Drift occurs when there is an undesirable displacement between the AFM probe and the sample. This displacement may be caused by thermal expansion mismatch of components, mechanical drifting of components, or other forms of instability such as piezo lag. For experiments that require accurate and precise measurements, it is important to understand the nature of drift and how it can affect the interpretation of data. By characterizing drift, an operator can make informed decisions to maximize the integrity of the data.
Thermal drift is measured by scanning the same sample over the coarse of hours. A suitable sample is prepared to facilitate automatic drift analysis. The sample comprise of a 500nm x 500nm x 100 nm (LxWxH) film on an SiO2 surface. The 256x256 point image is scanned over a 10 μm field at a scan rate of 2 Hz. The cantilever free air amplitude is approximately 150 nm and the set-point amplitude is approximately 120 nm. Driving the cantilever harder helps keeps the tip engaged over long periods of time. When we used a free air amplitude of 60 nm and a set-point amplitude of 48 nm, the tip loses tracking after a while.
The AFM automatically toggles between scan direction from top-down to down-top. This causes the scan lines between images to be out of sync. To overcome this problem, we use the Asylum Research Macro Builder. The macro builder allows us to force the AFM to always scan top-down. After each scan, the tip remains in contact. From previous experiments, retracting the tip after a scan or keeping it in contact does not affect drift results.
The Asylum Research AFM application is built using the Igor Pro platform. The AFM data is stored as an Igor binary wave file (*.ibw). To compare drift among the several hundred images requires programming. Matlab was used to analyze the dataset since a free library is available for importing Igor binary wave files. Each image is loaded and processed to determine the position of the sensor in the image.
Figure 3: Tracking the sensor position over time reveals drift in the AFM when using the VFM module.
The images for this experiment was stitched together to form a movie. Click here to view the movie. In this movie, the sensor moves towards the AFM left and AFM bottom, or -x and -y respectively. If you are facing the AFM stage, AFM left is your left and AFM bottom is towards you. The time each image is captured is annotated at the bottom right. At approximately 14 seconds into the movie, the image jumps significantly. This jump is the 7 hour rest period shown in Figure 3 where the AFM is idle.