Rapid, increasing growth has manufacturers producing about 70GW of solar cells per year. In addition, the fuel cell market saw growth of nearly $1 billion from 2013 to 2014 and more than 50,000 units shipped, totaling 180MW in 2014. Online monitoring can reduce manufacturing costs, provide tighter control on each process step, and save staff time by detecting potential problems during manufacturing. The Department of Energy’s (DOE) National Renewable Energy Laboratory (NREL) was recently issued a patent for a method achieving this dream for fuel- and solar-cell manufacturers.

Online monitoring

NREL’s technique characterizes materials via wide-angular illumination on a conveyor belt or roll-to-roll processing platform. Spectral imaging and reciprocal optics then assess material features such as thickness, surface conditions, and uniformity. A light source and in-line camera record the reflectance image of the wafer/cell moving at relevant speeds, such as 5ips. A high-speed computer then transforms the reflectance images into the appropriate parameter images.

“We started with solar cell process monitoring for silicon-based cells,” says Bhushan Sopori, principal engineer at NREL. “The wafers are typically carried on conveyor belts from one process step to another. While moving on the belt, one can rapidly measure the reflectance of each wafer.”

Different process steps can be characterized in terms of reflectance measured in different spectral bands, and these concepts can also be applied to fuel cells and many other thin-film products.

Though online monitoring techniques, such as minority carrier lifetime and defect distributions, have been previously available, these technologies tend to be complex, expensive, and limited to monitoring one process step – not fully integrated into production lines. NREL’s fast and cost-effective optical imaging monitoring also offers advancements in camera technology to provide high-speed data storage, high-speed data handling/manipulation, and fast image processing algorithms for high-resolution imaging within a broad spectrum. NREL’s characterization technology was established with support from DOE’s Office of Energy Efficiency and Renewable Energy, through both the Solar Energy Technology and Fuel Cell Technology Offices.

NREL’s fuel cell monitoring process.

Reflectance imaging

The process monitoring technology has evolved from a single-wafer reflectance imaging system called PV Reflectometer and uses the same principles:

Reciprocal optics illuminate a large-size sample. The test wafer/cell is illuminated with an optical source that has a large angular divergence. The reflectance normal to the wafer/cell is collected (as a true reflectance value).

Reflectance measured at specific wavelengths. The choice of wavelength is dictated by the fact that the reflectance at that wavelength is directly related to the amplitude of the parameter. For example, to monitor the thickness of anti-reflective (AR) coating, a filtered image of the AR coated wafer at 800nm is transformed into AR coating thickness image.

Reflectance imaging can be done in two modes: specular reflectance mode and diffuse reflectance mode, with illumination that can be either normal to the wafer or at an oblique angle. Each of these approaches offer various advantages. Normal illumination and scattered (diffuse) reflectance is used for measuring surface roughness. Illuminating large-area (156mm x 156mm) wafers through a broad angular spectrum and collecting the reflectance normal to the sample, is useful for solar cells (and large-area samples). Filtered images, at specific wavelengths, allow direct mapping of some material device parameters. These parameters include AR coating thickness, wafer thickness, and surface roughness.

“We measure scattered light or diffuse reflectance from the object. Hence, only a small part of light will be scattered in the direction of the camera,” Sopori says. “However, if the incident light has wide angular divergence, scattering from a wider angular reflection can be collected by the camera.”

Sopori explains this idea was originally used to develop a reflectometer that could measure reflectance of rough or textured surfaces.

“New techniques are engineering designs of this reflectometer that do only a few things with a lower accuracy but do them fast,” he continues.

Illumination mapping

In NREL’s SolarLine inline prototype, wafers/cells are transported over a conveyor belt, which carries them into the vision zone. A system of linear light sources (of wide angular dispersion) continuously illuminate the belt and the wafers/cells over it. A line camera, fitted with a filter assembly, records the reflected light of the wafer/cell as it moves within the zone.

The choice of the filter is based on the transmission of the reflected light so it can be proportional to the parameter value. The camera resolution (12,000 pixels/line) and the data acquisition speed are compatible with conveyor belt speeds up to 5ips. A high-speed computer collects the data for each line and then assembles them into a reflectance image using the encoder (for position) data.

A computer controlled driver manages the conveyer belt’s motor. The position signal required to synchronize the reflected image with the corresponding line in the object plane is obtained from an optical encoder that rides over the conveyor belt. The line camera has interchangeable lenses to match the width of the object. Typically, the system is aligned so that the imaged object width is slightly larger than the wafer/cell width. The image of the wafer/cell is then isolated from the larger image that includes a small part of the conveyor belt using the software’s reflectance criteria.

The software recognizes the wafer/cell reflectance and isolates the test device for data manipulation, and transforms the reflectance images into the appropriate parameter image.

For a multi-crystalline wafer, it can measure grain size and grain orientations and dislocation density. The results are obtained when running the system at 2ips. This speed is dictated by the mechanical stability of the wafer transportation system, not imaging system. Full images of the test wafers/cells for illustrations are active only when SolarLine is in manual mode. In automatic mode, the system will extract and save only the information pertinent to the parameter/process being monitored.

As fuel cell markets continue to develop and expand, and products like automotive fuel cells become commercially available, the need for high-speed inspection for manufacturing continues to increase. In particular, because fuel cell stacks contain tens to hundreds of cells in series, ensuring that each individual cell is free of defects is critical to stack lifetime. With this in mind, NREL scientists have taken the same reflectance methods originally developed for solar cell materials and applied them to a variety of fuel cell applications, including membrane electrode assembly components for polymer electrolyte membrane fuel cells.

NREL’s linescan camera for fuel cell monitoring.

The technique can perform thickness imaging and discrete defect detection for a wide range of fuel cell membrane materials. It also has been shown to be highly sensitive to the surface quality and morphology of fuel cell electrodes. The system for fuel cells uses the same optical system as the SolarLine with slight modifications to optimize the configuration, optics, and data processing for fuel cell monitoring. The technology can detect bubbles, scratches, and divots; find defects as small as ~10µm; and perform mapping of these blemishes during the fuel cell manufacturing process at speeds of up to 60ft per minute. Fuel cell line equipment includes:

  • Linescan camera with a 12" field of view
  • Fiber optic light source with cylindrical lens
  • Encoder for camera timing
  • High performance computer
  • NREL-developed software

Industry impact

The high-speed (2,000 wafers/hr in a single wafer line or line speeds of many tens of feet per minute for fuel cell component production), line camera-based reflectance mapping system for online monitoring of many material parameters can be adapted to any existing production line.

The patent is expected to be licensed and sold to solar and fuel cell manufacturers as well as other applications including photovoltaic technologies such as cadmium telluride (CdTe), and copper indium gallium selenide (CIGS).

Illumination mapping during fuel cell monitoring.

Mike Ulsh, manufacturing R&D project lead for NREL, sees the commercial benefits of implementing this invention in a number of industries including manufacturing of fuel cell components, semiconductor wafers, glass, and coatings.

“Introduction of this technique has the potential to help decrease the cost of producing materials in a variety of manufacturing industries,” Ulsh says. “It would likely have the largest impact on reducing cost in high-throughput environments, such as roll-to-roll processing facilities, because it can characterize materials at a speed of tens of feet per minute.”

National Renewable Energy Laboratory


About the author: Arielle Campanalie is the associate editor of TES and can be reached at 216.393.0240 or acampanalie@gie.net.