100mm ALD Prototype
In conventional ALD processes, wafers cannot move to another space during deposition. This is because wafers can be contaminated by gas or particles during transfer.
But if we could transfer wafers without contamination, process efficiency would increase dramatically. So, we decided to prove this hypothesis by developing a prototype device.
Goals
• Wafer transfer during process
• Rotating chamber
• Maintain vacuum while the chamber rotates
Challenges
There were two main technical challenges.
First, rotate the chamber stably.Second, maintain a consistent vacuum during rotation.
Since there were no existing examples, we had to find the right solutions ourselves.
Design
Selecting gears and motors for chamber rotation
To ensure stable chamber rotation, we decided to use a servomotor and internal gear. We selected suitable gears and a servomotor based on the chamber’s weight and calculated gear ratio.
Maintaining vacuum with O-rings and magnetic fluid
Designing the O-ring component to maintain vacuum was not difficult, but we could not use O-rings on rotating parts for one reason: O-rings on rotating parts generate many particles due to friction with the contact surface. These particles can damage the wafer and directly affect equipment reliability.
So, we needed a solution that met two criteria:
1. No particle generation
2. Maintain vacuum during rotation
Among various candidates, we judged that magnetic fluid would meet these criteria. When magnetized, the fluid is pulled into a certain shape by the magnetic field.
This allowed us to maintain a vacuum in the drive section without using O-rings. Similar solutions have been used by NASA for spacesuit sealing, so we believed it was applicable.
Final Solution
We designed the final product to meet all requirements. We applied magnetic fluid to the chamber groove and placed magnets below to hold the fluid in shape. The shaped magnetic fluid maintained the vacuum during rotation without generating particles.
Maintaining vacuum with O-rings and magnetic fluid
Manufacturing
We completed detailed designs considering four manufacturing processes:
• MCT
• CNC lathe
• Wire cutting
• Welding
After about three months of manufacturing, we assembled the final equipment.
Reflection
During validation, we found two problems. First, the magnetic fluid lost its properties at a certain vacuum level, causing the vacuum to break. Second, the wafer was deformed by the magnet due to insufficient shielding.
Since these issues were related to the nature of magnetic fluid, we had to find a different solution. After about eight months of further research, we designed a new rotating chamber using electromagnets.