Machine Vision Aids Solar Cell Production
by Paul Kellett
, RIA/AIA/MCA Director – Market Analysis, courtesy Automated Imaging Association
Robotic Industries Association Posted 04/20/2009
As seen on AIA's Machine Vision Online.Despite lower energy prices during the recession, the realization is growing that alternative energy sources are needed. An increasing number of people are convinced that alternative energy sources are essential to controlling energy costs, limiting dependence on foreign energy suppliers and protecting the environment. But few people are probably aware of the important role machine vision plays in making alternative sources of energy available. That in fact is now happening, as machine vision companies increasingly provide applications that improve the quality and efficiency of solar cell production. To understand the new sales opportunity represented by the solar cell market, let’s take a look at this market and how machine vision companies are leveraging their expertise to serve it.
The Solar (Photovoltaic) Cell Market
The most salient aspect of the solar cell market is its tremendous growth. Using available market research, we estimate a compound average growth rate (CAGR) of no less than 29.3 percent. While the global market is currently around $13 billion, we expect it to exceed $30 billion by 2012.
To under-stand this growth market, and the role machine vision plays in serving it, it is useful to consider not just the world’s hunger for energy but also the development and production of the solar cell. A solar (a.k.a. photovoltaic or PV) cell is a device that converts light directly into electricity based on the photoelectric effect. Solar cells were first invented in 1883 but were impractical, consisting of expensive materials and having only one percent efficiency. It wasn’t until 1946 that the modern solar cell was developed and patented by Russell Ohl. Eight years later in 1954, Bell Labs achieved another major milestone, when, experimenting with semiconductors, it developed the first PV cells consisting of silicon doped with certain impurities. The efficiency rate achieved by these cells was six percent. Since then, a number of subsequent discoveries improved the average efficiency of the solar cell to 12 – 16 percent.
Today, there are at least five major types of PV cells based on technology and materials:
Cadmium telluride (CdTe)
Copper indium (gallium) diselenide (CIS or CIGS).
PV cells based on crystalline silicon make up 93 percent of the market; newer thin-film cells comprise 7 percent. Monocrystalline PV cells are first generation crystalline cells; they are 12 – 16 percent efficient, the most expensive to produce, but also the most widely used in the market. Polycrystalline solar cells are second generation solar cells, which have medium production costs and medium levels of efficiency (11 – 13%). Thin-film cells are less expensive to produce than crystalline silicon but are also less efficient. Amorphous silicon cells, for example, are 8 – 10 percent efficient.
Regardless of their type, individual PV cells are typically connected and sandwiched between two layers of protective glass to form solar panels (a.k.a. solar modules or solar arrays). These solar panels may or may not be used with concentrating optics to increase light intensity and therefore power output, and they may or may not be on-grid (connected to the electricity grid) or off-grid (stand-alone). Off-grid solar systems employ batteries to store power, while on-grid systems feed power to the grid using inverters (which perform a number of functions including the conversion of DC into AC).
Within the solar cell market, cells and panels are manufactured by a multitude of companies but most are produced by five large manufacturers, based primarily in Germany, Japan and the US: Q-Cell, Sharp, Solar World, BP Solar and Kyocera.
How PV cells and panels are produced is of special importance to machine vision companies, since their applications must support these manufacturing processes. The specific production techniques employed depends on the type of solar cell produced. Poly-crystalline silicon wafers, for example, are made by sectioning silicon ingots and wire-sawing the sections into very thin slices or wafers. Doping then creates the photoactive p/n junction. Using silicon nitride, anti-reflection coatings are next applied. The wafer then receives a full area metal contact that is fastened to the back surface of the wafer. On the front side, a metal contact is applied by screen printing a grid-like pattern with a metallic paste that is fired in a furnace.
The Role of Machine Vision
According to Lasse D. Nergaard, CEO of Tordivel Solar AS, “(m)achine vision has multiple application areas in the PV industry, such as characterization & classifications of blocks, ingots, wafers, solar cells and solar panels.”
Some MV companies are leveraging their current expertise to serve the solar cell and panel industry. Because solar cells are semiconductor devices, they share many of the same processing and manufacturing techniques as other semiconductor devices. Machine vision (MV) companies are of course no strangers to the semiconductor industry, and so inspection tasks in the manufacture of solar cells and panels are a natural fit for them. Not surprisingly then, a number of MV companies are targeting the solar cell industry, including (in alphabetical order) Adept Technology, Automation Engineering Inc., Basler AG, ISRA VISION, ICOS Vision Systems (acquired by KLA Tencor), Tordivel Solar AS (an affiliate of Tordivel AS), Vitronic and others.
In the production of solar cells, machine vision plays a critical role. Since raw silicon has become more expensive and in shorter supply, silicon ingot sections are being sliced increasingly thin. This makes the wafers harder to separate and handle without introducing defects, and defects of course do occur. Enter machine vision. MV-based inspection systems sort out chipping, microcracks, incorrect thicknesses, warpage, saw grooves, finger prints and impurities in the wafers. This is more challenging than semiconductor inspection, because, at least in the case of poly-crystalline silicon, every wafer has a different crystalline structure, which requires vision systems to discriminate between ordinary crystal boundaries and defects as well as use a wider field of view. In fact, according to Dr. Gord Deans, VP of Business Development at Adept Technology, “…certain defects can only be discerned using machine vision, often combined with special lighting and/or non-visible spectrum sensors.”
Machine vision systems are also used to inspect solar panels. The many, specific applications machine vision offers illustrates the important role played by it in the solar cell industry:
Specific MV Applications in the Solar Cell Industry
Coatings, cracks, printing and edges of wafers
Defects, edges and format of cover glass
Soldering and spacing of cells in finished modules (panels)
Curved glass and mirrors
Wafers, wafer blocks and ingots
Control of Location
Recognize position and orientation of PV cells for robotic handling
Alignment recognition of soldering lugs for robotic handling
Having examined the solar cell market and the use of machine vision, what can we conclude? Based on the information presented, we find that demand for alternative energy will continue to drive solar cell and panel sales at impressive double-digit rates. This is very good news for the machine vision industry, particularly since current levels of solar cell and panel production lag demand, and machine vision offers a much needed productivity boost. According to Dr. Gord Deans, “the wafer, cell and module manufacturers have now grown to the extent that efficient automation of their manufacturing processes and plants is now required in order to meet their ongoing market demands."
As we have seen, several MV companies are positioning themselves to ride the wave of the solar cell industry. However, despite many opportunities for MV companies, market “entry (is nevertheless) complex due to high-tech production and very narrow qualification processes”, according to Lasse D. Nergaard. Still, further tweaking of MV applications used in the semiconductor industry could change that, resulting in sizeable market opportunities for the machine vision industry.
For more information on the MV market opportunity represented by solar cell production, please refer to AIA’s new 2009 market study, "Machine Vision Markets – 2008 Results and Forecasts to 2013," scheduled for release on March 31, 2009.