The position of the electron beam is driven by a scanner to draw patterns that are defined in a layout. When the electron beam is moved by any other mechanism, it results in errors in the replication of the patterns. The unwanted motion of the electron beam is called Beam Drift.
The primary cause of beam drift is temperature variation. When the components in an electron beam system heats up or cool down, their mechanical and electrical properties changes slightly. One component may become larger, another component may shift a little to the left, the resistance of another component may increase, and so on. The accumulated effect of all these changes will affect the position of the beam. When an electron beam writer is installed in a controlled environment and given time to stabilize, beam drift can be reduced below 50 [nm/hr].
The system administrator can characterize beam drift, by measuring the beam position at specific time intervals. In order to capture bad beam drift, the system was stabilized for at least 24 hours in Mode 4, the high resolution lens; then the system was switched to Mode 2, a different lens. Figure 1 shows a record of the beam position measured every minute. Within the first 30 minutes, the beam drifted beyond the range of the measurement window. At this point, the drift experiment was interrupted to center the beam before resuming the measurements. Nonetheless, it is evident that the beam drift is significant when a lens is turned on and it takes a couple of hours to stabilize. The bottom graph in Figure 1 highlights the stability of the beam position, revealing that after 2 hours, the drift rate never exceeds 100 [nm/hr].
Figure 1: The position of the beam drifts significantly right after energizing the lens (top), but it can be very stable once the lens has warmed up (bottom).
Figure 2 shows a record of the beam position measured every minute when the lens for Mode 4 is energized after the system has stabilized in Mode 2 for at least 24 hours. The drift rate in Mode 4 is approximately 10 times slower than Mode 2 since it has a 10 times smaller scan field. After a couple hours, the drift rate does not exceed 60 [nm/hr]. The bottom graph in Figure 2 highlights a very stable length of time where the drift rate does not exceed 20 [nm/hr].
Figure 2: The position of the beam drifts significantly right after energizing the lens (top), but it can be very stable once the lens has warmed up (bottom).
Whether or not beam drift is a problem depends on the amount of time needed to complete an exposure and the tolerances of the layout. Below is a table that lists the drift rate each hour after the mode is changed. These values are extracted from a single set of measurements, so they should only be used as crude approximations. But in general, our electron beam writer is relatively stable after a couple of hours.
|Hour||Mode 2 [nm/hr]||Mode 4 [nm/hr]|
|0 to 1||10800||1350|
|1 to 2||640||100|
|2 to ...||75||60|