- Factsheets
- Briefings
- General Sector Reports
- Economic Data
- Societal Issues
- Regulation & Standards
- HSE & Risk
- Business Tools
- Communication
reportBeyond CMOS
8.8.4 State of R&D
1. New logic and information processing devices
Extensions to CMOS: Low dimensional structures
A lot of research is currently devoted to extending traditional CMOS devices. One primary approach is to replace the FET channel with novel high carrier mobility materials. Some of these materials display a semiconducting band structure only under quantum confinement. The three main types of quantum-confined structures are carbon nanotubes (CNT), nanowires (NW), and graphene nanoribbons.
Recent research activities in nanowires and carbon nanotubes can be divided into three main categories: (i.) experimental growth and assembly, (ii.) CNT and NW device fabrication and characterization, (iii.) CNT and NW circuits and integration.
One key problem lies in the difficulty of separating different types of carbon nanotubes (semiconducting and metallic tubes) that are created together during material synthesis. Current research is investigating intensively various techniques for gaining better control over the chirality of nanotube materials (chirality is the property that determines wheter the nanotube is semiconducting or metallic).
Another key issue is to develop methods for controlling the assembly of nanostructures. Manipulating large numbers of nanotubes into position is very slow and no high-volume manufacturing processes exist today. New techniques for assembling parallel arrays of nanotubes or nanowires on substrates are currently investigated. Better control of the accuracy of these methods is necessary and requires further innovation in the assembly and fabrication technology.
Two-dimensional graphene films have generated a lot of interest recently as an interesting alternative for channel replacement material in FET structures. Graphene films are well known to behave as high mobility zero bandgap semiconductors with high carrier mobilities. When patterned to sufficiently small ribbon widths, the graphene ribbons begin to display a finite band gap resulting from quantum confinement.
New state variables
For a large part, future information processing will be done on information where the state variable is something other than electronic charge. This information includes optical images, image sequences, speech, and data sets derived from physical sensors.
Many different information carriers are currently researched. These include spin, molecular state, photons, phonons, nanostructures, mechanical state, resistance, quantum state (including phase) and magnetic flux.
Some of the novel devices may prove useful for various information processing tasks in addition to general purpose computing. Some of the more specialized tasks include associative processing, communication and multivalued logic. In general, these may require new a functional organization of the interconnected devices (new architecture).
Single-electron transistors (SETs)
SETs are switching devices that use a tunneling mechanism to transport electrons from source to drain. The latest standard MOSFETs that use thousands of electrons at any given moment in its on state, and the most sophisticated flash memory devices use roughly the same amount of electrons to store a bit of information. SETs are designed to do the same using a single electron.
Single electron devices hold the promise of ultra-low power electronics and further miniaturization. Potentially SETs can be applied to general purpose Boolean logic but significant progress on circuit and architecture development is required.
New applications and architectures that exploit the unique functionality of room-temperature operating SET circuits have been developed. Large threshold voltage variation impedes the realization of large-scale SET circuits. This currently makes it difficult for SETs to compete directly with CMOS devices used to implement Boolean logic operations.
The majority of the SET circuits demonstrated to date employ so called "voltage state logic" where a bit is represented by the voltage of capacitor charged by many electrons. Truly single-electron approaches, representing a bit by a single electron ("bit state logic") have been limited to laboratory demonstrations.
For information processing purposes, the non-linear current-voltage characteristics of SETs can be utilized effectively as the computing primitive in certain algorithms and applied to associative recognition systems that mimic the human cognitive function. Primitive associative processing (color identification) has been experimentally demonstrated using floating-gate SETs operating at room temperature.
Molecular devices
Molecular electronics is targeted at creating functional blocks at the molecular or supra-molecular level that could be assembled in more complex functions. Fully molecular-based complex systems including interconnected molecular logic and molecular memory devices have still to be demonstrated. Limited molecular logic, memory and interconnect functions have been shown, based on different types of molecules, but their integration into a single chip is still an issue.
The potential of molecular devices for computing applications is based on very high densities that can be attained, a large variety of molecular characteristics, ability to self-assemble, very low power consumption, and the ability to change state via electrical, optical, or chemical means.
A lot of progress has been made in improving the desirable characteristics of the molecules themselves and developing concepts for potential architectures that would utilize molecular elements. Significant problems exist in molecular synthesis, device and circuit fabrication and reliability. In addition, there is often a large gap between projected parameters and the actual observed ones.
The use of molecules as programmable diodes (molecular switches) is the core technology underlying most of the concepts for future applications. A molecular switch is a molecule which switches reversibly between two or more positions. Reproducibility and repeatability of experimental measurements still shows significant variations between different approaches and research groups.
A lot of research effort has been devoted to developing the concept of molecule on CMOS architecture (CMOL). These are hybrid CMOS / nanoelectronic systems that are based on conventional CMOS devices connected to nanowire arrays with molecular elements functioning as programmable diodes. These are predicted to have very attractive performance potential compared to scaled CMOS systems, but no successful demonstrations have been realised so far.
Perhaps the first potential application is using the bistable behaviour of certain molecules to produce memories with an extremely high density. Specific issues, such as contacting the molecule, carrying enough current to provide noise immunity and a reasonable fan-out, and the addressing and read out of specific blocks remain to be solved.
In addition to logic operations, molecular devices could potentially be used for several other applications. For example, molecular schemes for combinatorial logic have been envisaged. Computing has been demonstrated using DNA molecules utilizing self-assembly to perform computational steps in test-tubes.
The key challenges for molecular devices include the ability to electrically stimulate and measure response or the state. In some systems, protons have been used to communicate signals. Optical signal communication is also being investigated. Tunneling transport between molecular wires and devices is actively being researched and may be a viable option.
The subject of molecular and DNA switches is also discussed in the report on "Novel Biomaterials".
Spin devices
Spintronics (spin transport electronics, also known as magnetoelectronics) is an emerging technology which exploits the intrinsic spin of electrons and its associated magnetic moment, in addition to its fundamental electronic charge.
Spintronics has many potential advantages, including low power operation, non-volatility and co-localisation of data processing and storage.
Metal-based spintronics: The simplest method of generating a spin-polarised current in a metal is to pass the current through a ferromagnetic material. The most common application of this effect is a giant magnetoresistance (GMR) device. Metal-based spintronics is likely to be first introduced for data storage applications using either spin torque switching (spin torque transfer magnetic RAM, STT-MRAM) or domain wall effects (e.g. IBM Racetrack Memory).
Semiconductor-based spintronics: Spintronics using ferromagnetic semiconductor nanostructures holds promises for novel nanodevices sensitive to magnetic field. This could find application from information processing to sensors, though major breakthroughs are needed in materials (e.g. semiconductors with a higher critical temperature), devices (e.g. injection/detection trade-off), cointegration with CMOS or in exploring promising physical phenomena.
Spintronics using half-metals and molecules also need to be explored.
Ferromagnetic logic devices
Ferromagnetic logic devices are a class of alternative logic devices that use the local magnetization orientation of a domain of ferromagnetic material to store the computational state. Their operation is based on collective magnetic effects associated with the magnetic polarity of a nanodomain.
Ferromagnetic devices have the potential of being non-volatile and radiation hard, which is derived from the properties of the ferromagnetic materials themselves. They can be fabricated with ferromagnetic, metallic wires patterned to form Boolean logic devices.
The propagation of domain wall boundaries separating magnetic nanodomains has been shown to reach a velocity of several hundred meters per second. This has led to geometric realization of NOT gates, AND gates, fanout structures, cross-over structures, and shift registers using the domain wall movement driven by the external magnetic field.
Nanophotonics
A promising alternative for information transfer is to use light in the visible or infrared range. Nanophotonics allows the confinement and interaction of photons and electrons in a small volume which opens up the possibility of processing data at high frequency.
This subject is discussed in detail in ICT subsector report "Photonics".
2. New functionalities
Nanoelectromechanical systems (NEMS)
NEMS devices hold the promise to improve abilities to measure small displacements and forces at a molecular scale. The ITRS 2007 edition contains NEMS memory as a new entry for the Emerging Research Devices section.
Nanosensors
Nanosensors are any type of sensors (biological, chemical or other) that convey molecular-level information to the macroscopic world. Future multifunctional systems are envisioned to integrate nanoscale computing devices with sensing capabilities for use in various application areas ranging from communications to medical uses.
The potential of nanosensors as part of future computing systems lies in their capability to provide the link between other forms of nanotechnology and the macroscopic world. This would allow full exploitation of the potential of miniaturisation of computer chips while vastly expanding their storage potential.
However, before widespread implementation to consumer products becomes feasible, developers must overcome several major issues related to reliability, compatibility with CMOS technology, difficulties in mass manufacturing and high costs of production.
The topic of nanosensors and biosensors is also discussed in ICT sector report "Integrated circuits".
Organic / Plastic electronics
Organic electronics, plastic electronics or polymer electronics is an emerging branch of electronics that deals with conductive polymers, plastics, or small molecules. The field not only includes organic semiconductors, but also dielectrics, conductors and light emitters.
Conductive polymers are lighter, more flexible and less expensive than inorganic conductors. Potential application areas include consumer electronics, displays, packaging and photovoltaics.
Organic electronics is often associated with printed electronics. The processability of organic electronically functional materials in liquid form allows their use as functional inks in printing. Electronic thin-film devices are prepared by printing several functional layers on top of each other.
Printed electronics is expected to facilitate the establishment of "low-cost electronics" for application fields where high performance is not necessary. Examples of applications include flexible displays, light-emitting diodes, smart labels (RFID) and active clothing.
Electronic applications with high switching frequencies and high integration density ("high-end electronics") will be dominated for a foreseeable future by conventional electronics.
The development of displays is discussed in detail in ICT sub-sector report "Displays". Sector reports on Energy and Textiles include information on photovoltaics and active clothing.
3. Architecture and system level
Emerging architectures
Architecture refers to the functional arrangement of interconnected devices that includes embedded computational components. New computational schemes, such as neural networks or DNA computing, are being investigated as future alternatives to deterministic computing
Reliability will be one key issue for future nanoelectronics devices. The number of fabrication flaws is unavoidably increases when devices shrink towards the nanometer scale. Emerging devices are expected to be more defective, less reliable and less controlled in both their position and physical properties. It will be necessary to develop methods for error resiliency and trade off error rate against performance (e.g. speed, power consumption).
It is also important to pay attention to the "systemability" of emerging devices, i.e. the capacity of a device to be integrated into a complex system. Emerging technologies can eventually lead to paradigm shifts in the whole concept of computing and will therefore require completely different design flows in order to exploit them in systems. It will be essential to involve multidisciplinary teams of system architects and nanotechnology researchers in order to optimise the overall performance of a system.This research will link system-level objectives such as high performance and reliability to the development of advanced nanoscale devices.
Thermal management
The emerging field of phononics aims to control phonon movement by using engineered nanostructures. It brings new opportunities in the interaction between quasi-particles (e.g. electrons, photons, spins) and phonons, potentially allowing better heat removal, isolation from thermal noise and better carrier mobility.
It is much harder to control the flow of heat in a solid than it is to control the flow of electrons. However, researchers have recently demonstrated thermal diodes, thermal transistors and thermal logic gates. Such components also raise the possibility that heat could also be used to process information.
4. Manufacturing and design issues
Bottom-up techniques
The ability to manufacture billions of devices on chip with full control over their properties presents an overwhelming challenge that is threatening to lead to unbearable production costs.
New potential manufacturing techniques include various "bottom-up" approaches. These processes are attractive because they are fast and versatile, but they are generally incompatible with conventional fabrication.
Self-assembly and bio-inspired techniques are viewed as attractive concepts that need further investigation. Directed self-assembly is essential to fabricate complex structures composed of nanoscale building blocks. Bio-inspired processes may be used to develop self-adapting and self-repair properties.
Some examples of technologies that utilize bottom-up approaches exist today, such as self-aligned "optical fringing" nanotubes and self-assembled quantum dots. However, larger scale advances will require long term development. Such huge research effort has been devoted to the development and optimization of the CMOS platform that it is probable that novel bottom-up technologies are first combined with traditional approaches to achieve increased performance and cost effectiveness.
Full-scale utilization of bottom-up manufacturing would require that circuits be designed entirely differently. Research into new architectures may help on this path by relaxing the need for a deterministic approach to controlling the properties of the elementary devices.
Multiscale modelling and design tools
Next generation technologies will present major challenges to the design environment. It will be necessary to build suitable simulation methods and design tools that link nanoscale device models with higher-level heterogeneous system design environments.
Current research is putting a lot of emphasis on the physical properties and fabrication of new device structures. Much less effort is spent on how to design complex systems using these novel devices. This so-called "design gap" continues to grow. Therefore more research is needed to improve the ability to integrate new technologies into functional systems.
Numerical modeling will be an essential tool for integration of multiscale functionalities to future products. The key challenge is bridging the enormous gap in time and length scales when linking the nanoscale elements to macroscale systems.
Computer modelling of nanostructured materials needs to integrate a range of approaches from first-principles quantum mechanics, to forcefield-based molecular mechanics and mesoscale simulation methods. Further research is needed develop simulation methods that would allow modelling of complex nanoscale systems.
Increased system complexity represents another huge challenge: it is necessary to include elements of mechanics, hydraulics, chemistry, magnetics and (bio)sensors, together with the trend towards multiprocessor, defect-tolerant and power-managed implementation architectures. No applicable models, methodologies, or tools exist today.
Document details:
Visits: 1912, Published on: April, 29th 2009, 12:08 PM, Last edit: June, 1st 2009, 03:07 PM Size: 18 KByte