Is it possible that the rows in an AFM scan could be misaligned? For example, could the image of a square produced by an AFM end up being a rhombus? Also, how accurate is the aspect ratio of the image? For example, could a circle end up looking like an ellipse in the image? My apologies in advance if the answers to these questions are obvious --- I'm a theorist.
AFM imaging is subject to thermal and temporal drifts, and thus distortion. For example, if temperature of sample is appreciably different from the temperature of the AFM, or if sample is held by some not-very-stable adhesive, such as double-sided tape, then seeing a diagonal ellipse instead of circle or a rhombus instead of square would be very much expected...
I totally agree with Valery, if a squared structure appears as a rhombus, thus indicates the presence of drift, probably thermally induced drift. You could try to use image processing software (e.g. Gwyddion) to remove these error. Or try to wait for the whole system to stabilize!
Thanks Heinz, Valery and Florian for the input. How does thermal drift occur? Does the differing temperature of the sample and AFM lead to a change in the piezo temperature, and so to changes in the scan rate?
No, I don't think that this could be the problem. All components of the AFM, and also the sample, retract/extend differently due to thermal fluctuations and different material coefficients. This induces a time variant motion of the sample in relation to the AFM tip. This motion is slow, but can produce artifacts similar to those you described in your question. The relative movement of the tip to the sample does not really behave linear in the long-term, but for short time periods it may be almost linear thus producing rhombus-like structures in the AFM image for squared objects. You could also check what the image looks like when you change the scanning direction top-down vs bottom-up.
I would generally agree with all of the above. There are several centimeters (often 10 or more) along the solid path that holds the tip above the sample. So, even a small temperature change during the scan can cause several nm of drift. For a linear drift it is possible to correct for drift by taking scans in opposing directions (up/left and down/right). See J. Vac. Sci. Technol. B 16, 51-53 (1998) J. T. Woodward and D. K. Schwartz. There have also been a number of good papers since that are more extensive. I would also add as another possible source of distortion - tilt between the sdample and the plane of the scanner. There is both the geometrical effect of projecting the tilted sample onto the image plane and, for tube scanners, a coupling between the z position of the scanner and the x-y scan such that if there is a large change in z during the scan, this can distort the x-y scan.
With the above points I can add one thing that during the AFM scan while scan direction is changed from top-down to bottom-up if one finds a change in pattern orientation will indicate the presence of thermal drift. This drift may die out with time.
Hello, Mark.
In my practice such distortions appear when the sample is mounted by the double-side tap or maybe blu-tack. If you use such adhesives too, try to perform AFM measurements with unmounted sample — just put the sample on the holder, without any adhesive and start topography measurement. This experiment helps you to remove the effect of adhesive on the results.
If distorsions of AFM images correspond to vertical or horizontal direction strictly and are not dependent on scan direction (horizontal/vertical) and scan rate, check the scanner calibrations of Ox and Oy axes.