Stanford Researchers Make Photonic Components Faster, With Algorithmic Design

retroworks writes: Integrated photonic devices are poised to play a key role in a wide variety of applications, ranging from optical interconnects and sensors to quantum computing. However, only a small library of semi-analytically designed devices is currently known. In an article in Nature Photonics, researchers demonstrate the use of an inverse design method that explores the full design space of fabricable devices and allows them to design devices with previously unattainable functionality, higher performance and robustness, and smaller footprints than conventional devices. The designed a silicon wavelength demultiplexer splits 1,300nm and 1,550nm light from an input waveguide into two output waveguides, and the team has fabricated and characterized several devices. The devices display low insertion loss (2dB), low crosstalk (100nm). The device footprint is 2.8×2.8m2, making this the smallest dielectric wavelength splitter.

Editorial incompetence strikes again!

[aardvarkjoe]

The device footprint is 2.8x2.8m2

The actual measurement, from TFA [nature.com], is 2.8x2.8 square micrometers.

Apparently timothy is too busy burying unflattering stories about his employer [slashdot.org] to bother reading what he's posting to the front page.

Re:

[Whiteox]

Ummm... maybe posting special characters in Slashdot is disabled?

Re:

[rossdee]

"The actual measurement, from TFA [nature.com], is 2.8x2.8 square micrometers"

That doesn't make sense unless its 4 dimensional.

I think you mean 2,8 micrometres by 2.8 micrometres (which is 7.84 square micrometres)

Re:

[radarskiy]

The abstract litteraly says 2.8x2.8 square microns.

Yes, that makes no sense. No, aardvarkjoe is not incorrect since the intention was to quote from the paper.

I have access to the full text, and the body also says 2.8x2.8 square microns. However, looking at the microphotograph of the device it looks like what they meant was 2.8x2.8 microns square, i.e. a square that is 2.8 microns on each side.

Re:

[NoNonAlphaCharsHere]

2.8 x 2.8 m^2 is ~84 ft^2, which would be awesome for a solar cell.

Re:

[antiperimetaparalogo]

The device footprint is 2.8x2.8m2

The actual measurement, from TFA [nature.com], is 2.8x2.8 square micrometers.

Yes, it is square MICRO-meters [wikipedia.org] - (1 micro-meter = 1/1000000 of a meter; ~0.00004 of an inch, for our friends in USA who... but, don't get me started... just adopt the damn metric, you... you...!)

Apparently timothy is too busy burying unflattering stories about his employer [slashdot.org] to bother reading what he's posting to the front page.

Is it true "timothy"? Well, shame on you - not so much for failing to understand the metric system (or is it just because this site is not able yet to display unicode? The micrometer use the Greek "m" before the Latin "m"...), but because you bury a story with NEWS FOR NERDS, THINGS THAT MATTERS...

Re:

[Mr Z]

The Greek mu was probably there when it was copy/pasted. Slashdot silently eats characters outside the English alphabet though.

Re:

[Mr Z]

Ok, it allows a few, but it definitely strips mu. I tried it. Here's 15 mu characters:

Re:

[olsmeister]

Things like this are used a lot in telecommunications, they allow multiple wavelengths of light to travel over a single fiber, cutting down on the amount of glass needed and increasing the bandwidth.

Photonics?

[Livius]

Isn't that the technology behind the holodeck?

Re:

[ArcadeMan]

Every time I read "photonics" I think of the doctor [pinimg.com].

comment error

[beanfeast]

The summary seems to have been screwed up by some sort of comment glitch. It should read:

"low crosstalk (<11dB) and wide bandwidths (>100nm)"

Re:

[Mr Z]

The submitter probably forgot to use the HTML escapes for the less than and greater than (ie. &lt; and &gt;), and those got eaten (along with the mu character before m in micrometers.)

Quantum ?

[xafan]

GoogleTechTalks had a thing on optimizing quantum computers with error correction algorithms. I think it might be related, but then again: quantum? https://www.youtube.com/watch?... [youtube.com]

Algorithmic Design...

[johnw]

...but is it Heuristically Programmed?

Nanophotonic WG

[WhoBeDaPlaya]

The waveguide has a very small cross-section. Wonder how they coupled fiber to the ports. End-fire coupling directly to a fiber would be horrendously inefficient, since minimizing coupling loss requires both a good spatial overlap of the mode profiles and a good match of the effective refractive indices in the two waveguides.

"semi-analytically designed"

[pigiron]

Exactly WTF is that supposed to mean?