Graphene optical modulators could lead to ultrafast communications

May 08, 2011 By Sarah Yang
This is a schematic illustration of the graphene-based optical modulator. A layer of graphene (black fishnet) is placed on top of a silicon waveguide (blue), which is used as an optical fiber to guide light. Electrical signals are sent in from the side of the graphene to alter the amount of photons the graphene absorbs. Credit: Graphic by Ming Liu, UC Berkeley

(PhysOrg.com) -- Scientists at the University of California, Berkeley, have demonstrated a new technology for graphene that could break the current speed limits in digital communications.

The team of researchers, led by UC Berkeley engineering professor Xiang Zhang, built a tiny optical device that uses , a one-atom-thick layer of crystallized carbon, to switch light on and off. This switching ability is the fundamental characteristic of a network modulator, which controls the speed at which data packets are transmitted. The faster the data pulses are sent out, the greater the volume of information that can be sent. Graphene-based modulators could soon allow consumers to stream full-length, high-definition, 3-D movies onto a smartphone in a matter of seconds, the researchers said.

"This is the world's smallest optical modulator, and the modulator in data communications is the heart of speed control," said Zhang, who directs a National Science Foundation (NSF) Nanoscale Science and Engineering Center at UC Berkeley. "Graphene enables us to make modulators that are incredibly compact and that potentially perform at speeds up to ten times faster than current technology allows. This new technology will significantly enhance our capabilities in ultrafast and computing."

In this latest work, described in the May 8 advanced online publication of the journal Nature, researchers were able to tune the graphene electrically to absorb light in wavelengths used in data communication. This advance adds yet another advantage to graphene, which has gained a reputation as a wonder material since 2004 when it was first extracted from graphite, the same element in pencil lead. That achievement earned University of Manchester scientists Andre Geim and Konstantin Novoselov the Nobel Prize in Physics last year.

Zhang worked with fellow faculty member Feng Wang, an assistant professor of physics and head of the Ultrafast Nano-Optics Group at UC Berkeley. Both Zhang and Wang are faculty scientists at Lawrence Berkeley National Laboratory's Materials Science Division.

"The impact of this technology will be far-reaching," said Wang. "In addition to high-speed operations, graphene-based modulators could lead to unconventional applications due to graphene's flexibility and ease in integration with different kinds of materials. Graphene can also be used to modulate new frequency ranges, such as mid-infrared light, that are widely used in molecular sensing."

Graphene is the thinnest, strongest crystalline material yet known. It can be stretched like rubber, and it has the added benefit of being an excellent conductor of heat and electricity. This last quality of graphene makes it a particularly attractive material for electronics.

"Graphene is compatible with silicon technology and is very cheap to make," said Ming Liu, post-doctoral researcher in Zhang's lab and co-lead author of the study. "Researchers in Korea last year have already produced 30-inch sheets of it. Moreover, very little graphene is required for use as a modulator. The graphite in a pencil can provide enough graphene to fabricate 1 billion optical modulators."

It is the behavior of photons and electrons in graphene that first caught the attention of the UC Berkeley researchers.

Graphene optical modulators could lead to ultrafast communications
Shown is a scanning electron microscope (SEM) image magnifying the key structures of the graphene-based optical modulator. (Colors were added to enhance the contrast). Gold (Au) and platinum (Pt) electrodes are used to apply electrical charges to the sheet of graphene, shown in blue, placed on top of the silicon (Si) waveguide, shown in red. The voltage can control the graphene's transparency, effectively turning the setup into an optical modulator that can turn light on and off. (Ming Liu image)

The researchers found that the energy of the electrons, referred to as its Fermi level, can be easily altered depending upon the voltage applied to the material. The graphene's Fermi level in turn determines if the light is absorbed or not.

When a sufficient negative voltage is applied, electrons are drawn out of the graphene and are no longer available to absorb photons. The light is "switched on" because the graphene becomes totally transparent as the photons pass through.

Graphene is also transparent at certain positive voltages because, in that situation, the electrons become packed so tightly that they cannot absorb the photons.

The researchers found a sweet spot in the middle where there is just enough voltage applied so the electrons can prevent the photons from passing, effectively switching the light "off."

"If graphene were a hallway, and electrons were people, you could say that, when the hall is empty, there's no one around to stop the photons," said Xiaobo Yin, co-lead author of the Nature paper and a research scientist in Zhang's lab. "In the other extreme, when the hall is too crowded, people can't move and are ineffective in blocking the photons. It's in between these two scenarios that the electrons are allowed to interact with and absorb the photons, and the graphene becomes opaque."

In their experiment, the researchers layered graphene on top of a silicon waveguide to fabricate optical modulators. The researchers were able to achieve a modulation speed of 1 gigahertz, but they noted that the speed could theoretically reach as high as 500 gigahertz for a single modulator.

While components based upon optics have many advantages over those that use electricity, including the ability to carry denser packets of data more quickly, attempts to create optical interconnects that fit neatly onto a computer chip have been hampered by the relatively large amount of space required in photonics.

Light waves are less agile in tight spaces than their electrical counterparts, the researchers noted, so photon-based applications have been primarily confined to large-scale devices, such as fiber optic lines.

"Electrons can easily make an L-shaped turn because the wavelengths in which they operate are small," said Zhang. "Light wavelengths are generally bigger, so they need more space to maneuver. It's like turning a long, stretch limo instead of a motorcycle around a corner. That's why optics require bulky mirrors to control their movements. Scaling down the also makes it faster because the single atomic layer of graphene can significantly reduce the capacitance – the ability to hold an electric charge – which often hinders device speed."

Graphene-based modulators could overcome the space barrier of optical devices, the researchers said. They successfully shrunk a graphene-based optical modulator down to a relatively tiny 25 square microns, a size roughly 400 times smaller than a human hair. The footprint of a typical commercial modulator can be as large as a few square millimeters.

Even at such a small size, graphene packs a punch in bandwidth capability. Graphene can absorb a broad spectrum of light, ranging over thousands of nanometers from ultraviolet to infrared wavelengths. This allows graphene to carry more data than current state-of-the-art modulators, which operate at a bandwidth of up to 10 nanometers, the researchers said.

"Graphene-based modulators not only offer an increase in modulation speed, they can enable greater amounts of data packed into each pulse," said Zhang. "Instead of broadband, we will have 'extremeband.' What we see here and going forward with graphene-based modulators are tremendous improvements, not only in consumer electronics, but in any field that is now limited by data transmission speeds, including bioinformatics and weather forecasting. We hope to see industrial applications of this new device in the next few years."

Explore further: Scientists unveil new technology to better understand small clusters of atoms

Related Stories

Light-speed nanotech: Controlling the nature of graphene

Jan 21, 2009

Researchers at Rensselaer Polytechnic Institute have discovered a new method for controlling the nature of graphene, bringing academia and industry potentially one step closer to realizing the mass production ...

Producing graphene layers using crystallization

Mar 02, 2010

(PhysOrg.com) -- Ever since it's relatively recent discovery, graphene has generated a great deal of interest. Graphene is extracted from graphite in many cases, and consists of a sheet of carbon atoms bound together in a ...

Turning down the noise in graphene

Aug 06, 2010

(PhysOrg.com) -- Graphene is a two-dimensional crystalline sheet of carbon atoms - meaning it is only one atom thick - through which electrons can race at nearly the speed of light - 100 times faster than ...

Recommended for you

Bacterial nanowires: Not what we thought they were

Aug 18, 2014

For the past 10 years, scientists have been fascinated by a type of "electric bacteria" that shoots out long tendrils like electric wires, using them to power themselves and transfer electricity to a variety ...

Towards more efficient solar cells

Aug 13, 2014

A layer of silicon nanocrystals and erbium ions may help solar cells to extract more energy from the ultraviolet (UV, high-energy) part of the solar spectrum. Experimental physicists from the FOM Foundation, ...

User comments : 16

Adjust slider to filter visible comments by rank

Display comments: newest first

Quantum_Conundrum
5 / 5 (7) May 08, 2011
"Graphene enables us to make modulators that are incredibly compact and that potentially perform at speeds up to ten times faster than current technology allows.

...

The researchers were able to achieve a modulation speed of 1 gigahertz, but they noted that the speed could theoretically reach as high as 500 gigahertz for a single modulator.


The practical consequences of this revolutionary leap in computing and communication is really beyond anyone's comprehension. This will truly represent a distinct generation of computers in it's own right.

It is at least as revolutionary as going from vaccuum tubes to today's integrated circuits.

This is beyond Moore's law or even 3d architecture, and is also dealing directly with bottlenecks in the signal delay and bandwidth in networking and buses on the motherboard and add-on peripheral devices, such as secondary storage, monitor, printer, speakers, etc.
Intensero
not rated yet May 08, 2011
Absolutly exciting. Now all we need is room temperature super conductors and the ability to mass produce graphene and so many things will become avaiable. Can't wait for the advancement genetic mapping will get from this. I wonder how long we are to reaching the end of moores law, this leap seems to put us closer to the limit with current science.
I am loving this soooooooo much.
Quantum_Conundrum
4.5 / 5 (2) May 08, 2011
I wonder how long we are to reaching the end of moores law, this leap seems to put us closer to the limit with current science.
I am loving this soooooooo much.


Well, speaking strictly of Moore's Law:

We know the practical limit cannot be better than the ideal, which would be using a single atom as a transistor gate.

In the case of Silicon, this ideal limit is still more than an order of magnitude smaller than existing mass produced technologies in each dimension of measure.

We know that the experimental limit right now is already as good as 4nm wide transistor gates in 2d architecture, and that is much smaller than 22nm or 25nm in area...

For areas of transistor gates:

25^2 = 625
22^2 = 484

4^2 = 16

So a 4nm transistor is actually 30.25 times smaller area than is a 22nm transistor.

Again, we already know 4nm transistors are possible, as proven by experiment, we just don't know how to mass produce them in a chip or logic circuit just yet.
Quantum_Conundrum
4.5 / 5 (2) May 08, 2011
And so what I was alluding to is this INCREDIBLE observation...

Moore's law, 3d architecture, and materials sciences, such as graphene-based components, are different, unrelated drivers on improvements in computer technology, and they all "stack" multiplicatively with one another.

Moore's law arises as a consequence of the shrinking size of transistor gates and wires, but has only ever been applied to 2 dimensions.

The third dimension increases area by obviously changing the order of complexity from a power of 2 to a power of 3.

Assuming similar scaling of wires, cooling systems, scaffolding, and protective layers in chips, you can calculate the maximum density of transistors in the 3rd dimension by raising the existing number of transistors to the 1.5 power, which is the same as taking the square root and then cubing the result.

So for example, a 2d chip which has 2 billion transistors would be compared to a 3d chip which has 89 TRILLION transistors (eventually).
Quantum_Conundrum
4.5 / 5 (2) May 08, 2011
So the end result of computer technology, applying Moore's law and 3d concepts "should" be computers that have millions of times more trainsistors per unit volume, and each transistor operates ten to 100 times faster than existing technology.
Quantum_Conundrum
5 / 5 (1) May 08, 2011
Ok, so speaking of 3d architectures.

If acheiving the ideal of true 3d architecture isn't possible, then using Fractal terminology of fractional dimensions will be possible.

Here is how the increase in transistor density increases if we have a Fractal-like "fractional dimension", 2 < R < 3 for individual chips.

Dimension: Multiplier vs 2d

2.0: same as today.
2.1: 2.91777299
2.2: 8.5133992
2.3: 24.840166
2.4: 72.47797
2.5: 211.47
2.6: 617
2.7: 1800
2.8: 5253
2.9: 15327
3.0: 44721.36

In general, each tenth of a dimension represents a factor of 2.91777299069 greater than the previous tenth of a dimension for maximum theoretical transistor density.

These numbers assume a chip will be the same height as today's analogous chip's width, and assumes same-sized transistors vs same-sized transistors...
Quantum_Conundrum
4.7 / 5 (3) May 08, 2011
I know I'm sort of hung up on this thread, but the emerging technologies and the implications of their end results are absolutely mind blowing.

Take this, the memory company just announced 25nm process RAM.

So imagine we have 1 gigabit of 25nm process RAM.

Now imagine same scaling factor into the third dimension with same wastes for cooling, protection, and scaffolding, gives, with dimension R = 3:

3.16E13 25nm 3d transistors, or 31.6 Terabits.

Now, imagine the same area of 2d RAM, but eventually with 4nm process transistors, gives 39.06 gigabits of RAM in the same area, still 2d.

Finally, imagine 4nm transistors in 3d architecture, gives:

7.72E15 transistors in the same "space," or 7.72 Petabits = ~7720 Terabits = ~7,720,000 Gigabits.

Ok, not exactly since each prefix is bigger by 1024, not 1000, but that's close enough, damn...

As unimaginable as that is, 3d architecture with 4nm gates is theoretic 7.7 million times better than today's best RAM....
mrlewish
2 / 5 (1) May 08, 2011
3 dimensional chips are limited due to this thing you might have heard of called heat. The recent advancement by Intel looks to me to be just a fancy way of saying that they turned the junctions on their sides like books in a shelf instead of all flat and facing up. Yes there will be improvements, good ones at that, but it is not a panacea. Instead of a technological singularity what if there is a technological limit?
MIBO
not rated yet May 09, 2011
3D chips are a nice idea, but manufacturing of such devices would require building the device in layers.
Each layer requiring many processing steps and adding cost whilst reudcing yield.
Sadly I feel that it's not the theroetical technological limitations but the cost and time to manufacture such devices coupled with the low yield that will prevent the projections above becoming achievable in our lifetimes, if ever.
building a device 1cm square may take a few dozen process steps and have a yield of 99%,
now if a layer was just 1 micron thick and a stacked device built in layers, a 1cm cube would have 10000 layers, require 0.5 million process steps and have a yield of 0.99^10000, or
2.2e-44. Not to mention he several years in manufacture due to the time taken for each processing stage.
Husky
not rated yet May 09, 2011
regarding heat, it would be nice if the voltage of the graphene roof itself was tuned by light, for instance putting a solar cell like layer of Indium Germanium or something on top of to regulate the electrons in the graphene layer beneath, thus it would become allmost an all optical system, also the geometry of the waveguides could be exploited with metastructures to work with evanescent waves below the wavelength of light
Quantum_Conundrum
5 / 5 (1) May 09, 2011
MIBO:

That's why I discussed the notion of the fractal, or "factional dimension".

The key to this will be some sort of massively parralleled, "bottom-up" approach using molecular machinery to assemble chips from the atomic and molecular levels in the same way Ribosomes assemble protein chains to build cell structures.

Achieving dimensionality of 2.1 to maybe 2.3 or so at the chip level does not seem like too much of a stretch.

Full 3d will require the ultimate degree of nanotechnology.
Quantum_Conundrum
5 / 5 (1) May 09, 2011
3 dimensional chips are limited due to this thing you might have heard of called heat. The recent advancement by Intel looks to me to be just a fancy way of saying that they turned the junctions on their sides like books in a shelf instead of all flat and facing up. Yes there will be improvements, good ones at that, but it is not a panacea. Instead of a technological singularity what if there is a technological limit?


Yes, heat waste and energy consumption are easily the two limiting factors of 3d technology.

Fortunately, there are some technologies which will offset a significant portion of this.

1) Reduce heat waste

Spintronic and optical circuits should use orders of magnitude less energy, with orders of magnitude less heat waste, since you mostly only move spin currents or individual photons.

2) Thermoelectric devices can serve to recycle some of the waste heat back into electricity, which is then used as needed.
Beard
not rated yet May 09, 2011
The researchers were able to achieve a modulation speed of 1 gigahertz, but they noted that the speed could theoretically reach as high as 500 gigahertz for a single modulator.


Please excuse my misunderstanding; isn't the article talking about a faster replacement for fiberoptics? QC is talking about 3D chips though, is this a new max clock speed for transistors?
Quantum_Conundrum
not rated yet May 09, 2011
Please excuse my misunderstanding; isn't the article talking about a faster replacement for fiberoptics? QC is talking about 3D chips though, is this a new max clock speed for transistors?


Yes, but this is specifically the modulator for network communications, which is to say, an important component in networking devices such as modems, routers, and network adapters.

This technology would be likely to appear in large corporate networks, data centers, and Internet Service Providers within a few years of being manufactured. It would likely take 5 to 10 years or so to "trickle down" to end users in wired networks, though everyone would benefit from it immediately just because the server-side would be much faster internally in the "cloud" networks.

...
Yesterday's posts were inspired by this article and several other articles, and the potential synthesis and convergeance of these technologies in computer engineering.
Intensero
not rated yet May 09, 2011
QC where are you doing your research. You sound very knowledgable about all the things I want to study. Thanks for posting some figures for me to research. I agree with you that this is the best new I have read in some time.
Quantum_Conundrum
not rated yet May 09, 2011
QC where are you doing your research. You sound very knowledgable about all the things I want to study. Thanks for posting some figures for me to research. I agree with you that this is the best new I have read in some time.


All that is simple dimensional analysis.

Most of the component technologies to develop these "futuristic" computers have already been demonstrated in "one-of" laboratory proof of concept designs. The key is to optimize the designs and figure out a way to mass produce these components as needed.

I tend to pay a lot of attention to any developments in spintronics, optics, or 3-d trending architectures.

In 15 years, computers went from single-core, 32-bit, 200megahertz with like 32mb of RAM(factory), up to ~4 ghz, 6 core, hyperthreading, 64-bit with 6GB of RAM(factory), expandable to something like 24GB RAM.

In another 15 years, a desktop PC will have 384 of 64-bit processors, even without 3d architecture.