Maximum theoretical performance of photovoltaics?

[Running_with_Scissors]

I know photovoltaics are still in their infancy. Apparently, typical cells being produced today are at 20-25% efficiency.

Suppose that we could manufacture a cell that was 99.9% efficient. Ignoring energy used in manufacturing, and assuming optimal placement, what would be the maximum amount of power per square foot that this “ideal” photovoltaic panel could produce?

Or, in other words, how much power does sunlight have?

[Mangetout]

About 1kw per square metre (max, all factors being optimal)

[engineer_comp_geek]

Outside Earth’s atmosphere, the sun’s energy contains about 1,300 watts per square meter. About one-third of this light is reflected back into space, and some is absorbed by the atmosphere (in part causing winds to blow).

By the time it reaches Earth’s surface, the energy in sunlight has fallen to about 1,000 watts per square meter at noon on a cloudless day.

from here: How Solar Energy Works | Union of Concerned Scientists

[Chronos]

About 1kw per square metre (max, all factors being optimal)

Of course, the time-averaged value would be less than half of this, since half the time, the Sun’s not shining at all, and even when it is, it’s not always directly overhead. I think that the overall efficiency would be exactly 1/4 the maximum, since that’s the ratio of the area of a circle to a sphere.

[UncleBeer]

Also, the efficiency of PV cells is typically reduced as the temperature of the cell increases. Ironically, in the brightest sunlight, they may be less efficient than at other times of the day.

[Sunspace]

Not only that, but for maximum efficiency to capture that 1 kilowatt per square metre, your cells would have to absorb all incoming wavelengths of light (and thus they would look black).

[Omphaloskeptic]

@Chronos

Of course, the time-averaged value would be less than half of this, since half the time, the Sun’s not shining at all, and even when it is, it’s not always directly overhead. I think that the overall efficiency would be exactly 1/4 the maximum, since that’s the ratio of the area of a circle to a sphere.

This is true when averaged over all latitudes. At latitude L the time-averaged value is cos(L)/pi of the maximum, so 1/4 is also a reasonably good approximation for temperate latitudes such as the contiguous US.

[Chronos]

This is true when averaged over all latitudes.

Yes, I should have stated explicitly that I was also averaging over location. And I was also assuming that the cells were maintaining the same incredibly high efficiency under all circumstances, per UncleBeer’s comment.

[scm1001]

just a quick note on efficiencies. I have read that the maximum possible efficiency of an ideal (electric) solar sell is around 93% due to thermodynamic limitations.

The maximum possible for a single boundary solar cell is around 33%. This is because light which is too high in energy (say blue) wastes most of its energy, whereas light which is too low (say infrared) does not get absorbed. This is the limit therefore for any likely cheap PV material.

This limitation can be overcome through multiple layers which selctively extract different portions of the light. However, it will be very difficult all in to get past 50%, and expensive.

[Squink]

@scm1001

33%. This is because light which is too high in energy (say blue) wastes most of its energy, whereas light which is too low (say infrared) does not get absorbed.

Have you seen any action spectra for typical solar cells? I looked around a bit this AM, but didn’t find any.

[iamthewalrus_3]

@Sunspace

Not only that, but for maximum efficiency to capture that 1 kilowatt per square metre, your cells would have to absorb all incoming wavelengths of light (and thus they would look black).

Yeah, and not just black like when something is painted black. They’d look like a hole in reality, a total absense of any visual information.

It would probably be quite unnerving.

[Running_with_Scissors]

For what it’s worth, What Exit? posted this thread late last year regarding a new solar cell technology that approximately doubles current output (to about 40%)…

[Diceman]

@iamthewalrus(:3=

Yeah, and not just black like when something is painted black. They’d look like a hole in reality, a total absense of any visual information.

It would probably be quite unnerving.

Wow, that does sound freaky! It would look like a little patch of pure darkness, perfectly rectangular and impossible to illuminate no matter how big a flashlight you pointed at it.

[scm1001]

@Squink

Have you seen any action spectra for typical solar cells? I looked around a bit this AM, but didn’t find any.

this may be what you are looking for ?
http://www.semiconductor-sanyo.com/amorton/feature/index.htm

a nice page on ways to imcrease efficiencies is here
http://homepage.mac.com/nekins/web/research.html

[scm1001]

oh and more PV material spectra here http://www.rfcafe.com/references/electrical/ASTM%20G173-03%20Reference%20Spectra.htm

[Squink]

@scm1001

this may be what you are looking for ?
http://www.semiconductor-sanyo.com/amorton/feature/index.htm

That’s it!
It looks to me like we’ll need at least 3 different photactive materials in a cell, with peak efficiencies at different wavelengths, to pick up the full 1,000 watts per square meter from direct sunlight.

[Askance]

@UncleBeer

Also, the efficiency of PV cells is typically reduced as the temperature of the cell increases. Ironically, in the brightest sunlight, they may be less efficient than at other times of the day.

That’s also the time you need efficiency the least as the actual amount of power hitting the cell is highest.

IMO Photovoltaics cells are never going to be significant grid providers, they’ll mostly be on buildings and other structures to provide some degree of grid-independence. Solar thermal is the technology more likely to be able to replace fossil fuel generators.