Luận văn Thạc sĩ (Tiếng Anh): Simulation and fabrication of piezoelectric mems inkjet print head

al frequency. With membrane size of 300 um, the numerical and experimental values of the maximum displacement rate were ca. 0.053 um/V and ca. 0.059 um/V, respectively (Fig. 2-3). The simulation was also extended for various sizes of PIPH actuator membrane (i.e., 500um, 600um). Fig. 2-4(a) indicated the relationship between the maximum displacement of PIPH actuator membrane and thickness ratio of PZT to support layer at an applied voltage of 5V (thickness of support layer was fixed of 2.3 um). Increasing PZT thickness makes the PZT actuation strain increased and the stiffness of PIPH actuator membrane, thus, also increased, which decreased its displacement. Total actuation strain of PIPH actuator membrane finally depends on the competition between them. Therefore, maximum displacement of PIPH actuator membrane slightly increased from thickness ratio of 0.2 to 0.5 (corresponding to PZT thickness of 0.5 um to 1.15 um) and then decreased with thicker PZT film. According to Gere & Timoshenko, the maximum displacement of a pure bending membrane (with width a and Young’ s modulus E) under uniform load, P, can be expressed as [18] 18 3 4 max Et Paαδ = with α = 0.0138 (7) Therefore, with the maximum displacement δmax produced by PIPH actuator membrane, the maximum force can be inferred approximately as 2 3 max max a EtF α δ= (8) Figure 2-4(b) shows the relationship between the maximum force, the maximum displacement of PIPH actuator membrane and its width at applied voltage of 10V. Both actuator membranes with PZT thickness of 0.5 um and 1 um exhibited the similar values of maximum displacement and the parabolic dependence upon the width ( αmax ~ a2). However, the maximum force was enhanced significantly (i.e., 1.75 times) in case of the PZT thickness of 1 um. In addition, the maximum force reached the saturated values at membrane width of ca. 600 um. This marks a notice in selecting optimized sizes of PIPH actuator membrane which can produce both high maximum displacement and maximum force so that the highest value of the maximum work can be obtained. Finally, the fundamental frequency of PIPH actuator membrane (width of 300um) was predicted using FEMLABTM software and compared with an experimental result monitored by HP4194A impedance analyzer. Its simulation and experimental values were ca. 379 kHz and ca. 328 kHz, respectively (Fig.2-5). 2.4.2 Performance characteristics of PIPH actuator in liquid Amount of the fluid ejected through the nozzle is determined not only by the maximum actuation displacement, maximum actuation force of the membrane and the behavior of the fluid inside the system, but also by the deflection shape of the membrane [4]. The displacement of the membrane is due to the shear stress applied by the actuator and by the pressure of the fluid. This pressure was solved from Navier- Stokes equations by setting the membrane displacement as one of its boundary conditions. Therefore, the behavior of the PIPH is a set of electrical-mechanical-fluid couplings. 19 When the fluid-membrane interaction was considered, the maximum displacement slightly reduced (i.e., 0.049 um/V (in liquid-interaction) vs 0.053 um/V (in air)). The maximum force, thus, also reduced in accordance to equation (8). The maximum displacements of PIPH actuator membrane at various driving frequencies are listed in Table 2-4. And typical deflection shapes are shown in Fig.2-6. At low frequencies (100 Hz-25 kHz), the deflections were not sensitive to the driven frequency. The PIPH actuator membrane bends in one direction and it has only one peak (Fig.2-6 (a,b)). The maximum displacements in these cases were 0.245 ~ 0.267 μm. At higher frequencies (25 kHz-100 kHz), the deflections increased with the frequency and were different between bend-up and bend down modes. Moreover, because of the membrane-fluid interaction, the membrane deflection shape was changed at frequencies above 100 kHz. The deflection shape of PIPH actuator membrane was sophisticated and exhibited two or more peaks (Fig.2-6 (c,d)). The appearance of the deflection peaks becomes one of the disadvantages for the PIPH actuator performance. The behavior of fluid inside the chamber, thus, also changed unexpectedly. Below 125 kHz, outlet flow rate increased with increasing the frequency. The response of fluid and the vibration of membrane were in phase. The ratio between backflow and net flow reached the minimum value of ca. 3% at driving frequency of 25 kHz. Above 125 kHz, the response of fluid and vibration of membrane were out of phase (Fig.2-7(a)). This frequency is considered as a resonance frequency of the PIPH actuator membrane in liquid. And the experimental result showed the resonance frequency of membrane in liquid was ca. 90 kHz (Fig.2- 7(b)). Both numerical and experimental resonance frequencies of membrane in liquid were about 3 times smaller than those of membrane in air. The driving frequency of PIPH actuator must be smaller than this resonance frequency of membrane in liquid. Summary of maximum displacement and resonant frequency was shown in Table 2-5. The prediction for actuator membrane with width of 600 um was also given in this table. The simulation results agree well with experimental ones. This work along with another work studying microdroplet generation of PIPH (will be reported elsewhere) offers an effective guideline in designing the PIPH. 20 2.5 Conclusion Performance characteristics of PIPH actuator have been investigated numerically and experimentally. The maximum actuation displacement and the maximum actuation force are basic performance characteristics of a work-producing PIPH actuator. Natural frequency of actuator membrane was known as a limit of PIPH actuator operating at dynamic regime. Considering the electro-mechanical-fluid couplings indicated the inter- dependence between PIPH actuator membrane and fluid, which leads to significant changes such as the decrease of resonance frequency or the appear of multi-peak deflection shapes as well as out-of-phase response of ink liquid. The numerical results agree well with the experimental ones. These results play an important role in selecting the appropriate design parameters so that characteristics of PIPH actuator can be optimized. 21 2.6 References [1] M. Usui, Seiko Epson Corporation, Shiojiri, Nagano, Japan [2] Hermann Seitz and Joachim Heizl, J.Micromech.Microeng.14(2004) 1140-1147 [3] Steve Temple, Xaar plc, Cambridge [4] B Fan, G Song and Hussain, Smart Mater.Struct. 14 (2005) 400-405 [5] Vishal Singhal and Suresh V.Garimella, IEEE transactions on advanced packaging, VOL.28.N0.2, MAY2005 [6] J. KIM, ME608 Final project, Apr.26.2002 [7] J. S. Yahng and S. C. Jeoung, D. S. Choi and D. Cho, J. H. Kim, H. M. Choi and J. S. Paik, J.Korean Phys. Soc.Vol. 47, No. 6, December 2005, pp. 977_981 [8] K.Y.Lee, E.D.Case, Journal of Materials Science 31(1996)2253-2264. [9] Jun-Kyu Paik, Sanghun Shin, Sun-Woong Na, Nae-Eung Lee, and Jaichan Lee, Integrated Ferroelectrics, 69:383-390 (2005). [10] Frank M.White, Fluid Mechanics, 4th ed. (Mc Graw Hill, 1998). [11] Joel h. Ferziger et al, Computational Methods for Fluid Dynamics (Springer Verlag, 1999). [12] IEEE Standard on Piezoelectricity, ANSI/IEEE Std176-1987 [13] Szilard R, Theory and Analysis of Plates, Classical and Numerical Methods (Englewood Cliffs, NJ: Prentice-Hall,1974) [14] CFD-ACE+ Modules Manual version 2004 [15] J. R. Ahn, D. W. Kim, G. Y. Yeom, J. B. Yoo, and J. Lee, Ferroelectrics, 263, 244 (2001). [16] Jurgen Brunahl, Physics of Piezoelectric Shear Mode Actuators (Stockholm,2003). [17] J.E.Huber, N.A.Fleck and M.F.Ashby, Proc.Soc.Lond.A (1997)453,2185-2205. [18] Gere & Timoshenko, Mechanics of Materials, 3rd ed. (PWS-KENT, 1990). 22 Table 2-1. Fluid properties Property Density (ρL) Dynamic viscosity (μ) Unit kg m-3 x10-3 kg m-1 s-1 Value 800 2 Table 2-2. Support layer properties Property Density (ρ) Young’s Modulus (E) Poisson’s ratio (υ) Unit kg m-3 x109 Pa Value 3500 212 0.26 Table 2-3. PZT properties (PZT 52/48 ) Piezoelectric coefficients (x10-12 C/N) Density (kg m-3) Young’s Modulus (Pa) Poisson’s ratio (υ) d13 d23 d33 d42 d51 7500 1.7 x1011 0.3 93.5 93.5 223 494 494 Table 2-4. The displacement at various driving frequencies (voltage=5V) Frequency, f (Hz) Max. bend up (μm) Max Bend down (μm) 100 ~ 500 -0.2447 0.245 1k ~ 5k -0.2439 0.2451 10 k ~ 25 k -0.265 0.267 40 k ~50 k -0.296 0.302 100 k -0.371 0.365 125 k -0.416 0.533 200 k -0.595 0.677 Table 2-5. Summary of actuator performance characteristics Maximum displacement (in air) Maximum displacement (in liquid) Resonance frequency (in air) Resonance frequency (in liquid) Simulation (300 um) 0.053 um/V 0.049 um/V 379 kHz 125 kHz Experiment (300 um) 0.059 um/V x 328 kHz 90 kHz Prediction (600 um) 0.240 um/V x ~ 100 kHz ~ 30 kHz 23 Fig. 2-1. Model of a piezoelectric inkjet print head (PIPH) structure: (a) design, (b) CFD-ACE+ symmetric model with meshing grids. Fig. 2-2. Flowchart of fabrication process (a) and SEM images (b) of PIPH actuator. 24 Fig. 2-3. Maximum displacement of PIPH actuator membrane (300 um): (a) simulation and (b) experiment. Simulation was extended with membrane width of 500-600 um. Fig. 2-4. Dependence of actuator performance on geometrical parameters, (a) maximum displacement vs. thickness ratio (PZT/support layer) and (b) maximum force (Fmax) and maximum displacement (δmax) vs. membrane width Fig. 2-5. Resonance frequency (in air) of PIPH actuator membrane: (a) FEMLAB simulation and (b) experiment with HP4194A impedance analyzer. 25 Fig. 2-6. Deflection shape of actuator membrane interacting with liquid, dome shape with one peak at low frequencies (a) & (b), unexpected shape with more than one peak at higher frequencies (above 125 kHz < 379 kHz - resonance frequency in air ). Fig. 2-7. Resonance frequency (in liquid) of PIPH actuator membrane: (a) simulation and (b) experiment. 26 CHAPTER 3. SIMULATION OF MICRODROP GENERATION IN PIEZOELETRIC MEMS INKJET PRINT HEAD Abstract The simulation of microdroplet generation in piezoelectric MEMS inkjet print heads has been performed. The effects of actuating characteristics such as driving amplitude and frequency of the piezoelectric membrane, fluid characteristics including surface tension and viscosity, and geometrical parameters were investigated. From the simulation results, we obtained three regions for the formation of droplet with the actuating and fluid parameters, i.e., the formation of no-droplet, single droplet and satellite droplets. The single droplets can be obtained by applying driving voltage operating high frequencies with small amplitudes. The simulation results also indicated the competition between cohesive and disruptive forces in generating droplets. And the relative chamber size, aspect ratio of nozzle, nozzle shapes and the diffuser function were importantly geometrical parameters influencing on the droplet generation. This modeling offers a visual description, and an effective guideline to design inkjet print head structure with high quality printing. 27 3.1 Introduction Recently, the industrial ink-jet printing technology has been considered as a next generation process tool which replaces environmentally harmful, time-consuming and expensive process for the fabrication of precise electronic devices such as flat panel display (FPD) and various types of micro-devices. One of the most important elements in an ink-jet printing system is the ink-jet print head used to eject small amounts of fluid on target surfaces. The ink-jet print head can be classified into continuous-mode and drop-on-demand mode heads. In a drop-on-demand mode head, a piezoelectric layer or a thermal actuator component is typically incorporated into ink-jet head structure to create the dynamic force for ejecting the fluid from a nozzle outlet. Piezoelectric ink-jet heads normally offer a greater range of ink compatibility than thermal ink-jet heads, which are limited to water-based inks or require a new design for each different type of ink solvent [1-7]. Various designs and fabrication techniques of a piezoelectric inkjet print head have been reported in literatures [4-8]. However, it is necessary to obtain more intensive understanding of fundamental phenomena and principles in ejection process, which gives a useful guideline to design an inkjet structure. In order to reduce the computational demand, the piezoelectric actuator, which gives an actuating force, was neglected and replaced by a moving wall boundary. This treatment allows us to concentrate on the interaction between the liquid and surrounding medium at which the competition between the cohesive and disruptive forces occurs and droplets are generated. In this study, we have investigated the ejection process of droplets under various driving characteristics such as displacement amplitude and driving frequency, and fluid properties, i.e., surface tension and viscosity. We report the formation of fluid droplets at such various conditions. 3.2 Modeling and simulation settings In general, an acting function of a moving wall boundary (i.e. piezoelectrically actuated vibrating diaphragm) causes ink to be ejected from a nozzle or an orifice onto a surface. It is well expected that physical parameters determine fluidic characteristics of 28 ),,( wvuV =r 0=Vr the ejected ink, such as formation of a single droplet, decrease of satellite droplets or non-splashing, droplet size and so on. Those characteristics depend on driving force, fluid properties and geometric parameters of a print head [1, 2, 11]. The simulation is implemented with a simple piezoelectric inkjet structure consisting of a cylindrical chamber connected to a cylindrical nozzle. The diameters of chamber and nozzle are 300 μm and 30 μm, respectively. The piezoelectric actuator, which has a “forcing function” to produce droplets, was replaced by a moving boundary, as shown in Fig. 3-1. To simplify and reduce computational cost, 2-dimensional section of the inkjet structure was used in this simulation with orthogonal meshing. The simulation was performed using CFD-ACE+ package software known as a multiphysics modeling tool. For droplet generation, air region was added in the model (as shown in Fig. 3-1(b)). According to the Volume of Fluid (VOF) theory, the liquid volume fraction F is determined by solving the following passive transport equation (1) along with the Navier-Stokes equation (2) [9-15]. 0=•∇+∂ ∂ FV t F r (1) PVg Dt VD ∇−∇+= rr r 2μρρ (2) where is velocity vector, ρL is density and μ is dynamic viscosity of fluid. Isopropanol was used as an ink fluid in the simulation (ρ=800 kg/cm3, σ=22 mN/m, μ=2 cp). Boundary conditions for the fluid model are non-slip at the fluid-wall interfaces ( ) and far-field at the outlet where a constant pressure is set to the same as atmospheric pressure. Moving boundary condition is applied to the upper wall of the chamber. The movement of the upper wall can be expressed as follow: y(x,t)=Asin(2πx/2a)sin(2πft) (3) where A and f are amplitude and vibrating frequency of the driving displacement produced by piezoelectrically actuated membrane, respectively. a is the radius of the chamber. 29 Generally, droplet breakup in a flowing stream is governed by its surface tension, viscous forces, and dynamic pressure [1,2]. The effects of these quantities in terms of fluid properties and driving characteristics are considered in this simulation. 3.3 Results and discussion 3.3.1 Microdrop generation process The microdrop generation process was monitored at a driving displacement with amplitude and frequency of 5 μm and 30 kHz, respectively. At this driving displacement, a primary droplet is formed within one cycle, i.e., 33.33 μs. Figure 3-2 shows the four main steps of an ejection cycle including infusion, inversion, ejection and relaxation [1,4]. During infusion, called fluid jet formation, the displacement starts at zero or equilibrium state (Fig. 3-2(a)) and begins to increase, reaching a maximum value of 5 μm. This corresponds to the driving plate deflecting downward, drives the fluid into the inkjet orifice from the chamber and pulls the meniscus into the orifice through the nozzle. The meniscus grows until approximately 8.3 μs (Fig. 3-2(b)), begins to decrease in size and deform in shape. This step is called inversion corresponding to the decrease of displacement from positive to negative value (Fig. 3-2(c)). When the fluid jet is destabilized enough, the meniscus is broken off and a droplet is formed at 21.75 μs (Fig. 3-2(d)). During relaxation, the flow undergoes viscously-damped oscillations (Fig. 3- 2(e)) and approaches an equilibrium state before the next ejection cycle begins (Fig. 3- 2(f)). The destabilization of the fluid jet depends on the driving displacement, which produces discrete and free-flying drops or satellite droplets. For example, if clean break does not occur between the primary droplet and the fluid in the nozzle, satellite droplets are generated. The formation of satellite droplets needs to be suppressed for accurate control of ejection process such as the volume of the droplet and time duration of the droplet formation. 3.3.2 Effect of actuating characteristics The driving displacement includes its amplitude and frequency. Those quantities will affect the generation time and quality of droplets. Figure 3-3 shows the images of 30 single droplet and satellite droplets at various driving displacements. At the driving displacements with small driving amplitudes (e.g., 3.5 μm, 30 kHz –Fig. 3-3(a)), a single droplet is generated. At the driving displacements with large amplitudes and/or high frequencies, satellite droplets are generated (as typically shown in Fig. 3-3 (b) 3.5 μm, 70 kHz; (c) 4.5 μm, 50 kHz, (d) 5 μm, 30 kHz). In most situations, it is desirable to eliminate satellite droplets and increase ejection rates (or reduce droplet generation time). Figure 3-4 indicates the relationship between the time duration for droplet generation and actuating characteristics, i.e., the amplitude and frequency of the driving displacement. The droplet generation time decreases when the amplitude and/or frequency of the driving displacement increase. There exist three regions for the formation of droplets: no droplets, single droplets and satellite droplets. Droplets can not be formed at amplitudes below 2.75 μm (Fig. 3-4(a)) or frequencies below 20 kHz (Fig. 3-4(b)). Close to the boundary between no-droplet and single droplet regions, several push-pull cycles are required to generate a single droplet (i.e., 19 cycles at a driving displacement of 3 μm and 30 kHz). It’s clear that there is a strong competition between cohesive and disruptive forces in this region and cohesive forces are predominant. Thus it’s required numerous push-pull cycles to accumulate enough disruptive force. That’s the reason why time duration for droplet generation significantly increases with smaller driving displacements (i.e., 3 μm - 30 kHz, 2.75 μm- 40 ~ 80 kHz in Fig. 3-4 (a), or 3 μm - 30 kHz, 4 μm - 20 kHz in fig.4 (b)). Far from this boundary, the number of push-pull cycles for droplet formation decreases significantly and satellite droplet formation gradually occurs. Figure 3-4(b) indicates that only single droplet is generated at driving displacements with amplitude of 3 μm and frequencies above 30 kHz. However, satellite droplets are generated at most frequencies as the amplitude becomes large above 4 μm. The threshold of droplet ejection obtained from this simulation provides a guideline in choosing geometric parameters when the inkjet structure is designed. 3.3.3 Effect of fluid properties The surface tension of a liquid tends to pull the liquid into a form that exhibits the 31 minimum surface energy, while the stabilizing effect of liquid viscosity tends to oppose any disturbance in liquid geometry. External forces, such as driving displacement in this case, acting on the liquid surface may distort the bulk liquid and promote the disruption [2,3]. Typical viscosity and surface tension of a fluid to be used in a piezoelectric inkjet device are in range of 0.5-40 cp and 20-70 mN/m, respectively [1,2]. Figure 3-5 shows the droplet generation time versus fluid properties, e.g., surface tension and viscosity. The change in the time duration for droplet generation implies the competition between the cohesive and disruptive forces acting on the liquid surface. The competition leads to oscillation and perturbation in the liquid. Under favorable conditions, the oscillation may be amplified to such an extent that the bulk liquid disintegrates into droplets. For instance, when viscosity and amplitude of the driving displacement are fixed at 2 cp and 3.5 μm, respectively, the time duration for droplet generation increases with the surface tension from 20 mN/m to 70 mN/m and decreases with the frequency, as shown in Fig. 3-5 (a). The strong competition between surface tension and driving force produced by the driving displacement is clear at smaller frequencies, such as 40 kHz , 50 kHz and 60 kHz (as shown in Fig. 3-5 (a)). Surface tension is influencing to the time duration of droplet generation at driving frequencies. Increasing surface tension makes cohesive forces predominant. Therefore it’s more difficult (required longer time) to generate droplets with increasing surface tension while fixed driving displacement (i.e., increasing surface tension from 30 mN/m to 70 mN/m at fixed driving displacement of 3.5 μm - 30 kHz, or 50 mN/m to70 mN/m at 3.5μm – 40~50 kHz). At a driving frequency of 40 kHz, it takes 200 μs or 8 cycles to generate a droplet at a surface tension of 60 mN/m. However, no droplet is generated at a surface tension of 70 mN/m even after 20 pull-push cycles. On the other hand, the driving displacement becomes predominant when the driving frequency increases above 70 kHz, at which a droplet is formed within 1 pull-push cycle irrespective of surface tension. Time duration for droplet generation also increases linearly with the viscosity. It takes about 1 to 3 pull- push cycles to form droplets at investigated driving displacements, i.e., amplitude of 5 μm and frequencies of 20 kHz – 40 kHz, as shown in Fig. 3-5 (b). It is also observed 32 that less satellite droplets are generated with increasing the viscosity. Viscosity acts to dampen the instabilities that lead to satellite formation. 3.3.4 Effect of geometrical parameters Figure 3-6 shows three types of inkjet structure used to consider the effect of geometrical parameter. Effects of the relative chamber size X1/X2 (A-type), aspect ratio d/h (B-type) and diffuser function (C-type) are analyzed. Time duration for droplet generation reaches the minimum values at relative ratio of 0.3-0.5, as shown in Fig. 3- 7(a). Conical shape of nozzle inlet (B-type) reduces fluidic impedance compared with cylindrical shape (A-type). Therefore, droplet generation time of B-type is shorter than that of A-type, i.e., 55 um (B-type) versus 135 um (A-type) at the same driving characteristics of 2.5 um-30 kHz. It’s also observed that time for droplet generation decreases with decreasing the aspect ratio (AR=diameter of nozzle/height of nozzle). At relatively high driving characteristics, B-type and C-type have the same time duration for microdroplet generation (Fig. 3-7(b)). However, at lower characteristics (i.e., 2 um- 30 kHz), B-type can generate a single droplet after 5 pull-push cycles (165 us) while C- type generates no-droplet. This can be explained that adding a diffuser into microchannel makes backflow (flow from chamber to reservoir) increased and thus decreases the flow moving through nozzle inlet. Among there types of inkjet structures, B-type with optimized parameters such as relative ratio of 0.5 and conical nozzle shape, could be selected as a good candidate. Then its three regions of microdroplet generation are shown in Fig. 3-8. The microdroplet can be generated with applied voltages of 9V- 21V and frequencies above 15 kHz. 3.4. Conclusion Microdrop generation process and effects of basic forces on the process such as surface tension, viscous forces, and dynamic pressure, have been investigated. The correlation of driving displacement and fluid parameters is also analyzed in this simulation. Four distinct regimes have been identified in an ejection cycle: infusion, inversion, ejection and relaxation. In order to form droplets, the necessary and enough condition is that the driving force caused by the driving displacement is larger than the 33 surface tension force and it should be dynamic. The simulation results indicate that the driving displacement plays an important role to reduce the droplet generation time and improve droplet quality. The results show three regions divided into no-droplet, single droplet and satellite droplet regions. The droplet formation requires the following actuating characteristics: driving displacements with large amplitude and/or frequency. However, further high performance of an inkjet head, i.e., short duration time for droplet generation and high quality droplet (single droplet), needs driving displacements with relatively small amplitudes and high frequencies. Geometrical parameters significantly affect to microdroplet generation. Relative ratio of chamber sizes, aspect ratio and nozzle shape are important parameters which are optimized from this simulation. 34 3.5 References [1] Eric R.Lee, Microdrop generation, CRC Press (2003). [2] Liu, Huimin, Science and Engineering of Droplets - Fundamentals and Applications, William Andrew Publishing/Noyes (2000). [3] Nam-Trung Nguyen, Steven T. Wereley, Fundamentals and Applications of Microfluidics, Artech Hourse, Inc (2002). [4] Carl D. Meinhart and Hongsheng Zhang, J. Microelectromechanical Systems, 9 (2000). [5] John Collins, Yung-Chieh Tan, Abraham P.Lee, IMECE,41983 (2003). [6] MicroFab Technologies, Inc, www.microfab.com. [7] J.M. Meacham, M.J.Varady, F.L.Degertekin, and A.G.Fedorov, Physics of fluids, 17 (2005). [8] Steve Temple, Small fast inkdrop emission from a nozzle, Xaar plc, Cambridge. [9] B Fan, G Song and Hussain, Smart Mater.Struct. 14, 400-405 (2005). [10] Joel H. Ferziger et al, Computational Methods for Fluid Dynamics, 3rd Edition, Springer. [11] Joohan KIM, ME608 Final project (2002). [12] CFD-ACE+ Modules Manual version (2004). [13] Sanghun Shin, Jun-kyu Paik, Nae-eung Lee, Jaichan Lee, Jun-shik Park and Hyo- derk Park, J.Korean Phys. Soc. 46, 292-295 (2005). [14] Deuk Chul Kwon and N. S. Yoony, J. H. Kim, Y. H. Shin and K. H. Chung, J.Korean Phys. Soc. 47, 163-166 (2005). [15] J. S. Yahng and S. C. Jeoung, D. S. Choi and D. Cho, J. H. Kim, H. M. Choi and J. S. Paik, J.Korean Phys. Soc. 47, 977-981 (2005). 35 Fig. 3-1. Inkjet head geometry, (a) Three dimensional (3D) and (b) 2D symmetric section in CFD-ACE+. Fig. 3-2. Microdrop generation process at driving displacement with amplitude of 5 μm and frequency of 30 kHz. 36 Fig. 3-3. Droplet properties: no-droplet, single droplet and satellite droplets at various driving displacements (2~5um, 50 kHz). Fig. 3-4. Time duration for droplet generation at various actuating characteristics: (a) amplitude and (b) frequency. Droplets are generated in one cycle or several cycles. Fig. 3-5. Time duration for droplet generation with fluid properties: (a) surface tension and (b) viscosity. High surface tension or viscosity makes cohesive forces predominant. Single droplets Satellitedroplets Single droplets Satellitedroplets 37 Fig. 3-6. Geometrical parameters: (a) relative chamber X1/X2, (b) aspect ratio d/h and (c) diffuser. Fig. 3-7. Time duration for droplet generation vs.: (a) relative chamber size (A-type) and (b) aspect ratio (B-type & C-type). Fig. 3-8. Time duration for droplet generation vs. driving characteristics of the selected structure (B-type). Microdroplet can be generated at an applied voltage of 9V-21V and frequency above 15 kHz. 38 CHAPTER 4. FABRICATION AND CHARACTERIZATION OF PIEZOELECTRIC MEMS INKJET PRINT HEAD Abstract This report describes the fabrication and characterization of a piezoelectric inkjet print head (PIPH) structure, which is integrated with a reservoir, microchannel, and actuator membrane and fabricated by micro-electro-mechanical-system (MEMS) processing. Sol-gel derived Pb(Zr0.52Ti0.48)O3 thin film was used as a main component of the piezoelectric actuator membrane. And an improvement of inkjet structure (InkjetVer2) was performed with changing the shapes of nozzle inlet and nozzle outlet (orifice). The mechanical properties, such as maximum displacement and resonance frequency of actuator membrane, as well as the ink ejection were monitored. This work helps to verify the simulation results and standardize the fabrication processes. 39 4.1 Introduction The fabrication of electronic and/ or mechanical structures in micron scale is typically an expensive process and a wide variety of materials. An alternate approach is an ink-jet printing technology. Ink-jet printer is capable of depositing ranging from organic materials at cost-effective process [1-5]. The ink-jet print head is classified into continuous and drop-on-demand modes. The drop-on-demand printing is further divided into the followings: thermal, piezoelectric, electrostatic and acoustic actuations. The thermal ink jet printing is a method whereby ink drops are ejected from a nozzle by the growth and collapse of a water vapor bubble on the top surface of a small heater located near the nozzle. The simple design of the thermal ink jet print head and its fabrication process compatible with semiconductor processing allow the print head to be built at low cost but limited to water-based inks. As an alternate method, the piezoelectric ink- jet head typically offer a greater range of ink compatibility than thermal ink-jet head [6]. Moreover, the piezoelectric driving mechanism has several advantages over other mechanisms, such as high torque, fast response and low power consumption compare to other mechanisms. The piezoelectric actuating method also includes various deformation mechanisms such as squeeze, bend, push and shear modes. In this study, the PIPH structure operating in bend-mode was fabricated using two silicon wafers. The first one is used to make actuator membrane and ink chamber and another is used to form nozzle and microchannel. Both of them were bonded by Eutectic bonding technique. In this fabrication of InkjetVer2, the shapes of nozzle inlet and nozzle outlet were modified so that the performance of PIPH could be improved compared with InkjetVer1. In this paper, we describe its fabrication process and characterization including mechanical properties and ink ejection process. 4.2 Experiments Figure 4-1 shows the side view of a PIPH structure. It consists of several components integrated into two bonded silicon wafers, such as actuator membrane, ink chamber, nozzle, diffuser, microchannel and reservoir. Vibration of actuator membrane under an AC applied voltage causes a volume difference of ink chamber which directs 40 the fluid flow in microchannel from reservoir. The bending of actuator membrane, of course, applies pressure to the printing fluid in the chamber, which forces a droplet to be expelled from nozzle outlet (orifice) under favorable conditions. The fabrication process of PIPH required 10 masks (as shown in Fig. 4-2). Masks M1-M6 were used for fabrication of actuator-chamber plate and the rest was used for fabrication of microchannel-nozzle plate. Figure 4-3(a) shows the fabrication process of actuator-chamber plate. The low- temperature oxide (LTO) and low-stress SiNx (1.2 µm) layer were deposited on the double side polished 300 μm-thick p-type (100) 4 inch silicon wafer by low pressure chemical vapor deposition (LPCVD) system. Then 300 nm-thick LTO layer was additionally deposited as a buffer layer by plasma enhanced chemical vapor deposition (PECVD) system for the following deposition process of bottom electrode and PZT thin film. Thin (200 Å) Ta as an adhesion layer and a Pt (1500 Å) metallic layer as a bottom electrode were deposited by DC magnetron sputtering at 350 °C. PZT thin film (0.5 um) was coated on the prepared substrate (Pt/Ta/SiO2/SiNx/SiO2/Si/SiO2/SiNx) by sol-gel spin coating method with synthesized precursor solution, followed by fast annealing at 650 °C for 2 min in the rapid thermal process (RTP) system [8-9]. The PZT and bottom electrode layers were etched by the inductively coupled plasma (ICP) etcher. The ashing process was performed to remove the remaining photoresist after each dry etching process. The inter layer dielectric (ILD) was incorporated for the insulation between top and bottom electrode in the PZT capacitor. The silicon dioxide SiO2 layer (3000 Å), as an ILD, was deposited by PECVD and patterned by RIE. Then, the Ta/Pt thin films (200 Å/1000 Å) were deposited for the top electrode by DC magnetron sputtering. In order to form a vertical ink chamber and actuator membrane, the deep-reactive ion etcher (deep-RIE) was used. The SiNx/SiO2 layer at the bottom side of the actuator plate was etched in order to make a mask window of the deep-RIE process for the ink chamber, followed by the deep-RIE. Figure 4-3(b) shows the fabrication process of the channel-nozzle plate. The SiNx/Si/SiNx substrate was used for this plate. The nozzle inlet had pyramidal shape and 41 was defined by wet chemical etching with KOH solution (KOH: DI water=6:4, or 35% wt) at 80 °C [10]. Then the microchannel and diffuser were defined by the deep-RIE process. The SiNx layer on the bonding surface of the channel-nozzle plate was etched by RIE in order to expose the Si surface for the eutectic bonding process. The Ta/Au (200 Å /3000 Å) thin films were used as glue layers between top plate (actuator-chamber plate) and bottom plate (channel-nozzle plate) and were deposited by magnetron sputtering/e-beam evaporator on each bonding surface. Both plates were bonded by EV501 bonder at temperature of 400°C and an uniform force of 3000 N during 1 hour. 4.3 Results and discussion Figure 4-4 shows the SEM and optical micrographs of the fabricated PIPH structure. Multi-layered actuator membrane was stable without structural deformation due to the film stress or effects of fabrication steps. However, membrane size was exceeded 10~20% because of the DRIE process of chamber from back side. The nozzle inlet had pyramidal shape of silicon wet-etching. The structures of reservoir, channel, diffuser and nozzle outlet were well defined by deep-RIE process. The electrical properties of the PZT actuator membrane incorporated in the ink-jet head structure were measured by an impedance analyzer (HP4194A) and Radiant Technologies testing system (RT66A). The poling process was performed at final step to recover the PZT properties. After poling at 12V, 120°C, 60 min, PZT properties were recovered significantly. Figure 4-5(a) shows the polarization-electric field (P-E) loops of the PZT membrane under applied voltages of 3~12V. The remanent polarization (Pr) was ca.18µC/cm2 (compared with ca. 20µC/cm2 of initial state). The fundamental resonant frequency of the PIPH actuator membrane was ca.270 kHz (as shown in Fig. 4- 5(b)). With actuator membrane fabricated by KOH wet-etching, the membrane size was accurate. And its natural frequency was ca.330 kHz (see chapter 2). In this case PIPH actuator membrane was over-etched 10-20% by DRIE, its natural frequency of 270 kHz was reasonable. This is explained based on the relationship between the membrane’s resonant frequency, f, and its dimension, a, as following: 42 f ~ 1/a2 In order to measure the displacement of the PIPH actuator membrane, a LK-G10- KEYENCE non-contact laser displacement measurement was used. Figure 4-6 shows the maximum displacement (bending of the center point) of PIPH membrane actuator upon applied voltages. With the actuator membrane extended by DRIE process (over- etched 10-20%), the maximum displacement obtained from the simulation and experiment was ca. 0.77-0.112 um/V and 0.098 um/V, respectively. There are 7x12= 84 cells of PIPH integrated on 4-inch silicon wafer. Each cell consists of four PIPH structures. Cells of PIPH were separated by laser dicing (M2000 laser). Figure 4-7(a) shows an image of a PIPH cell. Electrical and fluidic systems were prepared for ejection testing as shown in Fig. 4-7(b). Ejection testing was performed using high speed digital camera system (Fig. 4-8). Ink fluid was supplied by a micropump which could control an accurate fluid flow. When fluid flow ≠ 0 (i.e., 0.5, 2, 10 ul/min), it was observed that there was liquid expelled from nozzle outlet. This confirmed that the possibility of the microchanel, having a diffuser with minimum size of 30 um, and the nozzle, with diameter of 30 um, could make the liquid moved inside the micro-scale system. When fluid flow =0 (stop the supply of external flow), an AC voltage with the amplitude of 5-10V and the frequency of 5-40 kHz was applied to the PIPH actuator membrane. No ejection or droplets of liquid existed. However, a vibration of the liquid meniscus was observed at the nozzle outlet (Fig. 4-9). This implied that the operation of PIPH actuator could not create enough work or “dynamic force” to expel the liquid out of the orifice. 4.4 Conclusion Piezoelectric inkjet print head was successfully fabricated by MEMS processing. The piezoelectric properties were maintained after a lot of fabrication steps. The existence of the motion of PIPH actuator membrane as well as the movement of liquid inside microchannel of PIPH structure exhibited the reality of the fabrication of piezoelectric MEMS inkjet. Although the ejection of liquid could not happened with current design of PIPH (InkjetVer2), the successful fabrication and its good mechanical 43 properties played an important role in confirming the simulation results and standardizing the process for a new design (InkjetVer3), which promises the possibility of mircrodroplet generation. 44 4.5 Rerefences [1] J. Priest, E. Jacobs, C. Smith, Jr., P. DuBois, B. Holt, and B. Hammerschlag, J. Microcircuits Electron. Packag, 17, 219–227 (1994). [2] D. J. Hayes,W. R. Cox, and D. B. Wallace, SPIE Micromachining & Microfabrication Conf, 22–25 (2001). [3] G. Duthaler, Master’s thesis, Massachusetts Institute of Technology, Cambridge (1995). [4] D. J. Hayes, D. B. Wallace, and M. T. Boldman, ISHM. (1992). [5] D. J. Hayes,W. R. Cox, and M. E. Grove J. Electron. Manufact, 8, 209–216 (1998). [6] J. Heinzl and C. H. Hertz, Advances in Electronics and Electron Physics, 65, 91–171 (1985). [7] Jun-Kyu Paik, Sanghun Shin, Sun-Woong Na, Nae-Eung Lee and Jaichan Lee, Ferroelectrics, 69, 383-390 (2005). [8] Caroline S. Lee, H.J. Nam, Y.S. Kim, W.H. Jin and J.U. Bu, J. Korean Phys. Soc, 45, 227 (2004). [9] Sanghun Shin, Jun-kyu Paik, Nae-eung Lee, Jaichan Lee, Jun-shik Park and Hyo- derk Park, J. Korean Phys. Soc, 46, 292-295 (2005). [10] Minhee Yun, J. Korean Phys. Soc, 37, 605~610 (2000). 45 Fig. 4-1. Schematic of piezoelectric inkjet print head structure (side view): (a) Inkjet version 1 and (b) Inkjet version 2 with the modified nozzle shape at locations marked 1 &2. Fig. 4-2. Masks used for fabrication of PIPH : M1-M6 (wafer 1) and M7- M10 (wafer2). 46 Fig.4-3. Fabrication process flow of PIPH: (a) wafer 1-actuator and chamber and (b) wafer 2-channel and nozzle. Both wafers are bonded by Eutectic bonding method. (a) Actuator and chamber (b) Channel and nozzle 47 Fig. 4-4. SEM and optical micrographs of the fabricated PIPH structure. Fig.4-5. Electrical properties of PIPH actuator membrane: (a) polarization-electric field (P-E) hysteresis loops at final step (Pr~18 uC/cm2) and (b) fundamental resonant frequency (membrane size was over-etched 10%-20% (330 ~ 360 um) by DRIE, 300 um-wide membrane fabricated by wet etching had f~ 330 kHz, agreed with f ~ 1/a2). 48 Fig. 4-6. Displacement of PIPH actuator membrane (membrane size was over-etched 10%-20% (330 ~360 um) by DRIE): (a) simulation 0.77~0.112 um/V and (b) experiment 0.098 um/V. Fig. 4-5. Preparing for ejection test: (a) 4-inkjet heads on 1 cell and (b) PCB-wire bonding and tube attachment. 49 Fig. 4-6. Ejection testing by high speed digital camera system. Fig. 4-7. Meniscus vibration under an applied voltage of 10V-40 kHz. 50 CHAPTER 5. CONCLUSION AND SUGGESTION 5.1 Conclusion Simulation, fabrication and characterization of piezoelectric MEMS inkjet print head have been investigated. Total analysis of the PIPH, including actuator performance characteristics and microdroplet generation, was performed in the simulation work with the aid of numerical tools (i.e., FEMLAB, CFD-ACE+). Basic performance characteristics of a work-producing PIPH actuator (i.e., the maximum actuation displacement, maximum actuation force and limited driving frequency (first resonant frequency)) were analyzed in details (chapter 2). Microdroplet generation and influences on it (such as effects of driving characteristics, fluid properties and geometrical parameters) were also considered carefully (chapter 3). The fabrication of the PIPH was carried out using MEMS processing. And the characterization of the PIPH was employed to confirm the simulation results as well as standardize the fabrication processes (chapter 4). The review of the simulation and fabrication offers a good guideline and an effective description in designing piezoelectric MEMS inkjet print head with possibility of droplet generation and high quality printing. 5.2 Suggestion (new design) Total analysis of the PIPH has been investigated during this thesis. A new design (InkjetVer3) is suggested as an candidate which promises good performance characteristics of PIPH actuator and possibility of mirodroplet generation. InkjetVer3 can be fabricated from three silicon wafers. Some main components (with indicated geometrical parameters) of InkjetVer3 are an actuator membrane with size of 600 um, two chambers with ratio of relative chamber size of 0.5 (300 um/600 um), a multi- diffuser microchannel with the minimum size of 30 um, a nozzle with pyramidal inlet and cylindrical outlet (diameter of 30 um) and a reservoir with size of 600 um. Model of InkjetVer3 and masks used to its fabrication are shown in Fig. 5-1 and Fig. 5-2, respectively. Fabrication of InkjetVer3 has been proceeding. 51 Fig. 5-1. Model of InkjetVer3 (3-silicon wafers). Fig. 5-2. Masks used for fabrication of InkjetVer3. 52 Appendix A. Python Source Script for simulation of microdroplet generation (effects of driving characteristics and fluid properties) # This file was generated by CFD-GEOM import GPoint import GCurve import GSurface import GLoop import GEdge import GFace import GFilament import GBlock2D import GBlock import GEntity import GManip import GGrid import GUnstruct import GInterface # Set Geometric Precision GEntity.SetPrecision ( 1E-006 ) a = 300 GEntity.CreateParameter (globals (), 'a') b = 360 GEntity.CreateParameter (globals (), 'b') c = 135 GEntity.CreateParameter (globals (), 'c') d = a-2*c GEntity.CreateParameter (globals (), 'd') e = 226+14 GEntity.CreateParameter (globals (), 'e') f = 1.2*b GEntity.CreateParameter (globals (), 'f') cgrid=15 dgrid=8 ggrid=25 egrid=20 fgrid=int(egrid*f/b) geom_point1 = GPoint.Create (0, 0, 0) geom_point2 = GPoint.Create (a, 0, 0) geom_point3= GPoint.Create (c, 0, 0) geom_point4 = GPoint.Create (c+d, 0, 0) geom_point5 = GPoint.Create (0, b, 0) geom_point6= GPoint.Create (a, b, 0) geom_point7= GPoint.Create (c, b, 0) geom_point8= GPoint.Create (c+d, b, 0) geom_point9= GPoint.Create (0, b+e, 0) geom_point10= GPoint.Create (a, b+e, 0) geom_point11= GPoint.Create (c, b+e, 0) geom_point12= GPoint.Create (c+d, b+e, 0) geom_point13 = GPoint.Create (0, b+e+f, 0) geom_point14 = GPoint.Create (a, b+e+f, 0) geom_point15 = GPoint.Create (c, b+e+f, 0) geom_point16 = GPoint.Create (c+d,b+e+f, 0) geom_curve1 = GCurve.CreateThroughPoints (geom_point1, geom_point3) geom_curve2 = GCurve.CreateThroughPoints (geom_point3, geom_point4) geom_curve3 = GCurve.CreateThroughPoints (geom_point4, geom_point2) geom_curve4 = GCurve.CreateThroughPoints (geom_point2, geom_point6) geom_curve5 = GCurve.CreateThroughPoints (geom_point6, geom_point8) geom_curve6 = GCurve.CreateThroughPoints (geom_point8, geom_point7) geom_curve7 = GCurve.CreateThroughPoints (geom_point7, geom_point5) geom_curve8 = GCurve.CreateThroughPoints (geom_point5, geom_point1) 53 geom_curve9 = GCurve.CreateThroughPoints (geom_point7, geom_point11) geom_curve10 = GCurve.CreateThroughPoints (geom_point11, geom_point12) geom_curve11 = GCurve.CreateThroughPoints (geom_point12, geom_point8) geom_curve12 = GCurve.CreateThroughPoints (geom_point12, geom_point10) geom_curve13 = GCurve.CreateThroughPoints (geom_point10, geom_point14) geom_curve14 = GCurve.CreateThroughPoints (geom_point14, geom_point16) geom_curve15 = GCurve.CreateThroughPoints (geom_point16, geom_point15) geom_curve16 = GCurve.CreateThroughPoints (geom_point15, geom_point13) geom_curve17 = GCurve.CreateThroughPoints (geom_point13, geom_point9) geom_curve18 = GCurve.CreateThroughPoints (geom_point9, geom_point11) #d -gird geom_edge1 = GEdge.CreatePowerLaw (geom_curve2,dgrid, 1.000000, 1) geom_edge2 = GEdge.CreatePowerLaw (geom_curve6, dgrid, 1.000000, 1) geom_edge3 = GEdge.CreatePowerLaw (geom_curve10, dgrid, 1.000000, 1) geom_edge4 = GEdge.CreatePowerLaw (geom_curve15, dgrid, 1.000000, 1) #c-grid geom_edge5 = GEdge.CreatePowerLaw (geom_curve1, cgrid, 1.000000, 1) geom_edge6 = GEdge.CreatePowerLaw (geom_curve3, cgrid, 1.000000, 1) geom_edge7 = GEdge.CreatePowerLaw (geom_curve7, cgrid, 1.000000, 1) geom_edge8 = GEdge.CreatePowerLaw (geom_curve5, cgrid, 1.000000, 1) geom_edge9 = GEdge.CreatePowerLaw (geom_curve12, cgrid, 1.000000, 1) geom_edge10 = GEdge.CreatePowerLaw (geom_curve18, cgrid, 1.000000, 1) geom_edge11 = GEdge.CreatePowerLaw (geom_curve14, cgrid, 1.000000, 1) geom_edge12 = GEdge.CreatePowerLaw (geom_curve16, cgrid, 1.000000, 1) #b-grid geom_edge13 = GEdge.CreatePowerLaw (geom_curve8, bgrid, 1.000000, 1) geom_edge14 = GEdge.CreatePowerLaw (geom_curve4, bgrid, 1.000000, 1) #e-grid geom_edge15 = GEdge.CreatePowerLaw (geom_curve9, egrid, 1.000000, 1) geom_edge16 = GEdge.CreatePowerLaw (geom_curve11, egrid, 1.000000, 1) #f-grid geom_edge17 = GEdge.CreatePowerLaw (geom_curve17, fgrid, 1.000000, 1) geom_edge18 = GEdge.CreatePowerLaw (geom_curve13, fgrid, 1.000000, 1) #meshing region 1 geom_face1 = GFace.Create ([geom_edge5,geom_edge1,geom_edge6],[geom_edge8,geom_edge2,geom_edge7],geom_edge14,geom_ edge13) geom_block2d1 = GBlock2D.Create2D (geom_face1) #meshing region 2 geom_face2 = GFace.Create (geom_edge2,geom_edge15,geom_edge16,geom_edge3) geom_block2d2 = GBlock2D.Create2D (geom_face2) #meshing region 3 geom_face3 = GFace.Create ([geom_edge10,geom_edge3,geom_edge9],geom_edge18,[geom_edge11,_ _geom_edge4,geom_edge12],geom_edge17) geom_block2d3 = GBlock2D.Create2D (geom_face3) GInterface.DTFWrite_2d ('VOF.DTF') 54 Appendix B. Pattern conditions for fabrication of Inkjetver2 AOE (PZT, BE), RIE (SiNx) - PR AZ1512 spin coating 500rpm/5’’, 2000 rpm/35’’(~2um) - Soft bake: oven 110oC, 3 min - Exposure: 14 sec - Develop: ~ 1min - Post bake: oven 110oC, 1 min Lift-off (Ta/Pt TE, Key 1) - PR AZ1512 spin coating 500rpm/5’’, 3500 rpm/35’’(~1.5um) - Soft bake: oven 110oC, 3 min - MCB : 3 min - Bake : oven 110oC, 2 min - Exposure: 14 sec - Develop: ~ 1min DRIE (chamber, diff, nozzle) - PR AZ9260 spin coating 500rpm/5’’,1000rpm/40’’ (15um) - Bake: hot plate 110oC, 2 min - Exposure : 12 mW/cm2, 180 sec (STD) - Develop : AZ500MIF, ~5-6 sec - Post bake: 1 min (Oven) 55 Appendix C. Dry etching conditions Advanced Oxidation Etcher (AOE) PZT (500 nm) Pt (1500 Å), Ta (200 Å Gas CF4 – 5 sccm Ar – 20 sccm C4F8 – 5 sccm C2H4 – 10 sccm He – 15 sccm Ar – 10 sccm Power Top power – 1200 W Bottom power - 400 W Bias – 400W Top power – 800 W Bottom power- 500 W Bias – 500W Pressure 5 mTorr 3 mTorr Time 3 min 10 sec 1 min 20 sec RIE (SiO2, SiNx) DRIE (Si) Gas O2 – 4 sccm CF4CHF3 – 36 sccm C4F8 - 150 sccm (passivation) O2 – 30 sccm SF6 – 300 sccm He - 9790 mTor (cooling) Power 190-230 W (RF generator) Top power – 1200 W Bottom power - 400 W Bias – 400W Pressure 40 mTor 45 mTor (SF6-O2) 18 mTor (C4F8) Etching rate 480 Å/min 60 um/min