Until recently, a truly wireless existence was beyond what silicon circuits could offer. The bands of the radio spectrum, such as Wi-Fi, that they could reach were too narrow to connect a high-definition TV to a high-definition DVD player. The chips that could do the job, made with exotic semiconductors, were too expensive for consumer electronics. But in the last two years, silicon circuits finally broke into the 60-gigahertz band, which has been shown to allow data-transfer rates of 5 gigabits per second over a distance of 5 meters.
Sixty-GHz radios, based on silicon or silicon-germanium chips, are expected to be integrated into TVs, set-top boxes, and other media-linked devices starting in 2009. But a new dark-horse candidate has emerged that claims to be able to make cheap 60-GHz technology without using any semiconductor materials at all--silicon or otherwise.
Boulder, Colo.based Phiar Corp. (pronounced ”fire”) uses a proprietary mix of insulators and metals to achieve quantum tunneling, which lets electrons zip through devices in mere trillionths of a second. ”We're at a tipping point,” says Adam Rentschler, Phiar's director of business development. ”Metal insulators are the first viable alternatives to semiconductors since the era of vacuum tubes.”
Normally you can imagine an electron as a ball and an insulator as a high hill. Given enough energy, the ball will make it over the hill; this is how an electron punches through insulation. But when the hill--the insulator--is only a couple of atomic layers thick, the rules of classical physics no longer apply: instead of a ball, the electron looks more like a wave. (The wave is actually the function that defines the probability of finding that electron in a specific place.) This wave is wider than the very narrow hill, so it stretches from one side of the insulator to the other. As a result, sometimes the electron simply appears on the other side, having ”tunneled” through the insulator. Tunneling happens all too often in the transistors of modern microprocessors, and is a serious problem [see ”The High-k Solution,” elsewhere in this issue]. But Phiar took advantage of the phenomenon by finding a way to make electrons tunnel easily in one direction and not at all in the other. The result is a diode made up of two very thin insulators sandwiched between two metal layers.
From the electron's perspective, Phiar's proprietary blend of metals and insulators makes the energy barrier between the metals thinner in one direction than the other. That happens because when the electron travels in the ”easier” direction, the interface between the two insulators creates a quantum well, a structure that confines electrons in two dimensions. The quantum well sits at the halfway point between the two metals, and once the electron has tunneled that far, the quantum well boosts the chances that it will make it through the second insulator. But the quantum well appears only when the electron is being pushed in one direction, called the forward bias [see illustration]. When the electron is being pushed in the opposite direction, reverse bias, there is no well, and without it, the insulation is too thick to tunnel through.
The metal-insulator-insulator-metal (MIIM) structure makes electrons ”10 billion times more likely to tunnel,” says Rentschler. Quantum tunneling lets an electron traverse a device in just 1 femtosecond, thousands of times as fast as an electron traveling through a typical semiconductor transistor. ”Semiconductors are called semiconductors because an electron doesn't move through them very well,” he says. ”It spends a lot of time bumping around through a slow, molasses-like atmosphere.” Rentschler says the devices are so fast that the 60-GHz band is the lowest frequency the company is interested in and that its circuits have been clocked at 110 GHz.
Phiar contends that its devices can be manufactured directly on top of a CMOS chip, potentially making them a simple addition to an already inexpensive technology. In fact, Rentschler maintains that Phiar devices can be fabricated on almost any substrate. ”Tree bark is probably a really bad choice,” he says, ”but we can pretty much deal with anything.”
Tree bark aside, the company is hoping to build its circuits on a variety of substrates in an effort to get around one of the peculiarities of the 60-GHz band. Radiation at that frequency is absorbed by oxygen in the atmosphere. To get a strong enough connection between, say a video player and an HDTV, the electronics must communicate in a tightly focused, highly directional beam. In some ways this attribute makes 60 GHz perfect for a wireless personal area network--you can beam your home movie from your DVD player to your TV without worrying about your neighbors watching, too. But it also may mean that people walking through the beam can disrupt the link.
Phiar's Rentschler says that instead of being confined to a single chip in a single location in a computer or TV, his firm's antennas and transceivers can be distributed in tiny strips all over a consumer device, allowing it to pick up a signal from any direction.
Of course, other radio developers, such as the leading 60-GHz silicon firm SiBeam, in Sunnyvale, Calif., have solutions for the band's directionality problem, too. SiBeam chief technical officer Jeffrey Gilbert says his company's RF chip looks for the receiver, and if it's not directly in sight, the receiver figures out the best path to ricochet the signal--off a wall or the floor--to get to the target.
Not everyone is convinced that Phiar's technology will make it into upcoming consumer devices. ”I would personally be very skeptical of anyone saying they will put semiconductors out of business anytime soon,” says John Cressler, a professor of electrical engineering at Georgia Institute of Technology, in Atlanta, who studies the 60-GHz band. Though Phiar's tunneling-based approach will produce fast circuits, he says, silicon-germanium and CMOS chips have the advantage at 60 GHz if for no other reason than that they are already nearing production.
IBM's Brian Gaucher agrees. His company is developing silicon-germanium-based 60-GHz chips. ”I don't doubt the device physics,” says the Yorktown Heights, N.Y.based research staff member, ”but I think that traditional silicon, due to its maturity, is the technology that will likely be leveraged to enable the HD-multimedia revolution.”
Behzad Razavi, an electrical engineering professor at University of California, Los Angeles, adds that while Phiar has demonstrated an ability to use quantum tunneling to make fast devices, it is not clear whether those fast devices can be integrated into actual applications. ”Everything they have in their products is single devices,” he says. ”Diodes, transistors--these are single components.” In an integrated system, countless components running at a 60-GHz frequency must work together flawlessly. But the metal interconnects between them are extremely difficult to manage at such high frequencies. ”You have these little components, each good for 60 GHz, but the wires have introduced their own problems,” Razavi says.