Photon frequency management can benefit from recent advances in photonics and this section provides a brief outline of different ways to increase the efficiency of collectors beyond the TIR efficiency limit.
Spectrally selective filters have been proposed for application to luminescent solar collectors in order to reduce the light losses through the TIR escape cone. The filters can be fabricated from photonic crystals. A photonic crystal has a spatial periodic variation in its dielectric constant and prevents light of certain energy propagating in certain directions.65 The top face of the collector can be covered by a photonic structure (a band stop) that reflects the fluorescent light (Figure 9.20) blocking much of the escaping light and reducing photon transport losses to a minimum. Simulations carried out by Goldschmidt and co-workers66’67 have shown 20% increases in efficiency with band edge filters. An increase in edge emission with the use of photonic opal crystals was also reported by Knabe et al.68 Debije et al.69,70 observed up to 12% more light collected on the edge of the collector by using
Figure 9.20 (a) Schematic of an advanced fluorescent collector where a photonic
band stop covers the entrance aperture, with the reflectance profile shown in (b). Reproduced from ref. 5.
wavelength-selective mirrors consisting of chiral nematic (cholesteric) liquid crystals applied on the top of the collector using an air gap and when used in conjunction with a separate white scatterer layer on the back of the collector.
Recently researchers have started attempting to improve photon collection efficiency by restricting the molecular orientation of the dye inside the collector. For example, a planar aligned dye with its absorbing axis parallel to the edge of collection will have an increased light output and a respective reduction edge light output in a direction perpendicular to the absorption axis. Using liquid crystal hosts to hold common dye molecules such as Coumarin 6 and DCM71 aligned, Debije et al. vo carried out measurements under polarised light and showed improvements up to 30% in collected light with planar dye alignment with respect to no dye alignment (isotropic). Baldo and co-workers71a in a similar experimental set-up saw improvement in collection efficiencies of 23% with respect to isotropic dye alignment. These results clearly demonstrate the potential of increasing collection efficiencies in fluorescent collectors by controlling the orientation of the dye molecules. However, doubts remain about losses introduced by a reduction in absorption72 and more work is clearly required to substantiate the potential for improvement using this approach.
Fluorescent collectors provide an elegant technique for exceeding the Shockley-Queisser limit by spectral splitting. Stacked collectors were introduced by Goetzberger and Greubel17 in the 1970s (Figure 9.21). Each plate absorbs part of the solar spectrum and re-emits it onto a small solar cell. Theoretical conversion efficiencies have been estimated to exceed 30%, but practical results remain well below this limit.
A radically new solution to enhance the photoexcitation of silicon by directing the illumination onto the edge of the solar cell by means of fluorescence energy collection has recently been proposed by our group.73 The generic structure of the new device is shown in Figure 9.22(a). A fluorescent collector, adjacent to a thin crystalline silicon solar cell, captures all or part of
Figure 9.21 (a) Schematic of a cross-section of a fluorescent collector stack. (b)
Example of absorption and emission spectra of the stacked layers. Reproduced from ref. 27.
Figure 9.22 (a) Solar cell illuminated from the edge by a fluorescent collector. (b) An
alternative ‘checkerboard’ arrangement, making use of strong silicon absorption at short wavelengths. (c) The absorption and re-emission paths of the red and blue parts of the spectrum. Reproduced from ref. 73.
the incident solar radiation and emits this energy in the form of fluorescence with frequency near the silicon bandgap. In an optimised structure shown in Figures 9.22(b) and 9.22(c), the collector absorbs only the red/near infrared part of the spectrum but the short wavelength radiation is absorbed directly by the thin silicon cell.
If attached to a 1 pm thick crystalline silicon solar cell with nominal 20% efficiency and illuminated from the edge at standard testing conditions, the edge illuminated ultra-thin silicon solar cells using fluorescent collectors can produce conversion efficiencies close to conventional c-Si solar cell but with greatly reduced material requirements as shown in Figure 9.23, depicting the quantum and total efficiencies. Higher efficiencies can still be achieved with thicker cells, different emission wavelengths or a different photon
Figure 9.23 (a) The spectral management of device operation illustrating the
absorption and fluorescence channels and a possible absorption spectrum of the dye. AM1.5 spectrum, W m~2 pm-1 or dye spectrum (arb. p.). (b) Calculated quantum efficiencies, showing contributions due to photons absorbed in different parts of the device. (c) Overall efficiencies (cell + collector) for 1 pm c-Si solar cell operating with a collector based on a Nd+3 or Yb+3 emission channel. Reproduced from ref. 73.
management strategy.7 In a practical setting, current developments in fluorescent collectors bear the promise of inexpensive practical devices with efficiencies in excess of 10%.
As suggested in Section 9.2.1, frequency management can be used not only to concentrate light but also to enhance light absorption by increasing the path length of light in the solar cell. Figure 9.24(b) shows a structure taking advantage of frequency shift to trap light inside a thin weakly absorbing c-Si solar cell, to be contrasted with a conventional light trapping schemes based on a textured rear surface as shown in Figure 9.24(a).5’74 We have shown that this photonic scheme has the potential to increases the photon path length by a factor proportional to the Boltzmann factor of the frequency shift.23 The photonic band stop now takes over as the principal bandgap that governs the operation of the solar cell. Somewhat surprisingly we find that a 1 pm thick c-Si solar cell with a photonic bandgap is not only highly effective in trapping light but can exceed slightly the efficiency of a standard c-Si solar cell!23
Figure 9.24 (a) Light trapping scheme with a textured rear surface, showing
external rays © within the ‘escape cone’ and trapped rays ©. (b) A photonic scheme where the absorbing/fluorescent layer at the back surface introduces a frequency shift. Reprinted with permission from ref. 23, Copyright 2011, AIP Publishing LLC.
In this chapter we have presented a unified overview of fluorescent collectors and down-shifting structures in application to solar cells. We have shown how different photon frequency management methods can be used to increase solar cell efficiencies and lower the cost significantly by reducing the size as well as the thickness of the solar cell. A simple two-flux model has been presented which can be used to describe the re-absorption losses and collection efficiency of the collector using as input only the absorption spectrum and edge fluorescence measurements. The application of photon frequency management materials that employ efficient Forster resonance energy transfer can be applied in the near term for increasing the efficiencies of commercial CdTe solar cells via down-shifting/light guiding or for the fabrication of fluorescent collectors with reasonable efficiencies approaching 10%. Looking further into the future, the employment of photonic structures with fluorescent collectors or into light trapping structures offers the promise for substantial increase in solar cell efficiencies even beyond the theoretical limits for single-junction solar cells with a significant lower material demand than for current conventional methods.
The authors would like to thank the Engineering and Physical Sciences Research Council for financial support through the PV21 Supergen consortium.
 Re-crystallisation of the CdTe resulting in grain growth
iii Passivation of the grain boundary defects.