The photovoltaic community is closer than ever to achieving ultra-high multijunction solar cell efficiencies (>50%). Subcells from III–V compound semiconductors are approaching ideal Shockley–Queisser behavior and emit significant radiation of photons with energies equal to or above the optical bandgap because non-radiative recombination has been minimized with advanced growth processes. The optical environment of a solar cell controls where the radiated photons from a subcell are directed, and this greatly affects its efficiency. Thus the optical design of multijunction architectures is crucial for maximizing performance. To date, light trapping and radiative coupling have been investigated as promising optical design strategies. Light trapping inhibits the radiative emission of a subcell in order to reduce the dark current and increase voltage.

Schematics of various multijunction cell architectures. Solid arrows denote photons that are radiatively coupled from the blue subcell to the green subcell; dotted arrows denote radiatively emitted photons that are trapped in the same subcell. Structures (a) - (c) represent traditional multijunction architectures that have been studied previously. The selective reflector structure (c) is the most efficient and has no radiative coupling. Structures (d) and (e) represent more novel spectrumsplitting architectures in which subcells are spatially separated from one another.
This work investigated different multijunction architectures to provide insight on how their optical environments affect overall efficiencies. A simple model was employed to understand how radiative coupling between subcells with back reflectors can improve multijunction performance. This was compared to the previously assumed maximum efficiency case. For cells that do not utilize radiative coupling, decreases in subcell voltages and efficiencies for architectures that incorporate back reflectors on all subcells were analytically derived and experimentally verified. Increasing the radiative coupling between subcells enables these incurred losses to be minimized. Finally, the effect of radiative coupling between subcells with back reflectors for spectrum-splitting architectures was determined, as well as the overall efficiencies for these devices.

Previous time-symmetric multijunction architectures include the traditional tandem stack, the air-gap tandem stack, and the selective reflector structure. In all of these structures, subcells are stacked in order of decreasing bandgap such that the incident spectrum is divided by above-bandgap absorption of the subcells. Both the traditional and airgap tandem stack structures can radiatively couple between subcells, but the airgap tandem stack can trap some of the radiative emission in the same subcell due to the refractive index contrast at the air–semiconductor interface on both sides of each subcell. By contrast, the selective reflector structure does not have any radiative coupling.

The selective reflector has the same benefit as a back reflector for a given subcell, but it can also restrict radiative emission for the next subcell if the difference between bandgaps, or spectral window, is small enough to reflect the radiative emission of the next lowest subcell. The selective reflector structure is more efficient than the tandem stack structures for finite numbers of subcells because the strong light trapping benefit from the selective reflector greatly outweighs the benefits from radiative coupling in the other structures.

In the architectures studied here, light is split and distributed onto a set of independently grown subcells either by an external optical element or by manipulating the packing of the subcells in the structure. This work only considers time-symmetric structures to provide the best comparison to the other geometries. However, spectrum-splitting structures can exceed the efficiencies of a selective reflector structure if radiatively emitted light can be coupled between subcells that have back reflectors. Because this geometry can recycle radiatively emitted photons between subcells that have back reflectors, a higher conversion efficiency beyond that in previously studied geometries was derived.

Analysis and experimental results show the important role of radiative coupling, and how spectral window and optical environment dictate the performance of subcells in a multijunction cell. As the number of subcells increases, the photon each subcell receives will decrease, reducing the power it converts, and this dependence is exacerbated when there is a lack of photon recycling between subcells. However, if subcells can radiatively couple into other subcells, this reduction is less significant.

This work was done by Carissa N. Eisler, Matthew T. Sheldon, and Harry A. Atwater of the California Institute of Technology; and Ze’ev R. Abrams and Xiang Zhang of Lawrence Berkeley National Laboratory for the U.S. Department of Energy. DOE-0001