Nanoprobe Network

Huabin - here are my replies, but I will also ask my group members to chime in since they know more about the hands on details - you should trust them more than me!
1) Yes and yes, but with one caveat - when the probe is very far from the surface, the normal force offset will be a little different than when it is close to (but not yet in contact with) the surface. This is because there is some light scattered off the surface and the amount changes with the surface-probe distance. So the best base-line value to take is when the probe is just pulled off from the surface. You'll see this value in a force curve. I believe the value will not be the raw photodiode normal signal, but the normal signal minus the setpoint.

For questions 2 and 3, are you referring to a regular friction measurement, or to scanning during the wedge calibration method?

4) Yes, changing the set point will change the deflection value because it is automatically subtracted. So if you are taking a series of discrete images at different set points, write down the value you used for each measurement and use this to reconstruct the true deflection. To get the baseline, do the force curve measurement with the set point set to zero. However, we use another method where we ramp the set point continuously during an image, so each line is taken at a different normal load. This requires using the signal access module. Do you have one?

5) If the feedback is on and optimized, there shouldn't be a large dependence in the deflection signal; you will see a "lag" as the probe goes uphill or downhill. This gets worse if you scan faster, or turn down the feedback. For scanning the wedge sample, it should be possible to have high enough feedback and to scan slowly enough that this does not lead to a large error.

(1) Did you mean that the signal of the baseline (out of contact) of a force curve can be taken as the offset value? Could the signal of the photo diode when the cantilever is off surface be used as the offset value?

I always took a series of force curves right before friction measurements and used the "setpoint zero" function to soft-code the out-of-contact PSD voltage into header of the friction data file (actual setpoint was set using a function generator). Of course you can always just write the value down, but that's not as elegant. Because I only took friction-vs-load measurements, took the liberty of taking the setpoint down below the pulloff force within each image so that some actual out-of-contact data were contained in each image, too.


(2) In your work, by which means did you get the deflection signals of the cantilever in a scanning process?

Torsional deflection was the "friction" signal, naturally. The "deflection" signal is actually the deflection error signal (deflection minus setpoint) so we used a breakout box, extracted the deflection signal, and ran it into an aux input on the back of the controller.

(3) According to the manual of Multimode AFM (please see pages 196-197 of the manual sent to you by email), the data of an image captured with the Data type set to Deflection could be used to calculate the force exerted on a sample (or deflection of the cantilever) if the feedback gains are set low when collecting the image. It seems a bit difficult here, how to set the gains ideally low?

I would not use that technique.

(4) As you know, in order to calibrate the cantilever, a series of set-point values are usually used in collecting the same image in the friction measurement. In my experiments, I found that the baseline of force curves changes with set-point values. If the offset value mentioned above is obtained by the consideration of the out of contact (baseline) of a force curve (collected immediately upon the completion of imaging), then, at which set-point should a force curve be collected, or how to collect a force curve of which the baseline can be used as the offset value?

The green horizontal axis of the force curve plots is not fixed at zero load (or any other load for that matter), it's just a representation of the setpoint. So as you vary the setpoint, the baseline will move up and down. This does not affect the measurement itself.

(5) Personally, I think that the deflection (load) of the cantilever in the same fast scan line (same set-point) not only depends on set-point, but also depends on the surface features of a sample. Does it make sense?

That's correct. In an ideal measurement, the cantilever deflection always equals the setpoint, and the scanner tracks the topography perfectly. In practice of course there is a lag between topog. changes > cantilever bending > feedback loop > piezo response > cantilever return to desired deflection. The method described in question (3) relies on and exaggerates this lag so that changes in topography force the cantilever into undesired deflections which are then recorded by the "deflection" channel -- which is really the error signal. It's a terrible method that relies on intentionally poor performance of the system as it responds to uncontrolled imperfections in the sample.

Like Rob said, we used an external function generator and a breakout box to ramp the setpoint indirectly -- this allowed the feedback loop to function properly. I forget exactly how it was done, but I know it also required a summing amplifier and physically adding a circuit to the feedback board in the controller. (We added a BNC port to the front of the controller with the center lead somehow connected to the setpoint signal.) I think this is because the error signal is actually calculated at the multimode/dimension scanner, but I forget the specifics.

Hope that helps
Matt Brukman