The Effect of Geometry on Nanoscale Tellurium Fuses for Solid- State Data Storage Anthony C. Pearson 1 , Bhupinder Singh 2 , Matthew R. Linford 2 , Barry Lunt 3 , Robert Davis 1 1. Department of Physics and Astronomy, Brigham Young University, Provo, UT 2. Department of Chemistry and Biochemistry, Brigham Young University, Provo, UT 3. School of Technology, Brigham Young University, Provo, UT 1. Introduction Due to the permanent nature of their programming process, write-once-read-many (WORM) memory devices are promising for long- term data storage, and nanoscale fuse geometries have been considered as WORM devices. 1-3 To program such devices, the fuses must be heated and blown as a voltage is applied. Accordingly, the fuse must contain a region of high resistance. Most of the power is dissipated across this region when a voltage is applied, and under a sufficient voltage this region will heat up and blow, causing electrical disconnection. The temperature a given fuse can reach is influenced by the magnitude of the applied voltage, the size and shape of the high resistance region, and the distance between the electrical contacts connecting across this region. Here we show results of a finite element analysis in which we have determined the effect of applied voltage and fuse geometry on fuse temperature. Additionally, we show comparison of these results to results from tellurium fuses we have fabricated. 2. Results and Discussion Figure 1a-b shows the fuse geometry chosen in simulations and in experiments. Figure 1a shows the layout of the fuse, which is connected between two gold contact pads. Figure 1b shows a closer view of the fuse structure, where the narrowest portion is the region of high resistance. Finite element analysis was done using COMSOL Multiphysics software version 4.2, to determine the effects of the width of the narrowest region of the fuse (w), the length of the narrow region (l), and the distance between gold contact pads (d). In all simulations the steady-state temperature was recorded as one parameter was varied. The ranges explored for each parameter were as follows: (1) applied voltage 1 - 15 V; (2) l: 500 nm - 10 μm; (3) w: 500 nm - 5 μm; and (4) d: 500 nm - 20 μm. As one parameter was varied all other parameters were fixed at default values which were chosen as follows: (1) applied voltage: 5 V; (2) l: 500 nm; (3) w: 500 nm; and (4) d: 10 μm. In simulations where the width was varied the length was also varied to keep the center region square. Literature values for the electrical and thermal conductivities of tellurium were used in the simulations. Figure 2 shows the results of simulations. As expected, as applied voltage increases (Fig. 2a), as the width of the fuse decreases (Fig. 2b), as the length of the fuse decreases (Fig. 2c), or the contact pad separation decreases, the fuse temperature increases, where the fuse temperature depends most strongly on the fuse width and contact pad separation. Thus fuses with smaller dimensions will generally be capable of reaching higher temperatures. The temperature at which the fuse blows will be dependent on the properties of the fuse material. An appropriate material should allow the fuse to blow at a fairly low temperature to minimize the programming power and time requirements. Fuses in geometries where high temperatures were achieved in simulations would likely blow before the simulated temperature is reached. Thus the goal of optimizing the fuse geometry is not to maximize the temperature, but to minimize the power and time required to heat the fuse to an appropriate temperature. From our simulations we therefore expect that smaller fuses should heat up to a sufficient temperature for programming at a much lower voltage and power, and in less time, than large devices.