Magnifying Margins With Microcells
Parallel printing of miniature multi-junction cells offers a low-cost, scalable approach to the production of CPV modules
Semprius 64 m2 system at Solar Technology Acceleration Center, Denver, Colorado, USA.
BY KANCHAN GHOSAL AND MATTHEW MEITL FROM SEMPRIUS
The CPV industry is in the midst of a transition. The days of building first-generation prototype systems are now behind us, and the manufacture of high-performance, reliable systems that can turn a profit are underway. Deployment is rising fast, with installations of 80 MW in 2013 following on the heels of about 40 MW in 2012, and suppliers such as Soitec have shown five-year field results with minimal degradation and an availability in excess of 99 percent.
However, despite all this success, CPV is still dwarfed by the incumbent solar technology, silicon "“ deployment of this totalled 40 GW in 2013. The plummeting prices of crystalline silicon over the last few years have made it challenging for advanced technologies that are not already at gigawatt levels to make a commercial impact.
One solar technology that can compete is a novel form of CPV developed by our team at Semprius, headquartered in Durham, NC. Our CPV technology is based on the parallel printing of thousands of cells, which are far smaller than those used by other CPV companies. Thanks to this move to greater miniaturization, alongside the selection of cost-effective, high-quality materials and processes, it is possible to address the challenges of scale, cost and reliability that must be overcome to have success in today's highly competitive solar market.
Figure 1: Micro-transfer printing using a rubber stamp enables the parallel transfer of many chips
The micro-assembly technology that we employ, which we refer to as micro-transfer printing, has its origins at the University of Illinois. Researchers working there developed a technique that enables the removal of semiconductor devices from the growth substrate and subsequent printing on another platform in a massively parallel manner (see Figure 1). Armed with this technology, it is possible to print thousands of devices simultaneously, each with a Â±2 Âµm placement accuracy. This capability allows us to work with devices that are too small, numerous, fragile, or otherwise difficult to handle by conventional methods.
Figure 2: The 600 Âµm cell on interposer lies at the heart of the Semprius CPV module
Using our printing technique, we transfer sub-millimetre solar cells from their native growth substrates onto low-cost interposers (see Figure 2) to form surface-mountable sub-receivers that serve as the engines of our modules. To make a module, hundreds of sub-receivers are mounted onto a backplane, each with a spherical secondary optic (see Figure 3). The backplane is then mounted inside a steel enclosure upon which a primary optic is attached.
Figure 3: Semprius' optical train enables high levels of concentration on an array of triple-junction cells
The steel enclosure protects the receivers from the environment and maintains the focal distance between the primary optic and the cell. The primary optic consists of a silicone lens array on a tempered front glass sheet, each lenslet perfectly aligned to a cell. The primary optic concentrates sunlight by a factor of 1,111 onto the micro-cells.
This module architecture is compatible with the most advanced III-V cell materials, which operate at the highest efficiencies and increase the competitiveness of CPV technology. Traditionally, cells used at high concentrations have been based on up to three junctions, but we, like several other CPV companies, are now progressing to devices based on four or more junctions that could lead to cell conversion efficiencies in excess of 50 percent.
These include those produced mechanically by combining sub-cells from different kinds of growth substrates. Recent work by us, in collaboration with colleagues at the University of Illinois and Solar Junction, has demonstrated how micro-assembly can facilitate this kind of mechanical stacking, using a micro-scale, heterogeneous integration approach.
Our team at Semprius have made significant strides with our printed micro-cell technology since we started work in this area in 2008. Our initial proof of concept module consisted of printed single-junction GaAs cells, measuring 0.1 to 0.3 mm on a side, which were designed for a concentration of a few hundred suns. After significant development and optimisation, we have progressed to 0.6 mm multi-junction cells that operate optimally at more than a thousand suns and are built from state-of-the-art epitaxial materials from partners such as Solar Junction. Cell size optimization is a complex process that maximises the performance-to-cost ratio. Cost is influenced by the part count, yield and wafer utilization. Performance considerations include passive thermal dissipation, current density and optical efficiency. A small cell size delivers many benefits, which are discussed below, including a reduced optical path, better thermal management, superior optics, and a lower series resistance for the cells.
Reducing the optical path unlocks the door to thinner modules. A thin module design has multiple benefits, including lower weight, a reduction of the material required to construct the modules, reduced wind loading, and greater packing density. This means lower cost modules, lower cost trackers and lower cost shipping.
Thermal management is a big issue for CPV, because focusing sunlight by factors of more than a thousand can lead to significant cell heating. Our design is superior in this regard, because the thermal load is distributed over a larger area. For example, each of our cells has to dissipate about 220 mW of heat, compared to about 60 W for a more typically sized cell with an area of 1 cm2. Thanks to the reduced requirement for heat dissipation, our modules don't require heat sinks, and this leads to significant cost savings.
Crucially, we can do this without compromising performance or reliability "“ even without heat sinks, our cells remain relatively cool during operation. Another strength of our design is that it allows the use of unique, high-performance secondary optics. In our case, the secondary optic is a tiny ball lens placed on top of the cell (see Figure 4). The advantages of this pupil-imaging optic are that it provides a wide acceptance angle, delivers a uniform distribution of light on the cells and produces minimal chromatic aberration. While micro-cells are compatible with this type of optic, it is impractical for larger cells because larger ball lenses require too much glass and are therefore too costly.
Figure 4: The fabrication of the module backplane with receivers employs processes based on surface mount technology
In order for the electrons generated in a solar cell to provide useable electricity, they have to traverse the lateral extent of the device. With smaller cells, electrons travel a shorter distance, leading to a lower series resistance and an increase in the power produced by the module. On top of all these benefits, our approach to module production has several other virtues, including substrate re-use, whereby the relatively expensive substrate can be cleaned after the micro-transfer printing process and re-used many times. This significantly reduces the cost of the solar cell.
Another benefit of our approach to module production is that it is similar to a standard microelectronics process. After we use micro-transfer printing to transfer thousands of cells from the source wafer to a ceramic substrate in a massively parallel manner, receivers are then attached to a backplane using standard surface mount technology. By adopting this approach, the only "˜assembly' part of the process is the attachment of the lens and the backplane to the enclosure. Alignment, though critical, is easily performed with adequate precision using existing technologies. As a result, the capital expenditure for setting up a manufacturing plant is among the lowest for PV module technologies and the process is highly scalable.
One major benefit of this manufacturing process is that it allows for a distributed manufacturing strategy. The receivers can be fabricated at a wafer fab, shipped as tiny die in tape and reel, while the modules can be assembled closer to the end markets, reducing transport costs related to materials such as steel and glass and addressing "˜local content' requirements, as necessary.
In 2008, we prototyped a module that operated at 1000 suns and was based on 4 single-junction microcells. This design evolved into an engineering prototype module with nominal dimensions of 14 inches by 14 inches, using 384 transfer-printed two-junction cells. Evaluation of this module's performance on-sun for more than 18 months confirmed the feasibility of this design and provided valuable insight for our production module. A small number of engineering prototype modules were integrated into a 1.3 kW research, development and demonstration (RD&D) system in Tucson, Arizona.
Thanks in part to a SunShot Incubator Award for $ 3 million from the US Department of Energy, we have refined our module design, making it better-suited to high-volume manufacturing and improving its performance-to-cost ratio. The latest version, which has now been in the field for more than two years, operates at 1111 suns and has a nominal surface area of 18 inches by 24 inches with a thickness of 2.7 inches. This module, manufactured at a pilot plant in Henderson, NC, features 660 transfer-printed triple-junction cells.
One of our overriding aims from a very early stage has been to develop cost-optimized modules that can be manufactured in high volumes and perform reliably in the field for more than 25 years. To accomplish this goal, we have taken great care to select materials that are already available and cost-effective in high volume, and have been characterized outdoors and in reliability chambers. We have used processes and equipment that are standard in the semiconductor, optoelectronic and automotive industries; the only unique process is the micro-transfer printer.
Reliability is paramount, and to address this in an appropriate manner, we have been running a rigorous outdoor and chamber testing programme for several years. The benefits of this programme includes being able to accurately characterize the impact of our development progress and to gain feedback on the suitability of the design to field conditions. Maturation of the CPV industry has also led to the development of IEC and UL standards that have been agreed to by the industry, vendors and national and commercial test labs. Additional characterization of our modules "“ both the early engineering prototype and the current design "“ has been performed at various independent labs, including NREL, Sandia National Labs and the Fraunhofer Institute for Solar Energy, providing valuable feedback to our design and characterization process.
Today's modules, which are produced in our pilot plant, have an efficiency of 34 percent to 36 percent at concentrator standard test conditions (CSTC): 1000 W m-2 direct normal irradiance, 25ËšC cell temperature and spectrum defined by the standard AM 1.5. Independent testing of one of our higher performing modules by scientists at Fraunhofer ISE revealed an efficiency of 35.5 percent (see Figure 5).
Figure 5: Independent test results from the Fraunhofer Institute for Solar Energy confirming a module efficiency of 35.5 percent
Encouragingly, after more than two years in the field, our latest generation of modules are showing no measurable deterioration in performance, or any other unwanted effects. Testing in this operating environment continues, with our modules currently under scrutiny in 15 pilot systems in eight countries. These systems are providing data from locations on three different continents, and are enabling us to understand the performance of our modules under different geographical and climatic conditions. The seven systems installed since late 2012 range from 14 kW to 24 kW and are suitable for future commercial deployments. They feature optimized trackers from mature vendors and industry standard inverters. This effort has involved working with tracker partners, who have partnered with us to design a system that optimises the performance-to-cost ratio. To alleviate any concerns relating to tracker reliability, we have only selected tracker partners with field experience spanning years and hundreds of megawatts of deployment. These systems have now been operating for more than a year, and they have a peak AC efficiency exceeding 30 percent.
High efficiency, low equipment cost and reliability are clearly important, but they are not the only issues relating to CPV deployment "“ there are also more underappreciated aspects, such as equipment transport, field installation, operation and maintenance. For example, it is critical that both the tracker and the modules are optimised for transport in standard ocean containers; otherwise transportation to the project site can add significantly to the costs. Thanks to the low profile of our module and the configuration of the module arrays, it is possible for us to realise a high packing density that is competitive with flat plate silicon modules and better than CPV modules made by our peers. It is also critical to devise installation methods that are standard in the construction industry, can be executed by construction labour, are safe and can be performed with high velocity and at a low cost.
We have also developed installation, operation and maintenance methodology and documented best practices that have drawn on our and our partners' experiences in civil construction, field assembly and commissioning. The experience gained from these activities will aid our next deployments, and lead to further refinements in the ways we approach fulfilling our customer's orders, as we deploy systems in larger projects.
Goals for the future
We have just completed the design and initial tests relating to our third generation module, and we unveiled the results at the 40th IEEE PVSC, which was held this June in Denver, CO. The overarching goal for this latest design was to cut costs further, while maintaining much of the existing design elements and making no impact on the pilot plant process and toolset. These considerations have led to a design that maintains the dimensions of the existing design, while increasing concentration to 1600 suns. Indoor and outdoor testing of the new module, which has 30 percent fewer receivers than the existing design and provides a significant reduction in module cost, shows it delivers a similar performance to its predecessor.
Another recent breakthrough that we have made is to demonstrate stacked cells that are fabricated by printing two micro-cells, grown on different substrates, on top of each other. In this project a top InGaP/GaAs/InGaAsNSb cell is grown on a GaAs substrate, released from it by etching a sacrificial AlInP layer and printed on a germanium bottom cell. A spin-cast, 300 nm-thick film of As2Se3 binds together these two cells together while providing a low-loss optical interface that is thermally conductive and electrically insulating. The triple-junction and germanium cells operate independently with separate sets of terminals, thereby avoiding current matching issues. Operating at 1000 suns, these multi-junction and single junction cells have efficiencies of 42.1 percent and 1.8 percent, respectively, and thus combine to deliver an efficiency of 43.9 percent.
This work represents the first demonstration of a four-junction, four terminal cell, and it lowers the barriers towards developing a cell operating at an efficiency greater than 50 percent, and a module operating at an efficiency in excess of 40 percent. Such cells can be fabricated on existing equipment and quickly integrated into our current module design, which will help to accelerate the adoption of CPV systems that are more competitive in sunny climates than those based on silicon.
Semprius has increased the concentration factor of its modules. This version operates at 1100 suns
Other opportunities for micro-transfer printing
Micro-transfer-printing is a powerful device assembly technology that can serve various compound semiconductor applications. For non-photovoltaic applications, it is being commercialized by X-Celeprint Limited of Cork, Ireland.
This micro-assembly technology offers a practical way to combine arrays of micron-scale, diverse, high-performance materials and devices with substrates that have vastly different properties and cost structures compared with traditional packages. It is a technology that opens new levels of component miniaturization, facilitates heterogeneous integration, and readily interfaces with epitaxial lift-off techniques for substrate re-use.