Plasmons: a review

Here is a review on Plasmon and some references can be found to dig it up.

IT'S a laser, but not as we know it. For a start, you need a microscope to see it. Gleaming eerily green, it is a single spherical particle just a few tens of nanometres across.

Tiny it might be, but its creators have big plans for it. With further advances, it could help to fulfil a long-held dream: to build a super-fast computer that computes with light.

Dubbed a "spaser", this minuscule lasing object is the latest by-product of a buzzing field known as nanoplasmonics. Just as microelectronics exploits the behaviour of electrons in metals and semiconductors on micrometre scales, so nanoplasmonics is concerned with the nanoscale comings and goings of entities known as plasmons that lurk on and below the surfaces of metals.

To envisage what as plasmon is, imagine a metal as a great sea of freely moving electrons. When light of the right frequency strikes the surface of the metal, it can set up a wavelike oscillation in this electron sea, just as the wind whips up waves on the ocean. These collective electron waves - plasmons - act to all intents and purposes as light waves trapped in the metal's surface. Their wavelengths depend on the metal, but are generally measured in nanometres. Their frequencies span the terahertz range - equivalent to the frequency range of light from the ultraviolet right through the visible to the infrared.

Gleaming eerily green, this laser is a single spherical particle just tens of nanometres across

In 2003, their studies of plasmons led theorists Mark Stockman at Georgia State University in Atlanta and David Bergman at Tel Aviv University in Israel to an unusual thought. Plasmons behaved rather like light, so could they be amplified like light, too? What the duo had in mind was a laser-like device that multiplied single plasmons to turn them into powerful arrays of plasmons all oscillating in the same way (see "From laser to spaser").

The mathematics of it seemed to work. By analogy with the acronym that produces the word laser, they dubbed their brainchild "surface plasmon amplification by the stimulated emission of radiation" - spaser - and published a paper about it (Physical Review Letters, vol 90, p 027402).

The spaser might have remained just a theoretical curiosity. Around the same time, however, physicists were waking up to the potential of plasmonics for everything from perfect lenses to sensitive biosensors (see "What have plasmons ever done for us?"). The spaser idea was intriguing enough that Mikhail Noginov, an electrical engineer at Norfolk State University in Virginia, and some of his colleagues set out to build one.

It was not an easy task. Light is long-lived, so it is relatively easy to bounce it around in a mirrored chamber and amplify it, as happens inside a laser. Plasmons, by contrast, are transient entities: they typically live for mere attoseconds, and cannot travel more than a few plasmon wavelengths in a metal before their energy is absorbed by the ocean of non-oscillating electrons around them. It was not at all clear how we might get enough of a handle on plasmons to amplify them at all.

In August 2009, Noginov and his colleagues showed how. Their ingenious solution takes the form of a circular particle just 44 nanometres across. It consists of a gold core contained within a shell of silica, speckled with dye molecules that, excited initially by an external laser, produce green light. Some of that light leaks out to give the nanoparticles their characteristic green glow; the rest stimulates the generation of plasmons at the surface of the gold core.

In the normal way of things, these plasmons are absorbed by the metal almost as soon as they are produced. But their tickling influence also stimulates the dye molecules in the silica shell to emit more light, which in turn generates more plasmons, which excites more light and so on. With a sufficient supply of dye, enough plasmons can exist at the same time that they start to reinforce each other. The signature of a laser-like multiplication of plasmons within the device is a dramatic increase in green laser light emitted from the nanoparticle after only a small increase in the energy supplied from the external laser - the signature Noginov and his colleagues reported last year (Nature, vol 460, p 1110).

And they were not the only ones. In October 2009, Xiang Zhang, a mechanical engineer at the University of California, Berkeley, and his colleagues unveiled a similarly tiny device that exploits plasmons to produce laser light (Nature, vol 461, p 629).

These innovations generated headlines at the time as an entirely new type of lasing device more compact than any yet seen and which, in theory, required a lot less power than a conventional device. That's an exciting development in its own right, but just one in a list of promising advances in the bustling business of laser technology.

Crucially, though, the development of spasers has sparked the hope that one of the great scientific disappointments of the past decades - the unfulfilled promise of optical computing - may yet be turned into triumph.

On the face of it, optical computers, which use light rather than currents of electrons to process information, are a great idea. Electrons are easy to manipulate and process, but they tend to get bogged down as they pass through metals and semiconductors, colliding with atoms and bouncing off them in ways that limit the speed and fidelity of information transmission. Photons, by contrast, can withstand interference, and are above all fast, in theory zipping around a chip at close to the cosmic speed limit.

In the 1990s, various groups claimed to be getting close to making the dream of optical computing a reality. That included a concerted effort at the world-famous Bell Laboratories in Murray Hill, New Jersey, where the building block of microelectronic circuits, the transistor, was invented in 1947. Researchers there and elsewhere hit a snag, however. The very fleet-footedness that made photons perfect for high-speed communications made them almost impossible to pin down and use for sensible processing of data.

"Optical computing has a chequered history, particularly the boondoggle at Bell Labs," says Harry Atwater, a physicist at the California Institute of Technology in Pasadena. All the efforts foundered when it came to producing anything like a transistor: a tiny, low-power device that could be used to toggle light signals on and off reliably.

In theory, a controllable laser would do this trick, if not for one problem - lasers devour power. Even worse, they are huge, relatively speaking: they work by bouncing photons around a mirrored cavity, so the very smallest they can be is about half the wavelength of the light they produce. For green light, with a wavelength of 530 nanometres, that means little change from 300 nanometres. Electrical transistors, meanwhile, are approaching one-tenth that size.

You see where this is leading. Spasers are a tiny source of light that can be switched on and off at will. At a few tens of nanometres in size, they are just slightly bigger than the smallest electrical transistors. The spaser is to nanoplasmonics what the transistor is to microelectronics, says Stockman: it is the building block that should make optical information-processing possible.

The spaser is to plasmonics what the transistor is to microelectronics

Inevitably, there will be many hurdles to overcome. For a start, Noginov's prototype spaser is switched on and off using another laser, rather than being switched electrically. That is cumbersome and means it cannot capitalise on the technology's low-power potential. It is also unclear, when it comes to connecting many spasers together to make a logic gate, how input and output signals can be cleanly separated with the resonant spherical spasers that have so far been constructed.

Mutual benefit

The most intriguing aspect of spasers, however, is the one that could make or break them as the basis of a future computing technology: they are made of metal. In one sense, that is a bad thing, because making a plasmonic chip would require a wholly different infrastructure to that used to make silicon chips - an industry into which billions in research money has been poured.

Silicon's predominance has not necessarily been a bar to other technologies establishing themselves: the radio signals used for cellphone communication, for example, are of a frequency too high for silicon chips to cope with, so an entirely separate manufacturing process grew up to make the gallium arsenide chips that can. To justify the initial investment costs, another upstart chip-architecture needs a similar "killer application": something it can do that silicon cannot.

Stockman reckons the extra processing speed promised by plasmonic devices will generate such applications in areas like cryptography. "Having faster processors than everyone else will be a question of national security," he says. And he points to another reason why the spooks might be interested. One problem with semiconductors is that their delicate conduction capabilities are vulnerable to ionising radiation. Such rays can send avalanches of electrons streaming through delicate electronic components. At best, this corrupts data and halts calculations. At worst, it fries transistors, permanently disabling them.

This is where the metallic nature of a plasmonic chip would come into its own. The extra electrons that ionising radiation can produce are mere drops in the ocean of free electrons from which plasmons are generated in a metal. A plasmonic device would be able to process and store information in the harshest radioactive environments: in orbiting satellites, in nuclear reactors, during nuclear conflict.

Perhaps the most likely outcome, though, is that rather than the one superseding the other, plasmonics and electronics come to coexist to mutual advantage in a single chip. As the transistors in chips become smaller, the wires that connect them over distances of just a few nanometres become a significant bottleneck for data. That is one reason why chips are currently spinning their wheels at speeds of about 3 gigahertz. "Wires limit the speed at which electrons can deliver information," says Atwater. "So an obvious solution is to replace them with photonic connections."

The problem with such connections to date has been converting electronic signals into photonic ones and back again with a speed and efficiency that makes it worthwhile. Plasmons, which owe their existence to the easy exchange of energy between light and electrons, could be just the things for the job, making a hybrid electrical-optical chip a genuine possibility.

As well as that, says Atwater, we should work out how to manipulate plasmons using devices that can be made in the same way, and on the same fabrication lines, as ordinary silicon chips. Early last year, he and his colleagues at Caltech revealed an electrically controlled device dubbed the plasmostor that can vary the intensity of plasmons as they pass through it, and which has an architecture very similar to that of conventional transistors (Nano Letters, vol 9, p 897). Just this month, a Dutch group has announced that they have produced an electrically powered source of plasmons fully compatible with existing silicon chip fabrication technology (Nature Materials, vol 9, p 21).

It's very early days, so such innovations have yet to match the performance of purely electronic components. The plasmostor, for instance, flips between its on and off states more slowly than a conventional transistor, and the signals have an annoying tendency to leak out of the device and get lost. There is still a long way to go to a computer that runs on anything other than electrons. But it is a start, says Atwater. "You're challenging a hugely successful technology. It's audacious to think that you can just replace it."

But if a tiny round green light isn't a signal to go ahead and give it a try, what is?

From laser to spaser

This year marks the golden jubilee of a ruby trailblazer: it was on 16 May 1960 that Theodore Maiman of Hughes Research Laboratories in Malibu, California, coaxed a synthetic ruby to produce the first ever laser light. The first laser to produce light from gas - a mixture of helium and neon - followed later that same year.

Half a century later, and there's hardly an area of human endeavour that doesn't depend on lasers in some way or another: CD and DVD players, metal cutting and welding, barcode scanners and corrective eye surgery to name but a few.

Early lasers were essentially made up of a mirrored box containing a "gain medium" such as a crystal or gas. Zapped with light or an electric current, electrons in this medium absorb energy, releasing it again as photons. These photons bounce around the box and stimulate further electrons to emit more photons. This self-reinforcing increase in light energy is "light amplification by the stimulated emission of radiation" - laser action, for short.

Spasers use the same principle, except rather than amplifying light directly, they amplify surface plasmons - the wavelike movements of free electrons on and near the surfaces of metals - using that in turn to emit light.

What have plasmons ever done for us?

Plasmons might sound esoteric, but it is not just with spasers (see main story) that they are seeing practical application.

Take molecular sensing. The amount and colour of light absorbed by a plasmonic nanoparticle is extremely sensitive to the surrounding molecular environment. This property has been exploited to build sensing devices that detect levels of anything from the protein casein, an indicator of the quality of milk products, to glucose in the blood.

What's significant about these plasmonic sensors is that they can make continuous measurements, unlike chemical tests which usually give a single snapshot. A plasmonic implant could one day help diabetics to monitor and control their blood glucose levels in real time.

Plasmons should also be useful for increasing the efficiency of certain kinds of flat-screen displays. In June 2009, Ki Youl Yang and his colleagues at the Korea Advanced Institute of Science and Technology in Daejeon showed how silver nanoparticles deposited onto organic light-emitting diodes used in some displays increases the amount of light they emit.

More impressive yet, plasmonic devices might also help to tackle cancer, if tests in mice are anything to go by. Plasmonic nanoparticles laced with antibodies can be made to latch onto tumours. When blasted with a focused beam of infrared light precisely tuned to the plasmon frequency, the nanoparticles heat up, killing the attached cancer cells while leaving the surrounding healthy tissue unharmed (Accounts of Chemical Research, vol 41, p 1842).