Inspection of Ceramics Incorporating Size Estimation Methods Using Conventional Ultrasonics

One of the principal characteristics of the inspection of ceramic materials is the small size of the critical defects involved. In order to meet this challenge, either very high frequency techniques must be developed or conventional (low frequency) techniques must be used in special ways. This paper addresses the latter approach by investigating the use of commercial transducers combined with focusing techniques in a water bath at frequencies in the range of 5 to 20 MHz as well as signal analysis techniques using plane waves in the 30 to 40 MHz range using special order commercial transducers. Although the high velocity of sound in the ceramics used (silicon nitride) put some restrictions pn the numerical aperture which could be obtained with acoustic lenses, a focused beam ultrasonic system for use in a water bath was designed and used to produce maps which showed the location and reflecting power of defects in flat ceramic specimens. Subsequent analysis of the reflectors discovered by this focused beam system was carried out by a hand-held ultrasonic probe which irradiated the sample with plane waves. By performing Fourier analysis of the echo signals from the "defects" and comparing the frequency spectrum observed with calculated spectra, it was possible to estimate the effective spherical size of the scattering objects. Analysis of the frequency dependence of the scattered energy in the 1 ong wave length limit also provides a measure of an effective spherical volume of the scatterer but this method was found to require additional signal processing methods which will be developed at a later date. BACKGROUND As described on Poster 1, the basic problem with inspection of ceramic materials arises from the fact that the critical defec,ts are small (of the order of 100 microns). Thus the ultrasonic reflections are expected to be small and near the noise level not only because of the defect's geometric size but also because the wave length of the sound wave (of the order of 500 microns at 20 MHz) is larger than the defect. Furthermore,· the estimation of the size and fracture mechanics parameters of the defect from analysis of the long wave length limit of the reflected energy requires accurate, quantitative data analysis that is a demanding process even with large signals. To overcome these drawbacks, two approaches are available. One is to increase the amount of ultrasonic scattering by increasing the ultrasonic frequency in order to make the wave length equal to or less than the defect dimensions. The other is to enhance the low frequency scattering by focusing more incident energy onto the dlfect. The former approach is described elsewhere while the objective of this paper is to investigate the latter approach and to demonstrate the capabilities listed on the bottom of Poster 1. Poster 2 describes the considerations necessary for implementing a focused ultrasonic beam technique for inspection of a ceramic plate. The demanding circumstance here is the unusually high ultrasonic velocity in the silicon nitride ceramic material which causes the outermost rays from the focusing transducer to be refracted strongly to a focus that is closer to the front surface than the focal point for the more centrally located sound rays. In order to minimize this effect (which is equivalent to spherical aberration in op~ics), the diameter or aperture of the transducer should be kept small or a liquid with a higher sound velocity than water should be used for immersion. For our experiments, it was found that a ~~~ diameter transducer had to be stopped down to 5/16 inches since a water bath was used. Poster 3 describe~ a second and very important consideration that appears when low frequency ultrasonic waves are used either for convenience or for obtaining scattering data in the long wave length limit.Z Under these conditions, the tail of the echo from the top surface of the sample may well still be present at the time of arrival of the defect echo so the two signals appear superimposed on one another and quantitative measurements of the defect echo are rendered very inaccurate. To circumvent this problem and to develop an accurate measure of the defect echo a 1 one, the waveform characterizing the tail of the front surface echo and any other background noise was measured by moving the transducer to a defect free region and the computer was programmed to subtract this background signal from the defect plus background waveform. An example of the waveforms observed, and the results of the subtraction process are shown on the right side of Poster 3. Once a "defect free" or background waveform had been established and stored in the computer, it could be used to subtract from all waveforms taken at different locations on the sample. Our computer was programmed to deduce and record tne peak-to-peak amplitude of any signal found in a time interval chosen to encompass the region in which the sound was focused along with the X-V coordinates of that region. From this data, a map such as the one shown in Poster 4 was made by printing X symbols with a density proportional to the signal amplitude at each coordinate location. Following the establishment of the locations and relative scattering powers of the defects, the