Nano-scale displacement measurement of MEMS devices using fiber optic interferometry C. W. Lee, X. M. Zhang, S. C. Tjin and A. Q. Liu Microelectronic Division, School of Electrical & Electronic Engineering, Nanyang Technological University, Singapore 639798 Email: eaqliu@ntu.edu.sg, Tel: (65) 6790-4336, Fax: (65) 6792-0415 ABSTRACT n this paper, the application of fiber interferometry in the nano-scale displacement measurement of microelectromechanical system (MEMS) device is being presented. Fiber optic interferometry combines the benefits of the optical fiber such as lightweight, small size and wide bandwidth with the high resolution, high sensitivity capability of the interferometry. t also provides easier setup and offers lower energy loss than the conventional free-spaced interferometry. The fiber optic interferometric system comprises a laser source and a 2 X 2 fiber coupler. The reference arm and the sensing arm of the interferometer are formed within a single output of the coupler. The resultant interference intensity is measured at one of the fiber coupler input. The fiber optic interferometry could be used for the MEMS with moving structure. A case study is being carried out to investigate the displacement of the micromirror in the MEMS Fabry-Perot Filter. The mirror is being driven by the comb drive actuator under the effect of applied voltage. t selectively reflects certain wavelengths while allows others to pass through determined by the air cavity length. The displacement under different applied voltages will be measured using the fiber optic interferometry. The experiment and the result will be demonstrated. Keywords: MEMS, Optic interferometry, MEMS Measurement 1. NTRODUCTON Microelectromechanical System (MEMS) is integrated micro system that combines electrical and mechanical components. They range from micrometers to millimeters in size, and are fabricated using technology developed from ntegrated Circuit (C) batch processes. As the devices function by small displacement or deformation, it is very important to have precise measurement and visual means to characterize the displacement of the MEMS. One such MEMS device is the optical crossconnect (OXC) by using drawbridge micromirrors 1. The OXC is an optical switch with many ports that interconnect multiple inputs and multiple outputs. Besides, the drawbridge structure has also been reported to be used in variable optical attenuator (VOA) 2. The correct operation of this device is very much depends on the precise out-of-plane movement of the micromirrors. Another device that requires the precise outof-plane displacement is the micromachined wavelength tunable laser 3, 4. The capability of tuning the laser s wavelength is obtained by adjusting the out-of-plane displacement of the micromirror. A nano-level position change can significantly influence the performance. For example, a displacement of 0.1 nm can make the laser output hop from one mode to another mode 3. Hence, it is important to develop a suitable method for the MEMS out-of-plane displacement characterization to ensure correct operation of these devices. Several methods have been developed for the above-mentioned purpose. One of them is the Electronic Speckle Pattern nterferometry (ESP). This method measures the displacements of the diffusely reflecting objects by analyzing the fringe pattern produced after subtracting two different speckle patterns on the same surface. However, this method has limitation that arises from decorrelation effects between corresponding speckles of two object states 5. Furthermore, the large in-plane displacement will worsen this limitation by lowering the fringe contrast. The spatial interferometer is also widely used. This method has been reported to be able to measure the MEMS in-plane and out-of-plane 196 Microsystems Engineering: Metrology and nspection, Christophe Gorecki, Editor, Proceedings of SPE Vol. 5145 (2003) 2003 SPE 0277-786X/03/$15.00
displacement, frequency response and velocity 6, 7, 8, 9. However, the experimental setup is rather complicated and the laser beam alignment may be time consuming. The fiber interferometry has been reported for measuring the velocity of a vibrating object 10. For fiber interferometry, the interferometer is formed inside an optical fiber. t inherits the benefits such as lightweight, small size and large bandwidth 11 from the optical fiber while maintaining the advantages of high-resolution and high-sensitivity capability of the spatial interferometry 12. Furthermore, it offers simpler and more flexible setup than the other approaches. The use of fiber pigtails in the laser source, photodetectors and other components has eliminated the hassle of laser beam alignment needed for the free space interferometer. The simple mathematical formulation has also provided an advantage over the ESP method. For the fiber optic interferometry, the phase difference between the reference arm and the sensing arm are used to extract the displacement information, which is similar to the spatial interferometer. The change of the air cavity length in the sensing arm provides the displacement results. Software has been developed for data processing. n this paper, the out-of-plane displacement of a MEMS Fabry-Perot filter is measured by using the fiber optic interferometry as a case study. The MEMS device consists of a micromirror that is actuated by the comb drive. The filter selectively reflects certain wavelengths while allows other to pass through others. The air cavity between the fiber end and the micromirror determines the wavelengths. The measurement principles of the fiber optic interferometry will first be presented, followed by the measurement setup. Finally, the out-of-plane displacement of the MEMS Fabry-Perot filter will be measured to examine the capability of the fiber optic interferometric system. 2. MEASUREMENT PRNCPLES The fiber optic interferometric setup consists of a fiber-pigtailed laser source, an isolator, a 2 2 50/50 optical coupler and two photodetectors. The schematic of the fiber optic interferometer is as shown in Fig.1. Laser Source solator 2 X 2 Optical Coupler MEMS Device Photodetector 1 Photodetector 2 Fig. 1 Schematic of fiber optic interferometer An input light is propagates towards the fiber end at the MEMS sample. The reference arm is formed by the Fresnel reflection at the fiber-air interface of the cleaved fiber end, while the sensing arm is formed by the reflection of the MEMS sample. By using only one output of the fiber coupler, the reference arm and the sensing arm are created for an interferometer. The reflected lights interfere with each other and are then divided equally into two beams. One is Proc. of SPE Vol. 5145 197
blocked by the isolator. The other is collected by the photodetector 1. The photodetector 2 measures the other output with the purpose of normalizing the reading at the photodetector 1. The irradiance of the interference beam T depends on the optical path difference between the two arms (twice the air cavity length), as given by 12. = + cos( 4πL / λ φ), (1) T avg amp + where avg is the dc offset of the sinusoidal wave, amp is the amplitude of the sinusoidal wave, ϕ is the initial phase difference and λ is the laser wavelength. The wave has a period of half of the source wavelength. Subsequently, the relationship between linear displacement and various light irradiances is being derived from Eq. 1, and the displacement x, can be expressed as λ -1 T0 avg -1 T avg x = [cos ( ) cos ( )] (2) 4π amp where TO is the resultant intensity at the reference position. Equation 2 implies that the displacement for a position can be determined by measuring the current irradiance and comparing it with respect to that of the initial position. The avg and amp can be obtained from the maximum intensity max and minimum intensity min as expressed by avg amp amp max + min = 2 (3) max min = 2 (4) The periodic nature of the irradiance of the interference may cause problem to the measurement. Hence, proper algorithms should be used as shown in Fig.2. n the algorithms, the number of turning points (i.e. maximum or minimum) is first determined. f no turning point is found, this implies that the displacement is less than a period and the displacement can be determined by the Eq. 2. Otherwise, each turning point contributes λ/4 to the displacement. This displacement is then added to the displacements of the initial position and the final position from their nearest turning point respectively. Start O btain the num ber of turning point betw een the initial and sam ple point, which equivalent to quarter wavelength No Any turning point? Yes Calculate the displacement betw een the initial point w ith the sam ple point Calculate the displacem ent betw een the initial point w ith the nearest turning point C alculate the displacem ent between the sample point with the nearest turning point Add the 3 results above together Fig. 2 Measurement algorithms End 198 Proc. of SPE Vol. 5145
3. EXPERMENT As a case study, a MEMS Fabry-Perot filter is investigated. The scanning electron micrograph (SEM) of the filter is shown in Fig. 3. The structures are fabricated in a silicon-on-insulator (SO) wafer by deep reactive ion etching (DRE) processes. The SO wafer has a silicon structure layer (75 mm thick) and a SiO 2 buffer layer (2 mm thick) on a handling silicon wafer (475 mm thick). The fabrication uses single mask and dry release. The depth of the mirrors, actuators and the grooves is 75µm, hence normal optical fibers can be used for optical packaging. n the final step, the optical fiber is integrated and packaged. The overall size is about 1.6mm 1.0mm (not including the fiber). Fig. 3 The MEMS Fabry-Perot Filter n the experiment, the laser (ANDO AQ4321D) and the 2 X 2 fiber coupler are used. The laser wavelength is set at 1550nm to match the operation frequency of the optical coupler. The MEMS Fabry-Perot filter has fiber groove, which is used to hold the fiber and also align the fiber to the micromirror automatically. At the receiver side, the photodetectors (Newport 818-S-1) are attached to the optical power meters (Newport 840), which convert the optical signal detected to the electrical signal. The MEMS micromirror is being driven by the comb drive actuator. The comb drive works as a series of capacitors, which produce electrostatic force when subjected to applied voltage. The electrostatic force is responsible for controlling the displacement of the micromirror. Two probes are used to apply the voltage from a DC power supply (TOPWARD Electric TPS4000) to the actuator. The reading of the optical power meter for different applied voltages are recorded and then analyzed by a data processing software developed by the authors on the LabVEW platform 13. Next, the measurement results are compared with the simulation results. The simulation results are obtained by using Eq. 5. 3 nεl 2 x = V (5) 3 4Egb where x is the displacement, n is the number of fingers, ε is the permittivity of free space, E is the Young modulus of the material, b is the suspension beam width, l is the suspension beam length, V is the applied voltage and g is the finger gap. The design value for the parameters of the Fabry-Perot Filter is summarized in Table 1. Proc. of SPE Vol. 5145 199
Table 1 Designed parameters value for MEMS Fabry-Perot Filter Parameter Value Number of fingers n 124 Suspension beam length l 700 µm Suspension beam width b 3 µm Free space permittivity ε 8.85 pf/m Finger gap g 2.5 µm Young modulus E 160 GPa 4.00 3.50 3.00 Simulation Measurement Displacement (um) 2.50 2.00 1.50 1.00 0.50 0.00 0 5 10 15 20 Voltage (V) Fig. 4 Comparison between measurement result and simulation result The relation of micromirror translation and the driving voltage is shown in Fig. 4. The square points represent the displacements in the loading process (i.e., driving voltage increases from 0 to 20 V). When the driving voltage is increased from 0, the comb drive does not have obvious displacement until the driving voltage reaches about 2 V. This is presumably due to the static friction between the micromirror structures and the substrate having some contact with each other. This often occurs in the structures released by wet etching. Besides this, the data follows the theoretical curve once the static friction has been overcome. A displacement of 3µm is obtained by using a driving voltage of 20V. The discrepancy between the simulation and the experimental result is probably due to the overetched of the comb drive 14. 4. CONCLUSONS n this paper, a fiber optic interferometric measurement system has been proposed for MEMS nano-scale displacement measurements. A MEMS Fabry-Perot filter has also been investigated to verify the capability of the measurement system. The measured displacement follows the simulation curve with marginal discrepancy. The system is able to measure the displacement up to a resolution of about 7nm with an optical power meter with resolution of 1µW. Hence, we can conclude that the fiber interferometric system is suitable for characterizing the MEMS devices. 200 Proc. of SPE Vol. 5145
REFERENCES 1. A. Q. Liu, X. M. Zhang, V. M. Murukeshan, Q. X. Zhang, Q. B. Zhou & S. Uppili. An Optical Crossconnect (OXC) using Drawbridge Micromirrors. Sensors Actuators A (97-98), 227-238, 2002. 2. A. Q. Liu, X. M. Zhang, C. Lu, F. Wang, C. Lu & Z. S. Liu. Optical and Mechanical Models for a Variable Optical Attenuator using a Micromirror Drawbridge. Journal of Micromechanics and Microengineering, 13, 400 411, 2003. 3. A. Q. Liu, X. M. Zhang, V. M. Murukeshan, C. Lu & T. H. Cheng. Micromachined Wavelength Tunable Laser With an Extended Feedback Model. EEE Journal on Selected Topics in Quantum Electronics, 8, 73-79, 2002. 4. A. Q. Liu, X. M. Zhang, V. M. Murukeshan & Y. L. Lam, A Novel ntegrated Micromachined Tunable Laser using Polysilicon 3D Mirror. EEE Photon. Technol. Lett., 13, 427 429, May 2001. 5. R. A. Matίnez-Celario, B. Barrientos, F. J. Sanchez-Marίn, L. M. Lόpez & J. A. Rayas. Out-of-plane Displacement Measurement by Electronic Speckle Pattern nterferometry in Presence of Large n-plane Displacement. Optical Communication, 208, 17-24, 2002. 6. J. Wylde & T. J. Hubbard. Measurement of MEMS Displacements and Frequencies using Laser nterferometry. Proceeding of the 1999 EEE Canadian Conference on Electrical and Computing Engineering, 1680-1681, 1999. 7. C. J. Tay, C. Quan, S. H. Wang & H. M. Shang. Determination of a Micromirror Angular Rotation Using Laser nterferometric Method. Optics Communications, 195, 71-77, 2001. 8. M. R. Hart, R. A. Conant, K. Y. Lau & R. S. Muller. Stroboscopic nterferometer System for Dynamic MEMS Characterization. Journal of Microelectromechanical Systems, 9, 409-417, 2000. 9. Rembe, C., Muller, L., Muller, R. S., & Howe, R. T. Full Three-Dimensional Motion Characterization of a Gimballed Electrostatic Actuator. Proc. of EEE nternational Reliability Symposium, Orlando, Florida, April 30- May 3, 2001. 10. V. Métivier, E. Morteau, E. Ouisse & H. C. Seat. "Dual-cavity Extrinsic Fiber Fabry-Perot nterferometer for Vibration-displacement Measurements. The 16 th European Conference on Solid-State Transducers, 383-386, 2002. 11. J. Hecht. Understanding Fiber Optics. 4 th Edition. Prentice Hall, 2002. 12. E. Hecht. Optics. 4 th Edition. Addison-Wesley, 2002. 13. LabVEW User Manual for Windows. National nstruments, 1993. 14. J. Y. Lee, S. H. Kim, H. T. Lim, C. H. Kim, C. W. Baek & Y. K. Kim. Electric Spring Modeling for a Comb Actuator Deformed by the Footing Effect in Deep Reactive on Etching. Journal of Micromechanics and Microengineering, 13, 72-79, 2003. Proc. of SPE Vol. 5145 201