Photovoltaic manufacturers are looking to cut costs and maximize performance, and lasers are starting to deliver in both areas
Interest in solar power is at an all time high, but the price/performance ratio of photovoltaic solar cells (PVs) is still not competitive with grid power. Achieving this parity will require a major (up to 3X) cut in manufacturing costs. Lasers are poised to play major roles in reducing cost and increasing yields, while enabling innovative designs and processing methods that will simultaneously boost performance. This article examines just some of the ways in which PV manufacturers are using lasers, as well as emerging applications that are at the R&D or pilot stage.
Generation I devices
There are two basic types of solar cells. Generation I devices are made of crystalline silicon (thin silicon wafers) where the biggest single cost factor is the silicon itself. This has led to the development of Generation II devices based on a thin layer of silicon or other semiconductor materials deposited on a glass or foil substrate.
Generation I devices are fabricated as square wafers (typically 156 mm x 156 mm) with each wafer becoming a cell; several of which are assembled and wired in series to produce a packaged module. Efficiency is paramount; typical cells have a (solar-to-electrical) power conversion efficiency of around 15-16 percent with a value of 12 percent for a typical assembled module. Manufacturers are striving to increase these values because premium performance commands a premium price and hence higher margins, so any technology that provides even a minor improvement in efficiency is of interest. However, cost is also critical because these are high-volume, commodity products.
At present, one of the largest laser applications is edge isolation. The p-doped wafers are coated with an outer layer of n-doped silicon to form a large area p-n junction, which ultimately generates the electrical power. But this thin (10-20 microns) layer coats the entire wafer, including the edges, and often the rear surface, creating an unacceptable recombination pathway between the front and back surface. This pathway can be eliminated by edge isolation, whereby a groove is continuously scribed completely through this n-type layer. In order to maximize cell active area and hence efficiency, this groove has to be as narrow and as close to the edge as possible.
The usual laser of choice is a Q-switched solid-state laser with a pulse repetition rate around 30 kHz. The majority of manufacturers use 1064nm lasers as these offer the highest power/cost and hence the fastest throughput. Typical scribe rates are around 500 mm/s. But some manufacturers are testing 532nm and even 355nm lasers for this application for two reasons. First, these lasers can scribe narrower grooves (see Figure 1). In addition, 1064nm lasers create microcracks that emanate from the scribed groove. If these reach the edge of the wafer, structural integrity is potentially compromised. So this limits how close the 1064nm-machined trenches can be placed to the edge of the wafer.
With thinner wafers, it can be electrically and thermo-mechanically advantageous to include a passivation layer between the rear surface aluminum electrode layer and the silicon. But this passivation layer is non-conducting. The Laser Fired Contacts (LFC) technique recently developed by Fraunhofer ISE now paves the way for full commercial production of this type of cell by using a laser to create localized contacts. At each site, a 1064nm laser pulse drives the aluminum through the passivation layer and several microns deep into the silicon, creating a localized Al/Si alloy.
Another fast-growing application is via drilling. The simplest solar cells have contacts on the front and rear surfaces to collect the negative and positive charge carriers. But the screen-printed metal comprising the front-side contacts blocks a significant area from receiving sunlight. (It’s that all-important efficiency issue again.)
Two newer high-value device architectures address this. In metal wrap through (MWT) devices, the thin metal “fingers” are moved to the rear surface. In emitter wrap through (EWT) devices, power-conveying busbars are moved to the rear surface as well, leaving the front free of metal. This is made possible by drilling tiny vias to connect the front surface with rear-surface contacts (see Figure 2). With MWT, this requires about 200 holes/wafer. With EWT, up to 20,000 of these vias are required on each wafer. Laser drilling is the only process with the potential to meet the commercial-scale speeds required here (see Figure 3). The laser of choice is currently a 1064nm laser with tens of watts of output. But again, manufacturers are already looking at 532- and 355nm alternatives to produce smaller holes with minimized thermal damage, both factors that increase efficiency.
Various designs have been investigated to build in the complex microstructures required for all-back-contact cells. For example, in the so-called RISE cell design, the negative and positive base contacts are located at different step heights, requiring laser ablation of relatively large areas of several individual layers. This process has been evaluated with a variety of lasers, including excimer (248nm), and solid state (1064, 532, and 355 nm). As with some other processes, the 1064nm laser produces some undesirable peripheral thermal damage.
Buried electrical contacts represent a different approach to minimizing the area obscured by screen-printed front-side metallization. This is a technology patented by BP Solar in the mid-1980s. After the surface has been antireflection coated with silicon nitride, a 532nm (or 355nm) laser is used to scribe narrow surface grooves, which are then plated. This results in electrodes with a high volume and collection surface, relative to their width. The grooves have a width and depth of 20-30 microns and are cut at 2-3mm intervals along the cell. In a recent adaptation of this approach, lasers cut only through the AR coating and the exposed semiconductor is electroplated.
Lasers are also being used by some manufacturers to cut the raw silicon itself. For example, RWE Schott Solar pulls its silicon in a hollow octagonal shape, each side of which becomes a wafer strip. 1064nm lasers cut each vertex of the octagon at high speed. Conversely, Evergreen Solar pulls its silicon from the melt as a single 8-cm wide ribbon. In this process, a 1064nm laser cuts or scribes the ribbon to convenient lengths.
Thin film devices
Most thin film devices are currently produced on glass substrates and utilize silicon with a layer thickness of only a few microns. These have efficiencies less than 10 percent. The value depends on the type of silicon, with amorphous silicon as low as 6 percent. This low value is the reason manufacturers are pursuing other thin film materials such as CIGS (copper indium gallium diselenide). But even CIGS can only reach 12-13 percent efficiency. Despite this low efficiency, there is great interest in thin film devices because of their potentially low manufacturing cost. The ultimate goal here would be web-based, roll-to-roll continuous processing, which requires the use of flexible substrates. Metal foils are just starting to be adopted, but no plastics have so far shown themselves to be robust enough for this application.
Thin film devices are still at an early stage and there are only three volume laser applications at this time. For thin films, Edge Deletion is the corresponding process to Edge Isolation. When the various layers are deposited on the substrate, the edges of each layer will not always terminate along the same line (rather like an overfilled sandwich). This can lead to electrical shorts and other functional problems. For this reason, high-power 1064nm lasers are used to scribe through all the layers down to the substrate.
Unlike crystalline silicon PVs, thin film devices are produced in large areas (typically 1 square meter), which allows monolithic integration. The panel may be subdivided into up to 200 cells that are electrically connected in series. This subdivision and connector patterning requires that the panel be scribed after each of the several layers are deposited (see Figure 3). This is done with both 1064- and 532nm lasers and requires depth control to avoid scribing layers below the target layer. This depth precision can be provided by both wavelength and careful dosage control. With glass substrates, the 532nm laser can even scribe from the rear surface, taking advantage of the substrate transparency.
Finally, the other thin film PV application is marking the substrate. Typically the mark is a 2D data matrix which then allows the substrate to be tracked and logged through the entire fab process.
Energy aid
Given the economic and environmental impact of our reliance on fossil fuels, it’s no exaggeration to say that the development of alternative energy sources may be the single most important technological effort underway. Lasers have been critical in enabling microelectronics manufacture, and it’s already clear that they will play a key role in aiding this development.
Corey Dunsky ([email protected]) is senior manager of strategic applications with Coherent Inc., www.coherent.com.