Manufacturing patterned media with step and flash imprint lithography
09/01/2008
G. M. Schmid, D. J. Resnick, and D. LaBrake Molecular Imprints, Inc, Austin, TX USA; G. Gauzner, K. Lee Seagate Technology, Fremont, CA USA
The increasing demand for hard drives with greater storage density has motivated a technology shift from continuous magnetic media to patterned media hard disks. Both discrete track recording (DTR) and bit-patterned media (BPM) approaches are expected to be implemented in future generations of hard disk drives to provide data storage at densities exceeding 1012 bits/in2. Step and flash imprint lithography (S-FIL), a low-pressure, room-temperature approach that incorporates drop-on-demand resist placement, is suited for high-volume manufacturing (HVM) applications such as patterned media. This article reviews the infrastructure required to enable an HVM solution???namely, fabrication of master templates, template replication, high-volume imprinting with precisely controlled residual layers, and dual-sided imprinting.
Data storage density of hard disk drives has increased 8 orders of magnitude in the last 50 years [1]. This remarkable progress has been enabled by thin magnetic film coating advances characterized by grain sizes as small as 7nm. The individual grains, typically made of a material such as CoCrPt for perpendicular recording [2], are separated by an oxide that forms at the grain boundaries. A cluster of grains with similar magnetization make up a single bit of stored data; many adjacent grains are required to form a volume large enough to be precisely written and read by the head element of a disk drive. In recent years, improvements in bit storage density have been driven by the development of deposition techniques capable of producing films with smaller magnetic grains. However, the superparamagnetic effect will eventually limit this progression; smaller grains eventually become magnetically unstable and the likelihood of a grain spontaneously flipping increases, ultimately resulting in loss of stored data.
The practical limitations of the superparamagnetic effect can be avoided by patterning boundaries between the magnetic domains. Patterning the magnetic material creates magnetic switching volumes with highly uniform size and shape, which can greatly facilitate bit reading and writing. Because the magnetic switching volume is defined lithographically???and not by the random grain structure of the deposited film???the magnetization of a single magnetic grain can be robustly addressed. The introduction of patterned media technology into manufacturing is expected to enable the next generation of hard disk drives with storage density exceeding 1012 bits/in2 (1 TB/in2). Realization of this technology transition will require industrial-scale lithography at unprecedented levels of feature resolution, pattern precision, and cost efficiency.
Two primary approaches to patterned media have been proposed to overcome the limits associated with superparamagnetism. In one approach, bit-patterned media (BPM), lithography and etch processes isolate each bit of data in a precise island of magnetic material, with each bit patterned individually. BPM patterns typically comprise dense pillars in close-packed arrays that represent a best-case scenario for data storage density (Fig. 1a). However, practical implementation requires a number of rapid advances in template fabrication, etching of magnetic material, addressing of the drive head element, etc. These challenges are being addressed by an intermediate approach in which individual tracks of data are patterned instead of individual bits. Thus, the dimensions of the magnetic domains are constrained in one dimension by patterning narrow gaps between discrete concentric tracks of magnetic material. This approach, DTR, features patterns that resemble densely packed arrays of lines (Fig. 1b). The processing demands of DTR are significantly relaxed in terms of resolution and dimensional uniformity, and it is expected that the process experience gained through the implementation of DTR will be valuable toward the introduction of BPM technology.
Figure 1. SEM images of a) BPM patterns at 40nm pitch (0.4 TB/in2), b) DTR track patterns at 70nm pitch, and b) a tilt image from a servo pattern region. |
Along with tracks or bits, additional features, called servo patterns, are included on hard disks to enable the head element to read and write data at precise locations [1]. The design of servo patterns plays a key role in the performance of the disk drive, and these patterns are typically proprietary to each drive manufacturer. Servo patterns contain features with dimensions that are typically 2?? to 10?? larger than the data features, and include arrays of lines, dots, and combinations thereof (Fig. 1c). Servo patterns are placed at regular angular intervals around the disk, with the effect of dividing the disk surface into a large number of wedge-shaped sectors. A typical disk might be divided into 100-400 sectors, with servo patterns occupying 5 to 20% of the disk’s active surface. Servo patterns are employed on conventional, un-patterned media by recording the patterns in the continuous media via a separate process that is similar to the normal operation of the disk drive. Servo recording is a time-consuming serial process, and so the ability to pre-pattern the servo features is significant.
Industrial-scale pattened media manufacturing poses a number of lithography challenges. S-FIL is the leading candidate to overcome these challenges. This article addresses key aspects of the S-FIL infrastructure and demonstrates a process capable of cost-effective, high-throughput double-sided patterning of hard disks at a rate of 180 disks/hr.
Master template fabrication
Fabrication of a master template for patterned media applications requires an electron-beam writing system with a rotating stage. This configuration is well-suited for defining the concentric layouts that are required for patterned media applications, and several suppliers now offer such systems (e.g. Crestec, Elionix, and Pioneer Electronics). Conventional electron-beam write tools have x-y stages and operate by stitching together adjacent exposure fields, but patterned media applications have very low tolerance for the stitching errors that inevitably occur at the boundaries between exposure fields.
Details of the imprint lithography process
For semiconductor device applications, a drop-on-demand S-FIL process has been used with a step-and-repeat imprint strategy to pattern fields on Si wafers. 18nm resolution and overlay performance better than 15nm, 3σ has been demonstrated [3,4]. With a suitable template, the same technology can be applied to imprint patterning of an entire wafer substrate in one step, with no need for a step-and-repeat approach.
Figure 2. Process flow for imprint patterning of hard disk substrates. |
Figure 2 demonstrates how patterning a hard disk can be performed. The liquid acrylate imprint resist is deposited with a multi-nozzle inkjet head across the active surface of the disk substrate. The template is lowered until contact is made with the substrate, and capillary action induces the imprint resist to completely fill the region between the substrate and the topography of the imprint template. The imprint material is then photo-polymerized via ultraviolet illumination through the fused silica template. The template is then separated from the disk, which now contains a relief image corresponding to the template pattern. The same process used to pattern disk substrates can also be employed to pattern template substrates, thus providing the ability to replicate the master template.
Figure 3. Uniform residual layers are obtained across the entire substrate, independent of pattern density and feature size variations. |
The inkjet-based drop-on-demand approach to dispensing imprint material provides several advantages over a traditional spin-coating approach. It is a straightforward and fast method for depositing material on surfaces of arbitrary shape, such as the annular disk-type substrates used in the hard drive industry. The process is inherently cleaner than spin-coating methods, and front- and backside edge bead removal is not required. Elimination of expensive two-sided coat-and-bake systems provides a compelling cost advantage. Drop-on-demand technology also allows the imprint tool to selectively place imprint resist to match the local pattern density of the template. The imprint material is dispensed as individual drops approximately 5pL in volume; roughly 2??104 drops are dispensed to pattern a disk surface with a total volume of ~100 nL. The precision of the dispense technique makes it possible to compensate for localized variations in pattern density across the template, and thus maintain a highly uniform residual layer across the entire substrate surface. The ability to form consistent residual layers is pictured in Fig. 3 for both nanoscale structures (such as discrete data tracks or bits) and microscopic patterns (such as servo regions). Because the imprint liquid is dispensed with high precision to match the local volume required by the template patterns, there is no waste stream of excess resist material or rinse solvent. Whereas a spin-coating process requires approximately 1 mL of resist to coat both sides of a single disk, the same volume of imprint resist is sufficient for patterning ~5000 disks.
Template replication using S-FIL
Industrial forecasts suggest that the market demand for hard disk recording media will reach 109 units in the next few years. Fabrication of patterned media to meet this demand will require a large supply of imprint templates: the lifetime of a single imprint template is anticipated to be ~104 imprints, suggesting that at least 105 templates will be required. It is not feasible to use electron-beam patterning directly to create this volume of templates. Instead, a master template???created by direct patterning with an electron-beam tool???will be replicated many times to produce the required supply of working templates for patterning disk media. Several replication schemes are being considered, including single-step replication (master template to working template) and two-step replication (master template to sub-master template to working template) [5]. In both cases, template replication can be accomplished using the same imprinting process described here.
Imprint tool
Two-sided imprinting of disk substrates was performed with an Imprio HD2200???a fully automated UV-nanoimprint lithography (UV-NIL) tool specifically designed for patterned media applications. The Imprio HD2200 provides the high patterning fidelity that is characteristic of UV-NIL[6], with automated double-sided disk patterning capability and throughput of 180 disks/hr.???with a roadmap to generate media at rates over 1000 disks per hour in the near future. Patterned media applications typically require a modest level of alignment (tens of microns) to ensure that the patterns are concentric to the spindle axis of the disk drive unit. The Imprio HD2200 provides alignment of the template pattern to the disk substrate within 10µm.
Figure 4. Photographs showing both sides of an imprinted disk with an inner diameter of 20 mm and an outer diameter of 65 mm. |
An example of a double-sided imprinted disk is shown in Fig. 4. The disk shown has an inner diameter of 20mm and an outer diameter of 65mm; the patterned area covers the surface in the radial span of 13.5 to 30.5mm. The example shown was imprinted using a test template with DTR line/space patterns at a track pitch of 300nm, with corresponding servo patterns.
Quality of imprinted pattern
Feature quality is critical for both DTR and BPM, and preliminary measurements have been performed to characterize each pattern. SEM images were acquired with a JEOL JSM-6340F field-emission SEM at 4 kV and a working distance of ~8mm. A thin layer (~2nm) of AuPd alloy was sputtered on the samples prior to microscopy. SEM image analysis was performed using Simagis software provided by Smart Imaging Technologies (Houston, Texas). Image processing analysis included normalization of image brightness and removal of angular tilt from line/space images, followed by a threshold function to locate feature edges. Typical analyses are shown in Fig. 5, with corresponding results in Tables 1 and 2. For the BPM example (Fig. 5a), data was collected across an array of 679 imprinted bits over a total area of 2.12µm2. The feature pitch is measured to be 50.0nm along the track axis, with 48.8nm distance between track axes. This array corresponds to a recording density of 0.25 TB/in2. Mean values of feature height (along track axis) and width were measured to be 35.7nm and 36.3nm, respectively, with standard deviation of ~2nm. As an example of DTR patterning, Fig. 5b shows imprinted lines with a design pitch of 70nm. The mean linewidth of the 15 measured lines was 43.47nm, with a standard deviation of 2.08nm; linewidth roughness was ~3nm (Table 2). The pattern quality typified by these examples is acceptable in the current phase of technology research, but improvements will be required to achieve robust data storage densities of 1 TB/in2 and above.
Figure 5. SEM image analysis for a) 50nm pitch BPM and b) 70nm pitch DTR patterns. |
null
|
null
Conclusions
The ever-increasing storage density of hard disk drives is approaching its limits imposed by the superparamagnetic effect. The practical limitations of superparamagnetism can be avoided by patterning the magnetic domains, but this patterning requires industrial-scale lithography at unprecedented levels of feature resolution, pattern precision, and cost efficiency. S-FIL has demonstrated the potential to fulfill these requirements. Rotating-stage electron-beam lithography tools have been developed to create large-area concentric patterns; ongoing tool development efforts address the stability of these tools for creating highly uniform features during continuous patterning processes that last several days. The master template thus patterned is replicated via imprinting to supply the working templates that are used to pattern hard disk substrates in high-volume manufacturing. An imprint lithography tool has been developed to meet the requirements of two-sided disk imprinting at a rate of 180 disks/hr. Further improvements in template mastering and replication processes, together with imprinting tools with increased throughput, will enable fabrication of patterned media for the next generation of hard disk drives with storage density exceeding 1 TB/in2.
Acknowledgments
The authors are grateful for assistance provided by David Kuo, Dieter Weller, Zhaoning Yu and Justin Hwu (Seagate Technology), XiaoMin Yang and Shuaigang Xiao (Seagate Research), Mike Miller, Cynthia Brooks, Niyaz Khusnatdinov, S. V. Sreenivasan, Gary Doyle, Steve Johnson, Chris Jones, and Paul Hofemann (Molecular Imprints, Inc.), and Scott Dhuey, Bruce Harteneck, Erin Wood, and Stefano Cabrini (Lawrence Berkeley National Laboratory). Portions of this work were performed at the Molecular Foundry at Lawrence Berkeley National Laboratory, which is supported by the Office of Science, Office of BES, of the US DOE under Contract No. DE-AC02???05CH11231.
References
- B. M. Chen, T. H. Lee, K. Peng, and V. Venkataramanan, “Hard Disk Drive Servo Systems”, 2nd ed. (Springer, NY, 2007).
- T. Oikawa, M. Nakamura, H. Uwazumi, T. Shimatsu, H. Muraoka, and Y. Nakamura, IEEE Trans. Magn., vol. 38, pp. 1976-1978 (2002).
- I. Yoneda, S. Mikami, T. Ota, T. Koshiba, M. Ito, T. Nakasugi, and T. Higashiki, Proc. SPIE ??? Emerg. Lith. Tech., vol. 6921, pp. 14-21.
- Colburn, S. C. Johnson, M. D. Stewart, S. Damle, T. C. Bailey, B. J. Choi, et al., Proc. SPIE - Emerg. Lith. Tech., vol. 3676, pp. 379-389 (1999).
- M. Miller, G. M. Schmid, G. F. Doyle, E. D. Thompson, and D. J. Resnick, Microelectronic Eng., vol. 84, pp. 885-890 (2007).
- F. Hua, Y. Sun, A. Gaur, M. Meitl, L. Bilhaut, L. Rotkina, J. Wang, P. Geil, M. Shim, and J. A. Rogers, Nano Lett., vol. 4, no. 12, pp. 2467-2471 (2004).
Gerard M. Schmid received his PhD from The U. of Texas at Austin and is currently a senior template scientist at Molecular Imprints; E-mail:[email protected]
Doug Resnick received his PhD in solid state physics from Ohio State U.. He is vice president of template technology at Molecular Imprints.
Dwayne LaBrake received his PhD in physical chemistry from Loyola U. of Chicago. He specializes in nanoimprint lithography processes and applications, and is director of applications at Molecular Imprints.
Gene Gauzner received his MSEE degree from the National Technical U. in Kiev, Ukraine. He is currently senior engineering manager at Seagate Technology.
Kim Y. Lee received a Ph.D. in electrical and electronics engineering from the U. of Glasgow in the U.K. He is currently a development senior manager at Seagate Technology.