Directed density multiplication for patterned disk templates

09/01/2008

R. Ruiz, E. Dobisz, and T.R. Albrecht Hitachi Global Storage Technologies, San Jose Research Center, San Jose, CA J. J. de Pablo, and P. F. Nealey, Department of Chemical and Biological Engineering, University of Wisconsin, Madison, WI H. Yoshida, Materials Research Laboratory, Hitachi Ltd., Hitachi, Ibaraki, Japan

Master templates for printing fine-pitch bit patterns on magnetic disk media can be manufactured using e-beam lithography to write lower pitch chemical contrast marks onto the substrate. Directed assembly of a block copolymer then interpolates the final TB/in² density pattern between the e-beam written registration marks, resulting in perfectly aligned island arrays with pitch below the e-beam resist resolution.

Innovative lithographic processes are critical to continue shrinking semiconductor device dimensions beyond the 22nm node, and to enable new technologies that require nanometer-scale features. This is particularly true for magnetic storage applications as the industry considers the introduction of patterned media to facilitate thermally stable magnetic recording at densities beyond 1TB/in² [1]. Because of the unique challenges of disk manufacturing, directed self-assembly using block-copolymer films may find its first large-scale lithographic application in the data storage industry.

Figure 1. Schematic of a magnetic storage disk with continuous granular media vs. bit-patterned media.Click here to enlarge image

In a magnetic bit-patterned media storage disk, the magnetic media is lithographically patterned into physically isolated single-bit islands (Fig. 1). On conventional sputtered granular media, each recorded bit is stored in a collection of 20 or more adjacent random grains. Scaling these grains down to accommodate higher recording density renders them magnetically unstable at room temperature, leading to long-term erasure of data. On bit-patterned media, each pre-patterned island stores one bit in a single magnetic grain. These islands are larger, and therefore more stable, than the individual random grains of conventional media, allowing higher density scaling.

A bit-patterned disk for a hard disk drive would have ~10¹³ islands???a significant fabrication challenge considering the nanometer-scale size of the islands. A suitable strategy for manufacturing patterned media starts with the fabrication of a master template mold for low-cost nano-imprint replication [2].

Figure 2. Feature densities vs. lattice pitch for various lattices. A 1Tdot/in2 requires a center-to-center spacing of 27nm for a hexagonal lattice or 13nm down-track center-to-center spacing for a rectangular lattice with an aspect ratio of 4.Click here to enlarge image

Lithography requirements for the master template are set by targeted storage density and geometry of the data patterns. There are various possibilities regarding choice of lattice geometries and the aspect ratio of the bit cell that can be used, all having an impact on the lithography process. Possibilities include hexagonal, square, and rectangular arrays. Fig. 2 shows the various bit densities that can be achieved for hexagonal, square, and two rectangular lattices with bit aspect ratios (BAR) of 2 and 4. The curves are plotted against the lithographically critical full pitch. Magnetic bit-patterned media is projected to potentially enter the market at densities beyond 1TB/in², which translates to bit cells with a full pitch under 27nm for a hexagonal lattice, or even smaller for other lattices. These dimensions are not contemplated by the ITRS Roadmap[3] until after they will be needed by the magnetic storage industry.

Lithographic quality requirements for patterned media master templates are extremely challenging. Feature size uniformity and placement accuracy all have to lie within 1σ = 5% while defect density has to be compatible with the addressability bit error rate [4]. Even though feature densities as high as 4.5Tdot/in² have been demonstrated by e-beam lithography under specialized conditions[5], patterned media specifications place serious challenges on e-beam lithography in terms of placement accuracy, size uniformity, and writing time. Under such ultra-high resolution e-beam lithographic conditions, the time to expose a 95mm disk template could take over a month.

We propose a directed assembly method that combines block copolymer self-assembly with e-beam lithography as an alternative to achieving dense patterns with tight size and placement tolerances, low defect densities, and reduced e-beam writing times. Block co-polymers such as polystyrene-polymethyl methacrylate (PS-PMMA) consist of two immiscible polymer chains covalently bonded to each other. In a film, the two ends of the molecule tend to self-segregate into (for example) an array of cylindrical PMMA microphases embedded in a PS matrix. In directed self-assembly, lithographic treatment of the substrate sets the equilibrium conditions to anchor features of the array in desired locations.

Block copolymer self-assembly possesses the attractive quality to produce periodic structures with critical dimensions in the range of 3-50nm.[6,7] Block copolymer patterns have also been proposed as lithographic masks to transfer the self-assembled patterns into a variety of structures by reactive ion etching[8], lift-off[9], electroplating[10], etc. However, block copolymer films alone have not yet demonstrated a direct analog to a conventional lithographic imaging layer in long-range order, uniform vertical profile, etch selectivity, feature placement accuracy, size uniformity, overlay, and flexibility to form arbitrary patterns. Given that block copolymers tend to form uniform patterns on a limited range of geometries over limited areas, an entry point for self-assembled films into manufacturing could be presented by applications such as patterned media that do not need a wide range of pattern geometries and do not require multiple lithographic steps.

Figure 3. Fabrication method. a) Substrate coated with a PS brush and e-beam resist; b)Chemical contrast on the substrate after removing the e-beam resist; c) The self assembled block copolymer cylindrical domains register with the pre-patterned sites and interpolates between the e-beam patterned dots. Click here to enlarge image

Feature density multiplication by directed assembly[11,12] onto chemically patterned substrates[13] establishes a path towards lithographic features with quality and resolution beyond the limitations of current lithography, and within more feasible e-beam writing times. We use cylindrical phase poly(styrene-b-methyl methacrylate) (PS-b-PMMA) block copolymers that form hexagonal-packed PMMA cylindrical domains embedded in a PS matrix. Our density multiplication method, as described previously[11], starts with a substrate treated with a polystyrene brush (Fig. 3).

We then apply a resist layer, and use e-beam lithography to define periodic features in an array close to a hexagonal lattice with a full pitch Ls=nLo commensurate to the natural pitch of the block copolymer pattern, Lo (n=2 in this work). The developed resist pattern masks the underlying PS brush during a brief exposure to an oxygen plasma, which creates a pattern of chemical contrast on the surface. The e-beam resist is removed, leaving the oxygen-treated PS chemically patterned substrate with pitch L_s.

We then spin-cast a block copolymer film, and anneal the sample in vacuum to form a pattern with near the preferential pitch L_o. The chemical contrast in the pre-patterned substrate dictates the final equilibrium configuration of the self-assembled film. The PMMA block preferentially wets the modified areas (arrays of dots) in the PS brush. In this manner, a cylindrical PMMA domain pattern forms that is registered with the pre-patterned sites. The pre-patterned lattice is nearly commensurate with the natural lattice of the block copolymer, and the self-assembled vertical cylinder array interpolates between the pre-patterned dots, multiplying the density of features by a factor of 4[11]. After the pattern is formed, the PMMA cylindrical domains can be selectively removed [14] and the remaining PS template can be used as a lithographic mask.

Figure 4. Feature density multiplication by directed assembly. a) and c) SEM micrographs of the e-beam patterned resist with Ls=78nm and Ls=54nm respectively. b) and d) block copolymer assembly the substrates of a) and c). with pitches Lo=39nm and Lo=27nm. The insets in a) and c) show direct e-beam patterned resist with Lo = 39nm and Lo = 27nm. Click here to enlarge image

Figure 4 shows representative scanning electron micrographs (SEMs) of both e-beam patterned resist and block copolymer patterns self-assembled onto chemically pre-patterned substrates at two different densities. In Fig. 4(a) and (c) the e-beam resist pattern was written for L_s=78nm and L_s=54nm, respectively. Patterns like those in Fig. 4 (a) and (c) were used to define the chemical pre-pattern for the block copolymer film of Fig. 4(b) and (d) with L_o=39nm and L_o=27nm, respectively.

The insets of Fig. 4(a) and (c) show e-beam resist patterns written at the same density as the block copolymer L_s=39nm and L_s=27nm. The block copolymer not only multiplies the feature density by a factor of 4, but also maintains a superior pattern quality compared to that of e-beam resist written at the same density???as seen in the insets where the pattern shows the noise in the e-beam tool. For patterns at 27nm full pitch, the resist approached its resolution limit, resulting in poor quality for any subsequent lithographic step, whereas the block copolymer film maintained a superior pattern quality.

The density multiplication and pattern rectification action performed by the block copolymer also provides a path towards feasible template writing times by enabling the use of faster e-beam resists and higher e-beam currents that usually impact resolution and pattern quality. The chemical pre-pattern can be done at lower resolution and lower quality, but higher speed, while the block copolymer can be used to enhance resolution and tighten quality standards.

Figure 5. Pattern Transfer. a) PS template after PMMA removal in a block copolymer thin film with Lo=39nm full pitch. b) Cr dots after lift-off. c) Si Pillars after reactive ion etching.Click here to enlarge image

The cylindrical domains (black dots in Fig. 4) of the PMMA blocks, which can be selectively removed and the pores left behind in the directed block copolymer film, have an adequate vertical profile that can be readily used to transfer the pattern into a template mold for imprint lithography (Fig. 5a). To test the pattern transfer, we deposited a 10nm layer of Cr on top of a developed block copolymer film (L_o=39nm like the one shown in Fig. 4b) followed by a liftoff in a piranha solution to form an array of Cr dots (Fig. 5b). We then used the Cr dots as an etch mask to form 25nm tall Si pillars (Fig. 5c) with a uniformity and long range ordering unprecedented in block copolymer-templated patterns [11].

Conclusion

The feature density multiplication and pattern rectification demonstrated by directed assembly provide a method to enhance the capabilities of e-beam lithography, enabling higher resolution with improved pattern quality and a reduction of writing time. This method provides a path towards master templates for patterned media applications that require feature sizes with tolerances beyond the limitations of current lithographic processes.

Acknowledgments

The authors gratefully acknowledge the contributions from D. S. Kercher and C. P. Henderson (Hitachi GST) and H. Kang and F. Detcheverry (U. Wisconsin).

References

1. D. Weller, A. Moser, IEEE Transactions on Magnetics 1999, 35, 4423.
2. S. Y. Chou, P. R. Krauss, L. Kong, The 40th annual conference on magnetism and magnetic materials AIP Philadelphia, Pennsylvania (USA), 1996.
3. Semiconductor Industry Association, International Technology Roadmap for Semiconductors Vol. 2005 Edition, San Jose, CA 2005.
4. M. Schabes, Journal of Magnetism and Magnetic Materials 2008, in press.
5. X. Yang, S. Xiao, W. Wu, Y. Xu, K. Mountfield, R. Rottmayer, et al., Journal of Vacuum Science & Technology B 2007, 25, 2202.
6. F. S. Bates, G. H. Fredrickson, Annual Review of Physical Chemistry 1990, 41, 525.
7. M. J. Park, K. H. Downing, A. Jackson, E. D. Gomez, A. M. Minor, D. Cookson, et al., Nano Lett. 2007, 7, 3547.
8. M. Park, C. Harrison, P. M. Chaikin, R. A. Register, D. H. Adamson, Science 1997, 276, 1401.
9. K. W. Guarini, C. T. Black, K. R. Milkove, R. L. Sandstrom, Journal of Vacuum Science & Technology B 2001, 19, 2784.
10. T. Thurn-Albrecht, J. Schotter, G. A. Kastle, N. Emley, T. Shibauchi, L. Krusin-Elbaum, K. Guarini, C. T. Black, et al., Science 2000, 290, 2126.
11. R. Ruiz, H. Kang, F. Detcheverry, E. Dobisz, D. S. Kercher, T. R. Albrecht, et al., Science 2008, 321, 936.
12. J. Y. Cheng, C. T. Rettner, D. P. Sanders, H.-C. Kim, W. D. Hinsberg, Advanced Materials 2008, 20, 3155.
13. S. O. Kim, H. H. Solak, M. P. Stoykovich, N. J. Ferrier, J. J. de Pablo, P. F. Nealey, Nature 2003, 424, 411.
14. T. Thurn-Albrecht, R. Steiner, J. DeRouchey, C. M. Stafford, E. Huang, M. Bal, et al. Advanced Materials 2000, 12, 787.

Ricardo Ruiz, received his PhD in physics from Vanderbilt U. and is a research staff member at Hitachi Global Storage Technologies San Jose Research Center, 3403 Yerba Buena Rd., San Jose, CA 95135 USA; ph.: 408-717-6000; email: [email protected].

Elizabeth A. Dobisz, received her PhD in materials science from the U. of Wisconsin and is a research staff member at Hitachi Global Storage Technologies San Jose Research Center.

Tom Albrecht received his BA in physics from Careleton College and a PhD in applied physics from Stanford U. and currently manages a team responsible for developing patterned media technology at Hitachi Global Storage Technologies San Jose Research Center.

Paul F. Nealey received his PhD in chemical engineering from MIT and is the founding director of the National Science Foundation-funded Nanoscale Science and Engineering Center in Templated Synthesis and Assembly at the Nanoscale, and is the Smith-Bascom Professor of Chemical and Biological Engineering at the U. of Wisconsin.

Juan J. de Pablo received his PhD from the U. of California, Berkeley, is the director for the NSF MRSEC on Nanostructured Interfaces and is the Howard Curler Distinguished Professor at the Department of Chemical and Biological Engineering at the U. of Wisconsin, Madison.

Hiroshi Yoshida, received Dr. of Eng. degree in polymer chemistry from Kyoto U. and is a unit leader of the Applied Electrochemistry Unit of Materials Research Laboratory, at Hitachi Ltd.