By Richard L. Anderson 
  

ORNL is playing a role in developing new measurement and control technologies, including smart sensors and intelligent measurement systems, and will benefit from using these magical tools in future research. 
 

Clark's 3rd Law: "Any sufficiently advanced technology is 
indistinguishable from magic." 

Advances in science and technology are causing remarkable changes in measurement and control. Microcomputers are becoming ever faster and more capable; sensors are getting smarter and cheaper; and microscopic instruments, mass produced by semiconductor manufacturing processes, can be packed onto a single chip. Also appearing on the scene are adaptive networks of intelligent sensors; sensors that analyze and control manufacturing processes; wireless communications for relaying data collected by sensors; and new measurement technologies with resolution at the atomic level, such as the atomic force microscope and its derivatives. Just as the industrial revolution has boosted our muscle power and the computing revolution has enhanced our brain power, we are on the brink of a "smart sensor" revolution that will extend our senses to enable more sensitive probes of natural and altered environments and the creation of smart machines.  

The most advanced research in electronics aims at developing quantum devices whose dimensions are comparable to the free path of the electron; as a result, their behavior can be described only by quantum mechanics. R. C. Ashoori recently reviewed the state of quantum electronics in an article entitled "Electrons in Artificial Atoms" in Nature. In this newest field of electronics, single electrons can be manipulated (e.g., by counting them) and a fractional electronic charge can be measured. A sensitivity of 10,000 times that of the best semiconductor electrometers has been demonstrated. Researchers at the National Institute of Science and Technology are developing standards of electrical charge and current by using sets of single-electron transistors to count the electrons one by one. In a seven-junction device, they have achieved uncertainties of fewer than 13 parts in one billion. Electrometers are often used with radiation detectors such as ionization chambers.  
 

Fig. 1. Dick Anderson studies a typical silicon microsensor developed at ORNL. Shown in the background is another example, a "critters on a chip" sensor that processes signals from bacteria attached to the chip that are genetically engineered to light up in the presence of certain pollutants.

ORNL is playing a role in developing new measurement and control technologies, including smart sensors and intelligent measurement systems, and will benefit from using these magical tools in future research (see Fig. 1). 

Many sensor technology advances are driven by the automobile industry's need for inexpensive, reliable measurement and control technologies. The electrical and electronic components making up the electromechanical systems of a modern automobile now account for about 25% of its cost, and the percentage is growing. For example, the industry needs an increasing number of smaller, smarter sensors—devices that receive and respond to a signal or stimulus, like ultrasonic sensors that enable grocery store doors to "see" you coming and open automatically to let you in. Older cars had only a few sensors, such as those connected to dashboard gauges that indicated when the car was low on oil or gasoline or the engine was running hot. In modern automobiles, sensors feed data to microprocessors controlling actuators that activate mechanical devices, ensuring smooth operation of anti-lock brakes, cruise and traction controls, and fuel injection systems. Even more advanced sensors, actuators, and computer controllers will be incorporated into future cars for satellite-based navigation systems to help drivers reach their destinations faster and radar collision avoidance systems to reduce the number of automobile accidents. 

Because of the increasing number of sensors, actuators, and controllers in multiple systems, the automobile industry is driving the development of fast sensor-bus technology that will reduce the weight and complexity of the wiring harnesses. For the many cars that are started outdoors in winter when the temperatures are below freezing, the electronic components under the hood must be environmentally hardened to withstand rapid heating from subzero temperatures to near 100°C in a few minutes. Because of the volumes required, meeting the automotive industry's need for cheap but sophisticated and robust electronics is sure to have a drastic impact on the pace and economics of intelligent microsensors development. As the number of applications grow, the prices will be driven down. 

Fiber-optic Sensors: 
Emerging Technology 

New sensor technologies, such as fiber-optic sensors, are starting to replace conventional sensors for many industrial applications. For example, electric utilities are beginning to use fiber-optic sensors to measure pressure, temperature, fluid level, and strain in power plants because such sensors are immune to electromagnetic interference. Fiber-optic strain sensors are being incorporated into "smart structures" that are then stiffened through feedback control. There is increasing interest in more reliable noncontact or nonintrusive measurements for which fiber-optic and other optical measurement methods are well suited. 

Fiber-optic sensors work by detecting changes in the features of light (e.g., wavelength or intensity) between the time it enters and exits optical fibers—light pipes the size of strands of hair—as a result of an external perturbation (such as pressure on the fibers). The source of the light may be laser diodes. An example of a medical technology that uses fiber-optic sensors is the lung diagnostic system being developed partly by ORNL to identify lung disease based on the sounds generated through a patient's breathing. Another ORNL-developed example of fiber-optic sensors is the weigh-in fiber-optic device, a "scale" for trucks. This technology being developed by Jeff Muhs and Steve Allison, both of ORNL's Engineering Technology Division, takes advantage of the unique properties of optical fibers made of silicone. Unlike a glass or plastic fiber, a silicone fiber is flexible—it can be squeezed or stretched, and the amount of compression or expansion can be measured by changes in light transmission through the fiber. In this way, silicone fibers embedded in roads can be used to weigh trucks. If a silicone fiber on a chip can sense pressure at various positions in the body, it may be used for monitoring blood pressure, pulse rate, breathing (chest expansion), knee bending during physical rehabilitation, and foot pressure distribution. 

Virtual Instruments 

In 1987 the "virtual instrument" (VI) was introduced into the measurement and control world through a software package called LabView. Using this bit of computer magic, a measurement system can be programmed graphically by simply drawing the wiring diagram connecting instruments, represented as icons, on the computer screen without writing a single line of code. Another powerful feature of LabView is its ability to encapsulate a complex measuring system into a single icon that can, in turn, be used as a component that can be connected to another component in another wiring diagram. This process may be repeated to produce a highly complex VI. Each VI thus created may be used as an interface to a set of physical instruments, or it can be used in a simulation to test a conceptual instrument or measurement system. Thus, many concepts of a measurement instrument can be tried out simply by simulating them on the computer rather than by constructing and testing hardware, saving time and money. 

VIs are in widespread use at ORNL. Doug Lowndes assembled one in the Solid State Division to control a system for processing silicon solar cells. I&C Division engineers, along with the Metrology Center at the Oak Ridge Centers for Manufacturing Technology, constructed several VIs for the Army Calibration Laboratory in Huntsville, Alabama. More recently, Bill Holmes was awarded first prize by National Instruments for his VI implementation of a remotely operated harsh environmental effects laboratory in ORNL's Chemical and Analytical Sciences Division. VIs are also being used for the remote operation of research electron microscopes at ORNL and elsewhere. 

Coupling Microprocessor and  
Sensors on a Chip 

Imagine that you are walking down a street in a residential area. You are carrying a cell phone. The wind is blowing. You see a house and a brush fire nearby. You notice that the fire is moving toward the house. You sense the house is in danger, so you dial 911 on your cell phone and report the situation. You have done three things: sensed the situation, processed the information (wind blowing, fire burning toward house), and transmitted a message wirelessly that communicated the danger to firefighters who could provide help. 

A part of the new magic in measurement and control technologies is the ability to combine sensing, processing, and transmitting capabilities on a single chip. It has not been easy. Most sensors generate analog signals from contact with their environment (such as rising temperatures, pressures, or levels of acidity) that vary continuously. Digital electronics in the form of microcomputers and signal-processors use digital, or discrete bits of, information (i.e., ones and zeros). To feed data from a chip's sensors to its signal-processing circuits, an analog-to-digital converter must be included on the chip. It converts the received analog information into digital information that can be processed and transmitted (by a communications module that converts the processed signal into a bus message leaving the chip). In this way, a chip can sound an alarm or actuate a valve when the temperature, pressure, or acidic level gets too high in a manufacturing process. A big advantage of on-board processing using single chips is that it crunches the raw data into a less dense stream of information that is less of a hardship on the rest of the system. In the language of electrical engineers, it reduces the bandwidth requirements on the communication network. Another advantage is that when the signals are converted to digital form for transmission, they are virtually immune from noise pickup. 

New semiconductor technology provides us with two kinds of magic: sensors can be combined with computing and transmission capabilities on a single silicon chip, and proven silicon chip manufacturing processes can be used to make sensor chips, reducing the manufacturing cost in the same way that improved semiconductor manufacturing technology lowered the cost of computer chips (by orders of magnitude). 
 

Fig. 2. Reversing classical economic rules, the magic of Moore's Law means that advances in semiconductor technology result in similar trends in instrumentation, ever-increasing complexity and capability, but with ever-decreasing size and cost.

Thanks to more magic using electronics, sensors are becoming smarter and smaller. Additional capabilities are being integrated into the sensor chip, and at the same time, new miniaturization technology crams these capabilities into smaller and smaller packages (See Fig.2). Eventually, stand-alone instruments will disappear and be replaced by measurement systems on a chip. 

More and more new measuring instruments and transducers (devices that convert input energy of one form to output energy of another form, as shown in Fig. 3) are controlled by silicon-based microprocessors, thanks largely to the increased availability of application-specific integrated circuits (ASIC). Many of these instruments also have silicon-based sensors. The reduced cost of sensors that are mass produced by semiconductor technology will make the introduction of larger numbers of sensors in manufacturing and process control systems economically possible. 

Fig. 3. A generalized picture of sensors as energy converters using a thermocouple as an example. Adapted from Expanding the Vision of Sensor Materials, Committee on New Sensor Technologies: Materials and Applications, National Materials Advisory Board, National Research Council, National Academy Press, Washington, D.C., 1995.

The very nature of sensors is changing. In addition to the primary sensing element, additional sensing elements are now incorporated with the signal processing on the same chip to measure and correct for ambient conditions. For example, if you measure the manifold pressure in your car's engine, you will need to correct for the temperature changes of the pressure sensor, so it's useful to have a temperature sensor as well as a pressure sensor on the chip. This combination of sensing elements is one feature of a "smart" sensor." In another concept, a measuring unit may employ many sensors integrated through the on-chip computer to measure parameters inferentially that cannot be measured directly. For instance, the outputs of several transducers could be integrated to provide an "efficiency" sensor on a heat pump that indicates whether it is operating at full efficiency. For example, by measuring changes in airflow and temperature across the heat exchanger, the sensor could detect that the airflow channels in the heat exchanger are clogged, resulting in lower efficiency. A measurement of indoor and outdoor temperatures would allow a calculation of maximum thermodynamic efficiency and provide a standard against which other measured factors that affect the heat pump's efficiency can be compared. Pressure sensors in front of the expansion valve and in the heat exchangers could be integrated into the efficiency meter to detect coolant leaks, indicating a loss of efficiency. 

Instruments-on-a-Chip 

Stand-alone instruments that were once commonly used in research laboratories are being replaced by instruments-on-a-card technology. A standard widely used for instruments-on-a-card today is the VXI standard, which defines card sizes, connector geometry, power supplies, etc. With the continued trend in miniaturization, the next logical step in this transition is the instruments-on-a-chip (IOC), and eventually the "lab on a chip." ORNL has developed a postage stamp-size lab on a chip that, compared with commercial analytical instruments, can analyze a DNA fragment 10 times faster and analyze a liquid sample 10,000 times smaller, saving materials. In addition, less labor is required because the device is computer controlled. 

Engineers in ORNL's I&C Division are working with Mike Ramsey of the Chemical and Analytical Sciences Division on the development of a lab on a chip for analyzing small molecules and DNA molecules. I&C personnel are integrating electronics into the chip while Ramsey's group works out the wet chemistry, under a Defense Applied Research Projects Agency contract. 

These developments will have a major impact on how instruments are used. Calibration intervals can be extended dramatically, because with redundant, inexpensive instruments on a chip, instruments can also check each other. IOCs will greatly reduce volume and power requirements. They will be fabricated from mostly standard, chip-level modules. Custom instruments will be made easily and affordable as short-run ASIC chips. 

The lab on a chip is one of several microelectromechanical systems (MEMS) being developed at ORNL. Other types are microcantilever sensors, medical telesensors, and pressure sensors. A MEMS chip is made by incorporating the mechanical and electronic parts into a silicon wafer in one smooth process. Microcantilever sensors will each incorporate one or more microscopic-coated "diving boards" on a silicon chip that bends or vibrates in a characteristic way in response to changes in pressure, temperature, moisture, chemical content, or other environmental variables. An array of more than 600 such tiny sensors etched on a semiconductor chip can form a "nose on a chip" for detecting potentially hazardous gases such as mercury vapor, natural gas, and carbon monoxide. Because these microcantilever sensors are so small, they are sometimes called "magic dust." Medical telesensor chips that can be attached to a soldier's skin or placed in the ear are being designed at ORNL to detect and relay wirelessly to medical personnel any alarming changes in body temperature, pulse rate, and blood pressure. The goal is to quickly identify wounded soldiers who should be able to return to fighting after rapid treatment.  

Dispensable Silicon Chips 

Throwaway silicon chips can already be found in singing greeting cards, credit-card-size pocket calculators, and talking toys. In the future, throwaway silicon labs on a chip the size of microscope slides may be used in hospitals and doctors' offices for single analytical measurements of body fluids (blood, urine, and saliva) to aid in disease diagnosis. 

By 2025 the U.S. Army plans to have unattended microsensors that can be dropped onto battlefields from fighter airplanes or helicopters. Such sensors may obtain and transmit information on (1) the identity of chemical or biological warfare agents that may be present or (2) the types and locations of enemy vehicles. Sensors sprinkled over the battlefield also could be used for battle damage assessment-determining if enemy soldiers or vehicles are still functioning after a targeted area has been attacked. (Only targets that survived the initial attack would need to be retargeted, saving ammunition and time.) 

Intelligent Measurement Systems 

Since the early 1980s computer-controlled instruments have become widespread; in fact, they probably represent the majority of measurement systems today. These so-called "smart" instruments have been commercially available since the mid-1980s. Like older instruments, they incorporate a simple microcontroller and continuously produce data. They often have digital communications capability, can be remotely adjusted for range and calibration, and are able to make local corrections for ambient conditions. 

The successor to smart instruments and data-acquisition systems is the intelligent measurement system. The intelligent measurement system will possess local analytical and decision-making capabilities. Instead of sending out a continuous stream of data, it may process and reduce the data and communicate this information only when interrogated or when a measured variable has changed. While the smart system produces mounds of data, the intelligent measurement system will communicate only meaningful information, significantly reducing the load on the communication system and central processing units. In this distributed intelligent system, individual sensors may pass information to intelligent actuators for direct action rather than passing the information up to a central processor and back down to the actuator. 

An intelligent measurement system will use more of the information available from a suite of sensors and transducers. For instance, with the tremendous computing power available in today's microprocessors and digital signal processing chips, fluctuations that are filtered out or suppressed in "dumb" sensors can be analyzed in real time. These process fluctuations can be analyzed to provide more immediate and more detailed information about the process (or sensors) than is now available. In many cases this integration of measurements can represent significant savings, because the output of a single sensor can be analyzed to provide multiple outputs. In the case of a fluidized-bed control, for example, a pressure transducer will do more than provide measurements of average system pressure. By analyzing the pressure fluctuations in the bed and by applying chaos theory, the intelligent pressure sensor can provide an output that can be used to maintain operation of the fluidized bed in a stable mode by modulating the fluidizing gas flow into the bed. 

An intelligent measurement system will likely contain multiple transducers whose outputs are integrated by the intelligent measurement system to provide measurements of inferred process variables (e.g., the heating value for a power plant, the efficiency of a heat pump). Such a system could also have readings corrected for environmental or process error sources (e.g., ambient temperature effects on the transducer output or corrections for emissivity in radiation thermometry). 

Fig. 4. A conceptual design of an intelligent process measurement and control system for the production of composite materials. Adapted from a National Research Council study on new materials for sensors.

Fig. 4 shows a conceptual design of an intelligent process measurement and control system for the production of composite materials. This example was adapted from a recent National Research Council study on new sensor materials. Along the bottom are the variables measured by the physical sensors. Along the bottom of the box underneath the dotted line are the process variables that are directly measured by the process sensors. Above the dotted line are process variables that are inferred from a process model and the measured variables. The goal of the intelligent measurement and control system is to tailor the desired properties of the composite material by controlling the process parameters inferred from the measured variables. A similar concept, described by the Next Generation Manufacturing Project, is to have a detailed process model running in real-time within the intelligent closed-loop controller. Of course, implementing such a plan will require a detailed understanding of the process through more extensive measurements that must be made to construct the model. 

An intelligent measurement system will also employ the distribution of appropriate processing power throughout the system, allowing control of nonlinear systems that are at present poorly controlled or not controlled at all. An intelligent measurement system will be designed with human safety and comfort, better access to information, and other factors in mind. Operator errors and mistakes can be minimized or eliminated by incorporating human factors into the system's design. Intelligent measurement systems will lead the way to better process control and, hence, to improved product quality.  

Sensors will be important in intelligent measurement systems that are designed to ensure that the operation of automated systems does not have unintended consequences. To understand the need for such an approach, consider the 1979 loss-of-coolant reactor accident at Three Mile Island. The accident got worse because of a problem with the reactor control systems. The sensor that indicated a rise in temperature of an overflow vessel was in an out-of-the-way location in the control room, so the reading was not seen by an operator until it was too late. In modern and future nuclear power plants, such a reading would be flagged on a reactor control computer screen as part of an intelligent measurement system. 

As great as the impact of computer technology has been, it is only the beginning. Computer-based measurement and control systems will take advantage of advances in semiconductor technology to provide more functions and information while driving down the cost per function, as has the semiconductor industry. Improved measurement and control technology has been highlighted as a major need in several industry road maps, and the needs of the automobile industry are driving the development of measurement and control technologies that could benefit almost all manufacturing industries. The next few years promise to be challenging and exciting for measurement and control research at ORNL and throughout the world as the Laboratory's engineers and computer programmers work their magic. 

BIOGRAPHICAL SKETCH  RICHARD (DICK) L. ANDERSON has been chief scientist for ORNL's Instruments and Controls (I&C) Division since 1992. He received a B.S. degree in chemistry from the Massachusetts Institute of Technology and a Ph.D. degree in physical chemistry from Rice University. He worked on temperature standards at the National Bureau of Standards and the Physikalisch-Technische Bundesanstalt (Braunschweig). In 1974 he joined the I&C Division to head the Metrology Research and Development Laboratory and then became leader of the division's Measurements and Sensors Group. In 1989, he joined the staff of SEMATECH, returning to ORNL three years later. He is a fellow of the Instrument Society of America and the American Physical Society. He has two patents and has won two R&D 100 Awards.