6.3.3 High Density Data Storage using AFM

Already in 1999, Cooper et al. [36] showed that by anodically oxidizing titanium with the AFM, an aerial density of 1.6Tbits/in$ ^2$ can be reached. In the same year, Namin et al. [158] used the atomic force microscope in order to manufacture high density data storage. The main idea behind the technique is generating nanodots in a pattern such that the presence of a nanostructure (nanodot) is read as a bitwise 1 while the lack of a nanostructure is read as a bitwise 0. The increase in the sharpness of the AFM needle tip down do approximately 10nm allowed for very high densities to be imprinted on silicon surfaces. Namin [158] achieved an areal density of 65Gbit/in$ ^2$ with readback rates larger than 10Mbit/s. More recently, storage densities of up to 3.3Tbits/in$ ^2$ have been reported [71]. For several years, the efficiency of this method was questioned due to the speed with which nanodots can be written with a single cantilever. However, many research groups are working on AFM nanopatterning using arrays of AFM needles, which could simultaneously generate multiple nanodots for a desired memory dot pattern [82], [207], [233], [235]. IBM has also developed a nanopatterning array, which spans 32$ \times $32 AFM cantilevers per pattern application [44], [167]. In 2004, Garcia [59] presented an imprinted image of $ \pi $, with 20-decimal place precision, written in binary code with oxide (SiO$ _x$) nanodots on a silicon surface, as shown in Figure 6.13a. Using the models presented in Chapter 5, the pattern depicted is reproduced using a non-contact AFM needle with a bias voltage of 24V, ambient humidity of 80%, and a pulse time of 0.16ms. Each oxide dot has a width and a height of approximately 25nm and 2.5nm, respectively, as in the original experiment from Garcia, depicted in Figure 6.13b.

Figure 6.13: Simulations of AFM-generated nanodots for ROM applications. (a) Image of $ \pi $ in binary code, written with oxide nanodots on a silicon surface from[59]. (b) Simulated image of $ \pi $ in binary code, repeating the experiment from[59], with inset of a proportional Figure 6.13c. (c) Simulated image of $ \pi $ in binary code. with improved aerial density.
(a) Image of $ \pi $ in binary code.
\includegraphics[width=0.44\linewidth]{chapter_applications/figures/Pi_min.eps} \includegraphics[width=0.45\linewidth]{chapter_applications/figures/Pi_min2.eps}
(b) Simulated image of $ \pi $ in binary code. (c) $ \pi $ with improved aerial density.

Using the models from Chapter 5 and assuming that the minimum readable height for the nanodots is 1nm, a higher aerial resolution can be achieved, when a pulse time of 0.1ms is used with an ambient humidity of 50% and an AFM needle bias voltage of 22V. This simulation results in nanodots which have an approximate height and width of 1nm and 2.1nm, respectively. The simulation results are shown in Figure 6.13c, where the improved nanodot density is evident.

L. Filipovic: Topography Simulation of Novel Processing Techniques