A Novel Method for Evaluating the Thickness of Silicon Membrane Using a Micromachined Acoustic Wave Sensor

Transcription

1 Tamkang Journal of Science and Engineering, Vol. 7, No. 2, pp (2004) 61 A Novel Method for Evaluating the Thickness of licon Membrane Using a Micromachined Acoustic Wave Sensor Chi-Yuan Lee 1 *, Ying-Chou Cheng 1, Tsung-Tsong Wu 2, Yung-Yu Chen 2, Wen-Jong Chen 2, Shih-Yung Pao 2, Pei-Zen Chang 2, Ping-Hei Chen 1, Kai-Hsiang Yen 2 and Fu-Yuan Xiao 2 1 Department of Mechanical Engineering National Taiwan University Taipei, Taiwan 106, R.O.C. 2 Institute of Applied Mechanics National Taiwan University Taipei, Taiwan 106, R.O.C. leecyu@mems.iam.ntu.edu.tw Abstract This work presents a novel method based on the micromachined acoustic wave sensor for evaluating silicon membrane thickness. Like pressure sensors, accelerometers, micro flow sensors and micropumps, many micro-electro-mechanical systems (MEMS) devices require that silicon membrane thickness be known exactly. Precisely controlling silicon membrane thickness during wet etching is important, because the thickness strongly affects device performance and post-processing. The proposed method for evaluating silicon membrane thickness is novel, simple to implement, and can be monitored in-situ and mass-produced. The spectral analysis of surface waves (SASW), detailed process flows, measurement set-up and the experimental results also are presented. Key Words: MEMS, Micromachined Acoustic Wave Sensor, SASW 1. Introduction Anisotropic wet etching is a key technology for fabricating microstructures on a single crystal silicon wafer [1]. Like pressure sensors, accelerometers, micro flow sensors and micropumps, many micro-electro-mechanical systems (MEMS) devices require that silicon membrane thickness be known exactly. Precisely controlling silicon membrane thickness during wet etching is important, because membrane thickness significantly influences device performance and post-processing. licon can be thinned using various techniques, including wet etching [2], dry etching [3] and mechanical polishing. Wet etching, which uses numerous anisotropic etchants, inclu- *Corresponding author ding potassium hydroxide (KOH), tetramethyl ammonium hydroxide (TMAH) and ethylenediamine-pyrocatechol (EDP), have been used [4] as a method that has been effectively implemented in mass production while dry etching precisely controls the thickness of silicon membrane. The advantages and disadvantages of wet and dry etching are well known [2]. A novel method, which differs from anyone described in previous works on etch-stop techniques [5 8], was developed herein for evaluating silicon membrane thickness. Whether ranging from a few ten µm to hundreds of µm, this simple setup has the capacity for in-situ monitoring and mass production. The spectral analysis of surface waves (SASW), detailed process flows, measurement set-up and experimental results are also presented.

2 62 Chi-Yuan Lee et al Al Intensity ? = I = 165,238 Figure 1. Schematic diagram of a surface acoustic wave sensor um θ (degree) Figure 3. X-ray diffraction scans of a film. Figure 2. A film with µm thick was confirmed by a surface profiler. 2. Design of Surface Acoustic Wave Sensor Surface acoustic wave (SAW), also called the Rayleigh wave, is essentially a couplin g between longitudinal and shear waves. The energy carried by the SAW is confined near the surface. An electrostatic wave associated with a SAW exists on a piezoelectric substrate, enabling electro-acoustic coupling via a transducer [9]. Zinc oxide () films have widely played an important role in the field of micro-electro-mechanical systems (MEMS) because of their good piezoelectric performance and excellent bonding on various substrate materials [10]. Hence, was selected as a piezoelectric film layer in this work. A surface acoustic wave sensor was made on a piezoelectric film based on a double polished (100) silicon substrate using three-ports of the interdigital transducers (IDT), as illustrated in Figure 1. A thin film was deposited of optimized µm thick by sputtering technology, as illustrated in Figure 2. Figure 3 shows the X-ray diffraction (XRD) scans of a film, which indicates that a film exhibits a good c-axis ori- (a) Top view (b) oss-sectional view Figure 4. SEM images of a film. entation. Figure 4 displays the scanning electron microscope (SEM) images. The top view in Figure 4(a) shows the good uniformity and compactness of the grains.

3 A Novel Method for Evaluating the Thickness of licon Membrane Using a Micromachined Acoustic Wave Sensor 63 Table 1 IDT design parameters Interdigital transducer (IDT) value Para meter Periodicity of IDT (?) 320 µm Length of IDT fingers (L) 9000 µm Overlap length of IDT fingers (W) 8000 µm Path length of acoustic propagation (l 1 ) 6000 µm Path length of acoustic propagation (l 2 ) µm Numbers of pairs (N) 3 Pad area 300 µm 300 µm parameters of the IDT design used in this experiment. L Figure 5. AFM micrographs of a film. λ l 1 Output IDT Piezoelectric substrate Input IDT Figure 6. The configuration of the IDT. l 2 Pad The cross-sectional view in Figure 4(b) indicates that, columnar grains are clear and vertical to the substrate with a slight angle, indicating that grains have good orientation. Figure 5 shows the atomic force microscope (AFM) scans of a film, which indicates that average roughness of a film surface is less than 50 Å. The piezoelectric effect is described as follows. When the AC voltage was applied to the input IDT and signal voltage variations were subsequently converted into a mechanical acoustic wave, then the others IDT were used as an output receiver to transform mechanical SAW vibrations back into output voltage. The output voltage or wave velocity changes interestingly with silicon thickness. Figure 6 illustrates the configuration of the IDT, where? denotes IDT periodicity; L represents the IDT finger length; W is the length of the overlap among the IDT fingers; and l 1 and l 2 denote the lengths of the acoustic propagation path and N is the number of pairs. Numerous parameters must be specified in the design of IDT. Table 1 lists the W 3. Device Fabrication This surface acoustic wave sensor was fabricated using the process illustrated in Figure 7. The process was begun with a µm thickness of 100 mm diameter double polished (100) silicon wafer. First, on the rear side of the silicon substrate, the Shipley S1813 photoresist was coated using a spinner that rotated at a rate of 1000 rpm for 10 seconds, and then 4000 rpm for 40 seconds. The coated substrate was then softbaked at 90 C for 120 seconds and set to ultraviolet exposure lithographically. To increase the adhesion of an film, a film of about 150 Å thick was evaporated before an film with a thickness of 850 Å was evaporated: both processes involved a thermal evaporator, as shown in Figure 7(a). film was used as an etching mask during wet etching in KOH solution. The unwanted and films were lifted off by soaking the substrate in acetone to form a rear side etching-hole, as shown in Figure 7(b). Care was taken to ensure the success of the lift-off process. Several causes of failure of lift-off may apply, such as a lack of an undercut photoresist pattern, non-vertical evaporation of, poor film adhesion and substrate surface contamination [11]. Then, a film, µm thick, was sputtered by an RF magnetron sputter, as shown in Figure 7(c). A film was prepared using an RF magnetron sputtering system, under the conditions described in Table 2. IDTs were patterned using semiconductor photolithographic techniques, by which Al was deposited to a thickness of 800 Å by thermal evaporation, as indicated in Figure 7(d). Finally, a silicon membrane was generated by soaking the rear side of the substrate with 73 C, 33 wt.% in KOH solution, the etching rate was 0.97 µm per minute, as shown in Figure 7(e).

4 64 Chi-Yuan Lee et al. Table 2 Sputtering conditions (a) Evaporating / film Target, 4 diameter Gas (sccm) Ar/O 2 = 1:1 Deposition pressure (Pa) RF power (W) 200 Substrate temperature ( C) 210 Sputtering rate (Å /hr) 4935 Working distance (cm) 13 RF Out (b) Defining the pattern on the rear side of the etching-hole Trigger R T (c) Sputtering a piezoelectric film (d) Evaporating Al film and pattern defining IDTs along the direction of 45 (e) KOH etch forming silicon membrane. Al Al Figure 7. Process flow of a surface acoustic wave sensor. Figure 8. Experimental setup of a surface acoustic wave sensor. 4. Experimental and Results The spectral analysis of surface waves (SASW) method is a nondestructive method based on the theory of elastic wave propagation in a layer medium [12]. This work adopted the SASW method for surface acoustic wave sensor testing [13]. By knowing the distance between the IDTs and the phase difference at each frequency, the dispersion relationships of phase velocity can be calculated. Figure 8 illustrates the experimental set-up, which involves a 200 MHz Pulser/Receiver machine that generates elastic waves of the SAW sensor and sends a trigger signal that is synchronized to the Leoy 9354CM digitizing oscilloscope. The wave signals in the time domain recorded on the surface of IDTs, which can be processed to obtain the two dispersion curves of phase difference against frequency in the two different thicknesses of silicon, as illustrated in Figure 9. Figure 9 reveals that the signals change owing to different silicon thicknesses. The Alpha-step 500 surface profiler was used to measure the different thicknesses of a silicon membrane, which were

5 A Novel Method for Evaluating the Thickness of licon Membrane Using a Micromachined Acoustic Wave Sensor 65 Phase Difference (radian) thickness Unetch (H=528.1 µm) Etch (H=310.8 µm) µm and µm, as illustrated in Figure Conclusion This study describes a surface acoustic wave sensor for evaluating the thickness of a silicon membrane by wet etching. The proposed method for evaluating the thickness of a silicon membrane is novel, simple to implement, and can be monitored in-situ and mass-produced. The spectral analysis of surface waves (SASW), detailed process flows, measurement set-up and the experimental results are also presented Figure 9. Experimental results of frequency with respect to phase difference um F (MHz) Acknowledgments This work was accomplished with much needed support and the authors would like to thank Yu-Ming Wang, Chih-Wei Liu, Shih-Yang Liu, Dr. Jen-Yi Chen, Dr. Chin-Wei Wu, Dr. Yuh Chung Hu, Dr. Chyan-Chyi Wu, Hun-Lin Chen, Jing-Hung Chiou, Tsung-Wei Huang, Shih-Chen Chang, Chun-Yuan Chi, Chia-Hua Chu, X. Y. Wang of the Institute of Applied Mechanics, National Taiwan University, Professors Lung-Jieh Yang, Ching- Liang Dai and Chien-Liu Chang for their valuable advice and assistance in experiment. In addition, we would, finally, like to thank the NSC Northern Region MEMS Research Center for kindly making their complete research facilities available. References (a) Photograph of an unetched µm thickness of silicon, obtained by a surface profiler um (b) Photograph of an etched µm thickness of silicon, obtained by a surface profiler. Figure 10. Experimental results for thickness of unetched and etched silicon membrane, obtained by a surface profiler. [1] Kovacs, G. T. A., Maluf, N. I. and Petersen, K. E., Bulk Micromachining of licon, Proceedings of the IEEE, pp (1998). [2] Williams, K. R. and Muller, R. S., Etch rates for Micromachining Processing, Journal of Microelectromechanical Systems, Vol. 5, pp (1996). [3] Kiihamaki, J., Kattelus, H., Karttunen, J. and Franssila, S., Depth and Profile Control in Plasma Etched MEMS Structures, Sensors and Actuators, Vol. 82, pp (2000). [4] Fujitsuka, N., Hamaguchi, K., Funabashi, H., Kawasaki, E. and Fukada, T., licon Anisotropic Etching without Attacking Aluminum with and Oxidizing Agent Dissolved in TMAH Solution, The 12th International Conference on Solid State Sensors, Actuators and Microsystems, pp (2003).

6 66 Chi-Yuan Lee et al. [5] Minami, K., Tosaka, H. and Esashi, M., Optical In-tu Monitoring of licon Diaphragm Thickness during Wet Etching, J. Micromech. Microeng., Vol. 5, pp (1995). [6] Madou, M., Fundamentals of Microfabrication, CRC Press (1997). [7] Ashruf, C. M. A., French, P. J., Sarro, P. M., Bressers, P. M. M. C. and Kelly, J. J., Electrochemical Etch Stop Engineering for Bulk Micromachining, Mechatronics, Vol. 8, pp (1998). [8] Chang, P. Z. and Yang, L. J., A Method using V-grooves to Monitor the Thickness of licon Membrane with µm Resolution, J. Micromech. Microeng., Vol. 8, pp (1998). [9] Uchino, K., Ferroelectric Device, Marcel Dekker, NY, U.S.A. (2000). [10] Xu, T., Wu, G., Zhang, G. and Hao, Y., The Compatibility of Piezoelectric Film with Micromachining Process, Sensors and Actuators, Vol. 104, pp (2003). [11] Sze, S. M., VLSI Technology, McGraw-Hill, NY, U.S.A. (1988). [12] Nazarian, S. and Desai, M. R., tomated Surface Wave Method: Field Testing, Journal of Geotechnical Engineering, Vol. 119, pp (1992). [13] Wu, T. T. and Liu, Y. H., Inverse Determinations of Thickness and Elastic Properties of a Bonding Layer using Lasergenerated Surface Waves, Ultrasonics, Vol. 37, pp (1999). Manuscript Received: Dec. 25, 2003 Accepted: Jan. 15, 2004