Silicon can now reconfigure itself with just a jolt of electricity

Semiconductor microparticles: Assemble!

The research points to a vaguely terrifying future of self-assembling and self-reconfiguring systems using a class of material known as active matter.

The team at Duke University and North Carolina State University developed a class of active semiconductor and diode particles able to draw energy and propel themselves in a controllable fashion.

osu javascript

The research, published in Nature Communications, showed the particles were able to assemble and disassemble on demand, which the team hopes could lead to self-healing networks, artificial muscles and smarter silicon.

“The idea is that eventually we’re going to be able to make silicon computational systems that assemble, disassemble and then reassemble in a different format,” said Nan Jokerst, the J. A. Jones Professor of Electrical and Computer Engineering at Duke.

Six types of thin film semiconductor silicon microparticles were fabricated on silicon-on-insulator (SOI) wafers using standard manufacturing processes, meaning that potentially millions of things could be made at a time.

The particles measure 3.5 x 10 x 20 microns (one thousandth of a millimeter), and can be scaled up and down as needed. The particles have positive (p) and negative (n) regions with some fabricated with both to create p-n junctions, allowing electricity to flow in only one direction.

By suspending the particles in fluid and applying varying magnitudes and frequencies of AC fields, the team were able to make the particles move and synchronise motions on demand.

The team then added metal to the surface of the particles to create p-n diodes with contacts.

Silicon Microparticles

While making particles propel themselves through fluid is undoubtedly neat, the scientists reckon this is only the groundwork, with more research needed into different geometry, field effects (optical and magnetic, rather than just electrical) and insulating properties.

With time, the team think that it will be possible to create a highly programmable form of active matter. Up to now, scientists have been able to assemble particles into structured networks but have been limited by an inability to controllably reconfigure those particles.

“This work provides a sense of the capabilities that are out there and is the first demonstration,” said Jokerst.

Such reconfigurable technology will doubtless be of great interest to chip vendors as they try to figure out a way of fixing bugs baked into processor architecture. However, it appears so far to be years away from leaving the lab bench. ®