NEWS FEATURE NATURE|Vol 459|28 May 2009 An eye for detail: John Pendry. IDEAL M. FINN-KELCEY/IMPERIAL COLLEGE M. FINN-KELCEY/IMPERIAL Before they were touted as invisibility cloaks, metamaterials promised a perfect lens. Geoff Brumfiel reports on the struggle for superior vision. s far as John Pendry is concerned, we are as good as blind. Sitting in his office at Imperial College London, the Atheoretical physicist gestures at the table in front of him. Assume, he says, that the fundamental limit of detail in the table is about the size of an atom. Then consider that the smallest feature the human eye can see, even using a high-quality optical microscope, is a few tenths of a micrometre across — roughly a thousand times bigger than an atom. That means that, even with 20–20 vision and the best optics, humans can access only 0.0001% of alluring idea yet proposed for metamaterials. “I Pendry recognized a loophole: the diffraction the information right before their eyes. “That’s think that the superlens will probably find more limit does not necessarily hold for metamateri- not much,” he observes. applications [than cloaking],” says Xiang Zhang, als. Although silver is not a perfect metamate- Nearly a decade ago, Pendry proposed a based at the University of California, Berkeley, rial, he showed that it would work well enough way to do thousands of times better using a who grabbed headlines in April for using meta- as a lens. If a silver film were placed close film of silver just tens of nanometres thick as a materials to create a primitive cloak2. enough to an illuminated object, it could catch ‘superlens’1. The film would behave as a simple The key to all this comes down to the struc- the ‘evanescent waves’ — short-range electro- metamaterial, a substance with a small-scale tures of metamaterials, which directly affect magnetic fields that carry the detailed, subdif- structure that allows it to manipulate light in their optical properties. The composition of fraction-scale information about the object, a way that no bulk composition could. This conventional substances, such as glass, is uni- but that cannot be picked up beyond a few tens means that the film could capture details that form at optical wavelengths, but meta materials of nanometres from the object’s surface. These elude conventional optical microscopes. In feature regularly patterned structures at those evanescent waves would be amplified by reso- theory, the superlens could also scales. When light of a certain nances in the silver film, and their information etch nano metre-scale patterns “It’s a very, very wavelength interacts within the would be carried through the film to create an on to a surface, which would simple technology. pattern, it sets up a resonance, exquisitely detailed image on the far side. make it very useful in the man- similar to the way a musical tone ufacture of microchips. It’s just a question triggers vibrations in a tuning Window of opportunity Pendry’s idea brought many of getting it right.” fork. The resonance can cause Richard Blaikie, an electrical engineer at the researchers into the field, and — John Pendry light beams to be deflected in University of Canterbury in Christchurch, New for a while expectations were the ‘wrong’ direction and it can Zealand, was excited by this concept — not as high. But nine years on, those expectations also enhance certain properties. a way to see, but as a way to write. He thought have yet to be fully realized. Although there Pendry’s initial work on metamaterials that the superlens could help chip-makers in have been proof-of-concept devices for both was aimed at beating a problem known as the their struggle to create subdiffraction-scale the superlens and the related ‘hyperlens’, indus- diffraction limit, which says that optical instru- circuit elements. But, as Blaikie soon learned, try has switched its focus elsewhere. And the ments such as telescopes and microscopes are what looked elegant on paper was messy in the field of metamaterials has itself been diverted subject to a fundamental fuzziness. No matter lab. “It’s very easy to do some modelling,” he by another high-profile potential application: how well they are made, their ability to resolve says, “but the experiments are very hard.” cloaking devices. fine detail is constrained by the wavelength of Blaikie did manage to make a prototype Making objects ‘invisible’ may have stolen the the light being used. For optical microscopes, superlens in 2005 that captured subdiffraction limelight, but one of the researchers behind that this means that any attempt to resolve features information3. But because the evanescent waves work believes that superlenses remain the most below around 200 nanometres will fail. fade away over such a short distance, the silver 504 © 2009 Macmillan Publishers Limited. All rights reserved 5504-50504-505 NewsNews FeatFeat SSuperlenses.indduperlenses.indd 504504 226/5/096/5/09 111:25:161:25:16 VolNATURE 459|28|Vol May 459 2009|28 May 2009 NEWS FEATURE film had to be fixed to the object being imaged. Superlenses do work in theory, he contends, and Similarly, whatever recorded the image had to many of the problems are “more of an engineer- be clamped tightly to the other side of the film. ing issue”. For his part, he believes that the field And even then, says Blaikie, “the images that may have to draw a little more from industrial (2005) 534–537 308, we got, although good in terms of their resolu- expertise. For example, he has used industrial- tion, were poor in terms of their fidelity”. Even grade germanium to deposit atomically flat a few nanometres of variation in the silver’s films of silver7 that can reduce the distortion surface created hotspots that made straight seen by Blaikie. ET AL. SCIENCE lines appear jagged. Blaikie says his group has Other groups are working to make hyper- struggled to follow up that 2005 demonstra- lenses more practical and, again, techniques N. FANG N. FANG tion. “You haven’t seen anything because we Superlens in action: a thin film of silver delivers from the semiconductor industry seem to haven’t had any success,” he says bluntly. a more detailed image (bottom) of nanometre- be helping. In April, Stefan Mendach and his scale lines. colleagues at the University of Hamburg, Ger- Hyper activity many, created a hyperlens by rolling up alter- In 2006, partly in reaction to the practical limi- nano structures required for the job. As a result, nating layers of semiconductor materials8. The tations of superlenses, two theoretical groups some scientists believe industry needs to start technique seems to be an easy way to make a independently developed the hyperlens4,5. investing in super- and hyperlenses to give that hyperlens, although Mendach needs to get Instead of a single film, the hyperlens features technology a boost. But for both imaging and more layers in his system before it can work. alternating layers of a metal and an insulator. circuit etching, the lenses would be entering And Vladimir Shalaev at Purdue University Experimentally, the layers are arranged to form markets in which competing subdiffraction in West Lafayette, Indiana, is trying to build a half-cylinder (see diagram). The object is technologies are already further along in devel- ‘flat’ hyperlenses. Although they are harder to placed in the centre of the cylinder and evanes- opment. Semiconductor researchers can use make — the individual layers can no longer be cent waves from it are caught nonlinear optics to write circuits uniform — they could prove easier to use, as by the lens and magnified as “You haven’t seen below the diffraction limit. And the specimen would not have to be precisely they pass through the various anything because for biological imaging, other positioned along the cylinder axis. layers, emerging as light that we haven’t had any groups are developing ways to Zhang, meanwhile, is pursuing a simpler but shines freely. The hyperlens is beat the diffraction limit using a related concept that he believes could revolu- more complex than the super- success.” combination of fluorescent pro- tionize data storage. Rather than using a single lens, but has a major advantage — Richard Blaikie teins and clever optics. superlens, he is using several ‘plasmonic’ lenses, in that it should be able to feed Biologists have shown only which focus evanescent waves. When normal the light emerging from its surface into con- tepid interest in superlenses, admits Nicholas light strikes the lens it interacts with electrons ventional optics. “I’m hopeful that this hyper- Fang, a researcher at the University of Illinois at on the lens surface. This concentrates evanes- lens could be a revolutionary instrument,” says Urbana-Champaign. Still, Fang and many oth- cent waves into a single point, effectively creat- Nader Engheta, a theorist at the University of ers remain fiercely devoted to the superlens idea. ing a subdiffraction hotspot that could be used Pennsylvania in Philadelphia and one of the If the lenses can be made to work, they would be to write a piece of information on to an optical originators of the idea. able to image living biological specimens with disk. Zhang has developed a prototype array of Again, experimentalists rushed into the lab unprecedented detail. Granted, he says, “there plasmonic lenses that does indeed write infor- to build hyperlenses, and again they met with are many open issues that we did not realize mation far below the diffraction limit. initial success. Less than a year after the idea at the very beginning”. But he points out that Back at his desk, Pendry is willing to wait for was proposed, two groups had made proof-of- this is hardly unusual with new technologies.