This radical discovery could turn semiconductor manufacture inside out
November 30, 2012
a, Au (gold) agglomerate formation; b, Au agglomerate size sorting using a differential mobility analyzer; c, Au agglomerate compaction into spherical particles in a furnace; d, nanowire growth; e, nanowire deposition (credit: Lars Samuelson et al./Lund University)
A completely new method of manufacturing the smallest structures in electronics could make their manufacture thousands of times quicker, allowing for cheaper semiconductors.
Instead of starting from a silicon wafer or other substrate, the idea is to grow gallium arsenide semiconductor structures from freely suspended nanoparticles of gold in a flowing gas. Semiconductor nanowires are key building blocks for the next generation of light-emitting diodes, solar cells, and batteries, according to Lund University researchers.
Behind the discovery is Lars Samuelson, Professor of Semiconductor Physics at Lund University, Sweden, and head of the University’s Nanometer Structure Consortium. He believes the technology will be ready for commercialization in two to four years. A prototype for solar cells is expected to be completed in two years.
“When I first suggested the idea of getting rid of the substrate, people around me said ‘you’re out of your mind, Lars; that would never work.’ When we tested the principle in one of our converted ovens at 400°C, the results were better than we could have dreamed of,” he says.
Nanowire (credit: Lars Samuelson et al./Lund University)
“The basic idea was to let nanoparticles of gold serve as a substrate from which the gallium arsenide semiconductors grow. This means that the accepted concepts really were turned upside down!”
Since then, the technology has been refined, patents have been obtained and further studies have been conducted. In the article in Nature, the researchers show how the growth can be controlled using temperature, time and the size of the gold nanoparticles.
Recently, they have also built a prototype machine with a specially built oven. Using a series of ovens, the researchers expect to be able to “bake” the nanowires, as the structures are called, and thereby develop multiple variants, such as p-n diodes. A further advantage of the technology is avoiding the cost of expensive semiconductor wafers.
“In addition, the process is not only extremely quick, it is also continuous. Traditional manufacture of substrates is batch-based and is therefore much more time-consuming,” adds Samuelson.
Aerotaxy
SEM image showing deposited nanowires on a Si substrate (credit: Lars Samuelson et al./Lund University)
At the moment, the researchers are working to develop a good method to capture the nanowires and make them self-assemble in an ordered manner on a specific surface. This could be glass, steel or another material suited to the purpose. The reason why no one has tested this method before, in the view of Professor Samuelson, is that today’s method is so basic and obvious. Such things tend to be difficult to question.
However, the Lund researchers have a head start, thanks to their parallel research based on an innovative method in the manufacture of nanowires on semiconductor wafers, known as epitaxy — so the researchers have chosen to call the new method aerotaxy. Instead of sculpting structures out of silicon or another semiconductor material, the structures are instead allowed to develop, atomic layer by atomic layer, through controlled self-organization.
The breakthrough for these semiconductor structures came in 2002 and research on them is primarily carried out at Lund, Berkeley and Harvard universities.
The Lund researchers specialize in developing the physical and electrical properties of the wires, which helps create better and more energy-saving solar cells, LEDs, batteries and other electrical equipment that is now an integrated part of our lives.
Comments (7)
by Damon Montano
….previous comment @ralph dratman
by Ralph Dratman
Damon Montano, thank you for your comment on my admittedly rash enthusiasm. I am quite aware of the need for caution in jumping to conclusions about new technical achievements. Nevertheless, this particular advance strikes me as well worth a hop or two.
That is because I’ve done a great deal of thinking about the power of dense 3D circuits (literally since I was a child), about the advantages of growing complex devices instead of building them (since young adulthood), and about the brain as a gargantuan graph embedded in 3 dimensions (since stumbling into a foolish middle age).
Today, on the edge of senescence, I try out new things on my iPhone and perceive that information technology has advanced farther than I ever expected in my lifetime. Then I consider that point-like nodes connected by curve-like wires are not subject to the routing constraints of circuitry limited to two dimensions. I recall that self-assembly means you don’t have to see or touch something to put it together. I summon today’s almost-hallucinatory information technology to learn that nerve impulses in our brain travel about 3 million times slower than light.
All this suggests to me that our species will be building brain-scale, brain-speed devices much sooner than I would have expected even two or three years ago.
Doddering old dreamer that I am, such a thought still gives me a shiver.
by Damon Montano
Talk about jumping to conclusions….
by David Ish Shalom
This article reminds me one of the almost forgoten pioneers of nanotechnology, K Eric Drexler, and his visionary book Engines Of Creation. Of course Kurzweil The Singularity… and all the notion of bottom up technology.
by Cybernettr
“He believes the technology will be ready for commercialization in two to four years. A prototype for solar cells is expected to be completed in two years. “
Usually such estimates tend to be hopelessly optimistic and I automatically tend to double or triple them mentally, but with the LOAR gaining steam, this may indeed not be as far from reality as one might think.
by Gorden Russell
Thank you for finding this article, Amara. Self-assembly is the yellow brick road to the Sing.
by Ralph Dratman
This is huge. Every entity we are familiar with that has evolved on its own has done so by growing, rather than by being built one step at a time by some other intelligence. If the proposed semiconductor technology works as this article suggests, we will enter the early phases of Advanced Civilization almost immediately, quickly achieving effective immortality for our machines, if not yet for our physical bodies.
The key is three-dimensionality. We already know that our brains are three-dimensional networks with both nanometer- and micrometer-thin wiring, and that some of the connections extend over centimeters. We also know that, by comparison with today’s electronics, the brain uses very slow switching and consumes correspondingly little power. Finally, we know that our brain is made of quadrillions of components operating in parallel. If you put all these together, it is clear that nanopower 3D electronics (or possibly spintronics) can be scaled to enormous component count, with huge interconnectivity and slow switching speed, thus eluding the problem of excessive power dissipation within each small volume of circuitry.