Power/Performance Bits: Oct. 17

Harvesting body heat
Researchers at the Georgia Institute of Technology developed a flexible, wearable thermoelectric generator that can harvest energy from body heat to power simple biosensors.

Thermoelectric generators have been available for decades, but standard designs use inflexible inorganic materials that are too toxic for use in wearable devices.

The team’s device uses thousands of dots composed of alternating p-type and n-type polymers in a closely-packed layout. By placing the polymer dots close together with an inkjet printer, the interconnect length decreases, which lowers the total resistance and results in a higher power output from the device.

An example of a circuitry pattern based on the Hilbert curve and printed on paper for thermoelectric conversion. (Source: Candler Hobbs, Georgia Tech)

“Instead of connecting the polymer dots with a traditional serpentine wiring pattern, we are using wiring patterns based on space filling curves, such as the Hilbert pattern – a continuous space-filling curve,” said Kiarash Gordiz, a postdoctoral fellow at the Colorado School of Mines. “The advantage here is that Hilbert patterns allow for surface conformation and self-localization, which provides a more uniform temperature across the device.”

The circuit’s fractally symmetric design allows the modules to be cut along boundaries between symmetric areas to provide exactly the voltage and power needed for a specific application. That eliminates the need for power converters that add complexity and take power away from the system.

“This is valuable in the context of wearables, where you want as few components as possible,” said Akanksha Menon, a Ph.D. student at Georgia Tech. “We think this could be a really interesting way to expand the use of thermoelectrics for wearable devices.”

So far, the devices have been printed on ordinary paper, but the researchers have begun exploring the use of fabrics.

With the design, the researchers expect to get enough electricity to power small sensors, in the range of microwatts to milliwatts. That would be enough for simple heart rate sensors, but not more complex devices like fitness trackers or smartphones. The generators might also be useful to supplement batteries, allowing devices to operate for longer periods of time.

Among the challenges ahead are protecting the generators from moisture and determining just how close they should be to the skin to transfer thermal energy while remaining comfortable for wearers.

Cheap sodium battery
Researchers at Stanford University developed a sodium-based battery that can store the same amount of energy as a state-of-the-art lithium ion, at less than 80% of the cost.

Materials constitute about one-quarter of a battery’s price, and lithium costs about $15,000 a ton to mine and refine. The team is basing its battery on widely available sodium-based electrode material that costs just $150 a ton.

“Nothing may ever surpass lithium in performance,” said Zhenan Bao, professor of chemical engineering at Stanford. “But lithium is so rare and costly that we need to develop high-performance but low-cost batteries based on abundant elements like sodium.”

This sodium-based electrode has a chemical makeup common to all salts: It has a positively charged ion, sodium, joined to a negatively charged ion. In table salt, chloride is the positive partner, but in the Stanford battery a sodium ion binds to a compound known as myo-inositol, found in baby formula and derived from rice bran or from a liquid byproduct of the process used to mill corn. Crucial to the idea of lowering the cost of battery materials, myo-inositol is an abundant organic compound.

The sodium salt makes up the cathode, while the anode is made up of phosphorous. The researchers plan to focus next on tweaking the anode to improve its performance.

While the focus so far has been on the cost of the battery compared to lithium, in the future they’ll look at volumetric energy density, how big a sodium ion battery must be to store the same energy as a lithium ion system.

Record optical switching capacity
Japan’s National Institute of Information and Communications Technology (NICT) demonstrated an optical switching capacity of 53.3 Tb/s for short-reach data center networks, a record.

The demonstration makes use of spatial division multiplexing (SDM) over multi-core optical fibers (MCFs) and a newly developed high-speed spatial optical switch system, enabling full packet granularity. According to the researchers, the new data center network provides a significant improvement of network efficiency and end-to-end energy consumption per bit compared to today’s optical circuit, fully-electronic packet switching networks.

NICT developed a high-speed 7-core-joint optical switching system that can switch all the cores of a 7-core MCF simultaneously with switching speed of 80 ns. The system consists of multiple electro-absorption (EA) optical switch elements with several nanoseconds switching speed. It also contains a switch controller, capable of reading the destination address of packets and controlling multiple EA switches simultaneously.

Concept diagram of high-speed 7-core-joint optical switch system. (Source: NICT)

Using this optical switching system, the team built a testbed of a time-slotted optical network, capable of achieving full packet granularity. This testbed used 64 wavelength channels, modulated at 32 Giga Baud with polarization division multiplexing (PDM) quadrature phase shift keying (QPSK). This delivered a nominal capacity of 53.3 terabits per second. In the testbed, three MCF segments were used: a 19-core 28 km fiber, a 19-core 10 km fiber, and a 7-core 2 km fiber. On each fiber, 7-cores were used in this demonstration to carry information signals.

In the future, the team will work to increase the network capacity by developing new optical switches with faster response, lower insertion loss and flatter frequency response, and to investigate coherent burst-mode receivers with high-order modulation formats for greater spectral efficiency.