Demouldability of microstructures in polymer moulding · 2007-05-04 · Demouldability of microstructures in polymer moulding A De Gravea, T. Eriksson , H.N. Hansena a Department

Demouldability of microstructures in polymer moulding A De Gravea, T. Erikssona, H.N. Hansena

a Department of Manufacturing Engineering and Management, Technical University of Denmark, DK-2800 Kgs.

Lyngby

Abstract Demoulding micro structured surfaces and micro parts is an important issue in replication of micro technology based components. In this paper a state of the art on ways to improve demoulding of microstructured parts both in polymer moudling and metal injection moulding is presented. The approach is described and simulation methods are shown in relation to previously existing studies. The approach chosen is the investigation of the influence of the demoulding angle on the demouldability of polymer mouldings. The design is based on a number of micro sized cones with different slope angle placed on a flat metal surface. A mould design is proposed and simulated. The design is based on a number of micro sized cones with different slope angle placed on a flat metal surface. A method for characterization of the mould using Laser Scanning Confocal Microscopy (LSCM) and Atomic Force Microscopy (AFM) to verify the slope angle and the straightness quality of the slope is discussed. Initial mould filling simulations using a 2½ D mesh show that a real 3D simulation were performed.

Keywords: demoulding of microstructures, polymer moulding, multi-scale simulation

1. Introduction Demoulding of parts with surface microstructures can cause difficulties since the relatively large forces might destroy or damage the micro structures. This is a problem both for moulding of polymers as well as metals. The problems arises partly from the fact that the relevant manufacturing techniques, especially injection moulding, are made at an elevated temperature resulting in a dimensional shrinkage when the part cools. The shrinkage pushes the moulded material against those parts of the mould sticking out. This is normally not a problem in macro sized manufacturing but since micro structures by definition are in the micro range, the moulded structures or sections are sensitive to high forces. Friction forces and sticking effect also influence the demoulding. In this paper a state of the art on ways to improve demoulding of microstructured parts both in polymer moudling and metal injection moulding is presented. The approach is described and simulation methods are shown in relation to previously existing studies. 2. State of the art Beginning with macro scale demoulding, Hopkinson et al. [1] investigated the injection forces in the AIMTM process using two different ejection methods; i.e. convectional ejector pins and using a conformal ejector pad. In general, ejector pins are positioned in a way to allow a clean ejection from the mould without damaging the moulding. Pad ejection is quite similar except that whole areas of the part are pushed by the pad instead of scattered points with pins. This gives a more even ejection and is normally preferred for larger parts which can be damaged by pins. The authors found that the ejection forces required were higher when using an ejector pad instead of ejector pins. They give two possible explanations. A possibility is that air is not able to fill

the void between the mould and the moulding at the early stages of ejection when using an ejection pad. When using pins, it is possible that the moulding bends while still on the core thus allowing air between the mould and the moulding. This would reduce the ejection force. Another possible reason is that the ejection profile using pads is steeper than the ejection profile using pins. Thus the moulding is pushed off at a faster rate when using pads giving a higher acceleration requiring a higher force. Worgull et al. [2] attempted to suggest a few modifications to the hot embossing process in order to reduce shrinkage of the moulded parts which has the largest influence on the demoulding forces. The suggestions are based on simulations. Two suggestions are made to improve the demoulding. Primarily, shrinkage and warpage can be reduced by using a frame around the microstructures. The frame is a boundary which gives an even pressure distribution in the melt resulting in less warpage and an even shrinkage which would lower the demoulding forces. Secondly, the authors suggest to add additional sacrificial microstructures at the very edges of the mould. It is shown that the contact stress decreases significantly inside the outermost microstructures thus lowering the demoulding forces. In an attempt to lower the adhesion and friction between moulding inserts and the thermoplastic material during the production of high aspect ratio microstructures, Peng et al. [3] compared the properties of demoulding of Ni and Ni-PTFE inserts. The moulding inserts are manufactured by electroforming. In order to create mould inserts with a lower surface friction, PTFE is added to the nickel galvanic bath. The inserts are manufactured with a gear pattern with a diameter of about 200 µm and a height of about 16 µm, The inserts are then used to carry out Hot Embossing experiments. The authors conclude that the Ni-PTFE inserts perform better than the Ni inserts due to their small frictional coefficients and low average surface energy. However, it is the belief of the authors of this paper that it is impossible

to achieve an even distribution of PTFE in the surface of the mould insert. Furthermore, the PTFE wears significantly faster than the Ni, which means that the lower friction material would disappear after some use. This means that this procedure is not suitable for industrial production. Michaeli et al [4] suggested a few alternative demoulding concepts in order to improve the demouldability of parts with surface microstructures. The demouldability of a part with surface microstructure depends both on its geometry, the material used and the nature and position of the force applied. The authors suggest new demoulding principles in which the demoulding force is applied evenly over the microstructures thus minimizing the forces acting to damage the microstructures. The demoulding principles were based on even break out of the structures, demoulding using vacuum and demoulding using ultrasound. The authors could not show any positive results for vacuum or ultrasound demoulding. However, it was found that precision cavity retraction was a successful method. 3. Approach The focus in this work is to investigate the effect of demoulding angles on the demouldability through the following approach:

• simulation of shrinkage in moulding and subsequent FEM analysis of the mechanical demoulding process to obtain likely places of breakage and deduce critical demoulding angles

• manufacturing and characterisation of a injection moulding insert with test structures

• characterisation of moulded parts and comparison with simulations

A mould with features in the form of pillars ranging from cylinders to cones is being manufactured. The idea is not to work on the optimization of replication but to focus on the specific demoulding step. Hence the aspect ratio is not higher than 2 with a depth of 100 microns, to avoid having trouble filling all the cavities. Two different materials are envisioned for the tests: PC and COC. PC is widely used in the CD and DVD market, COC is used for biofluidics and opto-MEMS. The principal design of the mould is shown in Fig.1. and Table 1. The geometry consists of a set of cones with 3 different aspect ratio (1, 1/2 and 2), the depth is kept constant at 100 µm thus giving a base diameter of 100, 200 and 50 µm. 5 different top angles (0, 1, 2, 3, 4 degrees) are used. For statistical reasons, 15 lines of ten identical cones are manufactured on a 3 mm thick plate with the dimensions 20 x 20 mm. Fig 2. shows the mesh but also shows that the structures are placed apart in order to avoid interferences, especially in terms of stress concentration at the bottom of the cones. Table 1. Summary of the geometrical features diameters (in µm) 100 50 200

aspect ratio 1 2 0.5

angles (in degrees) 0 1 2 3 4

Fig.1. geometry of the test microstructures

Fig.2. Example of mesh on two different aspect ratio cones (aspect ratio shown: 1 and 0.5) This design poses some issues. In particular, the making of the mould will be a challenge in itself and it is not sure that the degree of granularity required for the angle can be achieved. For example, in the case of a 1° cone with a diameter of 50µm and a height of 100 µm, the top diameter should be 96.51 µm. The manufacturing process used to obtain the mould is the micro-EDM milling technology. In theory, this technology has a resoution accuracy of 0.1 µm. This means that the slope from the top to the bottom of the cylinder will be manufactured in horizontal steps of 0.1 µm. Although, the EDM process will provide a continuous slope and not stairs due to its physical principle. Another issue is how to characterize the mould. A Laser Scanning Confocal Microscope (LSCM) can be used to measure the top and bottom of the cylinder. The confocal laser scanning microscope builds a 3D topography image by scanning a number of xy-planes at different z-level (optical slices). Only the surface features in focus is included from each z-level. The 3D image is then built from the different z-levels (optical slices). To ensure that only light in focus reaches the detector; a pin-hole is located on the focal plane and only accepts light from objects very close to the focal plane of the surface. Using this technology the average angle can be calculated. Atomic Force Microscopy (AFM) can be used to measure a profile of the slope from the top to the bottom of the cylinders. The AFM consists of a microscaled cantilever with a sharp tip (probe) at its end that is used to scan the specimen surface.

Typically, the deflection is measured using a laser spot reflected from the top of the cantilever into an array of photodiodes. Other methods that are used include optical interferometry, capacitive sensing or piezoresistive AFM probes. The information resulting from these two measurements will properly characterise the mould. The characterisation of the mould is important to be able to verify the influence of the cylinder angle on the demoulding. LSCM will be used to analyse the demoulded parts. The LSCM allows for a good three dimensional representation of the mouldings making it possible to detect any damaged structures. The polymers will be coated with aluminium to avoid difference in reflectability making it possible to compare the two materials. 4. Simulations The goal of the simulations is to analyse the forces that sticks the polymer to the mould. The shrinkage of the polymer during moulding forces the polymer towards the side walls of the structures. There is thus a stress concentration on the structures from the polymer. Furthermore, there is the friction between the polymer and the mould as well as the sticking effect. One of the challenges with such a simulation is that two different length scales are at play: the part in itself (i. e. the mould) is macro (in the order of tens of millimetres) whereas the features are micro (in the order of hundreds of microns). One way to cope with that problem is described by Eriksson et al. [5] by passing information from a macro-level simulation to the micro-level applied as local boundary conditions. Although these simulations are for the filling of polymer only, the principle is interesting. In [2], a two steps simulation approach is presented, applied to the hot embossing process of PMMA. Even though hot embossing and injection moulding differs in many aspects, the simulation approach used is of interest. Moldflow is used for the filling of the mould, using the injection compression moulding option. Then the shrinkage results are used for a mechanical simulation in ANSYS. The results show the contact stress between mould and microstructures during the demoulding step, they amount up to 2.86 MPa. A similar approach has been reported by Fu et al. [5] in the field of micro metal injection moulding. Stress distribution during cooling and demoulding of an array of micro pillars was studies using a finite element analysis. The result is an analysis of stress distribution during cooling and during the ejection step, based on time. It shows that the most likely time for breakage is during the onset of the ejection and in the section that is farthest away from the centreline of the mould. The purpose with the simulations in this work is to be able to simulate the demoulding procedure and compare it to the experiments. The simulations will be made in two steps:

1. Using Moldflow or similar softwares to get information about the shrinkage and/or warpage of the part.

2. Perform a demoulding simulation in a mechanical FEM software (such as for

example FEMLAB) using information from the filling simulations.

Up to this point, the polymer filling simulations have been initiated. The simulations have been performed using Moldflow using a shell (2 1/2D) simulation taking filling, packing and cooling phases into account which result in information about the shrinkage and warpage of the polymer. The material used for the simulation is standard Polycarbonate (PC). The displacement in z or x (principal plane of the part) seen in Fig.3. has a maximum of 60 µm (on the edges) and in y (normal to the plane) a maximum of 6 µm. These results shows the limitation of the method. To get an accurate input for further mechanical simulation, a real 3D mesh will be necessary. It is the only way to obtain the deformation of the geometry of the cylinders.

Fig.3. Shrinkage in the x direction (parallel to the

principal plane) maximum value is 60 um

Fig.4. Displacement shown at one of the edges, dimmed structures represent structures before

shrinkage

First of all, we can not be sure that the presence of the surface features affects the output of the simulation. Nonetheless the displacements obtained in x and z are in the order of the feature size, as seen on Fig.4. Therefore the pillars are not likely to be placed within whatever the tolerance will be. Thus, the simulations shows that there will be mechanical constraints due to the force applied between the polymer and the cones on the mould. 6. Conclusions and perspectives A design for testing demouldability has been proposed and simulated. The design is based on a number of micro sized cones with different slope angle placed on a flat metal surface. The mould will be used to perform experiments to investigate the demouldability. Simulations will be performed to verify the results of the experiments as well as obtain information on the demoulding process. The simulations will be performed using a polymer filling software to obtain information on the shrinkage and warpage of the polymer. This information will be used in the mechanical simulation of the demoulding. There are some other issues that will have to be taken into consideration:

• Is there a dimensional variation of the mould by thermal expansion to be considered?

• How to implement shrinkage and de-adhesion, what happens if there is a loss of contact between mould and plastic part?

• Is the tribology coefficient the same on the whole cone describing line?

These issues will be addressed during the finalization of this work. Acknowledgements Collaboration from MSc Alberto Gava at DIMEG (Department of Innovation in Mechanics and Management) at University of Padova is acknowledged for performing the numerical simulation. Also, this work was carried out within the framework of the EC Network of Excellence "Multi-Material Micro Manufacture: Technologies and Applications (4M)". References [1] N. Hopkinson, P. Dickens, Study of ejection forces

in the AIM process, Materials and Design 20 (1999), 99-105

[2] M. Worgull, M. Heckele, W. K. Schomburg, Large-scale hot embossing, Microsyst Technol (2005) 12, 110-115

[3] Z. Peng, L. Gang, T. Yangchao, T. Xuehong, The properties of demoulding of Ni and Ni-PTFE moulding inserts, Sensors and Actuators A 118 (2005) 338-341

[4] W. Michali, R. Gärner, New demolding concepts for the injection molding of microstructures, Journal of Polymer Engineering, vol. 26, Nos 2-4, 2006, pp. 161-177

[5] T. Eriksson, H.K. Rasmussen, The effects of polymer melt rheology on the replication of

surface microstructures in isothermal moulding, J. Non-Newtonian Fluid Mech. 127(2005) 191-200

[6] G. Fu, N. H. Loh, S. B. Tor, B. Y. Tay, Y. Murakoshi and R. Maeda, Analysis of demolding in micro metal injection molding, Microsystem Technologies, Vol. 12, 6, 2006