Conventional multi-junction cell measurements only yield the characteristics of the entire device. In stark contrast, a novel electroluminescence approach can probe far deeper, extracting the current-voltage curves for individual sub-cells that hold the key to optimising the overall conversion efficiency, say the technique’s pioneers, Raymond Hoheisel, Sebastian Rönsch, Frank Dimroth and Andreas Bett from the Fraunhofer Institute for Solar Energy Systems and Helmut Nesswetter and Claus Zimmermann from EADS Astrium.

Conversion efficiency is the key characteristic for solar cells. Increase this and photovoltaic installations become more attractive because fewer cells are needed to generate a given power. This makes the installation more affordable and also reduces the amount of space required for the photovoltaic system.

The most efficient solar cells utilize a very high proportion of the sun’s radiation, which spans a spectral range extending from the ultraviolet to mid-infrared. To capture this radiation and convert it to electrical power effectively photovoltaics employ a collection of sub-cells designed to operate in different spectral ranges. In these devices several p-n junctions made of semiconductors with different bandgap energies are stacked on top of each other.

One way to optimise the overall efficiency of these multijunction devices is to first characterise each sub-cell, and then optimise its contribution to the overall performance. Extracting this information is far from easy, but our partnership between the Fraunhofer Institute for Solar Energy Systems and EADS Astrium has pioneered a technique that can do just this, based on electroluminescence (EL) measurements. Today, no other technique can yield the information that is garnered by this approach.

We have used this EL technique to investigate multijunction solar cells made of the semiconductors GaInP, GaInAs and germanium, which together can yield record efficiencies of more than 42 percent under concentrated sunlight. Our technique is not restricted to this class of device and can also be applied to multi-junction cells based on organic materials, silicon and chalcopyrites.

The monolithically stacked, triple-junction photovoltaic cells that we are studying power satellites and generate electricity on earth, where they are deployed in systems that use mirrors or lenses to focus light by a factor of several hundred. For both these applications, characterizing individual sub-cells not only aids device development – it also enables qualification testing during processing (see Figure 1)

 



Figure 1: (a) GaInP top cell EL image of an 8 x4 cm2 triple junction space solar cell showing lateral inhomogeneities and (b) GaInAs middle cell EL image of a mechanically damaged cell. None of this information is accessible by (c) a standard optical inspection

 

Lateral and vertical characterization

Our EL technique involves forward biasing of the solar cell so that it operates as a light-emitting device. Each sub-cell has a different bandgap, so emits a different range of wavelengths (see Figure 2).

 

 



Figure 2: Schematic of a GaInP/GaInAs/ germanium lattice matched triple-junction solar cell. The bandgap energies and emitting wavelengths of each subcell are indicated

 

EL-emission can be detected in different ways. By recording the EL intensity distribution with a CCD sensor it is possible to expose lateral inhomogenities, a particularly important consideration in large space solar cells. Inserting output filters before the detector can select single sub-cell performance, allowing crystal defects that act as shunting paths to be detected, even if they only affect one particular sub cell. It is also possible to use this approach to provide qualitative quality control, because this technique can expose material issues, such as cracks (see Figure 1b). In comparison, standard optical inspection reveals none of the lateral and vertical information accessible via ELcharacterization (see Figure 1c). Even more detailed characterization of the cells can be realized by recording the EL-images of all of the sub-cells at a range of injection current densities.

Extracting cell I-V curves

The EL technique that we are pioneering also yields additional, incredible valuable information – the currentvoltage curves of individual sub-cells, information that can drive improvements in multi-junction device performance and enable the introduction of dedicated in-line inspection procedures.

Current-voltage curves for individual sub-cells cannot be accessed with a sun simulator because the cells are monolithically connected in series and metal contacts are only on the front and back sides of the device (see Figure 2). The sun-simulator is limited to characterising the current-voltage curve of the triple-junction, and can also extract values for the short-circuit current, the opencircuit voltage, the fill factor and the efficiency of the entire device.

The foundations of our EL technique are theoretical developments detailed in two landmark papers. Uwe Rau from Forschungszentrum Jülich wrote one of these; the lead author for the other was Rau’s colleague Thomas Kirchartz. Co-authors on this paper were Rau; Anke Helbig and Jürgen H. Werner from the Institute for Physical Electronics, Stuttgart; and Martin Hermle and Andreas Bett (an author of this feature) who both work at the Fraunhofer Institute for Solar Energy Systems (see Further Reading for paper references).

One cornerstone of this theoretical work is the so-called reciprocity relation. By measuring the spectrally resolved electroluminescence signal and the external quantum efficiency for each sub-cell, it is possible to access its dark current-voltage and illuminated current-voltage characteristics.

We have adopted this approach to study aGa0.50In0.50P/Ga0.99In0.01As/germanium lattice-matched triple-junction cell, extracting information that can be used to predict the power performance of this photovoltaic device under realistic operation conditions.

The spectral reciprocity relation connects the intensity of  the EL signal with the energy of an emitted photon, the internal voltage of the sub-cell and the external quantum efficiency of the sub-cell. This efficiency depends on the black body photon flux, which is described by Planck’s formula.

 

 



The first step is to measure the EL at a range of injection current densities when no light is incident on the cells. Once this is complete, the external quantum efficiency for each of the sub-cells is measured (see Figure 3). By applying the spectral reciprocity relation, the voltage of each sub-cell can then be calculated as a function of the dark-current density – this provides the dark currentvoltage characteristics for each sub-cell.

 



Figure 3: (a) Spectrally resolved EL spectrum at a constant injection current density of 37.5 mA/cm2 and (b) external quantum efficiency (EQE) of a 2x2 cm2 lattice-matched Ga0.50In0.50P/Ga0.99In0.01As/germanium triple-junction solar cell. Each EL peak is related to one of the sub-cells. The three EL peaks emerge at the same energy as the declining slope of the corresponding EQE because both are related to the band gap energy of the sub-cell

 

Once the ‘dark’ current-voltage characteristics of the respective sub-cells are known, the illuminated currentvoltage characteristics of individual sub-cells and of the whole multi-junction device can be easily calculated for any desired spectral condition using superposition principles.

Our studies have shown that there is excellent agreement between the resulting current-voltage curves derived by EL-analysis and those yielded by a sun-simulator (see Figure 4). This figure illustrates the great strength of the EL-characterization technique: In addition to predicting the combined current-voltage characteristics of the multijunction solar cell, it reveals, in detail, all sub-cell currentvoltage characteristics, including fill-factors and absolute voltage properties.

 

 



Figure 4: I-V characteristics with fill factor (FF) and efficiency (η) of the sub-cells and the triple-junction solar cell (3J) under AM0 illumination conditions. The plot includes also the directly measured I-V curve of the triple-junction solar cell (3J) by a sun simulator

 

In principle, it is possible to derive the current-voltage curves under any spectrum of interest with our approach. These curves are extracted by inserting values for the photocurrent density of each sub-cell, which can be found by integrating the measured quantum efficiency with the new spectral operation condition. Based on the resulting current-voltage curves it is possible to estimate the performance, efficiency and fill factor – both at the subcell and multi-junction level – under the desired spectral operation condition of interest (see Figure 5). Catering for resistance effects is easy, too.

 

 



Figure 5: I-V characteristics with fill factor (FF) and efficiency (η) of the subcells and the triple-junction solar cell (3J) under illumination conditions with reduced illumination in the absorption range of the top cell. The plot includes also the directly measured I-V curve of the triple-junction cell (3J) by a sun simulator.

 

Our development of the EL technique has shown that it has a great deal to offer to both developers and manufacturers of multi-junctions cells. It provides qualitative and semiquantitative information on lateral and vertical inhomogeneities, making this approach an attractive technique for fast inline inspection during the cell growth process and module assembly. In addition, our EL technique is unique in being able to yield detailed information concerning the currentvoltage curves of individual sub cells, which hold the key to precise device improvement. What’s more, fill-factor behaviour and the efficiency under arbitrary illumination conditions can be determined by our novel approach, aiding long-term energy harvesting analysis.

© 2011 Angel Business Communications. Permission required.

Further reading

M. A. Green et. al. Progress in Photovoltaics: Research and Applications 19 84 (2011)

H. Yoon et. al. Progress in Photovoltaics: Research and Applications 13 133 (2005)

M. Yamaguchi et. al. Solar Energy Materials and Solar Cells 90 3068 (2006)

C. G. Zimmermann IEEE Electron Device Letters 30 825 (2009)

C. G. Zimmermann Journal of Applied Physics 100 23714 (2006)

S. Roensch et. al. “Subcell I-V-characteristic analysis of GaInP/GaInAs/Ge solar cells using electroluminescence

measurements”, to be published.

U. Rau et. al. Physical Review B 76 085303 (2007)

T. Kirchartz et. al. Appl. Phys. Lett. 92 123502 (2008)