E vs delta point of contact

[SimoneBv]

Hi Paul,
I find your calculation of the E vs delta curves very interesting. Nonetheless, if I got it right, you use as point of contact your first guess, rather then the one that is determined by the fit. In the case of my curves, because of the reflectivity of the sample surface, I often have a detectable laser interference, that affects the most the first guess of the point of contact while is less problematic when it is determined by the fit. The question then is: would it be possible to update the POC position before the calculation of the E vs delta curves? By the way, when you calculate this curve, the POC is fixed for each indentation depth, right?
Thanking you in advance
best
Simone

[paulmueller]

Hi Simone,

thanks for taking the time and writing the issue. I finally have time to look into this.

My first idea would be to determine a better value for the POC by fitting a line and a constant (AFM-analysis/nanite#8) to the approach curve. This is probably faster than fitting e.g. the Hertz model and it could also be extended to a 2nd order polynomial for curves that exhibit a less-linear indentation part.

For that, could you please attach a few exemplary curves (which contain that laser interference) that I can use for testing?

Regarding you final question: Yes, the POC is identical for each indentation depth (It is where the tip position is zero in the plots).

[SimoneBv]

qi-data-2021.04.13-17.09.15.786_043_43.zip
Hello Paul, thanks for your answer to this issue.
I attached to the post an example force curve, acquired on a plant tissue with a spherical tip. In this curve the Hertz fit works well on the whole indentation range and the amplitude of interference is small compared to the maximum force. However interference becomes evident zooming on the baseline and may become troublesome if smaller indentation depths are considered.
For the estimation of the point of contact, what do you mean by fitting a line and a constant? Do you think about making a piece-wise fit using an horizontal and a tilted straight line, letting the intersection between them to be a free parameter of the fit and so and estimation of the POC? If this is the case, I think is a good idea, but only if you only take into account a small part of the contact region, while otherwise the power law behaviour of this part of the curve will have an impact in the positioning of the point of contact.
I think Manfred Radmacher at some point was doing something similar. Alessandro Podesta's group rather go for linearized curves (where F^2/3 vs delta is used, in the case of a spherical indenter) and then make a linear fit of only the first portion of the contact region (20/30% of the setpoint force) to determine the POC (DOI 10.1063/1.4915896). I don't know how fast this procedure can be.
In any case, I think that anyone of those approaches may be more efficient than an estimation of the POC by threshold, that can still be used for a first calculation of the indentation.
Thanks again
Simone
P.S.: If necessary, I can send you more curves or directly some QI maps.

[paulmueller]

Something is odd about this data file. For vDeflection (which translates to "force"), the encoder parameters (instructions on how to convert the vDeflection.dat binary data to Volts) are missing in the metadata. They are there for the other columns head-height, hDeflection, error, capacitiveSensorXPosition, capacitiveSensorYPosition, capacitiveSensorHeight, measuredHeight.

Do you happen to have another test file?

[SimoneBv]

Hello Paul,
I think I get the problem. I have probably exported this curve, acquired during a QI scan, from the analysis process window, so the axes of the curve were probably modified by the operations.
Here is a raw curve from the same QI, where again the Hertz fit works pretty well on the whole indentation.
Please, let me know if this one works out better for you.
Best
Simone
qi-data-2021.04.13-17.09.15.786_170.zip