IT’S a material so good they named a valley after it. And no wonder. Today’s connected society would be impossible without silicon. Chips made from it run everything from smartphones to pacemakers, with some 6.5 million square metres of the stuff rolled out every year. And the solar industry relies on vast quantities of silicon to make the photovoltaic cells that convert light into electricity.
Silicon is in such demand that you’d be forgiven for thinking its position at the top of the pile was untouchable. But its status owes more to the fact that it is the second most abundant element on the planet than to its performance. Crucially, silicon’s atomic structure limits its ability to conduct electricity. And that holds back computer processing speeds and the efficiency of solar panels. If electronic devices are to get faster, cheaper and more compact at the rate we’ve come to expect, silicon as we know it needs to be shown the door.
Advertisement
So the hunt is on for a replacement. Many elements and compounds have been proposed over the years, but it is starting to look like the solution might be closer to home. Ordinary silicon, imbued with certain superpowers, might be able to replace itself.
Silicon belongs to the semiconductor family of materials, whose ability to carry an electric current lies somewhere between that of a metallic conductor and an insulator. In a computer chip, applying a small voltage is enough to flip silicon’s state between conducting and insulating, producing the binary 1s and 0s of digital information. This control over the flow of electrons, combined with its low cost, stability and high availability, has made silicon the material of choice in electronics for over 60 years.
The trouble is, conventional silicon chips are about as good as they are going to get. A top-of-the-range chip today squeezes in around 5 billion transistors – the basic on-off switches that control the flow of electrons. That’s close to the upper limit. Try to pack in many more, and material defects combined with the heat produced by all the transistors switching simultaneously start to adversely affect a chip’s efficiency. This is the main reason why processor speeds have more or less stalled in the past decade. “Electronics has pretty much reached the peak of its performance,” says Lok Lew Yan Voon, a semiconductor physicist at the Citadel, the Military College of South Carolina in Charleston.
Place in the sun
When it comes to solar panels, silicon’s prospects look even dimmer. “Silicon is not very good at absorbing light,” says P. Craig Taylor, director of the Renewable Energy Materials Research Science and Engineering Center at the Colorado School of Mines.
To understand why light presents a problem, it helps to know a little about what makes silicon a semiconducting element in the first place. According to the rules of quantum mechanics, the electrons within a material can’t have just any old energy, but must occupy one of a set of well-defined energy levels. Electrons at lower levels remain bound within their individual atoms. Those at higher levels, meanwhile, are free to move around, allowing them to carry a current through the material.
In a metal such as copper, the atoms’ energy levels overlap, so electrons can move freely all the time. But in a semiconductor such as silicon, the electrons need a shove to lift them from a low to a higher energy level. In a chip, the energy required to bridge this “band gap” and generate a current can be supplied in the form of a voltage; in a photovoltaic cell, the energy comes from a photon of light.
But it needs to be the right sort of light. Some portions of the spectrum, such as infrared light, don’t provide enough energy, while others provide too much. This means that about half of the sunlight that strikes a silicon solar cell is essentially wasted.
And it’s not just about energy. Silicon is what is known as an indirect band-gap semiconductor, meaning its electrons don’t generally have quite the right amount of momentum to make the leap into the higher states unaided, making the transition even less likely to happen. The upshot is that conventional silicon solar cells are hampered by low efficiency (see illustration).
The pretenders to silicon’s solar crown are led by semiconductors that only need one jump to become conductive – those with direct band gaps such as cadmium telluride and gallium arsenide. But these materials have their weaknesses. Many of their constituent elements are rare or expensive, or are toxic heavy metals, such as cadmium and arsenic, that present a threat to the environment if not recycled carefully.
For computer chips, there is a similar lack of convincing replacements. A lot of money has been pumped into graphene, a wonder-material not only stronger and lighter than steel, but capable of transporting electrons across its surface at speeds far greater than in silicon. But graphene is difficult to make in quantity and doesn’t work as a semiconductor.
Expensive, toxic, hard to make: these aren’t exactly properties that recommend a material, especially if you’re going to need vast quantities of it. “For things like solar panels, if you’re going to cover half the state of Arizona, you need something that’s abundant and relatively inexpensive,” says Taylor.
A dream would be to find a way to turn ordinary silicon – non-toxic, readily available and equipped with a huge industrial set-up dedicated to working with it – into something that matches the best traits of these other materials. As it turns out, such a transformation may already be possible.
An element’s properties vary dramatically according to how its individual atoms are arranged. Graphene, for example, is a two-dimensional lattice of carbon. Arranged differently, the same atoms make dazzling diamond or dull, pencil-lead graphite.
Such different forms, or allotropes, of an element are by no means limited to carbon. Under ordinary conditions, silicon atoms adopt an essentially cubic arrangement similar to that of diamond. But up to a dozen alternatives are possible, according to some estimates, each with different and potentially useful properties.
Silicon enhancement
One researcher aiming to see if these silicon dreams can be made reality is Timothy Strobel at the Carnegie Institute of Washington. Last year, he and his team announced they had made a new silicon allotrope that could avoid the band-gap problem with just a squeeze. The discovery came almost by accident: the team had compressed elemental silicon and sodium together to create a shiny, blue-tinged crystal of Na4Si24 and wanted to measure the compound’s resistance to the flow of electrical current.
Getting accurate measurements meant attaching electrodes to the crystal with a glue, which required heating in order to set. “We put it in the oven, measured the resistivity, and we kept getting these crazy results we couldn’t understand,” says Strobel. The data suggested that temperatures as low as 40 °C were enough to bake sodium ions off the structure and change its electrical properties. This was an unexpected result, as similar compounds generally form networks of silicon cages in which the smaller sodium atoms rattle around, unable to escape even at high temperatures. But instead of cages, Na4Si24 forms corridors, allowing sodium ions to slide out easily as the heat rises (see illustration). Heating to 100 °C brought the sodium levels down to less than one atom in a thousand, making a bona fide allotrope of silicon – Si24.
“The really cool thing about the material we’ve made is that it’s the closest thing to a direct band gap material,” says Strobel. While it is still officially an indirect band gap semiconductor, applying a little physical strain is enough to open up a direct pathway across the band gap, allowing electrons to make the jump with no need for a change in momentum. “You just need to squish it by 2 per cent,” says Strobel. This could theoretically be done by growing Si24 on a template with slightly different dimensions, like a person being squeezed into an ill-fitting suit.
This has the potential to make a much more efficient solar cell, since more energy can go directly to moving electrons successfully. What’s more, Strobel and his colleagues think it will be possible to scale up production of Si24 to industrially useful levels – great news for the rapidly growing photovoltaics industry.
The very best silicon solar cells today convert just 25 per cent of solar energy into power, some way off the widely assumed upper limit for solar cells of 33 per cent. Si24 could bring cells closer to the upper limit, but who says we couldn’t go further still?
The figure of 33 per cent is based on an assumption that each incoming photon of light liberates just one conducting electron. But when quantum effects come into play, some materials can extract enough energy to excite more than one electron at a time. In 2013, of the University of Chicago and her colleagues proposed that nanoparticles of a silicon allotrope known as BC8 would be able to harness this property, converting up to 42 per cent of incoming sunlight into electrical energy. Strobel and others are now working to test this – although Galli cautions that they are exploring largely unknown territory as yet.
But it’s not all Si in the sky. In the world of computer chips, another silicon allotrope has been making breakthroughs. In February, Deji Akinwande and his colleagues at the University of Texas at Austin announced that they had built the first functioning transistor made from one of silicon’s most exotic forms – silicene.
Like graphene, silicene owes many of its desirable properties to being a single, two-dimensional layer of the element. But whereas graphene is entirely smooth, silicene’s structure buckles as its larger silicon atoms struggle to squeeze into the same regular arrangement. In both cases, the generally flat, honeycomb structure results in extra unbound electrons that hover above the surface, allowing them to travel faster than in the cubic lattice of ordinary silicon. “I’m not talking about two or three times faster,” says Lew Yan Voon, who named silicene and predicted its properties back in 2007. “I’m talking about a million times faster.”
“I don’t mean two or three times faster. I’m talking about a million times faster”
Travelling in these fast lanes across the surface would also lead to fewer collisions for the speeding electrons, significantly reducing the amount of heat that a densely packed chip would produce. And as transistors built from silicene can be made much thinner than those in existing chips, many more could be squeezed into the same space.
The device Akinwande’s team built is just a proof of principle, showing only modestly improved speeds. And there are major hurdles to overcome before we’ll find anything similar in our smartphones. The same two-dimensionality that should make silicene faster also means it readily falls apart; the transistor lasted just a few minutes. Nor is the material easy to make: synthesising it requires a high vacuum set-up and specialised expertise. “It’s more like an art,” says Lew Yan Voon. “Some people can do it without thinking, and some are still struggling to make it.”
Still, it’s a start. “Up until last year, people thought it wasn’t possible to make a device from silicene because of its instability,” Akinwande says. “So we were very excited and very lucky to get results.” For all the caveats, Lew Yan Voon agrees. “The fact that you can actually make a device with silicene is a big breakthrough”, he says.
Direct band-gap allotropes such as BC8 and Si24 could also have a part to play in the future of electronics, possibly enabling the integration of optical and electronic components onto a single chip. That is the dream scenario, Strobel says. Such hybrid chips could transmit signals using light as well as electrons, greatly increasing their speed as well as the amount of data they can carry.
It is too early, though, to know how these new forms of silicon will shape up. We have been here before, after all. A decade ago, there was great excitement over the potential of quantum dots, tiny crystals of regular silicon that harnessed quantum mechanical effects to upgrade their light absorption. But no one could get them to generate a current effectively and they soon fell out of favour.
This time it’s different, Taylor thinks. A few months after Akinwande’s success, he organised a conference on exotic forms of silicon to discuss advances in the field. “The feeling within the community, and certainly at our conference, was that the time was right to really push ahead,” he says.
If structures like silicene, BC8 and Si24 prove viable for solar cells and chips, they will tie together the promise of exotic materials with the safety, low cost and dependability of silicon. Swapping one form of the element for another might not sound like a big change. But this simple casting off of a disguise, like Clark Kent taking off his glasses, might superpower the world. And the valley wouldn’t even have to change its name.
Illustration: Spencer Wilson, Image: Justin Mott/Redux/Eyevine
This article appeared in print under the headline “Super Silicon Bros”
