Tuning The Triple-junction
Extracting the very best performance from a triple junction photovoltaic demands optimisation of the absorption edge of every sub-cell. Incorporating quantum wells into the cells can realise this, while allowing the device to be tailored for the spectral conditions where it will be deployed, say QuantaSol’s Keith Barnham, Alison Dobbin, Matt Lumb and Tom Tibbits.
Photovoltaic installations are going up and up. Despite the sharp downturn in the global economy, deployments continue to rise at an exponential rate in many countries, according to a recent report from the International Energy Agency’s Photovoltaic Power Systems Programme. Germany continues to lead the way in photovoltaic deployment thanks in part to the success of its feed-in tariff program, and in 2009 it installed nearly 4 GW. Now its recipe for success is being adopted by several other countries, including the UK, through the introduction of their own market stimulation packages.
Maintaining this rapid growth in cell deployment hinges on further cost reductions, which is possible with the introduction of third generation cells. Those that are tipped to make the first major contribution are the triplejunction cells based on GaInP, GaAs and germanium. This type of device is already the cell of choice to power satellites, and when it is used on earth it can realise efficiencies in excess of 40 percent – roughly three times that of second-generation, thin-film cells made from materials such as cadmium telluride and copper indium gallium selenide.
The key to realizing low costs compared to secondgeneration cells is to use the triple junction devices in concentrator systems that employ cheap plastic mirrors and lenses to focus sunlight by factors of around 500 onto the expensive III-V cells.
Like any new technology, the initial cost of these concentrating photovoltaic (CPV) systems is high. This partly accounts for their initial deployment in the very sunniest climes, where it is most cost-effective. Here engineers are honing the technology that is used to track the path of the sun across the sky and ensure that the sunlight is optimally focused on the cells. As time goes on, increased sales of this technology will drive down costs at both the system and cell level. This will increase the appeal of this technology, which can generate up to three times the electricity from the same system area as one based on second-generation cells.
High concentration systems are automated to track the sun and keep the sunlight focused on the small, high efficiency cells. The first systems have been deployed in locations with large amounts of direct solar insolation Image courtesy of Daido Steel
Quantasol’s quantum wells
Quantasol, a spin-out from Imperial College London that is based in Kingston upon Thames and was founded in 2007, is well placed to supply high-efficiency cells to the growing number of CPV module and system manufacturers. One of our key strengths is our exclusive, patent-protected, quantum-well (QW) technology that boosts the efficiency of triple-junction cells by around 3 percent in absolute terms (see Figure 1). This means that we can regularly realise 40 percent median efficiency across our production wafers, making the performance of our cells comparable to those made by the market leaders.
Figure 1. The blue histogram shows efficiency measurements at 200x concentration of 171 cells of standard triple junction design from one of eight wafers grown on a multi-wafer production reactor. The red histogram shows measurements of cells on a similar wafer with 50 quantum wells grown in the middle cell. The median efficiency has increased by 3 percent absolute. Measurements were made at 200x concentration on a WACOM solar simulator (1000 Wm-2 at 1 sun)
What’s more, our novel cells promise to set a new benchmark for efficiency that cannot be matched with conventional triple-junction cells, a factor that has helped us run a successful sampling campaign with customers through the US, Europe and Asia. This has swelled our order book, which stands in excess of 1MW of solar cells for delivery through the first half of 2011.
We have been pioneering the commercial development of III-V cells with QWs for several years, beginning with the incorporation of these layers into single and dual junction cells. Incorporating wells into the cell increases device efficiency, because it enables the cell to absorb light at longer wavelengths. However, it is only possible to benefit from this if the introduction of multiple quantum wells (MQWs) does not degrade the material quality of the entire cell. We meet this criterion by strain-balancing our structure. In these QW cells of high material quality, the only loss mechanism for current carriers generated by the incoming sunlight is radiative recombination back to photons, a process that cannot be prevented from happening. But we do not waste these photons. Instead, we confine a significant proportion of them within the device – so that they can be re-absorbed and generate electricity – by incorporating a reflector into the structure that is similar to the type used in vertical cavity lasers. This modification to the conventional device architecture justifies our claim to be the greenest of the solar cell companies – we even recycle waste photons!
Recently, we have been developing triple-junction cells with QWs in the middle junction. This structure gives us a major advantage over competitors, because our engineers can optimise the absorption edge of the middle cell during growth. Thanks to this, the absorption in the middle cell can be tuned to match the spectrum of sunlight at different times of day or year, and consequently optimise the cell for a given concentrator, at a given geographic location, for maximum electrical energy harvest over the year.
Optimising short wavelength performance
CPV systems rarely operate at their full potential at shorter wavelengths due to relatively high optical attenuation in this spectral range. Mornings and evenings, atmospheric pollution and turbidity, and the concentrator optics themselves all serve to reduce the energy available for the top two cells in triple-junction devices. However, by adding quantum wells and redesigning the top cell absorption appropriately, it is possible to increase the spectral envelope from which the top two cells harvest energy by about 5 percent, according to calculations that we presented at the 5th World Conference on Photovoltaic Energy Conversion in Valencia last September. These calculations employed a variety of atmospheric datasources, which were used to generate the solar spectra for each hour of the year in locations deemed favourable for the first CPV installations. The solar spectra were used to model the annual energy harvest of a traditional triplejunction device and another featuring MQWs. Efficiency gains stemming from the introduction of the QWs varied from 3.5 percent to more than 5 percent, with the greatest benefit coming in areas where the spectrum contains more red-light than the ASTM reference spectrum, such as Solar Village in Saudi Arabia (see Figure 2).
Figure 2. The simulated increase in power output of a MQW triple-junction device over a bulk triple-junction device under the ASTM G173-03 reference spectrum, and the increase in energy harvest calculated using hourly spectra specific to La Parguera in Puerto Rico, South West US, Gujarat in India and Solar Village in Saudi Arabia
Even higher gains are possible by modifying the band-edge of the middle cell and the transparencies of the top two sub-cells so that they are all tuned to the incident spectra. This results in roughly another 1 percent increase in energy harvest.
Deciding on the best design for a particular location requires a full-year analysis of the ratio of short circuit currents for the top and middle cells. The photocurrents from these cells must be closely matched to achieve the highest energy harvest efficiencies (see Figure 3). Interestingly, the highest daily energy output coincides with the peak in incident energy (blue and black lines in Figure 3). So, to extract as much electrical energy as possible from our cells, we are tailoring device design to maximize the energy harvest efficiency at the same time of the day and year as the peak in irradiance. This principle is also being applied to boost revenue generation by tuning cells to perform at their best during utility peak delivery times.
Figure 3 a) Daily incident energy density (blue), energy harvest efficiency (red), b) device energy output (black) and mean current match ratio (orange) for a simulated device under an hourly varying spectrum specific to the South West US
Doubling up on quantum wells
Looking to the future, we have started to investigate the advantages of MQWs in the top and the middle cells of a triple-junction device. One motivation behind this effort, which has been pursued in partnership with the Imperial College London QPV group that first developed the QW cell, is to optimise the absorption profiles of the cells. In a conventional triple-junction device, the absorption edgesof both the top and middle cell are at shorter wavelengthsthan ideal.
Our modeling indicates that efficiencies well in excess of 40 percent are possible with MQW based cells that shift the absorption edge to longer wavelengths (see Figure 4). These calculations – which are based on a concentration factor of 500 and have been plotted for the specific choice of the solar spectrum used in the standard test condition – show how optimising the absorption edge of our top and middle cells can yield overall efficiencies up to 45 percent. To illustrate what is feasible for a dual MQW device, a rectangular area bounded by a green dotted line is included in this plot and it can be seen that absorption edges reaching the 45 percent efficiency contour are possible.
Figure 4. The colored areas separated by black numbers show how the efficiency of a dual-MQW triple junction cell at 500x concentration varies with the absorption edge of the top cell and the middle cell. The green broken-line rectangle represents the variation in absorption edge possible for the dual-MQW cell. The black broken-line arrow represents the absorption edge variation possible for the competitor metamorphic cell. In all cases the recombination loss in the top and middle cell is assumed to be radiative. The blue contour lines (blue numbers) show the efficiency when radiative recombination photons from the top cell are absorbed in the middle cell. These contours are wider than when radiative coupling is ignored
As the solar spectrum changes by day and during the year, it shifts these contours that map out the conversion efficiency. The large rectangular area shows that the quantum wells add tremendous flexibility in the choice of absorption edges for both cells, indicating that it should be possible to optimise the design for maximum energy harvest over a very wide range of spectral conditions.
The primary rival to our dual-MQW triple-junction cell is the metamorphic triple-junction cell grown on a relaxed buffer layer, which is often referred to as a virtual substrate. With this design, the absorption edges of the top and middle cells can be extended to longer wavelengths. However, these adjustments cannot be made independently – the tuning of absorption that is possible with the metamorphic design is shown by the black arrow in Figure 4. This arrow can, in principle, be extended to higher efficiency contours. However, such a move pays the penalty of greater relaxation in the buffer, leading to more residual dislocations and ultimately an increase in efficiency loss. One assumption in these calculations is that the main loss mechanism in the top MQW cell is radiative recombination, which is known to dominate the loss in the middle cell. An intriguing possibility arises if this also happens in the top cell. In that case, the majority of the photons created by radiative recombination in the top cell will be emitted from the bottom of the quantum well, and most of these will then be re-absorbed in the middle cell, boosting its current output.
To understand how this affects the overall performance of our triple junction device, we have repeated the calculations for this scenario. Results are plotted in Figure 4, using blue contours and blue numbers. These results indicate that radiative recombination of photons from the top cell that are subsequently absorbed by the middle cell has a major benefit - a widening of the sweet spot for very efficient operation. This means that the efficiency of a triple-junction device with QWs in two of the cells will vary less with changes in spectral conditions than one incorporating wells into just one cell.
Realising such high efficiencies in real devices requires the top cell to be radiatively efficient. Initial results, which we presented in Valencia, are promising. The radiative efficiency, which was extracted from measurements of the dark-current of MQW top cells grown on a research reactor at the EPSRC National Centre at the University of Sheffield, increases with concentration (see Figure 5). The initial values for radiative efficiency for the top cells are not as high as for the more mature middle cell, but it does increase with the number of quantum wells. If around 50 of these are incorporated into the top well, the efficiencies predicted in Figure 4 should be realised.
Figure 5. The radiative efficiencies of two MQW top cells grown on a research reactor compared with a typical MQW middle cell. Efficiencies have been extracted from measurements of the cell dark-current.
Focusing sunlight by a factor of 500 or so onto a cell causes it to heat up. Cooling can reduce this rise in temperature, which is beneficial for conventional triple-junction cells, because as they get hotter, the cell efficiency falls. In our novel cells this is far less of an issue, because they can be designed with deep wells in the top and middle cells, a modification that allows efficiencies of more than 40 percent to be achieved at operating temperatures of 90 °C.
This cell heating can be put to good use, allowing this device to feature in integrated, combined heat-and-power applications that further reduce electrical generation costs. High temperature operation of the cells mean that the cooling water can be used in the building, because a temperature of 90 °C is high enough to run an absorption chiller providing air conditioning.
Meanwhile, the cells can generate electricity. With three times the efficiency of a second generation cell, they can be used on buildings in less than ideal locations, such as vertical walls in northern latitudes, and still generate more electricity per square meter over a year than secondgeneration -cells.
Our MQW cells are also ideal for use in smart, power generating windows. The blinds, which are transparent Fresnel lenses tracking the sun, focus the direct sunlight on a luminescent light bar while allowing the indirect sunlight into the room for internal illumination, thus reducing electricity demand. The QW cells are mounted in the window frame. The cooling water can be used for internal purposes including running air-conditioning, the demand for which is highest when the sun is shining.
Many other applications will open up for small-sized, high concentration units offering more than double the electrical efficiency of first- and second-generation photovoltaics. If first generation cells, which supplied the household’s electricity, were replaced by such a system covering the same roof area, the extra electricity produced could power the family electric car for the year, even in rainy England! In addition, this system could supply the household’s hot water requirements.
At Quantasol, we are well placed to capture a significant proportion of the ever-increasing orders for thirdgeneration cells, and we are destined to become a major player in this market. The versatility of our cells enables them to maximise electrical energy harvesting in varying spectral conditions as well as to cope with the challenges provided by the ingenuity of concentrator manufacturers.
Power generating window formed from transparent blinds feature Fresnel lenses that focus direct sunlight onto luminescent light bars which further concentrate the sunlight on cells in the frame. Diffuse light illuminates the room. (Courtesy Solarstructure Ltd.)
© 2011 Angel Business Communications. Permission required.
International Energy Agency, Photovoltaic Power Systems Programme. IEA-PVPS T1-19: 2010, http:/ /www.iea-pvps.org/
S. Kurtz, National Renewable Laboratory Technical Report, NREL/TP-520-43208, (2009), http:/www.osti.gov/bridge
K.Arthur and K. Barnham Compound Semiconductor, Jan&Feb 2008, 22
A. Dobbin et al. submitted to Solar Energy Materials and Solar Cells, February 2011
K. Barnham et al., 5th World Conference on Photovoltaic Energy Conversion (WCPEC5) , Valencia, 234 (September 2010).
K. Barnham et al. Nature Materials 5 161, (2006).