Previous Up Next

Tandem solar cells

The previous section showed that a single junction solar cell makes poor use of the solar spectrum for two reasons; first, photons with energies less than the bandgap do not contribute at all. Second, each photon with higher energy contributes one electron to the current, however, all the energy exceeding the bandgap is lost. For a better use of the photon energies we could try to use different semiconductor materials and guide each photon into an absorber where the bandgap matches the photon energy.

The simplest structure of this kind is the so called tandem where two absorbers are stacked. For the combination of a high and a low bandgap material the illumination should first strike the absorber with the high bandgap because there light with high energy will be absorbed with a high output voltage. Furthermore, this material will be transparent for low energy light wich can be passed on to the second absorber with the lower bandgap.

Technologically the fabrication of tandem devices is quite demanding because the extraction of the currents is not trivial. Basically two designs exist, the monolithically integrated tandem and the mechanically stacked tandem. The basic designs are shown in the figure below.

Figure 1: Schematic design of monolithic (left) and stacked (right) tandem devices. Irradiation enters from the top into a high bandgap absorber for high energy photons ("blue light"). Light with lower energy is passed through to a low bandgap absorber (red). The monolithic approach (left) requires a transparent tunneling junction. The stacked devices (right) consist of two isolated cells, here a thin film design for the combination of a superstrate and a substrate solar cell.

Stacked devices

Stacking of indvidual solar cells into a tandem structure seems plausible and straightforward. One would usually contact the top and the bottom cells separately which is the reason why these tandems are also called four terminal devices. Due to the separate connections of top and bottom cells they do not require current matching which makes the combination of bandgaps quite flexible. An estimation of the maximum efficiencies in dependence on the two involved bandgaps is shown below. The calculation is similar to the case of the single cell of the previous section; it was assumed that the top cell absorbs all photons with energies higher than the bandgap and is ideally transparent for sub-bandgap light. Thus, the illumination of the bottom cell is through the top cell. Contributions due to radiative recombination in the top cell were omitted for simplicity.

Figure 2: Efficiency map of a four terminal tandem device. Efficiencies of more than 30 % are expected for a broad range of bandgaps.

A problem of stacked cells is the requirement of transparent contacts. Tandems based on thin film technology were first manufactured by combining an a-Si top module and a CuInSe2 bottom device [1]. Other top absorber materials still require the development of transparent back contacts. This was shown for the case of CdTe solar cells [2] whereas transparent CuGaSe2 superstrate solar cells require still some research (some of the authors contributions toward this goal are presented on this website).

In the area of crystalline devices stacking is taken a big step ahead; usually it means putting a monolithic (In,Ga)P/GaAs tandem on top of a GaSb booster cell or even another monolithic (Al,Ga)(As,Sb) tandem. The resulting triple or quadrupel cells show efficiencies well above 30%.

Monolithic integration

In monolithically integrated tandems the two cells are series connected. For a given spectral distribution the absorber thicknesses must be adopted to yield equal currents. Spectral changes during the day or throughout the year (see the chapter about illumination) may give rise to spectral mismatch. In this case the subcell with less current acts as load on the second cell and the overall efficiency is lowered.
Figure 3: Efficiency map of a two terminal tandem device. The range for high efficiencies is narrower because the condition of current matching must be met.

High efficiency tandem cells are fabricated from crystalline materials where the required flexibility in band gap energies is best met by the III-V semiconductors. Lattice matched (In,Ga)P and GaAs yield the highest reported efficiencies. These materials can be grown epitaxially on Ge single crystal substrates. In a typical design of manufacturer Spectrolab Germanium is not only used as substrate for the epitaxy but also serves as IR sensitive bottom cell in a triple junction device. Single crystal substrates and epitaxial processes make these cells quite expensive, mostly they are used for space applications.

Monolithic integration also established in thin film technolgy. Amorphous and microcrystalline silicon with their respective bandgaps of 1.8 and 1.1 eV form an ideal pair. Such cells are often called micromorph tandems which first introduced by the PV-lab of the University of Neuchtel (since 2009 part of the EPFL). The technology was first commercialized by the Kaneka Corporation). Purely amorphous tandem cells are fabricated from a-Si and a-SiGe. The bandgap in the Si-Ge system can be varied in a wide range and allows for the series connection of absorbers with three different bandgaps. Xunlight uses this concept for their triple junction cells on flexible substrates.

Note: As in the previous section, more accurate assumptios made in the literature will yield different values. Based on purely thermodynamic limitations, DeVos finds a limiting efficiency of 42.3% for a four terminal device with band gaps of 1.0 and 1.9 eV [3]. Assuming material parameters of typical devices, Nell finds an optimum efficiency of 39.1% for a four terminal device with bandgaps of 1.16 and 1.91 eV [4].


[1] K. W. Mitchell, D. Willett, C. Eberspacher, J. Ermer, J. Proc. 21st IEEE PVSC 1990 p. 1481-6
[2] A. N. Tiwari, G. Krypunov, F. Kurdesau, D. L. Bätzner, A. Romeo, H. Zogg, Progress in Photovoltaics, 12 (2004), p33-38
[3] A. DeVos, J. Phys. D 13, p839 (1980)
M. E. Nell, A. M. Barnett, IEEE Trans. Electron Dev. 34(2) (1987)

Previous Up Next