The Pennsylvania State University. The Graduate School. Department of Engineering Science and Mechanics

Transcription

1 The Pennsylvania State University The Graduate School Department of Engineering Science and Mechanics DIMENSIONAL CAPABILITIES OF POWDER-BASED MICROCOMPONENTS FABRICATED BY UTILIZING FILLED PHOTORESITS A Thesis in Engineering Science by Kevin Robert Geist 2010 Kevin Robert Geist Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Science December 2010

2 ii The thesis of Kevin Robert Geist was reviewed and approved* by the following: Donald F. Heaney Associate Professor, Director of CISP Thesis Advisor Barbara Shaw Professor of Engineering Science and Mechanics Judith A. Todd P. B Breneman Department Head Chair Head of the Department of Department of Engineering Science and Mechanics *Signatures are on file in the Graduate School

3 iii ABSTRACT Experiments were performed in order to determine and quantify the effects of processing conditions on micro components created with filled photoresist. This process combines standard photolithographic procedures with that of standard powder metallurgy process. This technique offers two major benefits; the first being lithographic sized features in the 1-5 μm range because of the photolithographic portion of the process, and the second is the ability to use virtually any metal alloy to form components which is attributed to the powder metallurgy portion of the process. The filled photoresist process consists of filling a photosensitive polymer material, such as Micro Chem s Su-8 negative photoresist, with powder metal (stainless steel 316L). This mixture is then applied to a ceramic substrate through a spin coating process. The filled photoresist is then exposed to ultra violet light through a mask of desired geometries. The filled photoresist that is exposed to the ultra violet light through the mask becomes cross-linked, which enables it to stay affixed to the substrate. During the development step the non cross-linked filled photoresist is removed from the substrate leaving behind the micro components. The micro components are then debinded in air to remove the photoresist and lightly sinter the metal powder particles. Finally the micro components are sintered in hydrogen to densify. The effects of solids loading and powder particle size on dimensional variability were investigated. The effect of sintering temperatures on density and shrinkage was also evaluated. Dimensional variability was measured at two points during the processing steps. The micro components were measured after forming and after sintering. Six sigma precision was found to range from.0305 mm to.0524 mm depending on which dimension was measured. Based on studies the components displayed a slightly high dimensional variability. This variability was attributed to the inability to create high enough solids loading slurries of filled photoresist. It was shown that solids loading had a definite effect on dimensional variability of

4 iv both green and sintered components as well as an effect on feature size. Shrinkage was also characterized and was found to be isotropic. This ruled out sintering as a source for additional dimensional variation.

5 v TABLE OF CONTENTS LIST OF FIGURES... vii LIST OF TABLES... xi ACKNOWLEDGMENTS... xiv Chapter 1 Introduction... 1 Chapter 2 Background... 3 Market Overview... 3 Available Forming Technologies... 4 Lithographic Techniques... 4 Lithography... 4 LIGA... 6 Microtransfer Molding... 7 Suspension Casting... 7 Non-Lithographic Techniques... 8 Direct Ceramic Machining... 8 Electrical Discharge Machining... 9 Gel Casting... 9 Screen Printing Plastic Injection Molding Metal Injection Molding Powder Metallurgy Materials Solids Loading Debinding and Sintering Chapter 3 Experimental Procedure Experimental Design Powder Selection Powder Characterization Photoresist Selection Basic Process Spin Coating Experiments Exposure Experiments Debinding and Sintering Dimensional Analysis Qualitative analysis Chapter 4 Results Filled Photoresist Formation and Viscosity Variation... 28

6 vi Spin Thickness Exposure Depth Qualitative Analysis Dimensional variability Tooling dimension Green dimension Sintered dimension Chapter 5 Discussion Penetration Depth Analysis Viscosity Analysis Green component analysis Sintered Component Analysis Shrinkage Analysis Process Induced Variations Chapter 6 Conclusion Chapter 7 Future Work Appendix A Viscosity Data...91 Appendix B Spin Thickness Data...92 Appendix C Green Dimensional Data...94 Appendix D Sintered Component Dimensional Data Appendix E Sintered Component Density Appendix F Variability Data Appendix G Non-Technical Abstract References...114

7 vii LIST OF FIGURES Figure 2-1: Basic lithographic process of a negative photoresist... 6 Figure 3-1: Osprey 80% -5μm stainless steel 316 L powder batch No. 05D0280 at 1000x Figure 3-2: Osprey 90% -16μm stainless steel 316 L powder batch No. 03D0237 at 1000x Figure 3-3: Osprey 90% -22μm stainless steel 316 L powder batch No at 1000x Figure 3-4: Experimental Process Flow Diagram Figure 3-5: Laurell Spin Processor used in the spin coating process Figure 3-6: Spin cycle for coating ceramic substrates with filled photoresist Figure 3-7: OAI ultra violet exposure system used for exposing the filled photoresist Figure 3-8: Debind cycle for stainless steel filled photoresist micro components Figure 3-9: Picture of CM tube furnace used for sintering components Figure 3-10: Sintering cycle used for stainless steel micro components Figure 3-11: Micro components utilized for dimensional analysis Figure 4-1: Plot of the difference in viscosities between the three different powder sizes and Figure 4-2: Effect of powder size on spin thickness at 45% solids loading Figure 4-3: Effect of powder size on spin thickness at 40% solids loading Figure 4-4: Effect of powder size on spin thickness at 35% solids loading Figure 4-5: Effect of solids loading on spin thickness at 45% solids loading Figure 4-6: Effect of solids loading on spin thickness at 40% solids loading Figure 4-7: Effect of solids loading on spin thickness at 35% solids loading Figure 4-8: Effect of solids loading on exposure depth at the three powder sizes Figure 4-9: Effect of powder size on exposure depth at the three solids loadings Figure 4-10: Images prior to sintering of 35% and 40% solids loading -5μm SS 316L green components

8 Figure 4-11: Images prior to sintering of 35%, 40% and 45% solids loading -16μm SS 316L green components Figure 4-12: Images prior to sintering of 35%, 40% and 45% solids loading -22μm SS 316L green components Figure 4-13: Images after sintering of 35% and 40% solids loading -5μm SS 316L components Figure 4-14: Images after sintering of 35%, 40% and 45% solids loading -16μm SS 316L components Figure 4-15: Images after sintering of 40% and 45% solids loading -22μm SS 316L components Figure 4-16: Micro components utilized for dimensional analysis Figure 4-17: Box plot analysis of green location #1 (mm) dimensions for each powder size and solids loading Figure 4-18: Box plot analysis of green location #2 (mm) dimensions for each powder size and solids loading Figure 4-19: Box plot analysis of green location #3 (mm) dimensions for each powder size and solids loading Figure 4-20: Main effect (data means) plot analysis of different solids loadings and powder sizes at location # Figure 4-21: Main effect (data means) plot analysis of different solids loadings and powder sizes at location # Figure 4-22: Main effect (data means) plot analysis of different solids loadings and powder sizes at location # Figure 4-23: Main effects (data means) plot analysis of standard deviation of the green location #1 at Figure 4-24: Main effects (data means) plot analysis of standard deviation of the green location #3at different solids loadings and for three different powder sizes Figure 4-25: Main effects (data means) plot analysis of standard deviation of the green location #2 at different solids loadings and for three different powder sizes Figure 4-26: Plot of green location #1, mean values with six sigma tolerance limits at the eight forming conditions Figure 4-27: Plot of green location #2, mean values with six sigma tolerance limits at the eight forming conditions viii

9 ix Figure 4-28: Plot of green location #3, mean values with six sigma tolerance limits at the eight forming conditions Figure 4-29: Box plot analysis of location #1 (mm) dimensions for each powder size and solids loading Figure 4-30: Box plot analysis of location #2 (mm) dimensions for each powder size and solids loading Figure 4-31: Box plot analysis of location #3 (mm) dimensions for each powder size and solids loading Figure 4-32: Main effect (data means) plot analysis of the sintered location #1 at different powder sizes and solids loadings Figure 4-33: Main effect (data means) plot analysis of the sintered location #2 at different powder sizes and solids loadings Figure 4-34: Main effect (data means) plot analysis of the sintered location #3 at different powder sizes and solids loadings Figure 4-35: Main effect (data means) plot analysis of the standard deviation of the sintered location #1 at different powder sizes and solids loadings Figure 4-36: Main effect (data means) plot analysis of the standard deviation of the sintered location #2 at different powder sizes and solids loadings Figure 4-37: Main effect (data means) plot analysis of the standard deviation of the sintered location #3 at different powder sizes and solids loadings Figure 4-38: Plot of sintered location #1 mean values with six sigma tolerance limits at 7 molding conditions Figure 4-39: Plot of sintered location #1 mean values with six sigma tolerance limits at 7 forming conditions Figure 4-40: Plot of sintered location #1 mean values with six sigma tolerance limits at 7 forming conditions Figure 5-1: Plot of spin thickness versus the viscosity of the filled photoresist slurry that Figure 5-2: Plot of viscosity versus exposure depth of the three powder sizes. Notice that Figure 5-3: Plot of -5μm component dimensional variability for location # Figure 5-4: Plot of -5μm component dimensional variability for location # Figure 5-5: Plot of -5μm component dimensional variability for location # Figure 5-6: Plot of -16μm component dimensional variability for location #

10 x Figure 5-7: Plot of -16μm component dimensional variability for location # Figure 5-8: Plot of -16μm component dimensional variability for location # Figure 5-9: Plot of -22μm component dimensional variability for location # Figure 5-10: Plot of -22μm component dimensional variability for location # Figure 5-11: Plot of -22μm component dimensional variability for location #

11 xi LIST OF TABLES Table 2-1:Overview of lithographic and non lithographic process listed in order in which they appeared previously Table 3-1: Processing variables and their responses Table 3-2: Summary of powder characteristics Table 3-3: Summary of all samples tested for ultra violet light exposure depth Table 4-1: Summarization of the exposure mask master geometries Table 4-2: Summarization of the green components variability for all three powder sizes and solids loadings Table 4-3: Listing of forming conditions for Figure Table 4-4: Listing of forming conditions for Figure Table 4-5: Listing of forming conditions for Figure Table 4-6: Summarization of the variability of the sintered components for the different powder sizes at different solids loadings Table 4-7: Listing of forming conditions for Figure Table 4-8: Listing of forming conditions for Figure Table 4-9: Listing of forming conditions for Figure Table 5-1: Green t-test for pooled powder sizes, 35% versus 40% solids loading Table 5-2: Green t-test for pooled powder sizes, 40% versus 45% solids loading Table 5-3: Green t-test for -5µm powder t-test 35% versus 40% solids loading Table 5-4: Green t-test for -16µm powder t-test 35% versus 40% solids loading Table 5-5: Green t-test for -16µm powder t-test 40% versus 45% solids loading Table 5-6: Green t-test for -22µm powder t-test 35% versus 40% solids loading Table 5-7: Green t-test for -22µm powder t-test 40% versus 45% solids loading Table 5-8: Dimensional precision of plastic injection molding [36] and components fabricated in this thesis Table 5-9: Pooled powder sizes for sintered 35% versus 40% solids loading

12 xii Table 5-10: Pooled powder sizes for sintered 40% versus 45% solids loading Table 5-11: Pooled powder sizes for sintered 35% versus 45% solids loading Table 5-12: -5 μm powder t-test for sintered 35% versus 40% solids loading Table 5-13: -16 μm powder t-test for sintered 35% versus 40% solids loading Table 5-14: -16 μm powder t-test for sintered 40% versus 45% solids loading Table 5-15: -22 μm powder t-test for sintered 35% versus 40% solids loading Table 5-16: Dimensional Precision of MIM and P/S based on dimension size Table 5-17: Theoretical shrinkage values for each powdered size and solids loading. Calculated from the measured densities of the sintered components Table 5-18: Shrinkage and scale up factors for sintered 35 and 40 volume percent -5 m components Table 5-19: Shrinkage and scale up factors for sintered 35, 40 and 45 volume percent - 16 m components Table 5-20: Shrinkage and scale up factors for sintered 40 and 45 volume percent -22 m components Table A- 1: Viscosity Measurements Table B- 1: 35 volume % spin thickness of all powder sizes Table B- 2: 40 volume % spin thickness of all powder sizes Table B- 3: 45 volume % spin thickness of all powder sizes Table C- 1: Lithographic Mask Geometry Measurements Table C- 2: 35 Volume % -5 µm Green Component Measurements Table C- 3: 40 Volume % -5 µm Green Component Measurements Table C- 4: 35 Volume % -16 µm Green Component Measurements Table C- 5: 40 Volume % -16 µm Green Component Measurements Table C- 6: 45 Volume % -16 µm Green Component Measurements Table C- 7: 35 Volume % -22 µm Green Component Measurements Table C- 8: 40 Volume % -22 µm Green Component Measurements

13 Table C- 9: 45 Volume % -22 µm Green Component Measurements xiii Table D- 1: 35 Volume % -5 µm Sintered Component Measurements Table D- 2: 40 Volume % -5 µm Sintered Component Measurements Table D- 3 : 35 Volume % -16 µm Sintered Component Measurements Table D- 4: 40 Volume % -16 µm Sintered Component Measurements Table D- 5: 45 Volume % -16 µm Sintered Component Measurements Table D- 6: 40 Volume % -22 µm Sintered Component Measurements Table D- 7: 45 Volume % -22 µm Sintered Component Measurements Table E- 1: 40 Sintered Component Densities Table F- 1: -5 µm Component Variability Table F- 2: -16 µm Component Variability Table F- 3: -22 µm Component Variability

14 xiv ACKNOWLEDGMENTS I would like to thank Don Heaney for the wonderful opportunity he has bestowed upon me. Without his help and guidance I would not of had the resources or the ability to complete this thesis. I have genuinely enjoyed the opportunity I have been given to work at the university, in particular the CISP P/M Lab. I am also very grateful for the ability to access APP's facilities to conduct research. I would also like to thank Kristina Cowan-Giger, Michael Dissab-Miller and Derek Neupauer for their help as well as my committee member, Barbara Shaw. Everyone was tremendously helpful and cordial at CISP and APP, which made conducting research very pleasant. I would also like to express my gratitude towards family and friends for their encouragement and support throughout. Dad and Jan, thank you for your guidance and caring ways. Finally, I would like to thank my girl friend Karen for the generous amount of help, love and encouragement that she has given me.

15 1 Chapter 1 Introduction In today s world the need to miniaturize many mechanical components is ever increasing. Many of these components need to be created from structural alloys and other metals that there is no clear method for producing yet. With the trend towards miniaturization in many economic areas a gap of technology is created for parts of this nature. [35]. The method being researched uses powder filled photoresist to create parts of this nature. The powder filled photoresist method uses a powder metal alloy, 316L stainless steel, and the photolithographic process to create micro components. The integration of powder metal and photolithographic processing was accomplished by mixing a powder metal with a negative tone photoresist. The mixture or slurry is then processed using standard photolithographic techniques. The result of the processing leaves components on a substrate that are comprised of the powder metal that was used and photoresist as a binder holding the powders together. From this point, standard sintering techniques are used to sinter the components to their final density. Because the lithographic process is being utilized to form the green components, features can be created in the 1-5 μm range [34]. Conventional tooling fabrication methods such as wire electrical discharge machining (EDM) can only produce feature sizes of 50μm [13]. The unique aspect of this technology is that components can be fabricated with feature sizes from 1-5μm of any alloy. Nonpowder based technologies, such as electroplating, are limited to certain types of metals, like Ni and Cu, that can be used to process components. With this technology, the components can be created from a very large range of materials. They can be fabricated from a pure polymer to a ceramic, or in the case of this research, a metal alloy.

16 2 The objective of this research was to combine lithographic techniques with powder metal processing to create micro components with lithographic features of metal alloys. The importance of incorporating these processing technologies is high resolution features, low cost, high manufacturing rates and the ability to process different alloy powders. Since this method is relatively new, the effects of solid loading and particle size were examined to determine their effects on the finished micro components. These finished components were examined to determine dimensional variability, shrinkage and tooling scale-up factor.

17 3 Chapter 2 Background This section gives a detailed overview of the current technologies for creating micro components. This covers the forming process, materials used, and the capabilities of the process. This section also covers powder processing technologies. Market Overview There is a great need for miniaturization in many areas of industry. Because of this the need for MEMS (microelectromechanical systems) technology is greatly increasing. MEMS is the technology that involves devices that are 1 millimeter to 20 micrometers in size [40, 41]. MEMS are growing in popularity for many consumer devices, such as pressure sensors, microphones, accelerometers, and many other micro components for various different applications [42]. MEMS have become so popular because they offer many advantages when integrated into other functions to enable the miniaturization of an entire device. The growing demand for components of this nature has generated a need for a new fabrication process that can mass produce these components [43].

18 4 Available Forming Technologies There are currently many methods being utilized and studied to develop and fabricate MEMS and microcomponents. There are methods that utilize lithography and those that do not. Both methods are reviewed in the following sections. Lithographic Techniques These are forming techniques that are based around the lithographic technology. Lithography Lithography is the process of transferring a pattern to a photo-sensitive material by exposing it to ultra violet light. The photo-sensitive material that is used is typically called a photoresist. A photoresist is a polymeric material that either becomes cross-linked (negative photoresist) or un-cross-linked (positive photoresist) in the presence of certain wave lengths of light. The wavelength of the light plays an important role in the cross-linking of the photoresist, with thinner layers of photoresist corresponding to shorter wavelengths. This permits a reduced aspect ratio and a reduced minimum feature size. This is especially important in microelectronics for reduction in minimum feature size. The basic procedure used in lithography consists of: coating a substrate with a photoresist material, covering the photoresist with a patterned mask, exposing certain portions of the material through the mask, developing away the non-cross-linked photoresist, and typically etch the substrate although the photoresist structure could be used for other purposes[1-2]. This process is shown in Figure 2-1. The most common way of applying a photoresist to a substrate is by spin coating. Spin coating is when a liquid photoresist is placed on a spinning substrate and the centrifugal forces present are used to spread the photoresist

19 5 consistently over the substrate [3-4]. A typical mask consists of a chrome pattern on a piece of quartz glass. Exposure is regularly done with an ultraviolet light source; however some photoresists can be exposed with an x-ray or laser light source [11]. There are two main types of exposure techniques that are utilized in photolithography, and they are contact and projection lithography. Contact lithography is the simpler of the two. With this technique the mask is placed directly on the photoresist and then exposed to a uniform light intensity. Contact lithography offers ease of use but limits feature size of the components to that of which is obtainable by the mask which is typically chrome plated quartz glass. This process also makes the mask and photoresist susceptible to damage since the mask and photoresist come into contact [32]. Projection lithography uses a mask suspended above the photoresist that projects a shrunken version of the pattern onto the photoresist. Development occurs in an aqueous solution that breaks the weak bonds between the non cross-linked photo resist. The lithography technology is most commonly used in the integrated circuit industry; however it also has many uses in MEMS and microcomponent fabrication.

20 6 Resist Spin Coating Exposure Substrate UV Source Mask Resist Substrate Development Resist Substrate Plating Resist and Plating Substrate Strip Plating Substrate Figure 2-1: Basic lithographic process of a negative photoresist LIGA Lithographie galvanoformung abformung (LIGA) is a micro forming technology that uses lithography, micro electroplating, and micro molding. LIGA can typically produce micro components with a feature resolution between 1 and 20 µm. The first step in this process is to use the lithographic process to fabricate a three-dimensional structure in a thick layer of resist. The next step is to use the resist to create a complimentary metal structure through the use of electroforming. This plating process is limited to a narrow range of metals [5-6]. This metal structure can be used as the final product or a mold for a replication process. If the piece is to be used for a molding process, such as metal injection molding (MIM) or micro powder injection

21 7 molding (MPIM), there is a vast assortment of metals and polymers that could be used to create components. One drawback of this process is the need for high energy x-ray sources which are very costly [7,11]. There is also a potential for harm to the environment with this process because plating forms hazardous wastes [6]. Microtransfer Molding µtm (Microtransfer molding ) is a forming technology where an elastomeric mold is used to fabricate components. In this technique an elastomeric mold is created using Poly(dimehylsiloxanes), also known as PDMS, and filled with a polymer or metal slurry to create components. PDMS is a common elastomer used for creating the mold because it has several unique properties that makes it appropriate for this application [8]. The elastomer is poured over a master mold that has relief structures on its surface. The PDMS is then allowed to cure before removing it from the master mold. The cavities in the PDMS mold are filled with a polymer solution then brought into contact with a substrate. It is then typically heated or put under UV light to cure the polymer, and peeled away leaving the resulting micro components on the substrate. µtm has a feature resolution between 1 and 50 µm which is determined by the LIGA master mold. Excellent feature resolution along with reduced costs, compared to other processes, makes µtm highly desirable for construction of micro components [19]. Suspension Casting Casting suspensions into lithographic masks is a method used for creating components with features in the micrometer range. Heule et. al.[9] formed three-dimensional structures by exposing a thick positive photo resist (Clariant AZ4562) to a ultraviolet light source through a

22 8 chromium pattern on a glass substrate. The patterned photoresist was then developed with Microposit 351 basic solution. The voids left after development were then filled with a tin oxide suspension of 277 nanometer average particle size. The photoresist structure was then removed using tetrahydrofuran to prevent the thermal expansion that occurs during the pyrolysis of Clariant AZ4562. After sintering, ceramic structures with high resolution patterns were able to be obtained [9-10]. This could also be a viable method for creating metal microcomponents if a metal powder was used as a substitute for the ceramic powder. Non-Lithographic Techniques These are techniques in which the lithographic process does not play an essential part in the micro component manufacturing process. Direct Ceramic Machining Direct ceramic machining (DCM) is a method of rapid manufacturing of ceramic components in the.1 millimeter to 1 millimeter range. In this method, partially sintered ceramic bodies can be machined with enlarged contours to compensate for the shrinkage that occurs after being fully sintered [15]. After the ceramic bodies are machined, they are fully sintered. During the sintering the ceramic components shrink to full density and thus shrink to their final dimensions. The smallest linear feature resolution obtainable through this technique is 50 µm [2].

23 9 Electrical Discharge Machining Electrical discharge machining (EDM) is another manufacturing process that is used to for making complex and simple geometries in parts and assemblies. Wire EDM is capable of feature resolutions as small as 50 µm [13]. The resolution of this process is limited by the diameter of the wire used for the machining. EDM works by eroding material in the path of electrical discharges that form an arc between an electrode and the work piece [13]. The eroded material is then rinsed away by a dielectric fluid. EDM is very useful when working with difficult to machine materials, but it is not well suited for very small pieces and large quantities [14]. Also wire EDM can only make two dimensional components. Gel Casting Gel casting is a novel ceramic forming process in which a ceramic slurry is formed and then cast into a non-porous mold. The slurry is then allowed to solidify by evaporating the solvent from the slurry and a green body component is formed. The parts can then be further machined for additional dimensional accuracy. Finally the parts are sintered at high temperatures [16-18]. This process has been around since the 1990 s and has many advantages, such as: use of low cost molds, green body parts with high strength, machinability of green bodies for higher resolution, ability to make complex components, low organic content for ease of binder removal, and possible material selection is very flexible in that it is not limited to only ceramic powders [16-18]. This process is capable of producing components in the 500µm to millimeter range [16-18].

24 10 Screen Printing Screen printing is a simple powder based technique that uses a finely woven mesh screen to produce a pattern. In this process a paste or ink, consisting of the material of interest, a binder and a solvent, is pressed onto a substrate through the openings in the mesh screen. This method is very cost-effective and an efficient way of depositing virtually any ceramic powder that does not require complex equipment or processing [10]. However, through this process resolution is only in the range of 100 micrometers [12]. Plastic Injection Molding Plastic injection molding (PIM) is a commonly used manufacturing process in which melted plastic is injected into mold cavities of desired component geometries. The steps in the PIM process are as follows: heating the plastic so it will flow into a mold, solidification of the plastic in the mold cavity, and finally the ejection of the plastic component. This process is limited by the tooling, however it is capable of producing components varying in size from very large down to 500 µm [20]. Metal Injection Molding Metal injection molding (MIM) is a process of producing complex net-shaped metal parts. MIM uses the same process as PIM, however, instead of a plastic only feedstock, there is metal powders blended with the plastic feedstock [21]. MIM is a very attractive process especially when the following factors apply: thickness ranging from.2 to 20mm, mass ranging from.02 to 1000 g, moderate levels of shape complexity and smooth surfaces. This process is

25 11 limited by the tooling that can be created with standard techniques; however, MIM can typically produce tolerances in the range of.1 and 1mm [22]. An overview of the feature resolution obtainable with the lithographic and nonlithographic processes can be seen in Table 2-1. Table 2-1:Overview of lithographic and non lithographic process listed in order in which they appeared previously Method Smallest feature resolution (µm) References LIGA 1-20 [5-7,11] Microtransfer molding µtm 1-50 [8,19] Casting suspensions into 5-20 [9-10] photolithographic masks Direct ceramic machining 50 [2,15] (DCM) Wire EDM 50 [13] Gel casting 200 [16-18] Screen Printing 100 [10,12] Plastic injection molding 500 [20] (PIM) Metal injection molding (MIM) 500 [21-22] Powder Metallurgy Powder metallurgy is a process with the ability to fabricate near net-shape components and the versatility to tailor the material and microstructure to a desired property. The main factors affecting the manufacturing of a powder metal component are the shaping and sintering steps.

26 12 Materials The shaping step of the powder metal process relies on a mixture of metal powders and polymer binders or lubricants. This polymer is required to lubricate the flow of particles and tooling in compacting and molding processes and is sometimes used to temporarily act as a shaping agent in order to facilitate ease of handling during processing [23-24]. This binder should not have an influence in the final material composition; however, it serves a very important role in the powder metallurgy process [23-24]. Solids Loading One very important aspect of processing powder/binder mixture in powder injection molding, also known as feedstock, is the solids loading of the powder. Solids loading is the volume ratio of solid powder to the total volume of the powder and binder combination [24,26]. Solids loading is a very significant characteristic in powder processing. A feedstock with a very high solids loading leads to high feedstock viscosities making molding parts difficult or in some cases impossible. A solids loading that is to slow can lead to difficulties during the debinding process and slumping of parts due to low particle contact [24,26]. The amount of binder needed depends on the particle packing, particle shape, and particle size distribution since filling void space is necessary to maintain a low viscosity [22]. The two main concepts for solids loading are: critical solids loading and optimal solids loading. The critical solids loading corresponds to a feedstock composition where all of the powder particles are in close contact and the void space between the particles is filled with binder. The drawback of a critically loaded feedstock is that as it nears the critical solids loading, the viscosity goes to infinity [27]. An equation for calculating critical solids loading can be seen below. The optimal solids loading, the second concept is

27 13 usually a solids loading 4-6% below the critical solids loading. The optimal solids loading offers a lower viscosity making it ideal for forming process while still maintaining an acceptable amount of particle contact in order to maintain shape during debinding [27]. Φ c = Φ/ [(η b /η m ) 1/2 +1] 2.1 Where η m is the viscosity of the mixture, η b is the viscosity of the pure binder, Φ is the fractional solids loading, and Φ c is the critical solids loading. Debinding and Sintering The polymers that were used in the forming step must be removed before the components can be sintered to full density. The process of removing the binder from the parts is called debinding. Debinding can be done thermally or with a combination of thermal and solvent debinding techniques, through each process the polymers are decomposed into their constituents through the breaking of molecular bonds and removed from the part. Thermally debinding components involves chemical mechanisms with thermal degradation of organic compounds into volatile species. Some of these volatile products can be methane CH 4, carbon dioxide CO 2, carbon monoxide CO, or water H 2 O. A physical mechanism is also present during thermal debinding as diffusion of the volatile products takes place as well as changes in the binder distribution in the green body [28]. These changes in distribution are governed by diffusion and capillary migration. During thermal debind factors such as part position, part support, temperature cycle, gas flow, atmosphere, and furnace type can all have a profound effect on a components dimensional stability [22]. Solvent debinding can be affected by

28 14 solvent type, solvent temperature, debinding time, temperature gradient in the solvent bath, part position and support of the part [28]. Sintering, as defined by Hausner [31], is the bonding of particles in a mass of powders by molecular or atomic attraction in the solid state, by application of heat, causing strengthening of the powder mass and possibly resulting in densification and recrystallization by transport of materials. This bonding of particles during solid state sintering forms a metallurgical union between the particles, often referred to as a neck. This redistribution of mass leads to particle relocation, thus, moving the particles being bonded closer together and forming a larger neck. This process can be described as densification. This densification can occasionally lead to dimensional change and distortion [30]. Because densification is essential for superior mechanical properties, the sintering process must exhibit replicable and consistent shrinkage. Factors that have an effect on the shrinkage are: temperature, temperature gradient, temperature cycle, and gas atmosphere [30].

29 15 Chapter 3 Experimental Procedure Experimental Design This work investigated the processing condition effects on micro components created using powder filled photoresist. The responses being investigated and the processing conditions are listed in Table 3-1. The process used for creating micro components from powder filled photoresist consisted of coating a ceramic substrate with the filled photoresist and then exposing the wanted geometries to a UV light source, as described in the previous chapter for photolithography. Thirty components of each powder size and solids loading were fabricated to analyzed the effects of dimensional variability, and surface defects. Table 3-1: Processing variables and their responses Variables Levels Response Method Solids Loading Vol.% Spin Thickness Exposure Depth Defects Dimensional Variability Powder Size -5 μm -16 μm -22 μm Spin Thickness Exposure Depth Defects Dimensional Variability Metrology Metrology Optical Microscopy Optical Metrology Metrology Metrology Optical Microscopy Optical Metrology Gas atomized stainless steel 316L powders of varying particle sizes and distributions were mixed with SU-8 photoresist at different solids loadings to create separate filled photoresist slurries. Each slurry was individually prepared and spin coated onto a two inch diameter ceramic substrate. The micro components were sintered at a temperature of 1240 C to evaluate dimensional variability, shrinkage and feature retention.

30 16 Powder Selection Three different solids loadings were used: 35, 40, and 45 vol% as well as three different stainless steel powder particle sizes: -22 µm, -16 µm and -5 µm were used in this effort. The powders used were 80% -5 µm 316L stainless steel Batch No. 05D0280, 90% -16 µm 316L stainless steel Batch No. 03D0237 and 90% -22 µm 316L stainless steel Batch No Produced by Osprey Powders (Neath, Uk). These materials were selected because of their frequent use in the powder metal industry as well as their properties of biocompatibility for medical uses, which is a large micro component market. Powder Characterization The three powders were characterized for shape, size, and true density. A scanning electron microscope (SEM) was used to analyze the shape of the powder particles. Figures 3-1 to 3-3 show images of the particle shape of the two powders. The size of the particles was determined using a laser scattering particle size analyzer (model: Coulter LS230 Laser Scattering Particle Size Analyzer, supplier: Beckman Coulter Incorporated, Fullerton, CA). True density was determined using a helium pycnometer (model: Multipycnometer MVP-1, supplier: Quantachrome Corporation, FL). Particle size and true density can be found in Table 3-2.

31 17 Figure 3-1: Osprey 80% -5μm stainless steel 316 L powder batch No. 05D0280 at 1000x Figure 3-2: Osprey 90% -16μm stainless steel 316 L powder batch No. 03D0237 at 1000x

32 18 Figure 3-3: Osprey 90% -22μm stainless steel 316 L powder batch No at 1000x Table 3-2: Summary of powder characteristics. Powder Particle size distribution D10 (µm) D50 (µm) D90 (µm) Osprey Gas Atomized 80% -5µm SS 316L powder batch No. 05D ± 1.3% 4 ± 0.4% 6.8 ± 1.7% Osprey Gas Atomized 90% -16µm SS 316L powder batch No. 03D ± 0.6% 8.6 ± 3.1% 15.1 ± 4.2% Osprey Gas Atomized 90% -22µm SS 316L powder batch No ± 1.1% 12.1± 3.8% 18.6 ± 6.3% Pycnometer Density(g/cm 3 )

33 19 Photoresist Selection The photoresist utilized was SU from Microchem. This photoresist was selected for its wide use in the photolithography industry. SU-8 was also chosen for its high viscosity allowing thick coatings to be created on a substrate with only one spin coating. This photoresist also has a high transparency to ultra violet light, which is essential when working with a powder filled photoresist. This high transparency allows for components to be created from higher solids loadings and made to thicker dimensions. Basic Process This process of creating micro components uses a combination of powder metal and photolithography. In this process a photoresist (Microchem SU ) is used to shape components and to act as a binder for holding the component shape while handling prior to sintering. Creating components using the filled photoresists techniques involves blending powder with SU-8 photoresist to create a slurry, spin coating the slurry onto a ceramic substrate, soft bake to evaporate solvent and lightly cross-link the photoresist, expose the wanted geometries to ultraviolet light, post exposure bake (PEB) to ensure the photoresist is fully cross-linked, develop away unexposed material and finally debinding and sintering of the microcomponents. This process is depicted in Figure 3-4.

34 20 Metal Powder Photoresist Mixing Slurry Spin Coat Soft Bake Metal Component Sintering Exposure Thermal Debinding PEB Green Parts Develop Figure 3-4: Experimental Process Flow Diagram Spin Coating Experiments The spin coating process is the method of using centrifugal force to evenly spread a liquid over a substrate. In the spin coating step, filled photoresist is placed on the alumina substrate and then the substrate is spun at a predetermined ramp rate through the use of a Laurell Technologies spin processor, seen in Figure 3-5, to achieve an even coating of a targeted thickness. The ramp rate used for spin coating can be seen in Figure 3-6. The first 500 rpm step held for 10 seconds is necessary to ensure the filled photoresist spreads and coats the entire substrate. The second step is the final spin speed which determines the final thickness of the filled photoresist coating. Several experiments were performed at different spin speeds with each solids

35 21 loading and powder in order to determine the final thickness of the filled photoresist coating. Final thickness was determined by first measuring the thickness of the substrate before spin coating and then measuring it again after the soft bake and taking the difference. Figure 3-5: Laurell Spin Processor used in the spin coating process. Figure 3-6: Spin cycle for coating ceramic substrates with filled photoresist. Exposure Experiments The addition of powder to a photoresist greatly reduces ultra violet light transmittance. Because of this reduction in transmittance, it is necessary to determine how far the ultra violet light can penetrate the filled photoresist in order to establish the achievable component thicknesses. To do this, the ceramic substrates were coated with the filled photoresist in thickness

36 22 varying from 30μm to 600μm. The thickness was varied with the Laurell Technologies spin processor in increments of 500 rpm from 500 rpm to 4000 rpm for each powder and solids loading. Following the spin coating samples were soft baked and then exposed to 365 nanometer wavelength ultraviolet light with the OAI exposure system seen in Figure 3-7. Following the usual post exposure bake, all the samples were developed in aqueous SU-8 developer. The exposure depth was then determined by examining the samples and the thickest one that retained the component s features after development was concluded to be the maximum thickness that the ultraviolet light could penetrate. Table 3-3 summarizes all of the samples that were tested. Figure 3-7: OAI ultra violet exposure system used for exposing the filled photoresist.

37 23 Table 3-3: Summary of all samples tested for ultra violet light exposure depth. Powder Size (μm) % Solids Loading Spin Speeds , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, , 1000, 1500, 2000, 2500, 3000, 3500, 4000 Debinding and Sintering The debinding and presintering procedure was performed in a retort (Lindberg model 5686 Retort Furnace). For accuracy, the same debind cycle was used for all components and can be seen in Figure 3-8. The debind process is performed to remove the binder from the green component and also lightly sinter components for ease of handling during the sintering step. In the case of these experiments the binder of the micro components is the photoresist which has to be processed in an oxygen atmosphere to prevent the photoresist from turning to carbon. To do this the retort was sealed and then compressed air was flowed into the chamber. Also the components were created on a ceramic substrate which increased the ease of handling of the

38 24 micro components, because, the components did not need to be touched until they were fully sintered. Due to the ease of handling the components only needed to be debinded to a temperature of 650 C. To control the temperature during the debind cycle a programmable Lindberg digital controller was used. The debind cycle for the micro components is as follows: 5 C per minute ramp up to 650 C and hold for one hour then 5 C per minute ramp back down to room temperature. This cycle can be seen in Figure 3-8. Debind Cycle 650 C 1 hr 5 C/ min 5 C/ min Figure 3-8: Debind cycle for stainless steel filled photoresist micro components Sintering of the micro components was carried out in a CM horizontal alumina tube furnace (CM Furnace model , Bloomfield, NJ) Seen in Figure 3-9. The CM tube furnace has MoSi 2 heating elements located outside of the alumina tube which allows the furnace to reach a maximum temperature of 1650 C. The temperature was controlled by a programmable Eurotherm digital controller. The micro components were placed into the furnace on their ceramic substrates and sintered in 100% hydrogen. To do this the furnace was purged for 30 minutes with nitrogen to evacuate the chamber of all oxygen. After the 30 minute purge the nitrogen was shut off then the hydrogen was turned on. The sintering cycle of these components consisted of a 5 C per minute ramp up to 900 C with a 30 minute hold. This hold was essential for oxide reduction of the components before the final sintering stage where densification takes place. After the hold

39 25 the components were ramped at 5 C per minute to the final temperature of 1240 C and held for 1 hour before being ramped back to room temperature at 5 C per minute. This sintering cycle can be seen in Figure The components were not sintered to their full sintering temperature in order to minimize deformation in the components. Figure 3-9: Picture of CM tube furnace used for sintering components. Sintering Cycle hr 5 C/min 5 C/min

40 26 Figure 3-10: Sintering cycle used for stainless steel micro components. Dimensional Analysis Dimensional analysis was performed on the exposure mask, green components and finally finished sintered micro components. Twenty components of each powder size, solids loading and sintering temperature were measured. Measurements were performed with a Smart Scope FOV 200, with an accuracy of ± 0.4 µm. Figure 3-11 shows the three dimensions that were measured for each part

41 27 Figure 3-11: Micro components utilized for dimensional analysis. Qualitative analysis A qualitative assessment of the micro components was made using a scanning electron microscope (SEM) and the Smart Scope imaging system. The micro components were examined for defects and feature retention.

42 28 Chapter 4 Results The results from the experiments outlined in the previous chapter are presented here. The first two sections of this chapter show the results of the filled photoresist characterization. The next section focuses on the qualitative analysis of the green and sintered micro components. The remaining sections presented in this chapter are the analysis of the green and sintered components. This analysis includes dimensional variability, density and shrinkage. Filled Photoresist Formation and Viscosity Variation Creation of the filled photoresist slurry was the first step in the micro component forming process. After the slurry was created, the viscosity was characterized at the three different solids loadings for each powder type using a Zahn cup viscosity test following ASTM standard D4212.The Zahn cup that was used was manufactured by Paul N. Gardner Co. EZ Zahn and was a number 5 Zahn cup. Figure 4-1 shows the differences in the viscosities between powder sizes and between solids loadings. As expected a trend appeared, this trend is that as solids loading increases, the viscosity increases. A trend was also noticed that as the powder size decreased, the viscosity increased. The viscosity values measured are listed in Appendix A.

43 Viscosity (Pa-s) 29 Solids Loading Versus Viscosity um Powder -16um Powder -22um Powder Solids Loading ( Volume %) Figure 4-1: Plot of the difference in viscosities between the three different powder sizes and solids loadings. Spin Thickness The first stage in this research was to characterize the powder filled photoresist. It was first characterized for spin coating thickness at four different solids loadings, 35, 40, 35, and 50 volume percent, for the three powder types. The filled photoresist was then spun at speeds varying from 500 to 4000 revolutions per minute in 500 revolutions per minute increments. As expected, an increase in solids loading led to an increase in the coating thickness. Also, a smaller particle size at the same solids loading yielded an increase in spin thickness. Figures 4-2 to 4-4 show powder particle size versus spin thickness, and Figures 4-5 to 4-7 show solids loading versus spin thickness.

44 Thickness (μm) Thickness (μm) 30 Powder Size Versus Spin Thickness (45% Solids Loading) Micrometer -16 Micrometer -22 Micrometer Spin Speed (RPM) Figure 4-2: Effect of powder size on spin thickness at 45% solids loading. Powder Size Versus Spin Thickness (40% Solids Loading) Micrometer -16 Micrometer -22 Micrometer Spin Speed (RPM) Figure 4-3: Effect of powder size on spin thickness at 40% solids loading.

45 Thickness (μm) Thickness (μm) Powder Size Versus Spin Thickness (35% Solids Loading) Micrometer -16 Micrometer -22 Micrometer Spin Speed (RPM) Figure 4-4: Effect of powder size on spin thickness at 35% solids loading. Solids Loading Versus Spin Thickness (-5 μm Powder) % Solids Loading 40% Solids Loading 45% Solids Loading Spin Speed (RPM) Figure 4-5: Effect of solids loading on spin thickness at 45% solids loading.

46 Thickness (μm) Thickness (μm) Solids Loading Versus Spin Thickness (-16 μm Powder) % Solids Loading 40% Solids Loading 45% Solids Loading Spin Speed (RPM) Figure 4-6: Effect of solids loading on spin thickness at 40% solids loading. Solids Loading Versus Spin Thickness (-22 μm Powder) % Solids Loading 40% Solids Loading 45% Solids Loading Spin Speed (RPM) Figure 4-7: Effect of solids loading on spin thickness at 35% solids loading.

47 Exposure Depth 33 To continue the characterization of the filled photoresist, it was necessary to determine how far 365 nanometer ultraviolet light could penetrate the filled photoresist. To do this the ceramic substrates were coated with the filled photoresist in thickness varying from 30μm to 600μm. The thickness was varied with the Laurell Technologies spin processor in increments of 500 rpm from 500 rpm to 4000 rpm for each powder and solids loading. Following the spin coating samples were soft baked and then exposed to 365 nanometer wave length ultraviolet light with the OAI exposure system. Following the usual post exposure bake, all the samples were developed in aqueous SU-8 developer. The exposure depth was then determined by examining the samples and the thickest one that retained the component s features after development was concluded to be the maximum thickness that the ultraviolet light could penetrate. This too was performed with the three different solids loadings and three powder particle sizes. The 50 volume percent solids loading was left out because the ultra violet light was unable to penetrate any of the powder sizes. Figure 4-8 shows the effect that solids loading has on the exposure depth. It was found that the higher solids loadings greatly diminished the ability of the ultra violet light to penetrate the filled photoresist which was expected. This was expected because as the solids loading increases, the space around the particle occupied by the photoresist decreases which decreases the ability for light to penetrate the mixture. It was found that the 45 volume percent solids loading with the -5 µm powder did not allow adequate light penetration to form any components. In Figure 4-9 it can be seen that as the powder particle size decreases the exposure depth also decreases. This was also expected because of the particles ability to pack better and form a tighter matrix of particles allowing less light through.

48 Exposure Depth (um) Exposure Depth (um) 34 Solids Loading Versus Exposure Depth um Powder -16um Powder -22um Powder Solids Loading (%) Figure 4-8: Effect of solids loading on exposure depth at the three powder sizes. Notice that the -22 µm powder has a smaller depth at 35% solids loading, this is due to the fact that the thickest achievable coating of filled photo resist was 205µm which limited the thickness at which it could be tested. Powder Size Versus Exposure Depth % Solids Loading 40% Solids Loading 45% Solids Loading Powder Size (um) Figure 4-9: Effect of powder size on exposure depth at the three solids loadings. Notice that the 35% solids loading has a smaller depth at the -22µm, this is due to the fact that the thickest achievable coating of filled photo resist was 205µm which limited the thickness at which it could be tested.

49 Qualitative Analysis 35 The fabricated filled-photoresit micro components were qualitatively analyzed based on surface defects and feature resolution. After the development step of the forming process, defects were visually observed on the surface of the components. Examples of these defects can be seen in Figures 4-10 to The flaws seen in the figures are pores as well as agglomerations of powder from imperfections in the mixing procedures. It was noticed that solids loading had an effect on the surface of the components. An increase in solids loading lead to diminishing amounts of apparent surface defects. All three powder sizes followed this forming trend. The 50 volume percent components showed the least amount of defects. The defects associated with the agglomerations of powder on the surface could be due to the mixing procedure of the Su-8 and powder. Due to time and practicality restraints, the Su-8 and powder were simply mixed together by hand. Ball milling would have been the proper process for the mixing of the Su-8 and powder to ensure they were properly mixed and there were no agglomerations. However, ball milling 27 separate batches of su-8 and powder was not feasible. Green component features were imaged using a Smart Scope. The qualitative analysis of the green components included all of the different variations formed: 35 and 40 volume percent -5µm stainless steel 316L, 35, 40 and 45 volume percent -16µm stainless steel 316L and 35, 40 and 45 volume percent -22µm stainless steel 316L. Qualitative analysis was performed by counting the number of defects seen in each component. Figures 4-10 to 4-12 are of features of the green micro components. The images are separated by powder size and start with the lowest solids loading at the top of the figures and end with the highest solids loading at the bottom of the figures. All the figures show the trend that as solids loading increases feature resolution and retention increases, also the amount of defects decreases with increasing solids loading. A qualitative analysis was also performed on the micro components after sintering. When the 35 volume percent -22μm powder components were removed from the furnace it was noticed

50 36 that the components were cupped. They were cupped into and away from the sintering substrate, which rules out sintering as a cause of the cupping. If sintering were the cause, all the components would have cupped in the same direction. This indicates that there was an issue with the forming process. The features of the components were also imaged for qualitative analysis. This analysis included all of the components that were formed, except the 35 volume percent -22μm powder components due to the cupping. Figures 4-13 to 4-15 are of features of the sintered micro components. The images are separated by powder size and start with the lowest solids loading at the top of the figures and end with the highest solids loading at the bottom of the figures. Again a trend was noticed related to solids loading, where an increase in solids loading lead to better resolution and feature retention as well as fewer defects. Powder size also seemed to play a role, as the -5μm powder displayed the best parts.

51 Figure 4-10: Images prior to sintering of 35% and 40% solids loading -5μm SS 316L green components. Forming defects are more prevalent in the lower solids loading component. Pictures are ordered with the lowest solids loading at the top and highest at the bottom. There Is not a 45% component because the 45% filled photoresist was not able to be penetrated by the UV light. 37

52 Figure 4-11: Images prior to sintering of 35%, 40% and 45% solids loading -16μm SS 316L green components. Forming defects are more prevalent in the lower solids loading component. Pictures are ordered with the lowest solids loading at the top and highest at the bottom. 38

53 Figure 4-12: Images prior to sintering of 35%, 40% and 45% solids loading -22μm SS 316L green components. Forming defects are more prevalent in the lower solids loading component. Pictures are ordered with the lowest solids loading at the top and highest at the bottom. 39

54 Figure 4-13: Images after sintering of 35% and 40% solids loading -5μm SS 316L components. Figure 4-13 Images after sintering of 35% and 40% solids loading -5μm SS 316L Forming defects remained after sintering. Notice in the top picture of the 35% solids loading component that the defects are more prevalent than in the 40% solids loading component below it. There Is not a 45% component because the 45% filled photoresist was not able to be penetrated by the UV light. 40

55 Figure 4-14: Images after sintering of 35%, 40% and 45% solids loading -16μm SS 316L components. Forming defects remained after sintering. Notice in the top picture of the 35% solids loading component that the defects are more prevalent than in the 40% and 45% solids loading components below it. 41

56 Figure 4-15: Images after sintering of 40% and 45% solids loading -22μm SS 316L components. Forming defects remained after sintering. Notice in the top picture of the 40% solids loading component that the defects are more prevalent than in the 45% solids loading component below it. There is not a 35% component because they were to severely warped to form components after sintering. 42