Silicon Block Inspection with SWIR Imaging
- Friday, 02 December 2011
In a world where oil production is declining, and where both nuclear energy plant and spent-fuel storage safety have proven to be inadequate, there is increased pressure on solar power generation to fill the gap. In response to the increased demands for energy, the photovoltaic manufacturing industry has focused on two primary objectives: driving down the cost of solar panels and increasing panel efficiency.
That first objective is being addressed by casting silicon into blocks, which has reduced the input costs for cell fabrication. There are significant costs, however, resulting from blindly cutting the blocks into cells. Silicon-based CCD or CMOS sensors are not able to inspect the blocks, as silicon detectors do not absorb wavelengths beyond 1140 nm and thus cannot image the longer wavelengths where the bulk silicon becomes transparent enough to see the defects.
This article will illustrate how shortwave infrared (SWIR) imaging arrays uniquely find voids and inclusions in solid blocks before they can damage diamond wire saws or result in bad or inefficient cells.
The Inspection Process
For a long time, silicon solar-cell raw materials were supplied by the infrastructure created from the integrated circuit business. In 2008, the solar industry consumed about half of available supply, and spot prices soared as a result. The greater demand brought both an increase in the production of high-purity crystalline silicon using traditional growing methods, but also the development of other methods of forming more affordable solar-grade (SoG) silicon. The metallurgists recognized that cast metallurgical-grade silicon (MG Si) could be purified from 98 percent pure to the higher level needed by solar cell production, which is pure to 6 nines or 99.9999 percent. As a result, silicon began to be cast in huge vats, cutting raw material costs to the wafer processors by several times, and greatly increasing production capacity.
In order to cut the 680 x 680 mm (26.8") ingots into blocks of 156 x 156 mm (6.1 inches) and then into thin wafers, diamond wire saws must be used. If there is a carbon deposit in one of the silicon blocks, the saw could be damaged or dulled. In addition, if an air bubble is captured in a block, significant production time is wasted, the lifetime of the saw is reduced, product yield falls, and costs are driven up. Therefore, it is vital to identify bad blocks before cutting.
Silicon detectors are not usable for wavelengths beyond ~1140 nm (where the silicon becomes transparent), but shortwave infrared detectors and imaging arrays made from indium gallium arsenide (InGaAs) see through the silicon, even solid blocks. Figure 1 shows the 1140 nm peak wavelength of silicon electroluminescence (red curve), which marks the end of the material’s absorbance region (green curve). Beyond this wavelength, silicon becomes transparent, as shown by the green curve, since the fall of absorbance implies that transmission is rising. The bandgap of InGaAs occurs at the longer wavelength of 1680 nm (blue curve), and therefore has absorbance to that wavelength, making it usable to image through the now transparent silicon. The low dark-current and high uniformity of modern InGaAs, developed at Sensors Unlimited – Goodrich ISR Systems (Princeton, NJ), makes it a valuable material for detection and imaging throughout the shortwave infrared wavelength band.