Talk:Shockley–Queisser limit

From Wikipedia, the free encyclopedia
Jump to: navigation, search


In the "The Limit" section, one can read : "Since the act of moving into the conduction band requires energy, only photons with more than that amount of energy will create power..." ok.. and it continue : "and the power produced will be any leftover energy after boosting the electron to the conduction band". As I understand it, all leftover energy will be lost in heat, so it will not contribute to the power of the cell. I think this is a claim that must be check. —Preceding unsigned comment added by (talk) 00:20, 11 December 2008 (UTC)

It depends on your definition of "all". In practice, some of this energy is indeed captured, as some of the photoelectrons are released close enough to the depletion layer that they don't have time to loose energy into the structure as phonons. Of course this region is quite thin, and in practice it contributes very little (none) to the production. Maury Markowitz (talk) 21:39, 13 January 2009 (UTC)

The Notion of an Infinite Number of Layers[edit]

Would it be fair to state that, though the idea of an infinite number of layers is absurd, internal reflection will enable something approximating this to occur? In particular, internally reflecting light with a small angular offset ensures that more light can be converted to electricity (though niceties relating to frequency dependent reflection have to be considered).

ConcernedScientist (talk) 09:51, 2 February 2009 (UTC)

It's "less impossible" than you might think. Look up the work on optical rectennas. In practice a cell with 5 or 6 layers approaches the limit fairly closely, but the cost of those layers is very high and there's a serious diminishing returns curve. That said, if someone does come up with a working IR layer, I'd expect the GaAs cells to grow into the 5 to 6 layer depth very quickly. Maury Markowitz (talk) 11:28, 5 May 2009 (UTC)

Page is incomplete[edit]

I believe the detailed balance relates in part to carrier recombination, which has an unavoidable temperature-dependent extent. I know for sure the maximum single junction effiency is dependent on solar cell temperature. See The discussion on the main page only covers the mismatch between the solar spectrum and a single band gap absorber. —Preceding unsigned comment added by (talk) 14:11, 29 October 2009 (UTC)

Here's a link —Preceding unsigned comment added by (talk) 14:20, 29 October 2009 (UTC)

You're absolutely right. So I sat down with Ted Sargent at UofT in order to get a better understanding of these issues. I have extensively updated the article. Maury Markowitz (talk) 22:32, 17 January 2011 (UTC)

Any OR objections?[edit]

I did a reproduction of the Shockley–Queisser calculation ([1]). I was thinking of putting some of the plots onto this page, but don't want it to be deleted as original research.

There should be no problem with the efficiency plot, my plot agrees with the same plot in many textbooks. Also my ideal short-circuit current and ideal open-circuit voltage plots agree with the ones published in Practical Handbook of Photovoltaics, p128-9. So there should be no problem with those.

The only problematic one is the graph breaking down efficiency losses into below-bandgap-losses, etc. This is a straightforward calculation but I can't find it in a reliable source. If I put it up, would anyone plan to delete it as "original research"? If so, I won't bother in the first place. I consider it (marginally) a routine calculation... :-)

Thanks! --Steve (talk) 22:38, 6 February 2011 (UTC)

New Nanowire Research[edit]

This article ( points to a method of defeating the current limit by "a few percent" using nanowire technology. Food for thought, although just theoretical food at this point. (talk) 13:48, 25 March 2013 (UTC)Ben (mostly anon web user)

The article already mentions that the limit is slightly higher for concentrated light than unconcentrated. Shockley–Queisser limit#Light concentration. --Steve (talk) 20:45, 25 March 2013 (UTC)
This is a different twist as the power is increased without increasing the total surface area. Hcobb (talk) 14:08, 26 March 2013 (UTC)
Well, you are increasing the surface area: They are proposing to make a 1cm2 panel that has only 1mm2 of semiconductor material on it (or whatever the exact numbers are). You might think of it conceptually like starting with a 1mm2 solar cell and cutting it into nanowires and then "spreading them out". In other words, it is not fundamentally different than cutting up a macroscopic solar cell, spreading out the pieces, and putting a lens on top of each piece. But what the heck, I added a sentence to the article anyway mentioning this nanowire reference. --Steve (talk) 18:16, 26 March 2013 (UTC)


The Background section currently starts off with:

In a traditional solid-state semiconductor, a solar cell is made from two doped crystals, one an n-type semiconductor, which has extra free electrons, and the other a p-type semiconductor, which is lacking free electrons. When placed in contact, some of the electrons in the n-type portion will flow into the p-type to "fill in" the missing electrons, also known as "holes."

I believe this to be seriously misleading. A basic solar cell would be made from a single doped crystal, with (at least) two different dopant materials in different concentrations. With the exception of point contact diodes (which solar cells never are), I don't believe p-n junctions are ever formed by bringing two semiconductor materials together. Potatoj316 (talk) 21:17, 14 April 2014 (UTC)

What happens on the n side?[edit]

@Maury Markowitz: Regarding your edit (edit comment: "while true, this is covered in the solar cell article. as it contributes a *tiny* amount to production, it's not worth mentioning here as an aside"), why do you say that photon absorption on the n side is a tiny contribution? Doesn't that depend simply on the relative thicknesses of the n and p sides? Eric Kvaalen (talk) 18:01, 19 May 2016 (UTC)

The issue is that this is an article about the SQL, not the cell mechanisms. A basic description is all we need here, and in that spirit, I think it could use even more cutting. Maury Markowitz (talk) 18:56, 19 May 2016 (UTC)


@Maury Markowitz: Should the latest developments in Power conversion efficiency exceeding the Shockley–Queisser limit in a ferroelectric insulator be added to the sections on improving on the SQ limit? Hmoulding (talk) 22:55, 8 August 2016 (UTC)

This is an example of hot electron capture, so I would definitely think it should be added to that section. Maury Markowitz (talk) 12:06, 9 August 2016 (UTC)
I don't think it should be mentioned at all, it's a terrible article and none of the headline figures or conclusions are proven or even plausible. Solar cell metrologists recommend against trusting a solar cell efficiency measurement unless the device is 1cm×1cm, or a bit smaller is OK if you surround it by an opaque mask. They measured devices 100 million times smaller than that, with no mask. Not to mention, their efficiency calculation procedure involved multiplying the measured efficiency by a dubious factor of 30 (dubious because the 97% of light absorbed in ITO can also contribute to current, among other things).
The rottenness of their methodology is proven beyond a doubt by the fact that they claim a quantum efficiency of 18 (Fig. 3c), which is completely impossible under these circumstances (with photon energy a bit higher than the bandgap, the limit is 1 under normal circumstances or maybe 2 or 3 if there are exotic intermediate energy bands).
I'm familiar with these kinds of devices, indeed I'm a coauthor on two of the papers they cited (Refs. 6 & 7). Anyway, wikipedia policy always encourages not citing primary sources or press releases, and this is a case where that's wise. --Steve (talk) 15:18, 9 August 2016 (UTC)

BB radiation vs. recombination[edit]

Recent edits to the sections on blackbody radiation appear to be related to recombination, but the wording is unclear to me. Is Qc the rate of production due to recombination, or simply the long-tail of the BB curve? The wording suggests the former, which to my mind means it should not be in this section, but the next. Maury Markowitz (talk) 12:38, 9 August 2016 (UTC)

@Maury Markowitz: Qc refers to the tail of the blackbody curve (but I wouldn't call it the long tail – it's the tail on the visible and ultraviolet side, which falls off very rapidly compared to the tail on the far infrared and microwave side). So it's not just due to recombination. That would be tcQc. Shockley and Queissar assume that the recombination rate at zero bias is proportional to this total amount of blackbody radiation at energies above the band gap, and this plays an important role in the derivation of their equation for the amount of energy extracted per photon (per incoming photon above the energy gap that is).

As for which section this should go in, it's a little bit uncertain. Shockley and Queissar talk about blackbody radiation not so much as a loss but as a means for calculating the recombination. If in the section on blackbody radiation we only talk about it as a loss of energy, then that misses the point. It's actually good that there is blackbody radiation, because it helps keep the cell cool (along with conduction and convection of air), and a cool cell operates more efficiently. I tried to bring that out in my edit.

If you want to continue this discussion, please be patient with me because I am leavin tonight for three weeks of hiking in England and I don't know how often I will be able to respond! Eric Kvaalen (talk) 14:28, 9 August 2016 (UTC)

Well I'm leaving for the cottage too, so we'll have to do this like they did in the old days - weeks between messages! In any event, your first para clears this up. So then if Qc is "simply" the portion of the blackbody curve over and above the bandgap, it would seem that these photons are not due to recombination, correct? And if that is the case, I'm not sure what the connection is. Is it a case of the numbers being similar, so we can use that rate, or is there a direct physical mechanism I'm missing? Maury Markowitz (talk) 15:09, 9 August 2016 (UTC)
By time reversal symmetry / Kirchhoff's law / etc., absorption and radiation are related. Above-bandgap photons are absorbed almost exclusively by the process of creating an electron-hole pair. Therefore, above-bandgap photons are emitted almost exclusively by the process of an electron-hole pair recombining. Below-bandgap photons are more-or-less not absorbed at all; the material is transparent below its bandgap. Therefore, below-bandgap photons are emitted more-or-less never.
So, I would say talking about the "long tail" is misleading. It doesn't have a blackbody curve, because it's not a blackbody. It only emits above the bandgap. I guess you can draw a dashed line and talk about the spectrum that it would emit if it were a blackbody. But such talk is just liable to confuse readers I bet. --Steve (talk) 15:29, 9 August 2016 (UTC)

@Maury Markowitz and Sbyrnes321: I'm back. (In the end I only hiked six days!) Steve, whether the cell emits like a black body below the bandgap depends on how thick it is. Maury, if Steve is right that above-bandgap photons are absorb'd almost exclusively by creating an electron-hole pair, then (as he says) almost all photons emitted are created by recombination. However, according to the paper of Shockley and Queisser (end of page 514), the factor fc, which is the fraction of recombination events that lead to the emission of a photon, is quite small, like 10−10. So there are many more recombinations than photons emitted. The numbers are normally not similar as you suggest. But in any case, fc cannot be more than 1, and the upper limit (the Shockley-Queisser limit) requires taking fc = 1. Eric Kvaalen (talk) 19:05, 6 September 2016 (UTC)
Yes, virtually all above-gap photons come from recombination, but not all recombinations create above-bandgap photons. (I don't think the fraction is as low as 10−10 these days. Solar-cell efficiencies have improved since the 1960s! Best is the Alta Devices GaAs solar cells where the fc parameter actually gets quite close to 1. Maybe it's 0.8 or something, I forget. Not coincidentally, these same cells are very close to the S-Q limit.) --Steve (talk) 01:14, 7 September 2016 (UTC)