INTRODUCTION
The development of semiconductor integrated circuits (IC) and mass-data acquisition and storage systems set in motion the transformation of the human enterprise that will continue shaping world economies as well as political and social systems for at least the next half-century.[1,2] However, the end of the technology road map for semiconductors is rapidly approaching, and new technologies will be needed to cross over to a new era in integrated electronic systems.[3] Many highly innovative materials and device ideas have been proposed to help the transition, driven by various nanotechnology initiatives.[4-9] The realization of these ideas is subject to the availability of cost-efficient materials synthesis and device fabrication capabilities. While new nanofabrication technologies are constantly emerging, the economic reality calls for the adaptation of the vast IC fabrication tool set developed over the past several decades to fabrication at nanoscale.
Ion-beam technologies based on wide-area ion beams, focused ion beams (FIB), and shaped multiple ion beams have generated substantial interest in applications in nanostructured materials and nanodevice fabrication and prototyping.[10-16] In this entry, the utilization of ion beams for nanomaterials synthesis and fabrication of nanoscale devices is reviewed. Special emphasis is paid to FIB processing of nanoscale magnetic devices[17] because FIB has played a critical role in the successful development of magnetic recording at areal densities beyond 100Gbit/in2, which is one of the very few examples of a fully functional nano-technology with a significant economical impact.
ION-BEAM SYNTHESIS OF NANOSTRUCTURED MATERIALS
Ion-beam implantation has been one of the key IC manufacturing tools for controlled incorporation of impurities (dopants) into predefined regions on semiconductor wafers (matrix).[18] A remarkable feature of ion implantation is that nearly any element from the periodic table can be implanted into any matrix, making the technology highly versatile. Well-known examples include implantation of group V elements such as phosphorus or arsenic into silicon used for defining n-type Si regions and implantation of group III elements such as boron used for p-type Si regions. Ion bombardment is used for hardening of steels via formation of nitrides at the metal surface. Energetic processes based on ion-assisted deposition and ion-beam deposition have been used for the synthesis of metastable phases of various materials (e.g., diamond or cubic boron nitride).[19-21]
In the applications mentioned above, ion-beam implantation is used for the synthesis of homogenous materials where implanted species form chemical bonds with the host material. Ion implantation is typically followed by the annealing step to heal the ion-induced damage of the crystalline structure of the matrix and to activate the implanted species.
If the dose of ion implant exceeds the miscibility limit in the host materials or if the ion species and the host material are immiscible, the annealing step leads to the precipitation of a new phase in the form of clusters (or nanoclusters) as illustrated in Fig. 1. Such precipitates/nanoclusters can lead to significant modification of the electrical, mechanical, magnetic, and optical properties of the host material. Formation of nanoclusters, which is often referred to as ion-beam synthesis, has been documented for a large variety of ion species and host materials. The examples include elemental metal and semiconductor nanocrystals in inert hosts, compound nanocrystals formed as a result of chemical reaction between the ion implant and the matrix, metal alloys and compound semiconductors (II-VI, III-V, and IV-VI) formed by coimplantation of multiple reactive ion species into the host.[22-31]
Among potential applications of nanomaterials formed by ion-beam synthesis are all-optical switching that utilizes highly nonlinear optical response of metal nanoclusters embedded into nonconducting matrix, opto-electronic devices based on the unique optical response of semiconducting quantum dots, nonvolatile memory, spintronics, and single-electron transistor devices.
Fig. 1 A schematic of ion-beam syn esis process flow.
FOCUSED ION-BEAM FOR RAPID PROTOTYPING OF NANOSCALE DEVICES
The principle of operation of an FIB system is similar to the principle of operation of a typical electron-beam (e-beam) system used in electron microscopy (SEM, TEM) or e-beam lithography. The key difference is the source: FIB systems use a liquid metal ion source (LMIS). In such an ion source, a reservoir of liquid metal is in contact with the dull end of a sharp needle (usually made of tungsten). Metal from the melt flows toward the sharp tip of the needle biased at a relatively high voltage (~10kV) with respect to the extractor electrode. The source is designed such that the high electric field between the extractor electrode and the needle tip is sufficient for field ion emission. Extracted metal ions travel through the beam shaping/focusing column, which is similar to a typical e-beam column with the main difference being the polarity of the voltages applied to accelerating and focusing components. A typical LMIS uses gallium (Ga), which melts at 29.78°C (slightly above room temperature); this significantly simplifies the design and reliability issues.
The resolution limits of an FIB system are defined mainly by the tool design (such as finite penumbra due to finite source size and the precision of the beam shaping column) and not by diffraction limits, which are in deep subangstrom range.[39] The utilization of ions adds great versatility as compared to e-beam systems:
1. Focused ion beams can be used for direct etching of device patterns on a wafer using physical sputtering of materials with heavy energetic Ga ions.
2. Focused ion beams can also be used for highly localized material deposition where the ion beam is used as a material carrier. In this application, a vapor source (either gaseous or near-the-melting point solid) is inserted into the ion-beam path.
3. Focused ion beams can be effective for local ion implantation and ion mixing.
4. As in e-beam lithography, a pattern can be written into resist using secondary electrons for resist exposure.
5. An important feature of a typical FIB system is that it allows high-resolution imaging using either secondary electrons or ions.
Owing to the direct write capability, FIB is an ideal fabrication tool for the rapid prototyping of nanoscale devices. For example, with FIB, even in the academic environment, one complete iteration necessary for making a magnetic recording transducer (write/read head) with sub-100nm features takes less than 1hr.[40] For comparison, with e-beam lithography in the streamlined industrial environment, it takes months to go through one iteration cycle. In practice, it usually takes several iterations to complete the design of a next-generation recording head. Such tremendous time and resource savings with FIB is a critical advantage for prototyping at nanoscale.
An additional advantage resulting from larger ion mass, as compared to the mass of an electron, is the fact that the FIB is much less influenced by stray fields. For example, in the magnetic data storage industry, one of the main obstacles in implementing e-beam lithography in the fabrication of magnetic recording heads suitable for areal densities above 100Gbit/in2 is the frequent occurrence of the physical shift in patterned features due to the interaction between the e-beam and magnetic thin films.[41] This shift, depending on the properties of a thin film, could be as large as 100 nm (for Ni45/Fe55). For comparison with FIB, at equal conditions, this shift could be reduced to an unnoticeable level, as shown in schematic diagrams in Figs. 2A and B, respectively.
The FIB-based fabrication not only has the advantage of using fewer steps than lithography-based processing but also has the potential to pattern magnetic materials with higher resolution than what is achievable with e-beam lithography. The study on magnetic devices at Seagate showed that with 50% reproducibility, e-beam and FIB are capable of feature sizes of 30 and 10 nm, respectively, with individual devices made with FIB in sub-10 nm feature size range.
Fig. 2 Schematic diagrams showing trajectories of focused (A) electron and (B) ion beams.
The FIB is especially attractive for a small research laboratory/an academic environment. Patterns with nanoscale features can be defined without using complex and expensive fabrication facilities. With the FIB’s direct etch/write capability (no mask-based lithography is necessary), a university just needs to have one FIB system available to trim relatively large devices into nanoscale devices of different geometries for further experimentation and testing. The ability to rapidly fabricate at nanoscale for further characterization and optimization makes FIB an indispensable tool for the rapid development of emerging technologies in the field of nanotechnology.
ION-BEAM LITHOGRAPHY
Proximity and projection ion-beam lithographies (IBL) can be thought of as an extension of the FIB technology discussed above in that multiple ion beams are used to simultaneously write the pattern into resist.[42,43] Note, IBL is typically not used for the direct etching of the device patterns. In proximity IBL, illustrated in Fig. 3, a stencil mask is illuminated by a broad beam of energetic ions (e.g., helium) and the transmitted beamlets transfer the mask pattern to resist on a substrate. The theoretical resolution limit imposed by the range of secondary electrons and atom scattering within the resist is 1-2 nm (R. Kubena, unpublished data).[44-46] Using a small (a few micrometers) mask-to-wafer gap overcomes the practical limitations of diffraction and penumbra (due to a finite source size). Ion-beam lithography is on a fast track to practical application in the sub-10 nm domain by leveraging the many important developments of membrane mask technology in the proximity x-ray and projection ion and electron programs of the past two decades. It should be noted that it is also possible to generate a beam of neutrals that allows elimination of the negative effects of stray fields onto the beam.
A variation of IBL is projection IBL, which eliminates the need for stencil masks with ultrafine features. In projection IBL, the initial set of relatively wide beams is focused onto a substrate to achieve fine features.[47,48]
Fig. 3 A schematic of IBL: beamlets formed by an aperture plate write a periodic pattern in resist on a scanned substrate.
Fig. 4 M-H loops for 50 nm Co films with different Ga+ implantation doses: as-deposited Co film; 7 x 1015 ions/cm2; 3 x 1016 ions/cm2.
ION-BEAM APPLICATIONS IN MAGNETIC DATA STORAGE
Ion-Beam Modification of Magnetic Materials
Recently, there has been substantial interest in capitalizing on existing semiconductor processing technologies for magnetic thin film device fabrication. For example, e-beam lithography and chemical-mechanical polishing have been firmly accepted as next-generation tools for advanced magnetic recording applications.[49,50] Similarly, ion implantation and FIB processing, which are widely used in the semiconductor industry, are steadily gaining ground in magnetic thin film materials proces-sing.[12,51] To enable wide commercialization of these techniques in magnetic recording applications, a detailed study of these powerful fabrication tools with respect to their application to magnetic materials processing is necessary.[52-54] In this section, the effects of Ga+ ion implantation on magnetic properties of magnetic thin films and the application of FIB implantation for tailoring of magnetic properties at nanoscale are presented.[55]
Fig. 4 compares the M-H loops of an as-deposited Co film and two Co films with implant doses of 7 x 1015ions/cm2 and 3 x 1016 ions/cm2. Note, if all Ga+ ions were implanted into the film, 7 x 1015 ions/cm2 and 3 x 1016 ions/cm2 doses would correspond to ~1.5% (atomic %) and ^6.6% (atomic %), respectively. Substantial modifications to the magnetic properties can be observed. The coercivity of Ga+ implanted films increased by factors of 6 and 30 for 7 x 1015ions/cm2 and 3 x 1016 ions/cm2 implant doses, respectively. Also, the M-H loop squareness, S, defined as the ratio of the remanent magnetization, Mr, to the saturation magnetization, Ms, changed from a perfectly square loop in the case of as-deposited Co film with S = 1to S < 0.9 in the case of the higher implant dose. Reduced squareness and shearing of the hysteresis loop indicate a reduction in exchange coupling in the Ga+ ion implanted films. It should be noted that exchange coupling is one of the key parameters controlled during the development of the magnetic recording medium.
Fig. 5 compares atomic force microscopy (AFM) and magnetic force microscopy (MFM) images of the Co film implanted with a Ga dose of 7 x 1015ions/cm2. . Magnetization patterns in MFM images can be clearly tracked to the grain structure observed in AFM images. This suggests that ion implantation results in reduced intergranular exchange coupling. This is consistent with the observed shearing of magnetization loops (see Fig. 4). These data are similar to previously reported observations on permalloy thin films where an increase in coercivity was attributed to the introduction of extended defects.[56] In the present study such extended effects are represented by Ga-rich grain boundaries.
Fig. 5 AFM(left)/MFM (right) scans (3 mm x 3 mm) of Co films implanted at 3 x 1016 ions/cm2.
Fig. 6 Magnetic force microscopy image of (A) as-deposited Co film and (B) Ga+ ion implanted section in the Co film.
Local ion implantation can be achieved using FIB. Figs. 6A and B show MFM images of an as-deposited Co film and a Co film implanted with 50keV Ga+ ions at 1 x 1016ions/cm2 using FIB implantation, respectively. Similar to the above observations, a clear split of the film into a magnetic multi-domain state can be observed in Fig. 6B. At the same time, using AFM, no measurable modification of the film topography could be observed (RMS roughness in both cases was 0.9 nm). As discussed above, a dramatic change in magnetic domain patterns may be attributed to the diffusion of the implanted Ga+ ions ions to the intergranular boundaries, thus reducing the exchange coupling between the individual magnetic grains.
Fig. 7 Magnetic force microscopy images of four 300 nm x 300 nm Co squares formed by FIB etching in hcp Co film. Top left square is as-deposited Co; the remaining three squares are locally implanted with a ~1 x 1016 ions/cm2 dose using a 50keV beam.
Fig. 7 compares the MFM image of four 300 x 300 nm structures defined by FIB etching of the as-deposited Co film and of an intentionally ion implanted material. A transition to a magnetic multiple domain state attributed to ion implantation can be clearly observed.
In summary, ion implantation can be used for the modification of the properties of magnetic materials. For example, Ga ion implantation leads to the reduction of exchange coupling between magnetic grains and the formation of extended domain wall pinning defects. Implantation can be also achieved locally using FIB technology.
Nanofabrication of Magnetic Transducers
Magnetic recording heads
Magnetic recording has had very high areal bit density growth rates for the past 10-12 yr.[57] The resultant dramatic reduction of the bit size has led to the rapid scaling of the characteristic grain size and to the corresponding shrinkage of the thermal activation volume in conventional polycrystalline media. This, in turn, has led to challenging issues related to long-term data stability. Medium magnetic stability is controlled by the ratio of the magnetic anisotropy energy, KUV, to the energy of thermal fluctuations, kBT, where KU is the recording layer magnetic anisotropy energy density, and V is the thermal activation volume, which approximately corresponds to the volume of a single grain in a polycrystalline medium. The ratio KUV/kBT is kept at a value of 40-60 to ensure 10-15 yr data stability.[58] At the superparamagnetic limit, the scaling of the grain size necessary to maintain the adequate SNR can no longer be compensated by increasing Ku because of the limited write fields achievable with conventional write heads.[59,60] It has been predicted that the superparamagnetic limit for conventional longitudinal recording is an areal bit density of -150Gbit/in2.[61]
Fig. 8 (A) A schematic of a conventional longitudinal recording scheme employed in today’s hard drives. (B) A schematic of a single-bit transition in a polycrystalline medium. (C) A schematic of a perpendicular recording system. The soft magnetic under-layer acts as a magnetic mirror, doubling the amplitude of the write field.
A schematic of the conventional (longitudinal) recording paradigm is shown in Fig. 8A. During writing, the magnetic field generated by a write head aligns the magnetization along the track in either a positive or a negative direction. A magnetoresistive sensor detects the bit boundaries during playback. The lateral dimensions of a bit define the areal bit density that such a drive supports. As shown in Fig. 8B, the recording medium has a polycrystalline alloy structure where, to reduce noise, a bit is represented by the average magnetic moment of 50-100 grains. The SNR is typically 14-16 dB, measured with the autocorrelation SNR of a repetitive pseudorandom bit sequence. The 32-fold increase in bit density over the past 5 yr has been achieved by reducing grain size, while holding the number of grains per bit approximately constant.
A solution to extend the superparamagnetic limit was recently introduced in the form of perpendicular recording, where larger write fields enable the enhancement of the thermal stability ratio.[62,63] A schematic diagram of a typical perpendicular recording system is shown in Fig. 8C.[40] In this technology, the magnetization of the media is oriented normal to the disk surface and the use of a single-pole write head, mirrored in a soft underlayer, effectively doubles the available write field compared to longitudinal recording.
It should be noted that the transition of a highly competitive multibillion dollar industry such as the magnetic data storage industry to a different technology is not easy. Early demonstrations of ultra-high-density magnetic recording using an FIB fabricated magnetic recording head have played a key role in the process. An FIB image of a nanoscale trackwidth perpendicular magnetic thin film head fabricated of relatively wide (about 1 mm) longitudinal ring-type heads using FIB trimming from the air-bearing surface (ABS) is shown in Fig. 9A. As mentioned above, it takes approximately 1 hr to trim one head into the nanoscale dimensions. It would have taken more than several months to make similar changes to the design using conventional lithography-based methods. An MFM image of two 65nm-wide adjacent tracks recorded with an FIB-made single-pole perpendicular head with a trackwidth of 60 nm is shown in Fig. 9B.
Fig. 9 (A) Micrograph of an FIB-made perpendicular write head with a 60 nm wide write pole. (B) Magnetic force microscopy image of two adjacent 65nm-wide tracks recorded into Co/Pd multilayer medium using FIB writer.
At a 4 : 1 bit cell aspect ratio, such a narrow track corresponds to an areal density of the order of 1 Tbit/in2. With a series of follow-up experiments, this demonstration has resulted in a major shift to perpendicular recording in the entire multibillion dollar magnetic data storage industry today.[64]
Ultra-high-resolution MFM
As we step into the world of nanodimensions, the importance of high-precision microscopy is evident.[65] Scanning probe microscopy (SPM) is generally recognized as one of the most critical and unique techniques utilized for imaging with nanoscale resolution.[66] Many important discoveries of the last two decades would have been impossible without SPM. Some examples are carbon nanotube-based applications, new-generation magnetic media for data storage, detection of magnetic flux quanta in superconductors, and others.[67,68] With so many important discoveries resulting from the ability to directly see nanoscale objects, innovations for improving the resolution of SPM even further are critical.[69]
This section discusses MFM, which is used for measuring a magnetic field emanating from an object with sub-100nm resolution.[70] Today, MFM, unlike some other more popular modes of SPM, such as AFM and scanning tunneling microscopy, does not provide spatial resolution sufficient for study of the intergranular effects in different media. Most of these effects are caused by the quantum-mechanical exchange coupling with the characteristic length of interaction of the order of 1 nm. The best MFM system today provides resolution only of the order of 30-50 nm. Such a relative low resolution is attributed to the long-range magnetostatic interaction between a probe and a sample. However, it is believed that the resolution of MFM can be substantially improved with more advanced nanoscale fabrication methods with more flexible control of the probe geometry at the nanoscale (see Fig. 10). Recently, it was reported how FIB can be used to fabricate MFM probe tips that would provide not only improved resolution but also directional information about the magnetic field emanating from a sample.[71]
Magnetic nanotubes
The process and control of the magnetization switching in nanoscale magnetic probes is of critical importance for a number of emerging technologies, including magnetic random access memory (MRAM), magnetic nanoelectromechanical systems (NEMS), magnetic recording at densities beyond 1 Tbit/in2, and others. In this case, considering that the magnetic domain wall width in a typical ”soft” magnetic material is of the order of 100 nm, controllable switching of the magnetization becomes an issue with the reduction of the characteristic pole tip dimensions into the sub-100nm range.
Fig. 10 Electron image of an FIB-made nanoscale probe with a 40 nm x 40 nm x 10 nm apex.
In recognition of earlier theoretical conclusions, a rectangular probe with a magnetic void (tube-like structure of the probe tip) at the ABS in the form of a physical cavity, as shown in Fig. 11, was fabricated.[72] Magnetic nanoprobes with a rectangular cross section of 60 x 60 nm2 and a probe length (throat height) that varied in the range of 100-1000 nm were fabricated using FIB trimming of regular magnetic recording heads. Each magnetic nanotube had a 40 nm-deep cavity with a 40 nm diameter created at the ABS of the probe tip via FIB, as shown in Fig. 11. The FIB-made nanotubes were composed of a Ni/Fe (45/55) alloy with saturation magnetization of 1.6 T and with an anisotropy field of approximately 20 Oe. Also, it is clear that the shape of the nanotubes can be easily varied with FIB. These FIB-made nanotubes are perfect prototype devices to study the physics of magnetization switching in different types of MRAM and other magnetic devices.
Nanomanufacturing of Patterned Magnetic Recording Media
It should be noted that the perpendicular recording discussed above has it own superparamagnetic limit and is expected to top out of steam at 500-1000 Gbit/in2. If no alternative technology is developed to further extend the superparamagnetic limit, the high-areal density growth rate in magnetic recording technology will come to a halt in 2006-2007. Currently, there are two possible approaches to meeting the superparamagnetic challenge. One is thermally assisted recording on high-anisotropy media where local heating is used to temporarily reduce the magnetic anisotropy during writing.[73] The other is nanoscale-patterned medium recording (N-PMR), a single-grain-per-bit recording paradigm, where SNR is maintained by eliminating the randomness of the recording medium through lithographic patterning or self-assembly methods. The N-PMR would face a superparamagnetic limit for a bit size of 3-4 nm or a bit density of 20-50 Tbits/in2.[74] One of the promising approaches to N-PMR fabrication and prototyping is IBL, which is under development by the team of the University of Houston, Seagate, the National Institute of Standards and Technology, and Lawrence Berkeley National Lab. A stencil mask with an array of open apertures is used to produce trillions of parallel atom beams. These beams create an N-PMR medium pattern as the substrate is scanned across the unit cell. Because the openings of the aperture array are widely spaced, the ultimate resolution of e-beam lithography necessary for writing these masks is in the 5-8 nm range, not the 20-30 nm that would apply for patterning dense media directly.[75] The relatively low (e.g., 1%) pattern density on the stencil mask means the mask will be extremely durable and suffer minimal distortion.
Fig. 11 Micrographs of a 60 nm x 60 nm magnetic nano-tube with a 40 nm flux cavity fabricated using FIB etching of a magnetic recording head.
CONCLUSIONS
In summary, ion-beam technologies represent a versatile tool for next generations of material systems and nanoscale devices. Ion beams can be used for synthesis of various nanocomposite materials, materials properties modifications at nanoscale, and fabrication of record setting functional nanometer scale devices.
