Approaching the frontier of solar cell conversion efficiency

The conversion efficiency of a solar cell is an important quantity in photovoltaics (PV). This is because the efficiency eventually defines the amount of electricity generated per area, determines the size of the land needed for PV installations, and ultimately sets the cost of solar electricity. Doubling the efficiency of a solar energy system reduces the cost per kWh price by more than a factor of two.

But, of course this is not the full story. Today’s highest efficiency solar cells are made of III-V compound semiconductors and the manufacturing cost of these devices is approximately 50 times higher than for a state-of-the-art silicon solar cell. Even at 41% efficiency these solar cells will never become competitive with today’s flat-plate silicon based PV modules. The energy produced by a III-V flat-plate module would simply be too expensive.

However, scientists have found a way to get around this bottleneck and still benefit from the extremely high efficiencies by turning to concentrating optics. In high-concentration PV systems sunlight is collected from an area 500 – 2000 times larger than that of solar cell (see Fig. 1), which only covers a very small fraction of the module area. This slashes the amount of solar cell material needed and cuts cell costs significantly, assuming that the optics and the mandatory tracking units can be produced at low enough cost.

In fact, solar cells costs typically account for less than 10 percent of the total cost of high-concentration PV systems. In this case the benefit of a high efficiency can overcompensate the drawback of significantly higher material costs. Due to their unparalleled performance multi-junction solar cells made of III-V compound semiconductors are the product of choice for this application.

Fraunhofer ISE has a long tradition in the development of these highefficiency solar cells and concentrator systems[1] and this January it achieved a new record in solar electric conversion efficiency of 41% at 454 suns (1 sun corresponding to an intensity of 1000 W/m2). Spectrolab has since raised the bar even further to 41.6 percent.

The conversion efficiency of a solar cell made of a single semiconductor material such as silicon is limited due to two reasons: long wavelength photons are transmitted through the structure and carriers generated by short wavelength photons quickly thermalize to the band edge, loosing part of their energy by the generation of lattice heat.

These two fundamental losses limit the theoretical conversion efficiency of a solar cell made of a single semiconductor material to 40.8 % But these losses can be reduced i.e. through the use of several pnjunctions made of semiconductors with different bandgap energies stacked on top of each other. In this case short wavelength photons are converted more efficiently by a first high bandgap cell and longer wavelength radiation is transmitted to an underlying second pn-junction which can again be followed by further lower bandgap subcells in the same way. Fig. 2 shows the theoretical conversion efficiency for such a multi-junction stack calculated for an ideal combination of bandgap energies and at a concentration of 1000 suns (corresponding to 1000 times 1000 W/m2 incident intensity). The maximum efficiency increases to 55.9% for a dual-junction configuration to 63.8 % for an ideal triple-junction solar cell. The global efficiency maximum is achieved for an infinite number of pn-junctions but one can also see from Fig. 2 that the benefit of adding more subcells decreases with the number of junctions and at the same time the complexity of the solar cell device increases significantly.

In fact every subcell in a multijunction solar cell typically exists of approximately 6-10 individual layers forming the emitter and base, as well as surrounding barrier layers and the tunnel diode for the series interconnection of adjacent subcells. Today the highest efficiencies are reached for solar cells having three pn-junctions. An example of the complete layer structure of such a device is shown in Fig. 3.

In the future it is well possible that devices with four or more junctions will show better performances. However, this requires all the layers to be optimized and to show excellent material quality. In fact material quality is a key requirement for achieving high solar cell conversion efficiencies, which explains why III-V compound semiconductors have been more successful than multi-junction cells made of polycrystalline or amorphous materials.[2] Defects in the active region of a solar cell lead to non-radiative recombination of minority carriers and reduce the voltage as well as the current of the device. For this reason it was believed for a long time that high efficiency solar cells need to be built from semiconductors with the same lattice-constant to avoid the formation of defects due to lattice relaxation.

The global efficiency maximum for a triple-junction solar cell is found for a bandgap combination of 1.74, 1.17 and 0.70 eV. Germanium with a bandgap of 0.67 eV turns out to be a good candidate for the bottom junction but unfortunately for the middle cell there is no latticematched semiconductor with a bandgap between 0.67 and 1.41 eV (see Fig. 3 left). Therefore, until recently the most successful material combination was a lattice-matched Ga0.5In0.5P/Ga0.99In0.01As/Ge (1.88, 1.41, 0.68 eV) solar cell which has reached efficiencies up to 40.1%.[3]

At Fraunhofer ISE we have followed an alternative approach which enables the use of latticemismatched semiconductor materials. In this case the lattice constant is intentionally graded between Ge and - in our case - Ga0.83In0.17As with a bandgap energy of 1.17 eV.

The challenge of this structure is to completely relax the mismatched crystal layers in a region of the device where no photocurrent is generated. In the ideal case misfit dislocations are only formed inside a spatially defined buffer region (see Fig. 4) and threading dislocations are sufficiently suppressed. This approach is called metamorphic as starting from Ge the crystal is transformed into a virtually new lattice with a 1.2 % larger lattice constant. This new lattice acts as the template for the growth of a Ga0.35In0.65P top and Ga0.83In0.17As middle cell structure which are both lattice-matched to each other. The resulting solar cell is composed of semiconductors with bandgap energies of 1.65, 1.17, 0.68 eV which is close to the theoretical optimum for a device with three pn-junctions.

It has to be emphasized again that the success of the metamorphic approach relies on the material quality that can be achieved in the photoactive parts of the solar cell structure. Therefore, the graded buffer layer is a key for achieving excellent device performance.

Our best metamorphic triple-junction solar cells are reaching efficiencies of 41.1 % (see Fig. 5) under 454 suns concentration[4], and they demonstrate the high theoretical potential as well as excellent material quality which can be realized with metamorphic growth. Still there is sufficient room for further improvements of this structure. As an example some of the most important R&D topics are the development of:

- dislocation blocking layers surrounding the buffer structure to further reduce defect densities in the active solar cell layers
- high bandgap tunnel diodes between the subcells to reduce losses due to absorption and reflection
- back surface passivation of the Ge bottom cell to improve the photocurrent of this subcell
- a broad-band anti-reflection coating which covers the full spectral range between 300 – 1850 nm
- contact fingers with higher aspect ratio to minimize shadowing losses

The potential of the metamorphic Ga0.35In0.65P/Ga0.83In0.17As/Ge triple-junction solar cell has been shown to be high and excellent device characteristics can already be achieved.

For the application in a concentrator PV system, the influences of optical elements and the operating temperature of the cells have to be considered. In the future the development of multi-junction solar cells may be more directed towards a maximum power output of a specific concentrator system rather than increasing the cell performance under the standard AM1.5d spectrum at 25 °C. This will help III-V semiconductors to be more successful in the new solar concentrator PV market with the potential to become a major player in the industry and to lower the cost of solar electricity.