Image-processing in the near infrared spectrum is already well established in the photovoltaic industry where it is used in demanding optical inspection procedures that check solar cells for micro-cracks, which can occur during crystal growth and wafer sawing. But this technology has much more to offer and it is gaining in importance in other areas, as this article shows.
Industrial image-processing in the near infrared (NIR) spectrum is a powerful non-destructive analysis technique that is used on production lines for quality assurance and to increase productivity. This technology is proving to be a reliable manufacturing tool for inline inspection and classification of a range of products.
For instance, line-scan camera systems that are responsive in the NIR spectrum are currently used to check solar cells and printed circuit boards containing electronic parts, and for large web and coating inspection.
Cameras that operate in the visible spectrum generate data based on RGB color differences. The visible spectrum permits the analysis of the surface layers of a component, but electromagnetic radiation in the NIR spectrum (with a wavelength of 0.75–1.4 µm) can penetrate deeper into an object, providing information about its internal structure far from the surface.
This is a key reason why many companies are investing time and money in developing analysis tools and imaging systems that use NIR to ‘observe’ how an object interacts (absorbs or emits) with electromagnetic radiation at specific wavelengths to provide information and data about its material properties, characteristics and structural make-up.
SOLAR CELL MANUFACTURING
The photovoltaics industry is one sector that has been quick to see the benefits of using these tools.
Photovoltaic solar cells, fabricated from crystalline silicon wafers—which are typically 100–200 µm thick—can pick up defects at any stage of manufacture.
Whenever these delicate wafers are handled mechanically the applied stresses may create micro-cracks that can grow unpredictably, leading to mechanical failure of the wafer during subsequent processing stages.
It is important to remove wafers with critical micro-cracks early in the production process to prevent both wasted processing of already ruined wafers, and production-line stoppages caused by contamination from shards of wafers that shatter during processing.
Manufacturers of solar cells are constantly under pressure to find ways of increasing efficiency, improving quality and reducing costs.
‘‘A properly designed automated optical inspection (AOI) system can help meet these goals,’’ commented Xing-Fei He, Senior Product Manager, Teledyne DALSA Inc, a manufacturer of digital imaging products for various industrial, scientific and medical applications.
‘‘AOI systems are growing rapidly as the solar industry is reaching the stage of maturity in which machine vision undergoes mass adoption on the production floor.’’
There are numerous ways in which AOI aids in error detection and process control in a typical manufacturing environment. By using line-scan cameras, NIR is playing a major role.
One approach is backlight and luminescence inspection at NIR wavelengths. Under backlighting conditions, a crack will scatter light and create a dark line against a light background that is readily detectable. Sensors need a resolution of 2–8 k pixels with 7–14-µm pixel size to be able to detect these defects.
Another highly effective test technique uses electroluminescence or photoluminescence. When excited, either electrically or optically, silicon luminesces in a band around 1050 nm in the NIR spectrum. High-resolution machine vision can detect micro-cracks, which appear as fine, dark lines in images resulting from excitation from a NIR source.
However, both techniques are not perfect. ‘‘Sensor and camera selection are critical in developing an AOI system for backlit micro-crack inspection,’’ explained Xing-Fei.
‘‘The challenge is that charge-coupled device (CCD) image sensors normally have lower quantum efficiency (QE) at NIR wavelengths, resulting in a relatively weak signal. Moreover, camera systems vary considerably in their NIR sensitivity. Some cameras exhibit as much as 30–40% QE at 900 nm while others can be much lower.’’
Silicon NIR luminescence also can so inefficient that the output intensity is quite weak. To generate useful images, cameras need to integrate using times that are often too long for in-line inspection at today’s increasing production-line speeds. Sample wafers often have to be inspected off-line using stationary area-scan cameras.
To satisfy the photovoltaics industry’s need for high-speed imaging systems, sensor developers and manufacturers constantly seek better alternatives. One approach is to use a sensor based on indium gallium arsenide (InGaAs) —which has greater sensitivity at 1.1 µm—instead of one that uses a silicon CCD. Unfortunately, this technology is also very expensive.
A technique called time delay and integration (TDI), combined with a high-resolution line-scan machine-vision camera, creates a system that has the potential to meet many of the challenges that solar-cell manufacturers are now facing. Line-scan technology enables images to be captured of wafers that are moving steadily along a conveyor at production-line speeds. Using a ‘multiple exposure’ technique, TDI cameras effectively achieve higher responsivity and a better signal-to-noise ratio.
Teledyne DALSA cleverly combined key elements—enhanced QE in NIR and TDI technology—when it developed its Piranha HS NIR line-scan camera. Currently in production its responsivity is comparable to that of InGaAs-based systems, but it is more cost-effective (see boxes ‘Visible and NIR imaging using one camera’ and ‘Time delay and integration technology’).
“Improved sensitivity in the NIR region is critical to the success of equipment in a number of applications,” continued Xing-Fei.
“The Piranha HS NIR’s TDI technology enables multiple exposures for orders of magnitude increases in sensitivity while maintaining low noise performance—this is ideal for high-speed and low-noise applications.”
MUCH MORE TO OFFER
It is not surprising that technology which is having a major impact on inspection performance and efficiency in one industry is generating considerable interest in other areas.
NIR cameras are also being used by the printing industry to check the security features woven into paper that is used to make bank notes. Bank-note production is becoming increasingly more complex. Currency is often printed at multiple locations, but must still conform to a single set of stringent printing requirements.
Paper money also includes a range of security features, such as embedded magnetic strips and watermarks, and printing processes must be checked to ensure that they are complying with specific standards.
‘‘It is difficult to inspect watermarks on currency paper using visible light, so NIR is often used for this application,’’ explained Xing-Fei.
Typically systems use a combination of drive mechanisms to carry the sheets across a CCD linear array that captures visible and NIR images of patterns, which are checked against stored templates.
While it is necessary for currency paper to be rigorously inspected, checking packaging used to protect pharmaceuticals is also essential. A particularly challenging application is verifying the structure of Braille embossing.
According to UK-based PharmaBraille, since October 2005, European Union (EU) member states have been required to have in place legislation that conforms to EU Directive 2001/83/EC. This requires that all products authorized after October 30, 2005 carry Braille identification. In May 2009 the International Association of Diecutting and Diemaking announced ‘Can-Am Braille’—a set of guidelines and recommendations for the use of Braille on packaging in USA and Canada.
Small faults in the manufacture of packaging used for medicines can have dangerous consequences. For instance, a Braille dot that is too small, or not of sufficient height, can be confusing or even incorrectly represent medicine and dosage information. Manufacturers must, therefore, implement thorough quality control procedures to eliminate faults.
The Braille font dot height is indiscernible to untrained fingers so the characters are often assessed by mechanical inspection, but this can cause damage and be time consuming. As an alternative, optical inspection systems can reliably recognize Braille dots and evaluate their tactile quality.
‘‘Transparent dots deposited on types of pharmaceuticals packaging are particularly challenging,’’ said Xing-Fei.
“Visible light cannot be used to inspect these dots because some materials that are used are transparent, so one approach is to use NIR. In this way manufacturers can guarantee 100% inspection of their packaging.’’
MEDICINES AND HEALTHCARE
NIR technology is not only applicable to the packaging, but also to the drugs it contains. In the production of medicines it is used for quality control, as Brad Finney, Vice President Sales, North America, Teledyne DALSA, explains.
‘‘NIR is used to inspect tablets. It looks at where the active ingredient is physically located or how it is distributed throughout the tablet. Although it is not used at production-line speeds it is capable of analyzing samples taken from production runs.’’
Finney continued: ‘‘In healthcare, Optical Coherence Tomography (OCT) is a well established application of NIR, where it is used to acquire images of a patient’s retina. In virtual histology its use in areas that require deeper tissue penetration represent an up-and-coming market segment, and current research and development work is looking at taking the technology from the laboratory to a mainstream commercial environment where it can be used to diagnose melanoma and breast cancer by helping to detect and analyze tumors on, or near, the surface of skin. NIR sensitivity is also needed to image samples and slides in digital pathology, which relies on special fluorescent dyes.’’
In addition to its use in the semiconductor industry, NIR imaging could provide an effect way of checking the integrity of circuits based on organic materials, in the rapidly developing printed electronics industry, says Xing-Fei.
Printed electronics are electrically functional components or optical inks that are deposited on a substrate to create active or passive devices, such as thin-film transistors or resistors.
The production technique is expected to bring about widespread, low-cost, low-performance electronics for applications such as flexible displays, smart labels and animated posters and, therefore, has the potential to be a large market for companies currently developing and selling NIR imaging systems.
Camera systems based on NIR technology continue to evolve and are finding use in an expanding range of applications—some of which are well established, while others are still at the development stage.
Recently, researchers in Canada, at Université Laval, showed that NIR/SWIR inspection performed better than ultrasound thermography, and optical pulsed/lock-in thermography, in experiments that set out to analyze glass-fiber reinforced plastics (GFRP) for various defects—such as de-lamination, porosity, undesired inserts and material discontinuity—that may appear while this material is being manufactured. GFRP is commonly used in the aerospace industry—a sector where quality control is of paramount importance and, it appears, where inspection procedures based on NIR are again showing their usefulness.
Dale Deering is P. Eng, Senior Program Manager 1D Machine Vision. Brad Finney is Vice President, North American Sales. Dr. Xing-Fei He is Senior Product Manager. All are with TeledyneDALSA.
Xing-Fei He and Nixon O, Time Delay Integration speeds up imaging, Photonics Spectra, May 2012.
Xing-Fei He, Uniquely Challenging, Visions Systems Design, February 2010.
Reza Shoja Ghiass, Yuxia Duan, Kira Peycheva and Xavier Maldague, Inspection of glass fiber reinforced plastic (GFRP) using near/shortwave infrared and ultrasound/optical excitation thermography, International Workshop on Smart Materials, Structures & NDT in Aerospace (in conjunction with NDT in Canada 2011 Conference), 2–4 November 2011, Montreal, Quebec, Canada.