Computing Reaches for Terahertz Speeds
Alexander E. Braun, Senior Editor -- Semiconductor International, 5/5/2008 7:57:00 AM
 |
The work is headed by Ajay Nahata, associate professor of electrical and computer engineering, who has researched this area for almost two decades. Fiberoptic phone and data lines use near-IR and some visible light. The FIR region, usually defined as 0.1-10 THz, lies in the spectrum between mid-IR and microwaves, and is as yet unused for communications or any other practical purpose.
Over the past 15 or so years, Nahata’s research has been primarily aimed at developing sources and detectors, work that requires a hybrid optoelectronic approach. In this gray "Twilight Zone" area between electronics and optics, the issue has been how to create sources that are as coherent as a laser, because one is dealing with something that can propagate through free space.
Traditionally, the way to generate FIR energy has been through the use of a globar, a silicon carbide rod electrically heated to temperatures of 1800-3000°F. The resulting radiation is then detected through the use of a liquid-helium-cooled bolometer or something equivalent, which the University of Utah researchers consider unsatisfactory. “We’ve worked on developing optical approaches to generating terahertz radiation that exhibit laser-like properties, and in the production of room-temperature detection techniques that are relatively simple to use,” Nahata said, comparing the work to the building of the first laser: “Many asked, what can you do with this thing?”
In the terahertz regime, there are many things that can be done. The path of electronics is clearly headed toward increasingly higher frequencies. In the semiconductor area, the move is toward ever-faster computing speeds. However, this is getting increasingly difficult to do, and some limitations are already in sight. For the not-so-distant future, the issue is how to make circuits that will operate at terahertz frequencies. “We work at 300 GHz and 0.3 THz, but how do you get to 1, 2, 5, 10 THz?” Nahata asked.
So far, the focus of the Utah group has been on investigating new materials and structures appropriate for the terahertz regime. At present, the concentration has been on the development of waveguides, passive devices that are the equivalent of wires that will allow the efficient transportation of terahertz radiation. Besides this, the researchers are looking at the development of more complicated devices like a wide spreader and a 3-D decoupler. Last year, they reported on metal films perforated with holes, which acted as filters for terahertz radiation, demonstrating considerable design flexibility. The overall goal, however, continues to focus on new types of materials for terahertz applications and how to develop devices on them.
New materials become necessary at these frequencies because most traditional dielectrics tend to be lossy, and there is considerable absorption at the terahertz level. Silicon, the mainstay of the semiconductor industry, tends to be absorptive, as are some polymers and inorganic materials. Metals, however, when correctly set up, exhibit little loss at terahertz frequencies.
“We have made terahertz waveguides,” Nahata said, “and put various materials in between. This works reasonably well, but there is still loss. We plan to use polymers for the next step, where we will use a special class of polymers that have non-linear optical properties; this could enable us to produce active devices. Polymers such as Teflon and high-density polyethylene tend to have low loss.” Nahata added that the waveguides they have produced are passive devices that just transport radiation from one point to another. The group wants to make an active device, making it electrically possible (or in some other manner) to switch the radiation. If such an active device can be made and combined with passive devices, it would mark the beginning for all of the building block components needed to produce real circuits and get the desired functionality. However, that will take work.
 |
| This waveguide, made up of stainless steel foil ~4 in. long, 1 in. wide and 625 µm thick, with 500 × 50 µm rectangular perforations, “couples” terahertz radiation, moving it from one wire-like waveguide to another. Terahertz radiation within the semicircular etching is focused and sent down the lower waveguide (the lower X part). Where the waveguides are closest, half the radiation jumps from one to another with each half being emitted from the right-side end of each waveguide. (Source: University of Utah) |
Nahata expects to heavily borrow from semiconductor technology in terms of fabrication techniques and even some materials in the production of an active device. Although he is considering materials that are not ordinarily used in semiconductor technology, he is looking at high-resistivity silicon. “It can be used in the waveguides and then apply some sort of optical source to create electron whole pairs; this could actually be used as a sort of switch. If you create many electron whole pairs, the high resistivity silicon is absorptive or lossy. When the optically generated electron pairs are absent, the loss is relatively low.” According to Nahata, there are many materials like this, whether high-resistivity silicon, high-resistivity GaAs, or a variety of other readily available traditional semiconductor materials. The researchers are not beyond improvization, however — trash bags work well as a means of blocking optical light but letting the terahertz radiation to be transmitted with reasonably high efficiency.
Based on the current work, Nahata admits that it is difficult to theorize what a terahertz computer would look like or what its capabilities would be. As he puts it, “Originally, the laser was used for gas-phase spectroscopy. Nobody envisioned it being used for optical communications or micromachining applications.” As he sees it, the importance of the work is that it will give engineers everywhere new ideas.
The work being carried out at the University of Utah may find some applications, perhaps in five years, in augmenting some of the communications pathways. Making anything — much less a computer — that operates at terahertz frequencies is probably at least 10 or more years down the road.
