Power/Performance Bits: Jan. 28

Accelerator-on-chip
Researchers at Stanford University and SLAC National Accelerator Laboratory created an electron-accelerator-on-chip. While the technique is much less powerful than standard particle accelerators, it can be much smaller.

It relied upon an infrared laser to deliver, in less than a hair’s width, the sort of energy boost that takes microwaves many feet.

The team carved a nanoscale channel out of silicon, sealed it in a vacuum and sent electrons through this cavity while pulses of infrared light – to which silicon is transparent – were transmitted by the channel walls to speed the electrons along.

The accelerator-on-a-chip is just a prototype, but Jelena Vuckovic, electrical engineer and professor at Stanford, said its design and fabrication techniques can be scaled up to deliver particle beams accelerated enough to perform cutting-edge experiments in chemistry, materials science and biological discovery that don’t require the power of a massive accelerator.

“The largest accelerators are like powerful telescopes. There are only a few in the world and scientists must come to places like SLAC to use them,” Vuckovic said. “We want to miniaturize accelerator technology in a way that makes it a more accessible research tool.”

The team built a chip that fires pulses of infrared light through silicon to hit electrons at just the right moment, and just the right angle, to move them forward just a bit faster than before.

This image, magnified 25,000 times, shows a section of an accelerator-on-a-chip. The gray structures focus infrared laser light (shown in yellow and purple) on electrons flowing through the center channel. By packing 1,000 channels onto an inch-sized chip, Stanford researchers hope to accelerate electrons to 94 percent of the speed of light. (Image credit: Courtesy Neil Sapra)

To accomplish this, they turned the design process upside down. In a traditional accelerator, like the one at SLAC, engineers generally draft a basic design, then run simulations to physically arrange the microwave bursts to deliver the greatest possible acceleration. But microwaves measure 4 inches from peak to trough, while infrared light has a wavelength one-tenth the width of a human hair. That difference explains why infrared light can accelerate electrons in such short distances compared to microwaves. But this also means that the chip’s physical features must be 100,000 times smaller than the copper structures in a traditional accelerator. This demands a new approach to engineering based on silicon integrated photonics and lithography.

Vuckovic’s team solved the problem using inverse design algorithms that her lab has developed. These algorithms allowed the researchers to work backward, by specifying how much light energy they wanted the chip to deliver, and tasking the software with suggesting how to build the right nanoscale structures required to bring the photons into proper contact with the flow of electrons.

The design algorithm came up with a chip layout that resembles nanoscale mesas, separated by a channel, etched out of silicon. Electrons flowing through the channel run a gantlet of silicon wires, poking through the canyon wall at strategic locations. Each time the laser pulses – which it does 100,000 times a second – a burst of photons hits a bunch of electrons, accelerating them forward. All of this occurs in less than a hair’s width, on the surface of a vacuum-sealed silicon chip, made by team members at Stanford.

The researchers want to accelerate electrons to 94 percent of the speed of light, or 1 million electron volts (1MeV), to create a particle flow powerful enough for research or medical purposes. This prototype chip provides only a single stage of acceleration, and the electron flow would have to pass through around 1,000 of these stages to achieve 1MeV. But that’s not as daunting at it may seem, said Vuckovic, because this prototype accelerator-on-a-chip is a fully integrated circuit. That means all of the critical functions needed to create acceleration are built right into the chip, and increasing its capabilities should be reasonably straightforward.

The researchers plan to pack a thousand stages of acceleration into roughly an inch of chip space by the end of 2020 to reach their 1MeV target. Although that would be an important milestone, such a device would still pale in power alongside the capabilities of the SLAC research accelerator, which can generate energy levels 30,000 times greater than 1MeV. But the team believes that, just as transistors eventually replaced vacuum tubes in electronics, light-based devices will one day challenge the capabilities of microwave-driven accelerators.

Meanwhile, in anticipation of developing a 1MeV accelerator on a chip, electrical engineer Olav Solgaard, a co-author on the paper, has already begun work on a possible cancer-fighting application. Today, highly energized electrons aren’t used for radiation therapy because they would burn the skin. Solgaard is working on a way to channel high-energy electrons from a chip-sized accelerator through a catheter-like vacuum tube that could be inserted below the skin, right alongside a tumor, using the particle beam to administer radiation therapy surgically.

Quantum benchmark
Researchers at Oak Ridge National Laboratory developed a quantum chemistry simulation benchmark to evaluate the performance of quantum devices and guide the development of applications for future quantum computers.

While still in their early stages, quantum systems have the potential to be exponentially more powerful than today’s leading classical computing systems. But determining what applications are well suited to the unique architectures of quantum systems is important to a variety of fields, such as complex chemistry calculations.

“We are currently running fairly simple scientific problems that represent the sort of problems we believe these systems will help us to solve in the future,” said ORNL’s Raphael Pooser, principal investigator of the Quantum Testbed Pathfinder project. “These benchmarks give us an idea of how future quantum systems will perform when tackling similar, though exponentially more complex, simulations.”

The team calculated the bound state energy of alkali hydride molecules on 20-qubit IBM Tokyo and 16-qubit Rigetti Aspen processors. These molecules are simple and their energies well understood, allowing them to effectively test the performance of the quantum computer.

By tuning the quantum computer as a function of a few parameters, the team calculated these molecules’ bound states with chemical accuracy, which was obtained using simulations on a classical computer. The quantum calculations also included systematic error mitigation, illuminating the shortcomings in current quantum hardware.

An ORNL research team lead is developing a universal benchmark for the accuracy and performance of quantum computers based on quantum chemistry simulations. The benchmark will help the community evaluate and develop new quantum processors. (Below left: schematic of one of quantum circuits used to test the RbH molecule. Top left: molecular orbitals used. Top right: actual results obtained using the bottom left circuit for RbH). (Credit: Jacek Jakowski, Oak Ridge National Laboratory)

Systematic error occurs when the “noise” inherent in current quantum architectures affects their operation. Because quantum computers are extremely delicate (for instance, the qubits used by the ORNL team are kept in a dilution refrigerator at around 20 millikelvin (or more than -450 degrees Fahrenheit), temperatures and vibrations from their surrounding environments can create instabilities that throw off their accuracy. For instance, such noise may cause a qubit to rotate 21 degrees instead of the desired 20, greatly affecting a calculation’s outcome.

“This new benchmark characterizes the ‘mixed state,’ or how the environment and machine interact, very well,” Pooser said. “This work is a critical step toward a universal benchmark to measure the performance of quantum computers, much like the LINPACK metric is used to judge the fastest classical computers in the world.”

While the calculations were fairly simple compared to what is possible on leading classical supercomputers, quantum computers are expected to be able to more accurately and more efficiently perform a wide swathe of chemistry-related calculations better than any classical computer currently in operation.

“The current benchmark is a first step towards a comprehensive suite of benchmarks and metrics that govern the performance of quantum processors for different science domains,” said ORNL quantum chemist Jacek Jakowski. “We expect it to evolve with time as the quantum computing hardware improves. ORNL’s vast expertise in domain sciences, computer science and high-performance computing make it the perfect venue for the creation of this benchmark suite.”

Next, the team plans to calculate the exponentially more complex excited states of these molecules, which will help them devise further novel error mitigation schemes.

Lithium-sulfur battery commercialization
Researchers at Monash University, CSIRO, University of Liège, and Fraunhofer Institute for Material and Beam Technology say they are close to commercialization of a lithium-sulfur (Li-S) battery that could outperform current market leaders by more than four times.

Using the same materials in standard lithium-ion batteries, researchers reconfigured the design of sulfur cathodes so they could accommodate higher stress loads without a drop in overall capacity or performance.

Inspired by unique bridging architecture first recorded in processing detergent powders in the 1970s, the team engineered a method that created bonds between particles to accommodate stress and deliver a high level of stability.

“Successful fabrication and implementation of Li-S batteries in cars and grids will capture a more significant part of the estimated $213 billion value chain of Australian lithium, and will revolutionize the Australian vehicle market and provide all Australians with a cleaner and more reliable energy market,” said Mainak Majumder, a professor of mechanical and aerospace engineering at Monash.

In addition to ultra-high capacity, the team says their battery has less environmental impact than current lithium-ion products.

“This approach not only favours high performance metrics and long cycle life, but is also simple and extremely low-cost to manufacture, using water-based processes, and can lead to significant reductions in environmentally hazardous waste,” said Matthew Hill, associate professor of chemical engineering at Monash.

The researchers have an approved filed patent for their manufacturing process, and prototype cells have been successfully fabricated by German R&D partners Fraunhofer Institute for Material and Beam Technology.