Transcription
INSTITUTE OF PHYSICS PUBLISHINGJOURNAL OF MICROMECHANICS AND MICROENGINEERINGJ. Micromech. Microeng. 14 (2004) 415–422PII: S0960-1317(04)69783-XImplementation and analysis of polymericmicrostructure replication by microinjection moldingYu-Chuan Su1,2, Jatan Shah3 and Liwei Lin1,21Berkeley Sensor & Actuator Center, University of California, Berkeley, CA 94720, USADepartment of Mechanical Engineering, University of California, Berkeley, CA 94720, USA3Department of Mechanical Engineering, University of Michigan, Ann Arbor,MI 48105, USA2E-mail: yuchsu@me.berkeley.eduReceived 30 September 2003Published 17 December 2003Online at stacks.iop.org/JMM/14/415 (DOI: 10.1088/0960-1317/14/3/015)AbstractThis paper presents the adaptation of a conventional injection moldingprocess to the mass replication of polymeric microstructures withappropriate mold design and process control. Using wet-etched siliconwafers with microstructures on the surfaces as mold inserts, we havesuccessfully predicted, improved and optimized the replication results. Theflow behaviors of polymer melts in micro mold-cavities are characterized byboth simulation and experiments. Among various process parameters,temperature is identified as the key factor that decisively determines thequality of injection-molded microstructures. Based on the collectedexperimental and simulation results, process optimization is performed toimprove replication quality and to establish guidelines for potentialapplications. Because of its high speed and low cost, the adaptation of theinjection molding process to microfabrication will lead to a promisingtechnology for MEMS applications.1. IntroductionBecause of their unique properties, polymers have beenincreasingly used in a wide range of applications includingboth macro- and micro-devices. In order to expand thefield of MEMS to polymer-based devices, it is important tointroduce effective techniques for the fabrication of polymericmicrostructures at a low cost and with high precision. In recentyears, a number of technologies for polymeric microstructurereplication have been proposed, including the LIGA process[1, 2] that uses either hot embossing [3] or injection molding[4] to duplicate polymeric microstructures. Using mold insertsfabricated by x-ray lithography, the LIGA process providesthe possibility to manufacture microstructures with arbitrarylateral geometry and high depth for high aspect ratio devicesfrom a variety of materials such as metals, polymers andceramics by various molding processes. Among differentmolding techniques, injection molding is the most prominentone with advantages of low cost and high precision formass production. Successful results for the replication of0960-1317/04/030415 08 30.00polymeric microstructures have been achieved by using specialinjection molding processes [5–12] and conventional CDinjection molding techniques [13, 14]. However, the flowbehaviors of polymer melts in micro mold-cavities are notfully understood. It is believed that due to the large surface-tovolume ratio, surface effects will dominate the flow behaviorat the microscale [15].This paper aims to investigate the flow behavior ofpolymer melt in the micro mold-cavity and determine thenecessary strategies to adapt the traditional injection moldingprocess for the replication of polymeric microstructures. First,the direct application of the conventional injection moldingprocess in the replication of polymeric microstructures isanalyzed using a simulation software C-MOLD [16]. Differentcombinations of process parameters are then simulated toinvestigate the flow behavior of polymer melt, the relationshipbetween process parameters and the quality of moldedmicrostructures. Using these results, the most significantparameters can be identified and possible processing strategiescan be proposed and simulated to test the feasibility. Finally, 2004 IOP Publishing Ltd Printed in the UK415
Y-C Su et alzyzxyPolymer melt2bPressure andmaterial supplyVelocity profileFigure 1. Schematic of polymer melt flowing in a thin cavity.where v̄x and v̄y are averaged velocities over z, and b is halfof the thickness. After several derivative steps, the governingequation for the flow of the polymer melt can be reduced tothe celebrated Reynolds equation: P PS S 0(4) x x y ywhere S is the flow conductance which is defined as b 2zdz.S 0 η(5)The velocities and shear rate can be obtained as b bz1z1z vx xdz1dz1vy yγ ηηηzz(6)these strategies are applied in mold trials to evaluate theirvalidity.where2. Theoretical models x Because most injection molded polymeric parts havecomplicated three-dimensional (3D) configurations and therheological response of polymer melt is generally nonNewtonian and non-isothermal, it is extremely difficult toanalyze the filling process without simplifications. Thegeneralized Hele-Shaw (GHS) flow model introduced byHieber and Shen [17] is the most common approximation thatprovides simplified governing equations for non-isothermal,non-Newtonian and inelastic flows in a thin cavity, as shownin figure 1. The assumptions of the GHS flow model areBecause of the temperature difference between mold andpolymer melt and the viscous heating inside the flow, the fillingprocess should be treated as a non-isothermal case. Heatconduction in the direction of flow is neglected based on theassumption that the thickness 2b is much smaller than theother two dimensions. The energy equation in the melt regionbecomes T T T 2T vx vyρcp(7) k 2 ηγ 2 t x y z(1) The thickness of the cavity is much smaller than the otherdimensions.(2) The velocity component in the direction of thickness isneglected, and pressure is a function of x and y only.(3) The flow regions are considered to be fully developedHele-Shaw flows in which inertia and gravitational forcesare much smaller than viscous forces.(4) The flow kinematics is shear-dominated and the shearviscosity is taken to be both temperature and shear ratedependent.The detailed derivations have been developed by Hieberand Shen, and these assumptions apply well for the microinjection molding process. In view of these assumptionsand neglecting compressibility during the filling stages, themomentum equation in the Cartesian coordinate systemreduces to [17] vy vx P Pη η (1) z z x z z ywhere vx and vy are velocity components in the x and ydirections, respectively; P(x, y) is the pressure, η(γ , T ) isthe shear viscosity, γ is the shear rate and T is temperature.Under the present assumptions, γ is given by 1/2 vy 2 vx 2 γ .(2) z zApplying the lubrication approximation, the thicknessaveraged continuity equation results in (bv̄x ) (bv̄y ) 0(3) x y416 P, x y P yand 1/2 2x 2y.where the ηγ 2 is the viscous heating term, and ρ, cp and k aredensity, specific heat and thermal conductivity, respectively.For simplicity, it is assumed that the velocity and temperatureare symmetric in the z direction, the velocities of polymermelt on the mold surfaces are zero and the temperature ofmold remains at Tw during filling. The boundary conditionsare given by vy vx 0at z 0 z z(8) TT Twat z b 0at z 0. zAs can be seen, the equations of this model are nonlinear andcoupled. It is difficult to solve these equations analytically.In this paper, simulation software C-MOLD that employsnumerical solvers based on a hybrid finite element/finitedifference method is used to solve the pressure, velocityand temperature fields of the GHS model. Because of theseapproximations, a GHS model cannot predict the exact flowfield near the advancing flow front or at the edges of the mold.This might cause errors in predicting the flow behavior nearmicroscale mold cavities.vx vy 0at z b3. Design and fabrication of molding apparatusAn aluminum mold is manufactured for the replicationprocess. The schematic diagram and a photograph of thealuminum mold, which consists of cavity and core halves,are shown in figure 2. The cavity half incorporates thecavity in which a mold insert is kept. A 4-inch silicon
Implementation and analysis of polymeric microstructure replication by micro injection moldingMold insert(Silicon wafer)Cavityhousing plateCore housingplateBase plateSprueHeaterStripperplateMountingplateMounting plateInsulation layerFigure 2. Injection mold set-up.Figure 3. Microstructures on a silicon mold insert.wafer with bulk micromachined microstructures is used asthe mold insert. Figure 3 shows the silicon micro moldinsert that is etched to have a cavity depth of 110 µm. Thesquare cavities have openings of 320 µm, 160 µm, 80 µm and40 µm and are etched by means of anisotropic silicon etchingin TMAH (tetramethyl-ammonium hydroxide). A heater isinstalled in the injection mold to control the temperature duringthe molding process. To have better thermal conductivityand shorter cooling time, we employed an aluminum moldthat is also easier to manufacture and modify. In addition,with appropriate thermal insulation and a cooling system, theproblem of dimensional variation caused by thermal expansioncan be controlled and an aluminum mold can be used as a moreeconomical tool for the replication process.The molded component can be removed from the moldmanually or by using the ejection system. Unlike the processesdescribed in the previous literature, a silicon wafer that servesas the mold insert is placed in the mold cavity. Using siliconwafer as mold insert has the advantage of short turnaroundtime. In addition, the wear of a silicon mold insert is muchsmaller as compared to a traditional nickel tool [18]. However,a silicon mold insert is more brittle than a nickel one. To avoidthe breakage of the wafer during the molding process, theedge of silicon wafer should match the cavity boundary. Agap between the mold insert and cavity can allow polymermelt to solidify within, which would eventually lift the waferFigure 4. Arburg Allrounder 221M 350-75 injection moldingmachine.from the cavity during mold opening and result in the breakageof the wafer.4. ExperimentsAn Arburg Allrounder 221M 350-75 conventional injectionmolding machine, as shown in figure 4, with a singlecavity, cold runner mold is employed. The material usedfor mold trials is Bayer Makrolon 2205 polycarbonate (PC)thermoplastic resin. Because of its excellent optical, chemicaland mechanical properties, PC can be used in applications suchas medical instruments, biochemical sensors and data storagesystems. The polymer is injected into the mold cavity at apressure ranging from 40 to 50 MPa. The melt temperature inthe feeding zone is maintained at about 300 C. The moldtemperature is controlled by a heater and maintained at atemperature lower than 200 C. The cycle time of the moldingprocess is 65 s, and polymer melt and mold are allowed tocool down for 30 s after the filling stage. Figure 5 showsthe typical pressure versus time and corresponding flow rateversus time relationship of the molding process. For the micromolding process, injection pressure, mold temperature and417
Y-C Su et alFlow ratePressureInjectionInjection0Packing HoldingPacking Holdingt1t2t 3 Time0t1t2t3 TimeFigure 5. Typical pressure versus time and corresponding flow rate versus time relationships during the injection molding process.Figure 6. SEM micrograph of molding results (injection pressure45 MPa, mold temperature 25 C).injection velocity are recognized as the driving parameters.The depth-to-opening ratios of molded microstructures areused to measure the quality of molding results. A higherdepth-to-opening ratio means better filling status and moldingquality.The presence of voids plays a major role in themolding process. Preheating of the polymer prior to themolding process reduces the chances of entrapment of voids.Conventional venting methods are difficult to use for microinjection molding due to the high possibility of undesiredstructural changes in the molded component. Hence, anevacuated mold is recommended to obtain a good replicationprocess.In the first mold trial, ordinary injection moldingparameters were used and no additional control unit wasactivated. The molding result is shown in figure 6. As canbe seen, the molding results have a small depth-to-openingratio which means polymer melt cannot fill the micro moldcavity. In this situation, polymeric microstructures cannotbe successfully replicated. Before doing more mold trials toimprove the molding results, a simulation tool was used tounderstand the flow behavior of polymer melt in the micromold-cavity for feasible modifications to improve the moldingresults.5. SimulationIt is well known that computer-aided engineering (CAE)can improve the trial-and-error techniques, and computer418models can be relied upon to predict flow behavior andmold results. Ideally, CAE analysis provides insight thatis useful in designing parts, molds and molding processes.By using CAE analysis to iterate and evaluate alternativedesigns and materials, engineering know-how in the form ofdesign guidelines can be established relatively fast and costeffectively.The CAE software C-MOLD developed by ACTechnology is employed as the numerical computation tool.The mold filling process is modeled by the GHS modeldescribed in the previous section. The numerical solutionsare based on a hybrid finite element/finite difference methodto solve for the pressure, flow and temperature fields and acontrol volume method to track moving melt fronts. A finiteelement mesh is used to approximate the circular-shape baseplate with convex microstructures on one surface, as shown infigure 7. This finite element model is composed of 6008 nodes,2672 two-dimensional (2D) triangular elements and 4607 onedimensional (1D) part runner elements. The 2D triangularelements, which disregard the shear and cooling from sidewalls, are used to model the substrate plate. The 1D partrunner elements, which consider the shear and cooling fromall the contact surfaces, are used to model the microstructureson the surface. The following conditions are considered inthis work to control and investigate the injection moldingprocess: Filling time/injection pressure. In order to generateuniform molecular orientation throughout the part, itis recommended to maintain a constant velocity at themelt front. However, only advanced injection moldingmachines have the ability to exactly achieve this requiredvelocity profile. In C-MOLD, either filling time orinjection pressure can be used to control the processsequence. Mold temperature. It is believed that surface effects willdominate the flow behavior at the microscale, and melttemperatures are the key that determines the fluid propertysuch as viscosity, specific heat and thermal conductivity.However, high temperature might cause the degradation ofpolycarbonate, so a pre-defined maximal allowable melttemperature is used in the simulation process [19]. Thickness of the base plate. The base plate is employedto support microstructures and the thickness of the baseplate will affect the balance of polymer melt and thequality of molded results. Because the thickness of thebase plate is much larger than the individual opening of
Implementation and analysis of polymeric microstructure replication by micro injection moldingEntranceMicro mold cavityhSprueddddh 54 Base plateFigure 7. C-MOLD finite element model.0.3dDepth-to-opening ratio (h/d)0.25h54 0.20.150.1100um200um300umMeasured (100um)0.05001020304050Radial location (mm)Figure 8. Simulation results of molding quality versus radial location with various cavity sizes (mold temperature 25 C, filling time 1.5 s)and measured results of 100 µm cavities.the micro mold-cavity, the problem of hesitation occurs,which requires careful consideration to control the qualityof molded results. Depth to opening ratio. The dimensions of the micromold-cavity determine the flow resistance and heattransfer effect. In this paper, the depth to opening ratio ofthe molding structures (dimensionless) is used to describethe quality of the molding process. Radial location. It is desirable to generate microstructuresuniformly throughout the surface of the base plate. In thispaper, the radial location away from the center of the baseplate, which is the entering point of polymer melt, is usedto investigate the quality of uniformity of the injectionmolding process.Several simulation results have been obtained. Squaremicrocavities with openings of 100 µm, 200 µm, 300 µmand 400 µm are used to investigate the injection moldingprocess. First, a simple case is used to investigate thepossibility of using traditional injection molding techniquesto fabricate microstructures. In this case no cooling or heatingsystems are used so that the mold temperature equals ambienttemperature initially. The result of this case is shown infigure 8 where a filling time of 1.5 s is applied. As can be seen,the depth-to-opening ratios of molded microstructures aresmall and not uniform. Molded microstructures in the centralarea have larger depth-to-opening ratios because local melttemperature and pressure close to the entrance of polymer meltare higher than other areas and facilitate better filling of thepolymer melt. On the other hand, it is predicted that polymericmicrostructures close to the edge also have larger depth-toopening ratios. This is believed to be the result of backpressure when the polymer melt front hits the enclosed edge ofthe mold cavity. If the molding process is optimized, the depthto-opening ratio should be 0.707 for cavities etched by thesilicon anisotropic etching method. Generally, cavities withlarge openings have high penetration depth because it requiresless pressure to press polymer melt into these cavities. Thefilling process is incomplete because insufficient material wasinjected into the cavity as shown. The incomplete filling maybe caused by insufficient machine injection pressure (resultingfrom high melt resistance and a restricted flow path) or premature solidification of the polymer melt. The temperature ofpolymer melt in the micro mold-cavity could reduce rapidlyafter the polymer melt enters the cavity because of the highsurface-to-volume ratio as observed in the first mold trial to419
Y-C Su et al0.55Depth-to-opening ratio (h/d)0.5300um/0.4s300um/0.8s300um/1.5sMeasured (300um/1.5s)0.450.4d0.35dd0.3h54 0.250.201020304050Radial location (mm)Figure 9. Simulation results of molding quality versus radial location with various filling time (mold temperature 100 C) and measuredresults with 1.5 s filling time.d0.8Depth-to-opening ratio (h/d)dh0.7d54 l location (mm)Figure 10. Simulation results of molding quality versus radial location with various mold temperature (filling time 1.5 s).cause the phenomenon of short shot. The glass transitiontemperature of polycarbonate is about 145 C and the meltingtemperature is 225 C. Using a heater to keep mold cavitytemperature above the glass transition temperature can helpsolve the solidification problems. However, it will increasethe cycle time and reduce the efficiency.Some other factors including filling time and the thicknessof base plate are investigated. Figure 9 shows the simulationresults between the filling time and the depth-to-openingratios. As expected, shorter filling time, which means largerinjection pressure and faster injection velocity, will generatemicrostructures with larger depth-to-opening ratios. Theother possible modification is to change the thickness ofthe base plate. Based on simulation results, high injectionpressure is required to push polymer melt if the thickness ofthe base plate is small. This modification still fails to makeperfect replication. The next step to improve the moldingprocess is to increase the molding temperature. Figure 10420shows the simulation results between mold temperature anddepth-to-opening ratios when filling time is kept constant at1.5 s. As can been seen, the increase of mold temperaturedoes not increase the depth-to-opening ratios dramatically ifthe mold temperature is low or just a little higher than the glasstransition temperature. When mold temperature is at 200 C,polymer melt can fill the micromold cavity completely in 1.5 sas shown in the simulation results. Ideally, if mold temperatureexceeds the glass transition temperature of the polymer, nosolidified layer will be generated and polymer melt can fillthe micromold cavity completely if suitable injection pressureand filling time are applied. However, because the viscosityof polymer melt is relatively large when its temperature is justabove the glass transition temperature, the required injectionpressure is extremely large or the filling time is extremelylong. According to the simulation results, a mold temperature30 C to 40 C higher than the glass transition temperatureis recommended for the replication process. Also shown
Implementation and analysis of polymeric microstructure replication by micro injection moldingDepth-to-opening ratio (h/d)0.250.2d0.15h54 0.10.05100um/25C100um/100C001020304050Radial location (mm)Figure 11. SEM micrograph of molding results (injection pressure45 MPa, mold temperature 100 C).in the simulation results, the improvement on the moldingresults in the area close to the edge is less than those in otherareas. Because of the significant decrease of viscosity at highermold temperature, the required pressure for driving polymermelt into the micro-cavity is significantly reduced. Once themold temperature exceeds a certain limit, a polymer melt willbe able to fill all micro mold-cavities on the surface. Asthe temperature becomes the dominant factor, the issue ofback pressure becomes less important. It is believed that thetemperature far away from the center of the base plate will belower compared to the central region. As a result, the depth-toopening ratio becomes smaller as indicated in the simulationresults.When mold temperature is higher than the glass transitiontemperature of the polymer, the analysis is simplified byusing an isothermal model which assumes that the polymermelt temperature is fixed at the mold temperature during thefilling stage. This isothermal assumption is verified by anon-isothermal model, and the simulation results show thatthe difference between these two models is less than 10%in the worse case in predicting the depth-to-opening ratio.Consequently, the energy equation can be omitted and theflow behavior can be obtained by solving two nonlinear andcoupled equations instead of three.6. Experiment results and discussionsAs discussed previously, a temperature control unit wasfound to be most effective in achieving good molding results.Therefore, a heater capable of maintaining mold temperaturewas mounted on the backside of the mold insert to controlthe mold temperature. Figure 11 shows the results of moldtrials after the heater is set at 100 C. As can be seen,large microstructures (with openings larger than 100 µm)have flat instead of curved surfaces which means that thepolymer melt can flow deeper into the micro mold-cavityand replicate the pyramidal shape of the cavity. On theother hand, small microstructures (with openings smaller than100 µm) have better (but still curved) replication results ascompared to the previous results. The depth-to-opening ratiosFigure 12. Measurement results of molding quality versus radiallocation with various mold temperature (injection presssure45 MPa).of replicated microstructures are measured using a whitelight interferometer (Wyko NT 3300 profiling system) andthe results are shown in figures 8, 9 and 12. The comparisonof figures 8 and 9 indicates that the general trend, havinghigher depth-to-opening ratio around the center and edge,is predicted by the simulation results with 20% to 40%overestimation probably due to temperature variation acrossthe mold. Figure 12 compares the measured depth-to-openingratio against radial location at 25 C and 100 C and verifiesthe effects caused by raising mold temperature as simulationresults suggest.As predicted by simulation results, polymer melt cannotfill the micro mold-cavity completely. Based on figure 10, themold temperature should be higher than 200 C for completeduplication. However, in this prototype demonstration,the heating strip used is unable to maintain such a hightemperature. Therefore, one of the future works is to have amore sophisticated heating device and mold design to achievea higher mold temperature for optimized molding results. Onthe other hand, the high mold temperature will also increase therequired cooling period. A cooling system can bring down theoverall temperature effectively and reduce the cooling time.Another problem is air bubbles trapped in molded polymerstructures. In conventional molding processes, air vent slotsfacilitate the escape of the air in the mold during mold filling.However, these slots are close to the size of the microstructuresso they will be filled by the adapted micro-molding processand fail to vent the trapped air. Therefore, it is often suggestedthat an evacuation system should be constructed for the microinjection molding process.7. ConclusionsA conventional injection molding technique for the replicationof polymeric microstructures is investigated. According to thesimulation and experimental results, a conventional injectionmolding process with a specific process control strategy canbe applied in the fabrication of microstructures. Because ofthe large surface-to-volume ratio of the micromold cavity, thetemperature of the polymer melt reduces rapidly right after it421
Y-C Su et alreaches the entrance of the cavity. If the mold temperatureis lower than the transition temperature of polymer, boundarysolidification will prevent the melt from flowing into the cavity.This phenomenon contributes to the low depth-to-openingratio in the preliminary molding results. Two modificationshave been identified to improve the process. First, the moldtemperature must remain much higher than the transitiontemperature of polymer to reduce the viscosity during thefilling stage. In this situation, no solidified layer will begenerated and longer filling time and lower injection pressurecan be employed to reduce residual stress. Secondly, a vacuumis preferred in the micro mold-cavity to avoid possible airtraps. In addition, parameters such as injection velocity,injection pressure, holding pressure, cooling time and melttemperature must also be controlled appropriately, based onthe mold designs, to achieve optimal performance.AcknowledgmentsThis work is supported in part by a NSF award (DMI-9800434)and a DARPA/MTO/MEMS grant. The author would liketo thank Mr Kent Pruss for assisting the manufacturing ofaluminum mold and Mr Gerald Polashak for operating theinjection molding machine.References[1] Menz W, Bacher W, Harmening M and Michel A 1991 TheLIGA technique—a novel concept for microstructures andthe combination with Si-technologies by injection moldingProc. IEEE Micro Electro Mechanical Systems Workshoppp 69–73[2] Bacher W, Menz W and Mohr J 1995 The LIGA techniqueand its potential for microsystems—a survey IEEE Trans.Ind. Electron. 42 431–41[3] Both A, Bacher W, Heckele M, Muller K D, Ruprecht R andStrohmann M 1995 Molding process with high alignmentprecision for the LIGA-technology Proc. IEEE MicroElectro Mechanical Systems Conference pp 186–90[4] Ruprecht R, Hanemann T, Piotter V and Hausselt J 1995Injection molding of LIGA and LIGA-similarmicrostructures using filled and unfilled thermo-plasticsProc. SPIE 2639 146–57422[5] Weber L, Ehrfeld W, Freimuth H, Lacher M, Lehr H andPech B 1996 Micromolding—a powerful tool for the largescale production of precise microstructures Proc. SPIE2879 156–67[6] Rogalla A and Michaeli W 1997 Analysis of injection moldedmicrostructures Proc. SPE 55th Annual TechnicalConference pp 365–8[7] Piotter V, Hanemann T, Ruprecht R, Thies A and Hausselt J1997 New developments of process technologies formicrofabrication Proc. SPIE 3223 91–9[8] Despa M S, Kelly K W and Collier J R 1999 Injection moldingof polymeric LIGA HARMs Microsyst. Technol. 6 60–6[9] Piotter V, Bauer W, Benzler T and Emde A 2001 Injectionmolding of components for microsystems Microsyst.Technol. 7 99–102[10] Piotter V, Mueller K, Plewa K, Ruprecht R and Hausselt J2002 Performance and simulation of thermoplasticmicroinjection molding Microsyst. Technol. 8 pp 387–90[11] Madou M J, Lee L J, Koelling K W, Daunert S, Lai S,Koh C G, Juang Y-J, Yu L and Lu Y 2001 Design andfabrication of polymer microfluidic platforms forbiomedical applications Proc. SPE 59th Annual TechnicalConference pp 2534–38[12] Yu L, Koh C G, Lee L J, Koelling K W and Madou M J 2002Experimental investigation and numerical simulation ofinjection molding with micro-features Polym. Eng. Sci. 42871–88[13] Larsson O, Ohman O, Billman A, Lundbladh L, Lindell C andPalmskog G 1997 Silicon based replication technology of3D-microstructures by conventional CD-injection moldingtechniques Proc. Int. Conf. on Solid-State Sensors andActuators pp 1415–8[14] Ohman O 2000 Polymer replication of microfluidic CDdevices Proc. 2nd Annual Int. Conf. on Manufacturing &Commercialization Issues for Micro & Nano MedicalDevices[15] Ho C-M and Tai Y-C 1998 Micro-electro-mechanical-systems(MEMS) and fluid flows Annu. Rev. Fluid Mech. 30579–612[16] AC Technology 1998 C-MOLD version 4.0[17] Hieber C A and Shen S F 1982 A finite-element/finite-different simulation of the injection-molding fillingprocess J. Non-Newtonian Fluid Mech. 7 1–32[18] Becker H and Heim U 1999 Silicon as tool material forpolymer hot embossing Proc. IEEE Micro ElectroMechanical Systems Conference pp 228–31[19] Bottenbruch L 1996 Engineering Thermoplastics:Polycarbonates, Polyacetals, and Polyesters (Cincinnatti:Hanser Gardner)
molding process, injection pressure, mold temperature and 417. Y-C Su et al Injection Packing Holding Holding 0 t 1 t 2 t 3 0t 1 t 2 t 3 Injection Packing Time Time Pressure Flow rate Figure 5. Typical pressure versus time and corresponding flow rate versus time relationshi
