Nanoimprint lithography enables patterned tracks for high-capacity hard disks
03/01/2005
Advances in nanoimprint lithography (NIL) are being employed to introduce new manufacturing processes for next-generation thin-film disks, which use patterned substrates for land-and-groove structures that magnetically isolate individual data tracks. The concept, discrete track recording (DTR), promises to significantly increase data storage in hard-disk drives because recording heads will no longer be used to define data tracks. Physical tolerances for recording heads also will be relaxed, and other performance improvements are expected from DTR.
Intense competition in hard-disk drives (HDD) has resulted in data storage capacity increasing four orders-of-magnitude in the past 20 years, but at the same time, cost pressures have been tremendous on suppliers. During that 20-year period, HDD unit prices have declined twenty-fold from $2000 for a 10MB HDD to $100 for today’s 100GB drive. The compound annual growth rate (CAGR) for platter capacity in the hard-disk industry was about 100% in the 1990s, which exceeded the pace of Moore’s Law for transistor and bit integration on ICs in the semiconductor industry.
To meet ever-increasing storage demands, media bit size has been reduced by approximately a factor of 10, with bit size currently at about 0.034×0.16µm. For read/write heads and drive electronics, the signal-to-noise ratio (SNR) is related to the number of magnetic grains/bit on platter substrates. Therefore, the magnetic grain size must be reduced to maintain overall media recording performance. Ideally, the number of magnetic grains should be constant, but there are limits to grain size reduction. Thus, issues such as thermal stability have become more critical, and new technologies have been implemented to tackle these obstacles.
Although the pace of platter capacity increase has slowed in recent years to a more manageable 40% CAGR, intense demands on HDD suppliers continue. The disk-drive industry faces many questions about which direction to take and what technologies to pursue to keep pace with future storage demands. Some new technologies, such as synthetic antiferromagnetic (SAF) longitudinal media, have moved into volume production for 80-120GB platters, and others, such as perpendicular magnetic recording (PMR), are just beginning to see commercialization. Ultimately, the HDD industry views bit-patterned media technology as the most promising approach to achieve areal densities of 1Tbit/in.2 for 95mm disks (3.5-in.) with 1TB storage capacity in the next decade.
Looking for the next immediate step, Komag researchers have investigated DTR as a means to increase HDD storage capacity by patterning disks with magnetically isolated predefined data tracks. DTR is not so much an alternative to longitudinal or perpendicular recording techniques but a complementary concept that improves the performance and extends the lifetime of either recording technology. In addition, DTR affords a potential bridge from today’s recording technologies - which use write heads to format and define tracks on disks - to the long-term solution of bit-patterned media.
The concept of DTR and its associated benefits have been known since the early 1960s, but practical implementation of the technology in manufacturing has been elusive with no cost-effective means available to fabricate a DTR disk [1]. However, advances in NIL offer new opportunities to achieve DTR media in a practical manufacturing process. Komag has successfully demonstrated improved magnetic track isolation on hard disks by using lithographically defined tracks comprised of land-and-groove structures. The company plans to commercially introduce this technology for 200GB and greater platters by 2007.
How DTR disks work
In DTR media, land-and-groove structures are defined with lithography steps in a concentric or spiral pattern on the disk substrate. The “land” areas become the data tracks and the grooves provide for magnetic isolation between tracks as shown in Fig. 1. Conversely, the data surface for today’s state-of-the-art hard disk media is continuous with the write head being used to define the data tracks. Isolation between tracks is determined by the positioning accuracy of the head. The resulting erasure bands on the edge of the tracks cause noise in the read-back signal of disk drives. This noise is mitigated by using a read head 60% the width of the write head.
 Figure 1. a) A disk platter being processed in a NIL chamber; and b) the DTR disk concept. |
A primary advantage of DTR media is the elimination of these erasure bands because patterned grooves provide magnetic isolation between data tracks. The recording head cannot write in these grooves due to spacing loss. Consequently, the read head can be fabricated to the same width as the write head. This results in a 1.25dB improvement in SNR, which is extremely significant for an industry that often squeezes its technology for tenths of a decibel. An additional consequence of the land/groove structure is that the physical tolerances on the recording head are relaxed. As a result, head manufacturers also will benefit from DTR because it will offer higher yields, lower cost, and longer life for head-making technologies.
Furthermore, DTR media are expected to improve disk reliability by enabling manufacturers to embed servo information in the patterned track structures before products are shipped. In today’s drives, recording heads normally are used to write servo information, which is a costly, time-consuming step in hard-drive manufacturing.
Making DTR disks
The process sequence for manufacturing a cost-effective DTR disk using NIL technology is shown in Fig. 2. The first step is to create a master, which is then used to make Ni stampers for imprinting the desired track patterns on disk substrates. Initially, in development, optically mastered stampers with a track pitch (TP) of 800nm were used because they are readily available. However, it is difficult to produce small groove geometries below a 400nm TP using optically mastered stampers. Consequently, an electron-beam recorder (EBR) is being used to make masters on 8-in. silicon wafers. These masters are used to create Ni “fathers” with standard electroforming methods. Multiple Ni “mothers” are then formed, followed by multiple Ni “son” stampers. EBR-mastered Ni stampers have produced patterned disks with TP down to 127nm, which corresponds to a bit density >200Gbit/in.2
 Figure 2. The sequences of NIL fabrication process steps for a DTR disk. |
After stampers are made, the track pattern is transferred from the nickel stamper onto a polymer-coated substrate (a 95mm dia. NiP-plated AlMg disk) using a hot-emboss nanoimprint process. The polymer is then dry etched to remove any residual material left in the grooved area. The etching step also exposes the underlying NiP surface of the substrate. Using the remaining polymer layer as a stencil, an electrochemical wet-etching process is then used to form grooves into the NiP substrate. Subsequently, the polymer stencil is stripped, and the substrate is processed in a fashion similar to that used for today’s magnetic recording media. This final step sputters a hard magnetic storage layer and carbon overcoat on the disk substrate.
 Figure 3. A cross-section TEM micrograph shows a DTR disk with 800nm track pitch. The DTR disk has been sputtered with magnetic media layers and overcoat. |
Figure 3 shows a cross-section TEM micrograph of an 800nm TP DTR disk, which has been sputtered with magnetic media layers. The land width is ~500nm and the groove width is ~250nm with a depth of 60nm. Spin-stand recording tests were performed on these DTR disks to measure relevant recording characteristics [2]. As an example, Fig. 4 shows the SNR for a DTR disk (800nm TP) at 100kFCI (kilo flux changes/in.) and 500kFCI. At 100kFCI, there is a 25dB reduction in the SNR and a 35dB reduction at 500kFCI. The smaller reduction in SNR at 100kFCI results from the fact that the magnetic spacing loss in the grooves is less effective at lower linear densities, but in either case, the measurements demonstrate adequate rejection in the grooves.
 Figure 4. Signal-to-noise ratio is plotted for a DTR disk with 800nm TP at 100kFCI and 500kFCI as a function of cross-track position. |
Although NIL has been successfully used to fabricate a DTR disk with demonstrated recording performance advantages over standard hard-disk media, there are some major challenges to address before NIL becomes a cost-effective process for disk manufacturing. Consequently, Komag and EV Group have formed a joint-development effort to address key remaining issues for mass production of DTR disk platters. In December, EV Group also introduced an automated hot embosser system to support double-sided imprinting of patterned media.
Beyond overall cost and yield issues, unique challenges in applying NIL to DTR disk manufacturing are:
- simultaneous alignment and imprint of concentric or spiral tracks on both sides of a large-area substrate (i.e., a 95mm dia. disk) with eccentricity <5µm;
- process cycle times <10 sec or about 500 parts/hr;
- stamper lifetimes on the order of 10,000 cycles; and
- zero tolerance for asperity defects on the substrate >5nm in height.
Prototype tools have been developed for all of the DTR process steps. Single-sided DTR disks with a TP down to 380nm have been manufactured with uniform imprinting out to within 2.5mm of the disk edge and eccentricity at about 15µm. Defect testing has been performed on these disks by flying a head over the DTR surface and counting discrete contacts using an acoustic emission sensor attached to the head as the fly height is gradually reduced. Data for counts vs. fly height for several manufacturing runs over the course of a year showed that a standard state-of-the-art hard disk is capable of zero counts or hits even at fly heights approaching 0.2µin., or about 5nm, which is a necessary requirement for current hard-disk media.
Initially, DTR disks exhibited contacts in the 10,000 range even at much higher fly heights. However, improvements in contamination control led to fly heights for DTR disks comparable to a standard hard disk. In the manufacturing test runs performed in October 2004, 13 disks were pulled from a 100-disk run. Twelve of the disks showed no contacts down to 0.25µin., while one disk showed a single contact at 0.25µin. These results demonstrate that NIL can be used to fabricate DTR disks with fly height characteristics and hard defect densities comparable to state-of-the-art hard disks.
What’s next
The next crucial step in applying NIL to mass production of DTR disks involves further development of the NIL press. Current capabilities in all of the other process steps meet or demonstrate a path to high-volume manufacturing requirements. Komag and EV Group have entered into an agreement to develop, license, and sell a high-volume NIL tool for mass data storage. This production tool will be based on EV Group’s new automated hot embosser system, the EVG570PMI.
The manufacturing requirements for high throughput require fully automated, simultaneous processing and inspection of patterned disks. The hot embosser system has been designed for a simplified DTR process flow, which includes substrate heating, imprinting, and integrated imprint inspection. Based on the imprint alignment results and defined inspection criteria, the system is capable of deciding whether to unload substrates for additional process steps or to reject imprinted disks that fail inspection. The application requires 100% inspection because imprint accuracy is critical for fully functioning DTR disks. In the current stage of development, 30 sec cycle times are being achieved with integrated systems, resulting in a throughput of 120 disks/hr. The goal is to increase throughput to 500 disks/hr.
The hot embosser system employs on-line alignment inspection features to feed positioning data in a continuous loop to the imprinting head. This capability enables the system to accommodate stamp movement and changes in disk positioning during high-volume manufacturing. The NIL process requires precision temperature control and uniformity over the disk to allow rapid imprinting and demolding. The pattern replication fidelity is ensured with specialty material for high-throughput imprinting. Automated handling stations also are used to address requirements for positioning accuracy and high-volume manufacturing.
Summary
Nearly 40 years after initial conception, DTR technology appears to be moving toward commercialization, based on new advances in NIL and highly automated tool capabilities for volume manufacturing. DTR disks manufactured with NIL processes could become an important bridge to bit-patterned media, which could push the areal density of disks to the 1Tbit/in.2 realm by the next decade. Furthermore, stringent requirements for cleanliness and alignment in DTR disk production, along with high-manufacturing throughput, promise to drive NIL development for other applications outside of the HDD industry.
References
- L.F. Shew, IEEE Trans. Broadcast and Television Receivers, BTR-9, pp. 52-62, 1963.
- D. Wachenschwanz, W. Jiang, E. Roddick, A. Homola, P. Dorsey, et al., “Design of a Manufacturable Discrete Track Recording Medium,” IEEE Trans. Magn., Feb. 2005.
Thomas Glinsner is deputy chief technology manager, EV Group.
Peter Hangweier is a product manager at EV Group.
Helge Luesebrink is director of the Advanced Lithography Business Unit at EV Group, DI Erich Thallner Strasse 1, Schärding, A-4780 Austria; ph 43/7712-5311-4030, fax 43/7712-5311-4600, e-mail [email protected].
Paul Dorsey is a principal engineer for the DTR project at Komag Inc., 1710 Automation Parkway, San Jose, CA 95131; e-mail [email protected].
Andrew Homola is a senior technical staff member at Komag Inc.
David Wachenschwanz is executive director of product development and integration at Komag Inc.