August_EDFA_Digital

edfas.org ELECTRONIC DEVICE FAILURE ANALYSIS | VOLUME 22 NO. 3 32 suppress the attenuation, the length of the wire between the CMUT and the IV converter should be as short as pos- sible. Integration of a CMUT receiving probe with an IV converter can be the solution. HIGH-SENSITIVITY DETECTION OF ULTRASONIC SIGNAL USING PULSE COMPRESSION METHOD To improve the detection sensitivity of the received signal, we applied the pulse compression method that is commonly used to improve the distance resolution by detecting signals buried in noise with high sensitivity in radar signal processing. [8] Pulse compression is a method of computing a correlation function between transmit- ted signals to be input to an ultrasonic probe such as a chirp signal, or a pseudo random modulation signal and received signal reflected froma specimen and received at theultrasonic probe. The correlation function is calculated from the following equation. (Eq 8) Here, T is the sampling time interval, u ( nT ) is the received signal, and v ( nT ) is the transmitted signal. In this study, we applied frequency shift keying (FSK), which is a method that creates a modulated signal by combining waves of different frequencies for each assigned code. [9] FSK is an effectivemodulationmethod for improving the detection sensitivity because it has an advantage in that a high gain frequency can be selected to design the waveform. Todetect anultrasonic signalwithhighsensitivityusing pulse compression, it is necessary to design the transmis- sion waveform by optimizing the parameters of FSK: code strings, modulation frequencies, and the total length of the transmission signal. Further, in the piezoelectric element of the ultrasonic probe, distortion of the transmission waveformbasedon the second-order lagmodel occurs. [10] Therefore, it is important to consider the waveform distortion when optimizing the transmission waveform. In this study, the parameters of FSK using a simulated annealing method were optimized so that the full width at half maximum of the autocorrelation function of the distorted transmission waveform was minimized, and the defect-detection sensitivity was evaluated by applying the optimized waveform to the detection of ultrasonic signalswith amultilayer stacked specimen. HIGH-RESOLUTION IMAGING OF THREE-LAYER STACKED SPECIMEN WITH REFLECTION METHOD To evaluate the spatial resolution of SAT images, ultra- sonic inspectionwith the reflectionmethodwasmeasured with the conventional SAT using pulsewaves (hereinafter, referred to as pulse waves) and the proposed SAT using the pulse compressionmethod (hereinafter, referred to as pulse compression). The specimen was formed by stack- ing three layers of a 60-μm-thick silicon (Si) wafer and a 20-μm-thickpolyimide filmona 400-μm-thick Si wafer and processing holes as the artificial defects in the polyimide filmof the third layer fromthe top. Thedefectswere square holes with sizes of 100 (2 lines), 50 (3 lines), 40 (3 lines), 30 (4 lines), 20 (5 lines), 10 (5 lines), 8 (10 lines), 6 (10 lines), 4 (15 lines), 2 (30 lines), and 1 (30 lines) μm as shown in Fig. 5. In each measurement, an ultrasonic probe with a nominal frequency of 200 MHz was used, the scanning interval of the two-dimensional scanwas set to 2 μm, and theobservationareawas 720μm×2800μm. Thewaveform for FSKused in pulse compressionwas designed using 100 and 200 MHz sinusoidal waves with a high frequency gain in the selectedultrasonic probe. The optimized code string was the hexadecimal code 24CA4773, and the total length of the transmission signal was 225 ns. Figure 6 shows ultrasonic images with defects ranging from 6 to 100 μm in the third layer bonding interface obtained by using pulse waves and pulse compression. In the ultrasonic image obtained by using pulse waves (Fig. 6a), the 100-μm defects were observed, but smaller defects were difficult to detect, which is considered to be Fig. 5 Schematic diagram of three-layer stacked specimen.