4.5 Fluorescent concentrator systems
4.5.2 Systems with different materials
The presented organic fluorescent dyes achieve very high quantum efficiencies, but their absorption range is narrow in comparison to the solar spectrum. On the other hand, different dyes with different absorption ranges are available that cover at least the complete visible spectral range. Therefore, it is a rather obvious idea to combine different dyes to use a larger fraction of the solar spectrum. Already in the first research campaign in the 1980s, Wittwer et al. [51] achieved a conversion efficiency of 4% with a system that combined two 3 mm thick plates with different dyes in one stack with GaAs solar cells attached to the edges. The system was 40 cm x 40 cm in size and therefore quite large, so the achieved efficiency can be considered a very good result. The geometric concentration ratio, that is the ratio of the illuminated collector area to the solar cell area, was 16.7. The system produced around 3 times more energy than that the solar cells would have produced if they had been placed directly in the sun.
In section 4.3.2, I presented a method to determine the light guiding efficiency of different collector materials with a set of optical measurements. With this method, I selected promising materials, from which I realized systems of fluorescent collectors with attached solar cells. First, systems with only one material were realized and subsequently a stack with two materials from the most promising combination. The fluorescent collectors in these experiments had geometric dimensions of 2 cm x 2 cm and were 3 mm thick.
GaInP solar cells, as described in the previous section, were attached to the edges of the collectors. In these experiments, the used solar cells with 3 mm height all had efficiencies of 14.4±0.1%, under an AM1.5g spectrum. For the stack with two collector plates on top of each other, I used solar cells with 6 mm height. They all had efficiencies of 15.4±0.1%. The given accuracy reflects the efficiency distribution of the cells and not the absolute uncertainty. The purpose of the experiment was to find the most promising material combination. Therefore, it was important that the used solar cells had similar efficiencies so that the observed differences are a result of the material properties and not of the different solar cell efficiencies. At this point, the absolute height of the efficiency was of minor importance.
The intensity of the used sun simulator was calibrated with a reference solar cell to 1000 W/m2. No further mismatch correction had been applied. Under all systems, a
white bottom reflector made from BaSO4-coated aluminum was placed.
Fig. 4.53 displays the efficiencies of several realized systems. Only to one edge of each fluorescent concentrator solar cells were attached and the other edges were left open. The surroundings of the systems were covered with a black mask, so no light could enter the fluorescent concentrators from the side. As the purpose of this experiment was to compare different materials, the system had not to be optimized for the highest efficiency, e.g. with mirrors at the edges or by attaching solar cells to all four edges. The material denoted BA241 showed the highest efficiency of 2.5% in reference to the 4 cm2 area of the concentrator. The combination with a second material denoted
BA856 increased the efficiency to 3%.
Fig. 4.53: Materials identified as promising by the optical measurements were used to realize systems of fluorescent concentrators and solar cells. The material denoted BA241 showed the highest efficiency of 2.5% in reference to 4 cm2 area of the concentrator. The combination with a second material increased the efficiency to 3%. From this stack a system with four solar cells, one at each edge, was built. This system had an efficiency of 6.7%.
From this stack a system with four solar cells, one at each edge, was built. The single solar cells had heights of 6 mm, so every solar cell received the light from both collectors. The four solar cells were interconnected in parallel. This system had an efficiency of 6.7%. A similar system with only one concentrator made from BA241 with four parallel interconnected GaInP solar cells had an efficiency of 5.1%.
4.5 Fluorescent concentrator systems
Fig. 4.54: A photograph of the described stack system before the remaining three solar cells were attached.
Fig. 4.55: The External Quantum Efficiencies of a single GaInP solar cell measured under direct illumination, a system with only one fluorescent concentrator (BA241) and of a stack system with two materials. Both systems featured four parallel interconnected GaInP solar cells attached to the edges. The combination of the two materials significantly extends the used spectral range.
It is important to mention that with 4 cm2 these systems are comparatively small.
Therefore the concentration ratio is very small: 1.7 for the single fluorescent concentrator system and 0.8 for the stack system. So in fact, the stack system is a de- concentrator. Fig. 4.54 shows a photograph of the stack system before the four solar cells were attached. Fig. 4.55 presents the external quantum efficiency (EQE) of one
range. An additional side effect is also visible in Fig. 4.54. In the spectral range below 350 nm, the EQE of the fluorescent concentrator systems exceed that of the GaInP solar cell. The reason is that light absorbed in this range by the fluorescent dye is emitted at longer wavelengths where the EQE of the GaInP is higher.
Although this system achieved already a very high efficiency, it does not use the full potential of the material, because only GaInP solar cells were used. As we have seen in Fig. 4.52, GaInP does not absorb all photons emitted from the BA856 material. Therefore, I realized a system with GaInP and GaAs solar cells and the BA241 and BA856 materials. The efficiency of the used GaInP solar cells was 16.8±0.1%, and the efficiency of the GaAs solar cells 24.1±0.2%. To absorb all photons, the thickness of the fluorescent collectors was increased to 5 mm. No 5 mm thick sample was available from BA856, so a 3 mm and a 2 mm thick sample were optically coupled with silicone to form one 5 mm thick slab. Both collector plates were 5 x 5 cm2 in size. Two GaInP
solar cells were optically coupled with silicone to adjoining edges of the BA241 slab, and likewise two GaAs solar cells to the BA856 slab (Fig. 4.56). In front of the remaining two edges of each collector plate, white reflectors made from Polytetra- fluoroethylene (PTFE) were placed. In this configuration, light that is reflected without change in direction reaches a solar cell after a maximum of two reflections. If the solar cells were attached to opposing edges, it would be possible that light beams bounce back and forth between the solar cell free edges without ever reaching a solar cell. Therefore, a higher efficiency is expected from the chosen geometry.
Fig. 4.56: Sketch of the system setup. On the left the top view is shown. Two GaInP solar cells were optically coupled with silicone to adjoining edges of the BA241 slab, and likewise two GaAs solar cells to the BA856 slab. In front of the remaining two edges of each collector plate and under the whole system, white reflectors made from Polytetrafluoroethylene (PTFE) were placed. They can be seen in the cross section on the right.
4.5 Fluorescent concentrator systems
The geometric concentration ratio is 5x for a single system with two attached solar cells. In the stack with two collector plates, the aperture area is still 25 cm2 but the
number of solar cells has doubled, so for the stack the geometric concentration is 2.5x. The system was assembled step by step. The solar cells were attached one by one to the collector plates. In the final system, the two GaInP solar cells were interconnected in parallel, and the two GaAs solar cells were separately connected in parallel. At each point, the IV-characteristic and the efficiency were measured in different configurations: the Ba241/GaInP and BA856/GaAs system separate from each other, the Ba241/GaInP on top of the BA856/GaAs system, and the Ba241/GaInP underneath the BA856/GaAs system. Under all configurations, a white reflector was placed. The efficiency for all systems was calculated in respect to the energy that is incident on the 5 cm x 5 cm collector area, as well as the short circuit current density was determined in respect to this area. Table 4.1 summarizes the important results, for the systems with already two attached and interconnected solar cells.
Measured independently, the single system with the BA856 collector plate and the attached GaAs solar cells has a higher efficiency than the system with the BA241 collector plate and the GaInP solar cells. This can be understood by considering the results of the spectral collection efficiency measurements again (see Fig. 4.25). The BA856 absorbs over a wider spectral range. More photons are collected and hence the short circuit current density is significantly higher. The higher current over- compensates for the lower voltage in comparison to the BA241/GaInP system. However, because of this wide absorption, when the BA856/GaAs is placed on top of the BA241/GaInP system in a stack, nearly no photons that can be used arrive at the lower BA241/GaInP system. Therefore, while the efficiency of the Ba856/GaAs hardly changes in the stack, the efficiency of the BA241/GaInP drops to around 1%. The mathematical sum of these two independently measured efficiencies is therefore 6.6%. It is better to place the BA241/GaInP system on top. In this way, all photons that can be potentially used by this system are absorbed in the BA241 collector and their energy is converted into electric energy at a higher voltage. The BA856/GaAs uses the remaining photons and ensures reasonable spectrum utilization. Nevertheless, the efficiency of the BA241/GaInP system on top of the BA856/GaAs system is lower than when it is measured independently. Without the BA856/GaAs, photons that pass the BA241 collector are reflected at the white bottom and have a second chance to be absorbed in the BA241 collector. In a stack, these photons are absorbed by the BA856. The mathematical sum of the two independently measured efficiencies reaches a very
Table 4.1: Overview of the achieved efficiencies in different system configurations using different fluorescent collector materials and GaInP and GaAs solar cells. The value with * is corrected for the spectral mismatch between AM1.5g spectrum and the spectrum of the sun simulator.
System VOC / mV JSC/ (mA/cm2) FF K
BA241
2 GaInP solar cell attached 1366 3.8 87% 4.6%
BA856,
2 GaAs solar cell attached 1028 6.9 81% 5.8%
GaAs solar cells in configuration BA856 on
top of BA241
1023 6.7 81% 5.6%
GaInP solar cells in configuration BA856 on
top of BA241
1320 0.9 87% 1.0%
Combined efficiency 6.6%
GaInP solar cells in configuration BA241 on
top of BA856
1355 3.2 87% 3.8%
GaAs solar cells in configuration BA241 on
top of BA856
1008 4.3 82% 3.5%
Combined efficiency 7.3%
Combined efficiency with
mismatch corrected 6.9%*
The EQE measurement was performed with a filter-wheel monochromator that allowed illuminating the full fluorescent concentrator area. Therefore, the measurement yields the area-average of the EQE. It can be seen nicely that the two systems together cover a wide spectral range. In the EQE of the bottom GaAs/Ba856 system the effect of the absorption in the top system is visible as well. The bottom system shows also some response outside the absorption region (above 700 nm). This is an effect of the bottom reflector, which redirects transmitted light directly to the GaAs solar cells. The effects of the bottom reflector will be discussed in detail in section 4.5.4.
4.5 Fluorescent concentrator systems
Fig. 4.57: External quantum efficiency measurements of the two subsystems. The system made from BA241 with two attached GaInP solar cells was placed on top of the system made from BA856 with two attached GaAs solar cells. It can be seen nicely that the two systems together cover a wide spectral range. In the EQE of the bottom system one can clearly see the effect of the absorption in the top system.
The irradiance of the used sun simulator was calibrated with a reference solar cell to be equivalent to 1000 W/m2 under the AM1.5g spectral distribution. However, the
spectral response of the reference solar cell and the very special response of the two fluorescent concentrator systems do not match. With the EQE data the mismatch between the spectral response of the used reference cell and of the sub-systems could be calculated and a correction factor be determined that takes into account the differences in the spectral distribution between the spectrum of the sun simulator and the AM1.5g spectrum based on the IEC60904-3 Ed.2 (2008) norm spectrum for AM1.5g non concentrating conditions. It turns out, that with that mismatch correction the efficiency is only 6.9%, which is still a very good value but slightly less than the world record efficiency of 7.1% [40]. Because of the very special spectral characteristics of the systems, the mismatch correction has a high uncertainty and is very sensitive to the used spectrum. A mismatch correction calculated with an only slightly different, older norm spectrum yielded an efficiency of around 7.0%.
attached solar cells is bigger so that by adequate cell interconnections the current or the voltages of the GaInP and GaAs sub-system can be matched and an integration into one module is possible without any problems. Current matching could also be achieved by producing the GaInP and GaAs solar cells in different sizes.