[LINK] Crowdsourcing printable solar cell compounds

[stephen at melbpc.org.au]

Harvard global grid computing to help create printable solar cells

By Lucas Mearian, 16 April 2013.  http://www.arnnet.com.au/article/459247


In June, Harvard's "Clean Energy Project" (CEP) plans to release to solar 
power developers a list of the top 20,000 organic compounds that could be 
used to make cheap, printable photovoltaic cells (PVCs).

 http://cleanenergy.harvard.edu
 
The list, culled from about seven million organic molecules that a 
crowdsourcing-style project has been crunching over the past two-plus 
years, could lead to PVCs that cost about as much as paint to cover a one-
meter square wall.

"We're in the process of wrapping up our first analysis and releasing all 
the data very soon," said Alan Aspuru-Guzik, an associate professor of 
chemistry and chemical biology at Harvard

Today, the most popular PVCs are made of silicon and cost about $5 per 
wafer to produce. Silicon PVCs have a maximum solar conversion efficiency 
rate of about 12%, meaning only 12% of the light that hits them is 
converted to energy.

There is also a small niche market of organic PVC vendors, but their solar 
cells offer only about 4% to 5% efficiency rate in converting solar rays to 
energy. In order for a solar product to be competitive, each would need to 
cost about 50 cents, according to Aspuru-Guzik.

The Clean Energy Project, however, uses the computing resources of IBM's 
World Community Grid for the computational chemistry to find the best 
molecules for organic photovoltaics. 

IBM's World Community Grid allows anyone who owns a computer to install 
secure, free software that captures the computer's spare power when it is 
on and idle. http://www.worldcommunitygrid.org

By pooling the surplus processing power of about 6,000 computers around the 
world, the Clean Energy Project has been able to come up with a list of 
organic photovoltaics that could be used to create inexpensive solar cells. 

The computations also look for the best ways to assemble the molecules to 
make those devices.

Computational chemists typically calculate the potential for photovoltaic 
efficiency one organic molecule at a time. Over the past few years, 
computational chemists have identified a few organic compounds with the 
potential to offer around 10% energy conversion levels.

"But that's only two or three," Aspuru-Guzik said. "Through our project, 
we've identified 20,000 of them at that level of performance."

In fact, CEP's list of molecules include some that have upwards of 13% 
solar conversion efficiency rates, Aspuru-Guzik said.

The computing resources from IBM's World Community Grid are split for the 
CEP. Some of the computers in the grid are making mechanical calculations 
of molecular crystals, thin films and molecular and polymer blends; others 
are making electronic structure calculations to determine the relevant 
optical and electronic transport properties of the molecules.

Harvard has also constructed significant data storage facilities to capture 
the results of the computations. Each molecular computation produces on 
average about 20MB of data. In total, the global grid computing 
architecture generates about 750GB of data per day. So far, the data has 
grown to about 400TB.

Harvard has filled racks of servers with 4U-high hard drive arrays. Each 
array is filled with 45, 7200rpm 3TB hard drives from Western Digital 
subsidiary HGST.

"The data we're creating will ultimately benefit mankind with cleaner 
energy solutions," Aspuru-Guzik said. "Accordingly, we designed our Jabba 
storage arrays with built-in redundancies. But the key to the arrays' 
performance is the use of reliable, high-capacity, and low-power storage 
from HGST. We've filled nearly 150 HGST drives to this point and are 
currently building Jabba 5 and 6 to handle the enormous amount of data 
generated for the project."
--

Cheers,
Stephen Loosley