Tiny devices take on tough tasks in biomedicine.
Industries are becoming increasingly interested in the potential of diminutive machines. This is particularly true in biomedicine, where researchers of microelectromechanical systems (MEMS) are developing microsensors and microgrippers that are small enough to fit into blood vessels.
Biomedicine, in fact, is one of the few industries in which microstructures have already taken root. Disposable blood-pressure sensors made by companies such as Lucas Novasensor (Fremont, Calif.) and Motorola Inc. (Phoenix) have been on the market for more than 10 years.
Although roughly 10 million such devices are used each year, they are only a part of a growing market for micron-size sensors. "Revenues for silicon micromachined sensors rose from approximately $494 million in 1988 to about $746 million in 1991," according to Frost & Sullivan Market Intelligence (Silicon Valley, Calif.), an international technology research company. Strong growth is expected to continue throughout the forecast period (1988 to 1998), with growth rates near 20 percent during the mid-90s. By 1998, sales are expected to reach $2.26 billion, Frost & Sullivan projected.
In biomedicine, disposable blood-pressure sensors costing about $12 are displacing larger bonded strain gage instruments that are more expensive and cumbersome to use. At $200 to $500 apiece, the larger sensors are used several times before they are thrown away and must be sterilized after each measurement.
Disposable blood-pressure instruments are often used to monitor heart performance after surgery and in intensive-care centers. In the United States in 1984, medical professionals took roughly 6 million invasive pressure measurements. Roughly 1 million of these were performed with disposable pressure sensors. Today, most of the 10 million measurements are performed with disposable sensors.
To measure pressure the disposable sensor is placed in a tube of saline fluid connected to the patient's blood stream. Within the sensor the saline solution deflects a thin diaphragm. In Motorola's pressure microsensors, a piezoresistive pad ion-implanted into a silicon diaphragm receives up to 16 volts of direct current. As the diaphragm deflects under pressure from the saline solution, the piezoresistive pad strains. Voltage generated by the pad varies in proportion to the diaphragm's aphragm's deflection.
Motorola's standard sensor operates under pressures from 0 to 6 psi and between -- 15[degrees] and 40[degrees]C. After leaving the chip, voltage signals between 40 and 50 millivolts are amplified. Monitors display signal waveforms and digital values of systolic and diastolic pressure.
The Motorola pressure sensor is different from most others on the market in that the diaphragm has only one piezoresistive pad. Most other pressure sensors transmit measurements using four piezoresistive elements arranged in a Wheatstone bridge circuit. Since its sensor has a single piezoresistive pad, Motorola can put the mechanical diaphragm with electronic circuits that condition the signals on a single silicon chip.
Motorola introduced its pressure sensor in the early 1980s. Although its basic principle of measurement has not changed, the sensor package has been made less expensive and smaller so that it is less cumbersome in use. In reducing the size of the overall package, sensor designers were careful to develop a housing that does not put stress on the diaphragm, which can distort the accuracy of pressure measurements.
Another change to past designs is in the way saline solution contacts the diaphragm. In earlier generations, the saline fluid contacted the top surface of the diaphragm through a gel that provided hydraulic coupling and protected electronic circuitry. The fluid now contacts the bottom surface of the silicon diaphragm. Hence the electronic parts are less susceptible to corrosion from the salt in the saline solution. Motorola still coats the electronic circuitry with a gel.
Reducing size and cost are also goals for Novasensor, which for more than 10 years has been making disposable blood-pressure dies (the part of the sensor that directly senses pressure) and dedicated substrates that have signal-conditioning circuits. "Since these are to be used in millions of people and since they are disposable, one can easily understand that cost containment is an important issue," said Roger Grace of Roger Grace Associates (San Francisco), a technology marketing firm that works closely with Novasensor.
In the past 10 years, Novasensor has gradually been adding signal-processing capabilities to the sensor package. Recent designs include circuits that perform analog-to-digital conversion, amplify the sensor signal, refine the signal for accuracy under varying temperatures, and allow the device to be integrated with other types of sensors, Grace said.
Novasensor also produces a blood-pressure die that is small enough to fit on the tip of a thin flexible tube called a catheter. The pressure die is integrated into a sensor package and monitoring system by Millar Instruments Inc. (Houston) and sold to health-care centers. Certain heart-muscle diseases can be detected only by pushing a pressure sensor mounted on the end of a catheter through a blood vessel and into the patient's heart.
Making a Microstructure
To make the catheter-tip device, Novasensor uses bulk-micromachining technology, which integrates the semi-conductor manufacturing techniques of chemical etching and photolithography--methods common in the making of computer chips, in which only selected areas of a silicon wafer are machined.
In early designs, Novasensor also used anodic bonding in which a wafer of silicon is glued to a glass substrate that contains signal-processing circuits. Glass and silicon unite when the materials are heated to between 400[degrees] and 600[degrees]C and about 1000 volts is sent across the junction between them.
Subsequent generations of the sensor use silicon fusion bonding, which provides a smaller and less expensive chip. The technology developed by Novasensor allows silicon to adhere to silicon at the molecular level. Silicon molecules link up when they are heated to more than 1000[degrees]C.
This bonding technique helps to reduce the influence of thermal hysteresis on sensor accuracy, which is spurred by varying rates of thermal expansion between different materials. Use of silicon fusion bonding has helped Novasensor's designers reduce the size of the die. In the first-generation catheter-tip sensor, the length and width of the die were 1200 and 2400 microns; the current design is 900 by 400 microns.
Sensors of this size require complex precise prcessing. Automated pick-and-place robots and mechanical bonding machines deftly handle the parts, which are formed into silicon pressure dies and ceramic substrates. The creation of the catheter-tip sensor begins with finite element analysis. Novasensor uses Ansys, a finite element computer modeling and analysis system from Swanson Analysis Systems Inc. (Houston, Pa.), in the layout of several two-dimensional drawings.
Each drawing is a large-scale representation of a glass mask used in separate stages of die production. Drawing data are transferred to a magnetic disk, which is put in an automated mask-making machine. After the dimensions are reduced to scale, a step-and-repeat camera takes a photo of a full-size mask image and reproduces it thousands of times on a substrate. That way, the company can process 16,000 dies at a time on a silicon wafer that is 4 inches in diameter.
Masks are made from glass coated in certain areas with a thin layer of chromium that provides a barrier for rays of light. To generate the required shape in a wafer of silicon, the masks are used in a photolithograpic process.
In photolithography, a light-sensitive polymer film (called a photoresist) is deposited on a silicon wafer. The unit is baked to generate uniform material properties in the polymer. Then the wafer is put under a mask, which is irradiated with ultraviolet light.
The chromium pattern on the mask reflects the beams of light from the wafer, but the rest of the glass surface allows light to pass through. Beams of ultraviolet light alter the molecular structure of the polymer. Then a chemical such as potassium hydroxide, ethylene diamine pyrocatechol, or hydrazine selectively etches the material.
The manufacture of Novasensor's catheter-tip pressure sensor begins with the wafer being polished on its top and bottom surfaces. A polymer layer of silicon nitride, which is dielectric, is deposited on the two surfaces. Next, sensor makers diffuse or implant boron into the top surface of the silicon to generate piezoresistive junctions.
Aluminum is deposited on the same surface to provide voltage paths that stretch from the piezoresistive pads to electric junctions. Then the ultraviolet rays are beamed through a mask to open a cavity in the polymer layer coated on the bottom surface of the wafer.
The wafer is immersed in a chemical that begins etching a trapezoidal recess in the silicon. Etching begins at the bottom surface of the wafer and continues through the material toward the top surface. The process continues until a thin diaphragm is created in the wafer.
From the top of the wafer, the piezoresistive junctions form a Wheatstone bridge circuit that is used to generate the sensor's measurement. Two of the junction pads are in a region of the diaphragm subjected to relatively high stress; two are in a region subjected to lower stress.
During blood-pressure measurement, for example, the fluid generates loads that deflect the diaphragm. Diaphragm movement triggers the Wheatstone bridge circuit by generating varying values of strain in the piezoresistive pads.
Variations in strain from low stress to the high-stress pads are sensed by a voltage applied to the circuit. The sensor generates 0 volts when the diaphragm is at rest; it generates voltage values proportional to values of pressure when the diaphragm deflects.
After Novasensor's catheter-tip dies are formed, they are probed electronically. Those that are rejected are automatically marked and discarded. Then a computer-controlled cutting blade saws the wafer into separate dies. Next, robots pick up each die and automatically bond it to a ceramic substrate. Like the silicon wafers, the ceramic material is batch-manufactured into several individual but linked segments. After a wafer is bonded to each segment, the sections are broken apart to form individual units consisting of one silicon die and one ceramic substrate.
Sensor makers then connect thin gold wires to metal junctions on the die using thermal-compression bonding. Further, each sensor is subjected to thermal and pressure testing. A computer-controlled laser trims thick-film resistors on the ceramic substrate, which refines signals so that each sensor is similar to the one made before it. Sensor caps are bonded to the ceramic substrate using epoxy.
High-volume production of sensors is among the primary benefits of semiconductor-style manufacturing. Novasensor's plant in Fremont produces more than 6 million sensors each year.
Blood-pressure sensors are also used to make noninvasive measurements. Motorola makes sensors that connect to a cuff that wrap around the arm. The cuff is pressurized and then connected to an electronic sphygmomanometer to measure systolic and diastolic pressure.
Biomedical microstructures are not restricted to the monitoring of blood pressure. Health-care centers use the same piezoresistive-style design in other areas, such as the measurement of intracranial, uterine, and bladder pressures.
Teknekron Sensor Development Corp. (Menlo Park, Calif.) develops microstructures for use in biomedicine, which it licenses to biomedical systems manufacturers. One such device, which is now a prototype sensor, will be used in hospitals, physicians' offices, and homes to measure cholesterol. The sensor has deposits of enzymes that measure total cholesterol and its subfractions. Tekenekron said the device can help reduce the more than 500,000 deaths in the United States each year associated with the abnormal thickening and hardening of artery walls due to high levels of cholesterol, cigarette smoking, and high blood pressure.
Cholesterol analysis is conducted by placing a drop of the patient's blood on a chip. Reagents and steps that separate the blood sample are part of the sensor structure. In use the microsensor would link to a handheld processor/monitor. The monitoring system would reduce the risk of operator error and inaccuracies in measurement, according to Teknekron.
Separately the company developed a gas sensor that it claims can respond to measurements 100 times faster than traditional electrochemical detectors. The microsensors would be used to monitor a patient's metabolism in intensive care, ambulances, emergency rooms, and during surgery.
Teknekron calls the microstructure a Back Cell sensor because the sensed medium reaches measuring electrodes from the back side of a porous substrate. The device would fit near the patient's mouth and provide measurements of the levels of oxygen and carbon dioxide inhaled and exhaled. The monitoring system gives virtually instant feedback on the patient's breathing.
Detectors now on the market take a few seconds to sense sudden changes in breathing and provide only an average of what is inhaled and exhaled. Tekenkron is working with an undisclosed partner to integrate the sensor into a monitoring system. The MEMS developer said it hopes to see a product on the market within a year. The company must first seek the approval of the Food and Drug Administration (Washington, D.C.).
Microsurgeons Meet Micromachines
The micromachining center at the University of Wisconsin (Madison) makes micron-size surgical clamps that are being used by surgeons at the Rush-Presbyterian--St. Luke's Medical Center (Chicago) to sew blood vessels. In the suturing process, surgeons fold back a section of a vessel so that the inside surface faces out. The remaining loose end is then pulled over the folded vessel so that two inside surfaces are in contact.
Trouble emerges, however, when the folded section of the vessel tries to return to its former shape. A solution devised by Henry Guckel, a professor of electrical and computer engineering, and fellow researchers at the University of Wisconsin is a plastic ring about 150 microns in diameter that has three knife edges on its periphery, arranged 120 degrees apart.
When using the clamp, a surgeon threads the vein or artery through the ring and folds the vessel wall back on top of the knife edges. The surgeon then pulls the remaining end of the vessel over the top of the folded section and places it on the knife edges. Hence the blood vessel is held together and the surgeon is free to concentrate on sewing. Because of its tiny size and biocompatibility, the clamp can be left in the blood vessel wall.
Guckel is also developing microstructures with help from medical researchers at the university hospital in Madison. For example, he is designing a microcutter for use by neurosurgeons repairing abnormal dilations of blood vessels in the brain. If they are not reinforced, these aneurysms of the brain can lead to strokes when the blood vessel punctures.
Rather than slice into the cerebrum directly, surgeons prefer to install a blood vessel support by threading it through an artery that stretches from near the hip to the diseased site in the brain. The support is pushed through the arteries on the end of a catheter. Sometimes a balloon is used to support the blood vessel. Air is sent through the tube until the balloon inflates. Then the catheter is withdrawn and the balloon is left to support the blood vessel.
Detaching the balloon from the catheter, however, can generate problems. "Neurosurgeons here have found that as they try to detach the balloon they sometimes dislodge it within the blood vessel," Guckel said. Thus the medical researchers have built platinum coils that they use instead of balloons.
Like a balloon, a coil is pushed to the diseased site on the tip of a catheter, which is a stainless-steel cylinder inside a plastic tube. Unlike balloons, however, coils can be easily sprung open and released in the blood vessel by pushing forward on a sleeve inside the stainless-steel cylinder. The procedure is done with the patient under sedation, not unconscious. "Therefore the patient watches what is going on and can provide active feedback," Guckel said.
Coils are formed so that when they are released from the catheter they fit into the dilated blood vessels. One end of the coil remains connected, however, to the end of the stainless-steel tube. Surgeons now use electric current to dissolve the connection by electrolysis, but they prefer a method that is less harsh.
Guckel and his colleagues are working on a solution in the form of a microcutter whose shears move diagonally like the jaws of a chuck. The cutter would enter the body in parallel with the catheter. A piston driven by compressed air would actuate the jaws. A prototype will likely be ready within two years.
"What we are talking about is something that has jaws on it, can slice through the wire, and is operated by air," Guckel said. "In angioplasty, the catheter has a tube in it that supplies pressure to blow up the balloon. This pressure is enough to drive a pair of diagonal cutters. In our research we are getting to the beginning stage of developing actuators. Maybe by the end of the year we will demonstrate some of them."
Hydraulic and pneumatic actuators require geometries that are thicker than most micromachined parts. As a result, these devices are particularly suited to manufacture under the LIGA process (LIGA is a German acronym that translates into lithography, electroforming, and plastic molding). The technique is used to sculpt microstructures that are thicker than those made by other methods of micromachining (see "X-Rays with a Difference," page 64). The University of Wisconsin has already built prototype actuators using the process.
The Center for Retina Vitreous Surgery (Memphis, Tenn.) is also interested in using one of Guckel's microcutters. In this case a rotating or linear micromotor powered by pneumatics or hydraulics would drive the cutters. "This is a very complicated application because it's not only an issue of making a tool but also of locating the tool very precisely. You've got virtually no tolerances, and if there is a cut in the wrong place, the patient is blind," Guckel said.
"We tried to convince the center that a rotating device, something like a micron-size milling machine cutter, would work," according to Guckel. Surgeons at the center rejected the idea, however, based on the belief that a rotating tool would not cut the retina tissue. Instead, they believe the tissue would wind up on the cutter surface and tear. "Their contention is that you need something that cuts like a barber's scissor," Guckel said. The device is still in early development.
At the University of California's Berkeley Sensor and Actuator Center, researchers are designing a miniature gripper that might have several applications in biomedicine. They have already developed a prototype from silicon that is small enough to grip a single animal cell roughly 7 microns in diameter and 40 microns long.
The gripper loosens or tightens when 20 volts are applied to rows of cantilever beams, which are like combs with interlocking teeth that do not touch. The electric field generates an electrostatic force, measured in micronewtons, that runs parallel to the beams and is powerful enough to move the structure along with the gripper arms in the required direction. The arms move apart or together depending on where in the structure the voltage is applied. The silicon structure has grippers that measure roughly 10 microns by 2 microns on arms roughly 250 microns long.
A gripper like this would be extremely useful to doctors when mounted on the end of catheters and endoscopes, said Albert Pisano, a professor of mechanical engineering at the University of California in Berkeley. (Pisano directs the sensor and actuator center with Richard Muller, Richard White, and Roger Howe.) The microgripper needs to be bigger and more dexterous, however, to work on the end of a catheter inside blood vessels and body cavities. Hence the researchers are beginning to develop a prototype that could be 50 times larger than the cell gripper but still small enough to fit through holes and vessels less than 3 millimeters in diameter. They are making the second-generation gripper from nickel.
Pisano believes surgeons will be able to use the gripper to manipulate obstructions in fallopian tubes. Further, the gripper could function as a miniature tweezers for use inside the body. For example, two grippers and a miniature camera could enter the body--to take samples of tissues for lab analysis--through separate holes that are so small they might heal in a few days.
In addition to enlarging the gripper, the researchers are developing a feedback system that would tell what force the gripper is putting on tissue or other parts of the body. Such a device has the potential to help identify whether cysts are malignant or benign, Pisano said. The density of a cyst usually reflects its potential risk to patients' health.
Richard White is developing a microsensor for the same purpose. White's device has a miniature transmitter that shuttles high-amplitude Lamb waves across a hemispherical diaphragm to a tiny receiver. When human tissue is placed in contact with the diaphragm--between the transmitter and the receiver--the wave's speed is influenced. The tissue's density reflects directly on the length of time it takes the wave to travel across the diaphragm. Like the microgripper, this microsensor could be useful to doctors trying to determine if there is cancer in tissue.
Orthopedic researchers at the University of Wisconsin are investigating the feasibility of healing bones with help from microsensors. The tiny instruments would provide information on the evolution of the healing process. With such data, surgeons might speed the recovery process and reduce the costs of health care.
"We've found it is feasible to surgically attach a force sensor to a bone," Guckel said. "A more challenging issue is to develop a system to send and receive power and feedback signals without using wires or fiber-optic cables that would penetrate the skin--a process that could generate infection." Ideally a remote communication system would send power to the sensor and receive signals from it.
Microsensors are useful in this application because they require low voltage to operate, an asset that could permit the use of remote optical communications systems. The University of Wisconsin has developed microsensors that resonate differently in response to varying loads. The devices generate an electric signal proportional to force. The sensors operate when they receive [10.sup.-15] watts. University researchers have built components of the bone-measuring system that are undergoing tests, Guckel said.
Berkeley's Pisano, formerly an automotive engineer, said more mechanical engineers should work with microstructures. Many of the challenges associated with developing MEMS--such as reducing friction and increasing torque--are mechanically based, he said. Pisano presented the keynote address at the 1993 Bay Area Technical Conference on Microelectromechanical Systems at the University of California at Berkeley, May 1, which was sponsored by ASME.
Guckel favors including many disciplines in MEMS development. "It's not a luxury any more to say you are going to cooperate with another group," he said. "In many cases you cannot survive alone. My contention is that knowledge does not exist in one group. It exists in several groups, and we should combine it."
|Printer friendly Cite/link Email Feedback|
|Date:||May 1, 1993|
|Previous Article:||Taking a titanium bite.|
|Next Article:||Thinking the unthinkable.|