Squishy transistors—a device concept for fast, low-power electronics

Squishy transistors—a device concept for fast, low-power electronics
Credit: iStock.com/Volodymyr Krasyuk

An international team of researchers from the National Physical Laboratory (NPL), IBM, the University of Edinburgh and Auburn University have shown that a new device concept - a 'squishy' transistor - can overcome the predicted power bottleneck caused by CMOS (complementary metal-oxide-semiconductor) technology reaching its fundamental limits.

Moore's law predicted that the number of transistors able to fit on a given die area would double every two years. As transistor density doubled, chip size shrank and processing speeds increased. This march of progress led to rapid advances in and a surge in the number of interconnected devices. The challenge with making anything smaller is that there are fundamental physical limits that can't be ignored and we are now entering the final years of CMOS transistor shrinkage.

Furthermore, this proliferation is driving an increase in data volume, accompanied by rising demands on energy to process, store and communicate it all; as a result, IT infrastructure now draws an estimated 10 % of the world's electrical power. Previous efforts have focused on remediation by reducing the amount of energy per bit. However, soon we will hit a power barrier that will prevent continued voltage scaling. The development of novel, low-power devices based on different physical principles is therefore crucial to the continued evolution of IT.

A team from NPL, IBM, Edinburgh and Auburn have demonstrated the capabilities of the Piezoelectric Transistor (PET) as a post-CMOS technology that could overcome these issues and restore voltage scaling. In the paper, published in Applied Physics Letters, the team explain the physics underlying the PET's behaviour and use theory and simulation to predict its performance when optimised across a wide range of application spaces, spanning several different length scales: including radio frequency switches (on the micron scale) and devices such a smartphones and phased array radar.

Squishy transistors—a device concept for fast, low-power electronics
Schematic of the PET

The conceptual device is based on a pressure-driven insulator-to-metal transition, and has proved to be a promising, fast, low-power option for future IT infrastructure, with a performance that cannot be matched by CMOS transistors. These results should spur further research into piezoelectric scaling, and the PET fabrication techniques needed to realise this device in the future.

Explore further

Innovative technology to recover performance of CMOS devices damaged by hot carrier injection

More information: "The piezoelectronic stress transduction switch for very large-scale integration, low voltage sensor computation, and radio frequency applications." Appl. Phys. Lett. 107, 073505 (2015); dx.doi.org/10.1063/1.4928681
Journal information: Applied Physics Letters

Citation: Squishy transistors—a device concept for fast, low-power electronics (2015, September 4) retrieved 21 May 2019 from https://phys.org/news/2015-09-squishy-transistorsa-device-concept-fast.html
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Feedback to editors

User comments

Sep 04, 2015
Since the article here doesn't go into great detail: It works like this: When a voltage is applied across "gate" and "common," the piezoelectric (PE) material expands. The whole setup is enclosed in a fixed height fixture, so the PE expansion causes the piezoresistive (PR) material to compress. When the PR compresses, it undergoes a transition from insulating to conducting. Thus current can flow from the common to the "sense."

The published paper reports that the timescales involved here are on the order of femtoseconds, so it should be a fast enough transition to work in modern electronics.

Sep 04, 2015
Moore's law predicted that the number of transistors able to fit on a given die area would double every two years.

No it didn't. Moore's law was about number of transistors on a chip at the lowest cost. It has absolutely nothing to do with transistor density, or power consumption, or processing power, but simply what's the most economical number of transistors on a chip.

Every Single Time they get it wrong.

It's also why the charts counting the absolute number of transistors per chip as "proof" of Moore's law are wrong, since they don't also normalize them by cost.

The actual Moore's law has been long dead.

Sep 04, 2015
This comment has been removed by a moderator.

Sep 04, 2015
Shavera, thank you for the explanation. Also, I did a search on the web to find more about how it works - before checking the comments. And found that piezoeffect transistors were considered and studied for the last 20 years (and probably more). So, the concept isn't new.

docile, that's right, half-informed journalist. Reminds me how much hash another 'journalist' made with Hawkins new black hole hypothesis. Really no one cared about the journalist ruminate in attempt to understand what Hawking is saying and how it differs from hundreds other hypotheses that the journalist have no idea about. 'Journalist' ended up very pleased with himself, while managing to convey very confused picture.Next time, just simple facts, please, and no literary language.

Sep 04, 2015
This comment has been removed by a moderator.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more