There are many options for improving the performance of III-V solar cells, including inserting quantum wells and dots to extend spectral coverage and adding nanoparticles and diffraction gratings to boost light trapping. Insights into all these approaches are outlined by Sudha Mokkapati, Samuel Turner, Haofeng Lu, Lan Fu, Hark Hoe Tan and Chennupati Jagadish from The Australian National University

Efficiency is the key metric for solar cells. Increasing this ups the energy produced by the cell, and in turn cuts generation costs and trims the footprint of the cell required to deliver a given output.

One option for enhancing the performance of a conventional solar cell is to modify its design by inserting structures that stretch the spectral absorption to longer wavelengths. Adding quantum wells and dots can realize this, with greater spectral absorption increasing short-circuit current densities that could ultimately boost the overall efficiency of the solar cell. Gains in current resulting from the incorporation of wells or dots also open up new opportunities for bandgap engineering, allowing better current matching between different sub-cells of a tandem solar cell. For example, in a Ge-InGaAs-InGaP tandem solar cell, inserting InGaAs quantum wells or quantum dots into the InGaAs middle sub-cell transfers current from the current-overproducing bottom sub-cell to the current-limiting middle sub-cell.

A handful of research groups have pursued these approaches to improving solar cell performance. For example, Keith Barnham’s group at Imperial College, London, have increased the spectral response of an AlGaAs reference cell by turning to an GaAs-AlGaAs quantum well solar cells, while our team at The Australian National University have enjoyed similar success by replacing a GaAs reference cell with a InGaAs-GaAs quantum dot solar cell (see Figure 1 for details of the bandstructure of these novel cells).

Quantum wells and dots provide confinement of electron and holes in one and three dimensions, respectively. Both are grown epitaxially by techniques such as MBE and MOCVD. The growth of wells is very well established in academia and industry, while the formation of dots is predominantly carried out in university labs. 

Dots are produced by depositing a material with a larger lattice constant on a substrate with a smaller one – in our case we grow an InxGa1-xAs film on GaAs using a MOCVD reactor (see figure 1). Initially, the material with a larger lattice constant material tries to grow in a layer-by-layer fashion, adjusting its lattice spacing to match that of the substrate (see Figure 1). This leads to a build up of strain in the structure, which creates defects when it exceeds a certain value. To prevent this from happening, the thickness of the deposited film must be kept below a certain value, which is governed by lattice mismatch.





Figure 1: The formation of quantum dots in the Stranski-Krastanov growth mode (a) and the MOCVD reactor used for growing QWs/QDs (b). The relative bandgap energies of InxGa1-xAs quantum dot and quantum well (or wetting layer) with respect to bulk GaAs (c) .

When growth of the deposited film continues beyond this limit, strain in the system increases and beyond a certain threshold three-dimensional islands or quantum dots begin to form. That’s because for this configuration, the total surface energy is lower than the strain energy for two-dimensional growth. Beneath the quantum dots, a thin two-dimensional layer, known as a wetting layer, still exists. This is essentially a thin quantum well. Formation of quantum dot heterostructures by this approach is said to employ the Stranski-Krastanov growth mode.

A severe limitation associated with both quantum wells and dots is the very small absorption volume. For example, the absorption fraction of a single 7 nm In0.21Ga0.79As quantum well is only of the order of 1 percent beyond the bandgap of GaAs. So, to absorb a significant fraction of light beyond the band edge of GaAs, there needs to be a substantial increase in the absorption efficiency of the quantum wells and quantum dots.

One way to do this is to stack layers of quantum wells/dots on top of one another. However, the high level of strain in these quantum-confined absorbers means that when several of these layers are stacked together, excess overall strain can spawn the formation of dislocations, which act as non-radiative recombination centres that degrade device performance. What’s more, stacking too many quantum well/dot layers together may make it very challenging to extract carriers from the middle layers. Consequently, alternative approaches are needed to increase the absorption efficiency of long-wavelength light for efficient quantum well/dot solar cells.

A hike in the absorption probability is possible via a process known as light trapping - the ‘folding’ of light into a thin absorber layer to increase light-matter interaction times. Folding or trapping light in the absorber layer increases the optical thickness of the absorber layer to values far beyond its physical thickness. In an absorber layer that supports only a few waveguide modes, light is trapped by coupling it to the waveguide modes. This switches its propagation direction from perpendicular to the absorber to parallel to it. For relatively thick absorbers that support a continuum of optical modes, light is trapped by exploiting total internal reflection (see Figure 2). The strength of this approach is that light is coupled into the absorber outside of the escape cone (or at angles larger than the critical angle for total internal reflection at the absorber-air interface). We work with substrate-based structures that support a continuum of optical modes and feature either quantum well or dot absorber layers.

We employ two light trapping approaches to enhance the absorption efficiency of long wavelength light in our quantum well/dot solar cells: plasmonic light trapping and the addition of a diffraction grating on the surface of the solar cell.

Powerful plasmonics

Light trapping is possible using plasmonic nanoparticles – tiny metallic structures that interact with the incoming light and exhibit local oscillations in the density of their free electron gas. One consequence of this phenomenon is that incoming radiation is scatterred into the solar cell at angles outside the escape cone.

Good light trapping with plasmonic structures is possible by exceling in two areas: maximizing the scattering cross-section of the plasmonic nanoparticles, and realising a high efficiency for the coupling of scattered light  into the substrate. It is vital to maximise the scattering cross-section, because this ensures that the nanoparticles interact with most of the light incident on the solar cell and randomise its direction. Meanwhile, it is critical to achieve a high coupling efficiency of scattered light into the substrate of the solar cell, because this minimises reflection or transmission losses at each encounter between the weakly absorbed light and the plasmonic particles. It is also important to address parasitic losses, which are inherent to the resonant metallic nanostructures.

We form our plasmonic nanoparticles by depositing a thin silver film on the surface of a finished solar cell and annealing it in a nitrogen atmosphere. Nanoparticle arrays result from the difference in thermal expansion between the metallic layer and the substrate (see Figure 2). The fabrication process is relatively easy, since it does not require modification of the solar cell fabrication process and is scalable to large areas.

Last year we reported an 8 percent enhancement in the efficiency of a quantum dot solar cell through the addition of plasmonic light trapping. The improvement in power predominantly came from an increase in short-circuit current density (Jsc) by 5.6% − the open-circuit voltage (Voc) also went up, but just by 0.9 percent.

A good indicator of the effectiveness of a light trapping strategy is the enhancement in path length. This is defined as the ratio between the average distance travelled by weakly absorbed light in a solar cell featuring light trapping, to the distance travelled in a planar solar cell. Based on experimental results from our plasmonic quantum dot solar cell, we calculate that this path length enhancement is approximately 2 at a wavelength of 1000 nm. We can explain why this value is much smaller than expected from a good light trapping structure: Our solar cells are fabricated on n+ substrates, so free carrier absorption (parasitic losses) in this platform limits the benefits of this light trapping structure. We expect that parasitic losses in the device can be cut, leading to an increase in the short circuit current density and the efficiency of the solar cell, by fabricating solar cells on semi-insulating substrates.

Dielectric grating gains

 Another way to couple incoming light into diffraction orders outside of the escape cone in the solar cell with wavelength-scale diffraction gratings. These periodic structures are difficult to fabricate and require a lot more process optimisation, but they do not have the parasitic losses inherent to metallic structures. The latter strength indicates that they should be better for light trapping applications in solar cells, where every available photon is valuable.

When gratings are fabricated on the surface of a solar cell, they restrict the number of optical modes in the device to which the incident light can couple. The number of diffraction modes in a material depends on the periodicity of the grating and the refractive index of the cell. Employ a very small period grating, and only the principal diffraction mode in the solar cell that lies within the escape cone is supported. Turn to a very large period grating, and a continuum of modes can then be present in the solar cell and in air. However, a large fraction of the diffraction modes supported in the solar cell are within the escape cone and do not effectively trap light.

The sweet spot is to use wavelength-scale diffraction gratings (such as the ones shown in Figure 2), which support the principal diffraction mode in air, plus a few higher order diffraction modes that lie outside the escape cone. Good light trapping is then possible, thanks to efficient coupling of incident light to these higher order modes (see figure 2).



Figure 2 : Light trapping in thick solar cells by total internal reflection of weakly absorbed light (a). Scanning electron microscope images of a plasmonic nanoparticle array (b) and a diffraction grating (c) that are both fabricated on a solar cell. (d) shows how a plane wave front incident on a solar cell (left) is modified by the presence of a plasmonic nanoparticle (center) and a dielectric gratig (right) on the rear surface of the solar cell.

The key to efficient light trapping is two-fold: It involves maximising the relative number of diffraction modes outside the escape cone in the solar cell, with respect to the number of modes supported in air; and realising efficient coupling of incident light to these modes. The most efficient structures for light trapping are bi-periodic gratings with asymmetry introduced into the grating structure to satisfy these criteria. According to our calculations, the addition of an optimised asymmetric TiO2 grating structure at the rear of the solar cell can increase the short circuit current density of a ten-stack In0.21Ga0.79As-GaAs quantum well solar cell from 1.03 to 3.30 mA/cm2.

A common benchmark for assessing the performance of a light trapping strategy is the theoretical concept of Lambertian light trapping or the isotropic limit. This has been developed in the context of wafer-based silicon solar cells. The best that one can do is a path length enhancement of 4n2, (where n is the refractive index of the solar cell material), which is achieved using a Lambertian scatterer at the solar cell surface. It is possible to mimic a Lambertian surface by texturing the solar cell surface with random pyramids with feature sizes of the order of few tens of microns. 

To gauge the effectiveness of the plasmonic structures and the dielectric gratings − and compare with Lambertian light trapping − we investigate the relative short-circuit current-density enhancements (ratio of Jsc enhancement from a given light trapping structure to the maximum possible enhancement in Jsc assuming all of the incident light is absorbed in the quantum wells) from the quantum well region of a ten-stack In0.21Ga0.79As-GaAs quantum well solar cell. At the optimum geometry, a TiO2 dielectric grating can deliver a relative Jsc enhancement of 82 percent. This increase in current is much larger than the 28 percent that we obtained using plasmonic light trapping, and comparable to that of a Lambertian scatterer (95 percent). As mentioned earlier, the inferior result for the plasmonic light trapping structures is partly due to the absorption in the metal itself, which leads to very low short circuit current density enhancements compared to the dielectric structures that do not absorb the incident light.

We are now working on fabricating plasmonic solar cells on semi-insulating substrates, because this will minimize the free carrier absorption in the substrate. In addition, we are optimising the fabrication processes for wavelength-scale TiO2 diffraction gratings, so that these light trapping strategies can then be used in thin film III-V semiconductor solar cells, where epitaxially grown thin film absorbers can be lifted-off from the substrate and supported on an inexpensive substrate with light trapping strategies. This approach will allow the expensive substrates to be re-used, thereby reducing material cost in solar cell manufacturing.

 We acknowledge the Australian Research Council (ARC) for financial support, Australian National Fabrication Facility (ANFF), A.C.T. node for access to equipment and the National Computational Initiative (NCI) for computational resources used for this work.

 Further reading

K. W. J. Barnham et al. Physica E: Low-dimensional Systems and Nanostructures 14 27 (2002)

H. F. Lu et al. Appl. Phys. Lett. 98, 183509 (2011).

H. F. Lu et .al Appl. Phys. Lett. 100, 103505 (2012).

 S. Turner et al. "Periodic Dielectric Structures for Light-Trapping in InGaAs-GaAs Quantum Well Solar Cells," Optics Express (submitted).