How does a CIGS solar cell respond to sunlight?

Well, hopefully the solar cell will act like a battery when the sun shines on it. But, of course, I want to go into a bit more detail than that. I've already written a fair number of posts regarding the making and workings of CIGS solar cells. You'll find them by clicking on the "CIGS" label at the bottom of this post. If you're new to the CIGS world, I suggest you start there first.

By now, you must be tremendously curious about the wonderful graph my colleague Sebastian put together for me, my own computer skills being somewhat limited in this regard. First off, you probably can't read the labeling on the graph very well because it's too small. This is easily remedied by a single click on the graph itself to see it full size - a handy trick that's useful for websites in general. So go take a closer look now and I'll explain what you're seeing.

First off, I've actually got two graphs for the price of one. I'm "allowed" to put them together because they have a meeting ground in their shared x-axis, the one labelled wavelength. This refers to the wavelength of light. The units are nanometers, nm (1 nm = 0.0000001 centimeters - does that help?).

One graph is the solar irradiance aka solar spectrum. I'm pleased to display the visible part of the solar spectrum in the pretty colors. The y-axis on the left is for irradiance of the solar spectrum in units of watts per square meter per nanometer. The "square meter" part refers to sun power falling on a square meter of surface pointed directly at the sun. The "nanometer" part refers to the color or wavelength interval of incoming light so that you can see the relative power contribution by the various colors. To get the total solar power you'd have to integrate or add up all the individual contributions by wavelength interval over the entire spectrum including the wavelengths that lie outside the regions of my graph. Understandably a bit confusing, but I'm trying.

The graph of the solar spectrum is labelled very scientifically as "AM 1.5 global solar irradiance at sea level". This is a bit more information than I had intended to discuss, but, well, Sebastian is very much a scientist (I'm more of an engineer). Anyway, here goes. This spectrum is what's referred to as a "standard sun" for the purpose of comparing the performance of solar cells with each other under the same sunlight conditions. AM 1.5 stands for air mass 1.5 and refers to the sun when its zenith angle (the angle from vertical) is 48.2 degrees. This corresponds to sunlight that travels through an atmosphere that's "1.5 atmospheres thick". AM 1 would, of course, refer to a sun that's directly overhead with sunlight travelling through an atmosphere that's 1 atmosphere thick. The "global" part of the label refers to the fact that both direct and diffuse light are included. The diffuse light is the sunlight available to you when standing shaded from direct sunlight.

I think the spectrum must be what you might get in some fairly pristine environment and it's not the same as you would get where there are high pollutant levels. Humidity will also affect the spectrum. In fact, much of the "choppiness" in the spectrum is due to absorption by water molecules. In any case, the sun tends to be at some angle in the sky other than the zenith angle of 48.2 degrees. However, when you see a power rating of a solar module, it will have been calculated using this AM 1.5 spectrum. So a module rated at a certain power output will rarely produce that particular power. It's just a means of comparing solar module A to solar module B. Nevertheless, this particular spectrum has been chosen because it's a good indication of average solar power per square meter arriving at the earth's surface.

Now that I've gone through that lengthy discussion I might as well tell you how much sun power arrives at the earth's surface for the AM 1.5 spectrum. It's, conveniently, 1000 watts per square meter. Just out of curiosity, I went and read what the max sun power was according to our sensor (pyranometer) on the roof at the lab. Three days ago we had full sun and the power was impressively over 450 W for an hour or two. Being near the winter solstice, the sun rises only 6 degrees above the horizon.

Ok, onto the other part of the graph labelled quantum efficiency. The y-axis for this graph is on the right and its measure is percent. Quantum efficiency is also called spectral response, the former term being preferred by solar scientists. It's a measure of how well a solar cell can produce electric current from the incoming sunlight color by color, wavelength by wavelength. For scientists, the sunlight is "quantized" into photons, the minimum energy packets by which light can give its energy up. The quantum efficiency measures the percentage of incoming photons that will result in an electron coming out of the solar cell. For the units of incoming sun power, I could have replaced the "watts" with "photons per second".

You'll see that the top and relatively flat part of the quantum efficiency curve lies at about 90%. So if you shine pure red light, say, onto the solar cell 90% will be converted to electricity. If the quantum efficiency were 90% over all wavelengths, then you'd have a 90% efficient solar cell. But it isn't, and you don't. The reasons? Some light is reflected away and never even makes it into the cell. Some light produces overly energetic electrons that shrug off the extra energy as heat. That's not good when you'd rather just have electricity. And some light never frees up an electron at all, it just produces heat directly. And then, some freed electrons get trapped before making it out. They're forced to give up their energy and wait around for another photon to come by for another chance at freedom. The result is as you see in the graph. A solar cell has a fairly limited working range.

The particular solar cell in the graph is a pretty good one. It's a CIGS cell I made and measured and this one is my best ever at 18.5% efficient. In the graph you can see that where the sun is producing most of its power the cell is also at its most efficient, that is 90%. Going to the violet and ultraviolet there is less sun power available (thanks in part to ozone, but I digress) and what little there is is used poorly by the solar cell. For any CIGS savvy readers, you'll also notice that the characteristic absorption by the cadmium sulfide layer isn't there and you'd guess rightly that this cell is cadmium free. Going to the infrared part of the spectrum at longer wavlengths, again the sun power falls off, choppily and a bit more slowly. At a wavelength a bit over 1000 nm there's less than half the max sun power available while at the same time the quantum efficiency is down to about 50%. Half of a half is only a quarter as much power produced than at the max part of both curves (very approximately). Beyond 1200 nm, the solar cell is totally useless. Whatever sun power it picks up will only go as heat.

The information in the two graphs is sufficient to calculate the electrical current that a solar cell will (not just "can") produce, at least when the sun is looking like the AM 1.5 spectrum. Since the quantum efficiency curve is a real measurement of the current produced at each wavelength, this is considered a real measurement of maximum current output of the solar cell when the external circuit is in short circuit mode. Just take the incoming photons per second at each wavelength, multiply this by the quantum efficiency at each wavelength to get some reduced number, add up or integrate the contributions from all the wavelengths and there you have a measure of the short circuit current. This is considered a better measure of current than that measured under a solar simulator, because the simulator uses a light bulb whose spectral output differs from the sun's. The quantum efficiency method gives a wavelength by wavelength breakdown of the picture with soooo much more info.

Ok, so what's my point, anyway? Bit by bit, I'm hoping to explain why solar cells aren't better than they are. Why couldn't they be 80% efficient, or how about 60? 40% maybe? But not even 20?? That will be for another time. I'll try and make future posts much shorter than this one.

Hopefully my new blog contributors, Per-Oskar and Sebastian, will soon be spouting lots of CIGS wisdom of their own.