From Simulation to Concept: A 6-DOF MagLev Stage for Wafer Inspection

One of the most important findings from the project is that MagLev stage development cannot be treated as a standalone motor-design exercise.

In conventional motion systems, motors, mechanics, sensors, and thermal management can often be optimized somewhat independently. Magnetic levitation systems are fundamentally different. Changes to one subsystem immediately influence the others.

For example, modifying the permanent magnet architecture affects mover mass and inertia. Changes in inertia influence the forces and torques required from the motors. Increased force requirements change power consumption and thermal behavior, which can then affect positioning stability.

To address these interactions, the project employed an iterative mechatronic design workflow.

The process begins with a mechanical concept, payload requirements, and desired scanning trajectories. Mover mass properties are then integrated into an electromagnetic simulation model that predicts:

• Force generation
• Torque generation
• Coil currents
• Drive voltages
• Power dissipation
• Thermal behavior

Results from each simulation cycle are fed back into the design, allowing the mechanical and electromagnetic architectures to evolve together until performance objectives converge.

This simulation-driven development approach significantly reduces design risk while helping engineers identify performance bottlenecks before hardware is built.

The stage concept consists of a levitating mover and stationary stator; each divided into dedicated force-generation and metrology structures.

The overall platform measures approximately 580 × 580 × 82 mm and uses a mover with a mass of 14.3 kg prior to adding the 5 kg payload.

The metrology system directly measures all six degrees of freedom:

• Four laser encoders monitor X and Y translation as well as rotation about the Z-axis.
• Four laser interferometers measure Z position and angular motion around the X and Y axes.
• Integrated gratings and precision mirrors provide passive reference structures on the mover.

This architecture enables direct measurement rather than relying solely on motor commutation estimates, improving positioning accuracy and stability.

The separation between force generation and metrology structures also opens opportunities for future thermal isolation and structural optimization. Further studies are planned to evaluate whether compliant decoupling elements are required to minimize the transmission of thermal or mechanical disturbances between subsystems.

The motion platform uses four ironless planar motors arranged at 90-degree intervals.

Each motor incorporates permanent-magnet Halbach arrays, a highly efficient magnetic configuration frequently used in magnetic levitation applications to increase force density while minimizing stray magnetic fields. PI has previously highlighted the advantages of Halbach-based magnetic levitation architectures for precision positioning systems.

The motors operate according to the Lorentz-force principle and deliver highly linear force generation without the hysteresis often associated with iron-core motors.

The X-oriented motor pair primarily controls:

• X-axis motion
• Vertical Z force
• Rotation about X

The Y-oriented pair governs:

• Y-axis motion
• Vertical Z force
• Rotation about Y

Because all degrees of freedom are coupled, the system requires advanced multi-axis control. A proprietary electromagnetic simulation framework calculates the optimal current distribution across every coil based on real-time mover position and the desired six-dimensional force and torque command.

At the same time, the algorithm minimizes electrical losses while ensuring commanded motion does not create unwanted rotational disturbances.

For AFM inspection systems, thermal management is often just as important as positioning performance.

A stage may produce acceptable average heat levels but still introduce measurement errors if thermal loads vary significantly across the scan area. Position-dependent heating can create temperature gradients that distort structures and alter metrology measurements.

To evaluate this effect, the design team simulated a 15 × 15 grid meandering scan trajectory covering the full 50 × 50 mm² operating area.

The results were encouraging.

At a nominal 2 mm air gap:

• Average power loss was approximately 157.4 W.
• Standard deviation across the full trajectory was just 0.8 W.

This small variation suggests highly uniform heat generation across the scanning envelope, reducing one potential source of thermally induced drift. 

Reducing the air gap by 0.5 mm to 1.5 mm lowers the overall loss level by 23 W, while a vertical displacement of 100 μm leads to a small power consumption increase of only 4 W. Across the full X-Y travel range, the loss variation remains small.

The stability of the losses across variations supports the robustness of the design.

One concern with high-speed AFM motion is the additional power required for acceleration and deceleration.

To investigate this effect, dynamic simulations were conducted using 100 mm/s scanning velocity and 0.5 m/s² acceleration profiles.

Surprisingly, the additional motion demand had very little impact on overall power consumption.

Average dissipation increased by only about 1.2 W compared with the quasi-static case, while motor requirements remained below 22 V and 1.7 A.

For system designers, this is a significant result. It suggests that achieving higher throughput does not necessarily require a proportional increase in thermal load, removing a common tradeoff between speed and stability.

Although thermal losses are relatively uniform, approximately 160 W of dissipation still must be managed effectively. Thermal finite-element simulations compared passive cooling with liquid-cooled designs incorporating cooling channels beneath the motor arrays.

Without active cooling, simulated hot spots reached 56.7 °C. Introducing a modest coolant flow of only 0.2 l/min reduced the peak temperature to 38.4 °C, an improvement of roughly 18 K. Further increases in flow rate provided only marginal benefits, indicating that moderate liquid cooling is sufficient to maintain thermal stability while avoiding excessive system complexity.

The project highlights an important trend in precision motion engineering: success increasingly depends on system-level optimization rather than individual component performance. For magnetic levitation stages, the mechanical structure, electromagnetic design, sensor architecture, thermal management, and control algorithms must all be developed together.

Future work will include:

• Structural modal analysis
• Thermal-strain modeling
• Evaluation of ultra-low-expansion materials, such as Zerodur, silicon carbide, and aluminum nitride
• Further optimization of the commutation algorithm to minimize spatial variations in power dissipation

As semiconductor manufacturers push toward ever-higher throughput and tighter dimensional tolerances, these integrated design approaches will become increasingly important.