Cars, Clothes, and Computers: Help for Industry

ORNL has developed measurement and inspection techniques that may increase the competitiveness of the U.S. steel, textile, and semiconductor industries.

The white stripes of phosphor on this metal specimen glow red when illuminated by a black light. Some phosphors have no trouble surviving and functioning in high temperatures such as those produced by this propane torch.

Cars get us to work, clothes make us feel good at work, and computers help save us work. And it takes work on the part of American industry to manufacture the materials that provide us with these luxurious necessities. In making these materials into products, manufacturing operations are often plagued with inefficiencies. Too much defective material is being produced. Too much energy is being wasted. Too many workers are spending too much time looking for occasional flaws.

Automated measurement and inspection methods are needed to increase the production of high-quality material, reduce the production of defective material, save energy, and free up personnel to work on tasks that increase a company’s earnings. In other words, the goal is to raise production efficiency to reduce the costs of products, improve their quality, and increase their market share.

ORNL has the capabilities to develop measurement techniques that will help industry become more competitive. Two of our techniques are considered promising for increasing efficiency in the steel and textile industries and one is already helping the semiconductor industry. Our technologies may help increase the quality of cars, clothes, and computers.

Thermometry for the Steel Industry

Cars today don’t rust the way older vehicles did. The reason: The steel industry uses a “galvannealing process” to produce the corrosion-resistant sheet metal now used in virtually all the world’s automobiles. The process combines zinc atoms with iron atoms in a steel surface at high temperatures. The protective layer of zinc-iron alloy that is formed prevents the steel from rusting through. In fact, because of this galvanneal coating, lifetime guarantees against rust through can be offered by the automotive industry.

Getting the galvanneal coating right for automobiles and other products is not easy. First, a sheet of steel is dipped in a liquid bath of zinc at about 450°C. Then the steel sheet passes through a cascade of furnaces, raising its temperature to as much as 700°C. During heating, iron atoms from the molten steel sheet drift into the zinc coating to form the zinc-iron alloy. But, is the molten steel surface always at the right temperature to ensure formation of the best galvanneal coating? Making sure the temperature of galvanneal steel is on the mark has long been a problem for the steel industry, a problem that ORNL is helping to solve.

The galvannealing process of alloying zinc with iron at the surface must be controlled at production rates of 30 meters per second or higher to ensure the surface quality necessary for the automotive market. When the galvanneal coating is incorrectly formed, the material is rejected as second-rate steel, costing the U.S. steel industry $4 billion per year and reducing its competitiveness with steelmakers worldwide. Hence, getting these coatings consistently right was identified by the U.S. steel industry as the key to the future competitiveness of their galvanneal product line.

The problem is that the surface alloying process varies as the temperature of the metal surface changes, yielding a product of nonuniform quality. One challenge has been to devise a method that accurately measures the temperature of the molten material as it forms an alloy and cools. A second challenge has been to relay information instantly to steel producers so they can adjust furnace operation to get the right temperature—and best product.

To address these challenges, the American Iron and Steel Institute (AISI) accepted a proposal by ORNL and the University of Tennessee at Knoxville (UTK) to develop a totally new, first-principle-based technique for determining the surface temperature of galvannealed steel. The ORNL and UTK engineers designed and built a novel instrument system in collaboration with National Steel, the partner steel company. The project is part of the Advanced Process Control Program supported by 15 AISI-member steel companies and the Department of Energy’s Office of Industrial Technologies. Bailey Engineering of Mechanicsburg, Pennsylvania, is now developing the concept of the prototype instrument built at ORNL into a commercial product that will be available soon to the steel industry.

“Real-time steel temperatures cannot be measured precisely using the conventional method because it assumes that the surface properties are constant,” says Steve Allison of ORNL’s Engineering Technology Division, a principal developer of the technique. “The problem is that properties of the zinc-covered surface rapidly change as the coated steel cools from a molten to solid state, causing errors in the temperature measurement by as much as 40°C. Because our device uses a thermal phosphor method, it has demonstrated accuracy within better than 3°C. Clearly, it is more reliable than the conventional method.”

How does the thermal phosphor technique work? A steel sheet is partly dusted with white phosphor powder using a computerized phosphor-deposition system. Two optical fibers are positioned between the moving steel sheet and the temperature measurement equipment. As the sheet travels between the furnaces at up to 30 meters per second, short pulses of ultraviolet light are fired from a low-power nitrogen laser through an optical fiber leading to the molten steel. The laser pulses excite the phosphors, which emit light for a short time based on how hot they and the steel substrate are. The emitted light travels through the other optical fiber to a light detector (photomultiplier tube). It measures the time for the phosphorescence to decay, and a computer uses the real-time data to calculate the surface temperature of the galvannealed steel.

“To apply the phosphor to the moving sheet,” Allison says, “we had to solve some interesting problems in mechanical design, fluid mechanics, and optics. We had to figure out how to illuminate the phosphor and gather the light for temperature measurements. So we assembled a team of diverse skills and expertise from ORNL’s Engineering Technology and Instrumentation and Controls divisions, UTK, and National Steel.”

The team was asked to determine whether phosphor powder might damage the quality of the steel. Results of tests done by ORNL and National Steel indicated no adverse effects on either the coated steel’s surface appearance or its ability to be painted.

Other ORNL co-developers of the technique were Wayne Manges, Ruth A. Abston, William Andrews, David L. Beshears, Michael Cates, Eric B. Grann, Timothy J. McIntyre, Matthew B. Scudiere, Marc L. Simpson, David N. Sitter, and Todd V. Smith.

Early prototypes were tested at National Steel’s Midwest Steel Division in Portage, Indiana. On May 31, 1998, the final version developed at ORNL was successfully demonstrated on a galvanneal line at the Bethlehem Steel plant in Portage. The demonstration was part of DOE’s Technology Showcase held at this facility, where the system is permanently installed. 

The new process should result in less second-rate material and eliminate the need for costly off-line tests to determine if the galvanneal coating is correct. Accurate, reliable temperature measurements will ensure a quality product, reducing waste and saving energy. These improvements, if implemented throughout the U.S. steel industry, could save steelmakers as much as $70 million a year, increasing their competitiveness worldwide. And such a savings might lead to more affordable cars or, at least, larger earnings for the steel industry.

ORNL engineer Ruth Ann Abston adjusts the phosphor deposition device in front of the galvanneal sheet.
Schematic of galvanneal phosphor thermometry components. A thin phosphor layer deposited on the steel strip is illuminated using laser light. The duration of measured fluorescence from the excited phosphor layer indicates the steel’s temperature.

Inspection System for the Textile Industry

Defective material, a grudgingly accepted by-product of textile production, continues to be a problem in the American textile industry, long a source of our clothing, carpeting, medical dressings, protective outerwear, and automotive airbags. Erratic loom operation and errors in human inspection in U.S. textile mills result in excessive, reduced-price, second-grade merchandise. But technology may soon be a boon to textiles.

Reducing costs through technology is considered a key to making the industry more competitive. Currently, offshore competitors are a major economic threat. Since 1980, the U.S. textile industry has lost approximately 400,000 jobs to foreign manufacturers. If this trend continues, another 600,000 jobs may be lost by 2002.

To stem these losses, ORNL, the Oak Ridge Y-12 Plant, and other DOE laboratories have been working with the U.S. textile industry to weave new technology into textile operations. Under a collaborative research and development agreement (CRADA), a team of Oak Ridge researchers—including Glenn Allgood, Dale Treece, and John Turner, all of ORNL’s Instrumentation and Controls (I&C) Division, have already helped the American Textile Partnership. They have developed automated inspection for the weaving industry under the CRADA’s Computer-Aided Fabric Evaluation (CAFE) Project. Automated inspection is expected to improve quality and lower costs, increasing worldwide demand and market shares for U.S. textiles.

Today’s textile mills use high-speed looms to weave yarn into cloth. Inspectors in the mill manually feel and visually examine the cloth, looking for defects. Sometimes they become fatigued and miss the less obvious ones. Sometimes defective material passes through the entire textile system into the marketplace. If the fabric’s flaws are noticed by retailers or consumers, the material is marked down or returned as a loss.

One-inch image plane of 28 pick goods is captured electronically for display. The cloth has a horizontal and vertical structure typical of all woven material. The horizontal threads (fill yarns) are the ones imaged and analyzed by the pick measurement device.

Because of current off-line inspection methods, potentially thousands of yards of defective, off-quality material could be made before the problem is recognized. Therefore, to provide 100% reliable inspection, the CAFE industry partners asked the participating laboratories to invent new technologies to automate and improve the current process for inspecting fabric while it is being woven. This was an immense task, considering that the U.S. textile industry produces 5805 square miles of cloth a year.

Oak Ridge researchers led by Allgood first had to familiarize themselves with the complexities of cloth and textile manufacturing. “We had to know about cotton and polymers, different weave types, and yarn thicknesses,” Allgood says. “We observed the weaving process in which horizontal filling, or ‘pick’ yarn, is laid down at a right angle to the ‘warp’ yarns that run lengthwise through the fabric. We learned that the regularity and density of pick yarn—that is, the number of strands per inch—strongly affect fabric quality and the amount of yarn used.”

So they set out to develop an optical device to measure pick density and to locate in real time fabric flaws—unexpected variations in the measured pick density, such as missed picks or picks that are too close together or too far apart. The result was the pick measurement device, which is cheaper than conventional camera systems used for textile inspection and easier to install on existing looms. The device uses a laser to bounce light off the yarn with a set of cylindrical lenses arranged to gather the reflected light and build an image of only picks, not warp yarns. The image on the array appears on a desktop computer screen as a one-dimensional image of the fabric with each pick represented.

An ORNL-developed algorithm enables a computer to evaluate all the images, count the picks, calculate their density, and spot density variations. By matching these variations with known flawed patterns, it locates and names defects according to their classification. The computer will sound an alarm and can display an electronic defect map in the textile mill control room when flaws are first detected. Thanks to this nearly instant feedback, mill operators will be able to correct loom operation quickly to minimize the production of second-rate material. The map also provides data on the amount and quality of material headed for the mill’s “downstream processes” such as printing, dying, cutting, and sewing operations. As a result, material yields are optimized and second-rate material is not processed for sale as a high-quality product. The pick measurement device team received a Technical Achievement Award at the 1998 Lockheed Martin Energy Research Corporation Awards Night ceremony.

“We also developed an economic model so that our industry partners would know what the inspection system had to be able to do for the price to make it a worthwhile investment for the textile industry,” Allgood says.

The pick measurement device prototype tested in Oak Ridge was installed in 1997 at the Glen Raven plant in Burnsville, North Carolina. It was then moved to the Institute of Textile Technology (ITT) in Charlottesville, Virginia, for final testing. The device has been patented and licensed to ITT, which—with the help of Appalachian Electronic Instruments, Inc. in Ronceverte, West Virginia—will manufacture, market, and sell the device to the textile industry. The cost of the commercial device is estimated to be $1200 and the cost of using it is estimated to be 1% of that for a human inspector. It is believed the device will find widespread use in the textile industry and will help make U.S. textile manufacturers more competitive in both U.S. and world markets.

Defect Recognition for the Semiconductor Industry

Ken Tobin (left), Shaun Gleason, and Tom Karnowski show the display of results from the spatial signature analysis algorithm they developed.

The U.S. semiconductor industry is reducing waste, raising its productivity, and lowering its costs in producing electronic components for computers, thanks to a software tool developed at ORNL. Working with SEMATECH, which was created in 1987 as a partnership between the U.S. government and the semiconductor industry to make the U.S. semiconductor industry more competitive, ORNL developed an algorithm that recognizes defect patterns on silicon wafers and identifies the manufacturing problems causing the defects. This “spatial signature analysis” (SSA) tool has been licensed to 14 semiconductor manufacturers and equipment suppliers. Because the manufacturing process involves hundreds of steps, the opportunities for defects to form are many. Defects in the dies on the wafers—tiny luminescent squares dotting 8-inch black disks—mean that these dies for carrying traces of electrical current are unfit for use in microprocessor chips, the “brains” of desktop computers. A certain pattern of defects, or signature, usually indicates a particular manufacturing problem. For example, a scratch across many dies on a wafer could be a sign of mechanical mishandling of the wafer by an industrial robot.

“SSA rapidly extracts only meaningful information from huge amounts of data on the wafers obtained from lasers and microscopes,” says Ken Tobin, a senior research scientist at ORNL and one of the developers of the algorithm. “It quickly identifies defect patterns and traces them to manufacturing malfunctions, enabling industry engineers to find and fix the problem fast.”

“An excellent use of this tool is the detection of scratches caused from wafer handling in real time,” says Marylyn Bennett of Texas Instruments in Dallas, Texas. “If we could automatically detect and prevent scratches alone, the potential savings would be about $100,000 for every lot of wafers saved.” SSA was fully integrated into the Texas Instruments data management system in November 1997. Lockheed Martin Energy Research Corporation (LMER), which manages ORNL, recently licensed the ORNL-SEMATECH SSA technology to 14 companies, many of which are SEMATECH member companies. The licensees are eight semiconductor manufacturers—Advanced Micro Devices, IBM, Intel, Lucent Technologies, Motorola, National Semiconductor, Rockwell, and Texas Instruments—and six semiconductor equipment suppliers—Applied Materials, Defect and Yield Management, Inspex, KLA-Tencor, Knights Technology, Inc., and ADE, Inc.

SSA technology was developed by Kenneth Tobin, Shaun S. Gleason, and Thomas P. Karnowski, all of the Image Science and Machine Vision Group in ORNL’s I&C Division, under a CRADA between the Defect Reduction Technology Group at SEMATECH in Austin, Texas, and DOE, which provided half the funding.

A scratch-and-spray deposition pattern is shown on this wafer map.

In 1998, the developers won three awards for the SSA technology: a Technical Achievement Award from LMER, a Marketed Technology Award from the American Museum of Science and Energy in Oak Ridge, and a Department of Energy Federal Laboratory Consortium Award for Excellence in Technology Transfer.

Other industries that may benefit from this knowledge-based tool include manufacturers of textiles, flat panel displays, and optical and magnetic disks.

ORNL has developed measurement technologies that may help reduce or eliminate waste in the steel, textile, and semiconductor industries. For consumers, these advances may mean higher-quality and lower-cost cars, clothing, and computers.

Where to?

[ Next article | Search | Mail | Contents | Review Home Page | ORNL Home Page ]