invited paper for International Conference on Fracture, to be held Dec. 3-7, 2001, Honolulu, Hawaii, Hilton Hotel ADHESION TEST RESULTS Most cantilevers were free standing at lengths up to 1635 mm after the release and drying procedure as determined by interferometry. Some of these long cantilevers were contacting the substrate at their tips. In a few cases, cantilevers were attached over a relatively long length d, as indicated in Fig. 1(d). This latter group was excluded from further analysis. Knowledge of Young’s Modulus E and torsional support post compliance b are critical to assessing G [12]. We used a procedure previously described in detail [13] to determine E=163 GPa and b=1.25 mrad/(mN·mm). Also small curvatures k (caused by stress gradient through the thickness of the film) ranging from 0-1 m-1 were measured, and play a secondary role in determining adhesion values. These quantities are determined by electrostatically actuating the cantilevers, measuring the deflections and finding the best fit to finite difference models over a range of applied voltages. Using a fracture mechanics analogy, a cantilever’s adhesion to a substrate can be measured to high sensitivity and accuracy along its length [12,14]. In the adhesion testing procedure, free standing cantilevers are brought into contact with the substrate by modulating the voltage on the actuation pad. Using interferometry, the full deflection curve of the cantilevers is determined to better than 10 nm accuracy. At low to moderate voltages (up to 60V for this geometry), the deflections are highly sensitive to interfacial forces acting between the cantilever and the substrate. For different voltages applied to the actuation pad, corresponding to different points along the length of the beam, interferograms were taken and deflection curves were extracted. Knowing a and w from the mask layout and using the measured data for E, t, g, and b as input parameters, adhesion was determined by matching the model to each measured deflection curve. The only free parameter in the modeled curves is the adhesion G. A least squares fit between the model and measurement was used to determine its value. Typical minimum errors are less than 5 nm/pixel. Adhesion results for the different surface roughnesses are shown in Fig. 3, where the squares (data) correspond to the measured values of adhesion. The adhesion data is plotted versus av D , the average separation between the surfaces. For each value of av D , G values were determined from two different cantilevers at applied voltages of 0, 10, 20, 30, 40, 50 and 60 V. Small systematic errors in the input parameters limit the absolute accuracy of the G values to ~10%, but will not affect the relative values of adhesion in Fig. 3. ANALYSIS Adhesion testing was conducted in air at ~30% relative humidity (RH). A contact angle of 110° of the FDTS coating with water was measured, indicating a hydrophobic coating. We have observed no effect of RH on testing results up to 80% RH for these coatings [14]. Therefore, capillary condensation, which dominates adhesion of hydrophilic surfaces [15], does not play a role in these experiments. Furthermore, because the top and bottom surfaces are both grounded, electrostatic forces in the contact Fig. 3 Experimental and calculated results for adhesion verus average roughness 0.1 1 10 10 20 30 40 50 60 70 Meas. G (mJ/m2) D av (nm) data G = A/(12pD av 2) z-scaled AFM data mated AFM surfaces P2/P0 P0/P0 pairs P0/P0 (self) calculated values
RkJQdWJsaXNoZXIy MjM0NDE=