벌크 실리콘 미세가공을 이용한 고감도 Fabry-Perot 음향방출센서 및 위치추정 시스템 = High?sensitivity Fabry?Perot acoustic emission sensor and source localization system based on bulk silicon micromachining

다국어 초록 (Multilingual Abstract)

Piezoelectric and capacitive sensors are currently the most widely used electronic acoustic emission (AE) sensors. However, these sensors are vulnerable to electromagnetic interference (EMI) and pose challenges for long-distance monitoring and multiplexing. To overcome these limitations, we propose a MEMS-based fiber-optic Fabry–Perot acoustic sensor utilizing a 710 nm thin diaphragm fabricated by bulk silicon micromachining.
The diaphragm of the proposed sensor is designed as a multilayer structure composed of a 500 nm-thick Si3N4 layer, a 10 nm Ti layer, and a 200 nm Au layer. Si3N4 provides mechanical stability and high deposition reproducibility, even at sub-micrometer thicknesses. The Au improves the interference contrast of the Fabry–Perot spectrum through its high reflectivity, thereby enhancing the sensor's optical spectral performance. Furthermore, Ti acts as an adhesion layer between the Au and Si3N4, effectively preventing interface delamination even under long-term thermal and mechanical stress cycling. The fabrication process enables reproducible batch production of high-quality thin diaphragms, and alignment grooves were implemented to ensure precise alignment between the optical fiber and the diaphragm during assembly. Consequently, the four fabricated fiber-optic acoustic sensors exhibited excellent optical spectral characteristics, securing a uniform free spectral range (FSR) and an interference contrast exceeding 21 dB.
Performance evaluation showed the sensor achieved a high sensitivity of 179.3 mV/Pa at 140 kHz, approximately 16 times higher than that of commercial AE sensors, and a minimum detectable pressure (MDP) of 247.6 μPa/√Hz. The frequency response was almost flat over 20–150 kHz, and the sensor exhibited a cardioid-shaped directivity pattern with high forward sensitivity and reduced sensitivity in the rear direction.
To verify the performance of Time Difference of Arrival (TDOA) localization, an array system was constructed consisting of three sensors in 2D and four in 3D. A LabVIEW program was used to implement measurement and signal processing procedures. The core of TDOA localization is the precise calculation of the time difference of arrival between sensors. For this purpose, this thesis applied cross-correlation (CCR) and generalized cross-correlation with phase transform (GCC–PHAT). The 2D experimental results showed that CCR resulted in an average localization error of 2.2 cm and a maximum error of 4 cm. In contrast, GCC–PHAT exhibited a low average error of approximately 1 cm. This performance difference is attributed to CCR's sensitivity to the non-uniformity in the output signal amplitude caused by the sensor's spectral shift and directivity. In contrast, GCC–PHAT effectively suppresses the influence of these amplitude variations through phase-based weighting, enabling more stable and accurate time difference estimation even in the presence of noise and signal attenuation. Applying GCC–PHAT to 3D localization resulted in an overall error of less than 2.5 cm. While high accuracy was maintained in the X–Y plane, a larger deviation was observed along the Z-axis. This is attributed to insufficient altitude information, as only one sensor was placed in the Z-direction. Optimizing the sensor array or adding sensors along the Z-axis is expected to enhance the 3D localization accuracy.

목차 (Table of Contents)

• 1. Introduction 1
• 2. Application of AE sensors 5
• 2.1 Power equipment monitoring 5
• 2.2 Structural health monitoring 7
• 2.3 Underwater and marine structure 8
• 2.4 Medical imaging 10
• 3. Fiberoptic AE sensors 12
• 3.1 Sagnac interferometer 12
• 3.2 MachZehnder interferometer 14
• 3.3 Fiber Bragg grating 15
• 3.4 FabryPerot interferometer 16
• 4. Proposed MEMS FOAS 19
• 4.1 Bulk silicon micromachining 19
• 4.2 Fabrication of diaphragm 21
• 4.2.1 Diaphragm design and material selection 23
• 4.2.2 Verification of resonant frequency 23
• 4.2.3 Design of photomask 26
• 4.2.4 Diaphragm structure 27
• 4.3 Sensor fabrication 30
• 4.3.1 Design and fabrication of sensor frame 31
• 4.3.2 Alignment/bonding of diaphragm 34
• 4.4 Location estimation algorithm 38
• 4.4.1 Time difference of arrival 38
• 4.4.2 Crosscorrelation 39
• 4.4.3 Generalized crosscorrelation with phase transform 40
• 5. Experiments & Results 43
• 5.1 Sensor performance evaluation 43
• 5.1.1 Sensitivity 43
• 5.1.2 MDP 46
• 5.1.3 Frequency response 48
• 5.1.4 Directivity 49
• 5.2 TDOAbased localization experiments 52
• 5.2.1 2D Location 54
• 5.2.2 3D Location 60
• 6. Conclusion 62
• 7. Reference 65
• 요약(국문초록) 71
• Acknowledgement 74