Eff ect of Annealing on 50nm Gold Films
3Pages

Application Note Effect of Annealing on 50nm Gold Films Abstract different substrates and subjected to two different treatments. Ultra-thin gold films are used in many modern applications to The first substrate was the bare wafer—a fine-grained fusion impart essential optical qualities to surfaces. In this study, new of titanium carbide (TiC) and alumina (Al2O3). Figure 1 shows nanoindentation technology, NanoBlitz 3D and Express Test, are a scanning-electron micrograph of the bare wafer. The second used to evaluate the effect of annealing on 50nm gold films. substrate was an alumina film, 2.6μm thick, sputter-deposited Vacuum annealing for three hours (below 300°C) causes the on the bare wafer. The first gold film (A) was deposited directly hardness to decrease by 4.3% – 6.7% (99.9% confidence). on the bare wafer, and the second (A’) was the same film after Introduction Ultra-thin gold films impart essential optical qualities to many surfaces. Common applications include high-precision mirrors vacuum annealing for three hours below 300°C. The third gold film (B) was deposited on the alumina basecoat, and the fourth (B’) was that same film after vacuum annealing. and magnetic storage media. For the gold film to fulfill its purpose over time, its microstructure and properties must All samples were tested with a KLA Nano Indenter system, remain stable through fluctuations in temperature and stress. configured with the Express Test option. All tests were The purpose of the present work is to use a new performed using a low-load transducer head fi tted with a nanoindentation technique to assess how annealing Berkovich indenter. In all, 400 indents were performed on each affects the hardness of such ultra-thin gold films. sample, with each indent performed to a maximum By virtue of its speed, NanoBlitz 3D and Express Test options for the KLA Nano Indenter systems dramatically improves our ability to detect significant difference when employing the Student’s t-test to compare two measurement sets. Express Test is an advanced form of nanoindentation, performing one complete indentation every second. By contrast, typical nanoindentation technology requires at least 30 seconds per indentation. In other words, in a given time period, we can make 30 times the number of observations with NanoBlitz 3D and Express Test than with typical nanoindentation technology. This is important, because the value of the test statistic in the Student’s t-test depends directly on the square root of the number of observations; that is, more observations increase the magnitude of the test statistic and therefore increase the likelihood of detecting significant difference, if in fact there is a real difference to be detected1. Thus, NanoBlitz 3D and Express Test can reveal small differences in hardness which may not be discernible with fewer indents. Because annealing may have only a slight effect—if any—on hardness, Express Test is an appropriate measurement tool. Experimental Method Four gold-film samples of 50nm thickness were tested. The set of four samples comprised identical films deposited on two displacement of 20nm (40% of the thickness of the gold film). The testing protocol was purposefully designed to neutralize the effect of all physical variables that might influence the measured hardness other than the independent variable, which was annealing. Samples to be compared directly were mounted together on the same puck. Each sample was tested in multiple locations, and testing was systematically alternated between one sample and the other. The procedure may seem onerous, but it was necessary because four hundred independent observations yield extreme sensitivity, not only to the independent variable (annealing), but also to other uncontrolled variables. With four hundred observations on each sample, one can detect a significant difference of only 1% with 99.9% confidence, even if the standard deviation is 5%2. If four hundred observations were made at one site on one sample, followed by four hundred observations at another site on the other sample, one would likely discern a significant difference (according to the Student’s t-test) on materials that were in fact identical, with the difference being due to some uncontrolled (and presumably unknown) variable. But we have found that the protocol followed in our experiment yields no significant difference on materials which are identical. Thus, when a signifi

Application Note Results and Discussion Figure 2 shows the hardness for all four gold films. Annealing causes a slight but significant decrease in hardness. The hardness of the film deposited on the bare wafer decreased by 6.8% and the hardness of the film deposited on the basecoat decreased by 4.3%. These differences are so slight that detecting them with a high degree of confidence (99.9%) requires a large number of observations. By performing one indentation per second, the Express Test option allows such testing to be completed in an acceptable timeframe for industrial applications. The...

Application Note histogram, but the mean is lower (2.31 GPa) because it is skewed by the aberrant indent. NanoSuite or InView software A new nanoindentation technology, Express Test, was used to allow allows the experimenter to easily filter out aberrant tests; discern a small but significant decrease in the hardness of 50nm if this aberrant test is excluded, the mean increases to 2.39 GPa. gold films due to annealing. Because the decrease in hardness However, using the median to represent the set virtually caused by annealing was small, many indents were required in eliminates the need for...