"The only thing the two materials have in common,"
says Yu, "is that they both try to capture as much of the solar spectrum
In 2002, the researchers learned that indium gallium nitride (InGaN) would respond to different wavelengths of light if the proportions of indium and gallium in the alloy were adjusted. Thus it might be possible to create a photovoltaic cell sensitive to the full solar spectrum by stacking multiple negatively and positively doped layers to form several current-producing junctions.
In their latest discoveryówhat Yu calls "a totally new material concept"óthe researchers treat the alloy zinc manganese tellurium (ZnMnTe) in such a way that a single junction of the material may be able to respond to virtually the entire solar spectrum.
"This isn't a multijunction material," says Walukiewicz, "it's even more interesting: a multigap material"óa single semiconductor with multiple band gaps.
A solar cell with the simplest possible physical structure could achieve 50 percent efficiency or better, far higher than any yet demonstrated in the laboratory.
How solar cells work
Sunlight comes in many colors, combining low-energy infrared photons with high-energy ultraviolet photons and all the visible-light photons between. Each photovoltaic material responds to a narrow range of these energies, corresponding to its characteristic band gap.
The band gap is the amount of energy, expressed in electron volts (eV), required to kick an electron from a semiconductor's valence band, which is chock full of electrons bound to atoms, into its empty conduction band, where electrons are free to move. (The bands are graphical representations, not physical spaces.)
If the semiconductor is doped with impurity atoms to form an n-type, electrically negative material, it already has a few electrons in the conduction band; conversely p-type (positive) material has been doped to leave missing electrons, or holes, in the valence band. A junction between n- and p-type creates a voltage bias; when incoming photons are absorbed, electrons migrate toward the positive side of the junction and holes toward the negative side, forming an electric current.
Discovering the multigap phenomenon
"The concept of multiband cells goes back to solar-cell pioneer Martin Wolf, who proposed the impurity photovoltaic effect in 1960," says Walukiewicz. "The idea was that by introducing impurities with the right electronic properties into a semiconductor you could make a singleñjunction solar cell that absorbs more photons with different energies. Sounds easy, but nobody knew how to do it."
But in 1999, Walukiewicz and others at Berkeley Lab were working with solar-cell designers at DOE's National Renewable Energy Laboratory, who were trying to build a three-junction cell. The NREL researchers inadvertently created the first photovoltaic semiconductor with a split band gap. But at first they didn't realize it.
"They needed a new material with a 1-eV band gap and a crystal lattice structure that matched the other layers of the cell," Walukiewicz explains. ""They used gallium indium arsenide nitride alloys in which just a little nitrogen could achieve the desired band gapóand an almost perfect lattice match."
Since the band-gap reduction was unexpected, Walukiewicz set out to find out how it worked. The answer, it developed, was that the few atoms of nitrogen, which are much more electronegative than the host atoms (much more strongly attractive to electrons) produced a narrow energy band of their own, splitting the GaInAs conduction band into two parts. The gap to the lower of the two conduction bands was the desired 1 eV.
In the case of GaInAs, other characteristics of the split bands made for a poor solar cell material. Nevertheless, Walukiewicz and his colleagues continued to investigate the phenomenon and developed a model of the split-band phenomenon known as "band anticrossing."
A perfect mismatch
A so-called highly mismatched alloy results when a few of the host atoms of a semiconductor alloy in the III-V group, like GaInAs, are replaced with nitrogen atoms having very different electronegativity. (The Roman numerals refer to the columns in the periodic table in which the constituent elements are found.) Replacing atoms in the II-VI group of alloys with oxygen, also highly electronegative, produces highly mismatched alloys as well.
Split band gaps account for the electronic peculiarities of highly mismatched alloys. As with GaInAs, in most cases the split occurs inside the conduction band, with results that are of not much use in solar cells. In some materials, however, the band-anticrossing model predicts that the impurity atoms will produce a narrow band well below the conduction band. One such prediction was that adding oxygen impurities to the II-VI alloy zinc manganese tellurium, ZnMnTe, would produce well defined and widely split band gaps.
"Figuring out how to do this was not easy," Yu says. "It was important that the oxygen atoms be distributed evenly throughout the material. To trap enough oxygen ions you have to do it with the material in the liquid state and very fast. You can't just heat the material slowly, because the oxygen is rapidly driven out."
Walukiewicz adds, "That's why they're called highly mismatched alloysóbecause the impurity atoms and the host atoms don't like each other."
Yu says, "We did it in two stagesófirst we used ion beams to implant the oxygen, then we used pulsed laser melting to liquefy the ZnMnTe and recrystallize it rapidly. The whole laser process takes just a couple of hundred nanoseconds," a couple of hundred billionths of a second.
In this way the researchers were able to create single crystals of ZnMnTe whose top layeróonly 0.2 micrometers thick (a micrometer is a millionth of a meter)óheld enough oxygen impurity atoms to split the normal band gap.
A deck of energy levels
How can a split band gap convert a wide swath of the solar spectrum to electricity? Because two separate bands means the material efficiently absorbs photons of three different energies.
The difference between the material's valence band
and the lower of the split bands forms one band gap. In ZnMnTe incorporating
oxygen impurities (written ZnMnOTe), this first gap absorbs 1.8 eV photons.
The difference between the two split bands is a second band gap; in ZnMnOTe, this gap absorbs 0.7 eV photons. Finally, the difference between the valence band and the upper conducting band forms a third band gap; in ZnMnOTe, this gap absorbs 2.6 eV photons.
Together, these three gaps respond to virtually the entire solar spectrum. The calculated efficiency of a single-junction solar cell made with this material would be a remarkable 57 percent. But while the single-junction architecture is elegantly simple, many questions have to be answered before ZnMnOTe or any of its highly mismatched cousins prove they can do the job.
Making p-type and n-type versions of the split-band material does not appear to pose a problem. But the tricky process of ion implantation followed by pulsed laser melting is no way to manufacture semiconductors in bulk. And the oxygen-implanted layer must be at least 0.5 micrometer thick if the material is to absorb all the solar photons falling on itómore than twice the 0.2-micrometer thickness achieved so far.
Yu admits that forming highly mismatched alloys is "challenging from a crystal-growth point of view," but there is hope that crystals can be grown epitaxially. One good sign, he says, is that Japanese researchers have already grown thick oxygen-doped crystals of a related material, zinc selenium.
In the meantime, the Berkeley Lab researchers have teamed with Piotr Becla of MIT to manufacture a single junction of the material which gives a photovoltaic response. With it they have demonstrated the kind of three-band semiconductor needed for high-efficiency, single-junction solar cells.
There are many possible ways of varying the composition of these alloys to get the desired resultóso many that eventual success seems as highly likely as the alloys are highly mismatched.
"Diluted II-VI oxide semiconductors with multiple band gaps," by Kin Man Yu, Wladek Walukiewicz, Wei Shan, and Jeff Beeman of Berkeley Lab, Mike Scarpulla and Oscar Dubon of Berkeley Lab and UC Berkeley, and Piotr Becla of MIT, appeared in Physical Review Letters, 12 December 2003.
More about indium gallium nitride for solar cells
More about the structures of solar cells