The human brain is the most powerful and energy-efficient computer known and one of its unique properties is that it has a folded cortex, which makes it ultra-compact and able to house billions of neurons. Mimicking the brain’s architecture will be no easy task – even using the most sophisticated processing technologies available today. “If we consider that a single state-of-the-art transistor, made by Intel, for example, is equivalent to a single brain neuron, then it will take a minimum of 50 wafers that are 450 mm in size to even begin thinking about mimicking the human brain,” explains team leader Muhammad Hussain. “The only way to come any where near such miniaturization and obtain a compact structure will be to fold the wafers.”

However, folding traditional silicon (which makes up 90% of all electronics devices manufactured today) is just not possible because it is rigid and brittle. This is where flexible electronic materials could come into their own.

Although flexible electronics have considerably progressed in the last decade (with the advent of novel applications, such as electronic paper-like displays and artificial skin), most of these devices are made from organic materials like polymers, carbon-based nanowires and nanotubes. These materials are far from ideal in that they have much slower processing speeds than their silicon counterparts. They are not very robust to higher temperatures either and cannot be formed as continuous films – unlike bulk silicon that can.

The disadvantages do not end there: integration densities (the number of transistors that can be packed onto a single chip) for these materials are also much lower than those for the silicon-based transistors found in today's microprocessors. And although inorganic flexible electronics do exist, they rely on expensive substrates such as SOI, UTSOI and silicon (111), and it is not easy to fabricate ultrathin silicon or porous silicon.

Simple and cost-effective process

Hussain’s team new method overcomes these problems since it exploits traditional state-of-the-art CMOS-compatible processes to transform silicon-on-insulator (SOI) based FinFETs into flexible and semi-transparent silicon-on-polymer ones, while retaining silicon’s high performance and integration density. All device components can be fabricated prior to their release, which also simplifies the process to a great extent.

The researchers made their FinFETs using DUV lithography, high-dielectric/metal gate stack deposition and patterning, and ion implantation, to form source and drain regions and contact modules in the devices. They then employed specially engineered masks to cover the contact pads (since the devices are encapsulated with an interlayer dielectric) before etching through the SOI layers.

“Next, we covered the sidewalls of the etched trenches using vertical spacer layers and followed this step by a xenon fluoride-based isotropic etch to release the top 1 µm surface from the wafer,” explained Hussain. “The wafer can be easily bent (it has an unprecedented bending radius of 0.5 mm) and we can recycle the entire ensemble to release more wafer layers – something that makes the whole process cost-effective.”

"Outstanding electrical and physical properties"

The finished FinFETs have outstanding electrical and physical properties, he adds. For example, NMOS devices have a sub-threshold swing of 80 mV/dec and their PMOS counterparts a value of 70 mV/dec. Current on-off ratios are 4.6 and 4.78 decades for both respectively. And since the released fabric is only 1μm thick, and the host substrate is a polyimide sheet, the devices are semi-transparent.

“Our work confirms that our CMOS-compatible trench-protect-peel-recycle process can be used to make 3D non-planer flexible FinFETs,” Hussain told nanotechweb.org. “The technique is an important step towards brain-architecture inspired fully bendable and truly high performance and transparent computer chips and communications devices.”

Other applications for the flexible films include biomedical sensors and augmented, smart, electronic-skin.

The Saudi Arabia team says that it is now busy trying to fabricate such devices using its technique. “We are even looking at employing our method to make similar gadgets from non-silicon materials,” revealed Hussain.

The research is described in Advanced Materials.

Further reading

Quantum dots for superior solar cells (Aug 2012)
Flexible TEG breaks new power record (Dec 2013)