Machine Design

Precision moves with pint-sized motors

Innovative actuators and stages based on piezoelectric materials give superfine motion in a compact package.

Leland Teschler
Executive Editor

A typical use for PZT actuators is in the precision positioning of microscope objectives. The MIPOS 500 stage (top) from Piezosystem Jena Inc. is one example of a device designed for this application.-It travels up to 500 m and has a stiffness of 0.27 N/ m. General-purpose PZT actuators generally come in a stainless-steel case, but Nano-P-Series devices from Mad City Labs Inc. (bottom) are Invar, a low thermal-expansion alloy as strong as stainless steel. The coefficient of thermal expansion (CTE) of Invar is 23 times smaller than that of aluminum. This gives the Nano-P devices thermal stability and minimizes positioning errors from thermal drift. Range of motion can span tenths of nanometers to tens of micrometers depending on the model.

The typical PZT stack contains thin metal sheets between discs of PZT ceramic. The sheets serve as electrical contacts. PZT doesn't handle pulling forces well, so actuators typically incorporate a preload. Most manufacturers encapsulate the stack in a polymer to guard against contamination, though actuator maker PI has devised a ceramic coating it says better withstands humidity. PZT stack actuators can only tolerate axial stress, so they may use ball or flexible tips to help avoid tilting and shearing forces.

PZT motors from Nanomotion Inc. wiggle like an inch worm as they move loads. The stack has a hole through the center and rides on a shaft. The stack expands, a clamp on the top locks, the stack contracts, and the clamp on the bottom locks while the top clamp lets go.

PZT is good for more than just micrometerscale movements when it is built into linear motors. Motors from EDO Corp. use PZT stacks to push objects at top speeds of 300 mm/sec through a sequence of events depicted here. The action end of the motor consists of a shoe supported on two ends by PZT stacks. The device drives objects by driving one of the stacks on and off, forcing it to lengthen and contract, thus pushing the shoe against the object surface on each cycle. To push the object in the opposite direction, the motor applies a drive signal to the PZT stack on the other side of the shoe (top). The motor can produce rotary motion when it pushes on a disc (middle). Ordinary motion controllers such as those from Galil Motion Control Inc. and any number of position encoders can combine with the motors to define closed-loop systems (bottom).

Flexures that have been wire-EDMed into a stage are visible in this view of a device from Mad City Labs Inc. (top) Manufacturers can etch flexures to give stages movement in just one direction or in both X and Y. In the latter case the motion controller may use a software algorithm to compensate for any coupling of motion between the axes. A typical application for such stages is in positioning optical elements in microscopes. Alignment of optical fibers in telecom gear also makes use of specialized PZT stages. For example, a device targeting this area is the Pentor five-axis stage from Piezosystem Jena Inc. (bottom).


They can precisely move loads a few micrometers or spin a disc at a rapid clip. Piezomotors and piezoactuators are based on stacks of piezoelectric crystals. They got their start in vibratory transducers that generate sonar signals. As the technology has advanced piezoactuation has migrated into ultrasonic cleaners, remote level sensors, and to a variety of other uses characterized by minute but precise movements.

Indications are that the technology could play a bigger role in industrial and consumer products. The reason: Piezomotors can be compact and silent. And they've gotten less expensive in recent years. These qualities have aroused the interest of consumer electronics companies. Piezomotors are under consideration for use as drivers of DVD read heads and even for spinning DVDs.

All piezomotors synthesize motion by means of special polycrystalline material that exhibits a piezoelectric effect: The application of an electric field to the crystal in the direction of its main axis makes the crystal get thicker. This small increase in thickness can be harnessed for useful motion.

The material that serves as a basis for piezomotors is a compound of lead (Pb), zirconium (Zr), and titanium (Ti) oxide powder. The powder is fired and electrically polarized to align material domains along a primary axis. The finished material is a ceramic dubbed PZT.

There are several suppliers of PZT material. They can source PZT as raw powder, as pressed and fired discs, or as complete PZT "stacks." PZT stacks consist of PZT discs electrically connected so the displacements of the individual elementsadd up. The maximum stroke of a PZT stack is proportionalto the stack length. Stacks typically can deflect to about 1.5% of their length.

The majority of PZT devices used commercially are called low-voltage stacks. This moniker distinguishes them from PZT actuators that need in excess of 500-V drive signals. The lowvoltage versions have thin layers (about 0.1 mm) to permit deflection with electric fields of lower intensity. Use of thin layers, though, dictates that these devices be pressed and fired as one unit. This contrasts with high-voltage actuators comprised of thicker PZT layers that are assembled in a discrete manufacturing step.

The simplest actuators put the PZT stack in a metal casing together with a preload in the form of a spring. A plunger or rod atop the stack couples stack movement to a load. The actuator casing is usually stainless steel, but other metals may be used to promote heat dissipation in high-power devices that handle tasks such as vibration control.

PZT stacks generally get coated with a polymer to protect against humidity. One recent development is an actuator that replaces the polymer with a ceramic encapsulation. Its developer, Physik Instrumente (PI) L.P., Auburn, Mass., says the cofired ceramic encapsulation handles much more humid conditions than ordinary polymers. The structure also doesn't outgas in a vacuum, making it a candidate for use in semiconductor manufacturing processes. In addition, the ceramic coating also gives a better surface for position sensors that attach directly on the stack. And a Curie temperature of 320°C lets the device work in hot areas and under heavy loads that heat up the actuator.

The first applications for stack actuators were in generating sonar signals with enough power to bounce off submarines and echo back to the source. The motion created for this application is strictly oscillatory. Single-axis actuators deliver similar kinds of action when put to work controlling fuel valves in engines and other metering applications. Typical travel distances for these actuators are measured in hundreds of micrometers or less.

However, it is possible to configure PZT stacks so they produce more than just micrometer-scale motion. One technique for synthesizing larger movement configures a PZT stack to work against a surface via a friction shoe. Applying a positive voltage to the crystal lengthens it. This lengthening forces the shoe against the surface. Friction lets the shoe move the surface by some small increment. Removing the voltage from the crystal forces the shoe to retract away from the surface and repositions the shoe for the next cycle. Reapplying the positive voltage repeats the process.

The technique just described is that used by EDO Corp. of Salt Lake City. The company drives a PZT crystal at a 130-kHz rate in this configuration. It says this rate can produce linear velocities of about 130 mm/sec.

Users can gang EDO motors together to boost applied force if need be. It is also possible to convert this linear motion to circular motion by letting the EDO device push the outside radius of a disc. The piezomotor has an additional advantage over traditional motion equipment such as ball screws in that there is no damage to the device if an outside force stalls the load.

A single EDO motor delivers a nominal driving force of 1.3 N with about 5 N available through ganging. A related device from Nanomotion Inc. (Ronkonkoma, N.Y.), lists a driving force of 4 N for a single unit with up to 32 N possible with several combined. Its nominal speed is slower than what EDO provides, about 40 kHz.

The Nanomotion motor also drives loads with a friction shoe. But its mode of operation differs from that of the EDO device in that it contains PZT stacks having a hole through their center and which ride on a shaft. Their action is analogous to shimmying up a pole. The stack expands, a clamp on the top locks, the stack contracts, and the clamp on the bottom locks while the top clamp lets go. The overall effect resembles that of an inch worm. (See MACHINE DESIGN, 7/8/04, "Coarse and fine combined," p. S4, Semiconductor Manufacturing Equipment Supplement.)

PZT crystals behave quite differently over temperature and many developments from suppliers aim at compensating for these effects. For example, EDO says it will soon finish new driver electronics that allow for temperature-induced fluctuations in the resonant frequency that PZT exhibits. Ordinary means of measuring temperature such as RTDs or thermocouples are too slow for such applications. EDO says it instead devised a proprietary means of tracking how the piezo effect itself changes with temperature and feeds this information back into the controller electronics. The new scheme is expected to provide steady performance from 10 to 85°C.

On Stage
Though PZT actuators can give displacements in the range of tens of microns, they have trouble delivering such distances in a true straight line. The problem is that the actuator stack corkscrews, twists, and tilts as it expands. Though such effects are minute, they are enough to throw off high-precision-processes that can stand only about a nanometer of deviation from a straight trajectory. Moreover, the taller the stack, the more pronounced are these effects. (Makers of PZT stacks say an aspect ratio of about 10:1 for stack length-to-width is the maximum practical. Stacks exceeding this level are generally too compliant to control.)

The way manufacturers compensate for these material properties is to couple the actuator into a stage assembly carrying sophisticated guiding systems. One widely used technique is a flexure guiding system cut via wire-EDM ( electrical discharge machining) into a metal platform. The flexures are basically features etched into the stage that are designed to flex. They can be viewed as hinges that are stictionless and have no friction.

The usual approach is to put a flexure at each of the stage's four corners (though some devices use only two flexures). The hinges are designed so they are stiff in all directions except the direction of motion. So pushing on the center of the stage produces pure linear motion.

In general, stages built with a flexure at each corner have less side-to-side movement (runout) than two-flexure designs. But manufacturers may use active compensation schemes to mitigate such effects. In any case, the typical runout specs are in nanometers or tens of nanometers.

In many stages the PZT motion is amplified somehow before it drives the moving platform. One way is to have the actuator stack push against an integrated lever. The lever then amplifies movement by about three-to-one.

Manufacturers take two different approaches to producing two-axis motion with PZT stages. The most straightforward is to stack two stages with one moving in X, the other in Y. One problem with this approach is that movements in the bottom stage tend to be slower than those in the top simply because the lower stage carries the upper stage mass. Additionally, movement of the top stage induces vibrations in the bottom stage which may be difficult to compensate.

The alternative to stacked stages is to employ a single stage that carries parallel kinematics. These stages can produce both X and Y movements. They do so through use of platforms that have flexures that permit X and Y movements and PZT actuators mounted on two sides. This approach, dubbed a parallelogram configuration, better handles runout compared to stacked stages.

Two position sensors gauge X and Y movements of the stage. Their readings feed back to a servocontroller that can compensate for any lateral movements and keep the platform moving in a true straight line. In contrast, there is less opportunity for compensation in stacked stages. Here, the top stage is unaware of any runout in the stage on which it rides.

There are complexities associated with how parallelogram flexure actuators move. Specifically, the platform can move in a slight arc because of second-order cross coupling (parasitic motion) between two axes. This phenomenon can lead to out-of-plane errors on the order of 0.1% of the distance traveled.

Stage manufacturers may devise special guidance systems to handle such difficulties. For example, PI says it has a stage design that keeps flatness and straightness in the nanometer and micro-radian ranges respectively. To get even better figures, servocontrols can compensate for errors. A technique called active trajectory control, also devised by PI, measures and actively controls motion in six degrees of freedom to exhibit subnanometer and submicroradian tolerances.

Sensors Are Key
Most manufacturers use flexure designs in their stages that are geometrically similar. But there are numerous details in construction that are specific to individual vendors.

In particular, many stage makers view as proprietary the means by which they gauge stage position. For example, developers at Mad City Labs Inc., a company named for its roots in Madison, Wis., incorporate special piezoresistive sensors into the stage for position feedback. The sensor configuration gives the stages good linearity and a noise floor in the picometer range. Most Mad City stages go into scan probe microscopy applications working at speeds in the tens of kilohertz.

The sensor used to gauge stage position plays an important role in determining the resolution of the motion system. One typical way of measuring stage position is with a strain gage. The strain gage resides on the flexure. Developers may use an interferometer to calibrate strain-gage output with stage position. This information gets stored as a lookup table. Servoloop software then references the table as it gives the PZT motor position commands.

The problem with strain gages positioned this way is that they contribute a small amount of friction to the stage movement. This friction is enough to generate about 10 to 20 nm of noise over travel distances on the order of 100 m. And the noise scales up with distance. A stage traveling 400 m could see noise levels up to fives times higher.

The noise problems associated with strain-gage sensors have led some stage makers to devise capacitive position sensors. The benefit is that capacitive sensors can be set up so they don't physically touch the stage. Thus there is no perturbation from friction. For example, PI says the capacitive sensors it deploys on its stages can resolve distances to 0.1 nm and better. High-end stages carrying these sensors are said to provide a bidirectional repeatability of 1 nm.

EDO Corp.
(801) 486-7481
Mad City Labs Inc.
(608) 298-0855
Nanomotion Inc.
(800) 821-6266
Physik Instrumente (PI) L.P.
(508) 832-3456
Piezosystem Jena Inc.
(508) 634-6688

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.