Spectral Splitting

Statement: To substantially increase photovoltaics efficiency we need to move past the single junction cells used today. I have done this by dividing the colors of light and using different cells technologies for each spectral region.


Photovoltaic efficiency of single junction (bandgap) materials is inherently limited to little more than 25%. This because the solar spectrum has wide spectrum (as visible with the formation of rainbow) while a single bandgap converter (e.g. Silicon cell) can convert efficiently in a narrow spectral region, behaving poorly outside of it.

Vertical multi-junction cells (VMJ) , where multiple cells with different bandgaps (but strict lattice matching!) are stacked on top of each other, address this issue with demonstrated efficiency exceeding 40%, but their prohibitive cost relegates them to space application or to the higher region of solar concentration (and serious cost decrease has been a mirage in the last 10 years!)

Can we do things differently?

Alternatively we can divide the solar spectrum (via a prism, diffractive grating, hologram, dichroic mirror, Christiansen filter and many other approaches attempted in the years) into multiple spatial regions with limited spectral width (e.g.  “red”, “green”,“indigo” regions) and use different cells for every region!

These cells do not have the lattice requirement of VMJ (as they are all separated!) allowing different technologies (e.g. Silicon with CIGS and CdTe) to operate together. Efficiency can be as high (or higher) than with VMJ.

This is what I successfully attempted to do during several years.

First attempts

By thermoforming a flat faceted reflective concentrator, and adding a second, similar but with an angular ofsett  transparent shell lined with a plastic dichroic mirror film, we created our first spectral splitting system.

At that time, however, few alternative solar cells that could be used under concentration were available…

So I got involved in European Projects, Italian National Projects and, eventually, MIT-Masdar Institute co-sponsored projects to fill this gap.

During these projects I devoted myself to the optical spectral splitting design. I moved from Dichroic mirrors (that I deem expensive and with strong angular dependence) to the realization of a plastic concentrator system that, due to the natural dependency of the material refractive index from wavelength (dispersion) allows to simultaneously concentrate and spectrally split the incoming light.

Similarly to the flat faceted concentrator idea, I assembled multiple prims shaped and positioned to concentrate and split according to the design requirement. First I designed a linear concentrator, that, while of poor surface finish, performed according to design.

The, lately, I was abel to overcome some mathematical difficulties and proceeded by injection molding of policarbonate to realize the 2-d concentrator and splitter.

Again, while the device sort of works, it is affected by unacceptably low surface finish and I’m still trying to adapt the design to the manufacturers capabilities (and to get them interested in realizing few prototypes!).

On the side of the cells, while MIT provided cells where still based on a layer of III/V materials and were ultimately unsuccessful, my project with the Italian CNR resulted in the development of an horizontal graded junction of Cu(InxGa1-x)Se2  with composition varying from almost pure low bandgap CuInSe2 (B.G approx 1 eV) to Ga rich composition with bandgap above 1.6 eV. This horizontal junction perfectly matches the continuously split spectrum generated by the prismatic concentrator providing the sought after low cost alternative to vertical high concentration multijunctions.