THE CHARACTERIZATION OF OXIDE LAYERS IN ADHESIVE JOINTS USING ULTRASONIC REFLECTION MEASUREMENTS

The use of ultrasonic reflection coefficient measurements at both norm al and oblique incidence for the detection and characterization of oxi de layers at the adhesive-adherend interface in adhesive joints has be en investigated. Tests have been carried out on aluminium alloy-oxide- epoxy specimens with a minimal (< 0.1 mu m) thickness oxide layer, 43 and 15 mu m thick sulphuric acid anodized oxide oxide layers and an in dustry standard, 3.5 mu m thick chromic acid anodized oxide layer. A s uperpure aluminium-90 mu m thick oxalic acid anodized layer-water syst em has also been investigated. The results indicate that the cell wall density of the sulphuric acid anodized oxide layer is only 1990 kg m( -3), compared to 2850 kg m(-3) for the oxalic acid anodized layer. The cell wall density and Young's modulus are obtained from normal incide nce measurements. Predictions of the oblique incidence reflection coef ficients using the effective elastic constants of the oxide layer obta ined from a standard porosity model assuming these cell wall propertie s and the porosity measured from TEM images showed good agreement with the measurements made on both the sulphuric acid anodized and oxalic acid anodized specimens, only small adjustments to the properties bein g required to obtain a very good fit. The frequency range over which i t is necessary to take measurements increases as the thickness of the oxide layer decreases. Maximum frequencies of 15 MHz for the oblique i ncidence measurements and 70 MHz for the normal incidence measurement were adequate to obtain values of rho, C-11 and C-33 for a 43 mu m thi ck sulphuric acid anodized layer to an estimated accuracy of +/- 10% a nd to set a lower bound on C-55. In order for a 3.5 mu m thick oxide l ayer with similar properties to be detectable at normal incidence, the frequency would have to be increased to over 200 MHz, while the frequ ency of the oblique incidence measurements would have to be raised to about 25 MHz if a 5% change could reliably be detected, or 50 MHz if a 10% change is necessary in order to ensure detection.