Light Logic for 'Light'-ning Fast Computers
For some time now, the idea of building light-based devices to supplement semiconductor-based computing has attracted the interest of researchers and computer engineers alike. Why? Because, as eloquently put in a 2007 issue of Scientific American, “Light is a wonderful medium for carrying information.” While the unit particle of light waves, or photons, would not necessarily move much faster than the electrons that already speedily course through our computer microchips, light waves possess the power of larger bandwidths, and thus the ability to carry more data at one time, leading to higher throughputs for optical interconnects as compared with metal wire interconnects. Pushing even beyond the goal of merging optical interconnects with electronic logic gate components, some researchers are successfully attempting the unthinkable: creating nanoscale logic gates with unique light-activated nanowires, or the equivalent of optical transistors.
“But one vital building block is missing from Intel’s kit: an optical equivalent of the transistors that perform the logical operations at the heart of a computer.” – The Economist, 2010
Today’s computer microchips combine tiny electronic transistors (on the scale of 50nanometers, or 5×10-9 meters) with interconnecting copper wires, all wrapped up within a microchip-scale integrated electronic circuit. The role of this electronic circuit is to control the transport and storage of electrons (Science Review 2006), and the result is the magic of the laptop or desktop computer (or iPhone, or iPad!) which you are probably using at this very moment to read ‘Light’-ning Fast Computers. You might ask why, when you have the power of the internet at your fingertips in your lovely new iPad, researchers are still striving to create faster and smaller microprocessors by merging light-based components with electronic-based components? The answer is: we are always striving for faster and more efficient, especially when it comes to microprocessors and computer-based devices, in order to solve our most complex problems. While you and I may have the internet at our fingerprints (although my boyfriend probably wishes we could purchase even greater internet bandwidths for purposes of his online gaming), complex computational problems such as whole genome sequencing still require hours or days using our fastest and most powerful computers to date. While transistors continue to decrease in size and increase in numbers on our modern microchips, copper wiring interconnects are lagging in their ability to ‘keep up’ in terms of transporting electrons from one point to another on microprocessors. One way to fix this problem is to integrate optical interconnects or even actual optical logic components into semiconductor devices.
However, a major limitation to the use of optical interconnects or optical logic components is the diffraction limit of light (Nano Letters 2011). Diffraction describes the spreading out of light waves as they pass through a small aperture. If you watch the sunlight shining through the blinds covering your living room window, you might observe that a small spec of light coming through a hole in the binds becomes a much larger spot by the time it reaches the floor beneath the window or the wall on the opposite side of the room. Similarly, place a piece of paper with a pinprick hole over a flashlight, and shine your flashlight toward a wall in a dark room: you will observe the effects of diffraction, which worsen as you back further away from the wall (as the spot of light becomes larger and larger). Light has funny ways of spreading out when it passes through small holes, of bending when it travels from one medium to another (ex. from air to water), and of interfering with other light waves. Light diffraction is the fundamental obstacle that stands in the way of shrinking optical computing components down to the tiny sizes of today’s electronic devices in integrated circuits. The electronic components of computer chips can exist on the scale of tens of nanometers, length scales 50-100x smaller than the width of a light wave (i.e. the wavelength of light). The problem becomes, how can we squeeze light-based computing components down to scales much smaller than the wavelength of the light itself? A goal near impossible in traditional photonics, the ‘investigation of the emission, transmission, amplification, detection, and modulation of light’ (Wiki). Even with attempted miniaturization of traditional light waveguides (like optical fibers), these conduits for the transfer of light waves are still far too large to integrate with electronic components on tiny microchips.
An answer to the inherent limitation of diffraction with traditional photonics is the mind-boggling world of plasmonics, or surface plasmon-based photonics. Surface plasmons are light-induced, collective oscillations of the free electrons at the surface of a metal (conducting) surface (Science 2008). These surface plasmons occur after we excite a nanoscale metal structure, such as a silver nanowire, with light from, say, a laser source. The electron oscillations are confined to the surface of a thin metal strip or a metal nanowire (between the conducting metal surface and a surrounding dielectric or electrical insulator medium), and can travel down the strip or the nanowire in what is called a surface plasmon polarition. The surface plasmon polarition travels along the metal nanostructure very similarly to how traditional light waves travel down an optical fiber, rippling like the surface of a pool of water after you through a lucky pebble in. The lucky catch here is that unlike light waves inside an optical fiber, surface plasmon polaritions can travel within (or rather along the surface) of a nanowire MUCH smaller than the wavelength of the laser light used to excite the polarition! In other words, we can achieve spatial confinement of the original light into size ranges that more closely match tiny semiconductor electronic components (at the tens of nanometers range). Light at its free-space wavelength, of around 1000 nanometers, can be channeled into one end of a gold or silver nanowire with a diameter of only 1/4th or less of that original wavelength. The light then travels in a squeezed down form as a surface plasmon polarition along the tiny wire, and reappears on the opposite end of the wire back in its ‘full-form’, ready to interact with other traditional photonics components. Plasmonics thus offers the “prospect of combining the compactness of an electronic circuit with the bandwidth of a photonic network.” (Nat Photonics 2010)
“… plasmonics is expected to be the key nanotechnology that will combine electronic and photonic components on the same chip.” (Science 2006)
This is some amazing technology, but the mind boggling isn’t over yet. Hongxing Xu and colleagues researching out of China, Sweden, and Texas (USA), are not only forming networks out of silver nanowires, along which light can propagate in the form of sub-wavelength surface plasmons (Nano Letters 2011), but they are also combining these networks into optical logic gates, the equivalent of a series of electronic transistors. By tuning the properties of light ‘input’ lasers, the surface plasmon polaritions traveling down various branches of a silver nanowire network can be made to controllably interact and interfere with one another such that branch outputs can reliably be switched from OFF states to ON states, and vice versa. By creating branched silver nanowire structures, Hongxing Xu and coworkers demonstrated successful creation of a complete family of basic logic functions: AND, OR, and NOT gates. To further top their groundbreaking research into key components required for future optical computing devices, Hongxing Xu and his laboratory successfully formed a more complex cascaded optical logic gate by combining a series of their silver nanowire optical switches, as published just last month in Nature Communications. Xu et al specifically constructed the universal NOR gate, which results from combining OR and NOT Boolean logic operations (see figure below). The creative researchers were able to follow the path of electromagnetic energy through their silver nanowire networks by using tiny colorful fluorescent beads, or quantum dots, that when coated around the nanowire network will fluoresce or emit light of a characteristic color when excited by a passing surface plasmon wave. In this way, the researchers could prove that light was traveling through their nanoscale plasmonic networks in accordance to the rules of the Boolean logic gates that they had designed.
In summary, as modern day semiconductor devices approach their fundamental speed and bandwidth limitations (Nat Photonics 2010), and traditional photonics can only provide increased performance solutions at the micrometer scale due to the limit of diffraction, the field of plasmonics appears to hold the key to future improvements in computing. Engineered noble metal nanostructures, which “exhibit an unparalleled ability to concentrate light” (Science, 2010), fit the size criteria for nanoelectronics and fit the bill for desired improvements in speed and data-carrying capacity. Plasmonics certainly seems to have much to offer to the young field of optical computing. However, problem-solving is far from over for next-generation information technology: even innovative plasmonic nanowire networks suffer from worrisome limitations. With miniaturization of metal strips down to tiny nanowires, issues crop up including strong dissipation of light signals over relatively short network distances. Macroscale waveguides such as optical fibers are actually able to carry light signals over much longer distances without losing much of the original light signal. Also, smooth silver nanowires are very sensitive to structural imperfections: tiny bumps along the nanowire can disrupt the surface plasmon polarition and cause untimely emission of the formerly trapped light traveling along the nanowire surface. Despite limitations, on-chip integration of optical interconnects and sub-wavelength plasmonic logic units may pave the way to the future of optical computing and ‘Light’-ning fast computers.
References:
(1) Albert Polman. Plasmonics Applied. Science 322, 868-869 (2008)
(2) Brongersma M., Shalaev V. The Case for Plasmonics. Science 328, 440-441 (2010)
(3) D. K. Gramotnev, S. Bozhevolnyi. Plasmonics beyond the diffraction limit. Nature Photonics 4, 83-91 (2010)
(4) Ekmel Ozbay. Plasmonics: Merging Photonics and Electronics at Nanoscale Dimensions. Science 311, 189-193 (2006)
(5) Harry A. Atwater. The Promise of Plasmonics. Scientific American 56-61 (April 2007)
(6) James Martin. Photos: Intel Photonics Labs’ quest for blisteringly fast data rates. Silicon.com
(7) Optical computing’s bright future: Light without logic. The Economist, May 13th (2010)
(8) Wei et al. Cascaded logic gates in nanophotonic plasmon networks. Nature Communications 2:387 (2011)
(9) Wei et al. Quantum Dot-Based Local Field Imaging Reveals Plasmon-Based Interferometric Logic in Silver Nanowire Networks. Nano Letters 11, 471-475 (2011)
Wei H, Wang Z, Tian X, Käll M, & Xu H (2011). Cascaded logic gates in nanophotonic plasmon networks. Nature communications, 2 PMID: 21750541
Wei H, Li Z, Tian X, Wang Z, Cong F, Liu N, Zhang S, Nordlander P, Halas NJ, & Xu H (2011). Quantum dot-based local field imaging reveals plasmon-based interferometric logic in silver nanowire networks. Nano letters, 11 (2), 471-5 PMID: 21182282