Research sparks record-breaking solar cell performances

Nov 07, 2011 by Lynn Yarris
Contrary to conventional scientific wisdom, the key to solar cell efficiency is not absorbing more photons but emitting more photons. (Image courtesy of DOE NREL)

(PhysOrg.com) -- Theoretical research by scientists with the U.S. Department of Energy (DOE)’s Lawrence Berkeley National Laboratory (Berkeley Lab) has led to record-breaking sunlight-to-electricity conversion efficiencies in solar cells. The researchers showed that, contrary to conventional scientific wisdom, the key to boosting solar cell efficiency is not absorbing more photons but emitting more photons.

“A great solar cell also needs to be a great Light Emitting Diode,” says Eli Yablonovitch, the Berkeley Lab electrical engineer who led this research. “This is counter-intuitive. Why should a solar cell be emitting photons?  What we demonstrated is that the better a solar cell is at emitting photons, the higher its voltage and the greater the it can produce.”

Yablonovitch holds joint appointments with Berkeley Lab’s Materials Sciences Division and the University of California (UC) Berkeley, where he is the James and Katherine Lau Chair in Engineering, and also directs the NSF Center for Energy Efficient Electronics Science. He is the corresponding author of a paper describing this work for the Journal of PhotoVoltaics titled “Intense Internal and External Fluorescence as Approach the Shockley-Queisser Efficiency Limit.”

Co-authoring this paper with Yablonovitch were Owen Miller of Berkeley Lab, and Sarah Kurtz, at the National Renewable Energy Laboratory.

In their paper, Yablonovitch, Miller and Kurtz describe how  external fluorescence is the key to approaching the theoretical maximum efficiency at which a solar cell can convert sunlight into electricity. This theoretical efficiency, called the Shockley-Queisser efficiency limit (SQ Limit), measures approximately 33.5-percent for a single p-n junction solar cell. This means that if a solar cell collects 1,000 Watts per square meter of solar energy, the most electricity it could produce would be about 335 Watts per square meter.

Calculations by Miller, who is a member of Yablonovitch’s research group, showed that the semiconductor gallium arsenide is capable of reaching the SQ Limit. Based on this work, a private company co-founded by Yablonovitch, Alta Devices Inc., has been able to fabricate solar cells from gallium arsenide that have achieved a record conversion efficiency of 28.4 percent.

Thin film solar cells fabricated from gallium arsenide have achieved a record sunlight-to-electricity conversion efficiency of 28.4 percent. (Image courtesy of Alta Devices, Inc.)

“Owen Miller provided an accurate theory on how to reach the SQ Limit that for the first time included external fluorescence efficiency,” Yablonovitch says. “His calculations for gallium arsenide showed that external fluorescence provides the voltage boost that Alta researchers subsequently observed.”

Solar or photovoltaic cells represent one of the best possible technologies for providing an absolutely clean and virtually inexhaustible source of electricity. However, for this dream to be realized, solar cells must be able to efficiently and cost-competitively convert sunlight into electricity. They must also be far less expensive to make.

The most efficient solar cells in commercial use today are made from monocrystalline silicon wafers and typically reach a conversion efficiency of about 23-percent.  High grade silicon is an expensive semiconductor but is a weak collector of photons. Gallium arsenide, although even more expensive than silicon, is more proficient at absorbing photons, which means much less material is needed to make a solar cell.

“Gallium arsenide absorbs photons 10,000 times more strongly than silicon for a given thickness but is not 10,000 times more expensive,” says Yablonovitch. “Based on performance, it is the ideal material for making solar cells.”

Past efforts to boost the conversion efficiency of solar cells  focused on increasing the number of photons that a cell absorbs. Absorbed sunlight in a solar cell produces electrons that must be extracted from the cell as electricity. Those electrons that are not extracted fast enough, decay and release their energy. If that energy is released as heat, it reduces the solar cell’s power output. Miller’s calculations showed that if this released energy exits the cell as external fluorescence, it would boost the cell’s output voltage.

“This is the central counter-intuitive result that permitted efficiency records to be broken,” Yablonovitch says.

As Miller explains, “In the open-circuit condition of a solar cell, electrons have no place to go so they build up in density and, ideally, emit external fluorescence that exactly balances the incoming sunlight. As an indicator of low internal optical losses, efficient external fluorescence is a necessity for approaching the SQ Limit.”

Using a single-crystal thin film technology developed earlier by Yablonovitch, called “epitaxial liftoff,” Alta Devices was able to fabricate solar cells based on gallium arsenide that not only smashed previous solar conversion efficiency records, but can be produced at well below the cost of any other solar cell technology. Alta Devices expects to have solar panels on the market within a year.

“The SQ Limit is still the foundation of solar cell technology,” says Yablonovitch. “However, the physics of light extraction and external fluorescence are clearly relevant for high performance solar cells.”

Yablonovitch believes that the theoretical work by he and his co-authors, in combination with the performance demonstrations at Alta Devices, could dramatically change the future of solar cells.

“We’re going to be living in a world where solar panels are very cheap and very efficient,” Yablonovitch says.

Explore further: Going nuts? Turkey looks to pistachios to heat new eco-city

More information: For more information about the research of Eli Yablonovitch, visit the Website at optoelectronics.eecs.berkeley.edu/

Related Stories

Device can heat home, save money

Apr 19, 2011

(PhysOrg.com) -- A new polymer-based solar-thermal device is the first to generate power from both heat and visible sunlight – an advance that could shave the cost of heating a home by as much as 40 percent.

Hot Electrons Could Double Solar Cell Power Efficiency

Dec 18, 2009

Scientists have experimentally verified a theory suggesting that hot electrons could double the output of solar cells. The researchers, from Boston College, have built solar cells that successfully use hot ...

Solving the solar cell power conversion dilemma

Jan 24, 2011

(PhysOrg.com) -- "There is a lot of interest in creating more efficient solar cells that are also simpler than many of the designs common now," Wladek Walukiewicz tells PhysOrg.com. "We think that, throug ...

Recommended for you

Obama launches measures to support solar energy in US

Apr 17, 2014

The White House Thursday announced a series of measures aimed at increasing solar energy production in the United States, particularly by encouraging the installation of solar panels in public spaces.

Tailored approach key to cookstove uptake

Apr 17, 2014

Worldwide, programs aiming to give safe, efficient cooking stoves to people in developing countries haven't had complete success—and local research has looked into why.

User comments : 35

Adjust slider to filter visible comments by rank

Display comments: newest first

omatwankr
1 / 5 (1) Nov 07, 2011
Not much difference from this story in June
http://gigaom.com...ovation/
"Alta has achieved 28.2 percent efficiency"
Scottingham
1.6 / 5 (9) Nov 07, 2011
300 watts per square meter isn't that much...it adds up, sure...but I doubt the whole roof and whole parking lot of a Walmart could power the building itself.
axemaster
4.9 / 5 (18) Nov 07, 2011
300 watts per square meter isn't that much...it adds up, sure...but I doubt the whole roof and whole parking lot of a Walmart could power the building itself.


Really? Let's use my local Walmart as an example and do some estimating...

The building is probably around 100x100 ft - that's about 30x30 meters or 900 square meters. At 300W per square meter that's 270kW. That's a LOT of power. I doubt the entire Walmart uses more than 100kW.

The parking lot is probably 200-150ft or 2700 square meters. That gives another 800kW.

So all together if my Walmart was covered in solar panels it would put out a megawatt of power, putting it on the scale of a small power plant.

Learn to math.
Eikka
2 / 5 (5) Nov 07, 2011
It's not very counterintuitive if you know how the cell works internally.

The diode junction in a solar cell is forward biased by the electrons that the photons excite, and the recombination current of those electrons returning back to their original places flows through the P-N junction. The diode junction has an associated voltage drop which shows up as the external voltage of the cell.

Light emitting diodes just so happen to have a greater forward voltage drop, which means that a solar cell built on such a junction would naturally have a higher voltage, and when you have a higher voltage you get the same power out at a lower current, which leads to less heating of the panel, less resistive losses etc. etc.
Eikka
4.1 / 5 (7) Nov 07, 2011
So all together if my Walmart was covered in solar panels it would put out a megawatt of power, putting it on the scale of a small power plant.

Learn to math.


If you actually built it, you'd note that the system doesn't output a megawatt on average, but roughly 100 kW. It would do a megawatt during the middle of the day, but otherwise not.

And that's the problem with solar panels. To get the average up, you need to produce roughly ten times peak power. How do you smooth it out?
Koen
1 / 5 (1) Nov 07, 2011
Is there a way to optimize silicon solar cells by means of this counter intuitive principle?
dnatwork
3 / 5 (3) Nov 07, 2011
Eikka, why smooth it out?

Put panels over every roof and parking lot. Create huge excess capacity. Link up the whole globe on one grid. Add some storage capacity into the system (batteries, or pump water uphill, whatever). Improve efficiency of household and industrial devices (where I'd guess at least half the energy could be saved if we were smarter). Radically reduce the weight of vehicles by putting electric rails in the roads, so electric vehicles would not need to carry heavy batteries, and redesigning the traffic system to eliminate the ability of idiots to crash into one another, so safety systems would not have to be directly proportional to vehicle weight. Standardize on one or two voltages to reduce conversion losses.

Last week I read that solar panels get fried if you leave them in the sun without a load. The panels described here sound like they would not break down in that case, they would just glow.
Etreum
1 / 5 (1) Nov 07, 2011
This PV cells also degrade? How much per year?
210
2 / 5 (5) Nov 07, 2011

Learn to math.

The Inverter circuits are very inefficient as they exist today. Converting the DC to AC will waste a heck of a lot of power, I estimate 550 KWatts available in AC, best case scenario and some excellent cooling schemes, full day light in Arizona, Sahara desert etc, maybe mountain top, Andes.
word-
Burnerjack
4 / 5 (6) Nov 07, 2011
axemaster, your math does indeed appear sound. except for a couple of glaring ommisions. Average time duration of energy harvesting, power requirements AND time period of said consumption rate. But hey, a megawatt here and a megawatt there... I do agree that even if it isn't enough for the building to be self sustaining if replicated on a wide enough scale, serious grid load reductions and the associated benefits could be realized.
Now, about that money thing... The true "green" economy is the "green" that folds and fits into a wallet. Like it or not, it IS what "makes the world go 'round". When it becomes financialy beneficial, Al Gore's marketing plan will be unnecessary.
douglas2
not rated yet Nov 07, 2011
I think I remember reports over 15 years ago of a porous form of silicon that was efficient at emitting light. The intention there was to generate light right in silicon on an IC.
Arachnivore
5 / 5 (2) Nov 07, 2011

The Inverter circuits are very inefficient as they exist today.


On a Walmart scale installation, it might make more sense to simply use DC devices in the store. The cost of LED lighting is coming down fast and most other systems can be converted to DC.
CapitalismPrevails
1 / 5 (1) Nov 07, 2011
"This theoretical efficiency, called the Shockley-Queisser efficiency limit (SQ Limit), measures approximately 33.5-percent for a single p-n junction solar cell. "

Why is their a theoretical efficiency or ceiling limit for solar cell efficiency?
djr
5 / 5 (1) Nov 07, 2011
"Converting the DC to AC will waste a heck of a lot of power." Not so - inverter efficiencies are around 95%. You get a bigger drop if you convert back to DC - but that is no different than what we do today - converting grid AC back to DC for use in computers etc. http://howsolarwo...verters/
210
1 / 5 (3) Nov 07, 2011
"Converting the DC to AC will waste a heck of a lot of power." Not so - inverter efficiencies are around 95%. You get a bigger drop if you convert back to DC - but that is no different than what we do today - converting grid AC back to DC for use in computers etc. http://howsolarwo...verters/

So wherever 'inversion' or conversion must take place, that is power we do not get and since we cannot get solar power with the constancy of present fossil powered systems, the incentive to change is always argumentative and contentious. To send power over a distance, I need to raise voltage massively...send it, then step it down, convert it to DC locally @ the curb (1 to 40% loss!), or, send AC to the appliance and lose it there...humm..1 - 40% of the neighborhoods 250KW (AC/DC) or 10% of 10 watts (Mac Mini's and electronics, more for motors) 500 million times all day, a bit less at night. We need a paradigm shift - better electronics and motors??!

word-to-ya-mutha
nixnixnix
5 / 5 (2) Nov 07, 2011
@Eikka,

I don't know what shitty-ass town that axemaster lives in but in my shitty-ass town (in Canadia) the local walmart measures 330x400 feet and I don't think it is big by US standards so there you go. That's way more than a ten fold increase. Next challenge please! :)
laser
5 / 5 (1) Nov 07, 2011


And that's the problem with solar panels. To get the average up, you need to produce roughly ten times peak power. How do you smooth it out?


I agree that peak power vs actual created is very different and you could add where it is located in relation to the equator. But a thin film panel with 28% efficiency (beats silicon) is significant. If this has the same advantages as its thin film brothers it makes the silicon panel less desirable in all building applications. Longevity is an issue but given the material I would expect greater life (it is evident in LED light bulbs).

As to storage... Should their be excess power electric (battery based) cars could be depositories for excess using smart grid technology that already exists.

I agree that peak power vs actual created is very different and you could add where it is located in relation to the equator. But a thin film panel with 28% efficiency (beats silicon) is significant. If this has the same advantages as
Jimee
not rated yet Nov 07, 2011
Not very counter-intuitive? Too bad you weren't involved in the very first solar cell constructed so that all this time and energy wouldn't have been spent getting to this point.
laser
5 / 5 (1) Nov 07, 2011
Latest company news has them in a government study to have complete solar installations at a cost of $2 per watt. Currently the Chinese are flooding the market with silicon solar panels at $1.40 to $2 a watt killing the competition. This technology would put us in the drivers seat. Complete installations at this price would compete with any existing power plant.

axemaster
5 / 5 (1) Nov 07, 2011
Ok, I'll do a more extensive analysis:

The light incident on the earths surface during the day is equal to a sine wave:

300W/m * 30m^2 * Sin[theta(2Pi/(60*60*24))]

Integrating over 12 hours gives: 172 kW average power during the day.

Now, I admit my Walmart is crappy and small. Doing the numbers for nixnixnix's Walmart (330-400ft) gives 1.15 MW average daytime power.

I am not putting these numbers up as absolute in any way, I am just doing estimations. It's easy to see that lots of power generation is possible, even if actual output were only 1/2 or less.

The true "green" economy is the "green" that folds and fits into a wallet. When it becomes financialy beneficial, Al Gore's marketing plan will be unnecessary.


My parents have already installed solar panels in the roof of their business and saved a lot of money... We have discussed the lack of adoption quite a bit, and as far as we can tell, people simply aren't aware that it's actually financially practical to do.
muskrat66
not rated yet Nov 07, 2011
Just a thought,
If you tuned the external fluorescence frequency to the ideal wavelength of the solar cell and then redirected the fluorescence internally to another solar panel wafered behind the first, then 30% of the waste florescent energy could be collected by the second collector.
You could then increase the total output per meter/2 by another 10% approx.
Any waste fluorescence could then be radiated out the under side of the panel to further reduce heating.
Feldagast
3 / 5 (2) Nov 07, 2011
All these theories and papers are nice but when can you buy them??
Vendicar_Decarian
2.3 / 5 (3) Nov 08, 2011

"If you actually built it, you'd note that the system doesn't output a megawatt on average, but roughly 100 kW" - Eikka

Well, a bit more than that. More like 160 KWh.

But then like most everything else, WalMart doesn't run all day so they don't need to generate much energy when the sun goes down.

So if we just consider the daytime production, the average wattage is 300 KWh.

And these low numbers presume the panels aren't steerable.

You might notice a trend in Modern Retail stores where there is a ridge line of windows along the southern wall near the roof.

They exist to reduce lighting costs by bringing sunlight into the store.

These windows will continue to get larger as energy costs rise.

Vendicar_Decarian
3.9 / 5 (7) Nov 08, 2011
"Currently the Chinese are flooding the market with silicon solar panels at $1.40 to $2 a watt killing the competition." - laser

Lets assume $4.00 a watt with installation.

It is estimated that George Bush's war crimes in Iraq will cost the American People 4 trillion dollars.

If this money had been invested in solar panels installed at $4.00 a watt, then the U.S. would have a generating capacity of 1 trillion watts Solar.

Even if these panels operated at only 1/10th of their rated capacity on average, this would represeent 876 Trillion TWh of electric power generation.

This is about 40 times the current generating capacity in the U.S.

jcims
5 / 5 (1) Nov 08, 2011
@Eikka,

I don't know what shitty-ass town that axemaster lives in but in my shitty-ass town (in Canadia) the local walmart measures 330x400 feet and I don't think it is big by US standards so there you go.


We have a 'super walmart' in my town of 15K in Ohio, and the GIS map shows it at almost exactly 300'x600'. I would guess 8-10% of the surface is covered with skylights and HVAC mechanicals (it's only a few years old). As a side note, the roof is probably 20' high and collects over a million pounds of water for every inch of rain. That's 20 million foot pounds of energy, or an astonishing 7.5 kilowatt hours (lol).
Vendicar_Decarian
2.4 / 5 (5) Nov 08, 2011
"Why is their a theoretical efficiency or ceiling limit for solar cell efficiency?" - CapitalismFails

Multi-layer cells are getting 42 percent in the lab.
drewgrey
not rated yet Nov 08, 2011
About inverters. Ac is a given for non local distribution...but for electricity that is produced and used on the spot as it were why bother. Store it and use it as DC. Do LEDs use DC? Also a clever lad might arrange the solar panels so that the emited light could be used for building light ,especially if the hotter end of the spectrum is not in the mix of emmitted light.
Eikka
1 / 5 (1) Nov 09, 2011

"If you actually built it, you'd note that the system doesn't output a megawatt on average, but roughly 100 kW" - Eikka

Well, a bit more than that. More like 160 KWh.


Assuming that the panels have 100% fill and no frames, and no shade or dirt on them, and you don't use blocking diodes on the arrays because they're perfectly balanced.

There's practical problems with arrays of solar panels, especially when you connect cells in series to get a useful voltage out because if you shade one cell, it will effectively limit the output of the whole series and heat up considerably in the process, so a bit of dirt or a single leaf on a panel that blocks one cell in the series will have a surprisingly large effect on the whole.

That's also a reason why a higher voltage out of a single cell is nice, because then you don't have to put so many in series. Still, you need the blocking diodes to stop parallel unbalanced cells from discharging into one another.
Eikka
1 / 5 (1) Nov 09, 2011
(cont.)

And the blocking diodes sap some of the output voltage, reducing the total efficiency. For example, for a 12 volt panel you'd get a 6-7% reduction in efficiency in exchange for the ability to wire many panels in parallel. The diodes are somewhat expensive though, because they have to handle large currents, so you might want to increase voltage in series instead of putting more cells in parallel.

The series panel issue can also be solved with bypass diodes, which shunt the current around the weak cells, but again because the diodes aren't exactly cheap you can't have a diode per cell. More like one for every five.

Besides, wiring more cells in series diminishes maximum output anyways by increasing series resistance of the interconnecting wires and thus lowering efficiency, so the array design is always more or less a compromize.
drewgrey
not rated yet Nov 10, 2011
Is there a reason why the storage can not be integrated into the panel itself.Create a module that combines the solar panel with storage and pull of the DC as needed.
Vendicar_Decarian
1 / 5 (2) Nov 12, 2011
"Assuming that the panels have 100% fill and no frames" - Eikka

Completely irrelevant because surface area is free, unless you are designing a satellite.

kaasinees
3 / 5 (2) Nov 12, 2011
my roomate's sister makes $68 every hour on the computer. She has been without a job for 6 months but last month her pay was $8123 just working on the computer for a few hours. Read about it on this web site 133t.us/ab

is your roommates sister a camwhore?
kaasinees
1 / 5 (1) Nov 12, 2011
Does anyone know the efficiencies between a mirror-salt plant and a PV plant with the same ground surface area?
wealthychef
1 / 5 (1) Nov 12, 2011
"Were going to be living in a world where solar panels are very cheap and very efficient, Yablonovitch says.
Really? When will that be? 30 years from now? It just ain't happening yet. How close are we?
Eikka
1 / 5 (1) Nov 17, 2011
"Assuming that the panels have 100% fill and no frames" - Eikka

Completely irrelevant because surface area is free, unless you are designing a satellite.



But limited in the case we were discussing.

In the general sense, the more surface are you use the higher the installation, maintenance, and cleaning costs, so it's kinda not free.

More news stories

TCS, Mitsubishi to create new Japan IT services firm

India's biggest outsourcing firm Tata Consultancy Services (TCS) and Japan's Mitsubishi Corp said Monday they are teaming up to create a Japanese software services provider with annual revenues of $600 million.

Finnish inventor rethinks design of the axe

(Phys.org) —Finnish inventor Heikki Kärnä is the man behind the Vipukirves Leveraxe, which is a precision tool for splitting firewood. He designed the tool to make the job easier and more efficient, with ...

Atom probe assisted dating of oldest piece of earth

(Phys.org) —It's a scientific axiom: big claims require extra-solid evidence. So there were skeptics in 2001 when University of Wisconsin-Madison geoscience professor John Valley dated an ancient crystal ...