For Nonlinear Optical Processing, Things Do – and Don’t – Add Up

Hank Hogan, Contributing Editor

Sometimes you don’t want 2 + 2 to equal 4. Instead, it would be nice if the result were a lot more – or a lot less.

That’s the basis for nonlinear optical processing, where a material or device has an outsize response to incoming light. With that, it is possible to convert one frequency of light into another. Other outcomes can be optical parametric amplification, supercontinuum generation or fast switching. Nonlinear effects can lead to Raman lasers and other tools – including what could be the foundation for tomorrow’s computers.

To get a sense of where nonlinear optical processing is headed, a good place to look is the Optical Society of America (OSA)’s Nonlinear Photonics Topical Meeting, held this year in late June in Karlsruhe, Germany. Karsten Rottwitt, one of the program chairmen, reported 218 accepted papers, the largest number in years, and many more submitted. “It was very successful in terms of numbers of papers. I think that indicates a huge interest in the field.”

By exploiting nonlinear optical phenomena, researchers turned a tabletop oscillator (right) into a photonic chip (left). Courtesy of Marcello Ferrera and Dr. Luca Razzari, Ultrafast Optical Processing Group, INRS-EMT.

Rottwitt, an associate professor of photonics at Technical University of Denmark in Lyngby, noted that, in the past, the focus had been on optical communications. Today, applications in sensing and lasers are increasingly important, as are wavelengths other than the 1500-nm telecom standard.

What has not changed is the focus on new materials and new devices. Innovations in the first are particularly needed, Rottwitt said. An ideal material would be one with high nonlinearity, low loss, little two-photon absorption and easy processing.

On the device side, what is needed is compatibility with standard electronics and a nonlinear response. An example of this can be found in an integrated multiple wavelength laser source that was described in a paper at the OSA meeting. The device has a total conversion efficiency of almost 10 percent and frequency spacing of more than 6 THz from 1400 to 1700 nm, an important communication range.

It also has another important attribute.

“This material platform is the first CMOS-compatible platform that has been shown to be capable of very high nonlinear optical performance,” said research team member David J. Moss.

A physics professor at the University of Sydney in New South Wales, Australia, he noted that silicon has attracted considerable interest for a good part of the past decade as a nonlinear optical material. However, because it suffers from two-photon absorption, researchers have been looking for alternatives compatible with standard CMOS processing.

Moss, a group from the National Institute of Scientific Research – Energy, Materials and Telecommunications (Institut National de la Recherche Scientifique – Énergie, Matériaux et Télécommunication) in Quebec, and two researchers from Annapolis, Md.-based Infinera Corp. devised a solution based on a doped silica glass microring resonator. The ring measured 270 µm across and had a waveguide core about 1.5 μm square. To create that core, they buried a low-loss, doped glass with a refractive index of 1.7 inside standard silicon dioxide, which has a refractive index of 1.45.

Shown is output from a miniature multiple wavelength source (left) based on a ring resonator and an InGaAs camera image of the ring resonator in resonance conditions (right). Courtesy of Marcello Ferrera and Dr. Luca Razzari, Ultrafast Optical Processing Group, INRS-EMT.

The high-index glass is a proprietary blend, the rights to which are owned by Infinera. However, Moss said that any material with the right properties would do. What is needed, he said, are low linear and nonlinear loss, a high optical nonlinearity and a high linear refractive index contrast, as compared with normal optical fiber.

The proprietary formulation used by the researchers allows the creation of a low-loss film without the need for high temperature annealing. That is what makes the device and material CMOS-compatible.

The researchers measured the performance of their device, which works on third-order nonlinearities. They found the full width half maximum of the linear optical transmission spectrum to be 1.3 pm at 1544 nm, indicating a quality factor of about 1.2 million.

Moss reported that a number of companies have expressed interest in the device, which has possible application for next-generation telecommunications or as a means to optically interconnect computers. He cautioned, however, that research is ongoing as to what may be possible with such devices.

As for making devices such as this, it wouldn’t require a breakthrough or exotic manufacturing, Moss said. “A fairly standard semiconductor processing lab with cleanrooms and good lithography would suffice, so it is not that unique.”

Besides new devices, nonlinear optical processing also would benefit from new materials. In that regard, a nanowire design from researchers at Australian National University in Canberra showed a huge nonlinearity. They reported that they had achieved the highest nonlinearity ever obtained for a glass waveguide. What’s more, their construct had negligible two-photon and free-carrier absorption, leading to a nonlinear figure of merit 200 times that of silicon at 1550 nm.

They did this using chalcogenide glass, a mixture of germanium, arsenic and selenium. Coming up with the right formulation was a challenge, said research leader Barry Luther-Davies.

Nanowires (above) of chalcogenide glass generated the supercontinuum spectrum (right). Courtesy of Barry Luther-Davies, Australian National University.

“The physical properties of chalcogenide glasses are known to change under the influence of light or heat due to the propensity of these materials to allow switching of their chemical bonds when exposed to energy.”

This tendency would seem to preclude their use in all-optical devices, which require very stable materials. However, Luther-Davies said, the researchers found a small range of compositions where these effects were substantially reduced. The glass used in work covered at the OSA meeting was done with an example of this composition.

A professor of laser physics, Luther-Davies explained that another key to the material performance was its fabrication into nanowires. The researchers created 700 x 530-nm-high ribs on a silica substrate, using lithography to pattern and etching to reveal the nanowires. They coated them with a 5-nm layer of aluminum oxide and followed that with a 15-μm-thick polysiloxane cladding around the glass core of the waveguides.

One problem with chalcogenides is how to obtain very smooth etched sidewalls. That, said Luther-Davies, requires optimization of the gas chemistry used for etching and the film homogeneity.

He reported that the group is starting to work on this issue, trying a combination of thermal and optical annealing to smooth the sidewalls. Achieving the smoothest possible sidewalls will result in the best performance.

As for large-scale manufacturing, that likely will require a deep-ultraviolet stepper, similar to the type of device used for semiconductor lithography today. That should reduce the cost of manufacturing and might allow high throughput.

Long term, the material could find a home in a future communications technology based on all-optical devices. There, the absence of free-carrier and nonlinear absorption, particularly as compared with silicon, could be a significant advantage.

Luther-Davies is one of those who envision a three-dimensional hybrid structure, with light moving between layers with different properties. Some will be passive components, and others will be nonlinear devices.

An artist’s conception illustrates a 3-D hybrid structure with nonlinear and passive optical elements, enabling all-optical signal processing. Courtesy of Barry Luther-Davies, Australian National University.

“We believe this will lead to a practical method by which nanowires can be combined with more conventional circuitry without facing insurmountable alignment challenges that accompany the more conventional approach to hybrid circuitry,” Luther-Davies said.

A final example of nonlinear optical processing comes from the University of Stuttgart in Germany, where researchers demonstrated an all-optical control scheme of a hybrid plasmonic-photonics system, with their approach exploiting a nonlinear third-harmonic-generation process.

Using a control pulse injected after a start pulse, researchers demonstrated all-optical control of a plasmonic mode, raising the possibility of creating a whole new class of all-optical computers. Courtesy of Tobias Utikal, University of Stuttgart.

They did so by firing a sub-10-fs near-infrared laser pulse into a metal structure sitting on top of a quartz substrate. The metal was fabricated in 100-nm-wide ridges spaced 530 nm center to center.

The first pulse excited a polariton polarization in the nanostructure. The investigators then fired a second pulse into the device a few tens of femtoseconds later. This one either turned off the polarization or enhanced it, depending upon the length of the delay. To probe what was going on, they used a third pulse.

Tobias Utikal, a graduate student, was lead author of a March 2010 Physics Review Letters paper that described this approach. He noted that plasmonic structures could form the basis for all-optical nano-devices upon which an entirely new class of computers could be built.

This is a possibility because plasmonic structures combine optical responses in the femtosecond time range with nanoscale-size electric fields. Right now, however, such computers are a long way off, partly because there currently is no way to predict the nonlinear behavior of plasmonic devices.

Thus, one of the goals is to develop the knowledge needed so that devices can be designed for the lowest power and highest speed switching. When and if that happens, the results obtained by the Stuttgart researchers in their three-pulse setup could provide a scheme for building these devices.

As Utikal said, “The results show that the plasmon can indeed be switched off or reamplified at any time after its excitation. This method might be utilized in the future in active plasmonic devices for ultrafast optical switching.”