Fabrication of 1 Teradot/in.2 CoCrPt bit patterned media and recording performance with a conventional read/write head

Abstract

1 Teradot/in.2 (Td/in.2) CoCrPt alloy bit patterned media (BPM) disks were patterned by direct write e-beam lithography, and the recording performance was measured with a commercial recording head. Recording analysis showed a minimum error rate of 2 × 10−3, which was limited by the fraction of BPM patterning defects Continuous magnetic media disks were coated with a 20 nm thick carbon hard mask film by PECVD followed by a 8.5 nm thick hydrogen silsesquioxane (HSQ) resist by spin coating. A series of 1 Td/in.2 dot patterns were e-beam written in the HSQ, and the patterns were etched into the carbon hard mask by reactive ion etching. The underlying magnetic media was physically etched with 200 eV Ar. The carbon hard mask maximum thickness was limited by erosion of the HSQ dots during the carbon hardmask etch and shadowing of the mask during the magnetic media etch. The minimum carbon thickness and the maximum CoCrPt thickness were determined by erosion of the hardmask pillars during etching of the CoCrPt magnetic media. The optimal carbon hard mask thickness was determined to be ∼20 nm (for our PECVD carbon). The optimal CoCrPt magnetic media thickness was 6 nm, as determined by etch selectivity and magnetic properties. A Silvaco Monte Carlo 3D model simulation was used to describe the magnetic media etching process. Additional patterning steps formed physical support, surrounding the patterns, for the recording head that scanned in contact with the patterned magnetic media. Analysis of top down SEM micrographs of BPM patterns showed defect rates as low as 3 × 10−4 and a 1-sigma dot placement tolerance of 0.9 nm. Magnetic coercivity and switching field distribution width were measured from polar magneto-optic micro Kerr effect hysteresis loops (with a spot size of 20–50 μm). Patterning process conditions that produced a higher fraction of eroded or merged magnetic islands reduced the BPM coercivity and increased the relative width of the switching field distribution. Magnetic recording was performed with a commercial recording head of magnetic write width 90 nm in a shingled writing method. The recording error rate minimum varied with the fraction of defects in a similar manner as the magnetic switching field distribution width. A higher fraction of defects resulted in increased recording error probability due to data erasure by stray magnetic fields from the head.