Advances in thin films technologies are transforming traditional manufacturing processes and whetting the appetite of industry.
Multiple new techniques in thin films technology are vying for supremacy in two rapidly growing sectors—microelectromechanical systems (MEMS) and highly-efficient photovoltaic systems.
The silicon world is not alone in feeling a shake-up from advances in thin films, but MEMS and photovoltaics are hot right now. Some new processes in these fields are relying on a hybrid of well-established techniques to eliminate traditional problems and limitations in known methods.
The push is on to find new methods that blend low cost with high manufacturing volume.
Problem solving selenium  HelioVolt makes CIGS-based solar cells using a proprietary FASST process. Image: HelioVolt |
Cells made from silicon are stable, but have matured in both production levels and potential efficiency. Amorphous silicon, which can be readily fabricated, has subpar efficiency ratings compared to the best polycrystalline cells.
Silicon wafer-based cells, however, may soon give way to a new generation of photovoltaics. According to The Information Network, a New Tripoli, Pa.,-based market research company for the semiconductor industry, manufacturing methods using cadmium telluride, CdTe, and copper indium gallium selenide, CIGS, will see a tremendous boost in research dollars for future equipment manufacture. A 275% increase in investment by 2010 is expected, which is a testament to their anticipated impact.
CIGS and CdTe rely on the deposition of nanoparticles of the precursor materials on the substrate, followed by in-place sintering. The U.S. Dept. of Energy's National Renewable Energy Laboratory, Golden, Colo., says the best CIGS cell now matches the efficiency of the best polycrystalline-silicon cell, about 19.5% efficiency. Potential applications include newspaper-like roll-to-roll printing and CIGS-based inks.
One such company anticipating large demand for its CIGS process is HelioVolt, Austin, Texas, which applies CIGS films to building materials, such as metal and glass. Company founder B.J. Stanbery claims its CIGS solar cells are far more efficient, in relation to cost, than their silicon counterparts. He says an efficiency level of 12% to 15% should be commercially viable.
Stanbery says he began work on CIGS shortly after the first 10% efficient cadmium indium selenide (CIS) cell was developed at Boeing Aerospace in 1982. Stanbery, who was also at Boeing Aerospace, devoted his career to overcoming the limitations of the co-evaporation process then used in making CIS cells.
Besides reduced manufacturing costs, a big advantage of the CIGS solar cell, he says, is that like a silicon cell, it is intrinsically stable. However, the rapid thermal processing techniques used to create CIGS cells ran into repeated problems with selenium.
"Unlike silicon, it turns out CIGS is not homogeneous. It is nano-structured and has an alpha and beta phase. To get that structure, you have to get lots of selenium in there," says Stanbery. Selenium has good electronic properties, but it is very volatile and tends dissipate unless the process is done slowly.
In 2001, Stanbery developed a breakthrough that avoids excessive re-evaporation of selenium called field assisted simultaneous synthesis and transfer, or FASST. The CIGS film is synthesized in two steps, starting with the deposition of two materials layers on separate surfaces, one of which is a photodiode, and the other a printing plate. The precursor layers are pressed together to induce a chemical reaction. By applying an electrostatic current, the stack acts like a capacitor, trapping the selenium vapor. Because the compression force is inversely proportional to the square root of the gap, the process "pulls (the layer) down to an intimate contact on an atomic scale," says Stanbery."Conceptually, it came out of anodic wafer bonding," he continues, which is used extensively in MEMS fabrication. The advantage to his custom hybrid method is that it saves time and can be adapted to high-volume production methods, such as offset printing presses.
HelioVolt's approach is not the only one. Other companies are successfully manufacturing cells using CdTe, and even sulfur has been adapted to replace selenide in CIGS.
Last month, the Univ. of Delaware's Institute of Energy Conversion (IEC), Newark, developed a CIGS film on a flexible substrate of high-temperature polyimide. The multi-layer stack contains molybdenum, CIGS, cadmium sulfide, zinc oxide, and indium tin oxide, and boasts an efficiency conversion of 10%. The cell can be manufactured on long sheets of 25.4-cm wide polymer, and Robert Birkmire, director of IEC, reports they have made continuous sheets up to 34.3 m long. According to Birkmire, CIGS has numerous advantages over amorphous silicon, which is, cell for cell, far less efficient. He anticipates seeing the flexible CIGS cells achieve up to 15% efficiency in commercial applications.An important key to the method, says Birkmire, was finding a polymer to withstand high temperatures. The base material is exposed to temperatures of up 450° C during the deposition process, which lasts several hours. Stainless steel, which also flexible, could be substituted, he continues, but because it is conductive it is an unsuitable base for monolithic fabrication. Polyimide, on the other hand, has attractive electrical characteristics.
The process "has a lot of potential. The challenge is scaling up manufacturing," he says. IEC is working with Ascent Solar, Littleton, Colo., to develop ways to bring the material to market. The Univ. of Delaware is also part of an alliance of interested companies, including GE and Sandia National Laboratories, which is investigating commercial applications for these new cells.
Streamlining the micro etchIn the world of silicon, new microfabrication methods are paving the way toward ever-more sophisticated MEMS devices.
According to Shekhar Bhansali, associate professor of electrical engineering and nanomaterials at the Nanomanufacturing Research Center at the Univ. of South Florida (USF), Tampa, many of the newest approaches in thin films are "an art form." The hard part, he says, is moving them to an industrial process.
"People have been trying to build perfect micro for the last 20 years," he says.
"I would speculate in five years you would see some real complete systems hitting the market. There is still a lot of system development work."
Research by Bhansali and a team of graduate students from USF's Dept. of Electrical Engineering has resulted in a low-cost corner cube retroreflector (CCR) for potential use in a MEMS device. Rahul Agarwal, Scott Samson, and Sunny Kedia developed custom deposition techniques including a new two-stage deep reaction-ion etching process.
A CCR is a micro-device in silicon that features both vertical mirrored surfaces and a transparent window. Various MEMS dies require an optically transparent window to obtain optical access to the parts while providing protection from the environment. Optical switches, optical sensors, and tiltable mechanical mirrors are important elements to many MEMS devices, and successful microfabrication depends on etching, bonding, polishing, and deposition.
Optical CCRs permit transmission of data from sensors to a base station or network via a probing laser beam. The microdevice has three orthogonal flat mirrors—one active, two static—forming a concave corner. Under certain conditions, light entering the CCR is reflected back parallel to the source. Tilting and realigning a mirror, the CCR can intermittently reflect light using very little power. Modulating the mirror action transmits information.
The intent of the USF researchers was to provide a lower-cost solution to a packaged optical micro-device, which meant choosing an etching technique using a chemically-reactive plasma. The basics of reaction-ion etching (RIE) are well-established, but getting it to work properly, says Bhansali, is a challenge.
"It is very hard to get consistent results with deep RIE. We were not able to find a machine that is up to the level we need to get the results we want," he says.
Two other options were available, wet anisotrophic etching, and LIGA, a German acronym for lithography, electroforming, and moulding. Wet etching produces smooth surfaces, but the crystallographic nature of the process limits geometry. LIGA is prohibitively expensive because it requires x-ray exposure.
Taking smaller stepsLike some chemical vapor deposition processes, RIE depends on the creation of plasma using a radio frequency electromagnetic field. Ions created by the field react chemically with the substrate to produce an etch pattern.
Plasma enhancement is popular, Bhansali says, because it allows much lower temperatures. Some processes are already down to 250 to 300° C, but the integrity of the film can degrade with lower temperatures, he continues.
Similarly, controlling the etch pattern is difficult—vertical sidewalls, which are crucial to the mirror action of the CCR, are hard to achieve. And etched sidewalls are rough and scalloped, scattering light.
"The big question is, 'Can you get a perfect 90° wall?' We are going to buy another deep RIE machine that claims it can do just that. Once you get a true 90° wall, you are able to do a whole bunch of stuff. You can make active microdevices that work," he says. "If you have parallel surfaces you can make micromachines with gears that work and pumps that will not leak."
The answer was a two-stage approach. First, an initial deposition of a thin layer of 100-nm low-pressure chemical vapor deposition nitride was placed on a 10.16-cm silicon wafer. The nitride film was reaction-ion etched using plasma, followed by a very short silicon deep RIE of 10 µm, using a photoresist mask.
The silicon was then bonded to Pyrex for packaging purposes, and a 1,000 Å layer of aluminum was then applied to the mirrored surfaces using metallic sputter deposition, a common physical vapor application technique. A further hurdle was jumped when they circumvented harmful bubbling in the sputter process by switching from a conventional dropper to micropipettes.
Indenting and lithography was used throughout the fabrication process to define the etching areas, but yet another etching technique using potassium hydroxide was required to smooth the rough edges of the CCR sidewalls.
The two-stage deep RIE approach was crucial in minimizing micro-loading, and gave them a useful 89.8° sidewall angle. Completed CCRs were successfully tested for wireless communication to a distance of approximately 10 m.
Refining the process was the key to successful fabrication, says Bhansali, whose research has also focused on improving the reliability of various MEMS devices by introducing new sputter deposition methods. Although optical thin films will see the biggest surge in interest and funding in the near feature, he believes, silicon is making strides.
"Everything's going nano. The fun is going to be making them functional. There was a time when carbon nanotubes were the flavor of the month. The trick is to make money out of it," he says.
—Paul Livingstone