SoylentNews is people
SoylentNews
Study unlocks full potential of 'supermaterial' graphene
New research reveals why the "supermaterial" graphene has not transformed electronics as promised, and shows how to double its performance and finally harness its extraordinary potential.
Graphene is the strongest material ever tested. It's also flexible, transparent and conducts heat and electricity 10 times better than copper.
After graphene research won the Nobel Prize for Physics in 2010 it was hailed as a transformative material for flexible electronics, more powerful computer chips and solar panels, water filters and bio-sensors. But performance has been mixed and industry adoption slow.
Now a study published in Nature Communications identifies silicon contamination as the root cause of disappointing results and details how to produce higher performing, pure graphene.
The RMIT University team led by Dr Dorna Esrafilzadeh and Dr Rouhollah Ali Jalili inspected commercially-available graphene samples, atom by atom, with a state-of-art scanning transition electron microscope.
"We found high levels of silicon contamination in commercially available graphene, with massive impacts on the material's performance," Esrafilzadeh said.
[...] The article "Silicon as a ubiquitous contaminant in graphene derivatives with significant impact on device performance" is published in Nature Communications: DOI: 10.1038/s41467-018-07396-3
(Score: 2) by edIII on Saturday December 01 2018, @11:56PM (3 children)
That reminds me a conspiracy theory I heard a long time ago when we were using modems, and that was they had the tech for 28.8kb/s modems when they came out with 300 baud. It was just a bullshit game on part of the tech companies to force us to buy new modems every damn year.
What's the fear here though? That we wouldn't still buy new graphene computers every year? The neat thing about software, especially Microsoft, is that can seemingly expand in complexity and resource requirements to fit any computer you have.
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 2) by takyon on Sunday December 02 2018, @05:24AM
Functionality in software can actually increase to make good use of the hardware, even if some specific software like word processing feels laughably bloated.
And here's a classic:
Progress In Algorithms Beats Moore's Law [slashdot.org]
This has become even more striking in the age of machine learning:
Algorithms Outpace Moore’s Law for AI [eetimes.com]
The age of the algorithm — why AI progress is faster than Moore’s Law [medium.com]
One only has to check out a few videos at Two Minute Papers [youtube.com] to find that the pace of algorithmic progress is staggering, and that a new technique could be old hat in just a few months. And these algorithms can improve smartphone camera apps, voice assistants, games, photo/video editing software, you name it.
If you combine software advances with further hardware improvements, the results could be dramatic.
We still have some low-hanging fruit to collect with further process nodes, such as "7nm", "5nm", and "3nm". A couple of nodes below that are also possible. Single threaded performance can probably double along the way, while core counts could quadruple or better (counting a 2x from the "7nm" node alone).
In the pessimistic scenario, that's all we get and we're done aside from minor refinements. But optimistically, we could use new materials to boost clock rates to 100+ GHz or THz levels, or layer up with 3D integrated circuits, boosting conventional core counts from today's 4-16 to hundreds or thousands. Neuromorphic architectures could see big improvements as well.
With that kind of hardware, you will still see bloated web browsers, but you could easily see some mind-bending applications as well. And don't forget supercomputers, that need all the improvements that they can get.
The optimistic scenario could mean the difference between having strong AI at home, or only having it at big corporations.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Sunday December 02 2018, @07:13AM (1 child)
Was there a market for a 28.8kb/s modem that weighed 600 pounds, needed 70 Amps of 308-volt 3-phase power, occupied an entire 19" rack, and was priced accordingly?
(Score: 0) by Anonymous Coward on Monday December 03 2018, @03:30AM
A little excessive in your estimated resource requirements. But, yes, in 300 baud days, it would probably have taken a bunch of ECL chips to process 28Kpps because FET wasn't fast enough yet. It would be a little silly for you to pay far more for your modem than for your computer, not to mention how would ISPs cool racks with hundreds of those monsters. So it may have been technologically feasible, but not economically viable at that time.