Flat lens gives perfect image

August 26, 2012 11:57 AM

Flat lens gives perfect image by causing instantaneous phase shift at surface rather than relying on phase shift of regular lens happening by traveling through different thickness of material (like glass) causing different phase shifts to focus light.
posted by aleph (29 comments total)

Cool! So it's like a fresnel lens, but shaping the wavefront by using interference as it passes through a flat surface?
posted by Joe in Australia at 12:01 PM on August 26, 2012

That's absolutely amazing. It's something that has the potential to completely transform all optics, imaging, and everything in between. Wow. We can bend and focus light with a completely flat piece of glass. That's nuts.
posted by disillusioned at 12:02 PM on August 26, 2012 [7 favorites]

Neat!
posted by Sticherbeast at 12:04 PM on August 26, 2012

This sounds like something that would have helped with either Hubble's main mirror, or during the Tholian incident.
posted by dhartung at 12:07 PM on August 26, 2012 [4 favorites]

You could make a single pane of glass that was both the lens and the mask (and perhaps the light source) for a projector! Projectors everywhere! Embed them in glasses and you'd have something like 3D goggles, but lighter and much less visually offensive!

Or actually, make regular optical glasses that are both thin and powerful, so people don't look as though they're peering through Coke bottles!
posted by Joe in Australia at 12:08 PM on August 26, 2012

I was about to comment how 'Tholian incident' sounds even more like science fiction than this technology itself, but I notice there's a good reason for that.
posted by Anything at 12:10 PM on August 26, 2012 [4 favorites]

Too bad they call it a meta-suface. Since this thing is modifying the light as it passes through it, it wouldn't have been too much of a stretch for them to call it a... well... you know...
posted by Hicksu at 12:22 PM on August 26, 2012 [37 favorites]

I'm not sure if I'm more surprised by the technology or that it's from Harvard...or that Harvard a school of engineering.
posted by zippy at 12:24 PM on August 26, 2012

I think this effect is going to be really chromatic, though -- in other words, only useful for imaging with monochromatic illumination (although it would certainly be possible to fabricate flat lenses that would individually work well at any particular wavelength).

Of course, you don't much care about chromaticity for photolithography. I suspect that may drive the interest here. I'll read the Nanoletters article at work tomorrow.
posted by samofidelis at 12:27 PM on August 26, 2012 [1 favorite]

"Near infrared to teraHz" doesn't include visible light, sadly.
posted by seanmpuckett at 12:32 PM on August 26, 2012

The flat lens eliminates optical aberrations such as the “fish-eye” effect that results from conventional wide-angle lenses.

This isn't correct. A projection from 3D to 2D cannot preserve all of the qualities of the 3D environment. Distortion from lens imperfection is fundamentally different from projection distortion. A fisheye image is no more distorted than a rectilinear wide angle shot - it's just distorted in a different way. Distortion is impossible to avoid when projecting onto a 2D surface, no matter how good your lens is.
posted by jimmythefish at 12:41 PM on August 26, 2012 [3 favorites]

This is the same field of tech that's been creating all those "invisibility cloak" headlines over the past few years. As I understand it (which is poorly), it's all about using nanoscale periodic structures on the surface of materials to direct and manipulate electromagnetic waves. I think samofidelis is right—generally these materials have to be built to work within a narrow range of wavelengths, though tunable metamaterials are being investigated.
posted by dephlogisticated at 12:59 PM on August 26, 2012 [1 favorite]

Yeah, the infrared and "can be tuned for specific wavelengths of light" means this method might not be applicable to conventional photography any time soon. But cool for scientific instrumentation at least. NASA will likely love a lens that's a tiny fraction of the normal weight.

"The flat lens eliminates optical aberrations such as the “fish-eye” effect that results from conventional wide-angle lenses."
I don't get how that would work. Fish-eye is really caused by the fact that we normally capture and display images on flat surfaces. A new kind of lens can't get around the mathematical limitations with projecting an image onto a flat surface, can it?
posted by RobotHero at 1:02 PM on August 26, 2012 [1 favorite]

(Or what Jimmythefish said.)
posted by RobotHero at 1:03 PM on August 26, 2012

Question from the nano-science challenged: does this lens only function as a normal lens? Is it possible to create, using this technology, flat telephoto or wide-angle lenses?
posted by Hollywood Upstairs Medical College at 1:04 PM on August 26, 2012

As mentioned above, the problem with this sort of thing is that its intrinsically narrow band, meaning it can only be "perfect" at a single frequency. A wideband response requires time delay, rather than just phase shift, and moreover, the antennas they are using will not have a constant response over frequency. They mention the ability to re-tune it for different wavelengths, but that doesn't work if you have a broadband signal. It could be that in the THz regime, that the operational bandwidths needed are negligibly small (I work at lower microwave frequencies, so I'm not sure), but I don't see anything in their paper which even addresses the bandwidth of the lens.
posted by jpdoane at 1:06 PM on August 26, 2012 [2 favorites]

It would be possible to design a lens of any focal length simply by adjusting the phase delay profile accordingly. One could even produce non traditional focal patterns useful for non point source emitters or sensors, or to correct distortion, or tuned for certain emission patterns.
posted by seanmpuckett at 1:11 PM on August 26, 2012 [1 favorite]

The full article is behind a paywall, and the link doesn't give the size lens they made, but a nanometer layer of gold, with tiny strips peeled away, is probably a fraction of an inch in diameter. Practical eyeglasses, let along a Hubble-sized lens, will present large manufacturing problems.
posted by KRS at 1:15 PM on August 26, 2012

The full article is behind a paywall, and the link doesn't give the size lens they made, but a nanometer layer of gold, with tiny strips peeled away, is probably a fraction of an inch in diameter. Practical eyeglasses, let along a Hubble-sized lens, will present large manufacturing problems.

I have access to the paper. The lens they built is 0.9mm in diameter. But again, this would not be used for those type of applications, but according to the paper could be used in:

"....telescopes, autocollimators, microscopes, laser surgery, and optical trapping"

Also, they don't mention this in the PR piece on the Harvard site, but the efficiency is currently only 1%.
posted by jpdoane at 1:47 PM on August 26, 2012 [3 favorites]

I don't think paywalls are an issue for the article - is it not this one?

http://arxiv.org/abs/1207.2194

Though this article says they are focusing ~10% of the incoming light beam as opposed to the 1% jpdoane cites, so maybe it's different. . . still a lot less than you'd like for most purposes.
posted by janewman at 2:31 PM on August 26, 2012

Will this mean faster internet with fiber-optic, or a higher quality signal, or will it allow fiber-optics to carry more information?

Also the talk of a fish-eye distortion makes it sound like this lense will work in the visible spectrum, but the range given doesn't include visible light, and I right?


This is heady stuff...really really amazing technology.
posted by Skygazer at 3:22 PM on August 26, 2012

I don't think the focusing optics are the limiting factor in fiber optic data rates. I guess it has the potential to reduce prices at some point in the future.
posted by ryanrs at 3:35 PM on August 26, 2012

"... by adjusting the phase delay profile accordingly. One could even produce non traditional focal patterns..."

Though using a different technique (ordinary refraction), phase adjustments to a beam of light are capable of creating apparently curved beams of light.
posted by Phyllis Harmonic at 3:38 PM on August 26, 2012

Also, they don't mention this in the PR piece on the Harvard site, but the efficiency is currently only 1%.

Yeah, all the metamaterial stuff has terrible throughput. I built a Vis-NIR polarization interferometer to test negative index of refraction metamaterials -- the efficiency is garbage, and to do all the Penrose superlens stuff requires lossless materials. I got out of the field, but I understand that the new jam in metamaterials is working to build devices that have gain -- eg embedding quantum dots in a bulk metamaterial.

By the way, 'bulk' is a terribly abused term in this field. Other than some holographic techniques, you seem to be looking at largely photolithographically assembled media, and I've seen a pretty sloppy argument used to claim that the observed negative index effects can be attributed to bulk properties, rather than some quirky thin film thing.
posted by samofidelis at 7:29 PM on August 26, 2012 [1 favorite]

Parallel write ebeam lithography is going to take over high resolution stuff anyway. No more serial write bottlenecks.

I was reading about a sub wavelength resonate cavity with metamaterial coatings on the boundaries...very cool.
posted by Chekhovian at 8:20 PM on August 26, 2012

Yeah, it might be interesting to look at all those cavity QED experiments and ask yourself if there were any effects you could observe by playing some kind of game with the cavity density of states. Of course, once again you'd need lossless metamaterials to even think about it. So, not going to happen.
posted by samofidelis at 8:58 PM on August 26, 2012

I'm more interested low frequency maybe 100 Mhz and below type stuff. 3 m cavities aren't as practical as 3 cm ones.
posted by Chekhovian at 9:33 PM on August 26, 2012

I'm more interested in sub-lambda at visible; it's hard to get anything other than classical wave dynamics out of Rf systems, right? The only RF engineering I know is what I needed to build drive circuitry for a bunch of acousto-optic modulators with approx. 100 MHz central frequency. I've barely ever messed with GHz stuff, and it's a hole in my education for sure.
posted by samofidelis at 9:38 PM on August 26, 2012

RF from Mhz to Ghz is basically always a pain in the ass. There are folks these days building muwave resonators and getting them into the single photon limit. Usually that's for qbit stuff so far.
posted by Chekhovian at 9:52 PM on August 26, 2012