Through-thickness Embossing Process for Fabrication of Three-dimensional Thermoplastic Parts.
INTRODUCTION The standard hot embossing process is designed for imprinting surface features onto polymer substrates , . The substrate is relatively thick as compared with the feature size. During the process a heated embossing master, typically a metallic stamp with to-be-replicated features, is pressed against the substrate, either heated or not heated, for feature transfer. Owing to this simplicity in tool and process setup, hot embossing has been widely used in polymer microfabrication. The process, however, is subjected to some process flaws, greatly limiting its capability. Because of the open-die nature of the process, the polymer squeezes out in the lateral direction during the embossing stage. This is generally considered to be acceptable in replication of low aspect ratio surface features. It may, however, result in incomplete replication when a high embossing pressure is needed, e.g., during embossing of high aspect ratio features. Further, separable or discrete features, e.g., microgears, waveguides, micro thorough holes, among others, are difficult to fabricate. If one defines surface features as 2.5D features, these aforementioned discrete features/parts are indeed more three-dimensional (3D). To emboss such precision 3D features out of a polymer sheet, an appropriate through-thickness action is needed. At the moment, precision 3D parts are mainly produced using precision injection molding. The molding results, however, are often compromised, because of the complex tool setup and the high amount of stresses introduced to the part during injection molding. It is thus advantageous to use hot embossing, a low-stress process with a simpler tool and process setup, for precision fabrication. To date, however, there has been limited effort in adapting the hot embossing process for fabrication of through-thickness features. Heckele and Durand developed a technique for producing through-holes by hot embossing. They used a substrate with two layers of different materials. After embossing, the tool features protruded through the upper layer into the lower layer. After removal of the lower layer, through holes were left on the upper layer. Werner described a process involving identical top and bottom mold halves, both containing pins, whose top surfaces are attached to each other upon mold closure. By this process, through holes, with only a thin residual layer remaining, can be embossed. Mazzeo et al. developed a tool set for punching thin plastic films. They were able to emboss holes as small as 500 [micro]m in diameter. The methods described earlier all involve a through-thickness action for fabricating through holes. Similar ideas may be developed for hot-embossing discrete 3D parts out of polymer sheets. In this study, a hybrid punching and embossing process was investigated for through-thickness embossing of 3D parts. The embossing tool includes a punching head and to-be-replicated features in the socket behind the punching head. The built-in punching head allows for a through-thickness action and provided a close-die environment for embossing pressure buildup. The method was used to successfully emboss multichannel waveguides which require uniform edges and accurate dimensions. [FIGURE 1 OMITTED] EXPERIMENTAL Waveguide Design Millimeter waveguides, because of their low signal attenuation, are widely used in remote sensing applications . Currently, the industry uses metal as the waveguide material and relies on precision machining for waveguide fabrication. The resulting waveguides are expensive, thus demanding new waveguide materials and their net-shape manufacturing processes. Polymers, for their versatile properties and mass-production capabilities, would be promising materials in such applications . In this study, a multichannel waveguide, as shown in Fig. 1, was chosen. The original design was a shell structure (Fig. 1a). The channel dimensions were determined using a commercial finite element method (FEM) software package, Ansoft HFSS[TM] from Ansoft Corporation, to achieve a cut-off frequency between 40 and 75 GHz. To facilitate manufacturability, this geometry was modified into a two-piece assembly comprising a grooved disc (Fig. 1b) and a solid disc, which can be electroplated and assembled together to form sealed grooves. Mold Design and Fabrication The hot-embossing mold for the waveguide comprises a mold insert assembly and a pair of hot and cold plates, attached to a platen set. The mold insert was designed as an assembly of three mating pieces: a punching cutter, a shaper, and a disc spacer, as shown in Fig. 2. The three-piece design of the mold was adopted to achieve (1) complete fill of the mold cavity during the embossing process, (2) a mold with sharp edges and tight dimensional tolerances, (3) easy cleaning of residual polymers after embossing, and (4) evacuation of trapped air during the embossing process. [FIGURE 2 OMITTED] [FIGURE 3 OMITTED] The critical component in the mold insert assembly is the shaper, which is designed as the negative pattern of the geometry to be embossed. Slots cut at the end of the four arms assist in the mating of the shaper with the holder. Close dimensional tolerances were employed during machining of the shaper, the shaper holder, and the disc spacer to prevent leakage of the polymeric material during the embossing process. The contour of the shaper was produced using micro electrical discharge machining with a 50 [micro]m diameter tungsten wire. The shaper holder is necessary for aligning the shaper and providing open channels on the embossed part. The shaper holder and the punching cutter are a combined single component in the present design. The polymer blank is cut along the circumference of the circular cutter as the punching head move down during the embossing process. Full mating of the shaper and the shaper holder creates a cavity of 2-mm thickness below the shaper, used to form the base of the waveguide during the embossing process. The disc spacer prevents the polymer leaking from the shaper holder during embossing. Standard hot embossing often utilizes vacuum to assist in complete fill of cavities. This, however, could result in more complicated mold designs. The use of vacuum during waveguide embossing was eliminated by employment of a small clearance on the order of 5 [micro]m between the spacer and the shaper. This small clearance creates a path for the trapped air to escape. The assembled mold insert was fastened to a hot plate with heating and cooling elements embedded, as shown in Fig. 3. A precision platen set was used to mount the hot and cold plates, as shown in Fig. 4. The hot plate was mounted to the top platen, the moving platen, and the cold plate to the bottom platen, the fixed platen. Thermal insulation composites were inserted between the plates and the platens. Embossing Setup A pneumatic press was modified and used as an embossing apparatus. The stationary platen of the platen set was fastened to the base of the pneumatic press by means of a lock screw. The moving platen was connected to the piston of the pneumatic press by means of a threaded stainless steel stud. The daylight opening between the mold and the base plate was set to 12.5 mm by adjusting the threaded stud connecting the platen and the piston. This was done to ensure that there was no contact between the mold and the polymer substrate prior to the embossing process. The stroke length of the piston was adjusted to achieve complete fill of the mold cavity and control of the cutter position during the embossing process. The stroke length of the piston can be adjusted by varying the position of the piston lock screws at the top of the cylinders. The optimum stroke length for the embossing process was calculated by trial and error. The fabrication process consists of a series of sequential operations, i.e., mold preheating, embossing, cooling, mold opening, and finally part ejection from the mold, as illustrated in Fig. 5. Before embossing, a machined polymer slab, 75 x 37.5 x 6.4 [mm.sup., was placed on the lower plate, and at the same time the top plate was heated to reach the designated embossing temperature. The embossing stage was commenced by activation of the optical switch and movement of the ram on the pneumatic press, resulting in contact of the cutter with the polymer substrate. It should be noted that the heated cutter causes the polymer to soften, thus assisting in the cutting action. Further movement of the punching cutter results in heating, softening, and then squeezing of the polymer billet inside the mold cavity. The embossing stage was constant force controlled. After a designated embossing period elapsed, the heated top plate was cooled by circulating with tap water. At the end of the cooling cycle, the embossing force was released. The embossed waveguide was attached to the polymer substrate at the circumference with a ring in a thickness of 0.5 mm, similar to a ring gate in injection molding. The waveguide was manually ejected from the mold by cyclic loading on both sides of the polymer blank. Care was taken to ensure that the vibrations were within the elastic range of the material to prevent any damage to the embossed part. The ring connector was mechanically trimmed off to disconnect the embossed waveguide from the polymer sheet. In production, the cutter could cut all the way throughout the entire blank thickness, thus eliminating the trimming step. This, however, requires a more sophisticated ejection mechanism and redesign of the mold, and was therefore not investigated in this study. [FIGURE 4 OMITTED] [FIGURE 5 OMITTED] EMBOSSING RESULTS AND DISCUSSION ABS was chosen as the polymer for through-thickness embossing of the multichannel waveguide. The polymer has a glass transition temperature ([T.sub.g]) at 105[degrees]C. Design of experiments was carried out to optimize the major process parameters, including tool temperature and embossing force, so as to obtain complete replication of the waveguide. The optimum embossing condition was found to be 140[degrees]C (embossing temperature) and 4000 N (embossing force). The waveguides fabricated under these conditions demonstrated complete mold fill, as well as sharp edges and smooth surfaces along the channels, as shown in Fig. 6. The entire embossing stage, starting from the contact between the cutter and the polymer and ending at the beginning of the cooling stage, lasted for 3 min. The constant force control scheme, rather than constant speed control, is considered advantageous in the present application. In the constant force case, the softening process and the deformation process are automatically synchronized in a creeping mode; that is, deformation carries on as the polymer at the contact softens. This helps protect the cutter's blade, on one hand, and create a stable process, on the other hand. The small contact area between the cutter blade and the polymer, as opposed to a much larger contact area in the standard embossing process, helps maintain a small deformation zone on the polymer substrate. The nonisothermal embossing setup, rather than an isothermal setup in the standard embossing process, is also considered beneficial. This is understandable since the material away from the cutter is maintained below the glass transition temperature, thus reducing the amount of squeezing flow toward the surrounding area. At the end of the embossing stage, tap water was circulated inside the heated plate, while the same embossing force was applied on the tool to produce a holding pressure. The force was removed when the heated plate was cooled to about 70[degrees]C, well below the [T.sub.g] of the polymer. The total cycle time was 7 min. [FIGURE 6 OMITTED] [FIGURE 7 OMITTED] The embossing force and temperature were found to have a profound influence on the quality of the embossed waveguide. To study such effects, the embossing process was carried out at varied embossing temperature and force. At a lower embossing temperature, e.g., 120[degrees]C, complete fill of the mold cavity was not obtained even with the application of a higher embossing force, e.g., twice higher, as shown in Fig. 7a. At a higher embossing temperature, e.g., 160[degrees]C, incomplete fill also occurred, as shown in Fig. 7b. The former case is easy to understand considering the increased resistance, i.e., increased viscosity, of the polymer to deformation at a lower temperature. The latter may be attributed to the increased lateral flow to the adjacent area, thus lowering the pressure in the cavity. This increased outflow was actually observed in the experiment. [FIGURE 8 OMITTED] The effect of the embossing force was also investigated. The embossing temperature was kept at 140[degrees]C while the embossing force was varied. Lower embossing forces, e.g., 3000 N, resulted in surface irregularities and incomplete fill of the post on the waveguide, as shown in Fig. 8. Particularly, a large void was observed on the waveguide surface. These surface irregularities suggest that the waveguide was not sufficiently packed during the embossing and holding stage. Similar parametric studies were conducted for PMMA. For PMMA, complete cavity fill was achieved at a mold temperature of 180[degrees]C and an embossing force of 4000 N. Similar to the ABS case, considerably lower or higher embossing temperature than the optimum temperature resulted in incomplete cavity fill. The PMMA waveguide was found to be more difficult to eject and often fractured during ejection. The increased ejection difficulty can be accredited to the relatively brittle nature of PMMA as compared to ABS, a toughened copolymer. MODELING OF THROUGH-THICKNESS EMBOSSING The through-thickness embossing process that involves a punching mechanism as described earlier is highly nonisothermal. The process starts from a line contact between the tool and the polymer. The polymer that contacts with the punching cutter is instantaneously heated, creating a sharp temperature gradient near the cutter edge. This, in turn, creates a large gradient in viscosity, as polymer's viscosity is extremely sensitive to thermal changes. The large gradient in viscosity results in a localized deformation zone surrounding the cutter edge. From previous investigations in nonisothermal embossing processes , , such localized deformation greatly influences the cavity filling process. Since a hot mold is employed in standard hot embossing, it is difficult to produce frozen partial fills experimentally. This is different from injection molding. In injection molding, partial cavity fills can be readily created, because the polymer in contact with a cold mold instantaneously freezes when flow is stopped. To study the thermomechanical changes during the through-thickness embossing process and generalize the findings for other process conditions, a thermal flow model for the process is needed. Hot embossing involves deformation of polymer near the glass transition temperature. Rheological behavior at such a meso-temperature range is quite complex. Considering the long embossing stage, 3 min long, in the waveguide embossing process as described earlier, one may adopt a creep flow model. As such, a viscous material model with a temperature shift factor can be used to approximate the more complex rheological properties. Further, the deformation rate during the prolonged embossing stage is estimated to be low, and thus one may drop the strain rate dependence in the viscosity model. The thermal creep flow model used in the waveguide embossing simulation solves the simplified conservation equations, as follows: [nabla] x [[nu].[bar]] = 0; (1) -[nabla]p [nabla] x [[eta]([nabla][[nu].[bar]] [nabla][[nu].bar.sup.T])] = 0; (2) [rho][c.sub.p] ([[partial derivative]T/[partial derivative]t] [[nu].[bar]] x [nabla]T) = [nabla] x (k[nabla]T) [eta]([nabla][[nu].[bar]] [nabla][[nu].bar.sup.T]): [nabla][[nu].[bar]]; (3) where [[nu].[bar]] is a velocity field, T is temperature, [eta] is viscosity, [rho] is density, [c.sub.p] is specific heat, and k is thermal conductivity. The inertia and body force effects are neglected in the momentum equation. This is justified by the high viscosity of the polymer near the glass transition temperature. The temperature dependency of viscosity is modeled using the following equation: [eta](T) = [eta]([T.sub.g])exp[-[[[C.sub.(T - [T.sub.g])]/[[C.sub. T - [T.sub.g]]]], (4) where [C.sub. and [C.sub. are material constants in the temperature shift factor. Below [T.sub.g], an infinite viscosity was assigned to the polymer. The representative material parameters for ABS (Moldflow Plastics Insight[R], Moldflow Corporation) were chosen for the embossing simulation, as listed in Table 1. A similar model was used previously by the authors in simulating a nonisothermal microlens embossing process and was able to reasonably predict the embossing behavior . Compared with the microlens embossing process, the through-thickness embossing process in this study involves much larger deformation, particularly in the vicinity of the punching cutter. Thus, special care was taken to ensure convergence of the solution. Specifically, an adaptive meshing method was used to remesh the geometry in the vicinity of the contact at each time step. A small radius was also added to the sharp cutter edge to improve convergence of the solution. The geometry and boundary conditions are shown in Fig. 9. An axisymmetric geometry was used to approximate the more 3D geometry of the waveguide. The flow boundary conditions are: BC.1 (axisymmetry), BC.2 ([[nu].sub.n] = 0 and [f.sub.s] = 0, where [[nu].sub.n] is normal velocity and [f.sub.s] is tangential stress), BC.3 (periodic symmetry), BC.4 (free surface with contact detection), and BC.5 (mold surface to be contacted). The thermal boundary conditions are: BC.1, BC.2, and BC.3 (zero heat flux), BC.4 (zero heat flux before contact but convective after contact is detected), and BC.5 (constant mold temperature). After contact is detected on BC.4, the convective heat transfer coefficient is equal to the thermal contact conductance between the polymer and the mold. Three regions, R.1, R.2, and R.3, are labeled in the figure to study the flow behavior in these different regions. The cutter edge is denoted as Point C. [FIGURE 9 OMITTED] [FIGURE 10 OMITTED] The above model was implemented using Polyflow[R], a commercially available FEM software package for polymer processing from Fluent Corporation. SIMULATION RESULTS AND DISCUSSION Isothermal through-thickness embossing was simulated first. In this case, both mold and polymer were set to a constant temperature at 140[degrees]C. The embossing time was set to 180 s. Figure 10a shows the simulated flow pattern. The normalized time in the figure is defined as [.t] = t/[t.sub.p], where t is the actual time and [t.sub.p] is the processing time. Complete fill of the waveguide geometry is not achieved in this isothermal case, even when the cutter cuts through the whole substrate thickness. This filling difficulty may be explained by the unique flow mechanism in isothermal through-thickness embossing. At the initial embossing stage, e.g., [.t] = 0.25, the cutter contacts with the polymer and leaves an indentation on the polymer. After that, significant flow occurs in all three regions, i.e., R.1, R.2, and R.3. The squeeze flow results in gradual filling of R.1 and R.2, but also increases the thickness of the substrate, i.e., flows into R.3. By the end of filling, the substrate has experienced a thickness increases above 15% in R.3. This significant outflow surrounding the waveguide cavity causes a difficulty in pressure buildup and consequently incomplete fill of the cavity. One could study if changes in process conditions, e.g., changes in the isothermal temperature and embossing speed would help with the cavity filling process. However, these adjustments do not improve the percentage of cavity fill in the isothermal embossing case. This can be theoretically proved by rewriting the momentum equation in a process condition independent format: -[nabla][.p] [nabla] x [([nabla][.v.[bar]] [nabla][.v.[bar].sup.T])] = 0, (5) where [.v.[bar]] is defined as [.v.[bar]] = [[partial derivative][x.[bar]]]/[[partial derivative][.t]] = [t.sub.p][[nu].[bar]], and [.p] is defined as [.p] = [[t.sub.p]p]/[[eta](T)]. From Eq. 5, it can be seen that the material is at the same position, [x.[bar]], at a given [.t]. Therefore, the same degree of cavity fill is obtained at the end of embossing, i.e., at [.t] = 1. In the nonisothermal case, the degree of cavity fill is expected to be dependent on process conditions. In this case, the variation of the thermal field results in a gradient in the viscosity, and therefore Eq. 5 is not valid. Two nonisothermal embossing cases were simulated. The interfacial conductance between the mold and polymer was taken to be infinite. This represents a perfect thermal contact with zero thermal resistance. The simulation results were compared with those in the isothermal case. The comparisons are shown in Figs. 10-12. Figure 10 compares the advancement of the flow front under different thermal conditions. The mold temperature was set to 140[degrees]C for all three cases, but the initial temperatures of the polymer were set, respectively, at 140, 120, and 100[degrees]C. For the two nonisothermal cases, one has an initial temperature above [T.sub.g] and the other below [T.sub.g]. It can be seen that, as the polymer temperature is decreased, the degree of fill at a given time instant is improved. At a polymer temperature of 120[degrees]C, the cavity can be completely filled at [.t] = 0.93. As compared with the isothermal case, the squeeze flow is more confined at the tool-polymer contact. There still exists outflow surrounding the cavity, but the outflow is more confined in the vicinity of the cutter. The increase in substrate thickness is around 5% at the end of complete cavity fill. This substantially reduced outflow can be used to explain the increase in cavity fill in this nonisothermal case. As the polymer temperature is further reduced to 100[degrees]C, below [T.sub.g], the above difference becomes more drastic. The time for complete cavity fill further reduces to [.t] = 0.65. The outflow in this case is more confined, and essentially becomes a wall-climbing flow. At the end of the cavity fill, no change in substrate thickness can be detected. It is also seen that a relatively large cushion layer of polymer is retained at the end of the cavity fill. This cushion layer is considered important, as it supplies additional polymer during the holding/cooling stage to compensate for the thermal shrinkage. The above filling results are summarized in Table 2. The vast change in flow pattern can be understood by examining the velocity field during the embossing stage, as shown in Fig. 11. Under isothermal embossing, the velocity field is distributed quite evenly in the entire substrate, indicating a uniform squeezing flow across the entire substrate. At a polymer temperature of 100[degrees]C, the velocity field is confined near the contact, indicating a more localized squeezing flow. [FIGURE 11 OMITTED] Figure 12 provides simulated temperature distributions in the nonisothermal embossing cases. For generality, the temperature was normalized, defined as [.T] = (T - [T.sub.i])/([T.sub.m] - [T.sub.i]), where [T.sub.i] is the initial temperature and [T.sub.m] is the mold temperature. The evolution of the thermal field can be used to explain the unique flow pattern in the nonisothermal embossing case. At the initial embossing stage, [.t] = 0.25, high temperature is localized at the contact point only, thus resulting in a localized flow surrounding the cutter edge. As time elapses, the high temperature zone enlarges, and the flow field inside the cavity becomes more uniform, particularly for the higher initial temperature case. An attempt was made to simulate through-thickness embossing at a lower initial polymer temperature, e.g., T = 25[degrees]C. However, convergence of the solution was not achieved, since an extremely fine mesh size was unaffordable for simulating the extremely high gradient of the field variables near the contact. Nevertheless, one may use the dimensionless temperature obtained in Fig. 12 to estimate the temperature evolution in other cases with a lower polymer temperature. For example, given [T.sub.i] = 25[degrees]C, one can use [.T] = 0.7 to estimate the size of the zone near the contact that has reached the glass transition temperature. In Fig. 12, this corresponds to a zone with a color of red and yellow. This method should offer a quick estimation as long as the process is speed controlled with the same embossing time. In general, the results in Figs. 10 and 11 suggest that, as the polymer temperature reduces, the zone with temperature above [T.sub.g] reduces, thus resulting in a more confined flow field near the polymer-cutter contact, i.e. Point C in Fig. 9. The more confined flow field promotes the flow inside the cavity and consequently results in the complete fill of the cavity. However, it should be mentioned that a reduction in temperature causes an increase in viscosity, and therefore a higher embossing pressure is needed. [FIGURE 12 OMITTED] CONCLUSIONS A four-channel polymer waveguide was fabricated using a single-step through-thickness embossing process. An embossing insert with an integrated punching cutter and embossing shaper was designed and fabricated. Selection of ABS as an embossing polymer, an embossing temperature of 140[degrees]C, an embossing force of 4000 N, and a total cycle time of 7 min resulted in complete cavity fill and wave-guides with sharp, uniform edges. The effects of the tool temperature and the embossing force were studied. At both lower and higher tool temperatures, incomplete cavity fill was observed; the former was attributed to the increased resistance from the material at a lower temperature, given the same embossing force, while the latter was believed to be caused by the increased outflow at a higher temperature. A computer model was set up to study the filling process in the waveguide cavity during through-thickness embossing. It is found that, under isothermal embossing conditions, significant outflow exists, thus resulting in a difficulty in obtaining complete cavity fill. As the difference in temperature between the tool and the polymer increases, the flow becomes more confined in the vicinity of the contact. This reduces the outflow into the surrounding region while promoting localized squeezing flow into the cavity. The nonisothermal embossing setup with the combined punching and embossing tool essentially creates a closed-die embossing environment for filling cavities in a single-step through-thickness action. ACKNOWLEDGMENTS The research was sponsored by Delphi Research Laboratories. DY acknowledges the support of NSF CAREER Award under Grant No. DMI-0503138. REFERENCES 1. M. Heckele and W.K. Schomburg, J. Micromech. Microeng., 14, R1 (2004). 2. H. Becker and C. Gartner, Electrophoresis, 21, 12 (2000). 3. M. Heckele and A. Durand, in The Proceedings of 2001 Euspen's 2nd International Conference, Torino, Italy, May 27-31, 196-198 (2001). 4. M. Werner, "Hot Embossing of Through-Holes in Cyclo-Olefin Copolymer," Diploma Thesis, Technical University of Denmark, Denmark, 2005. 5. A.D. Mazzeo, M. Dirckx, and D.E. Hardt, SPE ANTEC Technical Papers, Cincinnati, May 6-10, 2977-2981 (2007). 6. J. Kraus and D. Fleisch, Electromagnetics with Applications, McGraw-Hill, New York, 456-468 (1999). 7. F. Sammoura, Y.-C. Su, Y. Cai, C.-Y. Chi, B. Elamaran, L. Lin, and L.-C. Chiao, Sensor Actuator A, 129, 270 (2006). 8. D. Yao, V.L. Virupaksha, and B. Kim, Polym. Eng. Sci., 45, 652 (2005). 9. Y.-J. Juang, L.J. Lee, and K.W. Koelling, Polym. Eng. Sci., 42, 551 (2002). 10. D. Yao, P. Nagarajan, L. Li, and A.Y. Yi, Polym. Eng. Sci., 47, 530 (2007). Pratapkumar Nagarajan, (1) Donggang Yao, (1) Thomas S. Ellis, (2) Reza Azadegan (2) (1) School of Polymer Textile and Fiber Engineering, Georgia Institute of Technology, Atlanta, Georgia (2) Delphi Research Labs, Shelby Township, Michigan 48315 Correspondence to: D. Yao; e-mail: