Non-Newtonian Fluid Mech., 1987, 24, 121-60. Xu, G. and K. W., Koelling. SPE ANTEC Tech. Papers, 2001, 47, 604-7. Xu, G. and K. W., Koelling. SPE ANTEC Tech. Papers, 2002, 48, 521-5. Xu, G., and K. W. Koelling. SPE ANTEC Tech. Papers, 2003, 49, 566-70. Yang, X., Ishida, H. and S., Wang. J. Rheol., 1998, 42(1), 63-80. Yesilata, B., Oztekin, A. and S., Neti. J. Non-Newtonian Fluid Mech, 1990, 34, 145-80. Yesilata, B., Oztekin, A. and S., Neti. J. Non-Newtonian Fluid Mech, 1999, 85, 35-62. Yokoi, H., Nagami, S., Kawasaki, A. and Y., Murata. SPE ANTEC Tech. Papers, 1994a, 40, 368-72. 237

Yokoi, H. PPS Tenth Annual Conference, April 5-8, Akron, OH, USA, Presentation, 1994b, 171-2. Yokoi, H., Nagami, S., Kawasaki, A. and Y., Murata. SPE ANTEC Tech. Papers, 1994c, 40, 829-32. Yoshii, M., Kuramoto, H. and K., Kato. Polymer Engineering and Science, 1993, 33(19), 1251-60. Yoshii, M., Kuramoto, H., Kawana, T. and K., Kato. Polymer Engineering and Science, 1996, 36(6), 819-26. Yu, L., Koh, C. G., Koelling, K. W., Lee, L. J. and M. J., Madou. SPE ANTEC Tech. Papers, 2001, 47, 785-9. Yu, L., Koh, C. G., Lee, L. J., Koelling, K. W. and M. J., Madou. Polym. Eng. Sci., 2002, 42, 871-88. Yu, L., Lee, L. J. and K. W., Koelling. Polym. Eng. Sci., 2004a (In press). Yu, L., Xu, G., K. W., Koelling and L. J., Lee. SPE ANTEC Tech. Papers, 2004b, 50 (In press). Zahavich, A. T. P., Takacs, E., Latto, B., and J., Vlachopoulos. SPE ANTEC Tech. Papers, 1992, 38, 1186-1188.

238