Controlling the quality of thin films by surface acoustic waves

Abstract

Abstract Techniques based on laser-induced surface acoustic waves have been developed with the potential for characterizing thin films. They enable variations of elastic parameters, density and film thickness to be detected. Previous work has demonstrated that from the measurement of the frequency dependence of the surface wave velocity, film parameters may be solved by using an inverse dispersion relation. In the present paper, a method is described that can be extended to the use of surface acoustic waves to the control of film quality in an on-line process. The method exploits the phenomenon that surface waves in coated materials show dispersion, giving rise to a characteristic deformation of wide-band surface wave impulses. Having traversed a fixed measuring distance on the test piece sample, an impulse signal is detected and compared with a reference signal from a standard sample. Signal processing yields two test functions: a cross-correlation of the test signal with a reference signal and the envelope of this cross-correlation function. If the film quality of the test sample agrees with that from the reference sample, the cross-correlation function is symmetric and its maximum coincides with the maximum of the envelope. Deviations from the required quality causes both test functions to shift against each other and gives rise to non-symmetric shape of the cross-correlation. The shifts and the deformation of the test functions are shown to contain information about the variations of film quality. As an example, results are presented for the case of TiN-coated steel samples having different film thickness. The sensitivity of the method is discussed and found to be dependent on signal-to-noise ratio (SNR) and bandwidth of the measurement. Tests were performed with two different types of laser-ultrasound instrumentation. They both used a pulse laser to generate the surface wave impulses. In the first type of instrumentation, ultrasonic impulses were detected with a piezoelectric transducer. It showed good signal-to-noise ratios (SNR) and high-frequency (60 MHz) bandwidth, but required mechanical coupling to the surface. The second type of instrumentation used a Fabry-Perot interferometer, which enabled a complete non-contact measurement to be performed up to a 15 MHz frequency bandwidth. The principle of the measurement method was again confirmed using the interferometer.