Lifting limitations in photovoltaics
MicroLink’s solar cell arrays, which are made with its proprietary epitaxial lift-off technology, are an attractive option for powering electric unmanned aerial vehicles, thanks to their high efficiency, low weight and excellent flexibility. With an aerial mass density of less than 350 g/m2 and a power density exceeding 250 W/m2, these arrays have doubled the endurance of the Raven. This small, hand-launched, remote-controlled aerial vehicle that is powered by a lithium ion battery can fly at altitudes of 10,000 feet to 15,000 feet and reach speeds of 60 miles per hour
Developers of multi-junction solar cells tend to try and improve device performance by inserting new materials to boost efficiency. But there are many other ways to make the cells more appealing, such as trimming weight, increasing robustness, cutting material costs and enhancing flexibility. One company that has developed a technology that allows high-efficiency cells to excel in all these regards is MicroLink Devices, which is based in Niles, a northern suburb of Chicago, Illinois. Founded in 2000, MicroLink is best known for its MOCVD growth of epitaxial wafers for handsets and other wireless products, but over the last five years it has also been developing a wafer-scale, epitaxial lift-off process for manufacturing ultra-light, photovltaics with one, two or three cells. The performance of these devices, which are mounted on a flexible metal backing, is very promising. Measurements replicating the sun’s spectrum yield efficiencies for MicroLink’s best 1 cm2 cells of 31 percent in space (AM0) and 33.9 percent on the ground (AM1.5). Scale the cell size to 20 cm2, and efficiencies fall by just one percentage point. These high-efficiencies, combined with an incredibly low cell weight, make these photovoltaics attractive candidates for powering satellites. These cells are also flexible, so they can wrap around the topside of the wings of battery-powered, unmanned aerial vehicles (UAVs), increasing their endurance. One organization starting to look at doing just this is the US Air Force Research Laboratory, and it is funding a project involving MicroLink. If this project has widespread success, it could do far more than just aiding the military – those working in the real-estate sector, for example, view UAVs as an attractive approach to surveying vast areas of land. MicroLink’s cells could also be a competitive product for the terrestrial concentrating market, thanks to relatively low production costs that stem from multiple re-use of GaAs substrates, which just require a polish before they re-enter the MOCVD chamber. Using re-polished substrates makes no impact on device performance. “The substrate is a considerable fraction of the bill of materials for the device,” explains Chris Youtsey, Fab Director at MicroLink. “The limitations on how many times you can re-use the substrate come down to how much material you remove from the polishing and [the frequency of] breakages.” Youtsey claims that a reasonable goal is to re-polish the wafer ten times. He points out that polishing removes as little as 10 µm, and says that thinning a GaAs substrate by 100 µm or so should have no impact on the deposition of a high-quality epitaxial structure. Even higher rates of substrate re-use are theoretically possible, but they produce ever diminishing returns. Enter this regime and the rewards do not justify the efforts. Peeling it off The epitaxial lift-off technology that MicroLink uses is certainly not new. Reports of such efforts date back to 1978, when a Japanese team from Tokyo Institute of Technology published a paper on the fabrication of high-efficiency GaAs solar cells with a ‘peeled film’ technology. Progress in the 1980s and 1990s included efforts by Eli Yablonovitch from Bell Communications Research, who revealed the extreme etching selectivity of AlAs compared to GaAs. And further strides on epitaxial lift-off technology have been made during the last 15 years, with John Schermer’s group from Radboud University reaching a new level of understanding of the process, and devising new ways to increase the etch rate. However, epitaxial lift-off is still an intrinsically slow process, according to Youtsey: “It’s not seconds or minutes – it’s hours. So if you want to get high throughput, you have do it in batches.” Switching to batch processing is one of the two big breakthroughs in epitaxial lift-off technology made by MicroLink Devices, whose team has a background in high-volume GaAs fabs, including TriQuint. Running in pilot product, MicroLink processed 1400 4-inch wafers in 2011, and it could increase this throughput substantially – each etching bath that it uses is capable of processing 1500 wafers per month. Youtsey says that another significant breakthrough made by his team is the simplification of the process – no longer are weights used to separate substrates and epilayers: “Our approach uses a proprietary support layer, which allows us to efficiently batch process.” At MicroLink, photovoltaic production begins with MOCVD growth of an inverted metamorphic structure featuring InGaP, GaAs and InGaAs cells. This epitaxial stack contains a 5 nm-thick AlAs layer sandwiched between the substrate and triple-junction cell. After attaching a thin, flexible metal carrier layer to the uppermost epitaxial layer, the resulting composite is immersed in a bath of concentrated hydrofluoric acid, which selectively dissolves the release layer – the etch selectivity of AlAs relative to the GaAs epitaxial structure exceeds 105. After etching for 12 hours or so, the metal carrier and solar cell epilayers are completely separated from the GaAs substrate. Engineers at MicroLink mount these epitaxial lift-off foils to a temporary, rigid carrier, so that they can process the wafer into devices. This involves evaporation and lift-off of a metal ohmic contact, wet etch isolation, evaporation of an anti-reflection coating made from a bilayer dielectric stack, and dicing of the processed wafer into individual devices. Following that, the solar cells are removed from the temporary carrier.
MicroLink forms its metamorphic cells on a GaAs substrate. Etching a sacrificial, AlAs layer about 5 nm-thick with hydrofluoric acid allows separation of substrate and epilayer
Epitaxial lift-off does not degrade the material quality of the cells. Transmission electron microscopy images from the National Renewable Energy Laboratory fail to uncover any delamination, cracking, threading dislocations or voids in the cells. An absence of cracks and defects holds the key to the fabrication of cells with very large areas. After a flexible metal carrier is attached to the solar cell epi-structure, the GaAs substrate is removed by etching in hydrofluoric acid A better base MicroLink’s metamorphic structure features: An InGaAs bottom cell with a bandgap that can be tuned from 0.9 eV to 1.1 eV; a 1.42 eV GaAs middle cell; and an (Al)InGaP top cell with a bandgap that is adjustable from 1.88 eV to 2.00 eV. “I think that’s pretty close to optimum,” claims Youtsey. To evaluate the impact of re-polishing on device quality, engineers at Micolink have compared the efficiency of 1 cm2 cells formed on 25 substrates that were initially pristine, and then re-polished once, twice, three and then four times. Their conclusion: The average cell efficiency on the three re-polished device populations is comparable to that of cells grown on original, prime GaAs substrates. If these cells are to be used in space applications, they need to be able to withstand bombardment from various forms of radiation. Youtsey believes that the cells can meet this demand: “The radiation hardness of individual epitaxial lift-off junctions, such as InGaP, GaAs and InGaAs, follows the expected trends for these materials.”
Processing begins with the growth of an AlAs release layer and a solar cell structure on a GaAs substrate. After etching in hydrofluoric acid the GaAs substrate is re-polished and used again, while the solar cell and back metal composite is temporarily bonded to a carrier wafer for device processing
Scaling cell size also appears to have no impact of device performance – typical conversion efficiencies of 29 percent were recorded for 1 cm2 cells and 20 cm2 equivalents, which were both measured at NASA Glenn using the AM0 spectrum. Flexible solar sheets have been produced at MicroLink by interconnecting 30 large cells with silver-based foil ribbons and laminating the structure between transparent sheets to yield flexible solar sheets. The resulting composite is highly flexible, and can wrap around curved structures, such as the wings of solar-powered planes. One great attribute of this solar sheet is its incredibly high specific power, which exceeds 400 W/kg. According to Youtsey, triple-junction germanium cells with a 150 µm thickness are at least four times as heavy: “From an application point of view, that is a big deal.” In addition, cells formed on germanium substrate are very brittle. “You have to build a wafer fab to accommodate them,” explains Youtsey.
A 4-inch film produced by MicroLink’s epitaxial liftoff process features two, 20 cm2 solar cells on a thin, flexible metal backing
Bigger and better MicroLink has also fabricated 61 cm2 cells from 6-inch GaAs wafers. Part of the motivation behind producing such large cells is to demonstrate the quality of MicroLink’s process. “But there is interest, in terms of panels assembly, in working with large cells: Fewer part counts, fewer interconnects,” explains Youtsey. Today, however, MicroLink's primary focus is not on making bigger and bigger cells. Instead, it is setting its sights on increasing efficiency and streamlining process flow. Success will simplify ramping to production volumes and it will also increase yield. To boost efficiency, efforts will focus on increasing the open-circuit voltage and fill factor. Improvements in material quality will underpin these programmes. Further efficiency gains could result from increasing the number of junctions to four or five, but Youtsey acknowledges that this step is not easy to execute in high-volume production. MicroLink is currently in pilot production, and it plans to progress to full commercialisation within the next two years. When it hits that milestone, customers of multi-junction cells will have some head-scratching to do. Up until that point, selecting a device has been based on efficiency, reliability and price, but from then on factors such as weight, flexibility and robustness will come into the play. Two 61 cm2 triple-junction cells formed by applying MicroLink’s epitaxial lift-off process to a 6-inch GaAs wafer
Thermophotovolatics MicroLink’s epitaxial lift-off process is an attractive approach for making any type of device that has a large proportion of its cost associated with that of the substrate. One such example is thermo-photovoltaics, devices built on InP substrates that convert heat into electricity. Reducing the number of InP substrates needed to make thermo-photovoltaics could lead to massive cost savings, because InP substrates are nearly an order of magnitude more expensive than GaAs equivalents.