Skip to Content

The Best Photon Detector Yet

Cheap, high-performing detectors could boost Internet bandwidth and speed up computers.
December 8, 2008

Silicon is used throughout the electronics industry to make countless components and devices. This cheap and abundant chemical could also be about to transform the photonics industry, by making vital communication equipment cheaper and better. Engineers have already made a number of optical devices with silicon. On Sunday, researchers at Intel announced another significant advance: a silicon-based photodetector that achieves higher speed and sensitivity than other devices, including those made using exotic, expensive materials.

Avalanche array: The circles shown in the above array are silicon photodetectors fabricated at Intel. The detectors were made different sizes so that the researchers could compare their performance.

Light detectors are typically made of III-V semiconductors–materials composed of multiple elements such as indium and phosphorous. Although these semiconductors have great optical properties, they are more difficult to manufacture in high quantities than silicon, which makes them expensive. In the quest to make silicon photonics a reality and to match III-V semiconductor performance, researchers have sought to exploit the way electrons behave in silicon. And Intel’s research, detailed yesterday in Nature Photonics, shows that silicon can make a detector that goes above and beyond expectations.

“This is the first time a silicon photonics device has [shown] better performance than any recorded for a III-V-based material,” says Mario Paniccia, an Intel fellow and director of the company’s photonics technology lab in Santa Clara, CA.

Detectors are crucial components in optical networks, receiving information transmitted via the light sent through fiber optic cable. Previously, Intel has built silicon-hybrid lasers, silicon-based modulators (which encode data on light), and detectors that use silicon as a channel to guide light.

Intel’s new device is based on a pre-existing design: an avalanche photodetector. In this type of detector, light is received and amplified by electron behavior within the material used. Avalanche photodetectors are found inside bulky optical networking equipment and currently cost $200 to $300 dollars apiece. If these detectors could be made out of silicon, Paniccia says they could cost less than $10. This would translate into higher bandwidth at cheaper prices and make it feasible to add optical networking to computers themselves, replacing the slow copper wiring that connects circuit boards and microchips with superfast optical connectors.

A simple photodetector converts photons to electricity similar to a solar cell. Once inside the detector, a photon creates a quasi-particle made up of a negatively charged electron and positively charged “hole” (an electron vacancy). When voltage is applied to the detector, the electron and the hole separate, producing an electrical current. The characteristics of this current can reveal information encoded onto the light.

Wire it up: A packaged silicon avalanche photodetector (the gray square in the center of the image) is wired to several electrical contacts. The setup lets researchers test the device.

Avalanche photodetectors are based on similar principles. The difference is that a special region of the detector creates additional electron-hole pairs, so a single incoming photon can produce tens or hundreds of electrons, effectively amplifying the signal. Avalanche photodetectors use III-V materials for the amplification, but Intel’s researchers turned to silicon instead. The new detector consists of an absorption region made of germanium and a “multiplication region” made of silicon. Previously it had been tricky to combine these two materials due to the spacing of each material’s atoms. But a special method for laying germanium developed at Intel helped overcome this challenge. “The benefit of doing this in silicon is that it’s a much better multiplication material because of its inherent properties of low noise and crystalline purity,” says Paniccia.

The performance of a detector is measured by a characteristic called gain-bandwidth, which measures in gigahertz its potential speed and sensitivity. Traditional detectors operate with a gain-bandwidth of around 120 gigahertz, but Intel’s detector reaches 340 gigahertz. What this means, explains Paniccia, is that it can produce a larger electrical response, for a given speed, than traditional detectors. Ultimately, this gives engineers the freedom to build network components that operate faster and more efficiently.

“We now have the opportunity to build high-speed optical links in networks and bring low-cost optical communication in and around our PCs,” Paniccia says.

Intel’s work is “a significant development,” says Bahram Jalali, professor of electrical engineering at the University of California, Los Angeles. “Avalanche photodetectors provide the amplification needed to detect very weak optical data,” he says. “Normally, the downside is that the avalanche process that leads to amplification also adds noise to the data,” an undesirable characteristic that must be cleaned up using extra electronics. “However,” he says, “silicon’s material properties lead to very low avalanche noise.”

Paniccia stresses that the new detector is not yet ready to appear in products. “We still have work to do in reducing dark current,” he says, referring to stray current that leaks from the device even when it’s not absorbing photons. But he says that he expects to see commercial silicon photodetectors from Intel in the next couple of years. For one thing, it should be possible to make the device with the equipment used to make other kinds of silicon electronics. “This was fabricated in our production fabs alongside other products,” Paniccia says. “There’s nothing fundamental there that we couldn’t commercialize in the future.”

Keep Reading

Most Popular

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

It’s time to retire the term “user”

The proliferation of AI means we need a new word.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Sam Altman says helpful agents are poised to become AI’s killer function

Open AI’s CEO says we won’t need new hardware or lots more training data to get there.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.