Power/Performance Bits: June 29

Persistent photoconductivity
Researchers at the U.S. Department of Energy’s National Renewable Energy Laboratory (NREL), University of Wisconsin Madison, and the University of Toledo, discovered a unique effect in metal-halide perovskite semiconductors that could be used in neuromorphic computing systems.

Perovskites are currently being investigated as highly efficient solar cells. In fact, the team was looking into a combination of perovskite nanocrystals with a network of single-walled carbon nanotubes they thought could have interesting properties for photovoltaics or detectors.

When they shined a laser at the combined material, it displayed persistent photoconductivity, a form of optical memory where light energy hitting device is stored in ‘memory’ as an electrical current.

“What normally would happen is that, after absorbing the light, an electrical current would briefly flow for a short period of time,” said Joseph Luther, a senior scientist at NREL. “But in this case, the current continued to flow and did not stop for several minutes even when the light was switched off.”

Persistent photoconductivity can be used to mimic brain synapses that store memory. But the phenomena was usually limited to low temperatures or high operating voltages, with the current spike lasting for fractions of a second.

The new device needed only low voltages and low light intensities at room temperature and was capable of producing an electrical current for more than an hour after the light is switched off. The team tried three formulations of perovskites (formamidinium lead bromide, cesium lead iodide, and cesium lead bromide) and found that reach produced persistent photoconductivity.

Such devices could be used in image recognition systems as artificial synapses, the researchers said.

“There are many applications where sensor arrays can take in images and apply training and learning algorithms for artificial intelligence and machine-learning-type applications,” said Jeffrey Blackburn, a senior scientist at NREL. “As an example, such systems could potentially improve energy efficiency, performance, and reliability in applications such as self-driving vehicles.”

However, there is more work to be done, Blackburn added. “What we made is only one of the simplest devices you could make from combining these two systems, and we demonstrated a simplistic memory-like operation. To build a neural network requires integrating an array of these junctions into more complex architectures, where more complex memory applications and image processing applications can be emulated.”

Speedy magnetic switching
Scientists at Trinity College Dublin recorded the fastest yet magnetic switching in a recently developed material.

The material, an alloy of manganese, ruthenium, and gallium (MRG), was developed by the group in 2014. Instead of using a magnetic field, a femtosecond laser system was used to hit thin films of MRG with bursts of red laser light, which delivered megawatts of power in less than a billionth of a second.

This heat transfer switches the magnetic orientation of MRG in a tenth of a picosecond. Importantly, the researchers were able to switch the orientation back again 10 picoseconds later. They noted that this switching and re-switching is six times faster than previously recorded.

Trinity researchers Jean Besbas and Karsten Rode stated: “Magnetic materials inherently have memory that can be used for logic. So far, switching from one magnetic state ‘logical 0’, to another ‘logical 1’, has been too energy hungry and too slow. Our research addresses speed by showing that we can switch MRG from one state to another in 0.1 picoseconds and crucially that a second switch can follow only 10 picoseconds later, corresponding to an operational frequency of ~ 100 gigahertz — faster than anything observed before. The discovery highlights the special ability of our MRG to effectively couple light and spin so, that we can control magnetism with light and light with magnetism on hitherto unachievable timescales.”

Michael Coey, a professor in Trinity’s School of Physics and CRANN, added, “This demonstration will lead to new device concepts based on light and magnetism that could benefit from greatly increased speed and energy efficiency, perhaps ultimately realizing a single universal device with combined memory and logic functionality. It is a huge challenge, but we have shown a material that may make it possible. We hope to secure funding and industry collaboration to pursue our work.”

Thin shortwave infrared imager
Engineers from the University of California San Diego, University of Southern Mississippi, and Samsung Advanced Institute of Technology built a compact shortwave infrared imager. Capturing infrared light wavelengths from 1000 to 1400 nanometers, just outside the visible spectrum, the imager could be used to inspect boards through silicon, among other applications.

The imager combines sensors and display into one device, avoiding the bulk of traditional solutions. It is made from organic semiconductors, making it flexible and safe for use in biomedical applications. It can also see more of the shortwave infrared spectrum compared to typical imagers. The team said it is fabricated using thin film processes, making it easy and inexpensive to scale up.

The new infrared imager is thin and compact with a large-area display. (Credit: Ning Li / UC San Diego Jacobs School of Engineering)

The device is constructed of multiple stacked semiconducting layers. A photodetector layer, an organic light-emitting diode (OLED) display layer, and an electron-blocking layer in between are each made of a different organic polymer.

The photodetector layer absorbs shortwave infrared light (low energy photons) and then generates an electric current. This current flows to the OLED display layer, where it gets converted into a visible image (high energy photons). An intermediate layer, called the electron-blocking layer, keeps the OLED display layer from losing any current.

In contrast to traditional imagers, the process of converting low energy photons to higher energy photons, called upconversion, is electronic. “The advantage of this is it allows direct infrared-to-visible conversion in one thin and compact system,” said Ning Li, a postdoctoral researcher at the UC San Diego Jacobs School of Engineering. “In a typical IR imaging system where upconversion is not electronic, you need a detector array to collect data, a computer to process that data, and a separate screen to display that data. This is why most existing systems are bulky and expensive.”

It can also provide both optical and electronic readouts. Li noted that when researchers shined infrared light on the back of a subject’s hand, this multifunctionality enabled the imager to provide a picture of the subject’s blood vessels while recording the subject’s heart rate. They also tested the imager’s ability to see through smog, which could be useful for autonomous and assisted driving, and through silicon wafers, which could aid inspection. The team next plans to work on improving the imager’s efficiency.