A team of researchers believe they might have found a way to produce ultra-low power chips using ferroelectric materials, helping to get a leg up over a Moore’s Law barrier.
The researchers at the University of California, Berkeley, have been adding up the numbers and dividing by their shoe size to work out how to reduce the minimum voltage needed to store charge in a capacitor. This would mean reducing the overall power and heat generation of current electronics.
And it is heat which is one of the main stumbling blocks to the move to smaller and smaller processes.
Common sense says that it is difficult to move past the around a seven nanometre process for chip production with the conventional electronics. This has lead to a variety of exotic technologies being researched, such as spintronics and quantum computing.
But it appears that the Berkeley boffins have a way of fine tuning more conventional transistor systems.
This is down to the use of ferroelectric materials which can improve the efficiency of transistors. Ferroelectric materials exhibit some unusual and beneficial qualities and can hold both positive and negative electrical charges.
It is even possible to hold an electrical charge even when a voltage is not applied to it, while its electric polarisation can be reversed when an external field is applied.
The team were able to demonstrate for the first time that a capacitor made using ferroelectric materials combined with a dialectric insulator, the charge accumulated for a voltage can be amplified. This is known as negative capacitance, and the team reckon it is a “viable strategy to overcome the power draw of today’s transistors”.
The method involved the use of the ferroelectric material lead irconate titanate (PZT ) and the dialectric strontium titanate (STO), stacked together. A voltage is then applied to the PZT-STO structure and a layer of STO on its own, comparing the amount of charge stored in both. With the ferroelectric material there was a “two-fold voltage enhancement” for the same voltage, which they believe could go even higher.
While it is currently necessary to have at least one volt to operate a transistor, what the researchers at Berkely have essentially done is reduce the amount of voltage that is needed to generate a larger charge from a smaller voltage.
This means less heat dissipation, and therefore allows more transistors to fit onto a chip. While a bottle neck has been reached since 2005, with a plateau in Moore’s Law as the researchers say, this could allow for even greater advances in clockspeeds once again.
Unfortunately the effect that the team found works at a higher temperature than would be found without slowly roasting you laptop, working at around 200 degrees Celsius. Usually 85 degrees Celsius is the limit for modern processors. The researchers are looking at room temperature negative capacitance now though.
Also the ferromagnetic materials used are not likely to be commonplace or cheap, so the technology may still be someway from being commercially viable.