Airbearing Simulation of Discrete Track Recording Media

Heat assisted magnetic recording (HAMR) or patterned media recording is expected to be used in future disk drives in order to increase the areal density above 500 Gbit/in2. In HAMR laser light is used to heat individual bit cells, thereby lowering the media coercivity during the write process. In patterned media recording, individual tracks or individual discrete bits are fabricated on the disk surface. In bit patterned media (BPM) each bit forms a single surface entity that is physically separated from neighboring bits in the circumferential as well as radial direction. Thus, transition noise between adjacent bits is absent. In discrete track recording (DTR) technology the bits are stored on single tracks which are physically separated from each other. Thus, the transition noise is eliminated in the radial direction but not in the circumferential direction. Sliders flying over BPM or DTR media "see" a disk surface that consists of ridges and grooves in the case of DTR, or individual distributed island-like regions in the case of BPM. The flying behavior of a slider over such a surface is different from that for flying over a "smooth" disk. A reduction in the steady state flying height coupled with a different flying attitude can be expected for DTR or BPM media unless planarization of the disk surface is performed. To investigate the effect of discrete tracks we have modified an existing finite element based air bearing simulator that takes the characteristics of the grooved disk surface into account. A finite element approach was chosen, because of the ease of implementation of mesh size variations for different areas of the air bearing surface. In areas of high pressure and low spacing a very fine mesh is selected to investigate the effect of the ultra fine grooves appropriately. The average element edge length in areas of high pressure and low spacing was chosen to be one fifth of the groove width of the DTR media. In areas of higher spacing, we have chosen a coarser mesh, assuming that the influence of surface features is negligible at large spacing. This approach keeps the overall model size small enough to be solvable on a PC. A uniform high mesh density would result in a forbiddingly large numerical problem that could not be handled on present day computers.