Undergraduate Squeezes More Out of Memory Technology

Research conducted by Embry-Riddle Aerospace Engineering undergraduate Ambroise Juston could one day help produce smaller, more efficient memory storage — a crucial component to powering the future of machine learning.

Nearly every electronic device uses solid-state memory. Juston’s research into a developing technology called memristors is finding that this technology can increase the amount of digital information that can be stored compared to complementary metal-oxide semiconductor (CMOS) memory, the most common kind of silicon computer chip technology, said Dr. John F. Sevic, associate professor of Electrical Engineering at Embry-Riddle Aeronautical University and the advisor on the research.

Sevic said Juston’s findings have implications for what is known as Moore’s Law, which predicts that the transistor count on integrated circuits doubles roughly every two years. Proposed by Intel co-founder Gordon Moore, Moore’s Law has been key to driving semiconductor innovation. However, as the technology nears its physical limitations, experts question how long it will hold.

“We are working around the limits of Moore’s scaling law,” Sevic said.

Juston, a junior from Lyon, France, presented the research on memristor memory technology — a product of a collaboration with the University of California Santa Cruz (UCSC) — at the Society of Photo-Optical Instrumentation Engineers (SPIE) conference last August in San Diego, California.

He had no idea what memristors were when he began studying them, he said. “But as I became more curious, I realized the incredible potential this research holds for the future of memory technology.”

To conduct the research, Juston employed cutting-edge engineering resources available to the Undergraduate Multiphysics Computation Group (UMCG) on the Prescott Campus.

“We are extremely fortunate at Embry-Riddle to have a rich set of industry-standard simulation tools widely adopted by our employers and in graduate schools for our students’ use,” said Sevic, the group’s director. “The research we are doing breaks the myth that undergraduate students must be exposed to simple or trivial problems. This is untrue.”

Aerospace Engineering undergraduate Ambroise Juston and Dr. John F. Sevic, professor of electrical engineering, show off a poster summarizing their recent research on memristors. Juston recently presented the research at a conference. (Photo: Embry-Riddle/Wilson Van Ness)

Smaller, Faster and More Efficient

A memristor is a two-terminal solid-state device with two persistent resistance states: a “1” and a “0”. Juston used computer simulations to guide the physical design and optimization of memristor-based memory elements to reveal their full potential.

“Simulation is ubiquitous in the modern electrical engineering workflow,” Sevic said.

“Many industries would not exist in their current state without simulation, impacting the chips used in modern cell phones or high-performance aircraft. A significant advantage of simulation generally is a marked increase in accessibility to impactful engineering analysis and design.” 

An example of conducting filaments spanning from the top to the bottom of a memristive-thin film. (Photo: Embry-Riddle/Ambroise Juston)Juston’s studies focus on Multiphysics phase-field modeling, a powerful technique that reveals how nanoscale materials alter their structure to produce a memory function in a computer.

“Our simulations focus on understanding how the geometry of the memristor impacts the formation of memory states,” explained Juston. “We discovered that the amount of conducting filaments increases linearly, which had been theorized but not tested.” 

Juston’s research shows that memristors fabricated at UCSC increase memory storage and density compared to current methods. This could help engineers and scientists manufacture smaller, more powerful devices with increased memory capacity and propel vast cloud computing data centers used in artificial intelligence beyond their current limits.

Juston said that presenting his research paper at the SPIE conference was intimidating but rewarding. He looks forward to continuing his research next year, exploring thermal effects on memristors — a critical aspect of the technology’s future development.

“I am still inexperienced in memristors and computer simulations, but I am happy to be able to contribute something of value to the greater community,” Juston said.

Posted In: Engineering | Research