Introduction

Nanotechnology is a broad set of technologies aimed at developing novel materials, processes, and devices. The previous article in this series built a taxonomy to segment the basic technologies. This article explores a few applicable nanotechnology markets and outlines the opportunities and technologies, some not purely nano-scale, that challenge traditional approaches.

The companies and researchers in nanotechnology target breaking fundamental limits of today's proven technologies. The markets highlighted here include computing, data storage, and medicine.

Computing

The search to define a replacement to today's silicon process is timely. Over the next 5 years the limits of silicon-based computing will be increasingly approached. These limits arise from the combination of the photo-lithographic process used to produce chips and the physical scale limits of today's silicon-electron devices. As much as the physics challenge, an opportunity for a cheaper process will force the market to move before the true limits of silicon are reached in, perhaps, 2017. Silicon has been pushed at great cost; today's new generation of fabrication facilities are estimated to cost $2 billion.

Opportunity

All of the commercially produced silicon processors and memory produced today is based on exposing a photosensitive layer coated on a silicon wafer. Bulk materials are deposited in patterns on the wafer in multiple stages. Two fundamental scaling limits of this set of technologies are the wavelength of the light used in fabrication, and the physics of electron transport in the finished silicon.

In 2003, IBM, Intel, and others will release a microchip fabrication process that operates at 90 nanometers (nm). This deep ultraviolet (UV) process will not scale down much beyond this level. The following process, based on xtreme UV? will scale features down to 65 nm. Microchips fabricated with this process should appear in early 2005.

Optical photolithography appears impractical much past this point, and other methods, such as electron-beam lithography, which uses electrons rather than light to produce a pattern in silicon, have not proven attractive. While efforts in electron-beam lithography were well funded for the last the last two decades, more than $100M in Japan alone, it does not appear this approach will replace photolithography. For example, IBM has stated it believes it cannot happen at this time, while Motorola has abandoned the approach entirely.

Fundamentally, as the scale decreases, the twin specters of quantum effects and power consumption begin to choke development of conventional silicon devices.

Quantum Effects

The flow of electrons making up electric current no longer behaves continuously, like water in a river current. Instead, each single electron causes a noticeable behavior change and inter-electron quantum interactions become significant. Strange and challenging effects result.

For example, electrons may be transported in significant quantity from the source to the drain of a transistor by tunneling through the potential barrier. Ballistic transport, also comes into play; charge may travel without scattering off of atoms in the bulk of the structure.

Below 50nm, fundamentally different physical effects begin to manifest themselves. The result? Much of the design methodology used to develop today's chips will begin to fail and chips must be designed with quantum effects in mind.

Power Consumption

today's fast desktop processors consume about 60 watts (W). Increasing the transistor count leads to ever-greater power density and increases the challenge of pulling the generated heat away from the CPU. Assuming transistor count and densities quadruple, as manufacturers project, then power density will increase to over 100W even with lower cell voltage. The power consumption of CPUs creates another barrier to current design practices.

Technologies

Companies are targeting this opportunity using approaches that involve molecular, mechanical, and quantum computing.

Molecular Computing

Molecular computing is one path to lowering the number of electrons required for logic, and generally still uses the motion of electrons for communication. This approach combines traditional technologies with both self-assembling and manufactured nanoscale organic elements, yet quantum effects and electron leakage can be accommodated.

Among the groups focused on molecular computing are HP/UCLA's teams, led by James Heath and R. Stanley Williams. Molecules called rotaxanes are used to switch current between cconventional fine conductors. They are currently exploring the use of carbon nanotubes to replace conventional fine conductors. At this time these devices are largely constructed rather than self-assembled.

The Rice/Yale team, let by Mark Reed and James Tour, is pushing ahead with a different chemistry based on ethynylphenyl. This approach uses molecular self-assembly of logic elements onto a silicon substrate. They have formed a company called Molecular Electronics to commercialize their discoveries.

Both are conservative about commercialization timeframe, and each promise to demonstrate working systems in 2003.

By replacing macro-electronic effects with quantum effects and replacing photolithographic features with molecular devices, these methods can defeat both the thermal and scale limitations of silicon. Work in molecular computing and quantum transistors is progressing surprisingly well, and early commercial devices targeted first at memory applications are expected by 2005.

Mechanical Computing

One way of avoiding the quantum limitations of electron-based computing is by replacing electrons with physical position and motion. The Babbage engine of the 1800's was the last 'great' foray into complex mechanical computation. Babbage's last design, the Analytical Engine, was never produced. It is interesting to contemplate; just as electronics proved far smaller and more easily fabricated in the 20th century, so the mechanical computer may rise again in the 21st.

Inspected more closely, this may be a late 21st century innovation since the theoretical advantages of mechanical logic are counterbalanced by the challenge of fabrication. Codesta is not aware of any company presently building a nanomechanical computer. Work by Ralph Merkle, currently at Zyvex, on helical logic represents one such approach. His work, which focuses on inorganic assembled nanoscale devices, explicitly depends on manufacturing technology that is in the early research phase and is therefore viewed as decades from fruition.

Quantum-Based Electronics

An elegant approach to the quantum limit barriers in conventional silicon technology is to embrace the quantum effects and design devices around them. Single-electron devices and quantum spin devices are under investigation in many labs.

The single-electron transistor, designed with a small number of electrons in mind, is one example of a nano-technological approach combining a new scale factor with reduced power consumption. Room-temperature single-electron transistors fabricated using conventional silicon chemistry have been demonstrated and several other approaches are being pursued. The technology is challenging to commercialize for many reasons, and estimates place the timeframe about 10 years out. Companies actively involved in this work include Toshiba, Hitachi, NEC and NTT.

Recent advances in using quantum spin effects have spawned efforts in spintronics. Spintronics relies on a feature of electronics known as spin, rather than on the transport of electrons themselves. Spin can contain state information much as charge does.

The above two quantum-based approaches use inorganic assembled nanoscale devices. Considered as a refinement of today's silicon manufacturing methods, they will not, however, favorably improve the growing cost of building fabrication facilities.

Bulk Persistent Data Storage

Bulk persistent data storage is served primarily by magnetic media today in the form of hard disk drives. Hard disks have shown tremendous cost effectiveness and performance flexibility in the last decades, but fundamental limits will cause a plateau to be reached. Solid state flash memory is another option, but primarly where smaller amounts of faster data access is required, and where price isn't as much of an issue.

Opportunity

Hard disks, which have 200 million units shipped a year, are currently manufactured with data densities of up to 6 gigabytes (GB) per square inch, or 60 GB per 3.5" platter, and this number continues to increase. IBM, for example, has been growing areal density (bits per unit of area) by 100% per year for the last several years. They recently demonstrated a data density of 12 GB per square inch.

Theory suggests the traditional approach to magnetic data storage will limit density to 60 GB per square inch. Beyond this density the magnetic domains become so small that data is lost through random thermal flipping of bits.

While 60 GB per square inch, which could be reached by 2006, offers a significant amount of storage, it may not be enough to cover demand. For example, personal digital video recorders (PVR) are gaining popularity as a replacement for the VCR. However, a library of 300 hours of high definition television (HDTV) video will consume about 3,000 GB even with MPEG-2 compression. The typical four-platter hard disk will not be able to reach this storage size using traditional magnetic media, and optical media, such as DVD, have far lower densities.

The technological vision for persistent storage is driven by cost, size, and write/read speed. Hard disks have maintained a retail price of $100-$300 for popular high-volume units over the past several years. The most popular sizes for hard disks are a 2.5" form factor for laptops and a 3.5" form factor for desktop computers and PVRs. And today's better hard drives transfer data off the hard disk at approximately 50 MB/s.

Any successive technology will have to achieve a similar or better unit price, be similar or smaller in physical size and sustain a faster transfer rate. Continuing the above example, 250 MB/s is the minimum rate needed to manage a 3,000 GB data volume. This allows for a 3.5 hour window to backup the entire data set.

Technologies

Companies approaching the storage market opportunity are using a variety of technologies including mechanical, optical, and molecular storage.

Mechanical Storage

The size of the smallest possible useful magnetic region for recording on hard disk is believed to be approximately 20 nm. Mechanical writing, also knowns as micro-electromechanical systems (MEMS), has recently been demonstrated on a 10 nm scale by IBM. This allows for densities of approximately 120GB per square inch

Writing is accomplished using a series of mechanical needles, numbering in the thousands, that punch holes in a thin plastic film. Harkening back to the days of punch cards, this approach has major challenges in read and write speed, with today's lab devices demonstrating a write speed of less than 1 MB/s. If the needles can be operated at speeds approaching one million cycles per second as researchers believe, a 4000 needle array can read or write at the aforementioned 250 MB/s.

The fundamental limits of this technology are the nature of the polymer film and the area the needles can traverse. For example, it has been discussed that densities of 12.5 terabytes (TB) per square inch are possible with other materials.

While the technology is still in the lab, and the commercialization timeframe is not known, this or a similar approach could well reach the market by 2006. Depending on the price point, this technology could appear as both a fixed and removable media solution.

Optical Storage

In order to achieve the densities necessary to replace magnetic media, holographic technologies are also being explored. Using volumes rather than surfaces to store data, nanoscale engineering is not required.

Holograms work by recording the interference patterns of multiple coherent light sources. Holographic storage achieves high density by recording these patterns through the depth of the media, effectively using 3 dimensions for storage. Densities of 40 GB/square inch have been claimed by Lucent, and they have established InPhase Technologies to develop and commercialize the technology. Tapestry, their first product, is to be released next year with an initial disc capacity of 100 GB, with 1 TB targeted for successive products. Optware, a Japanese company, also has plans to introduce a 1 TB product based on this technology.

An alternate approach involves using a bioactive molecule, typically bacteriorhodopsin, embedded in a transmitting matrix as a recording medium. While an intriguing, it does not appear to have real advantages over the holographic methods which can use solid molded media.

Molecular Storage

The densities achievable by both holographic and micromechanical storage can be eclipsed by molecular storage. Molecular storage often relies on a bistable molecule, representing a bit of data, with a significant energy barrier separating the two states. An electrical charge can be used to flip the molecule from one state to the other.

While there there is significant overlap within molecular computing, the prime differentiator is the data integrity time in the unpowered state. The following approaches combine self-assembled organics to form the memory-retaining layer and manufactured inorganic processes to build the electronic interfaces.

Rolltronics is in the process of creating a molecular storage device that targets the flash memory market. Their device is based on porphyrin molecules assembled using a rolling process between electrode layers and uses a combination of light and charge to write to the memory location. The company's first product concept, a 64 GB PC Card, is projected for 2004 release.

Another company, Zettacore, is also using porphyrins as a base, though with a different manufacturing method. They plan on a product similar to Rolltronics, but will focus on straightforward compatibility with silicon-based circuits.

Molecular Electronics, referred to previously, has disclosed a more volatile memory system that has a power-off retention time of several seconds. Presumably their methods could be used to produce working nonvolatile memory.

The densities these methods target is roughly 100-1000 times that which is currently achieved in silicon, or roughly a 10 nm linear scale per bit. If molecular memory can demonstrate a cost and density advantage over mechanical memory in the next several years it's naturally greater headroom should gain favor in the market.

Medicine

The human body functions via a collection of chemical, mechanical, and electrochemical processes. All medical treatments are based on modifying elements of these processes, but are unable to do so in a coordinated fashion at the molecular level. Some of the greatest medical challenges relate to the targeting of modifications, and to access affected tissue.

Nanotechnology offers so much promise in improving targeting and access that some visionary researchers believe that most of the symptoms of disease and aging may be eliminated and near-immortality achieved. Rather than debate the ultimate result of nanotechnology on the human condition, this work examines some of the larger current efforts under way in the medical space.

Opportunity

The pharmaceutical space alone is estimated at slightly over $400 billion in annual sales. With an R&D investment of about $45 billion per year, and an average cost per new drug of $500 million, pharmaceutical companies have tremendous resources to support innovating and delivering new treatments.

In the process of bringing a drug to market, candidate drugs are tested for efficacy (ability to control or cure an illness), absorption and delivery (ability to reach a drug target), toxicty (harmful effects), and for their metabolism in and elimination from the body. Most drugs are delivered orally or intravenously and then make their way to a drug target through the bloodstream. The ability to deliver a drug to a targeted tissue in sufficient concentration is a major challenge, particularly if the drug has toxic effects on other tissues.

Nanotechnology has potential impact upon drug delivery and targeting, with the benefit of significantly reducing potential toxicity issues. Nanotechnology also has long-term promise for targeting disease by physical action at the sub-micron scale. This could be considered true micro-surgery?

In the near-term, the safety-mandated conservatism of the medical industry will focus on relatively more conventional strategies. In truth, biotech already operates at the molecular scale with the current focus on proteomics (the study of the set of protein structures and functions in living organisms) and genomics (the study of the set of genes within chromosomes of organisms.)

Technologies

Nanoencapsulation

Drug delivery and absorption, estimated to be a $30 billion market in 2002, represent large elements in the success of a pharmacological treatment. Cancer drugs are well known for producing serious side-effects as non-targeted tissues are impacted. Other challenges relate to neurological products which need to cross the blood-brain barrier to have a desired effect.

The crux of the issue is that a drug's activity is exposed to every tissue it encounters as the body circulates and metabolizes it, even though often only one tissue is targeted. By encapsulating a drug's active component in a relatively inert 'nanocapsule', which binds and opens in response to a target tissue site, many toxicity issues could be sidestepped. Research in this area is occurring at a number of companies and universities, as you can see by the following examples.

The Wooley group at University of Washington is developing a broad set of nanocapsules, as is the Atwood group at the University of Missouri. Both groups employ self-assembling organic processes.

The company Advectus, together with the University of North Carolina at Chapel Hill, is developing a self-assembling nanoscale polymer to carry anticancer drugs across the blood-brain barrier. This treatment is a candidate for fast-track investigation by the FDA.

FeRX is applying drugs into magnetically coupled inorganic manufactured nano-particles which can be drawn to a drug target via an externally applied magnet.

Microfluidics

Microfluidics research deals in the handling of extremely small volumes of liquid, typically using MEMS (Micro-ElectroMechanical Systems) based technology developed in silicon. Put together with in-device chemical sensors these devices can address dosage fluctuation issues by varying drug levels in the body in real time.

These devices represent a very attractive alternative to the spikes in drug concentration often seen from periodic dosing. The spikes represent a high risk for drugs with slim margins between efficacy and toxicity. Particularly promising, though not necessarily strictly nanotech in scale, are devices for handling drugs targeting cancer, heart disease, and diabetes.

University labs conducting research include UC Irvine Center for Biomedical Engineering, the MD Cancer Center of Lawrence Livermore Labs, and the University of Virginia. Companies involved include iMedd, Microchips, Debiotech, and Issys. Products are expected to emerge by 2004 and are already in the approval process.

Nanostructure Materials

This is a broad space, and the focus in this article is on current and near-current technologies. Much of the progress in the treatment of burns and bone replacement over the last decade has been at least in part due to the development of novel materials which act as templates for tissue replacement.

Integra artificial skin, a major advance over animal or cadaver skin, is a bilayer of purified cartilage (a polymer) that sits below a silicon protective layer. The polymer acts as template that speeds the regrowth of new tissue. This treatment is unassailably significant. In 2001, the first year the artificial skin was approved for use, the rate of death from serious burns dropped from 100% to under 40% in patients with burns covering more than 70% of their bodies. This product combines a self-assembled (actually cow-assembled) organic structure with a manufactured inorganic layer.

Encouraging the growth of bone tissue is another area where complex structural templates have benefit. A polymer template produced using a molding process acts as a scaffold for the growth of new bone tissue. Still in the experimental stages, this work could drastically reduce the number of bone grafts currently performed. An Australian team is trying a different approach using a moldable template that includes a growth-encouraging sugar. This would speed healing from large-bone breaks. These methods use inorganic and organic manufactured compounds respectively, with nanoscale features and properties.

This family of technologies promises to lead the way to the growth of more complex structured organs such as hearts and kidneys. These are distant goals at present, but more sophisticated artificial scaffolds and processes are under development.