Thermal behaviour of piezo actuators: the 60°C stator shift problem

Every ultrasonic piezo motor is a resonant device. It operates at a specific mechanical resonance frequency, typically between 20 kHz and 200 kHz, where the stator vibration amplitude peaks and the motor produces useful force and velocity. That resonant frequency is not fixed. It shifts with temperature, and the shift is large enough to take the motor out of its operating band if the drive electronics cannot track it. This is the 60 °C stator shift problem: at roughly 60 °C above ambient, many PZT-based stators have shifted their resonant frequency by 1 to 3%, enough to halve the motor's force output or stall it entirely. Understanding and managing this behaviour is essential for any piezo motor application that operates outside a narrow temperature window.

Image: Nanomotion Ltd.

Why resonant frequency changes with temperature

The resonant frequency of a vibrating stator depends on its geometry, mass, and elastic stiffness. For a simplified beam-type stator, the fundamental frequency scales as:

f ∝ √(E / ρ) × (t / L²)

where E is the effective Young's modulus, ρ is the density, t is the thickness, and L is the length. Temperature affects both E and ρ (through thermal expansion), but the dominant effect is the change in the elastic constants of the PZT ceramic.

The softening of PZT with temperature

PZT (lead zirconate titanate) is a ferroelectric material. Its elastic, dielectric, and piezoelectric properties are all temperature-dependent, and the dependencies are not linear. As temperature increases from room temperature toward the Curie temperature:

• The elastic compliance (s₁₁, s₃₃) increases, meaning the material becomes softer.
• The piezoelectric coefficients (d₃₁, d₃₃) increase, meaning the material becomes more responsive to applied voltage.
• The dielectric permittivity (ε₃₃) increases, meaning the material draws more current at a given voltage and frequency.
• The mechanical quality factor (Qm) decreases, meaning losses increase and the resonance peak broadens.

The net effect on resonant frequency is a downward shift. For common "hard" PZT compositions (PZT-4, PZT-8, and their commercial equivalents), the resonant frequency decreases by approximately 0.02 to 0.05% per degree Celsius. This sounds small, but it accumulates:

Temperature rise (°C)	Frequency shift (%)	Shift for 40 kHz stator (Hz)
10	0.2 to 0.5	80 to 200
30	0.6 to 1.5	240 to 600
60	1.2 to 3.0	480 to 1200
100	2.0 to 5.0	800 to 2000

A 40 kHz motor with a mechanical Q of 500 has a 3 dB bandwidth of only 80 Hz. A frequency shift of 500 Hz moves the operating point completely off the resonance peak, reducing vibration amplitude by an order of magnitude. The motor effectively stops working.

Contributions from the metal stator body

The PZT ceramic is bonded to a metal stator body (typically stainless steel, phosphor bronze, or titanium). The metal's elastic modulus also decreases with temperature, but by a smaller fraction: roughly 0.02 to 0.04% per degree Celsius for stainless steel. Since the stator is a composite structure, the overall frequency shift is a weighted average of the PZT and metal contributions. In practice, the PZT dominates because its temperature coefficients are 2 to 5 times larger than those of the metal.

Thermal expansion of the stator body changes its dimensions and therefore its resonant frequency. For stainless steel with a CTE of 16 × 10⁻⁶ /°C, a 60 °C rise changes the length by roughly 0.1%. This contributes a frequency shift of approximately 0.05%, small compared to the PZT stiffness change, but not negligible.

The Curie temperature ceiling

Every ferroelectric material has a Curie temperature (Tc) above which it loses its piezoelectric properties permanently. The ceramic undergoes a phase transition from the ferroelectric (tetragonal or rhombohedral) phase to the paraelectric (cubic) phase, and the spontaneous polarization vanishes. For common PZT compositions:

Material	Curie temperature (°C)
PZT-4 (hard)	328
PZT-5A (soft)	365
PZT-8 (hard)	300
PZT-5H (soft)	193
Lithium niobate	1150
BaTiO₃	120

For ultrasonic motors, hard PZT compositions (PZT-4, PZT-8) are preferred because of their high mechanical quality factor, which enables large vibration amplitudes at resonance. Their Curie temperatures of 300 to 330 °C provide substantial margin above normal operating temperatures.

However, the usable temperature limit is well below Tc. As the temperature approaches Tc, the piezoelectric coefficients become unstable, hysteresis increases, and aging accelerates. A practical rule is to limit the maximum PZT temperature to half the Curie temperature on the absolute scale, or roughly Tc minus 100 °C, whichever is lower. For PZT-8 (Tc = 300 °C), this means a maximum operating temperature of approximately 200 °C. For PZT-5H (Tc = 193 °C), the limit drops to roughly 100 °C, which is uncomfortably close to the temperatures reached during high-duty-cycle operation.

Depolarization below Curie temperature

Partial depolarization can occur at temperatures significantly below Tc, especially under sustained mechanical stress or strong AC electric fields. This is a concern for ultrasonic motors because the PZT operates simultaneously under high AC drive (10 to 200 V peak, at resonance) and mechanical stress from the preload force. Combined thermo-electro-mechanical loading can cause domain reorientation that partially depolarizes the ceramic over time. The depolarization is not instantaneous; it manifests as a gradual loss of motor performance over thousands of hours.

The practical implication: a motor that works fine at 80 °C during a short test may exhibit measurably degraded performance after 1000 hours at the same temperature due to cumulative partial depolarization.

Image: Nanomotion Ltd.

Thermal expansion and its effect on positioning

Stator dimensional changes

Thermal expansion of the stator directly affects the contact geometry between the friction tip and the runner. As the stator expands, the preload changes, which affects both force output and friction-interface wear. For a stator with 10 mm between the mounting point and the friction tip, a 60 °C temperature rise causes:

• Stainless steel: 10 mm × 16 × 10⁻⁶ /°C × 60 °C = 9.6 µm
• Titanium: 10 mm × 8.6 × 10⁻⁶ /°C × 60 °C = 5.2 µm
• Invar: 10 mm × 1.2 × 10⁻⁶ /°C × 60 °C = 0.7 µm

A 10 µm change in the preload gap translates directly to a change in preload spring deflection. If the preload spring has a rate of 5 N/mm, this is a 0.05 N change, which can be significant for miniature motors with nominal preloads of 0.5 to 2 N.

Stage-level thermal drift

Positioning accuracy is often limited not by the motor itself but by thermal expansion of the stage structure. An aluminium stage base 200 mm long expands by 200 × 23 × 10⁻⁶ × 1 °C = 4.6 µm per degree Celsius. For nanometre-level positioning, even fractional-degree temperature changes are significant.

This problem is common to all motor technologies, not unique to piezo. But it interacts with piezo-specific effects in two ways:

1. The motor's self-heating contributes to stage thermal drift. A piezo motor dissipating 0.5 W in a thermally isolated stage can raise the stage temperature by several degrees.
2. The motor's resonant frequency shift due to self-heating can cause the drive electronics to compensate by changing drive amplitude or frequency, which alters the motor's heat dissipation, creating a feedback loop.

The 60 °C problem in practice

The "60 °C stator shift problem" refers specifically to the scenario where a motor, designed and tuned for room-temperature operation, is deployed in an environment 60 °C warmer (or heats itself to 60 °C above ambient during operation). At this temperature rise, the combination of resonant frequency shift, reduced Qm, and preload change can reduce motor performance by 30 to 70%.

Case study: semiconductor wafer handling

A piezo linear motor driving a wafer transfer arm operates in a process module at 80 °C ambient (room temperature plus process heating). The motor was characterized at 25 °C, where it achieved 5 N force at 50 mm/s. At 80 °C (ΔT = 55 °C):

• Resonant frequency dropped by 1.8% (from 42.0 kHz to 41.24 kHz).
• The drive electronics, using a fixed-frequency oscillator, were still driving at 42.0 kHz.
• Vibration amplitude dropped to 35% of its room-temperature value.
• Motor force dropped to approximately 1.8 N, insufficient for reliable wafer transport.
• The solution: replacing the fixed-frequency driver with a phase-locked loop (PLL) that tracks the resonant frequency automatically.

This case illustrates why frequency tracking is not optional for any application with temperature variation greater than 10 to 15 °C.

Drive electronics: frequency tracking strategies

Phase-locked loop (PLL)

The most common approach uses a PLL to lock the drive frequency to the motor's resonant frequency. The PLL monitors the phase relationship between the drive voltage and the motor current. At resonance, the phase difference is near zero (for a series resonance) or reaches a characteristic value (for a parallel resonance). The PLL adjusts the drive frequency to maintain this phase relationship as the resonant frequency drifts.

PLL tracking works well for slow temperature changes (minutes to hours). It can track frequency shifts of several percent, more than sufficient for thermal drift. The main limitation is that the PLL can lose lock during rapid temperature transients or if the Q factor drops so low that the phase transition at resonance becomes too gradual to detect reliably.

Typical PLL tracking bandwidth: 100 to 1000 Hz update rate, capable of following resonant frequency changes of up to 100 Hz/s.

Admittance (impedance) tracking

An alternative approach measures the electrical admittance of the motor at multiple frequencies and identifies the resonant peak directly. This is more robust than PLL in low-Q conditions but requires a more complex excitation scheme (frequency sweeps or multi-tone excitation). It is used in high-reliability applications where PLL lock loss is unacceptable.

Digital frequency scanning

Modern digital drive electronics can periodically sweep the drive frequency across a window around the expected resonance, measure the motor response (vibration amplitude, current draw, or velocity), and adjust the operating frequency to the optimum point. This approach is robust and flexible but introduces brief interruptions in motor operation during the sweep. For intermittent positioning applications, this is acceptable; for continuous scanning, it is not.

Temperature feedforward

If the stator temperature is measured (by an embedded thermocouple or thermistor, or inferred from the drive current), the expected frequency shift can be predicted from the known temperature coefficient and applied as a feedforward correction to the drive frequency. This reduces the burden on the PLL or impedance tracker and improves transient response. The temperature coefficient must be calibrated for each motor design, as it depends on the PZT composition, stator geometry, and bonding method.

Thermal management strategies

Reducing heat generation

The primary heat sources in an ultrasonic piezo motor are:

1. Dielectric losses in PZT. The PZT dissipates power proportional to the square of the drive voltage, the frequency, and the loss tangent (tan δ). For hard PZT at 40 kHz and 100 V peak, typical dissipation is 0.1 to 1 W.
2. Friction at the contact interface. The sliding contact between the stator tip and runner generates frictional heat proportional to the normal force, sliding velocity, and friction coefficient. This can be 0.1 to 0.5 W during active motion.
3. Structural damping. Internal friction in adhesive layers, bolted joints, and other structural elements contributes a smaller amount.

Strategies to reduce heat generation:

• Minimize drive voltage. Operate at the lowest voltage that provides adequate force. Dielectric losses scale with V².
• Use hard PZT. PZT-8 has a loss tangent roughly half that of PZT-4 and one-fifth that of PZT-5A.
• Reduce duty cycle. If the application permits, move the motor quickly and then hold position with the power off (the motor's self-locking property provides holding force without power).
• Optimize preload. Excessive preload increases frictional heating without proportionally increasing useful force.

Removing heat

• Conductive paths. Mount the stator on a thermally conductive base with good contact (thermal grease or indium foil interface in vacuum). Aluminium and copper brackets can conduct heat to a heat sink or the machine frame.
• Forced air cooling (in atmospheric applications). Even low-velocity airflow dramatically reduces stator temperature rise. A small fan delivering 1 m/s airflow can reduce the temperature rise by 50 to 70%.
• Thermal straps (in vacuum applications). Flexible copper braids or straps connect the motor mount to a temperature-controlled surface.
• Peltier coolers. For precision applications where stator temperature must be held within 1 to 2 °C, a thermoelectric cooler can actively regulate the motor temperature. This adds complexity and power consumption but eliminates thermal drift.

Design for thermal stability

• Symmetric stator design. A thermally symmetric stator expands uniformly, maintaining the mode shape even as the dimensions change. Asymmetric heating causes mode distortion that reduces motor efficiency beyond what frequency shift alone would predict.
• Invar or titanium stator bodies. Low-CTE materials reduce dimensional changes and preload shifts. Titanium (CTE = 8.6 × 10⁻⁶ /°C) is roughly half the expansion of stainless steel.
• Temperature-compensated preload. A preload mechanism designed so that thermal expansion of the motor is offset by expansion of the preload spring mount can maintain constant preload over a temperature range. This requires careful mechanical design and material selection.
• Thermally decoupled mounting. Isolating the motor from the stage structure with low-conductivity spacers (e.g., ceramic washers) prevents motor self-heating from affecting the stage's positional stability. The tradeoff is that the motor itself runs hotter.

Temperature ranges for different PZT compositions

The choice of PZT composition determines the usable temperature range:

Standard hard PZT (PZT-4, PZT-8): -40 °C to +200 °C

Suitable for the vast majority of industrial and scientific applications. Performance degrades gracefully with temperature; no cliff-edge failure below 200 °C. Resonant frequency tracking is required above ΔT = 15 °C from the calibration temperature.

Soft PZT (PZT-5A, PZT-5H): -40 °C to +80 °C

Soft compositions are rarely used in ultrasonic motors because of their low Qm, but they appear in some sensor and actuator hybrid designs. Their lower Curie temperatures and higher temperature sensitivity make them unsuitable for elevated-temperature service.

High-temperature compositions: up to 300 °C

Bismuth titanate (Bi₄Ti₃O₁₂), modified PZT with manganese or antimony doping, and other specialized compositions can operate at higher temperatures. These are used in downhole oil and gas applications, jet engine monitoring, and nuclear instrumentation. Their piezoelectric coefficients are lower than standard PZT, so motor force and velocity are reduced.

Cryogenic operation: down to 4 K

PZT retains piezoelectric activity at cryogenic temperatures, but the coefficients decrease significantly. At 77 K (liquid nitrogen), d₃₃ is typically 60 to 70% of the room-temperature value. At 4 K (liquid helium), it drops to 40 to 50%. The resonant frequency increases by 2 to 4% due to stiffening of the ceramic. These changes are predictable and can be compensated, making cryogenic piezo motors practical for applications in cryostats, dilution refrigerators, and space instruments.

Worked example: predicting stator frequency shift

A Langevin-type ultrasonic motor has a room-temperature resonant frequency of 40.00 kHz. The stator uses PZT-8 ceramics bonded to a stainless steel body. The temperature coefficient of resonant frequency has been measured at -0.03%/°C.

Problem: The motor will operate in an enclosure where ambient temperature varies from 20 °C to 50 °C. During operation, stator self-heating adds another 15 °C. What frequency tracking range must the drive electronics support?

Solution:

• Minimum stator temperature: 20 °C (ambient, motor off)
• Maximum stator temperature: 50 + 15 = 65 °C (worst-case ambient plus self-heating)
• Temperature range: 65 - 20 = 45 °C
• Frequency shift: 40,000 Hz × 0.03% × 45 = 540 Hz
• Required tracking range: 40,000 Hz to 39,460 Hz (1.35% below nominal)

The drive electronics should support a tracking range of at least ±1.5% (±600 Hz) to provide margin. A PLL with a capture range of 39,000 to 41,000 Hz (±2.5%) would be conservative and is readily achievable.

Implications for Qm: If the mechanical Q at room temperature is 800, the 3 dB bandwidth is 40,000/800 = 50 Hz. The PLL must track frequency shifts 10 times larger than the bandwidth, confirming that fixed-frequency drive is completely impractical for this application.

Thermal effects on motor lifetime

Elevated temperature accelerates several degradation mechanisms:

1. Adhesive bond degradation. Epoxy bonds between PZT and metal weaken at elevated temperatures. Most structural epoxies begin to lose strength above their glass transition temperature (Tg), which ranges from 80 to 200 °C depending on formulation. Operating within 20 °C of Tg accelerates creep and fatigue.

2. PZT aging acceleration. Piezoelectric ceramics exhibit aging (gradual loss of polarization and properties over time). Aging rates follow an Arrhenius relationship, roughly doubling for every 20 to 30 °C increase above the poling temperature. At 100 °C, a PZT-8 element may age at 5 to 10 times the rate observed at 25 °C.

3. Friction interface wear acceleration. Ceramic wear rates increase with temperature due to changes in the surface chemistry and mechanical properties of the contacting materials. Alumina, for instance, transitions from predominantly mechanical wear to tribochemical wear above approximately 200 °C in humid environments.

4. Thermal cycling fatigue. Repeated heating and cooling cycles stress the PZT-metal bond due to CTE mismatch. PZT has a CTE of approximately 4 × 10⁻⁶ /°C; stainless steel is 16 × 10⁻⁶ /°C. The resulting shear stress at the interface is proportional to the CTE difference, the temperature swing, and the bond area. Over thousands of cycles, this can cause delamination.

Summary

Temperature is the single most important environmental variable affecting ultrasonic piezo motor performance. The resonant frequency shift caused by PZT softening at elevated temperatures, roughly 0.03%/°C for hard compositions, can move the operating point off resonance and stall the motor within a 60 °C temperature rise. Effective thermal management combines three strategies: minimizing heat generation through proper drive parameters and duty cycle management; removing heat through conductive paths and (where available) forced convection; and tracking the resonant frequency electronically using PLL or impedance methods. With these measures in place, piezo motors operate reliably across temperature ranges from cryogenic to 200 °C, covering the vast majority of industrial, scientific, and aerospace applications.