Issue



Electron beam and nanoimprint lithography for patterned media


09/01/2008







B. Heidari, M.Beck and K.Mason, Obducat Technologies Malmo. Sweden

The drive for higher capacity in storage devices has forced disk drive manufacturers to develop technologies to improve storage density. Media patterning allows the magnetic isolation to be improved, thereby increasing the recording capacity of the disk. The patterning requires a mastering technology to write a template and a printing technology to mass produce the patterning onto disk substrates. The lithographic demands, however, far outstrip the dimensions obtainable through the widely accepted UV lithography approaches. In this article, we look at two systems developed to meet this manufacturing need.

The drive for storage memory is endless. Computers and multimedia abilities continue to add sophistication, and the desire to store and instantly access huge volumes of data continues to accelerate???disk media storage at the density of a TB/in.2 is a near-term reality.

This presents complex technology challenges for the disk drive manufacturer. The physical property limitations of the disk itself ???the polycrystalline nature of the magnetic medium???sets a limit for the maximum storage density of the written information. At very high densities, above 500GB/in.2, the small dimension of the grains (~10nm) is not sufficient to stabilize the magnetization direction over long time periods due to issues such as thermal stability. A way to circumvent this limitation is to use patterned media, where discrete track recording (DTR) and bit-patterned media (BPM) are the most promising technologies. This patterning is done by lithography, however the requirements are not at all similar to those of conventional CMOS lithography. Pattern resolution exceeds the current semiconductor lithography roadmap by more than one generation, and the common stepper technology would not be suited to this application due to stitching error???the small misalignments that occur when one field is placed next to another, feature to feature. Patterned magnetic media, in combination with other techniques, is expected to extend the data densities up to 50TB/in.2 [1]. Toshiba Corporation recently announced a prototype hard disk drive (HDD) that uses discrete track recording (DTR) technology, which enabled a capacity of a record-breaking 120GB on a single 1.8-in. disk. Toshiba plans to release product to the market with DTR at the end of 2009.

By using patterned media, data densities can be increased beyond the superparamagnetic limit. Ordinary thin-film media consist of small, single domain magnetic grains that are exchange-decoupled from one another. The magnetic axes of these grains are randomly distributed. For a good signal-to-noise ratio (SNR) in a written bit, each bit must contain a number of grains. To increase storage densities, bit sizes (and therefore grain sizes) must be reduced to much less than 10nm in diameter. However, the grain size cannot be reduced arbitrarily, because as the grain volume decreases, thermal effect becomes significant. When the magnetic grains are decreased in size, the magnetization directions of the grains can reverse spontaneously. This leads to loss of recorded data.

Manufacturing patterned media presents new challenges and its implementation requires two technologies, a writing technology (electron beam recorder) and a printing technology (nano-imprint lithography). We will look at these two systems in greater depth.

First, an electron beam recorder (EBR) is used to define the pattern on a master disk and second, this master is used as a stamp (template) for replication on the media disk substrate using nano-imprint lithography (NIL).

A new electron beam lithography system has been developed that has a rotating spindle on a linear slider that moves the substrate in a radial direction. The software and encoders allow data conversion to be performed on-the-fly while the pattern is written, thereby eliminating conversion time losses and memory requirement known from conventional vector scan e-beam systems

The process used for replication of the master onto magnetic media disks is NIL, employing a two-step imprint process. First, the generation of an intermediate polymer stamp (IPSR)[2] and (second) a simultaneous thermal and UV (SUV) imprint process[3]. Both imprint tool and process are capable of high-volume manufacturing and the combination of EBR-mastering with NIL enables high-volume manufacturing with a 600 HDD disk/hr. per lithography system.

EBR

EBR is based on a thermal field emission cathode with high current stability and brightness. It operates as a paraxial gaussian beam system at an acceleration voltage of 50kV. The lithography of pattern structures is derived from the combination of disk rotation, translation, high-speed sub-field deflection, and beam blanking.

The electron beam column was designed mainly for high-resolution, high-speed beam blanking, high-speed beam deflection, and long-term stability. A thermal field type emitter (TFE) was used to obtain a high-intensity and stable electron beam. The beam energy of 50keV was selected to decrease influence of the back scattering. To ensure beam stability, the control system of the electron beam column is feedback-coupled for stable and accurate control. The high-speed beam blanker and a stigmator installed above the beam blanker were adopted to write precise pits without shape-distortion caused by beam blanking. The rise and fall times of beam blanking were measured to <5ns. In addition, a high-speed sub-beam deflector with resolution of 1nm was added for improved definition positioning of servo patterns. A dynamic focus system allows for stable recording through the exposure area of the substrate. It consists of a height sensor and focusing lens in the electron beam column. The height sensor measures the fluctuation of the substrate height, and the focal distance of the objective lens is controlled in accordance with the height information.

The substrate is attached directly to the turntable to minimize rotation imbalance. A laser interferometer and a rotation laser encoder ensure the substrate positioning accuracy in nanometer level, while the stage is rotated and moves in a radial direction. In addition, an active vibration isolation system isolates the work chamber from undesirable disturbance.


Figure 1. SEM picture shows the Blu-ray master disk made by EBR. The linewidth is 150nm.
Click here to enlarge image

The system is controlled by software that was developed to generate the pattern for both magnetic and optical medium writing. The patterns are calculated on-the-fly during the exposure to minimize the memory requirements and total exposure time. The software addresses accuracy of 1nm in both down- and cross-track directions. The writing strategy for optical disk is fast exposure speed. The writing strategy for HDD application is accuracy, therefore the HDD pattern is written at lower speed to maintain the required accuracy.


Figure 2. The graph shows the track pitch analyses made on an optical disk master made by the EBR, showing that the track pitch variation is <6nm.
Click here to enlarge image

As an evaluation method of the recording stability, we fabricated a nickel stamp and measured the uniformity of pattern by AFM and FIB analyses (Fig. 1). A summary of the track pitch accuracy on the Blu-Ray nickel stamp is shown in Fig. 2. In the HDD application, the structures are measured in several ways to determine the accuracy and resolution. A summary of these measurements is shown in the Table.

Click here to enlarge image

null


Figure 3. SEM images of the DTR structure in nickel stamp.
Click here to enlarge image

In these experiments, silicon substrates coated with the resist (ZEP-520A, Nippon ZEON Co., Ltd.) were used in the recording process. The recording parameters were chosen dependent on speed or resolution requirements. The beam currents were chosen in the range of 0.9nA to 20nA. Constant linear velocity (CLV) writing strategy was used in both optical and HDD disks to ensure the dose uniformity. The exposed substrates were developed and sputtered with a metal seed-layer as a conductor for subsequent Ni electroplating process. All measurements made on the resulting nickel stamps showed a total thickness of 300µm (Fig. 3).

NIL

Thermal NIL[4-6] and UV NIL[7,8] are well known as process technologies in many applications. Thermal NIL is a widely distributed method to transfer a mold pattern to a polymer layer. The polymers used in the process are mostly thermoplastic materials of varied composition, molecular weight, and a variety of glass transition temperatures. Some of the most common thermoplastic polymers are polyacrylate-based materials and their derivatives. Others are available and one may choose the material fitting best to the process and its final application.

Intermediate polymer stamp

For manufacturing, an intermediate polymer stamp (IPS), which is a consumable copy of the master stamp, is made within the system. The IPS extends the life of the master stamp by using a relatively soft material. Additionally, it facilitates easy separation of the stamp and printed substrate. Because each IPS is used once, any particulate or contaminant that may have entered the process is carried off by its disposal. Fig. 4 illustrates the reduced contaminants over the processing of 250 sample wafers with the IPS process.


Figure 4. Reduced contaminants over the processing of 250 sample wafers with IPS process.
Click here to enlarge image

null

Simultaneous thermal and UV process

The simultaneous thermal and UV process (STU) has all the benefits of a pure UV process and allows the user to select spin-coatable pre-polymers that are in a dry state after the soft bake.


Figure 5. Process flow scheme for a simultaneous thermal and UV nanoimprint lithography process (STU-NIL).
Click here to enlarge image

These pre-polymers have thermoplastic properties, a very low glass transition temperature, and can be printed at temperatures ranging from room temperature up to 100??C. The pre-polymers have a sufficient number of reactive sites that can be activated for cross-linking by UV radiation, which takes place during a post-exposure bake that is executed at the same temperature as the other process steps. A typical process flow comprises the following steps and is illustrated in Fig. 5.

  1. Preparation of substrate by cleaning process
  2. Coating of pre-polymer layer on substrate
  3. Soft bake
  4. Temperature conditioning of stamp and substrate
  5. Imprint (pressure phase)
  6. UV-exposure
  7. Curing phase
  8. Pressure release
  9. Stamp/substrate separation
  10. Ashing of residual layer with O2-plasma (typically 5-15nm)
  11. Pattern transfer (user specific). Steps 5-9 are executed at constant temperature thereby eliminating the risk for pattern distortion known from hot embossing processes due to the presence of a heating and cooling phase.


Figure 6. AFM cross-section analysis of the residual layer thickness (top) and imprint uniformity (bottom). The residual layer thickness was measured to 3.6nm ?? 2.1nm and the FFT spectrum plot revealing a sharp peak at a spectral period of 62.5nm.
Click here to enlarge image

An example of an STU imprint obtained on a 2.5in. HDD substrate is given in Fig. 6. The two AFM-images show the data track with 30 nm half-pitch. The upper image shows a residual layer analysis scan obtained after scribing the imprinted substrate with tweezers followed by measurement of the step height. The residual layer analysis for this pattern gave 3.6 ?? 2.1nm on the entire pattern area. The lower image shows a non-scribed area with an FFT spectrum plot revealing a sharp peak at a period of 62.5nm. The mean imprint depth is 37nm.

Acknowledgments

STU, IPS, and Soft Press are registered trademarks of Obducat Technologies.

References

  1. M. H. Kryder, Carnegie Mellon U. Magnetic Recording Roadmap, Technical Symposium IDEMA, Dec. 2007.
  2. M. Beck et al., 50th EIPBN (2006) 167 (IPS).
  3. M. Beck et al., 50th EIPBN (2006) 167 (STU).
  4. NTT (Nippon Telegraph and Telephone Public Corp.) Molded Mask Method, JP patent 54-22389, 1979.
  5. Kuwabara et al., U.S. Patent 5 259 926 , 1993.
  6. S.Y. Chou et al., Applied Physics Lett. 67, 1995, 3114.
  7. NTT (Nippon Telegraph and Telephone Public Corp.), Molded Mask Method, JP patent 53-22427, 1978.
  8. M. Colburn el al., SPIE, 3676 (1999) 379-389.

Babak Heidari holds a PhD in education for Solid State Physics from U.of Lund and has experience in process development of micro and nano lithography, design of electron beam lithography and imprint lithography systems, fabrication methods and process development. He is currently CTO at Obducat Technologies AB, Sweden.

Marc Beck recieved a PhD in Technical Physics at the Department of Solid State Physics from the U. of Lund and specializes in nano-imprint lithography. He os a Senior Application Engineer at Obducat Technologies AB.

Ken Mason, holds an AA in Marketing from Bucks County Community College and attended Philadelphia Biblical U. and has published papers and articles relating to nanoimprint and X-ray lithography. He is Business Development Manager for Obducat in North America E:mail: [email protected].