In This Computer Component, Data Slides Through Honey

This article is part of The State of Science, a series featuring science stories from public radio stations across the United States. This story, by Jes Burns, was originally published by OPB.

A honey bear is probably one of the weirder things you’d see in a science lab, especially in a lab making computer parts.

“It’s just processed, store-bought honey,” said Ph.D. student Zoe Templin. “Off the shelf — a little cute bear so we can put it in photos.”

But for Templin and her colleagues at Washington State University, Vancouver, the honey is key.

“It is cheap and it is easily accessible to everyone,” said master’s student Md Mehedi Hassan Tanim.

The honey also has natural chemical properties that make it a promising foundation for a new kind of environmentally friendly computer component — one that could make computing faster and more energy efficient while reducing the impact on the environment.

“For me electronics in general produce so much toxic waste. It really is a devastating amount of electronic waste that occurs,” said Templin.

In fact, the world produces about 50 million tons of electronic waste per year. Only 20% is recycled, meaning the other 80% ends up in landfills.

Using honey instead of materials like silicon in computers would make recycling easier and less toxic, because honey breaks down in water.

“I would love to see this become a stepping stone to having electronically clean waste,” said Templin. “Actually getting organic materials out into your phone, into commercial use.”

Memristors

Templin and Tanim have spent a good deal of time in the University’s cleanroom fabricating electronic components called memristors.

“We generally fabricate devices at the start of the semester for like 2-3 weeks and then we test them later,” said Tanim. “I’ve fabricated like 40 to 50 samples, but there are hundreds of devices on each sample.”

The devices are extremely small so the dust-free space of the cleanroom and head-to-toe protective gear is necessary to create the pristine metal and honey layers that comprise the chips.

“We have to keep our devices as clean as they are because our honey film is so thin. Any particles that land on that film… [are] going to be a defect and it is going to affect the performance of our device,” Templin said.

She places a glass slide in a centrifuge and drips a few drops of a specially diluted honey mixture on the surface. She starts the machine, which will spin the honey into an even layer.

“It’s thinner than our hair,” she said.

They’ll make the memristors by sandwiching the honey between two metal electrodes.

Memristors are relatively new in the world of electronics. The term is a portmanteau of the words “memory” and “resistor” — both essential for computers.

The memristor’s function was first theorized in the 1960s, but no one managed to actually make one until 2008.

Its combination of memory and electricity resistance is unique in electronics, but there’s another place where this combo is quite common — in our brains.

Our brains are full of nerve cells called neurons, and we learn when our neurons create connections between each other. The points where the neurons connect are synapses.

When we hear a new piece of information, a small bridge is built across the synapse. The bridge is our memory. The more we hear the information repeated, the more robust that memory bridge becomes.

“So we have our neurons and the synapse, they process data and they store data at the same time,” said WSU Vancouver electrical engineering professor and project head Feng Zhao.

This dual function is a huge deal and helps make our brains the most efficient computer on the planet.

The team’s memristor works the same way. It builds bridges (microscopic filaments of metal atoms) through the honey when exposed to certain levels of electricity — that’s the data processing. When the electricity stops flowing, the bridge remains intact — that’s the memory.

If you reverse the flow of electricity, the memory is wiped clean.

“We are building systems of memory that are as efficient as how the brain stores memory,” said Templin.

Zhao’s lab at WSU Vancouver discovered that honey has the chemical characteristics needed to make memristors work.

“They have big molecule chains in the material. And we need those molecule chains in order for us to build up those bridges,” said Zhao.

The team also focused on honey because it’s stable and doesn’t spoil. And unlike other traditional computer materials, producing honey actually benefits the environment because of the key role bees play in pollination.

“Using honey as the synapse element is great. It’s very clever. It’s a nice advance as far as getting things to be biodegradable,” said Purdue University physicist Erica Carlson. Carlson’s research has similarly focused on finding new materials to create faster, more efficient computers, but she is not involved in the WSU Vancouver project.

A New Branch Of Computing

So if you have a computer component that mimics a brain cell, could you put a bunch of them together and get a computer that mimics a brain?

The answer is yes.

“Right now our device simulates a single neuron, a single synapse, but we want to integrate those devices together,” said Zhao. “And then this will behave more like our brain.”

This is the idea behind neuromorphic computing — a new revolution in computer design.

“We’re coming up against some hard technological issues to where we need fundamentally different forms of computation if we’re going to keep the world moving forward the way that we want to keep moving forward,” Carlson said.

Pretty much every computer we use is based on a design that’s 80 years old: The von Neumann architecture.

In this design, the part of the computer that processes information is separate from the part that stores it. And it takes time and energy for the information to travel in between.

“This conventional computing system consumes a tremendous amount of energy,” Zhao said.

Neuromorphic computers solve this bottleneck, in part, by having the processor and storage all in one place — the memristor.

“It’s a building block,” said Tanim. “By combining millions of devices like this, I can build a neuromorphic … chip, which can actually do the computation that a supercomputer can do — but 100 times faster and using 1000 times less energy.”

These neuromorphic computer systems are still in their infancy as far as development goes, but that could be changing with the seemingly overnight emergence of artificial intelligence.

“Right now, the way we’re doing AI is we’re using software to mimic the brain, and we’re putting that software on computer architectures that weren’t designed for it. So that’s why it’s so enormously energy intensive,” said Carlson.

It’s estimated that within four years, AI servers like ChatGPT will use as much electricity annually as countries the size of Argentina.

Neuromorphic computer systems could eventually reduce that energy use significantly.

“Now that we’re in the midst of the AI revolution, it would not surprise me if we started to see a lot more funding,” she said. “I mean, we’re trying to do something fundamentally different, and nobody knows which is going to be the best method.”

With honey-based devices, the WSU Vancouver team thinks they’re on track to help solve two problems at once.

By developing a biodegradable computer component made of honey, for a computer system that’s more energy efficient than any we’ve ever seen, the team is taking computers into a faster and cleaner new age.

“I’m actually extremely fortunate to have been raised and grown up in the Pacific Northwest, that cares about the environment to this level,” said Templin. “Where I can do research that could … truly help the environment in a way that really hasn’t been explored before.”