Luminescent solar concentrators: the road to low-cost energy from the sun

Photovoltaics (PVs) enable conversion of solar light into electricity without detrimentally affecting the environment. Unfortunately, solar cells require high-cost semiconductors, which makes them expensive. Concentrator systems help to minimize the coverage needed, but normal concentrators only work for a narrow angular range of solar radiation. Moreover, they need tracking devices and are poorly suited to capturing diffuse radiation. Luminescent solar concentrators (LSCs) avoid these disadvantages.^1,2 An LSC consists of a glass or plastic plate containing or coated with luminophores (dyes or phosphors) that absorb sunlight and emit light at longer wavelengths (see Figure 1). A substantial part of the longer-wavelength light is trapped by total internal reflection and guided to the edges of the LSC plate, where it is absorbed by small-area PV cells.

The idea of using LSCs already dates back 30 years,^3,4 and much research has been carried out in this area.^5–8 However, up to now, the technology has not delivered on its promise. Although theoretically concentration could be huge,^1,9 the best LSC-based PV system to date¹⁰ has a conversion efficiency of 7.1% and saves a factor 2.5 of PV-cell area. The reason for these low numbers is primarily losses of various kinds. For example, light is absorbed by the luminophore, but no luminescent light is emitted because of limited quantum efficiency (QE), i.e., the ratio of absorbed to emitted number of photons. Another important means of loss is escaping luminescent light from the concentrator at angles larger than the critical angle for total internal reflection (see Figure 1). One effective way to prevent this is by applying a wavelength-selective filter on top of the light guide,^1,11,12 which keeps the luminescent light inside. However, such filters may also prevent solar radiation incident at high angles from entering the light guide. We explored the effect on LSC efficiency through ray-tracing simulations.¹³

Figure 1. (left) Luminescent solar concentrator with luminophores (dots) absorbing sunlight (dashed) and emitting longer-wavelength light (solid line) that is absorbed by a photovoltaic (PV) cell. (right) Luminescent plates with light guided towards the edges. (Photo courtesy of P. P. C. Verbunt.)

Figure 2(a) shows the part of the solar spectrum (photon irradiance) that can be converted into electrical energy by a silicon solar cell (band edge 1100nm). In our simulations, we use an idealized luminophore with a rectangular absorption spectrum and a narrow emission line. Figure 2(a) shows the reflection band of a filter that reflects at and longwards of the wavelength of emitted radiation. Suitable filters are Bragg reflectors based on dielectric stacks⁷ or cholesteric liquid crystals¹² with period d. According to Bragg's law—λ=2nd cos θ—at larger angles θ, the reflection band shifts to smaller wavelengths λ. (n is an integer defining the interference order.) We consider the case that the filter reflects the emitted radiation at all angles smaller than the angle of total internal reflection (42^°). If the absorption band of the luminophore is completely outside this range, all sunlight of interest will be transmitted to the light guide. However, for a small distance between absorption and emission bands, part of the incident light will be reflected. Simulations show that the optimal distance between emission-line and absorption edge is 100–150nm, since at short wavelengths not much sunlight is available.

Figure 2. (a) Solar spectrum (photons/m²/nm), absorption and emission spectra, and reflection bands at 0 and 42^°for a wavelength-selective filter of reflection width 280nm. (b) Collection probability versus self-absorption of emitted radiation relative to that of absorbed radiation, with and without a filter. QE: Quantum efficiency.

Figure 2(b) shows additional simulation results. We consider the collection probability of the LSC, that is, the ratio of the number of photons collected by the solar cells to the total number of incident photons. We varied the amount of self-absorption, in other words, the ratio of the absorption coefficient for the luminescent radiation to that of the incident absorbed radiation. We also varied the QE. With a wavelength-selective filter, escape of the emitted radiation is prevented and, for QE=1, the collection probability is independent of self-absorption (assuming that absorption by the plate material is negligible compared with that by the luminophore). Without a filter, the collection probability approaches 75% for small self-absorption and decreases with increasing self-absorption. For QEs <1, the collection probability decreases both with and without a filter, but the enhancement provided by the filter can still be as large as a factor of 3.

We conclude that high collection probabilities can be realized if certain conditions are met. The most important is small self-absorption. Collection probabilities above 70% are possible if the self-absorption is a factor of 1000 or more lower than the absorption of the incident radiation. The effective path length in the light guide can be increased further by applying a wavelength-selective filter. It should be possible to collect 50% of the incident solar flux up to the PV-cell absorption edge while saving a factor of 10 or more in PV-cell area. This would make LSCs an interesting alternative to other types of solar panels.

In summary, the prospects for LSCs are good. They can provide inexpensive PV energy and, because they consist of colored sheets of material, be attractive objects to integrate in built environments or consumer appliances. To generate solar energy with appreciable efficiency, suitable luminescent materials with low self-absorption should be developed, for example, rare-earth-containing line emitters.⁸ Applying wavelength-selective filters makes it possible to reach high concentration factors. Future work will be directed towards obtaining suitable luminescent materials and filters and combining them with appropriate PV cells into efficient LSCs.

This work was partly supported by SenterNovem (project IS073014).