Design of piezoelectric micromachined ultrasonic transducers (pmuts) for high pressure output

Transcription

1 Microsyst Technol (2017) 23: DOI /s TECHNICAL PAPER Design of piezoelectric micromachined ultrasonic transducers (pmuts) for high pressure output Mingjun Wang 1 Yufeng Zhou 1,2 Received: 20 November 2015 / Accepted: 23 March 2016 / Published online: 8 April 2016 Springer-Verlag Berlin Heidelberg 2016 Abstract A novel design of piezoelectric micromachined ultrasonic transducers (pmuts) with a fully free edge structure by introducing a deep trench between cells was proposed, and its performance was evaluated by finite element method in both time and frequency domains for the resonant frequency and acoustic pressure output. In comparison to current cell configurations of clamped and simply suspended boundaries, our design has a great increase in the output pressure. The effect of the covering area of piezoelectric material (i.e., AlN) was also investigated. It is found that when the piezoelectric layer has the same size as the cavity the generated acoustic pressure will reach its maximum value. Altogether, the newly designed structure of pmuts has the potentials of high pressure applications (i.e., ultrasound therapy) if driven properly. 1 Introduction Ultrasound transducers made of bulk piezoelectric materials with a variety of shapes have been dominant ultrasound applications for the past 50 years (Wong et al. 2008), but they have restricted applications in some special conditions such as endoscopic ablation which has strict requirements on the probe size and the output pressures. Microelectro-mechanical system (MEMS) device has been extensively studied and applied in various aspects, such as * Mingjun Wang WANG0683@e.ntu.edu.sg 1 2 School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore, Singapore Key Laboratory of Modern Acoustics, Nanjing University, Nanjing, China radio frequency (RF) communication, gyroscope, Lamb wave resonators and film bulk acoustic resonator (FBAR), because of its batch fabrication of high-density transducer array as well as single elements and a high level of integration and scalability for a wide range of devices at different frequency spans and regimes (Guedes et al. 2011). Currently the piezoelectric based micromachined transducer has become attractive in acoustic MEMS device for its promising application in medical imaging (Yang et al. 2013; Wang et al. 2011; Dausch et al. 2010) and therapeutic purposes (Nyguyen-Dinh et al. 2012). Capacitive micromachined ultrasonic transducers (cmuts) are made of small and thin membranes (tens to hundreds of micrometers) that are suspended over a conductive silicon substrate by insulating posts to essentially create small capacitors. This structure results in very efficient transducers that can compete with the piezoelectric counterparts in terms of efficiency and bandwidth. With the advancement of the fabrication technology and improved design methodology, the cmuts for high power application (1.8 MPa output pressure) have already been fabricated and tested (Yamaner et al. 2012). Furthermore, cmuts could also work in a dual mode (both diagnosis and therapy) due to its wide bandwidth and high output pressure (Meynier et al. 2012). Piezoelectric micromachined ultrasonic transducers (pmuts) use a piezoelectric layer adhered to an inert membrane and operating in flexural mode, exploiting the piezoelectric coefficient d 31 for vibration (Chao et al. 2006). By applying an electrical field to the film, the resulting strain (due to the direct piezoelectric effect) causes the structure to bend (Akasheh et al. 2004). In comparison to cmuts, pmuts have advantages of low driving voltage, very high sensitivity and simple fabrication process (Sammoura et al. 2013). It is well known that the output pressure of cmuts is highly dependent on the separation distance of two

2 1762 Microsyst Technol (2017) 23: electrodes and the applied voltage. However, the high driving voltage may break down the insulation layer and lead to static charge accumulation (Shelton et al. 2009). The output pressure of pmuts, nevertheless, could be easily manipulated because the generated acoustic pressure only relies on the material used, the boundary condition of the membrane and the excitation voltage. But most research on pmuts focuses on its application in ultrasound diagnosis, such as catheter probe for real time 3D imaging (Yang et al. 2013; Wang et al. 2011; Dausch et al. 2010). Although pmut has been tried for high pressure output, the acoustic pressure generated in the air was only around 2 kpa (Nyguyen-Dinh et al. 2012), which is far from the therapeutic requirement. In this paper, the feasibility of pmuts particularly for high pressure application was investigated using finite element method (FEM) simulation. To further increase the output pressures, the traditional clamped (Chao et al. 2006) and flexurally-suspended membrane (Guedes et al. 2011; Muralt et al. 2005) has been modified to a totally free edge boundary condition. Simulation results of an aluminum nitride (AlN) based pmut show that such a modification could improve the output pressure significantly. In summary, our novel design of pmuts has the capability of high pressure output and the potentials in the ultrasound therapy such as high-intensity focused ultrasound (HIFU) ablation for cancer and tumor. 2 Design of AlN based pmuts Aluminum nitride (AlN) was chosen as the piezoelectric material because of its metal-oxide-semiconductor (CMOS) compatibilities (Shelton et al. 2009), although its piezoelectric coefficient is relatively smaller than the other piezoelectric material such as ZnO or lead zirconate titanate (PZT). The schematic diagram of AlN-based pmuts is shown in Fig. 1. pmuts usually consist of a membrane of piezoelectric material, a supporting silicon layer and a cavity. A pure AC signal is applied to the top electrode while the bottom one is grounded. Due to the piezoelectric effect, driving signals cause the piezoelectric layer and the attached supporting layer to vibrate, subsequently Fig. 1 Schematic diagram of the AlN-based pmut. Si Silicon, SiO 2 silicon dioxide, Al aluminium, AlN aluminium nitride generating acoustic waves to the surrounding medium. The design goal for high pressure pmuts is to increase the displacement of the membrane. In order to achieve it the piezoelectric layer and the attached silicon layer are isolated from the neighbors by introducing a deep trench around each cell, which can free the membrane in the dynamics and reduce the cross-talks. It is noted that the trenches used in this design are conceptually different from those in the flexurally-suspended mode, in which the trench was located at the inner part of the cavity and stopped at the oxide layer (Guedes et al. 2011; Muralt et al. 2005). Another key parameter in the design of pmuts is the size of the membrane, including the piezoelectric and supporting layer. Theoretically, a thin supporting layer is preferred because of the high flexibility. However, a too thin layer may break down during the operation, and it is technically difficult to fabricate a layer with thickness no more than 0.1 μm (Nakagawa et al. 1991). Considering the factors above and the specifications of commercially available silicon on insulator (SOI) wafer, 5 µm standard silicon supporting layers is used, which is the trade-in between a relatively good performance and a simple fabrication process. The size of the membrane determines the resonant frequency of pmuts, which is different from the popular thickness mode of the bulk piezoelectric material. In addition, the multilayered structure of pmuts provides flexibility of tuning the operating frequencies by adjusting the size ratio of piezoelectric to silicon layer (Sammoura and Kim 2012). The piezoelectric layer mainly acts as a driver to excite the silicon membrane into vibration instead of being an acoustic emitter itself (Choi et al. 2010). Therefore, 1 µm AlN was used in this study mostly owing to the fabrication convenience (see Fig. 2). 3 Simulation and results Performance of the novel pmuts design and the effect of membrane size on the output pressure was simulated using a commercial FEM software (COMSOL 4.2A, Burlington, MA, USA). The piezoelectric module and pressure acoustic module were coupled by setting the acceleration at the surface of the membrane as the acoustic source of the pressure subdomain and the acoustic pressure as the loadings for the piezoelectric subdomain. The boundary of the pressure domain was set as spherical radiation wave, which means no reflection but only transmission occurring on it. Because of the geometrical symmetry, 1.5D model was used in the simulation to tremendously reduce the computation time. The material property and parameter values used in the simulation are listed in Table 1. In order to determine the resonant frequency of the pmuts immersed in water a static analysis followed by

3 Microsyst Technol (2017) 23: Fig. 2 Finite element model for simulation of pmuts Table 1 Materials and sizes of pmuts used in the simulation Isolation silicon thickness 5 µm Cavity depth 20 µm Cavity radius 45 µm Substrate material Silicon Electrode material Aluminium Thickness of electrode 0.2 µm eigen-frequency analysis were carried out in the simulation with no driving signal applied to it. One of the results, pmuts with the radius of 50 µm, is shown in Fig. 5b. It is found that the resonant frequency has a complicated relationship with the size of the membrane (Fig. 3). The maximum resonance existed at the radius of 50 µm. When the cell has a smaller radius, the membrane vibrates at the mode of (0, 1), whose resonance frequency is almost constant, 5.8 MHz (Benaroya 2006). The mode will change to asymmetric mode of (1, 1) if the radius is larger than 50 µm, whose resonant frequency decreases with the increase of the membrane radius (Benaroya 2006). The pressure simulation was conducted by a transient analysis with a sinusoidal driving signal being applied onto the top electrode of the membrane. The average pressure across the surface of the cell was calculated as the output pressure. A peak-to-peak acoustic pressure of 3.0 MPa was obtained for the membrane with radius of 50 µm at a driving voltage of 100 V with excellent symmetry on its compressional and rarefractional pressures (see Fig. 4). It usually takes about 5 6 cycles to reach the steady pressure output. Although the cmuts can achieve the output of 1.8 MPa, its large radius (280 µm) and high driving voltage (125 V) may not outweigh our novel pmut structure (Yamaner et al. 2012). The effect of boundary conditions of the membrane on the performance of pmut is an important concern in the Fig. 3 Relationship between the resonant frequency and the radius of pmut Fig. 4 Acoustic pressure output of pmut with a membrane of 50 µm in radius driven by 100 V continuous wave at its resonant frequency of 6.2 MHz design. The following conditions have been considered: (1) clamp on both the piezoelectric layer and the supporting layer, (2) fix the supporting layer but free the piezoelectric layer, (3) free both the supporting layer and the piezoelectric layer while gradually reducing the size of AlN. Figure 5 shows the responses of the membrane driven by 100 V AC signal, and the simulated results are summarized in Table 2.

4 1764 Microsyst Technol (2017) 23: Fig. 5 The resonant frequency and pressure of pmuts with 50 µm in radius at for various boundary conditions: a fully clamp both the piezoelectric and silicon layer, b fix the silicon layer but free the piezoelectric layer, c g free both the silicon and the piezoelectric layer with radius of 50, 45, 40, 35 and 30 µm, respectively

5 Microsyst Technol (2017) 23: Table 2 Simulation results of pmuts at various boundary conditions Boundary conditions It shows that the vibration mode of the membrane is dependent on the boundary condition. Figure 5a and b have a mode of (0, 3) while the rest is in the mode of (0, 1). The mode shape of (0, 3) has a much smaller average displacement and subsequently low output pressure. Therefore, in order to obtain high output, the boundary condition should be elaborately studied for the appropriate mode. The free edge boundary condition has a 6.8-fold increase of output pressure compared to the partially clamped one (silicon fixed and AlN free). In addition, coverage of piezoelectric layer also has an influence on the acoustic pressure. As the AlN layer goes from fully covered (Fig. 5c) into partially covered (Fig. 5d g), the output pressure first increases from 1.5 to 1.7 MPa (45 μm, exactly the cavity radius) and then gradually reduces to 1.2 MPa. The optimized geometry resembles the flexurally-suspended design. 4 Discussions Resonant frequency (MHz) Both clamped Silicon fixed, AlN free Both free, AlN (50 µm in radius) Both free, AlN (45 µm in radius) Both free, AlN (40 µm in radius) Both free, AlN (35 µm in radius) Both free, AlN (30 µm in radius) Output pressure (MPa) In this study, the resonant frequency and the output pressure for a newly proposed pmuts were investigated by finite element method. If our pmuts are working in the suspended model (Guedes et al. 2011; Muralt et al. 2005), it has a resonant frequency of 4.81 MHz and output pressure of 1.25 MPa, which is below our optimal design. In the flexurally-suspended design, the stiffness of the membrane is mainly determined by the ring-happed suspending silicon oxide layer, which is usually µm in thickness (Chao et al. 2006; Muralt et al. 2005). Since the silicon oxide has a much smaller Young s modulus (80 GPa) than that of the AlN (340 GPa), when the electrical field was applied, the induced stress can only cause the overlapped membrane to bend. In pulse-echo B-mode imaging, this approach could enhance the transmission efficiency (Dausch et al. 2010). But for large amplitude vibration, although this design can reinforce the uniform output, vibrating amplitude will be limited. If the membrane is completely clamped at the periphery, resonant frequency will be relatively high. However, acoustic pressure will be seriously affected because the central part of the membrane has the largest displacement. It has been shown that the flexurally-suspended membrane has almost twice of the output pressure as the clamped one in the air (Guedes et al. 2011). In our new design, pmut has a higher resonant frequency than the suspended design. The presence of the maximum resonant frequency may be due to the mode transition. Vibration mode depends on the membrane size, and there is a nonlinear relationship between the stiffness of the membrane and the size in the multiple layered membranes, which was also observed in flexure-mode pmut arrays (Nyguyen-Dinh et al. 2012). Trench geometry (depth and distance to the side wall of the cavity) has a great impact on the performance of pmut. Further investigation in theoretical simulation and experimental measurement is in a great need to enhance our understanding in order to optimize the design. For cmuts in higher power transmission, the driving signal is usually at half of the resonant frequency due to the fact that the electrostatic force is proportional to the square of the driving signal. However, the stress caused by the piezoelectric effect is linear proportional to the frequency used by the driving signal. Thus, the working frequency should be set at the resonant frequency for pmuts. Furthermore, linear theoretical models, based on the assumption of small amplitude vibration, are used to simulate the performance of either cmuts or pmuts at high output. In such a case, nonlinearity effects may be considered for more accuracy (Benaroya 2006), which will be investigated in future work. 5 Conclusion Introduction of deep trenches between pmut cells would completely free the boundaries of both piezoelectric and silicon supporting layers. As a result, the acoustic pressure could be increased significantly in comparison to the conventional clamped and suspended conditions. Adjusting the size of the piezoelectric layer provides a strategy of tuning the operating frequency of payments. High pressure output makes the application of pmuts in the ultrasound therapeutic ablation possible. Device fabrication and characterization will be carried out in the near future to prove this novel design.

6 1766 Microsyst Technol (2017) 23: References Akasheh F, Myers T, Fraser JD, Bose S, Bandyopadhyay A (2004) Development of piezoelectric micromachined ultrasonic transducers. Sens Actuators A Phys 111(2 3): Benaroya H (2006) Mechanical vibration, analysis, uncertainties and control. Marcel Dekker, New York Chao C, Lam TY, Kwok K-W, Chan HLW (2006) Piezoelectric micromachined ultrasonic transducers based on P(VDF-TrFE) copolymer thin films. In: 15th IEEE international symposium on the application of ferroelectrics, pp 1 4 Choi H, Aderson MJ, Ding JL, Bandyopadhyay A (2010) A twodimensional electromechanical composite plate model for piezoelectric micromachined ultrasonic transducer (pmuts). J Micromech Microeng 20(1): Dausch DE, Gilchrist KH, Carlson JR, Castellucci JB, Chou DR, von Ramm OT (2010) Improved pulse-echo imaging performance for flexure-mode pmut arrays. In: 2010 IEEE international ultrasonics symposium, pp Guedes A, Shelton S, Przybyla R, Izyumin I, Boser B, Horsley DA (2011) Aluminum nitride PMUT based on a flexurally-suspended membrane. In: 16th International Conference on Solid- State Sensors, Actuators and Microsystems (TRANSDUCERS), Beijing, 2011, pp Meynier C, Yamaner FY, Canney M, Nyguyen-Dinh A, Carpentier A, Chapelon J-Y (2012) Performance assessment of CMUTs in dual modality imaging/hifu applications. In: 2012 IEEE international ultrasonics symposium, pp Muralt P, Ledermann N, Baborowski J, Barzegar A, Gentil S, Belgacem B, Petitgrand S, Bosseboeuf A, Setter V (2005) Piezoelectric micromachined ultrasonic transducer based on PZT thin films. IEEE Trans Ultrason Ferroelectr Freq Control 52(10): Nakagawa A, Yasuhara N, Baba Y (1991) Breakdown voltage enhancement for devices on thin silicon layered silicon dioxide film. IEEE Trans Ultrason Ferroelectr Freq Control 38(7): Nyguyen-Dinh A, Wang D, Meynier C, Tyholdt F, Vogl A, Tofteberg H, Østbø NP, Flesch E (2012) pmut for high intensity focused ultrasound. In: IEEE international ultrasonics symposium, pp 1 5 Sammoura F, Kim S-G (2012) Theoretical modeling and equivalent electric circuit of a bimorph piezoelectric micromachined ultrasonic transducer. IEEE Trans Ultrason Ferroelectr Freq Control 59(5): Sammoura F, Smyth K, Kim S-G (2013) Optimizing the electrode size of circular bimorph plates with different boundary conditions for maximum deflection of piezoelectric micromachined ultrasonic transducers. Ultrasonics 53(2): Shelton S, Chan M-L, Park H, Horsley D, Bernhard B, Izyumin I, Przybyla R, Frey T, Judy M, Nunan K, Sammoura F, Yang K (2009) CMOS-compatible AlN piezoelectric micromachined ultrasonic transducers. In: IEEE international ultrasonics symposium, pp Wang Y-F, Ren T-L, Yang Y, Chen H, Zhou C-J, Wang L-G, Liu L-T (2011) High-density PMUT array for 3-D ultrasonic imaging based on reverse-bonding structure. MEMS Wong SH, Watkins RD, Kupnik M, Pauly KB, Khuri-Yakub BT (2008) Feasibility of MR-temperature mapping of ultrasonic heating from a cmut. IEEE Trans Ultrason Ferroelectr Freq Control 55(5): Yamaner FY, Olcum S, Oguz HK, Bozkurt A, Koymen H, Atalar A (2012) High-power CMUTs: design and experimental verification. IEEE Trans Ultrason Ferroelectr Freq Control 59(6): Yang Y, Tian H, Wang Y-F, Shu Y, Zhou C-J, Sun H, Zhang C-H, Chen H, Ren T-L (2013) An ultra-high element density pmut array with low crosstalk for 3-D medical imaging. Sensors 13(8):